Predation in the Hymenoptera: An Evolutionary Perspective by Research Signpost

PREDATION IN THE HYMENOPTERA: AN EVOLUTIONARY PERSPECTIVE EDITOR CARLO POLIDORI

Cover picture Picture credits, from up on the left to down on the right: D. Santoro, A. Andrietti, A. Dejean, D. Santoro.

Predation in the Hymenoptera: An Evolutionary Perspective

Editor

Carlo Polidori Dipartimento di Biologia, Sezione di Zoologia e Citologia, UniversitĂ degli Studi di Milano, via Celoria 26, 20133 Milano, Italy

Transworld Research Network, T.C. 37/661 (2), Fort P.O., Trivandrum-695 023 Kerala, India

Published by Transworld Research Network 2011; Rights Reserved Transworld Research Network T.C. 37/661(2), Fort P.O., Trivandrum-695 023, Kerala, India Editor Carlo Polidori Managing Editor S.G. Pandalai Publication Manager A. Gayathri Transworld Research Network and the Editor assume no responsibility for the opinions and statements advanced by contributors ISBN: 978-81-7895-530-8

Preface Will you believe it [â&#x20AC;Ś] not even the very smallest mistake had been made by the wise Wasp. What can we not learn from this intelligent Industry in so tiny an insect! (J. H. Fabre, The Hunting Wasps) Natural world is as we know it because of the many types of mutualistic and antagonistic interactions between its living parts, the organisms. Individuals of each species share their environment with many other individuals of the same and of different species, with which they collaborate for mutual benefits (e.g. pollination, mutualistic host-endosymbiont systems, sociality) or that they eat and by which they are eaten (e.g. grazing, predation, parasitism). Predation, defined as the antagonistic interaction in which one organism captures and feeds upon another (called prey) [1], is certainly one of the ecological interactions that contributes more in structuring populations, communities, and whole environments. In insects, a vast number of extant species is actually composed of individuals who prey upon other organisms at least during some life stages, and almost every order of insects includes some predatory species, with some orders exclusively composed of predators [2,3]. In the classic definition [1], this means that individuals of these species attack other organisms, kill them, and feed upon them to acquire the necessary energy to guarantee their metabolic functions. However, such definition does not apply exactly to all the organisms that we call â&#x20AC;&#x153;predatorsâ&#x20AC;?. The Hymenoptera, a large order of holometabolous insects, include many species of predators [2,4]. For example, in literature, digger wasps, paper wasps and army ants are all named predators [5]. But do they match the classic definition of predation? Actually, individuals (typically females) of these species hunt other organisms primarily (though not exclusively in some cases) to feed their immature and almost immobile offspring, and only rarely they enjoy the meal they collect. Instead, adult females acquire energy mainly from nectar sources, and then spend a large part of such energy to convert themselves in formidable killers. From a certain point of view, such life-history trait is similar to parasitoidism, where females search for a host and oviposit on or into it to guarantee food for the offspring [6], but with an important difference: predatory Hymenoptera actually carry their prey to the offspring rather than carry their eggs to the hosts [7].

This peculiar meaning of the term â&#x20AC;&#x153;predationâ&#x20AC;? for Hymenoptera fascinated plenty of students and researchers through centuries, and contributions on the knowledge of predatory and foraging behaviour of these insects continuously occupy an important amount of pages in the scientific journals all over the world. However, in the last 20 years, reviews on predatory Hymenoptera only were included in books strongly oriented to specific taxa (i.e. apoid wasps, ants, social wasps) or to certain behavioural categories (e.g. solitary groups vs. social groups) [5,8,9,10,11,12,13]. This strongly contrasts with the much more numerous volumes published in the same period on the foraging ecology of parasitoids [e.g. 7,14,15 and references therein]. The reason behind this bias is clear: parasitoids can be used to control insect pests much more efficiently than predators, and thus attract more attention. The idea for this book was to concentrate exclusively on predatory Hymenoptera as a whole, that is, discussing those groups in which females search actively for prey, i.e. any animal or plant individual falling victim of the predator, and then carry it to their offspring. Parasitoids are not considered except for comparative purpose with predatory species or in an evolutionary context. This criterion in defining a predatory hymenopteran ultimately leads to the discussion of the following groups of aculeate Hymenoptera (Aculeata): Apoidea (e.g. digger wasps) and Vespoidea (e.g. social wasps and ants). Given this, the book aims, from one side, to critically review what we know and what we should know about predation in the Hymenoptera across the different taxa and, from the other side, to try to answer to a number of evolutionary, ecological and/or behavioural questions about predation in this group of animals. Predation has evolved independently several times in the order, but how many times it evolved and which organisms were first used as prey? (Chapter 1). Wasps which carry repeatedly their prey in flight to the nests (central-place foragers) face the problem to not exceed the maximum load theoretically possible, but do they maximize their load-lifting capacity while hunting a prey in accordance with general foraging theory? (Chapter 2). Ants are exceptional predators which co-evolved morphology and foraging behaviour with their prey: how ant phenotypic diversity is structured in relation to prey types? (Chapter 3). Seed-harvesting ants are undergoing a long-time coevolution with their prey plants, so why some species are generalist predators and other specialized? (Chapter 4). Despite wasp populations may appear generalist in the choice of prey, they can be composed of specialized individuals: is individual specialization widespread in wasps? And does wasp size account or even can predict the level of inter-individual diet variation? (Chapter 5). Once a prey is found, wasps have to sting it and paralyze it in order to provide alive-but-immobile fresh meal for their offspring: how the

stinging patterns behaviourally differ among species hunting for different prey taxa? (Chapter 6). Colonies of social wasps may have different food demands compared to solitary relative species: does prey taxonomic spectrum vary with social organization in digger wasps? (Chapter 7). And finally, predatory hymenopterans not only hunt, but they are also hunted by a very different spectrum of organisms: Do the spectrum and diversity of such natural enemies differ among hymenopteran groups? And how hymenopterans evolved in response to risk of predation and parasitism? (Chapter 8). I would like to convey my deep gratefulness to all people who participated in the achievement of this book. First, the authors of the chapters of this volume have been extremely cooperative and did a great job, writing and revising their chapters within reasonable times. Second, the referees contacted by me for chaptersâ&#x20AC;&#x2122; reviews were extremely kind, gave many important suggestions, and also did their work in reasonable times. Third, I had the luck, during the book editing, to spend periods in different universities where colleagues constructively and critically discussed with me about the book idea, its organization, and its aims. Carlo Polidori

References 1. 2. 3. 4. 5. 6. 7.

8. 9.

Elewa, A.M. T. (Ed.) 2006, Predation in Organisms. A Distinct Phenomenon, Springer-Verlag, Heidelberg, Germany. Foottit, R.G., and Adler, P.H. (Eds.) 2009, Insect Biodiversity: Science and Society, Wiley-Blackwell, Canada. Schowalter T.D. (Ed.) 2006, Insect Ecology. An Ecosystem Approach, Academic Press, London, UK. Austin, A.D, Dowton M. (Eds.) 2000, Hymenoptera: Evolution, Biodiversity and Biological Control, CSIRO Publishing, Canberra. Oâ&#x20AC;&#x2122;Neill, K.M. 2001, Solitary wasps: Behavior and natural history. Comstock Publishing Associates, Ithaca, New York. Reuter, O.M. 1913, Lebensgewohnheiten und Instinkte der Insekten. Friedlander, Berlin. Wajnberg, E., Bernstein, C., Van Alphen, J.J.M. (Eds.) 2008, Behavioral ecology of insect parasitoids: from theoretical approaches to field applications, Blackwell Publishing, Oxford, UK. Evans, H.E., O'Neill, K.M. 2007, The Sand Wasps: Natural History and Behavior, Comstock Publishing Associates, Ithaca, New York. Holldobler, B., and Wilson, E.O. 1990, The Ants, Springer-Verlag, Berlin, Germany.

10. Keller, L., and Gordon, E. 2010, The Lives of Ants, Oxford University Press, UK. 11. Lach, L., Parr, C., Abbott, K. (Eds.) 2010, Ant Ecology, Oxford University Press, UK. 12. Jarau, S., and Hrncir, M. (Eds.) 2009, Food Exploitation By Social Insects: Ecological, Behavioral, and Theoretical Approaches, CRC Press, Boca Raton, FL. 13. Hunt, J.H. 2007, The Evolution of Social Wasps, Oxford University Press, UK. 14. Godfray, H.C.J. 1994, Parasitoids: Behavioral and Evolutionary Ecology. Princeton University Press, New Jersey, USA. 15. Hochberg, M.E., and Ives A.R. (Eds.) 2000, Parasitoid Population Biology. Princeton University Press, New Jersey, USA.

Contents

Chapter 1 Origin and occurrence of predation among Hymenoptera: A phylogenetic perspective Gabriel A. R. Melo, Marcel G. Hermes and Bolivar R. Garcete-Barrett Chapter 2 Effects of prey size and load carriage on the evolution of foraging strategies in wasps Joseph R. Coelho Chapter 3 Predation by ants on arthropods and other animals Xim Cerdá and Alain Dejean Chapter 4 A fable on voracious and gourmet ants: Ant-seed interactions from predation to dispersal Francisco M. Azcárate and Pablo Manzano Chapter 5 Individual prey specialization in wasps: Predator size is a weak predictor of taxonomic niche width and niche overlap Carlo Polidori, Davide Santoro, Josep Daniel Asís and José Tormos

101

Chapter 6 The art of managing weapons: The stinging behaviour of solitary wasps in the eyes of past, present and future research Francesco Andrietti Chapter 7 The role of increased prey spectrum and reduced prey size in the evolution of sociality in Cerceris wasps Carlo Polidori Chapter 8 Predators as prey: Top-down effects on predatory Hymenoptera Heike Feldhaar

123

199

217

Transworld Research Network 37/661 (2), Fort P.O. Trivandrum-695 023 Kerala, India

Predation in the Hymenoptera: An Evolutionary Perspective, 2011: 1-22 ISBN: 978-81-7895-530-8 Editor: Carlo Polidori

1. Origin and occurrence of predation among Hymenoptera: A phylogenetic perspective Gabriel A. R. Melo, Marcel G. Hermes and Bolivar R. Garcete-Barrett Departamento de Zoologia, Universidade Federal do Paraná, Cx. postal 19020 81531-980, Curitiba, PR, Brazil

Abstract. Predation in its narrowest sense—where the predator attacks and consumes its own prey— is proportionally less important in Hymenoptera, being restricted to adult feeding behaviors such as host feeding with lethal consequences in a few ichneumonids and dryinids, and direct predation by adult tenthredinids hunting other insects for self nutrition. Predation in Hymenoptera occurs mostly for larval feeding and is carried out by adult wasps acting as provisioning predators. Provisioning predation has undoubtedly evolved from the parasitoid mode of life and one of the key features in this transition involves the relocation of the host. Host relocation—a term used here to characterize a series of behaviors associated with movement of the host (or prey), commonly to a concealed site, from the place where it was originally found and subdued—is found only among the aculeate Hymenoptera. Here, the evolutionary pathways that probably led some hymenopterans to adopt a predatory way of life are briefly summarized and the number of times in which prey relocation and provisioning behavior evolved within the order is investigated. Not taking into consideration the pompilid wasps, prey relocation might have originated from one to three times independently, the different Correspondence/Reprint request: Dr. Gabriel A. R. Melo, Departamento de Zoologia, Universidade Federal do Paraná, Cx. postal 19020, 81531-980, Curitiba, PR, Brazil. E-mail: garmelo@ufpr.br

Gabriel A. R. Melo et al.

scenarios dependent on the phylogenetic hypothesis and mode of character optimization. A separate analysis, at genus level, carried out for the Pompilidae postulates two separate origins for prey relocation involving two main clades, one within the Pepsinae and the other within the Pompilinae. It is concluded that further advances in understanding should come with more robust phylogenies. Additional phylogenetic reconstructions at genus, or species level, within polymorphic clades such as the Pompilidae and Scoliidae, are also necessary for more precise groundplan estimations. More stable and unambiguous reconstructions at family level within the Aculeata, however, will require data from clades for which we currently lack any biological information. Knowledge on the biology of some key groups, such as the Bradynobaenidae, remains elusive and will not be obtained without concentrated effort.

Introduction In ecology, predation has been defined as a competitive biological interaction where a consumer (predator) feeds on another organism (prey) which is always alive at the time the predator first attacks it and this interaction usually results in the death of the prey [1]. Predation represents one of the major evolutionary forces [2] and has evolved many times at different taxonomic levels, including within insects where several groups have independently evolved to be predators. Most predatory insects feed on other arthropods and, in this way, are involved in regulating their communities by reducing the prey’s population to a lower level than would occur in the predator’s absence [3]. In its classical sense, a predator kills its prey more or less immediately after attacking it, and during its lifetime, the predator will kill several prey individuals, often consuming them entirely. In a broader sense, predation can also result from interactions like grazing and parasitism, and includes more subtle interaction processes like parasitoidism [1]. Grazers attack several prey during their lifetime, but only take a portion of each one, usually affecting them in a limited way and rarely being lethal in the short term [2]. Mosquitoes are an example of grazers amongst insects and some kinds of host feeding in parasitoid Hymenoptera [4] can be accounted as such too. A parasite is a symbiotic organism that feeds on a single prey (host)— rarely a few more under special circumstances—being harmful in the long term but rarely causing death in the short term. The parasite benefits at the expense of the host, reducing its fitness in several ways: it reduces the host’s capability of reproducing, may induce different degrees of pathology and also modify the behavior of the affected organism [1]. Parasitoidism involves the traits described for parasites, but the interaction between the parasitoid and its hosts unavoidably results in the

Origin of predation in Hymenoptera

death of the latter or—less frequently— in the loss of its reproductive viability [5]. In some sense, parasitoidism may be considered a somewhat intermediate mode of life between classical parasitism and predation [6, 7]. Similarly to predators, parasitoids will directly impact their host populations by reducing their numbers. Among insects, the parasitoid lifeway has arisen most commonly in the Diptera, where it has evolved independently in 21 families (probably over 100 times). It has arisen in 11 coleopteran families but probably only once in the Hymenoptera [8]. In most cases host searching is by the adult female parasitoid but in a significant minority of cases host-searching is by the first instar larva of the parasitoid, or (more rarely) by the host ingesting the parasitoid egg [8]. Predation in its narrowest sense —where the predator attacks and consumes its own prey— is proportionally less important in Hymenoptera, being restricted to adult feeding behaviors such as host feeding with lethal consequences [9, 10] and direct predation by adult tenthredinids hunting other insects for self nutrition [11; see below]. In Hymenoptera, predation is carried out not only for self nutrition, and its peculiarities make its definition and distinction from parasitoidism a matter of controversy [12]. The peculiarities of predation in Hymenoptera are directly inherited from a series of evolutionary steps dating back from the origins of the order as such. The groundplan life-style of Hymenoptera is phytophagy, probably endophytic into highly nutritious tissue [12] as exhibited by xyeline Xyelidae, this family being recognized as the sister lineage to the remainder of the Hymenoptera [13, 14]. Although several basically exophytic phytophagous lineages (Tenthredinoidea, Pamphiloidea and Cephoidea) appeared afterwards as sideshoots within the order, an important behavioral consequence of endophytic nutrition is direct oviposition within the food source, a feature that will be maintained throughout the order [5] with just some minor modifications and losses (notably in Perilampidae, Eucharitidae and Trigonalidae, which just lay their eggs in the substrate that will potentially be transited or consumed by a host). At some point, the main lineage of Hymenoptera—in the grade containing the superfamilies Siricoidea and Xiphydrioidea—adopted an important shift in larval feeding habits as endophytic symbiotic mycoxylophages [15, 16, 17]. It is in the clade containing Orussidae and the Apocrita that entomophagy appears as idiobiont parasitoidism a single time in the evolutionary history of the Hymenoptera. Theoretical pathways to parasitoidism have been suggested from classical predation, saprophagy and mycophagy [8]. This last pathway seems to have happened in Hymenoptera, starting in the endoxylophagous scenario

Gabriel A. R. Melo et al.

characteristic of the ancestral lineage of Orussidae plus Apocrita [8]. Orussids and some rather basal families of Apocrita are parasitoids of woodburrowing insects (mostly Coleoptera, but also larvae of Siricidae and Xiphydriidae) [12, 18]. Some authors have clearly demonstrated that species of Orussidae develop as parasitoids of the larvae of Buprestidae and Siricidae [18, 19, 20, 21, 22]. There has been some discussion whether the oviposition is in protected burrows of stems, in which the young larva searches and feeds on larvae of other arthropods, or directly on the host. Nevertheless, the idiobiont way of life seems to have evolved in the lineage that gave rise to the Orussidae plus Apocrita, with many changes along the evolution of the derived members within the order. At its origin, parasitoid entomophagic behavior in larval Hymenoptera probably evolved from facultative carnivory in a cleptoparasitic species, in which its larva fed both on the host immature (or even on a conspecific larva), as well as on plant substrate [8]. Here, the concept of predation as applied to Hymenoptera is discussed and definitions for complementary terms are introduced. We also briefly summarize the evolutionary pathways that probably led some hymenopterans to adopt a predatory way of life, compile information on the predatory lineages and investigate how many times this behavior has evolved within the order.

Predation in the Hymenoptera Predation in Hymenoptera occurs mostly for larval feeding and, more rarely, as adult self nutrition. Most cases of predation for larval feeding are carried out by adult wasps acting as provisioning predators, while a few records of direct predation by larvae are known in many families of otherwise typical parasitoids [7]. Adult self nutrition derived from provisioning predation seems to be restricted to social vespids, that malaxate their prey before providing it to the larvae [23], and to some solitary eumenine and apoid wasps that bite or pierce the prey to suck up fluids, sometimes also partly consuming prey tissue [24, 25]. The few cases of adult predacious feeding, not related to larval provisioning, are known in tenthredinid sawflies and some apocritan parasitoid wasps preying directly on their hosts.

Adult predacious feeding Although the basal hymenopteran lineages have phytophagous larvae, a large number of adult sawflies in the family Tenthredinidae have been observed preying on other insects [11, 26]. At least 63 species in 27 genera have been reported to hunt and feed directly on a number of holometabolous insects, including adult flies (Bibionidae, Calliphoridae, Empididae,

Origin of predation in Hymenoptera

Fanniidae, Scatophagidae, Stratiomyidae, Syrphidae, Tachinidae, Tipulidae), adult sawflies (Tenthredinidae), adult beetles (Cantharidae, Nitidulidae), larvae and pupae of chrysomelid beetles and immature Lepidoptera [11, 26]. Another remarkable case of adult predation is known in one species of Ichneumonidae, which uses a wide range of moth larvae as hosts [9]. Adult females are not able to oviposit without first consuming some host tissue, usually hemolymph. This could be included simply as a case of host feeding [27], except that very small hosts can be entirely consumed by the females during the feeding process [9]. The Dryinidae comprises another case where adult females are known to also prey on their auchenorrhynchous hosts [10, 28]. Females of three species kept under confined conditions were reported to feed partially on their hosts, which subsequently were either discarded or parasitized [10]. They usually consumed the first host captured in the day. These cases of adult predacious feeding are based either on anecdotal field observations or on behaviors exhibited under experimental conditions. It must be taken into account that laboratory factors might have influenced the results observed for Ichneumonidae and Dryinidae, although it seems likely that these wasps behave similarly in nature. Unfortunately the limited information available for Hymenoptera precludes an evaluation of the importance and impact of this kind of predation in the wild.

Idiobiont parasitoidism and host relocation In Hymenoptera, provisioning predation has undoubtedly evolved from the parasitoid mode of life and one of the key features in this transition involves the relocation of the host, usually to a site far away from where it was subdued. Parasitoidism is a mode of life exhibited by a vast portion of hymenopteran lineages, of which the most diverse are the Ichneumonoidea and the Chalcidoidea [29]. Hymenopteran parasitoids have long been classified in a dichotomy, distinguishing where their development on the arthropod host takes place, either internally or externally, respectively endoparasitoidism or ectoparasitoidism. However, these traits are somewhat vague and must be used for particular cases only. For major tentative classifications, the terms idiobiont and koinobiont (more appropriately transliterated as â&#x20AC;&#x2DC;cenobiontâ&#x20AC;&#x2122; [30]) should be adopted [5, 31]. Idiobiosis is commonly related to ectoparasitoid behavior, but by no means all idiobionts are ectoparasitoids. The term idiobiont is better related to the influence that the parasitoid exercises over its host, that is, its larva consumes the host while the latter has its development interrupted by the adult female, which paralyzes the host prior to or during oviposition [5]. Koinobiosis (or

Gabriel A. R. Melo et al.

better, â&#x20AC;&#x2DC;cenobiosisâ&#x20AC;&#x2122;), on the other hand, is commonly related to endoparasitoidism, but yet again this is no absolute rule. This is an apomorphic condition compared to idiobiosis, and the parasitoid larva consumes the host only after having allowed it to develop to certain stages of its lifespan [5]. Most idiobionts attack hosts living in concealed places [5, 32] and they can be divided into two basic groups depending on how the female wasps reach the hosts. In the first group, the parasitoid females reach the hosts deploying an elongate ovipositor that both injects substances that paralyze and alter the host physiology and transfers the parasitoid egg to its body surface. Under this situation, the parasitoid female does not get within close reach of the host. This is assumed to be the primitive condition among parasitoid Hymenoptera, having evolved from mycoxylophagous ancestors that inserted the eggs and their symbiotic fungi deep into the plant substrate. Alternatively, in other groups of idiobionts, the female moves through the substrate in search of hosts and needs to enter in direct contact with them for paralysis and oviposition. Host stinging usually takes place through a short, modified ovipositor, and the female attaches its egg on the external surface of the host body. This mode of attack is characteristic of the aculeate hymenopterans that behave as idiobiont ectoparasitoids [12]. In their direct contact with the host, these parasitoid females are repeatedly confronted with the decision to leave it in the same place or to move it somewhere else. In the latter situation, females may relocate their host, usually for considerable distances, before ovipositing. Host relocation is exhibited by many aculeate wasps (see below) and this term is used here to characterize a series of behaviors associated with movement of the host (or prey), commonly to a concealed site, from the place where it was originally found and subdued. Host relocation has probably evolved as a means to reduce host usurpation, predation, superparasitism or hyperparasitism of the developing immature [7]. In the short term, it is likely to be most effective against competition from conspecific females, which supposedly would not forage in places where the hosts are not expected to thrive. Independent of the putative advantages related to relocation of the hosts, fixation of this behavior in several aculeate lineages has undoubtedly opened up major avenues leading to complex behaviors associated to nest construction and sociality.

Host relocation, nest building and provisioning predators The most widespread mode of predation within Hymenoptera is carried out by adult wasps capturing other arthropods that are taken to and stored in nests as food for their larvae. Wasps with this life history have been termed provisioning predators [7], since they will only rarely feed on the prey

Origin of predation in Hymenoptera

themselves, and therefore do not behave as typical predators. This kind of predation, destined to larval feeding, occurs only among aculeate lineages and involves host/prey relocation in a first stage and nest building in a second stage [33]. Host/prey relocation in aculeate wasps has almost always led to evolution of behaviors associated with active construction of a nest, a place especially prepared by females to which the prey are transported and stocked. Nests are most commonly built as tunnels and chambers dug in the ground, although more elaborate and complex nests are found in all nest-building aculeate groups, such as aerial nests built with plastered material (mud, resins or plant fibers mixed or not with glandular secretions produced by the adult insects or their larvae). Most nests also have as their basic unit the brood cell, primarily a cavity built and provisioned for the development of a single immature. Predatory wasps can be single or multiple provisioners depending on the number of prey items fed to each immature. A multiple provisioner can be either a mass or a progressive provisioner. Mass provisioning occurs when the adult stocks all of the food necessary for the development of the larva in a single process before sealing the containing cell. The egg is normally laid on the food after the stocking is complete, but in some groups it is laid prior to provisioning. Progressive provisioning occurs when the adult provides fresh food to its larva from time to time until its complete development. The cell is either permanently open or the adult opens and re-seals it each time a new load is provided. The thin boundary between an idiobiont parasitoid and predatory Hymenoptera has traditionally been considered to reside in the behavior of the female towards its host/prey: a female wasp that does not relocate its host/prey is considered a parasitoid, while one that relocates it is considered a predator [7, 25]. There is, however, a seamless gradation among aculeate wasps from species that behave as parasitoids to those that move the paralyzed prey only for short distances and hide it in pre-existent crevices to those that prepare a nest previously to prey hunting [25]. Therefore, any attempt to label species strictly according to behavioral categories is necessarily arbitrary. For this reason, we focus here in a specific behaviorâ&#x20AC;&#x201D; host relocationâ&#x20AC;&#x201D;instead of on behavioral categories.

Occurrence of prey relocation in Hymenoptera All known cases of host/prey relocation in Hymenoptera are restricted to the Aculeata, a large clade in which the females have their ovipositor modified as a sting apparatus used to inject venom into their host or prey, and therefore no longer used for laying eggs. In their groundplan, the aculeates

Gabriel A. R. Melo et al.

behave as idiobiont parasitoids, attacking beetle larvae living in concealed places. From this parasitoid groundplan, host/prey relocation and predatory behavior have evolved a number of times within the Aculeata. The formicids, vespids, most apoid wasps and, to some extent, the pompilid spider wasps behave as typical provisioning predators, paralyzing their prey and transporting it to their nests for larval feeding. Nevertheless, cases of limited host relocation are known in Bethylidae, Tiphiidae and Scoliidae, families that otherwise behave as typical parasitoids (see below). A summary of the main biological features for the aculeate families is given in Table 1. Biological information is lacking for three key basal lineages in the Aculeataâ&#x20AC;&#x201D;Plumariidae (Chrysidoidea), Sierolomorphidae (Vespoidea) and Heterogynaidae (Apoidea). Based on morphological and some behavioral evidence, the first two families are likely to exhibit the groundplan biology of the aculeates, behaving as parasitoids upon beetle larvae living in the ground. As regards the Heterogynaidae, it has been suggested that their biology is probably similar to that exhibited by the Ampulicidae [34]. Also, data on the biology of Bradynobaenidae is virtually lacking, except for anecdotal information on a species of Typhoctinae [35].

Bethylidae Bethylids as a whole seem to be adapted for searching hosts living in cryptic situations [39] and many species have developed host relocation and nest closure similar to that observed in other predatory aculeates. It is likely that these behavioral traits have evolved independently and possibly several times within the family, but the knowledge of the biology and phylogeny of bethylids is still by far too incomplete as to get a good picture of their evolution. Host carriage to a bembicine wasp nest burrow used by the female bethylid as a starting point for her own adventitious and poorly elaborated nest has been reported for Epyris eriogoni [40]. Also females of Epyris extraneus hide their host in a temporary shelter, dig a burrow and then retransport it to the burrow [41]. Both species of Epyris subdue and transport tenebrionid larvae much larger than their own size. Host relocation has also been reported for other genera such as Allepyris, Bethylus, Cephalonomia, Holepyris, Laelius, Parascleroderma and Trachepyris [40, 42]. Apoidea Besides the bees (Apidae sensu Melo [34]), Apoidea contains a varied assemblage of predatory wasps in the families Ampulicidae, Sphecidae and Crabronidae, and rare cases of parasitoidism (e.g. genus Larra in the

Table 1. Summary of main biological features for the families of aculeate Hymenoptera (character states are assumed to be part of the family’s groundplan). Classification according to Brothers [36], Carpenter [37] and Melo [34]. A “?” indicates lack of information, and a “—”, an inapplicable character.

Origin of predation in Hymenoptera 9

There are reports of limited host relocation (see text for details). 2Some species also lay a single egg per host. 3Amiseginae and Loboscelidiinae parasitize eggs of Phasmatodea. 4Ampulicidae does not build nests in a strict sense. 5Groundplan states are based in Melo [34] and an unpublished analysis. 6There are reports of limited host relocation (see text for details). 7Diamminae is an exception and uses Gryllotalpidae as hosts [38]. 8There is no available reconstruction for the groundplan states of Pompilidae; therefore it is considered polymorphic for characters of columns 1 and 4 (see text for details). 9Based only on anecdotal information for a species of Typhoctinae [35]. 10Food items in Formicidae are very diverse; here all states, except â&#x20AC;&#x153;pollenâ&#x20AC;?, are attributed to the family. 11Formicids do not lay their eggs directly onto the prey; however, they are likely to have originated from a species with gregarious larvae. 12There are reports of limited host relocation and preparation of a rudimentary chamber (see text for details).

Table 1. Continued

10 Gabriel A. R. Melo et al.

Origin of predation in Hymenoptera

Crabronidae) [34]. All ampulicids are known to attack only Blattaria and to behave rather uniformly: after capturing and stinging a roach, the female waits until it partially recovers from the paralysis, drags it by the antenna to a suitable natural cavity (in the soil, in wood or in hollow stems), lays an egg on it and then finally blocks the opening of the cavity with pieces of various kinds of debris [43, 44, 45, 46]. Therefore, there is no true nest construction among ampulicids, only hiding of the prey in a suitable pre-existing cavity. Sphecidae and Crabronidae, on the other hand, use as prey a large number of insect orders, as well as spiders, and most species build their nests before prey hunting [25, 47, 48]. Nesting behavior is most diverse in the Sphecidae, with nests being constructed before or after prey capture, being made underground, in wood cavities or made of plastered mud, containing single or multiple brood cells, and provisioned with single or multiple prey items. Pompilidae The pompilid wasps are well known predators of spiders [25, 26, 49, 50, 51]. Behavioral diversity in the family includes both idiobiont and cenobiont ectoparasitoids, obligatory cleptoparasites and a variety of nest-building predators, ranging from those that make simple, unicellular burrows in the ground only after prey capture to those that prepare elaborate multicellular mud nests before prey hunting. Despite this variability, all known pompilids attack only spiders and always provision a single prey item per larva.

Tiphiidae Most female tiphiids seek for the hostâ&#x20AC;&#x201D;usually scarabaeid grubsâ&#x20AC;&#x201D; underground, temporarily paralyze it by means of several stings, oviposit on it and immediately leave it alone without relocation or modification of its original place [25, 26, 52]. In some cases, as in the genera Methocha (Methochinae), Diamma (Diamminae) and Pterombrus (Myzininae), the host is stung and buried in its own burrow by the female wasp, the host burrow serving as a sort of rudimentary nest [53, 54, 55]. Nevertheless, special cases of relocation and burrow digging to hide the host when it is located above the ground have been described for some species of Methocha [55], Myzinum (Myzininae) and Tiphia (Tiphiinae) [52]. Secondary evidence of host relocation and cell conditioning has also been recorded for females of Cosila chilensis (Anthoboscinae) as the paralyzed and oviposited larvae of its host (the scarabaeoid Pseudadelphus ciliatus) were found away from their food source and all surrounded by earth while active individuals of the same species were found near plant roots and surrounded by a mixture of earth and degraded wood [56].

Gabriel A. R. Melo et al.

Scoliidae Although there has been very little published on the biology of scoliids, the available information indicates that these wasps attack only scarabaeoid larvae living in soil [25, 26, 41, 52]. Females searching for hosts spend most of the time underground and apparently require direct contact with host body odors and feces to locate them [57]. Relocation of the host, sometimes several feet underground, followed by preparation of a cell, where oviposition on the host takes place, has been assumed as part of their groundplan [26, 52]. Some species, however, apparently do not move their host from the place in which it is paralyzed [45]. Vespidae All vespids are nest-building predators and one of the key features of their biology is oviposition into an empty cell built before prey capture [58]. As far as is known, the basal vespids are mass multiple provisioners, with the Euparagiinae (extant subfamily sister to the remaining Vespidae) using larvae of Coleoptera as prey [59]. Although most of the remaining lineages utilize Lepidoptera larvae as food resource for the immatures, some genera of Eumeninae also provide their offspring with Coleoptera larvae as well [23, 24]. Nevertheless, no generalization can be made for the entire subfamily, since information regarding their biology is by far the least known among vespids. The Masarinae are an exception within the family, providing their nests entirely with pollen and nectar, as bees do [60]. Formicidae While extremely diverse in many morphological, behavioral and ecological features, ants are rather uniform when compared to other provisioning aculeate predators: all species are eusocial, rear their larvae in communal chambers and feed them progressively [61]. Nests are most commonly a series of tunnels and chambers dug in the soil or in decaying wood. Although primarily predators, ants have evolved to explore a wide range of food items, among them cultivated fungus mycelia and harvested seeds [61, this book].

Origins of predation in Hymenoptera How many times has provisioning predation arisen among the aculeate hymenopterans? Only an approximate answer to this question is currently

Origin of predation in Hymenoptera

possible and we discuss here some putative scenarios based on hypothesized changes in behavioral features— prey relocation in particular—optimized on available phylogenies for the Aculeata [34, 36, 37, 62]. The analyses focus only on the lineages showing typical provisioning predatory behavior—Apoidea, Pompilidae, Formicidae and Vespidae—and employs the information of the first three columns of Table 1. The outgroup, assumed to behave as an idiobiont ectoparasitoid on beetle larvae, had its character states based on the biology of Orussidae and other apocritan groups, such as Stephanidae and Megalyridae. Postulated character reconstructions are shown in Figs. 1 and 2. At family level and not taking into consideration the Pompilidae (see below), prey relocation might have originated from one to three times independently, the different scenarios dependent on the phylogenetic hypothesis and mode of character optimization (either accelerated or delayed state changes). A single origin (Fig. 2A), at the ancestral lineage of Apoidea plus Vespoidea, is recovered under Pilgrim’s et al. hypothesis [62] when applying accelerated optimization. Two origins (Fig. 1), one at the ancestral lineage of Apoidea and the other in the lineage leading to Formicidae and Scoliidae plus Vespidae, result from Brothers’ hypothesis [36], independent of the optimization. Three separate origins (Fig. 2B), once each at the ancestral lineages of Apoidea, Formicidae and Vespidae, is recovered under Pilgrim’s et al. hypothesis when applying delayed optimization. No resolution can be attained for Pompilidae at family level when treating it as polymorphic for prey relocation (see Table 1). A separate optimization was carried out using an available phylogeny [63] having as terminals the genera for which there is biological information [51]. As shown in Fig. 3, current phylogenetic hypotheses favor two separate origins for prey relocation involving two main clades within Pompilidae, one within the Pepsinae and the other within the Pompilinae. This scenario, as opposed to a single origin, is favored because all basal lineages within the family behave either as cleptoparasites (Ceropalinae) or as spider parasitoids (Minagenia in Pepsinae, the Ctenoceratinae and basal Pompilinae). Considering the low resolving power of the resulting reconstructions at family level (Figs. 1 and 2), one can hardly exert preference for a given scenario over the others. Nevertheless, it is possible to evaluate the different hypothesis against the available evidence. Among the three main scenarios, a single origin for prey relocation seems the most unlikely, since it requires two reversals, one involving a large clade of Vespoidea and another for the Rhopalosomatidae. Reversal to an idiobiont behavior has occurred quite infrequently and the few known cases are restricted to Apoidea and Pompilidae, with no occurrence in Vespidae or Formicidae.

Gabriel A. R. Melo et al.

Figure 1. Ancestral state reconstruction for selected behavioral characters in Aculeata Hymenoptera. Characters 1 to 3 (see Table 1) were optimized on a combined phylogeny for Chrysidoidea [37], Apoidea [34] and Vespoidea [36]: (1) Relocation of larval food, (1.0) absent/ (1.1) present; (2) Larval hosts or provisions, (2.0) orthopteroids (Embioptera, Blattaria, Mantodea, Orthoptera and Phasmatodea)/ (2.1) immature Hemiptera/ (2.2) immature Coleoptera/ (2.3) immature Lepidoptera/ (2.4) immature Hymenoptera/ (2.5) pollen/ (2.6) arachnids; (3) Number of larvae per host, (3.0) single/ (3.1) multiple (gregarious larvae). A, acctran optimization; B, deltran optimization.

It has been commonly assumed that nesting behavior, and in the present context prey relocation, has originated independently among the aculeates [e.g. 33, 64]. Perhaps the main evidence for an independent-origin scenario rests in the distinct behavioral features exhibited by the provisioning predators

Origin of predation in Hymenoptera

Figure 2. Ancestral state reconstruction for selected behavioral characters in Aculeata Hymenoptera. Characters 1 to 3 (see Table 1 and legend of Fig. 1 for character description) were optimized on a combined phylogeny for Chrysidoidea [37], Apoidea [34] and Vespoidea [62] (see their Fig. 8B). The Chrysidoidea are not shown (see Fig. 1). A, acctran optimization; B, deltran optimization.

Gabriel A. R. Melo et al.

Figure 3. Ancestral state reconstruction for selected behavioral characters in Pompilidae. Phylogenetic relationships and subfamilial classification from Pitts et al. [63]; biological information from Shimizu [51; see his Fig. 378]. Characters: (1) Relocation of larval food, (1.0) absent/ (1.1) present; (2) Nest building (applicable only to those taxa receiving state 1.1), (2.0) after prey capture/ (2.1) before prey capture. Reconstruction under delayed optimization.

Origin of predation in Hymenoptera

in the Aculeata, each clade having a number of idiosyncrasies not found in the others. A separate origin in the Apoidea has support in their somewhat isolated phylogenetic position. The lineage that gave rise to them probably behaved similarly to how the extant ampulicids do (see above), with prey relocation by foot and only a rudimentary nesting behavior. Evolution of true nesting behavior, involving construction of a nest before prey capture, occurred in the ancestral lineage of the clade formed by Sphecidae and Crabronidae plus Apidae [34]. Indeed, some of the synapormorphies from the adult morphology shared by this clade may have evolved in conjunction with changes in their nesting biology. For example, the modifications of the midand hind coxae, as well as adjacent areas of the mesepisternum and metepisternum, may be correlated with changes in prey transport mechanisms and/or with the evolution of nest digging behavior [65]. The enlargement of the coxal bases is probably associated with larger and stronger coxal muscles that are required for using the legs for carrying prey and for digging underground nest tunnels [65]. In Sphecidae and Crabronidae, the prey is always actively carried and not just guided to the nest site as ampulicids do. They transport their prey in flight or in case of a large prey over the ground, but always moving forward with the prey secured by the legs and/or the mandibles [6, 66]. As regards the formicids and vespids, two independent origins also seem more likely for them. The single-origin scenario for this lineage (Fig. 1) depends somewhat on the groundplan behavior of scoliids, which were here scored as polymorphic, with prey relocation both present and absent. The rudimentary prey relocation exhibited by some scoliids, however, might not have been present in their ancestral lineage. Unfortunately, biological information for the scoliids is scant and fragmentary, with less than ten species investigated [25] and no data available for the putative archaic genus Proscolia. Also, the peculiar biology of formicids, as compared to vespid wasps and other nest-building aculeates, favors a scenario of parallel evolution for prey relocation and nesting behavior. Little can be said of the behavior of the ancient lineage that gave rise to formicids, but it has been often assumed that ants evolved from wasps that subdued large, soil- and wood-dwelling arthropod hosts, rearing the larvae gregariously on them [61, 67]. Host/prey relocation might not have been present in the initial stages due to the large size of the hosts. Alternatively, ancestral ants might have behaved as some basal ants of the subfamily Amblyoponinae still do [68, 69], dragging entire chilopods to their nests and then placing the larvae to feed directly on them [61].

Gabriel A. R. Melo et al.

Diversity of host/prey use within aculeate Hymenoptera The lineage that gave rise to the aculeate Hymenoptera diversified around 160-150 mya in the late Jurassic [70, 71], with the majority of the extant lineages being already established by the early Cretaceous (~130 mya). Although these extant lineages use a wide variety of arthropod groups as host/prey for their immature (Table 1), the ancestor of the Aculeata probably behaved as an idiobiont ectoparasitoid on beetle larvae. Aculeate wasps, however, differ from other Apocrita groups also primitively associated with beetle larvae in having evolved from a lineage that specialized in hunting for hosts in the ground [70], living in association with dead plant matter either lying over the surface or buried underground. Use of ground-living host/prey is widespread within the three aculeate superfamilies and is reflected in the behavior exhibited by those families that hunt for beetle larvae as well as those that attack other arthropods. As detailed below, all switches for host/prey living in above ground substrates occurred later in the evolutionary history of these clades and are scattered among independent aculeate lineages. In Chrysidoidea, Plumariidae probably exhibits the groundplan condition for the aculeates, since the morphology of the females (apterous) and their collecting sites (under rocks) suggest that they search for hosts on the soil surface or underground [72]. In the remainder of the superfamily, there are four main host switches (Fig. 1): immature Lepidoptera in Bethylinae (Bethylidae), immature Hymenoptera in Chrysididae, Embioptera in Sclerogibbidae and Auchenorhyncha (Hemiptera) in Drynidae + Embolemidae. Also, a switch in host choice to phasmid eggs in the ancestor of Amiseginae + Loboscelidiinae occurred in Chrysididae, but apparently maintaining the plesiomorphic search behavior in regard to the host substrate. A switch in prey choice, from beetle larvae to orthopteroid insects (most likely cockroaches), has occurred early in the history of Apoidea (Figs. 1 and 2), but this switch also did not involve a major change in the prey substrate. Nothing is known about the biology of Heterogynaidae, but due to the brachypterous condition of the females, it has been suggested that they might behave as idiobiont parasitoids [73]. On the other hand, despite their brachyptery, females might drag their prey in a way similar to that of Ampulicidae [34]. There are two additional main host switches within Apoidea (Fig. 1): Hemiptera in the ancestral lineage of Crabronidae + Apidae, and pollen in Apidae. The further diversifications in food choice and substrate search behavior in Apoidea have been enormous and a detailed discussion of these changes is beyond the scope of the chapter.

Origin of predation in Hymenoptera

Seeking for host/prey living in the ground is also the basic pattern for the entire Vespoidea, with only the Vespidae diverging early to above-ground prey searching behavior. Use of Coleoptera is maintained as a groundplan feature for Tiphiidae, Scoliidae and probably also for Vespidae and Sierolomorphidae. Although the biology of sierolomorphid wasps remains unknown, specimens have been collected after emerging from decaying logs on the ground [74]. Host/prey choice is diverse in this superfamily and the main switches are: immature Hymenoptera in Sapygidae + Mutillidae, Araneae in Pompilidae, Orthoptera in Rhopalosomatidae, and immature Lepidoptera and pollen within Vespidae. Other switches involve Bradynobaenidae and Formicidae, but in these cases, there is no unambiguous state at family level. The basic condition for Bradynobaenidae is hard to define, since only anecdotal information is available for one species of Typhoctes which parasitize Solifugae (Arachnida). Formicidae seems to have gone through multiple early switches to a broader repertoire of prey, followed by later diversification in their diet, including non-animal food sources (fungi, harvested seeds, etc.). In their groundplan, however, they are clearly associated with ground-living prey, many groups being specialized in diverse arthropod groups, including chilopods, diplopods, diplurans and beetle larvae.

Concluding remarks Despite the major adaptive radiations opened up in the aculeate Hymenoptera by the acquisition of a predatory way of life, our understanding of its early evolution is still precarious for many clades. Current phylogenetic hypotheses offer only limited resolution in pinpointing in which lineages prey relocation and provisioning predation first evolved. The traditional view that provisioning predation evolved independently in the main predatory groups has only partial support from ancestral-state reconstructions. Not considering the pompilid wasps, possible scenarios involve from one to three independent origins. Nevertheless, those postulating convergent evolution seem more plausible in view of the distinct behavioral features exhibited by the different clades of predators. Further advance should come with more robust phylogenies, derived from morphology and/or molecular data. Additional phylogenetic reconstructions at genus or species level, within polymorphic clades such as the Pompilidae and Scoliidae, are also necessary for more precise groundplan estimations. More stable and unambiguous reconstructions at family level within the Aculeata, however, will require data from clades for which we currently lack any biological information. Knowledge of the biology of some key groups, such as the Bradynobaenidae, remains elusive and will not be obtained without concentrated efforts at sites in which these wasps are most abundant.

Gabriel A. R. Melo et al.

References 1. 2. 3. 4.

5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30.

Begon, M., Townsend, C.R., and Harper, J.L. 2006, Ecology, from individuals to ecosystems (4th Ed.), Blackwell Publishing, Malden. Bengtson, S. 2002, Origins and early evolution of predation, M. Kowalewski and P.H. Kelley (Eds.), The Paleontological Society Papers 8, 289. Koul, O., and Dhaliwal, G.S. 2003, Advances in Biopesticide Research volume 3 â&#x20AC;&#x201C; Predators and Parasitoids, Taylor and Francis, London. Jervis, M.A., Kidd, N.A.C., and Sahragand A. 1987, Host-feeding in Dryinidae: its adaptive significance and its consequences for parasitoid-host population dynamics, C. Vidano and A. Arzone (Eds.), Proceedings of the 6th Auchenorrhyncha Meeting, Turin, 591. Gauld, I.D., and Hanson, P.E. 1995, Carnivory in the larval Hymenoptera, P.E. Hanson and I.D. Gauld (Eds.)-, Oxford Univ. Press, Oxford, 40. Evans, H.E., and West-Eberhard, M.J. 1970, The Wasps, Univ. Michigan Press, Ann Arbor. Godfray, H.C.J. 1994, Parasitoids: Behavioral and Evolutionary Ecology, Princeton Univ. Press, Princeton. Eggleton, P., and Belshaw, R. 1992, Phil. Trans. R. Soc. Lond. B, 337, 1. Leius, K. 1961, Can. Entomol., 93, 1079. Waloff, N. 1974, J. Entomol. (A), 49, 97. Hobby, B.M. 1932, Proc. Entomol. Soc. Lond., 7, 14. Gauld, I.D., and Bolton, B. 1988, The Hymenoptera, Oxford Univ. Press, Oxford. Schulmeister, S. 2003, Biol. J. Linn. Soc., 79, 209. Vilhelmsen, L. 2001, Zool. J. Linn. Soc., 131, 393. Deyrup, M.A. 1984, Great Lakes Entomol., 17, 17. Parkin, E.A. 1942, Ann. Appl. Biol., 29, 268. Stilwell, M.A. 1966, Forest Sci., 72, 121. Vilhelmsen, L. 2003, Zool. J. Linn. Soc., 139, 337. Gourlay, E.S. 1951, Bull. Entomol. Res., 42, 21. Nuttal, M.J. 1980, Forest and Timber Insects in New Zealand, 47, unpaginated. Powell, J.A. and Turner, W.J. 1975, J. Kansas Entomol. Soc., 48, 299. Rawlings, G.B. 1957, Entomologist, 90, 35. Spradbery, J.P. 1973, Wasps: an Account of the Biology and Natural History of Solitary and Social Wasps, Sidwick & Jackson, London. Cowan, D.P. 1991, The solitary and presocial Vespidae, K.G. Ross and R.W. Matthews (Eds.), Cornell University Press, Ithaca, 33. Oâ&#x20AC;&#x2122;Neill, K.M. 2001, Solitary Wasps: Behavior and Natural History, Comstock Publ. Assoc., Ithaca. Iwata, K. 1976, Evolution of Instinct: Comparative Ethology of Hymenoptera, Amerind Publishing Co., New Delhi. Jervis, M.A., and Kidd, N.A.C. 1986, Biol. Rev., 61, 395. Clausen, C.P. 1962, Entomophagous Insects, Hafner Publ. Co., New York. Sharkey, M.J. 2007, Zootaxa, 1668, 521. DalMolin, A., and Melo, G.A.R. 2005, Informativo Soc. Entomol. Brasil, 30(2), 3.

Origin of predation in Hymenoptera

31. Askew, R.R., and Shaw, M.R. 1986, Parasitoid communities: their size, structure and development, J. Waage and D. Greathead (Eds.), Academic Press, London, 225. 32. Quicke, D.L.J. 1997, Parasitoid Wasps, Chapman & Hall, London. 33. West-Eberhard, M.J., and Hanson, P.E. 1995, Nesting behaviour and the evolution of sociality, P.E. Hanson and I.D. Gauld (Eds.), Oxford Univ. Press, Oxford, 67. 34. Melo, G.A.R. 1999, Scientific Papers (Nat. Hist. Mus. Univ. Kansas), 14, 1. 35. Brothers, D.J., and Finnamore, A.T. 1993, Superfamily Vespoidea, H. Goulet and J.T. Huber (Eds.), Agriculture Canada, Ottawa, 161. 36. Brothers, D.J. 1999, Zool. Scr., 28, 233. 37. Carpenter, J.M. 1999, Zool. Scr., 28, 215. 38. Rayment, T. 1935, A Cluster of Bees, The Endeavour Press, Sydney. 39. Evans, H.E. 1964, Bull. Mus. Comp. Zool., 132, 1. 40. Rubink, W.L., and Evans, H.E. 1980, Psyche, 86, 313. 41. Williams, F.X. 1919, Proc. Hawaii. Entomol. Soc., 4, 55. 42. Azevedo, C.O. 1999, Família Bethylidae, C.R.F. Brandão and E.M. Cancello (Eds.), FAPESP, São Paulo, 169. 43. Gess, F.W. 1984, Ann. Cape Provincial Mus. (Nat. Hist.), 16, 23. 44. Maneval, H. 1939, Ann. Soc. Entomol. Fr., 108, 49. 45. Williams, F.X. 1919, Hawaii. Sugar Planters’ Assoc. Entomol. Bull., 14, 19. 46. Williams, F.X. 1929, Proc. Hawaii. Entomol. Soc., 7, 315. 47. Bohart, R.M., and Menke, A. S. 1976, Sphecid Wasps of the World, Univ. California Press, Berkeley. 48. Evans, H.E., and O’Neill, K.M. 2007, The Sand Wasps: Natural History and Behavior, Harvard Univ. Press, Cambridge. 49. Evans, H.E. 1953, Syst. Zool., 2, 155. 50. Evans, H.E., and Yoshimoto, C.M. 1962, Miscell. Publ. Entomol. Soc. Am., 3, 67. 51. Shimizu, A. 1994, Tokyo Metrop. Univ. Bull. Nat. Hist., 2, 1. 52. Clausen, C.P., Gardner, T.R., and Sato K. 1932, United States Department of Agriculture, Techn. Bull., 308, 1. 53. Palmer, M. 1976, Proc. Entomol. Soc. Wash., 78, 369. 54. Williams, F.X. 1928, Hawaii. Sugar Planters’ Assoc. Entomol. Bull., 19, 128. 55. Wilson, E.O., and Farish, D.J. 1973, Anim. Behav., 21, 292. 56. Janvier, H. 1933, Ann. Sci. Nat., 10(16), 209. 57. Inoue, M. and Endo, T. 2008, J. Ethol., 26, 43. 58. Carpenter, J.M. 1982, Syst. Entomol., 7, 11. 59. Trostle, G.E, and Torchio, P.F. 1986, J. Kansas Entomol. Soc., 59, 641. 60. Gess, S.K. 1996, The Pollen Wasps, Harvard Univ. Press, Cambridge. 61. Hölldobler, B., and Wilson, E.O. 1990, The Ants, Belknap Press, Cambridge. 62. Pilgrim, E.M., von Dohlen, C.D., and Pitts, J.P. 2008, Zool. Scr., 37, 539. 63. Pitts, J.P., Wasbauer, M.S., and von Dohlen, C.D. 2006, Zool. Scr., 35, 63. 64. Evans, H.E. 1966, Ann. Rev. Entomol., 11, 123.

Gabriel A. R. Melo et al.

65. Melo, G.A.R. 1997, Phylogenetic relationships and classification of the major lineages of Apoidea (Hymenoptera), with emphasis on the crabronid wasps (Ph. D. dissertation), Univ. Kansas, Lawrence. 66. Evans, H.E. 1962, Evolution, 16, 468. 67. Wilson, E.O. 1971, The Insect Societies, Belknap Press, Cambridge. 68. Gotwald, W.H., and LĂŠvieux, J. 1972, Ann. Entomol. Soc. Am., 65, 383. 69. Traniello, J.F.A. 1982, Psyche, 89, 65. 70. Rasnitsyn A.P. 2002, 2.2.1.3.5. Superorder Vespidea Laicharting, 1781. Order Hymenoptera LinnĂŠ, 1758 (=Vespida Laicharting, 1781), A.P. Rasnitsyn and D.L.J. Quicke (Eds.), Kluwer Academic Publishers, Dordrecht, 242. 71. Grimaldi, D., and Engel, M.S. 2005, Evolution of the Insects, Cambridge Univ. Press, Cambridge. 72. Evans, H.E. 1966, Psyche, 73, 229. 73. Day, M.C., 1984, Syst. Entomol., 9, 293. 74. Vanderwel, M.C., Malcolm, J.R., Smith, S.M., and Islam, N. 2006, For. Ecol. Manage., 225, 190.

Transworld Research Network 37/661 (2), Fort P.O. Trivandrum-695 023 Kerala, India

Predation in the Hymenoptera: An Evolutionary Perspective, 2011: 23-37 ISBN: 978-81-7895-530-8 Editor: Carlo Polidori

2. Effects of prey size and load carriage on the evolution of foraging strategies in wasps Joseph R. Coelho Institute for Franciscan Environmental Studies, Biology Program, Quincy University 1800 College Ave, Quincy, IL 62301, USA

Abstract. Load-carriage limits should impose constraints on foraging for wasps that carry prey in flight. However, application of a simple foraging model to a variety of such wasps reveals a diversity of evolved solutions to the problems of load carriage, as well as often greater constraints imposed by the environment. Some wasps are able to carry loads greater than that with which they should be able to take off. Some wasps are limited in prey size optimization by the availability of prey sizes, others by prey stealing behavior of conspecifics or by the location of their nests. Even species which appear to be nearly perfect in utilizing their foraging capacity may demonstrate low precision upon closer inspection. The ability of wasps to assess prey size and to switch prey species should be explored further. Studies of foraging behavior in the field, rather than at the nest site, might provide greater insight into waspsâ&#x20AC;&#x2122; abilities to optimize loading.

Introduction Wasp foraging behavior has been the subject of several classic natural history studies [1, 2, 3, 4], as it is fascinating to observe. It is only recently, Correspondence/Reprint request: Dr. Joseph R. Coelho, Institute for Franciscan Environmental Studies Biology Program, Quincy University, 1800 College Ave, Quincy, IL 62301, USA. E-mail: coelhjo@quincy.edu

Joseph R. Coelho

however, that formalized quantitative efforts have been applied to wasp foraging [e.g. 5, 6]. In this regard, wasps have proven excellent models, yielding interesting insights into the evolution of foraging behavior. Being relatively small, abundant and, frequently, returning to the same site (nest) repeatedly has made wasps convenient study organisms. If energy supplies are limited, selection should favor individuals most effective at acquiring energy [7]. One model of foraging efficiency (modified from [8]) is as follows: E/T=(NeE-Cs)/(1+NeH) where Ne=prey encounter rate E=energy gain/prey Cs=cost of searching H=handling time T=time Hence, selection should favor predators that maximize E/T. All other things being equal, larger prey yields larger E return. Hence, a predator should maximize E by taking the largest prey items available (although increasing load size may also increase the energy cost of flying [9] and decrease foraging lifespan [10]). This principle should be particularly applicable to predaceous wasps, as most are specialized predators of restricted taxa, and all prey of a given taxon are likely to have a similar energetic value. Thus, E does not vary so much with prey type as prey size. For many predatory wasps, prey size maxima may be limited by the physical constraints of flying with the prey. One method used to examining foraging decisions in wasps is to predict the optimal load size of the prey, and to compare this prediction to the size of actual prey loads. Fortunately, the optimal load size is simple to calculate for species that carry the prey in flight, as the force production of flying animals is primarily dependent on their flight muscle mass. The maximum force produced must equal the weight of the wasp plus that of the prey if the wasp is to fly. At the maximum load mass, the ratio of flight muscle mass to body plus prey mass (or flight muscle ratio) is 0.179 for Hymenoptera [11]. In other words, if the wasp has a prey item so large that flight muscle ratio < 0.179, it cannot take off, and the foraging bout fails. Use of this threshold, the marginal flight muscle ratio, as a metric for flight capability has several advantages, primarily its independence of the size of the wasp, allowing it to be used in cross-species comparisons.

Loading constraints on foraging

This simple model and methodology have been applied to several species of wasps not so much as a formal test of foraging theory but as a means of examining behavioral alternatives that have evolved in each species. The ecological context of each species imposes limitations and opportunities that affect its behavior.

Methods In each study essentially the same techniques were used. Wasps carrying prey were captured, usually as they returned to the nest. The body masses of the wasp and prey were measured fresh or frozen (never dried). The thorax mass of the wasp was determined by cutting away legs, wings, head and abdomen and weighing the remaining thorax on a balance. As flight muscle comprises 95% of the thorax in Hymenoptera [11], the flight muscle need not be dissected out and weighed separately.

Discussion Yellowjackets (Vespula) are social wasps that prey upon or scavenge a variety of foodstuffs. Normally, undertaker honey bees (Apis mellifera L.), carry their dead far from the hive [12]. However, when the thorax temperatures of honeybees are measured (by stabbing with a thermocouple probe, e.g. [13]), discarded dead bees rapidly accumulate near the hive. The investigator noticed opportunistic yellowjackets consuming them with considerable zeal. When foraging on dead honey bees, yellowjackets chew body parts off the carcass until they are able to take off with the remaining portion of it. The first parts chewed away are generally not very nutritiousâ&#x20AC;&#x201D; legs and wings. Still, one would predict that a yellowjacket should chew away just enough material to reduce its FMRo to 0.179, thereby maximizing the caloric intake of the colony. This hypothesis was tested in three species of Vespula [5]. In neither case did a wasp species achieve the ideal FMRo (Table 1). The nearest to approach the optimal loading was the largest, the German yellowjacket (V. germanica [Fab.]), at FMRo=0.220. V. squamosa (Drury) achieved only 0.289, while the smallest, V. maculifrons (Buysson), declined to forage on dead honeybees at all. The reason for the apparent suboptimality of the first two species appears to lie in the technique used to dismember the honey bee corpse. V. germanica, being the largest species, was occasionally able to carry an entire bee corpse; however, when it was unable to do so, it chewed through the narrowest portions of the bodyâ&#x20AC;&#x201D;the petiole and cervix. Hence, discrete tagmata of the bee corpse were taken. Such a body segment, while carryable,

Joseph R. Coelho

Table 1. Average body mass (Mb) of wasps and prey, and resulting flight muscle ratios of wasps while unladen (FMR) and laden with prey (FMRo). All prey are paralyzed adults unless otherwise noted. The marginal flight muscle ratio, the minimum at which the wasp can just take off, is 0.179 (11). Wasp species

Wasp

Prey type

Mb (g)

Prey

FMR

FMRo

Mb (g)

Bembix troglodytes

[26] 0.099

Monobia quadridens

Diptera

0.045

0.358

0.285

Pyralidae 0.218

(larvae)

[22] 0.108

0.385

0.263

Sceliphron caementarium

[14] 0.129

Aranea

0.034

0.423

0.364

Sceliphron spirifex

[24] 0.132

Sphecius convallis

Aranea

0.077

0.409

0.258

Tibicen 0.991

parallelus

[14] 1.127

0.401

0.187

Sphecius speciosus

[16] 0.822

Sphecius speciosus

Tibicen spp.

1.541

0.433

0.142

Neocicada 0.557

hieroglyphica

[14] 0.298

0.380

0.248

Sphex ichneumoneus

[20] 0.303

Conocephalinae

0.331

0.462

0.238

Sphex pennsylvanicus

[14] 0.489

Phaneropterinae

0.697

0.425

0.189

Stizus continuus

[6] 0.184

Acrididae

0.188

0.380

0.200

Tachytes chrysopyga

[26] 0.052

Vespula germanica

Orthoptera

0.050

0.334

0.177

Apis mellifera 0.092

Vespula squamosa

Reference

(dead)

[5] 0.045

0.330

0.220

Apis mellifera 0.056

(dead)

[5] 0.022

0.400

0.289

might not meet the maximum load. To do so perfectly (to maximize E) would require chewing through the full thickness of the thorax or abdomen. The lengthy H accrued by such a task would reduce foraging efficiency. V. maculifrons would not forage on honey bee corpses in spite of a variety of efforts to induce them to do so. Calculations indicate that they would have

Loading constraints on foraging

difficulty carrying a bee thorax. While carrying a head was possible, the low nutritional value might not be worth their time. Larger wasps carried larger loads within species and among species, indicating some degree the waspsâ&#x20AC;&#x2122; ability to maximize E through prey size. Interestingly, V. germanica foraging on honey bees were 36% larger than a sample collected at a nest. Similarly, foraging V. squamosa were 10% larger by body mass, suggesting the possibility of polyethism, or that workers may specialize on foods that are suitable to their body size. The largest V. germanica, which were sometimes able to carry entire bee carcasses, both optimized the prey size and minimized handling time. The ability to macerate prey probably contributes greatly to the broad generalist foraging capabilities of social yellowjackets. By contrast, all the species discussed below are not only solitary, they forage on a narrow group of taxa. They carry their prey intact, and never chew off parts of it. The first of these species to be examined was the eastern cicada killer (Sphecius speciosus Drury). Cicada killer females, as their name implies, forage exclusively on cicadas (Hemiptera: Cicadidae). Finding their prey usually in a tree, they sting it. Complete, permanent paralysis results in less than 60 s [14]. The wasp grasps the cicada ventral side up and flies to its burrow, where it sequesters it in a cell and lays an egg on it, or obtains another cicada [15]. The implausible sight of a female cicada killer, itself nearly one gram in body mass, attempting to carry a cicada even larger than itself, immediately raises questions of load carriage and difficulty in flight.

Figure 1. Sphecius speciosus using midlegs to drag Tibicen spp. overland in the vicinity of the burrow. Will County, IL, USA (photo by Joe Coelho).

Joseph R. Coelho

At sites in Indiana and Illinois (USA) eastern cicada killers foraged on species of Tibicen, reducing their FMRo to levels well below those with which they should have been able to carry the cicada [16]. These populations of cicada killers had the lowest mean FMRo of all wasps examined. Only 10% of cicada killers should have been able to take off with their prey. However, cicada killers are prodigious foragers, securing between seven and eight cicadas per day on average [17]. Part of the reason for the foraging success of cicada killers lies in the plasticity of their load carriage behavior: they can climb trees while carrying prey. Once the wasp has flown as far as it can with the cicada, presumably downward even under maximum power, it crawls about and finds a vertical object. It ascends the object, usually a tree, until it reaches a height, and takes off again. By repeating this sequence, the wasp eventually gains its burrow. Often the wasp reaches the burrow in flight, but occasionally the wasp is seen hauling the prey overland. This process, as well as climbing, is facilitated by the wasp clasping the cicada with the middle pair of legs, leaving the forelegs and hindlegs for walking. The careful observer will see that the pretarsi of the midleg closely fit a ridge at the base of the cicadaâ&#x20AC;&#x2122;s wing, providing a firm grip while also holding the wings of the cicada closed. Any vertical object will serve for climbing purposes, including humans willing to hold still. On one occasion a female cicada killer with prey was observed climbing a vertical pipe to its top, approximately 2 m, and flying to the ground. However, her wings were so thoroughly worn that she could not generate much horizontal displacement. In searching for another vertical object, she climbed the same pipe again and again. Eventually, she dropped the cicada, and it was taken by another female [14]. Prey carriage is often difficult for eastern cicada killers. Paralyzed cicadas are dropped from trees for reasons unknown or left on the ground when obstacles between wasp and burrow are seemingly insurmountable (e.g. dense, high grass). These prey are usually abandoned, and devoured by scavengers such as ants and yellowjackets. In cicada killers, long powerful legs, climbing ability, and abundant prey allow them to be even more efficient foragers than predicted. Behavioral plasticity in foraging behavior also appears to contribute to the success of cicada killers. Again, dropped cicadas are usually abandoned, except apparently at large, high-density nesting sites (described in [18]), where other females will pick them up (as described above). Prey stealing may also occur at such high densities. While such behaviors might appear to improve foraging by shortening the distance traveled and time spent returning with prey, there must be disadvantages as well. At numerous other sites observed by these investigators over a large geographical area, these behaviors were rarely observed. Probably, the frequency of cicadas dropped

Loading constraints on foraging

or available for theft is much lower at sites of average or lower density. Recently, an unusual population of cicada killers was discovered in Florida (USA), in which the primary prey species, Neocicada hieroglyphica (Say), is rather small [19]. The female eastern cicada killer is also reduced in size, but not as much. As a result, the wasp enjoys a much higher FMRo (0.25) than populations using the much larger Tibicen. None of the females in the sample (n=14) had FMRo < 0.179, and none was observed climbing with prey to achieve altitude [14]. Loading is less than ideal, and females provision each cell with more cicadas to compensate for small prey size. Hence, variation in prey species availability can affect the foraging behavior and efficiency at the population level. The great golden digger wasp (Sphex ichneumoneus L.) also digs a burrow, but forages on orthopteran prey. Although somewhat smaller in body size than Sphecius, it has the highest FMR of any wasp, at 0.46 [20]. This mass allocation is favorable to foraging on quite large prey, which are primarily tettigoniids. When thus burdened, the FMRo of S. ichneumoneus is reduced to 0.238 on average. However, 25% of wasps had FMRo < 0.179, indicating that Sphex ichneumoneus is frequently overloaded (Fig. 2). The great golden digger wasp arrives at essentially the same solution to this problem as Sphecius: climbing. The prey of Sphex ichneumoneus, however,

Figure 2. The distribution of operational flight muscle ratios in three solitary wasps (Tachytes and Bembix from [25], Sphex from [20]). The arrow indicates the marginal flight muscle ratio, where take-off is just possible [11].

Joseph R. Coelho

is normally found in low vegetation, making more frequent climb-and-fly cycles necessary. When climbing is not an option, Sphex ichneumoneus beats its wings while dragging its prey overland. Prey that are too large to be carried are abandoned [21]. As the nesting season progresses and prey species grow larger, Sphex ichneumoneus may switch to smaller prey [21]. However, no change in FMRo occurs over time [20]. Prey size and wasp size are correlated. Various lines of evidence suggest the ability of S. ichneumoneus to learn to assess prey size during foraging, and adjust choice of prey size (see discussion in [19]). This phenomenon should be studied further to determine its pervasiveness and capacity to improve foraging efficiency in solitary wasps. Sphex pennsylvanicus L., the great black digger wasp, was found foraging on various katydids (Orthoptera: Phaneropterinae), including Amblycorypha, Microcentrum, and Scudderia (Fig. 3). Females flew in with prey, landing very close to their destination, a crack in a concrete porch, beyond which individual burrows presumably lay. Prey were frequently dropped, and

Figure 3. Sphex pennsylvanicus with katydid prey. Lewis County, MO, USA (photo by Joe Coelho).

Loading constraints on foraging

prey theft was common. In spite of being 61% larger than its aforementioned congener, Sphex pennsylvanicus had 8% lower FMR. Moreover, the prey were 110% larger, contributing to a 24% lower FMRo [14]. Although the FMRo was near optimal, the range was quite broad (0.131-0.230) and the sample size low (n=9). Some females (4 of 9) were apparently overloaded. Though climbing was not observed, it could have occurred in large nearby trees. The carpenter wasp (Monobia quadridens L.) is an aerial-nesting eumenid, using primarily vacant carpenter bee (Xylocopa) borings for nests [22]. It also readily uses wooden trap nests with 1-cm diameter holes. The carpenter wasp mass provisions a cell with paralyzed lepidopteran caterpillars [22], lays an egg, seals the cell with mud, and begins provisioning the next cell in the series. In west-central Illinois (USA), prey are mostly pyralid moths. As an aerial nester, the carpenter wasp cannot easily use the same tricks as Sphecius and Sphex (climbing) to carry oversized prey, at least not for the final approach to to its nest. Not surprisingly, the mean FMRo of carpenter wasps carrying caterpillars was 0.263, well above the marginal FMR. The size of caterpillars taken, however, increased linearly over time, approximately doubling during the May to September active season, resulting in a corresponding linear decline in FMRo. These trends result in caterpillars that are too light early in the season and too heavy later in the season. Behavioral flexibility allows the carpenter wasp to compensate in part for this apparent environmental constraint. When caterpillars are small, the wasp may carry two at one time. When caterpillars are large, FMRo may fall below the marginal level (5.6% of all cases). Some wasps attempt to drag the oversized caterpillars up the woodwork to the nest. These efforts often fail, however, and the prey is dropped and abandoned [22]. Seasonal growth of prey occurs in other wasps [21, 23], and some switch prey species to compensate [21]. Perhaps small caterpillars are simply unavailable at the end of summer, disallowing prey switching. In the case of the carpenter wasp, the nesting habit and environmental effects appear to constrain foraging capacity. Mud daubers of the genus Sceliphron (Crabronidae) also are aerial nesters, but form their nests almost entirely from mud, and mass provision their larvae with spiders (Fig. 4). Two species, Sceliphron spirifex L. and Sceliphron caementarium (Drury) have been examined [14, 24]. Though native to different continents, the two mud daubers are remarkably similar in size and FMR. In both cases, FMRo was well above the marginal level, and no individual carried a spider that brought FMRo below the marginal level. Unfortunately, sample sizes are small in both cases (n=4 for Sceliphron spirifex). For Sceliphron caementarium (n=29), the black and yellow mud dauber, a 12x range of prey Mb was observed, but even the largest spider

Joseph R. Coelho

Figure 4. Sceliphron caementarium, the black and yellow mud dauber, carries spider prey with its mandibles. Its nests are constructed in a sheltered area, in this case, a garden shed. Lewis County, MO, USA (photo by Joe Coelho).

(Mb=0.127g) depressed FMRo only to 0.25. Sceliphron caementarium has the highest mean FMRo of wasps examined. Hence, it appears tentatively that aerial nesting limits mud daubers to prey items below the maximal loadlifting capacity at take-off. However, growth of spider prey may make it more difficult for them to carry late in the season [23]. Hence, this genus should be examined more thoroughly, especially at different times of the nesting season, to determine whether the pattern of light loads holds. Bembix troglodytes Handlirsch is a ground-nesting, progressively provisioning crabronid wasp that forages on flies (Diptera). Bembix is adept at exploiting the most abundant flies by learning sites with high densities of flies and visiting them repeatedly [25]. This behavioral flexibility suggested that B. troglodytes might be able to select flies of ideal size to optimize load carriage on the return trip to the nest [26]. On the contrary, foraging females reduced their FMR (0.36) to an FMRo of only 0.29 at a site in Big Bend National Park (Texas, USA), resulting in the second highest average FMRo among species examined herein. In fact, no foraging female had FMRo equal to or lower than the marginal level (Fig. 2). The possible reasons for lack of optimization were readily apparent to the observers. Nearly every female returning to a burrow with prey was immediately pounced upon by conspecific females. Prey were often dropped during the attack, then stolen

Loading constraints on foraging

by another female. Many aspects of the females' foraging behavior seemed to be adapted to minimizing prey theft. The fly was tucked tightly under the body of the wasp and held with all legs, making it difficult to determine whether the female was carrying prey at all. The high FMRo of prey-laden females made them quite maneuverable and difficult to capture. In a related wasp (Stictia heros [Fabr.]), the prey size is directly related to the probability of attempted prey theft [27]. Abundant brood parasites, such as velvet ants and satellite flies, also made speed in securing the prey in the burrow advantageous. Therefore, being an apparently suboptimal forager relative to prey size may actually optimize foraging by discouraging prey theft and brood parasitism, while attempted prey stealing was rarely observed in the other species treated in this review. Tachytes chrysopyga obscurus Cresson is a small (the smallest discussed herein) solitary wasp that mass provisions its earth burrows with various orthopteran prey (Fig. 5). It packs each cell of its complex burrow with up to 10 prey items before laying an egg within [28]. Tachytes chrysopyga in Will County, Illinois (USA), were found in a large colony within an even larger colony of eastern cicada killers [25]. Tachytes chrysopyga females' FMR was a modest 0.33. However, when laden with prey, the FMRo of such females averaged 0.18, almost exactly the marginal level. Could T. chrysopyga be a true optimal forager? Inspection of the distribution of FMRo reveals otherwise

Figure 5. Tachytes chrysopyga carrying grasshopper prey over the substrate, occasionally making short, hopping flights. Will County, IL, USA (photo by Joe Coelho).

Joseph R. Coelho

(Fig. 2). Although the marginal FMRo falls within the 95% confidence interval of the mean FMRo, the range of FMRo is very broad. Wasp behavior matched the variation in FMRo. Some prey-laden females were fast and manueverable on the wing, while others used short, hopping flights to return to the burrow. While weak stabilizing selection may be acting on prey size selection in T. chrysopyga, its behavioral plasticity allows it to forage on a ten-fold range of prey masses. Like many of the wasps described herein, T. chrysopyga is probably opportunistic, attempting to forage on the first suitable prey it encounters rather than wasting time and energy looking for prey of ideal size. Nearly ideal loading was observed in another ground-nesting, solitary, orthopteran-hunting wasp, Stizus continuus (Klug). Loaded females (Fig. 6) had mean FMRo of 0.20 [6]. Fifteen individuals (38%) were overloaded, having FMRo < 0.179. With a standard error of 0.05 (n=39), variation was also considerable. The similarity to Tachytes ends there, as Stizus continuus is 3.5x the body mass of Tachytes chrysopyga. The authors suggest that overloaded wasps may climb objects in the environment to gain height. Prey species selection is nonrandom in Stizus continuus, and appears to be related to preferred hunting habitat [6], large bushes, from which wasps could descend with large prey, which was observed, though infrequently [Polidori, pers. comm.].

Figure 6. Stizus continuus carrying prey in flight. El Saler, Valencia, Spain (photo by Davide Santoro).

Loading constraints on foraging

Recently, the Pacific cicada killer, Sphecius convallis Patton (Fig. 7), was studied in Ruby (Santa Cruz County), Arizona [14]. This population preyed exclusively on Tibicen parallelus Davis, resulting in a mean FMRo of 0.187 (n=46). This value lies just above the marginal level, and only seven wasps (15%) were below 0.179. The low standard error (0.002) suggests that S. convallis is the most ideal flying predatory wasp so far examined with respect to loading. The fact that this population preys on adults of a single cicada species may allow stabilizing selection to adjust the size of adult females to a nearly optimal level.

Figure 7. Sphecius convallis carrying Tibicen parallelus in flight casts a shadow on sandy mine tailings where it nests. Santa Cruz Co., AZ, USA (photo by Joe Coelho).

Conclusions Studies of load carriage have provided interesting insights into the foraging behavior of wasps. It appears that few wasps are â&#x20AC;&#x153;optimalâ&#x20AC;? foragers in the sense that they achieve the ideal FMRo of 0.179 during foraging. In most cases, the biotic environment imposes limits or conditions which

Joseph R. Coelho

prevent optimality from being achieved. Wasps appear to have evolved toward optimal load carriage, but usually without achieving it because of the particular constraints to which each is subjected. There is an incredibly diverse array of behaviors exhibited by wasps to circumvent limitations imposed by load-lifting, prey diversity, prey developmental stage, prey theft and other environmental stressors. It cannot be doubted that continued examination of wasp foraging behavior in this manner will reveal additional, unexpected behavioral adaptations.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24.

Fabre, J.-H. 1879, Souvenirs entomologiques - Livre I. Étude sur l'instinct et les moeurs des insectes. Librairie Delagrave, Paris. Rau, P., and Rau, N. 1918, Wasp studies afield. Princeton University Press, Princeton. Teale, E.W. 1962, The strange lives of familiar insects. Dodd, Mead & Company, New York. Evans, H.E. 1973, Wasp farm. Doubleday, New York. Coelho, J.R., and Hoagland, J. 1995, Funct. Ecol., 9, 171. Polidori, C., Mendiola, P., Asís, J.D., Tormos, J., García, M.D., and Selfa, J. 2009, J. Nat. Hist., 43, 2985. MacArthur, R.H., and Pianka, E.R. 1966, Amer. Nat., 100, 603. Molles, M.C. 2010, Ecology: concepts and applications. 5th ed. McGraw-Hill, New York. Wolf, T., Schmid-Hempel, P., Ellington, C.P., and Stevenson, R.D. 1989, Funct. Ecol., 3, 417. Wolf, T., and Schmid-Hempel, P. 1989, J. Animal Ecol., 58, 943. Marden, J.H. 1987, J. Exp. Biol., 130, 235. Visscher, P.K. 1983, Anim. Behav., 31, 1070. Coelho, J.R. 1991, Environ. Entomol., 20, 1627. Coelho, J.R., Hastings, J.M., and Holliday, C.W., unpublished data. Dambach, C.A., and Good, E. 1943, Ohio J.Sci., 43, 32. Coelho, J.R. 1997, Oikos, 79, 371. Coelho, J.R. and Holliday, C.W. 2008, Ecol. Entomol., 33, 1. Hastings, J.M., Coelho, J.R., and Holliday, C.W. 2008, J. Kansas Entomol. Soc., 81, 301. Hastings, J.M., Holliday, C.W., and Coelho, J.R. 2008, Florida Entomol., 91, 657. Coelho, J.R., and LaDage, L.D. 1999, Ecol. Entomol., 24, 480. Brockmann, H.J. 1985, J. Kansas Entomol. Soc., 58, 631. Edgar, P.K. and Coelho, J.R. 2000, J. Hymenoptera Res., 9, 370. Brockmann, H.J., and Grafen, A. 1992, Behav. Ecol. Sociobiol., 30, 7. Polidori, C., Federici, M., Trombino, L., Barberini, V., Barbieri, V., and Andrietti, F. 2009, J. Zool., 279, 187.

Loading constraints on foraging

25. Evans, H.E. 1957, Studies of the comparative ethology of digger wasps of the genus Bembix. Cornell University, New York. 26. Coelho, J.R., Hastings, J.M., Holliday, C.W., and Mendell, A.M. 2008, J. Hymenoptera Res., 17, 57. 27. Villalobos E.M., and Shelly, T. 1996, J. Insect Behav., 9, 105. 28. Evans, H.E., and Kurczewski, F.E. 1966, J. Kansas Entomol. Soc., 39, 323.

Transworld Research Network 37/661 (2), Fort P.O. Trivandrum-695 023 Kerala, India

Predation in the Hymenoptera: An Evolutionary Perspective, 2011: 39-78 ISBN: 978-81-7895-530-8 Editor: Carlo Polidori

3. Predation by ants on arthropods and other animals 1

Xim Cerdá1 and Alain Dejean2,3

Estación Biológica de Doñana, CSIC, Avda. Américo Vespucio, 41092 Sevilla, Spain 2 CNRS, Écologie des Forêts de Guyane (UMR-CNRS 8172), Campus Agronomique 97379 Kourou, cedex, France; 3Université de Toulouse, 118 route de Narbonne 31062 Toulouse Cedex, France

Abstract. Ants are the most widely distributed and most numerically abundant group of social insects. First, they were ground- or litter-dwelling predators or scavengers, and certain taxa evolved to adopt an arboreal way of life. Most ant species are generalist feeders, and only some ground-nesting and groundforaging species are strictly predators. Ants are central-place foragers (with the exception of army ants during the nomadic phase) that may use different foraging strategies. Solitary hunting is the most common method employed by predatory ants. Cooperative hunting, considered more evolved than solitary hunting, is used by army ants and other ants such as Myrmicaria opaciventris, Paratrechina longicornis or the dominant arboreal Oecophylla. Army ants are predators with different levels of specialization, some of which focus on a particular genus or species, as is the case for Nomamyrmex esenbeckii which organizes subterranean raids on the very large colonies of the leaf-cutting species Atta colombica or A. cephalotes. Arboreal ants have evolved predatory behaviors adapted to the tree foliage, where prey are unpredictable and able to escape by flying away, jumping or Correspondence/Reprint request: Dr. Xim Cerdá, Estación Biológica de Doñana, CSIC, Avda. Américo Vespucio, 41092 Sevilla, Spain. E-mail: xim@ebd.csic.es

Xim CerdĂĄ & Alain Dejean

dropping. The weaver ant, Oecophylla longinoda, for example, hunts prey diurnally in groups. They detect prey visually from a relatively long distance and the workers adhere to the plant substrate by means of very powerful adhesive pads and claws. On some occasions, during prey retrieval, the prey can be stolen by other ants; foodrobbing is more frequent in ground-dwelling than in arboreal species. Many predatory ants are engaged in a kind of arms race: they have evolved morpho-physiological adaptations to foil prey defense or escape mechanisms. Mandible shapes have changed and powerful venoms have been developed by different species. Depending on their prey specialization ants can have many different mandible shapes: trap-jaw mandibles, nutcracker mandibles adapted to hunting long prey, pitch-fork mandibles, falciform mandibles and long mandibles. Other ant species are specialized in hunting a certain prey type, but do not have a mandible shape particular to that specialization; these species are egg predators, collembolan predators, or social insect predators. Some ant species are either specialized or occasional termite predators. All of these ants play a role in the equilibrium of ground- and litter-dwelling detritivorous arthropods and the herbivorous insects living in these strata.

Introduction Ants, which represent the family Formicidae, have a stinging apparatus and so belong to the aculeate suborder within the insect order Hymenoptera (some evolved taxa later lost the ability to sting). The oldest ant fossils date from â&#x2030;&#x2C6;100 mya, meaning that ancestral ants most likely appeared during the early Cretaceous Period (144-65 mya) [1]. Ants were firstly ground- or litter-dwelling predators or scavengers, which are plesiomorphic traits. Furthermore, the first ants may have behaved much like todayâ&#x20AC;&#x2122;s army ants, based on what we know from molecular phylogenies that show that the subfamily Leptanillinae is a sister group to all other extant ants [1,2]. Initially, most ant species had spherical or ovoid heads and short mandibles with small numbers of teeth [3]. As ants evolved, worker morphology changed in two main ways: (1) their heads changed shape and their mandibles became more elongated as they became specialized in predation; and (2) their claws became well-developed and their tarsa acquired adhesive pads, thus permitting them to adopt an arboreal way of life [3,4]. Both of these changes occurred in certain taxa. Indeed, the arrival of angiosperms created more complex habitats on the ground and in the leaf-litter when compared to the gymnosperms that had previously dominated the flora. In addition, some angiosperm species provided ants with food in the form of extrafloral nectar, food bodies and the elaiosome on their seeds. Consequently, ant diversification closely tracked the rise of angiosperms and the ecological dominance of ants was notable by the mid-Eocene (50 mya) with nearly all extant subfamilies and most genera

Ant predation

already in place, suggesting an explosive radiation just prior to this period [1,3]. Moreover, as the angiosperms proliferated, the “higher” termites (which comprise 84% of the species) and major herbivorous insects became more diverse, and so there was an increase in the abundance and diversity of potential prey. Hemipterans also became more diverse and numerous taxa developed mutualistic relationships with ants [3,5-6]. The family Formicidae is extremely diverse with 12,651 known ant species [7], and an estimated 3,000 to 9,000 additional species as yet unknown to science. The phylogeny of the family is clearly separated into three clades divided into 19 subfamilies [1]. Ten subfamilies are almost entirely composed of ground-dwelling predatory or scavenging ant species. Although many Myrmicinae, Formicinae and Dolichoderinae species are ground-nesting, their workers forage mostly on plants to gather exudates or attend Hemipterans, the same being true for some less diverse subfamilies such as the Paraponerinae, Myrmeciinae, Heteroponerinae, Ectatomminae and Pseudomyrmecinae [1,8]. Finally, the very abundant group of canopydwelling ants, here also mostly Myrmicinae, Formicinae and Dolichoderinae species, represent a large proportion of the overall animal biomass in this habitat where the irregular availability of prey means that they are omnivorous [9].

Foraging Ants as ecosystem engineers ... and ground-dwelling predators Ecosystem engineers are defined as “organisms that directly or indirectly affect the availability of resources for other organisms through modifications of the physical environment” [10]. Among soil dwellers, earthworms, termites and ants have been identified as the main soil engineers [11-12]. Food storage and the accumulation of feces, corpses and food remains by ants have been shown to rapidly and extensively change the soil conditions within the nest area [13] by affecting: soil texture [12]; chemical composition (i.e., C, P, N and K content) [11, 12,14,15]; and microbial and microfaunal communities [16]. As a result, ants create biogenic structures (e.g., nest, galleries and waste chambers) that influence decomposition dynamics at scales of time and space that exceed their life-span [17]. However, most soil-dwelling ant species are predators (or scavengers; see Box 1) that prey on invertebrates or arthropods participating in the cycle of litter and wood degradation: earthworms, acarids, isopods, different kinds of myriapods (e.g., iulids, chilopods, polyxenids), collembolans, termites, other ants, and “other insects” (e.g., beetles, bark lice, lepidopterans).

Xim Cerdá & Alain Dejean

Box 1. WANTED...dead or alive? Scavengers or predators? A methodological commentary In Webster's Online Dictionary (http://www.websters-online-dictionary.org) predation is defined as “the act of preying by a predator who kills and eats the prey”; and scavenger as “any animal that feeds on refuse and other decaying organic matter”. It seems easy to distinguish between predatory ants, which collect (and kill) live prey, and scavenger ants which collect dead items. But a problem arises because most ant species are omnivorous and generalist, meaning that, even if they are predators, they can also collect arthropod corpses and very frequently scavengers may act as predators when prey are small (e.g., arthropod eggs [21] and/or with very reduced or no mobility (e.g., pupae, small Hemiptera or injured arthropods). In his monograph about fire ants [22], Walter Tschinkel wrote, "In fire ants, the largest portion of their animal-matter diet is insects. How important is predation, as opposed to scavenging? Again, the answer is probably a matter of opportunity. Fire ants are clearly effective predators, often suppressing prey populations, and therefore control agricultural pests. Fire ants are truly omnivorous, feeding on fluids derived from plants or animals, acting as both predators and scavengers and at times even primary consumers." When Robert Jeanne studied the rate of ant predation along a latitudinal gradient [23], his method was to assess predation on the brood of social wasps (by using wasp larvae baits). Predation was more rapid in tropical than in temperate sites, in fields (open habitats) than in forests, and on the ground than on vegetation and these differences were much more pronounced in the temperate zone than in the tropics. Because the wasp larvae were alive, it was a study on predation (excellent, beyond all doubt), but the question is how many of the species that fed on these wasp larvae were truly predators and how many were mostly scavengers? It might be relevant to the results of Jeanne's study if ants are scavengers or predators, but, in many other cases, it can be very important, especially when considering the role of ants in the ecosystem (e.g., nutrient cycling). Predacious species are secondary consumers; whereas, scavengers, like many detritivorous microarthropods, are decomposers. Trophic relationships are fundamental to understanding the structure and function of terrestrial ecosystems. The data in most studies on ant diet are gathered in the field by collecting the prey that workers transport to the nest. However, fresh prey may come from predation or from scavenging. Very recent studies on food webs use stable isotopic analysis (N isotopic values) to assess the main type of food sources for each species [9,24]. But without direct observations, it is impossible to disentangle scavenging/predation categories because the isotopic values are the same whether the insect has been collected dead or alive. The study of PCR-based techniques (i.e., DNA gut-content analysis) are highly efficient and sensitive, both in fresh and carrion prey detection [25]. One approach to disentangling active predation from scavenging might be to use isoenzyme electrophoresis, as this method relies on active prey enzymes which may become altered after death [25].

Ant predation

Ants may act as ecosystem engineers, but they can have a negative effect on other soil engineers (see sections below): the ponerine African ant Psalidomyrmex procerus is a specialist predator of earthworms that may have an intense impact on its prey populations [18]. The African army ant Dorylus molestus preys mainly on earthworms. After swarm raids, there is an immediate decrease in earthworm numbers (the estimated proportion of earthworm prey biomass extracted by driver ants and swarm-attending birds was about 2%), but 8 days later the number returned to pre-raid levels [19]. On the other hand, earthworms seem to be associated with red wood ants (Formica aquilonia) in Finnish forests [20]. The ant nest mound surface (the uppermost 5-cm layer) harbors a much more abundant earthworm community than the surrounding soil. Earthworms are not preyed upon by the ants because their mucus repels the ants (suggesting a chemical defense against predation) [20].

A bit of optimal foraging theory Robert MacArthur and Eric Pianka are considered to have presented the first paper on the Optimal Foraging Theory [26] aimed at understanding the determination of diet "breadth" (i.e., the range of food types eaten by an animal) within a given habitat. These authors argued that, in order to obtain food, any predator must expend time and energy, first in searching for its prey and then in handling it (i.e., pursuing, subduing and consuming it). While searching, a predator may encounter a wide variety of food items and have different responses. Generalists collect a large proportion of the prey types they encounter, while specialists continue searching until they encounter the prey of their specifically preferred type [27]. For each foraging strategy, the predator has a "problem" to solve: if it is a specialist, then it will only pursue profitable prey items, but it may expend a great deal of time and energy searching for them; whereas, if it is a generalist, it will spend relatively little time searching, but will pursue both more and less profitable types of prey. An optimal forager must balance the pros and cons so as to maximize its overall rate of energy intake [27]. Although the consequences of individual foraging decisions on fitness are relatively straightforward in most solitary animals, they may be more complicated to disentangle in social animals. Hence, while the errors made by solitary animals have an impact on them alone (in terms of reduced fitness), those by social animals have negative consequences at both the individual and colony levels. To prevent such errors, insect societies have developed efficient, collective mechanisms for reducing the risk of error and conveying information about food sources [28].

Xim CerdĂĄ & Alain Dejean

Since ants are social insects, the decision to exploit a food source is made at two different levels: at the individual level (when the worker carries the food to the nest and communicates this to nestmates) and at the colony level (when social strategies such as recruitment using chemical trails are employed to collect the food source) (see Box 2). In many species, foragers are able to â&#x20AC;&#x153;measureâ&#x20AC;? food characteristics (e.g., quality, quantity and transportability), deciding whether or not to recruit accordingly [28-31]. The social integration of individual information about food emerges as a colony decision as to whether to initiate or to continue recruitment when the food patch is rich.

Central-place foraging With regard to animals that carry the food they find to a fixed place (i.e., central-place foragers, as is the case for ants that carry the food to the nest), the Central-Place Foraging Theory (CPFT) explains foraging behavior and food choice [32]. The CPFT predicts that if food size is independent of the costs of manipulating the item, optimal foragers will choose larger items far from the central place. CPFT also predicts that workers will be more selective in their food choice when farther away from the central place. There are four basic assumptions to this model [32]: (1) organisms behave in ways that maximize the net rate of energy gain; (2) all prey types are encountered randomly and, thus, foragers can look for different food types simultaneously; (3) there are no additional time costs in traveling with a food item; and (4) the energy costs for traveling with a load are greater than the energetic costs of traveling without a load. Some studies on foraging in seed harvester ants have suggested that these organisms routinely violate two CPFT assumptions [see 33]. First, the energetic costs of foraging are minimal and do not appear to have a significant role in seed choice. Second, food size is not independent of time costs. Larger seeds cause an increase in the time required to return to the nest. Workers regularly choose loads that are large in relation to their body size, and this causes a substantial increase in the time required to take an item to the nest. Another unstated assumption of classical CPFT is that all foragers behave independently of one another. In individually-foraging ant species, the net energy gain of individuals may be relatively unaffected by nestmates; but in species where workers cooperate for food retrieval, individual actions are not independent of one another [33]. Most studies about optimal foraging theory and CPFT in ants have focused on seed-harvesting and leaf-cutting ants. It is easier to manipulate or to offer workers a seed or a leaf than a prey (which can escape). There are

Ant predation

very few studies analyzing predation by ants in a CPFT framework. This could be the case for Pachycondyla tarsata (formerly Paltothyreus tarsatus), a generalist ponerine predator that strongly prefers termites and very large prey such as giant diplopods and crickets to other choices within its diet. Its workers show an adaptive predatory strategy compatible with CPFT: CPFT predicts that single-prey loading is an extension of the optimal diet choice since the greater distance from the central place enhances selectivity, while multiple-prey loading behavior would correspond to optimal patch use. According to the kind of prey discovered, P. tarsata workers properly choose one of the two foraging strategies predicted by CPFT. While hunting its favorite large prey, the recruitment of nestmates enhances the efficiency of total predation (single-prey loaders). The strategy for capturing small, aggregated prey (grouped termites) is characterized by the loading of multiple prey at a single time (multiple-prey loaders) through a concentrated search in a restricted area (optimal use of patches) and by an optional recruitment of nestmates from starved colonies [34]. Another study on CPFT focused on the wood ant, Formica rufa, but not on its prey choice or predation behavior, but its trail use. Using experimental colonies in the laboratory, it set out to test the CPTF assumption that colony efficiency is expected to be maximized by minimizing the lengths of established trails [35]. Wood ants made clearly adaptive behavioral adjustments in their choices of foraging trail routes and tended to use the shortest route whenever possible. The conclusions of this study contradicted some previous conclusions for harvester ants [36], showing that the theory may be flawed if it cannot be extended to other groups.

Foraging strategies of predatory ants Foraging, that is the collection of resources from the environment, has two phases: the search for the resource and its recovery. Both phases account for the costs, but only recovery produces a tangible benefit [37]. The foraging strategies of predatory ants (see Fig. 1 for different examples) fall mainly into two categories. Small prey items are captured by either single workers using their mandibles or sting (solitary hunting), or groups of ants foraging cooperatively, forming large raiding groups or swarms, thus enabling them to overwhelm large prey items or other social insects (cooperative hunting) [24]. Solitary hunting is the most common method employed by predatory ants, in some cases coupled with the recruitment of nestmates when necessary in order to transport the prey (see Box 2). The ponerine ants Gnamptogenys moelleri and G. sulcata, for example, hunt solitarily, but can retrieve both solitarily (small items) and in a group of recruited workers (large

Xim CerdĂĄ & Alain Dejean

Figure 1. Examples of different ant strategies for preying on arthropods. A. Solitary hunting. A Platythyrea conradti worker is capturing a locust by sliding its gaster under the preyâ&#x20AC;&#x2122;s thorax in order to sting it ventrally. This permits the venom to act on the ventral neural chain. B. Group ambushing. Oecophylla longinoda workers spreadeagling a praying mantid; they never use their venom during prey capture. C. Coordinated group ambush. With their mandibles wide open, Azteca andreae workers ambush side-by-side under the leaf margins of their host tree, the myrmecophyte Cecropia obtusa. Insects alighting on the leaves are seized and then spread-eagled. These workers are able to capture comparatively large prey thanks to their hookshaped claws and the velvet-like structure of the underside of the leaves, both combining to act as a natural VelcroÂŽ. D. The use of a trap. Allomerus decemarticulatus workers build gallery-shaped traps by manipulating their host-plant trichomes and fungal mycelium that they use to form a composite material pierced with holes. They ambush under the holes, and seize the extremities of insects landing on their trap and pull backward, immobilizing them. Recruited nestmates then use their venom to paralyze these prey. (Photo credits: Alain Dejean).

items) [31,38]. Similarly, foragers of the ant Formica schaufussi search for prey individually and recruit nestmates to large arthropod prey and cooperatively transport them to the nest [39]. Cooperative hunting is considered more evolved than solitary hunting because it implies cooperation between workers and results in a greater range

Ant predation

of prey sizes that a species can exploit. Among generalist predatory ants, group hunting is known principally in army ants, which are swarm raiders, and in dominant arboreal ants, such as Oecophylla. In both cases, the colonies are very large. For army ants, workers forage in groups during swarming, while for Oecophylla they stalk [40]. In both cases, they have developed a system of short-range recruitment that permits numerous workers to overwhelm large prey by spread-eagling them. Army ants generally carve up large prey on the spot, while Oecophylla always retrieve large prey whole. Myrmicaria opaciventris is an African myrmicine ant with very large, polydomous and polygynous colonies. M. opaciventris uses a group hunting strategy enabling the species to overwhelm very large prey items [40]. Paratrechina longicornis (Formicinae) workers also participate in a type of group hunting. Each individual forages, surrounded by nestmates behaving in the same way and within range of a recruitment pheromone. They detect prey through contact with successful workers; then, they recruit nestmates at short range and all together they spread-eagle the prey and retrieve them whole [41]. An example of a highly-evolved cooperative strategy is army ant teams [42]. Army ants form groups, with a definite structure, to retrieve large prey. These groups have a distinct caste (worker size) distribution. They typically consist of a large front runner, often a submajor, which is assisted by smaller workers that prevent the prey item from dragging on the substrate [42-43]. The workers are able to assess their own performance and their potential contribution to a group effort, and they act as a superefficient coordinated team for optimizing large food item retrieval [43]. A foraging strategy may be modulated by the workers of the colony. Edward Wilson was one of the first to study in depth how ant colonies modulate foraging according to the food source. In his already classical study, he showed that the fire ant, Solenopsis saevissima, organizes worker recruitment through trails as a function of food quality [46]. From an adaptive perspective, the more flexible the foraging behavior, the more readily the colonies may adjust to environmental changes [47]. Deborah Gordon [48] considers behavioral flexibility to be the process by which an animal changes its behavioral patterns when the environment changes. In ants, for example, an individual worker may change from individual retrieving to grouprecruitment. According to the weight and size of their prey, Ectatomma ruidum workers can employ different recruitment systems (e.g., solitary hunting, cooperative hunting and group hunting with recruitment) [47]. Pheidole pallidula ants shape their recruiting behavior simply according to the preyâ&#x20AC;&#x2122;s tractive resistance [29]. Some ant species are able to â&#x20AC;&#x153;measureâ&#x20AC;? food size or patch richness and recruit accordingly: Formica rufa scouts that find baits with six larvae recruit more workers and more rapidly than for baits

Xim Cerdรก & Alain Dejean

Box 2. Different prey transport strategies by ants The transport of a prey is a crucial phase of foraging. Depending on the species and the prey size, different strategies may be employed. These strategies can be summarized as individual or social (recruitment). Social strategies are the different types of recruitment used when the worker is not able to transport the prey individually, and the different communication systems used to recruit nestmates to the food source.

Figure 2. Main strategies employed by ants to collect and transport small prey. (Modified from [44]). 1. Individual: the successful solitary forager collects the food that she is able to transport alone. There is no transmission of information about prey discovery to nestmates. 2. Tandem-running: when the forager comes back to the nest, she recruits a nestmate and leads her from the nest to the food. The recruiter (in grey in the figure) and the recruit (walking behind her) keep in close antennal contact. This is considered the most primitive recruitment system because it only allows the recruitment of one worker, but it is a prey-size-dependent type of recruitment well adapted to small colonies [45]. 3. Group-recruitment: the recruiter (in grey in the figure) first lays a temporary chemical trail as she returns to the nest and subsequently leads a small group of recruits along this trail to the source. It is mainly used to collect solid food items by recruiting a few workers. 4. Mass-recruitment: in the most evolved strategy, recruiters returning from a food source to the nest lay a chemical trail that guides their nestmates to the source. While group-recruitment involves a leader, mass-recruitment is "anonymous". Moreover, in mass-recruiting species, chemical signaling prompts the formation of a permanent trail and the recruitment of hundreds of workers that monopolize the source. During food collection, for all the recruitment systems, the recruit may become the recruiter and activate a positive feed-back process. (Modified from [28,44]).

Ant predation

with only two larvae [30]. However, the modulation of foraging behavior is not limited to recruiting species. Decamorium decem and Serrastruma lujae are myrmicine ants that hunt solitarily for collembolans and other small arthropods in the leaf litter of African rainforests. During the dry season, collembolans aggregate in wet patches in the dry litter where numerous single workers hunt from their nest. When foragers reach a wet patch, they use areaconcentrated searching: they control the sinuosity and the speed of their food searching paths in order to concentrate their activity on these areas of high prey density [49,50].

Food robbing by ants Bert Hรถlldobler [51] proposed that the term "food robbing" include only those cases in which prey or any other food was directly taken away from the body of the forager ants by the robber ants. Food robbing is a widespread form of interference competition by means of which many animals reduce the costs of searching for, handling, and obtaining food [52]. This behavior is relatively common in predacious and scavenger ants, but social wasps also rob food from ants; Polybioides tabida (Polistinae; Ropalidiini), for example, rob pieces of large prey from Tetraponera aethiops (Pseudomyrmecinae) whose colonies live in the hollow branches of the plant Barteria fistulosa [53]. Food robbing has been frequently observed in the North American desert ant Myrmecocystus mimicus, which waylays the returning foragers of several Pogonomyrmex species at their nests and takes insect prey, particularly termites, away from them [51]. The tropical ponerine Ectatomma ruidum also robs food, and it is able to use the foraging trails of other species (e.g., Pheidole radoszkowskii) to find the returning workers and remove bits of prey from their mandibles [54]. One very original case involves Ectatomma tuberculatum and the myrmicine Crematogaster limata parabiotica, both of which are sympatric arboreal ant species that forage on the same pioneer trees. Most of the E. tuberculatum workers coming back to the nest carrying a droplet of liquid food (of Hemiptera honeydew) between their mandibles were robbed by C. l. parabiotica. This is not solid prey robbing, but rather a case of sugary food robbing [52]. Georges Oster and Edward Wilson [55] inferred from elementary mathematical models that the relationship between prey size and the probability of interference (e.g., food robbing) is somewhat sigmoidal: as prey size increases, so does the probability of interference competition. This prediction has been confirmed through different field studies (see an example in Box 3) with insect prey and baits: ant species that hunt large prey are

Xim Cerdรก & Alain Dejean

subject to higher levels of interference during foraging than species that hunt small prey [56-59]. A recent comparative study about the prey retrieval strategy (fragmentation vs. transportation whole) in 44 Asian ant species [59] showed that the workers of most arboreal species cut up large prey at the site of capture, and individual workers retrieved the smaller pieces. In contrast, in ground-dwelling species, the most frequent strategy was for a group of workers to retrieve large prey cooperatively without fragmentation. Moreover, one of the most interesting results was that, on the ground, parts of the large prey item were often robbed by other ant species during transport, while such interference was rare on trees [59]. Box 3. Food robbing in a guild of Mediterranean scavenger ants In open habitats in temperate or semi-arid ecosystems, predation by ants on arthropods is much rarer than in tropical ecosystems. In these habitats, the ant species that feed on arthropod items are mainly scavengers. This is the case for the thermophilous Cataglyphis cursor, C. rosenhaueri or C. velox, the group-recruiting Aphaenogaster senilis or A. iberica, and the mass-recruiting P. pallidula, Tapinoma nigerrimum or Tetramorium semilaeve [58,60]. Some of these species composed the guild of scavenger ants in a Mediterranean town (Canet de Mar, Barcelona, Spain) where interspecific interference interactions were studied [58]. Prey of different sizes (i.e., small: fruit flies; medium: cockroaches; large: crickets; and very large: baits) were offered to foragers from each species at different times of the day (i.e., morning, afternoon and night). For most of the species studied, prey loss (through food robbing) varied according to prey size, but dissimilarly so at the different periods of the day (when different ant species were present). Figure 3 shows the percentages of prey loss for medium and large prey for the morning and afternoon periods. First, food robbing is not only interspecific but also intraspecific: in the morning 8% of medium-sized Cataglyphis cursor prey and in the afternoon 20% of Aphaenogaster senilis were robbed by other workers from the same species. Second, prey size does matter: 38% of the cockroaches offered to C. cursor were snatched by A. senilis workers and, inversely, 14% of cockroaches offered to A. senilis were snatched by C. cursor workers; but when the prey offered were larger (i.e., a field cricket), only A. senilis was a successful robber, and C. cursor lost 56% of its prey. Third, the ecologically dominant species that were the mass-recruiting ants (i.e., P. pallidula, T. nigerrimum and T. semilaeve) were relatively unsuccessful facing the subordinate A. senilis; they lost most of the medium-sized prey (between 76 and 94%) and nearly one-third of the large prey (between 20 and 36%). A. senilis dissected the prey inefficiently, but transported whole prey to the nest in a single, highly cooperative and very rapid action by several workers. In contrast, dominant species were better able to defend larger prey by recruiting a large number of workers before their competitors were able to intercept and carry them to the nest. There is a trade-off between dominance at food resources and speed of food location and transport.

Ant predation

Figure 3. Percentages of thefts of medium (i.e., German cockroaches: Blattella germanica) and large (i.e., Mediterranean field crickets: Gryllus bimaculatus) items among different scavenger ants in a grassland (Canet de Mar, Barcelona, Spain). The species abbreviations are: Aph, Aphaenogaster senilis; Cat, Cataglyphis cursor; Phe, Pheidole pallidula; Tap, Tapinoma nigerrimum; and Tet, Tetramorium semilaeve. Thin arrows indicate the direction of prey robbing, including intraspecific robbing; tiny arrows indicate prey robbing by ant species not included in the study; thick arrows indicate the percentages of items carried back to the nest from those offered to each species. Two periods of the day were considered separately: the morning (n=50 items of each prey type) and the afternoon (n=25 items of each prey type). (Modified from [58]).

Army ant behavior Army ant adaptive syndrome is defined as “a life-history” characterized by group predation, nomadism, permanently wingless queens, and dependent colony founding [61]. Classically, three ant subfamilies are considered “true” army ants: Aenictinae, Ecitoninae, and Dorylynae. However, none of the army ant traits are restricted to these families, as they also occur in distantlyrelated ant species, including members of the subfamilies Amblyoponinae, Cerapachynae, Leptanillinae, Leptanilloidinae, Myrmicinae and Ponerinae. Army ants are characterized by the raids they conduct on large arthropods or social insect prey (see section "Preying on social insects" below; that is, they are ants preying on ants). The adaptive value of nomadism seems clear; by continually moving into fresh hunting grounds, only predatory ants are able to build large colonies [62]. Army ant colonies have a rigid temporal pattern of activity associated with the development of their brood, which, to a large extent, dictates their foraging pattern [37,63] (see Box 4). The relevance of most army ants to ecosystem functioning remains globally poorly understood [61]. However, in the ecologically best-known army ant species, Eciton burchelli, a raid triggers (by preying on other ant

Xim Cerdรก & Alain Dejean

colonies) a process of change, similar to succession, in the ant community and also favors the establishment of prey species colonies: incipient prey species colonies are founded in greater abundance in recently-raided areas [37]. Box 4. Foraging and migration pattern in the army ant Eciton burchelli In a noteworthy study conducted at Barro Colorado Island, Panama, Nigel Franks and Charles Fletcher mapped, on a daily basis, the position of each swarm's principal trail to describe the changes in the spatial pattern and raid systems of Eciton burchelli colonies [63]. This species inhabits the tropical American lowland rainforest and its colonies stage the largest army ant raids: a single swarm raid may contain up to 200,000 ants and average 6 m wide [63]. The raid moves as a phalanx of ferocious workers; the swarm front proceeds in a zigzag pattern, so that the overall course of a raid is roughly a straight line [37]. Only one swarm is produced per colony per day and the raiding ants return with their prey to the nest by a principal trail, while others move out to join the swarm [63]. The colonies of this species maintain, throughout their lives, a 35-day cycle of activity (see Fig. 4). Colonies alternate bouts of centralplace foraging (statary phase) with periods of nomadism. During the statary phase, the colony uses the same nest site during an average of 20 days, and produces raids like

Figure 4. The 35-day behavioral cycle of Eciton burchelli in Barro Colorado Island, Panama. The foraging and migration pattern of the brood cycle is diagrammatically represented above. Numbers indicate the raid sequence. See text for further explanation. (Modified from [37] and [64]).

Ant predation

the spokes of a wheel from the hub of the central bivouac. Successive raids are separated on average by 123Âş. This system maximizes the separation of neighboring foraging paths in time as well as in space, allowing time for their general arthropod prey to recover before the next raid [37]. During the statary phase, the brood is mainly composed of eggs and pupae and workers have only themselves to feed. At the end of this central-place foraging phase, the eggs and pupae hatch into larvae and callow workers, respectively, and the nomadic phase begins. Emigration is constrained and determined by the foraging patterns because it follows the principal trail of the nomadic raid. To feed its voracious larvae, the colony conducts raids every day; after a 15-day nomadic phase, the larvae all pupate and the colony enters another statary phase [37,63].

Generalist predatory species Generalist predators and biological control Most ground-dwelling ant species are omnivorous, feeding on nectar or plant exudates, Hemiptera honeydew, and preying (or scavenging, see Box 1) on other arthropods. In a Scottish forest, the total daily colony intake for the wood ant, Formica rufa, was composed of 44% honeydew and 66% solid food. [65]. In other British F. rufa populations, caterpillars were 3-4 times more abundant on sycamore trees not explored by ants than on those where workers foraged [66]. A large colony of F. rufa in Germany has a daily intake of 65,000-100,000 caterpillars (Gosswald 1958 in [67]). This heavy predation of herbivores by wood ants has led to breeding them for biological control purposes [68]. The type of response that predators show towards a prey population can have a marked effect on the population dynamics of the prey. Egg predation by ants can reach 71%, and be an important mortality factor for Cactoblastis cactorum in South Africa. Because ants are polyphagous, their population dynamics would probably only be negligibly affected by fluctuations in the density of C. cactorum eggs as these would form only a small component of the antsâ&#x20AC;&#x2122; diet [69]. An interesting case of adaptation to prey availability is that of a North American seed harvester ant, Pogonomyrmex rugosus, which forages intensively on the seeds of herbaceous annuals and annual grasses, but exhibits a "pulse" of predation in response to a short-duration episodic event (e.g., the emergence of a large number of prey; in this case, grass cicadas). This pulse of predation demonstrates the importance of protein to seed harvester ant colonies [70]. Predatory ants can significantly affect the behavior of prey and depress the size of potential pest populations (see Box 5) [71]. The published literature has emphasized seven genera of dominant ant species that are either

Xim Cerdรก & Alain Dejean

beneficial or potentially beneficial predatory ants: Oecophylla, Dolichoderus, Anoplolepis, Wasmannia, and Azteca in the tropics; Solenopsis in the tropics and subtropics; and Formica in temperate environments (see reviews in [21,72]. Box 5. Ants as biological-control agents Victor Rico-Gray and Paulo Oliveira consider that the most important attributes of ants, which make them potentially useful biological control agents, are [71 page 217]: 1. 2. 3. 4. 5. 6.

Their diversity and abundance in most tropical and temperate ecosystems, and the fact that most can be considered predators; Their response to changes in the density of prey; Their ability to remain abundant even when prey is scarce because they cannibalize their brood and/or use plant and insect exudates as stable sources of energy; Their ability to store food and hence continue to capture prey even if it is not immediately needed (i.e., predator satiation is not likely to limit the effectiveness of ants); That they can, in addition to killing some pests, deter many others including some too large to be successfully captured; and That they can be managed to enhance their abundance, distribution, and contact with prey.

Defenses against generalist predatory ants Arthropod prey may develop different defense mechanisms to protect themselves from ant predation. Many lepidopteran larvae have chemical defenses against ants. The accumulation of sulfur amino acids in the caterpillars of the leek moth, Acrolepiopsis assectella (a pest moth with different economic host plants: leek, onion, garlic, etc), makes plausible the role of alkyl-cysteine sulfoxides of Allium in the protection of A. assectella from Formica ants [73]. In a wider study [74], Lee Dyer offered 70 species of lepidopteran larvae extracts to the predatory ant Paraponera clavata to examine the effectiveness of larval antipredator mechanisms and to test the assumption that diet breadth and chemistry are important predictors of predation responses. He offered caterpillar extracts paired with sugar water controls to the P. clavata colonies, and then measured the degrees to which the extracts and the controls were consumed. The extracts were considered unpalatable when the ants consumed more control than extract. Prey with unpalatable extracts were frequently rejected by P. clavata, while prey with palatable extracts were rarely rejected. He concluded that plant specialist caterpillars were better protected than generalists.

Ant predation

In the tropics, where the majority of social wasp species dwell and the greatest diversity in nest architecture occurs, predation by ants on the brood is believed to be a major force in nest evolution [75]. The brood of social wasps is particularly vulnerable for several reasons: it occurs in large concentrations, is exposed in open cells, and the nests have a long durability, making the chance of discovery by predators relatively high. Tropical social wasps can be divided into two major groups according to the type of nest and the evolution of the type of defensive behavior used against ants: a single comb of cells suspended from a petiole or nests enclosed by an envelope [75]. Most solitary founding species construct small uncovered nests, where the petiole is built of a tough material of glandular origin and may have antrepellent properties [75,76]. However, nest protection is not only provided by physical or chemical barriers against ants. They have developed behavioral (e.g., different alarm signals) and ecological mechanisms such as the association with plant-ants. As a result of strong predation pressure from Ecitoninae army ants, the wasp Parachartergus apicalis nests mostly on Acacia trees occupied by Pseudomyrmex colonies. They benefit from the protection provided by the Pseudomyrmex that are very aggressive towards Eciton [77]. Similarly, other wasp species nest mostly on trees occupied by aggressive dolichoderine ants since this strategy is considered to be the only truly efficacious protection against army ants [78].

Specialized species and predator-prey arms race To enhance their efficiency by reducing the time and energy necessary to overwhelm their prey, many predatory ants are engaged in a kind of arms race where they have evolved morpho-physiological adaptations to the different potential means of prey defense or escape. Potential prey in this case include animals contributing to the cycle of degradation of the leaf litter and wood fallen to the ground such as earthworms, isopods able to roll into a ball or to escape swiftly, centipedes, millipedes and polyxena among the Myriapoda, collembolans able to jump thanks to their furca, termites and other ants, and even some newly-emerged flies. Note that the phylogenetic distribution of the most specialized predatory ant species is disproportionately concentrated in the morphologically primitive subfamily Ponerinae. Yet oligophagy is a derived character as most genera in this subfamily are polyphagous predators or include both polyphagous and oligophagous species [79]. Like for other insect predators, the success of foraging predatory ants depends on several factors including prey detection (at a distance rather than by contact), effectiveness of prey seizure (i.e., the role of the mandibles) and prey immobilization (i.e., the role of the venom) increasing the speed at

Xim Cerdá & Alain Dejean

which the prey are mastered. Ants have evolved in different ways in relation to their predatory activity; for example, the shape of their mandibles has changed and they have developed increasingly powerful venoms generally correlated with a specific type of behavior, whereas the group hunting strategies that some ant species use are known to be an evolved trait [79].

Mandible morphology, mechanics and neurophysiology of mandible closure Trap-jaw mandibles Three tribes of ants belonging to three different subfamilies – namely, the Odontomachini (Ponerinae), the Dacetini (Myrmicinae), and the Myrmoteratini (Formicinae) - have independently evolved the ability to strike prey extremely rapidly using their hypertrophied mandibles thanks to a socalled “trap-jaw mechanism” [79]. Here, the mandible closure in a fast strike is similar in design to a catapult and results from the release of stored energy that overcomes the constraints of the muscles. It is controlled by a monosynaptic pathway of giant neurons and a trigger muscle specialized in high contraction velocity. First, by contracting their adductor muscles, which are composed almost entirely of long sarcomeres or slow-contracting fibers, the ants store mechanical energy in their mandibles. Then, a structure, which acts as a kind of “latch” (i.e., the labrum or mandible protrusion, depending on the species), is suddenly released, so that the mandibles close quickly, striking and sometimes locking onto a prey. These structures keep the mandibles blocked open at approximately 180° in Odontomachus, Anochetus and Strumigenys, and up to 280° in Myrmoteras [80,81]. The mandibles close through a reflex mechanism that is triggered when the long sensory hairs located on the inner edge of the mandibles or on the labrum make contact with the target. They in turn monosynaptically stimulate giant motor neurons (with axons that have a particularly large diameter and that conduct information very quickly) commanding muscles specialized in high speed contraction that release the latch and thus trigger the strike [82-90]. The 4-10 millisecond latency of the entire reflex corresponds to one of the most rapid movements made by an animal [83,85]. Note that Myrmoteras toro does not have sensory hairs, but M. barbouri does [80]. After leaving their nest to forage, Odontomachus workers open their mandibles, and are ready to react when they encounter a potential prey. Although some Odontomachus species can feed on sugary resources such as extrafloral nectar and the honeydew produced by Hemipterans, and even if

Ant predation

they can capture a large variety of small ground-dwelling arthropods, they are mostly termite predators. The mandible strike permits them to numb the termite soldiers during head-on encounters in galleries. Anochetus species (Odontomachini) are also mostly specialized in termite predation, while ground-dwelling Dacetini are specialized in the capture of Collembolans, something also noted in Myrmoteras barbouri [79,80,91,92].

Nutcracker mandibles adapted to capturing long-shaped prey and the role of snapping Whereas, among the long-mandibled Ponerinae, the stenophagous Psalidomyrmex capture only earthworms, Plectroctena are predators of a relatively wide range of arthropods, but need millipedes in their diet for their colonies to be able to produce adult individuals [93, 94]. When they come upon small earthworms (4-cm-long individuals), Psalidomyrmex procerus seize and sting the anterior parts of the prey, immediately paralyzing the distal parts and enabling the ants to retrieve the entire prey. Large earthworms, in contrast, are seized by the part with which the ants first come into contact. When this is the anterior part, the workers sting the worm and - because autotomy is rather exceptional - the process is similar to the one described above, and the entire worm is retrieved. In the other cases, the worm undergoes autotomy, and the workers retrieve the part of the worm that they seized [93]. Hunting Plectroctena minor workers seize spirostreptid millipedes of up to 4 mm in diameter by their anterior part. The antsâ&#x20AC;&#x2122; mandibles slip on the exoskeletal coils of the millipedeâ&#x20AC;&#x2122;s body and are caught between two segments that are slightly separated by the strong pressure, allowing them to sting the millipede, mostly on the ventral surface in the soft intersegmentary space of the seized zone. The venom acts quickly on the ventral neural chain, immediately paralyzing the distal parts of the millipedes and thus permitting the ants to easily retrieve them. The workers seize larger millipedes by an appendage before stinging them or by wrapping themselves around the prey to form a kind of collar. Once the millipede is paralyzed, they recruit nestmates to help retrieve it. Very large millipedes (95-to-105 mm-long individuals; approximately 8 mm in diameter) can be captured only if encountered in galleries (or test-tubes during experiments). During head-on encounters, the workers grip one of the millipedeâ&#x20AC;&#x2122;s antennas or mandibles and sting the very end of its body, triggering rapid paralysis and so singly mastering and then retrieving it. This is the highest body weight ratio between a prey and a predator ever noted for ants that hunt solitarily as these millipedes weigh 94 to 117 times as much as a worker. It is difficult,

Xim Cerdá & Alain Dejean

however, for the hunting worker to seize the hind part of the millipede, so that it recruits nestmates in most cases; nevertheless, the process of mastering the prey is relatively long [95]. Plectroctena (Ponerinae) [96] and Mystrium (Amblyoninae, another subfamily of the Poneromorphs) [97], have mandibles that are able to snap, something that is mostly known in termite soldiers and used in nest defense. The ants snap their mandibles by contracting the adductor muscles while bracing the tips of their long mandibles against one another, storing mechanical energy much as a catapult does. It is supposed that an imbalance in the energy stored in the left and the right mandibles is released when one of the mandible tips pivots, and that the strike is initiated when there is contact with the mechanosensory hairs on the mandibles [89]. Snapping can occur in Plectroctena minor during prey capture, especially in the case of termite soldiers that are stunned. Nevertheless, because prey are classically captured through seizure and stinging, this snapping behavior is mostly used for colony defense as intruders are snapped at if they are encountered close to the nest entrance [96].

The pitch-fork shaped mandibles of the genus Thaumatomyrmex Colonies of Thaumatomyrmex spp. contain only a few workers equipped with slightly asymmetric, pitch-fork shaped mandibles [98,99]. They are stenophagous, specialized in the capture of Polyxenidae (Diplopoda, Penicillata, Polyxenida) and known to use a very efficacious anti-predator strategy based on the projection of detachable barbed and hooked trichomes [100]. A foraging worker encountering a polyxenid first palpates it with the tip of the antennae, and then seizes and stings it on the intersegmentary membranes, rapidly immobilizing the prey. The worker then seizes its prey at the base of the head, avoiding contact with the prey hairs, and transports it over its own head. Once inside the nest, the successful hunting worker places its paralyzed prey on the ground, and grasps it with its mandibles whose teeth penetrate through the layer of trichomes permitting the workers to get a good grip on the prey’s body. Then, they strip the trichomes down to the polyxenid’s integument using the short, stout setae on the tarsi of their forelegs. The worker then eats its prey, starting from the head. Sometimes it shares it with a nestmate, or feeds the entire prey or the remains to larvae. Then, using their forelegs, the workers transport the piles of polyxenid trichomes away from the nest [98,99]. Although they are not equipped with pitch-fork shaped mandibles, workers of the Asian ant Probolomyrmex dammermani also specifically hunt polyxenids and strip the prey’s trichomes inside their nest [101].

Ant predation

The falciform mandibles of Leptogenys Leptogenys workers are armed with long, thin, curved mandibles articulated at the extreme corners of the anterior margin of the head. Most of the species, whether they are African, American, Oriental or Indo-Australian, have small colonies with solitary foragers specialized in the capture of isopods. This stenophagy seems to correspond to a plesiomorphic character as dietetic diversity is associated with species having larger colonies and hunting in a group, two characteristics considered as derived in ants [79, 102-104]. Note that among the species with solitary hunters, L. benghazi is specialized in termite predation and Leptogenys sp. 13 in earwig predation [105,106]. Oniscoid isopods, protected by a shell and tegumental gland secretions, present three types of behavioral and morphological defenses: “rollers” can roll into a ball, “spiny forms” have long spines that stick out when they roll, and “runners” are capable of rapid escape. Based on the size of their mandibles and the size of the prey, Leptogenys workers can seize these prey by their body whether or not they are rolled up (they do not sting the prey) or, alternatively, by the edge of the shell; they then turn the prey over and sting it on the ventral surface. Because this ventral surface is membranous, it is easily penetrated, and the venom quickly reaches the ventral neural chain, facilitating paralysis. Species with long mandibles can easily seize their prey by the body, while large prey are more often seized by the edge of the shell [104]. Spiny isopods are easily captured as the workers can seize them by the edge of the shell, negating in this way the defensive role of the spines. Runners that generally escape swiftly from antennal contact are seized by the edge of the shell. After they have successfully escaped once, many prey are finally captured thanks to the so-called “reserve behavior” of the ants whereby excited workers find them a second time, seize them by the edge of the shell and sting their ventral surface. Bathytropid isopods are attracted to the nest of L. mexicana colonies. They enter the nests where they move slowly or remain immobile for a long time if not detected by the workers, and so they go "into the mouth" of their predator [104]. Among species with larger colonies, two types of group hunting have been noted. “Megaponera-like” foraging takes place when a scout, having successfully discovered a group of prey, lays a recruitment trail while returning to the nest; this trail is then followed by a group of nestmates [107,108]. Army ant-like mass foraging behavior is known in the Asian species Leptogenys distinguenda that has very large colonies (several thousands) [109,110].

Xim Cerdรก & Alain Dejean

Long mandibles, detection by sight and jumping Workers of the Indian ponerine species Harpegnathos saltator have long mandibles equipped with trigger hairs on their inner surface. When the mandibles are open, the contact of these hairs with the prey body triggers a strike. Workers hunt by sight thanks to their compound eyes with ca. 1,600 ommatidies permitting binocular vision centered on the mandibles. This makes it possible to detect prey at relatively long distances, something that is helped by an adaptive behavior enabling them to jump. Indeed, by using both their median and hind legs, workers can jump and hence capture prey escaping by running or even flying away, such as cockroaches or dipterans. Even 10% of the prey located approximately 10 cm from the workers are captured, and the rate of successful capture attempts reaches 80% for prey detected at 3-4 cm. Furthermore, workers use their jumping ability to escape from their own enemies. In this case, they can jump distances of up to 20 cm [111,112].

Specialized predatory ants with behavior not automatically related to mandible shape The evolution of prey specialization globally followed the diversification of ant subfamilies during the geological ages. Indeed, specialization has mostly concerned ground-dwelling species, with poneromorphs being the most represented and some Myrmicinae and dorylomorph subfamilies also concerned. The Formicinae are represented by the genus Myrmoteras (preying mostly on collembolans), while the Dolichoderinae are not represented [79]. Egg predators Although they are generalist predators, several ant species can opportunistically gather arthropod eggs, but the African Ponerinae Plectroctena lygaria, Proceratiinae of the genera Discothyrea and Proceratium as well as the Myrmicinae Erebomyrma and Stegomyrmex can be specialized in preying on arthropod eggs [79,113,114]. Several species, such as Plectroctena lygaria, are specialized in gathering myriapod eggs and in storing millipede eggs in their nest chambers [115], while Stegomyrmex vizottoi workers are specialized in gathering diplopod eggs that they retrieve by holding them between the ventral surface of the opened, curved mandibles and the hairy anteroventral region of the head [114]. Proceratium silaceum tuck the slippery eggs between their opened mandibles and their downward-pointing gastral tip during transport [79].

Ant predation

The African species Discothyrea oculata is specialized in spider egg predation. Both founding queens and workers are able to open spider oothecas and even manipulate the crimped silk that protects them so as to line and obstruct cavities where the colonies settle thanks to forelegs that are equipped with parallel claws and a comb-shaped tibial spur opposite a brushlike zone on the first article of the tarsa. Furthermore, both founding queens and adult colonies can settle in the oothecas where they find both shelter and food. Yet, hunting workers can also forage for new oothecas that they open easily with their forelegs. Spiderlings emerging from eggs are stung and eaten [93,116]. Collembolan predators The tribe Dacetini is composed, among others, of two genera, Strumigenys and Pyramica, whose representatives live in the leaf litter of subtropical or tropical countries. Although some Asian species prey mostly on soft-bodied ground- and litter-dwelling arthropods [117,118], most Strumigenys and Pyramica species are specialized in collembolan predation [92,117, 119-125]. Yet, when the colony is starved, hunting workers have a tendency to capture alternative prey [126]. Typically, hunting Strumigenys workers forage in a slow, erratic movement. When detecting a prey at a very short distance (a few mm), they stop, open their long mandibles to approximately 180°, exposing the pair of trigger hairs that rise up from the labral lobes and extend forward from the ant’s head. Special teeth at the base of the mandibles catch on the lateral lobes of the labrum while the adductor muscles tense. Then, the workers approach the prey very cautiously and when the two trigger hairs touch the prey’s body, the labrum drops, triggering a violent strike. The prey’s body is generally impaled on the mandible’s apical teeth. Stinging is not always necessary, but struggling prey are lifted and stung into paralysis [92,117,119]. Among the species now regrouped under the genus Pyramica, one can distinguish medium- from short-mandibled species. The mandibles only open to 90° maximum and the prey is seized by an appendage or, for Pyramica benten, even at the outer circumference of the cephalic cavity from which the mouthparts protrude, so that stinging is necessary [117-120]. Species with very short mandibles have a tendency to halt when perceiving the presence of a prey, or even ambush it. In both cases, they have a “pointing phase” where they crouch by lowering their head against the substratum, fold their antennae into their scrobe while they open their mandibles between which are the trigger hairs that protrude from the labrum. These workers approach the

Xim CerdĂĄ & Alain Dejean

detected prey very slowly, placing their mandibles on both sides of a prey appendage; seizure occurs when a prey movement makes its appendage come into contact with the trigger hairs [119,123]. Prey seizure for both mediumand short-mandibled species is followed by the struggling of the prey which is stung on the seized appendage. Other examples include the ability of short-mandibled Asian species to camouflage themselves by smearing their bodies with organic material that they gather from the ground with their mandibles. The workers then use their forelegs to scrape the dorsum of their head and thorax with this material [118]. In African species of the former order Smithistruma, the workers attract collembolans. The source of the attractant, unknown, could be the spongiform appendages of the petiole or secretions from the labrum [122]. Several studies have illustrated the fact that the predatory behavior of these ants is not stereotypical. Prey anaesthetized with carbon dioxide are not stung if encountered far from the nest, but are stung if found next to the nest entrance, illustrating the influence of territoriality. They are also stung if the colony was previously starved. If, during a capture attempt, the prey successfully escapes, the worker uses the â&#x20AC;&#x153;reserve behaviorâ&#x20AC;? consisting of an intensive searching process where both the sinuosity of its trajectory and its speed increase, facilitating the retrieval of the prey. If the prey (or another insect) is encountered, it is immediately attacked, seized and stung [92,123,126-129]. This reserve behavior has also been found in all groundhunting ant species studied so far [47,50,81,130,131]. Colonies of short-mandibled species that are bred in the laboratory are fed with alternative prey (Psocidae) that escape when the workers are in their pointing phase. The workers only capture these prey using the reserve behavior once the colony begins to starve. Then, as the colony ages, the workers starting to hunt for the first time will only encounter Psocidae that they attack directly, so without going through a pointing phase. Several months later when confronted with collembolans, their principal prey, these workers never go through the pointing phase. Thus, pointing is acquired only if the workers passing from the internal to the external service of the colony encounter collembolans as they begin to hunt. Otherwise, they act as generalist predators, which enables them to capture alternative prey when the dry season arrives and collembolans are more and more rarely found in the leaf litter [129]. Myriapod predators We have already seen how, by using their pitch-forked mandibles, Thaumatomyrmex capture and strip polyxenids of their protective hairs.

Ant predation

Moreover, Plectroctena workers, thanks to their nutcracker mandibles, are specialized in the capture of millipedes; encounters in restricted environments such as galleries permit them to capture large items. This notion was illustrated again in a study on the dacetine ant species Pyramica hexamerus which is, on the other hand, a centipede predator. The workers hunt in galleries or crevices, crouching and remaining immobile when encountering a centipede. When the centipede steps on its lowered head, the worker attacks from below, and, in an upward strike of its mandibles, impales the prey trunk on its long, apical teeth [117]. Leptanilla (Leptanillinae) and Amblyopone (Amblyoponae) are predators that mostly capture centipedes [132]. Amblyopone are equipped with long mandibles placed at the extreme corners of the frontal part of the cephalic capsule, which allows them to seize the bodies of prey with a relatively large diameter. Like for trap-jaw mandibled species, two trigger hairs emerge from the clypeus, enabling the workers to adjust the degree of mandible closure [132]. Workers detect these prey from a distance and approach them cautiously so as to place themselves over the prey body and seize it. They then bend their gaster, whose extremity comes under the prey body, and sting the ventral surface where the neural chain passes. Because paralysis is not immediate, the workers move forwards along the prey body, their mandibles slightly opened and placed on both sides of the prey body, ready to seize it if necessary. Upon reaching the anterior part of the prey body, the workers bite it and sting it again. They then lick the paralyzed prey before dragging it to the nest [132,133]. In Amblyopone and Myopopone, larvae can be transported to the prey rather than the reverse [79].

Preying on social insects Attacking a social insect colony: Importance of the Lanchester theory of combat Unlike cases of lestobiosic species whose colonies live in the ant hills or termitaries of their host, predatory ants must penetrate the colonies of social insects to prey on them or on their brood, while vulnerable species have developed several means of defense (e.g., walls for termitaries, the presence of soldiers or major workers blocking the nest entrances, chemical defenses). The Lanchester theory of combat was proposed as a theoretical framework to explain combat between ants [134]. The Linear Law predicts that fighting ability contributes more towards victory than the number of combatants when a restricted combat area forces individuals to engage in a series of duels. Here, the presence of termite or ant soldiers blocking the

Xim Cerdá & Alain Dejean

entrance to a gallery leading to their nest is very important, particularly for termite soldiers that generally have a large, sclerotized cephalic capsule, powerful mandibles that can snap in certain species, and defensive anti-ant chemicals produced by their frontal gland. On the other hand, the Square Law predicts that when combatants can mix freely, numerical superiority is the deciding factor. Lestobiosis including termitolesty The colonies of some ant species with small workers take shelter in the walls of the large nests of other ants or termites. They enter the host nest chambers to steal food, or to prey on the host eggs and/or larvae. There are no confrontations between individuals, or they are much reduced. The relationship is therefore both “parasitic” with respect to nest sharing and “predatory” with respect to the brood thief [79]. Colonies of Solenopsis of the subgenus Diplorhoptrum (Europe and North America) generally nest next to larger ant species; the workers enter the other ant species’ nests where they prey on the brood. Diplorhoptrum fugax workers produce repellent chemicals (trans-2-butyl-5-heptylpyrolidin) permitting them to avoid attack by workers of the host colony. Species of Carebara, Caberella, Diplorhoptrum and Eberomyrma nest in the vicinity of termite mounds if not in their walls. They enter the termitary galleries and pouches where they steal termite eggs [79,135]. Workers of the ponerine ant Hypoponera eduardi, whose colonies develop inside societies of Reticulitermes (lower termites), have a kind of chemical camouflage through similarities between their cuticular hydrocarbons and those of the termite host [136]. Workers of the termitolestic species Tetramorium termitobium do not trigger an alarm among different higher termite species of the subfamily Macrotermitinae thanks to secretions of the mandibular gland [137]. In the African Ponerinae Centromyrmex bequaerti, entire colonies settle in the chambers and galleries of the termitaries of diverse Termitinae and Macrotermitinae species instead of in the walls. In this species, several evolved traits have been noted, such as: a strong dimorphism between the queens (large) and workers, something relatively rare in the Ponerinae where queens and workers are generally of a similar size; oligogyny (multiple queens, but mutually isolated by living in different chambers of the termitary); a polymorphic worker caste, as the workers are blind (a trait related to living in the termitaries); and relatively large colonies. Although all kinds of workers can hunt, this task is mostly limited to media individuals while, on the other hand, majors act as guards, blocking the entrances of the chambers where the colony is established [138].

Ant predation

Anochetus traegordhi colonies lodge in rotting logs exploited by Nasutitermes spp., so that hunting workers are at least partially cryptic to their prey. They attack termite soldiers head-on, and the mandible strike numbs them in more than 60% of the cases. They sting non-numbed soldiers on the thorax, while termite workers, approached from behind, are seized and stung on the abdomen. Anochetus workers are therefore able to distinguish termite soldiers from workers and to adapt their capture behavior to the situation [139]. “Lower” termites (i.e., the Termopsidae, Kalotermitidae, Prorhinotermes in the Rhinotermitidae), are one-piece nesting or wood dwellers since they inhabit and spend their entire life cycle in pieces of wood that also serve as feeding sites. Their nesting habits indirectly protect the colonies; soldiers exist and have likely evolved to defend the colonies from other termites [140]. In “higher” termites (i.e., the Serritermitidae and Termitidae) and most Rhinotermitidae, individuals must forage outside the nest and so are exposed to ant predation. Their defensive strategy consists of their termitary structure with thick walls and chambers connected by easily-defended galleries, foraging in subterranean or covered galleries leading to food sources and investment in soldiers [140]. Occasional termite predators The workers of many non-specialized ant species occasionally prey on foraging termites or on termite individuals exposed after their termitaries are broken open by a vertebrate or a falling tree. Consequently, to locate ant nest entrances in the tropical rainforest, researchers scatter portions of termitaries on the ground and track the ants that retrieve the termites [79]. In this case, the foraging strategy is very similar to hunting other kinds of insects, save for the fact that finding an individual termite is generally correlated to the presence of nestmates in the vicinity, resulting in concentrated searching (increased sinuosity and decreased speed) after discovering a first termite or detecting only termite tracks (acting here as kairomones), the successive capture of several individuals if the ant mandibles are long enough, and the rapid recruitment of nestmates [50,79,141,142]. Workers of the short-mandibled basicerotine ant Eurhopalothrix heliscata ambush solitarily in rotting wood where termites forage. They seize termite prey by an appendage and sting them [143] in a manner very similar to that described for short-mandibled Dacetini with regard to collembolans. The defensive mechanisms of termites are generally efficacious enough against most army ants, with only subterranean African doryline ants being specialized in termite predation [64]. Nevertheless, there is the exceptional

Xim Cerdá & Alain Dejean

case of the epigaeic species Dorylus rubellus, which successfully attacks fungus-growing termites of the genus Macrotermes [144]. Specialized termite predators The elimination of termite soldiers in order to gain access to workers or brood is therefore a challenge that specialized ant predators have to surmount. Chemical crypsis has been noted in several specialized termite predators including Crematogaster sp.C, and Decamorium uelense. Like in the termitolestic species Tetramorium termitobium, the mandibular glands of these species produce non-repellent aliphatic alcohols, whereas unspecialized congeneric species produce repellent ketones and aldehydes [137,145]. The elimination of soldiers of the fungus-growing termite Macrotermes bellicosus by workers of the African ponerine ant Pachycondyla analis was observed in laboratory conditions by using large pieces of termitaries whose galleries, connected to chambers, were opened laterally to form a narrow window permitting direct observation [146]. These termite soldiers guard gallery entrances through phragmosis (plugging the galleries with their large sclerotized head) aided by powerful mandibles and the secretion of toluquinone by the salivary gland [147]. Detecting these guards from a distance, the ants flexed their gasters under their thoraces and heads, extending their stingers towards the counter-attacking soldiers. The latter closed their mandibles, which then slipped on the tip of the ant’s fusiform, sclerotized gaster, and then remain closed due to the tetany of the adductor muscles. This behavior is effective as regards non-specialized enemies which are then killed by the chemical defenses, and, furthermore, participate in plugging the galleries. The Pachycondyla analis workers then, stingers extruded, deposited venom on the mouthparts of the now inoffensive termite soldiers due to the fact that their mandibles were locked in a closed position. The venom seemed to have an immediate topical effect, probably owing to the numerous, thin intersegmental membranes located there. The ants then pulled the termite soldier backward to to gain access to the termite workers [146]. According to the Linear Law of the Lanchester theory of combat, minor Pachycondyla analis workers have developed a specific behavior that grants them easy victory during combat in galleries. Pachycondyla analis (previously Megaponera foetens) has a typical foraging behavior now reported as ‘Megaponera-like behavior’ when noted for other ants specialized in termite predation. The scouts are major workers looking for foraging termites or even termitaries at as far as 95 m from their nests. After discovering a group of termites, thanks to the odors emanating from their foraging galleries and serving as kairomones, the scouts return to

Ant predation

their nests while laying a scent trail with their sting partially extruded. They recruit nestmates with 5-12 major workers and approximately 40 minor workers forming a column guided by the recruiting individual following its return path. All of the workers in the outgoing column in turn lay a scent trail while stridulating loudly enough to be audible by humans. When the column reaches the termite foraging area or the termitaries, the major workers break open the galleries. Only the minor individuals enter the galleries, and the attacks last approximately 9 minutes during which time the minor workers, in a series of exits and entrances, create piles of paralyzed termites around the gallery entrances. At the end of the raid, the major workers gather up to 10 termites that they pack between their mandibles; the minor workers gather fewer termites or none at all. The major workers, including the recruiting individual, lead the way along the return path that follows the same route as the outbound path [146-150]. Megaponera-like foraging behavior, with some variants, has been noted in other ant species such as the myrmicine ant Decamorium uelense that preys mostly on Microtermes, Leptogenys chinensis on Odontotermes or Hypotermes, Pachycondyla commutata on Syntermes, and Pachycondyla marginata on Neocapritermes [107,151-153]. Ant predators Melophorus anderseni (Formicinae) prey on the brood of Iridomyrmex sanguineus (Dolichoderinae). During encounters with Melophorus workers, Iridomyrmex foragers cower as they do when faced with large competing ant species (but Melophorus are small). This enables the Melophorus to rub their bodies against those of the Iridomyrmex, and thus to acquire their cuticular hydrocarbons (colony odor). Then, the â&#x20AC;&#x153;made upâ&#x20AC;? Melophorus individuals safely enter the Iridomyrmex nest where they steal larvae that they retrieve to their own nest [154]. New World army ants, or Ecitoninae, are ant predators with different levels of specialization. Some of them are even specialized in a particular genus or species [64,155,156]. When army ants begin to enter a nest, they release an allomone that triggers panic among the assaulted workers that then leave their nests, some of them carrying brood, so that fighting is avoided in most cases. The army ant workers then capture nearly all of the brood, callow workers and winged sexuals of the attacked colonies. Yet, they do not attack other workers or the queen(s), so that the attacked colony can reconstitute itself in most cases, particularly when the colonies have polydomous nests [64,155-159]. Nomamyrmex esenbeckii colonies organize subterranean raids on the very large colonies of the leaf-cutting, fungus-growing ants Atta colombica or

Xim Cerdรก & Alain Dejean

A. cephalotes. Both the perpetrator and the attacked colonies contain several million individuals. Nomamyrmex soldiers always attempt to penetrate the targeted colonies, while the entrance galleries of the Atta nests are defended by major workers that are rapidly recruited. The confrontations therefore take place in a restricted area where the ability of the fighters is primordial in a series of duels, according to the Linear Law of the Lanchester theory of combat [134]. While major Atta workers defend their nest entrances using their large mandibles and are helped by minor workers that attack vulnerable parts of the Nomamyrmex soldiers, the latter also use both their mandibles and their venom. There is, therefore, a complex distribution of tasks in the defensive strategy, so that the success or failure of a raid attempt will depend mostly on the rapidity with which the defense is organized. If Nomamyrmex soldiers are successful, all of the other workers will enter the Atta nest, so that this time the number of combatants is primordial, corresponding to the Square Law of the Lanchester theory of combat. In fact, two Nomamyrmex esenbeckii raids out of three are successful and approximately 60,000 Atta larvae are captured, representing one-third to the half of the brood in addition to the hundreds of major workers killed. In extreme cases, the Atta colony can die. Therefore, a two-stage strategy, where both the Linear and Square Laws come into play, exists in this case [160]. Territorially-dominant African ant species such as Oecophylla longinoda and Crematogaster spp. have workers that forage on the ground around the base of their host trees. They even prey on Dorylus spp. workers when a column of these army ants passes close to the base of their host tree. These arboreal ants, probably helped by allomones, lower the level of aggressiveness of the Dorylus that are preyed upon while surrounded by thousands of nestmates [161-163].

From ground nesting and foraging to relationships with plants From the ground (or underground) to the trees Initially, ants formed a group of soil-dwelling predators or scavengers, as still occurs in the vast majority of ant species belonging to "primitive" subfamilies. A second step was to have a nest on the ground and forage on plants. And the acquisition of an arboreal life in ants probably developed secondarily; strictly arboreal species belong to the most "advanced" subfamilies. Species of foliage-dwelling ants include both "true" canopy inhabitants that nest only in plant organs, and species that commonly nest on the ground, but that are also able to form colonies in hanging soil or are

Ant predation

associated with the epiphytes and hemi-epiphytes that abound in the canopy of a tropical forest [71]. Ants numerically dominate the canopy fauna of tropical rainforests where they are considered to be key predators. Studies of foliage-dwelling arthropods have shown that ants may represent 86% of the arthropod biomass and up to 94% of the arthropod individuals living in the rainforest canopy [71]. A conspicuously low abundance of less mobile holometabolous insects (e.g., Lepidoptera larvae) corresponds to this ant dominance. This is in contrast to temperate regions where ants are mostly absent from trees and holometabolous larvae are frequent [164]. Davidson [165] has suggested that the high abundance of liquid food sources (i.e., extrafloral nectaries and honeydew-producing Hemiptera) on foliage plays an important role in shaping the food-web structure of tropical forests by fueling costly prey-hunting activities by foliage-dwelling ants, especially if the ants are physiologically adapted to a low-nitrogen diet.

Arboreal ant species (nesting and foraging on the trees) Predation on tree foliage and biotic protection of the plants It is likely that ground-nesting, foliage-foraging species constitute the first line of defense in the plants’ biotic protection thanks to their predatory activity. If defoliating insects have frequently developed the means for resisting plants’ chemical defenses, they rarely possess successful counteradaptations against ants, except by escaping through dropping, jumping or flying away [166,167]. Some ant lineages developed tight evolutionary bonds with plants and became arboreal-nesting and foraging. Most arboreal ants have evolved diffuse relationships with plants, the latter inducing different ant species to patrol their foliage by producing energy-rich food rewards such as extrafloral nectaries or food bodies. Furthermore, the relationship can be indirect with ants attending sap-sucking Hemipterans whose role can be similar to that of extrafloral nectaries when the host plant is not affected or only a little [79]. Nonetheless, the relationship between myrmecophytes and ants is necessary to the survival of both partners, with myrmecophytes offering a nesting place (i.e., hollow structures called domatia such as hollow twigs and thorns or leaf pouches) and frequently extrafloral nectar or food bodies to specialized “plant-ants”. In return, plant-ants protect the myrmecophytes from a broad range of herbivores plus competitors and fungal pathogens, and/or provide them with nutrients [168,169].

Xim CerdĂĄ & Alain Dejean

In the exceptional cases when food bodies are protein-rich, hence rendering hunting unnecessary, the ants protect their host-tree foliage from herbivorous insects and mammals only through their territorial aggressiveness. This is, for example, the case of the association between Central American Acacia spp. and ants of the genus Pseudomyrmex, or Piper and Pheidole [170,171]. Other plant-ants are predatory, while intermediary cases have been noted in plant-ants that retrieve only a part of the insects that they capture. They discard the other individuals, but can eventually consume part of their haemolymph [172,173].

Predatory behavior in the trees The canopies of tropical forests and tree crop plantations are occupied by â&#x20AC;&#x153;territorially-dominantâ&#x20AC;? species characterized by (1) extremely populous colonies (several hundred thousand to several million individuals), (2) the ability to build large and/or polydomous nests (carton builders, carpenter ants and weaver ants), and particularly (3) a highly developed intra- as well as inter-specific territoriality that causes their territories to be distributed in a mosaic pattern in the forest canopies [163,174]. These territories are marked with persistent landmarks that can last for over a year and are recognized by other ants that avoid them or adapt their behavior so as to avoid encountering the occupying ants [163,175,176]. Since the availability of prey in tree foliage is unpredictable and most prey are insects able to escape by flying away, jumping or dropping [163], arboreal ants have evolved predatory behaviors adapted to this restricted foraging area by optimizing their ability to capture such insects. The predatory behavior of the weaver ant, Oecophylla longinoda, the first species studied in this context, is well adapted to the fact that prey are likely to escape. Workers hunt diurnally in groups. Prey detected visually from a relatively long distance are seized by an appendage and immobilized by a first worker that then releases a pheromone to attract nestmates. Recruited nestmates, in turn, seize a prey appendage and pull backward, spread-eagling the prey. This behavior, used even for relatively small prey, also permits the ants to capture large insects and even other animals [79,177,178]. Entire prey are retrieved cooperatively, including, in some cases, heavy prey such as small birds [179]. This form of prey capture and retrieval requires the workers to adhere to the substrate by means of very powerful adhesive pads and claws, a characteristic that seems common in arboreal species [180]. Other dominant ants exhibit relatively similar behavior based on the spread-eagling of prey. Detection may occur at a short distance or even by

Ant predation

contact; venom is generally used to subdue the prey prior to cutting it up and transporting it in small pieces. This concerns African Myrmicinae Atopomyrmex mocquerisii, Crematogaster sp., and Tetramorium aculeatum and the Neotropical dolichoderine ant Azteca chartifex (the Dolichoderinae do not have a sting) [173,181-183]. Azteca lanuginosa and A. andreae workers possess an elaborate hunting technique consisting of ambushing side-by-side under the leaf margins of trees with their mandibles wide open. When an insect lands on their leaf and moves toward the edge, all of the members of the ambushing group attack it simultaneously, rushing onto the upper surface of the leaf to spread-eagle it [184,185]. Plant-ants can capture prey in a similar way as territorially-dominant species by spread-eagling them [173,186], but some plant-ants use a more evolved behavior. Tetraponera aethiops (Pseudomyrmicinae) and Azteca bequaerti workers, hidden in their host plant domatia, react to the vibrations transmitted by an alien insect landing on a leaf, making it unnecessary for them to forage for prey [172,173]. Furthermore, plant-ants of the genus Allomerus collectively ambush prey by building galleries pierced with numerous holes serving as traps. When a prey lands on the gallery each worker waiting in a hole near the landing site seizes an appendage and pulls backward, moving deeper into the trap. With its appendages caught in the trapâ&#x20AC;&#x2122;s different holes, the prey is immobilized and recruited workers sting it repeatedly [187].

Conclusions This study has focused on the predatory behavior of ants, mostly on prey capture, taking into account the fact that most ant species are generalist feeders. Only ground-nesting and foraging species are strict predators (certain species are even predators specialized on one prey taxa); among them, some ecitonine species can climb trees to hunt for other ants. All of these ants play a role in the equilibrium of ground- and litter-dwelling detritivorous arthropods and the herbivorous insects living in these strata. Note that, in this general context, human perturbation, through greater agricultural activity, plays an important role in the balance between termites and army ants; for instance, in the Ivory Coast, Macrotermes spp. termitaries were massively destroyed by Dorylus dentifrons after farmers changed their agricultural methods [188]. Many ground-nesting ant species are also arboreal foraging, exploiting extrafloral nectar, sometimes food bodies, and mostly Hemipteran honeydew, while arboreal-dwelling ants rely primarily on these foods. Because they are

Xim Cerdá & Alain Dejean

also predators, certain territorially-dominant species have been used as biological control agents [24,163]. Certain arboreal species do not even hunt or scavenge, so that their nitrogen requirement is provided by their host plant – as is the case for Acacia-associated Pseudomyrmex [169,171] - or through endosymbiont bacteria [189,190].

Acknowledgements We would like to thank Carlo Polidori for encouraging us to write this synthesis and for his patience as an editor, and two anonymous referees for their constructive comments. We are also grateful to Andrea Dejean for proofreading early versions and the final version of the manuscript, to Ana Carvajal for formatting the references and to Susan Rutherford for her editorial assistance.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

Moreau, C.S., Bell, C.D., Vila, R., Archibald, B. and Pierce, N.E. 2006, Science, 312, 101-104. Brady, S.G., Schultz, T.R., Fisher, B.L. and Ward, P.S. 2006, Proc. Natl. Acad. Sc. USA, 103, 18172-18177. Wilson, E.O. and Hölldobler, B. 2005, Proc. Natl. Acad. Sc., USA, 102, 74117414. Federle, W., Rhiele, M., Curtis, A.S.G. and Full R.J. 2002, Int. Compar. Biol., 42, 1100-1106. Grimaldi, D. and Engel, M.S. 2005, Evolution of the Insects, Cambridge University Press, Cambridge, UK. Engel, M.S., Grimaldi D.A., and Krishna K. 2009, Amer Mus Nov., 3650, 1-27. http://osuc.biosci.ohio-state.edu/hymenoptera/tsa.sppcount?the_taxon=Formicidae (5/12/2010). Tillberg, C.V. and Breed, M.D. 2004, Biotropica, 36, 266-272. Davidson, D.W., Cook, S.C., Snelling, R.R., and Chua, T.H. 2003, Science, 300, 969-972. Jones, C.G., Lawton, J.H. and Shachak, M. 1994. Oikos 69:373-386. Lavelle, P., Bignell, D., Lepage, M., Wolters, V., Roger, P., Ineson, P., Heal, O.W. and Dhillion, S. 1997. Eur. J. Soil Biol., 33,159-193. Folgarait, P.J. 1998. Biodivers. & Conserv., 7,1221-1244 Dauber, J. and Wolters, V. 2000. Soil Biol. Biochem., 32,93-99. Petal, J. 1978. In: Production Ecology of Ants and Termites, M.V. Brian (Ed.), IBP 13. Cambridge University Press, Cambridge, 293-325. Petal, J. 1998. Appl Soil Ecol 9:271-275. Dauber, J., Schroeter, D. and Wolters, V. 2001. Eur. J. Soil Biol., 37,259-261.

Ant predation

17. Paris, C.I., Polo, M.G., Garbagnoli, C., Martínez, P., Somma de Ferré, G. and Folgarait, P.J. 2008. Appl. Soil Ecol., 40, 271-282. 18. Dejean, A., Schatz, B., Orivel, J. and Beugnon, G. 1999, Sociobiology, 34, 545554. 19. Schöning, C., Csuzdi, C., Kinuthia, W. and Ogutu, J.O. 2010. Insectes Soc., 57, 73-82. 20. Laakso, J. and Setälä, H. 1997. Oecologia, 111, 565-569. 21. Way, M.J. and Khoo, K.C. 1992, Annu. Rev. Entomol., 37, 479-503. 22. Tschinkel, W.R. 2006, The Fire Ants, Harvard University Press, Cambridge, Mass. 23. Jeanne, R.L. 1979. Ecology, 60, 1211-1224. 24. Blüthgen N. and Feldhaar, H. 2010, Ant Ecology, L. Lach, C. Parr and K. Abbot (Eds.), Oxford University Press, Oxford, 115-136. 25. Juen, A. and Traugott, M. 2005. Oecologia, 142, 344-352. 26. MacArthur, R.H. and Pianka, R.H. 1966. Am. Nat., 100, 603-609. 27. Begon, M., Townsend, C.R. and Harper, J.L. 2006, Ecology. From Individual to Ecosystems. 4th edition. Blackwell, Oxford. 28. Cerdá, X., Angulo, E., Boulay, R. and Lenoir, A. 2009. Behav. Ecol. Sociobiol., 63, 551-562. 29. Detrain, C. and Deneubourg, J.L. 1997. Anim. Behav., 53, 537-547. 30. Lenoir, L. 2002. Eur. J. Soil Biol., 38, 97-102. 31. Cogni, R. and Oliveira, P.S. 2004. J. Insect Behav., 17, 443-458. 32. Orians, G.H. and Pearson, N.E. 1979. Analysis of Ecological Systems, D.J. Horn, R.D. Mitchell and G.R. Stains (Eds.), Ohio State University Press, Columbus, 154–177. 33. Morehead, S.A. and Feener, D.H. Jr. 1998. Oecologia, 114, 548-555. 34. Dejean, A. and Beugnon, G. and Lachaud, J.-P. 1993, J. Ethol. 11, 43-53. 35. Denny, A.J., Wright, J. and Grief, B. 2001. Anim. Behav., 61, 139-146. 36. Weier, J.A. and Feener, D.H.Jr. 1995, Behav. Ecol. Sociobiol., 36, 291-300. 37. Sudd, J.H. and Franks, N.R. 1987. The Behavioural Ecology of Ants, Blackie, Glasgow and London. 38. Daly-Schveitzer, S., Beugnon, G. and Lachaud, J.-P. 2007. Insectes Soc., 319-328. 39. Traniello, J.F.A. and Beshers, S.N. 1991, Behav. Ecol. Sociobiol., 29, 283-289. 40. Dejean, A., Schatz, B. and Kenne, M. 1999. Sociobiology, 34, 407-418. 41. Kenne, M., Mony, R., Tindo, M, Njaleu, L.C.K., Orivel, J. and Dejean, A. 2005, C.R. Biol., 328, 1025-1030. 42. Powell, S. and Franks, N.R. 2005, Proc. R. Soc. London B, 272, 2173-2180. 43. Franks, N.R. 1986, Behav. Ecol. Sociobiol., 18, 425-429. 44. Cerdá, X. 1999. Bol. S.E.A., 26, 676-692. 45. Franks, N.R., Dechaume-Moncharmont, F.-X., Hanmore, E. and Reynolds, J.K., 2009, Phil. Trans. R. Soc. B, 364, 845-852. 46. Wilson, E.O. 1962, Anim. Behav., 10, 134-147. 47. Schatz, B., Lachaud, J.-P. and Beugnon, G., 1997, Behav. Ecol. Sociobiol., 40, 337-349. 48. Gordon, D.M., 1991. Am. Nat., 138, 379-411.

Xim Cerdá & Alain Dejean

49. 50. 51. 52. 53.

Durou, S., Lauga, J. and Dejean, A. 2001, Behaviour, 138, 251-259. Dejean, A. and Benhamou, S., 1993, Behav. Process. 30, 233-244. Hölldobler, B. 1986, Oecologia, 69, 12-15. Richard, F.-J., Dejean, A. and Lachaud, J.-P. 2004. C.R. Biol., 327, 509-517. LaPierre, L., Hespenheide, H. and Dejean, A. 2007. Naturwissenschaften, 94, 997-1001. Perfecto, I. and Vandermeer, J. H. 1993, Insectes Soc., 40, 295-299. Oster, G.F. and Wilson, E.O., 1978. Caste and Ecology in the Social Insects. Princeton University Press, Princeton, NJ. Traniello, J.F.A. 1983. Oecologia, 59, 94-100. Savolainen, R. 1991. Behav. Ecol. Sociobiol., 28, 1-7. Cerdá, X., Retana, J. and Cros, S. 1998. Oikos, 82, 99-110. Yamamoto, A., Ishihara, S. and Ito, F. 2009. J. Insect Behav., 22, 1-11. Cerdá, X., Retana, J. and Manzaneda, A. 1998. Oecologia, 117, 404-412. Kronauer, D.J.C., 2009. Myrmecol. News, 12, 51-65. Wilson, E.O. 1958. Evolution, 12, 24-36. Franks, N.R. and Fletcher, C.R. 1983. Behav. Ecol. Sociobiol., 12, 261-270. Gotwald, W.H., 1995, Army Ants. The Biology of Social Predation, Cornell University Press, Ithaca. Holt, S.J. J. Anim. Ecol., 24, 1-34. Warrington, S. and Whittaker, J.B. 1985, J. Appl. Ecol., 22, 775-785. Stradling, D.J. 1978. Production ecology of ants and termites, M.V. Brian (Ed.), IBP13, Cambridge University Press, Cambridge, 81-106. Pavan, M. 1959, Collana Verde, 4, 1-80. Robertson, H.G. 1988, Ecol. Entomol., 13, 207-214. Whitford, W.G. and Jackson, E. 2007, J. Arid Environ., 70, 549-552. Rico-Gray, V. and Oliveira, P.S. 2007, The Ecology and Evolution of Ant-Plant Interactions. University of Chicago Press, Chicago. Peng, R. and Christian, K. 2010, Ant Ecology, L. Lach, C. Parr and K. Abbot (Eds.), Oxford University Press, Oxford, 123-125. Nowbahari, B. and Thibout, E. 1992, J. Chem. Ecol., 18, 1991-2002. Dyer, L.A. 1995, Ecology, 76, 1483-1496. Jeanne, R.L. 1975. Q. Rev. Biol., 50, 267-287. Kudo, K. and Yamane, S. 1996. J. Ethol., 14, 83-87. Dejean, A., Corbara, B. and Lachaud, J.-P. 1998, Sociobiology, 32, 477-487. Chadab-Crepet, R. and Rettenmeyer, C.W. 1982, The Biology of Social Insects, M.D. Breed, C.O. Michener and H.E. Evans (Eds.), Westview Press, Boulder, 270-274. Hölldobler, B. and Wilson, E.O. 1990, The Ants, Harvard University Press, Cambridge, Mass. Moffett, M.W. 1986, Insectes Soc., 33, 85-99. De la Mora, A., Pérez-Lachaud, G. and Lachaud, J.-P. 2008, Behav. Proc., 78, 64-75. Gronenberg, W., Tautz, J. and Hölldobler, B., 1993, Science, 262, 561-563. Gronenberg, W., 1995, J. Comp. Physiol. A, 176, 391-398.

54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83.

Ant predation

84. 85. 86. 87.

Gronenberg, W., 1995, J. Comp. Physiol. A 176, 399-408. Gronenberg, W., 1996, J. Exp. Biol. 199, 2021-2033. Gronenberg, W. and Ehmer, B., 1996, Zoology, 99, 153-162. Gronenberg, W., Paul, J., Just, S. and Hölldobler, B., 1997, Cell Tiss. Res., 289, 347-361. 88. Gronenberg, W., Brandão, C.R.F., Dietz, B.H. and Just S. 1998, Physiol. Entomol. 23, 227-240. 89. Gronenberg, W., Hölldobler, B. and Alpert, G.D., 1998, J. Insect Physiol., 44, 241-253. 90. Just, S. and Gronenberg, W. 1999, J. Insect Physiol., 45, 231-240. 91. Wilson, E.O. 1953, Ann. Entomol. Soc. Amer., 46, 479-495. 92. Dejean, A. 1986, Insectes Soc., 33, 388-405. 93. Dejean, A., Grimal, A., Malherbe, M.C. and Suzzoni, J.P. 1999, Naturwissenschaften, 86, 133-137. 94. Suzzoni, J.P., Schatz, B. and Dejean, A. 2000, C. R. Acad. Sc., 323, 1003-1008. 95. Dejean, A., Suzzoni, J.-P. and Schatz, B. 2001, Behaviour, 138, 981-996. 96. Dejean, A., Suzzoni, J.-P., Schatz, B. and Orivel, J. 2002, C. R. Biol., 325, 819825. 97. Moffett, M.W., 1986, Biotropica, 18, 361-362. 98. Brandão, C.R.F., Diniz, J.L.M. and Tomotake, E.M. 1991, Insectes Soc., 38, 335344. 99. Jahyny, B., Lacau, S., Delabie, J. H.C. and Fresneau, D. 2008, Sistemática, Biogeografía y Conservación de las Hormigas Cazadoras de Colombia, E. Jiménez, F. Fernández, T.M. Arias and F.H. Lozano-Zambrano (Eds.), Instituto de Investigación de Recursos Biológicos Alexander von Humboldt, Bogotá, 329348. 100. Eisner, T., Eisner, M. and Deyrup, M. 1996, Proc. Natl. Acad. Sc. USA, 93, 10848-10851. 101. Ito, F. 1998, Insectes Soc., 45, 79-83. 102. Traniello, J.F.A. 1989, Annu. Rev. Entomol., 34, 191-210. 103. Peeters, C. 1997, The Evolution of Social Behavior in Insects and Arachnids, J. Choe and B. Crespi (Eds.), Cambridge University Press, Cambridge, 372-391. 104. Dejean, A. and Evraerts, C. 1997, J Insect Behav., 10, 177-191. 105. Maschwitz, U. and Mülenberg, M. 1975, Oecologia, 20, 65-83. 106. Steghaus-Kovac, S. and Maschwitz, U. 1993, Insectes Soc., 40, 337-340. 107. Maschwitz, U. and Schönegge, P. 1983, Oecologia, 57, 175-182. 108. Duncan, F.D. and Crewe, R.M. 1994, Oecologia, 97, 118-123. 109. Maschwitz, U., Steghaus-Kovac, S., Gaube, R., and Hänel, H. 1989, Behav. Ecol. Sociobiol., 24, 305-316. 110. Witte, V. and Maschwitz, U. 2000, Insectes Soc., 47, 76-83. 111. Maschwitz, U., Hahn M. and Schönegge P. 1979, Naturwissenschaften, 66, 213214. 112. Mustahk Ali, T.M., Baroni-Urbani, C. and Billen, J., 1992, Naturwissenschaften, 79, 374-376. 113. Wheeler, W.M. 1936, Proc. Amer. Acad. Arts Sci., 71, 159-243.

Xim Cerdá & Alain Dejean

114. Diniz, J.L.M. and Brandão, C.R.F. 1993, Insectes Soc., 40, 301-311. 115. Bolton, B., Gotwald, W.H. and Leroux J.-M. 1976, Ann. Univ. Abidjan sér. E, 9, 371-381. 116. Dejean, A. and Dejean, A. 1998, Insectes Soc., 45, 343-346. 117. Masuko, K. 1984, Insectes Soc., 31, 429-451. 118. Masuko, K. 2009, J. Nat. His., 43, 825-841. 119. Brown, W.L. jr. and Wilson, E.O. 1959, Quart. Rev. Biol., 34, 278-294. 120. Dejean, A. 1980, Ann. Sc. Nat., Zool., 2, 131-143. 121. Dejean, A. 1980, Ann. Sc. Nat., Zool., 2, 145-150. 122. Dejean, A. 1985, Insectes Soc., 32, 158-172. 123. Dejean, A. 1985, Insectes Soc., 32, 244-256. 124. Masuko, K. 2009, J. Kansas Entomol. Soc., 82, 109-113. 125. Dietz, B.H. and Brandão, C.R.F. 1993, Rev. Brasil. Entomol., 37, 683-692. 126. Dejean, A. 1987, Sociobiology, 13, 119-132. 127. Dejean, A. 1987, Sociobiology, 13, 191-208. 128. Dejean, A. 1987, Sociobiology, 13, 295-306. 129. Dejean, A. 1988, Sociobiology, 14, 325-339. 130. Dejean, A. 1991, Entomol. Exp. Appl., 58, 123-135. 131. Dejean, A., Beugnon, G. and Lachaud, J.-P. 1993, J. Insect Behav., 6, 271-285. 132. Masuko, K. 1993, Bull. Ass. Natl. Sc., Senshu, 24, 35-43. 133. Wild, A. 2005, Notes from Underg., 11, 27. 134. Franks, N.R. and Partridge L.W. 1993, Anim. Behav., 45, 197. 135. Lepage, M.G. and Darlington, J.P.E.C. 1984, J. Nat. Hist., 18, 293-302. 136. Lemaire, M., Lange, C., Lefeuvre, J. and Clément, J.-L. 1986, Actes Coll Insectes Soc., 3, 97-101. 137. Longhurst, C., Baker, R., and Howse, P.E. 1979, Experientia, 35, 870-872. 138. Dejean, A. and Fénéron, R. 1999, Behav. Proc., 47, 125-133. 139. Schatz, B., Orivel, J., Lachaud, J.-P., Beugnon, G. and Dejean, A. 1999, Sociobiology, 34, 569-580. 140. Scholtz, O., Macleod, N. and Eggleton, P. 2008, Zool. J. Linn. Soc., 153, 631650. 141. Dejean, A., Moreau, C.S., Uzac, P., Le Breton, J. and Kenne, M. 2007, C. R. Biol., 330, 701-709. 142. Dejean, A., Moreau, C.S., Kenne, M. and Leponce, M. 2008, C. R. Biol., 331, 631-635. 143. Wilson, E.O. and Brown, W.L. 1984, Insectes Soc., 31, 408-428. 144. Schöning, C. and Moffett, M.W. 2007, Biotropica, 39, 663-667. 145. Longhurst, C., Baker, R., and Howse, P.E. 1980, Insect Biochem., 10, 107-112. 146. Corbara, B. and Dejean A. 2000, Sociobiology, 36, 465-483. 147. Longhurst, C., Johnson, A. and Wood, T.G. 1978, Oecologia, 32, 101-107. 148. Prestwich, G.D. 1979, J. Chem. Ecol., 5, 459-480. 149. Longhurst, C. and Howse, P.E. 1979, Insectes Soc., 26, 204-215. 150. Bayliss, J. and Fielding, A. 2002, Sociobiology, 39, 103-122. 151. Longhurst, C., Johnson, A. and Wood, T.G. 1979, Oecologia, 38, 83-91. 152. Mill, A.E. 1984, J. Nat. Hist., 18, 405-410.

Ant predation

153. Leal, I. and Oliveira P.S. 1995, Behav. Ecol. Sociobiol., 37, 373-383. 154. Agosti, D. 1997, J. NY Entomol. Soc., 105, 16-1691. 155. Perfecto, I. 1992, Psyche; 99, 214-220. 156. LaPolla, J.S, Mueller, U.G., Seid, M. and Cover, S.P. 2002, Insectes Soc., 49, 251-256. 157. Droual, R. 1983, Behav. Ecol. Sociobiol., 12, 203-208. 158. Rettenmeyer, C.W., Chadab-Crepet, R., Naumann, M.G. and Morales, L. 1983, Social Insects in the Tropics, P. Jaisson (Ed.), Presses de l'Université Paris-Nord, Paris, 59-73. 159. Le Breton, J., Dejean, A., Snelling, G. and Orivel, J. 2007, J. Appl. Entomol., 131, 740-743. 160. Powell, S. and Clark, E., 2004, Insectes Soc., 51, 342-351. 161. Gotwald, W.H. 1972, Psyche, 79, 348-356. 162. Dejean, A. 1991, Entomophaga, 36, 29-54. 163. Dejean, A., Corbara, B., Orivel, J. and Leponce M. 2007, Funct. Ecosyst. Communit., 1, 105-120. 164. Floren, A., Biun, A. and Linsenmair, K.E. 2002, Oecologia, 131, 137-144. 165. Davidson, D.W. 1997. Biol. J. Linn. Soc., 61, 153-181. 166. Coley, P.D. and Kursar T.A. 1996, Tropical Forest Plant Ecophysiology, S.S. Mulkey, R.L. Chazdon and A.P. Smith (Eds.), Chapman and Hall, London, 305336. 167. Dejean, A., Delabie, J.H.C., Cerdan, P., Gibernau, M. and Corbara, B. 2006, Biol. J. Linn. Soc., 89, 91-98. 168. Beattie, A. and Hughes, L. 2002, Plant-Animal Interactions: an Evolutionary Approach, C. M. Herrera and O. Pellmyr (Eds.), Blackwell, Oxford, 211-235. 169. Heil, M. and McKey, D. 2003, An. Rev. Ecol., Syst. Evol., 34, 425-453. 170. Fischer, R.C., Richter, A., Wanek, W. and Mayer, V. 2002, Oecologia, 133, 186-192. 171. Heil, M., Baumann, B., Krüger, R. and Linsenmair, K.E. 2004, Chemoecology, 14, 45-52. 172. Dejean, A., Djiéto-Lordon, C. and Orivel, J. 2008, Biol. J. Linn. Soc., 93, 63-69. 173. Dejean, A., Grangier, J., Leroy, C. and Orivel, J. 2009, Naturwissenschaften, 96, 57-63. 174. Blüthgen, N. and Stork, N.E. 2007, Austral Ecol., 32, 93-104. 175. Beugnon, G. and Dejean, A. 1992, Insectes Soc., 39, 341-346. 176. Offenberg, J. 2007, Insectes Soc., 54, 248-250. 177. Dejean, A. 1990, Applied Myrmecology, a World Perspective, R.K. Vander Meer, K. Jaffe and A. Cedeno (Eds.), Westview Press, Boulder, 472-481. 178. Dejean, A. 1990, Physiol. Entomol., 15, 393-403. 179. Wojtusiak, J., Godzinska, E.J. and Dejean, A. 1995, Trop. Zool., 8, 309-318. 180. Federle, W., Rohrseitz, K. and Hölldobler, B. 2000. J. Exp. Biol., 203, 505-512. 181. Djiéto-Lordon, C., Richard, F.J., Owona, C., Orivel, J. and Dejean, A. 2001, Sociobiology, 38, 765-775. 182. Richard, F.-J., Fabre, A. and Dejean, A. 2001, J. Insect Behav., 14, 271-282. 183. Kenne, M., Fénéron, R., Djiéto-Lordon, C., Malherbe, M.-C., Tindo, M., Ngnegueu, P.R. and Dejean A. 2009, Myrmecol. News, 12: 109-115.

Xim Cerdá & Alain Dejean

184. Morais, H.C. 1994, Insectes Soc., 41, 339-342. 185. Dejean, A., Leroy, C., Corbara, B., Roux, O., Céréghino, R., Orivel, J. and Boulay, R. 2010, PloS ONE, 5, e11331. 186. Dejean, A., Solano, P.J., Orivel, J., Belin-Depoux, M., Cerdan, P. and Corbara, B. 2001, Sociobiology, 38, 675-682. 187. Dejean, A., Solano, P.J., Ayroles, J., Corbara, B. and Orivel, J. 2005, Nature, 434, 973. 188. Bodot, P. 1967, Insectes Soc., 14, 229-258. 189. Sauer, C., Dudaczeck, D., Hölldobler, B. and Gross, R. 2002, Appl. Environ. Microbiol., 68, 4187-4193. 190. van Borm, S., Bushinger, A., Boosma, J.J. and Billen, J. 2002, Proc. R. Soc. London B, 269, 2023-2027.

Transworld Research Network 37/661 (2), Fort P.O. Trivandrum-695 023 Kerala, India

Predation in the Hymenoptera: An Evolutionary Perspective, 2011: 79-99 ISBN: 978-81-7895-530-8 Editor: Carlo Polidori

4. A fable on voracious and gourmet ants: Ant-seed interactions from predation to dispersal Francisco M. Azcárate and Pablo Manzano Departamento de Ecología. Universidad Autónoma de Madrid,C/ Darwin 2. Madrid. E-28049, Spain

Abstract. Ants and seeds show a variety of interactions, from generalist seed harvesters to more selective elaiosome-eating ants. Seed harvesting by ants is an interaction typical of non-forested systems from mid and low latitudes. Most aspects of the foraging strategy of harvester ants can be interpreted as a time-saving policy. Ants take advantage of the limited periods of time in which activity outside the nest is feasible to collect as many seeds as possible, using in many cases a trunk trail system to access to the foraging patches. This voracious strategy inflicts severe seed losses to certain plant species. However, harvester ants also behave as accidental seed dispersers (dyszoochory) by dropping or mislaying seeds in foraging areas, foraging trails, chaff piles and granary chambers. The alteration of local microsite conditions caused by harvester ants can be beneficial for the dispersed seeds. In contrast to the less-discerning seed harvesters, some seed-feeding ants behave as true gourmets. Myrmecochory is an interaction that requires selective ants to collect elaiosome-bearing seeds and disperse them after consuming the elaiosome. Ants obtain nutritional benefits from myrmecochory, valuable in quality rather Correspondence/Reprint request: Dr. Francisco M. Azcárate, Departamento de Ecología. Universidad Autónoma de Madrid, C/ Darwin 2. Madrid. E-28049, Spain. E-mail: fm.azcarate@uam.es

Francisco M. AzcĂĄrate & Pablo Manzano

than in quantity. Plants can benefit from an increase in dispersal distance, predator avoidance, reduction of kinship among seeds and arrival to favourable microsites. The interaction can be vital for plants, and is highly dependent on the ant species that collects the diaspore. A third type of interaction involving seeds and ants is diplochory. In this case, seeds experience two phases of dispersal, usually a combination of a coarse and a fine event. Ants can participate in diplochory by collecting coarsely dispersed seeds (e.g. ballistic dispersal, endozoochory, etc.) and transporting them to favourable microsites. Many diplochorous plants offer a reward for frugivores in the form of fruit pulp, as well as a reward for ants in the form of an elaiosome. In other cases, the seeds can be secondarily dispersed by dyszoochory rather than by myrmecochory.

Introduction Imagine an ant and a seed. Most non-biologists will instantly imagine a tiny, hard-working black insect carrying a huge and delicious plant diaspore under the summer sun while a grasshopper plays violin on a verdant leaf in the background. Aesopâ&#x20AC;&#x2122;s fable contains some unquestionable truth; however, the interactions between ants and seeds are much more complex than Aesop portrayed (Fig. 1).

Figure 1. The classic depictions of ants usually portrait harvester ants. Illustration by Milo Winte (public domain). Source: Wikipedia.

Ant-seed interactions

Seed predation by harvester ants is indeed the best known of ant-seed interactions. However, harvester ants not only depredate seeds, but also disperse them when the seeds are lost in trails, accidentally discarded in chaff piles or abandoned in granary chambers. This process is generally known as accidental dispersal, or dyszoochory. Additionally, not all ants that utilize seeds are harvester ants. Myrmecochory refers to seed dispersal, mediated via specific adaptations on the seed, which is generally conducted by ants that exhibit non-harvesting behavior. Those seeds generally bear a lipidic body called an elaiosome. This kind of mutualism is successful if the ant that collected the seed drops the seed in a suitable microsite (often the ant hill itself) while removing the elaiosome and gathering it for the sake of the colonyâ&#x20AC;&#x2122;s nourishment. Finally, there are other ways of interacting with seeds that are even more complex than myrmecochory. For example, some ants collect seeds from vertebrate feces, which, in turn, can be further dispersed or consumed.

â&#x20AC;&#x153;Voraciousâ&#x20AC;? seed harvesting by ants By harvester ants we mean ants, generally belonging to the Myrmicinae family, whose diet consists mainly of seeds. This does not mean that harvester ants collect only seeds. In fact, most species include in their diet other types of food such as invertebrates, vegetative plant parts, flowers, feces and even vertebrate carrion [1, 2, 3, 4, 5, 6, 7]. Seed harvesting by ants has been intensely researched in North America and, to a lesser extent, in Australia, South Africa, South America and the Mediterranean Basin. Even if there is a likely geographic bias in the origin of the research, it is nevertheless clear that harvester ants are found in open systems (deserts, steppes, pastures and open scrublands) where a high proportion of net primary production is assigned to seeds. In North American arid and semiarid environments, ants and heteromyid rodents share their role as the main seed predators [8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18]. Pogonomyrmex and Messor are the main ant genera involved, although some species belong to Aphaenogaster, Ephebomyrmex, Pheidole, etc. [16]. The relatively unstudied South American harvester ants which consist mainly of Pogonomyrmex species, may be more important seed predators than birds or rodents in some arid ecosystems [6, 19, 20, 21]. In the Old World, most studies have focused on Mediterranean countries, where Messor is by far the main genus [3, 5, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33]. African and Asiatic systems have received less attention, with some research being carried out in the South African Karoo [34, 35, 36, 37, 38], the Namib [39, 40] and other non-forested systems [41, 42, 43]. In Australia, ants

Francisco M. Azcárate & Pablo Manzano

are important granivores in savanna, arid, semiarid and mediterranean ecosystems [44, 45, 46, 47, 48, 49, 50]. They are represented by a diverse array of species in the genera Meranoplus, Monomorium, and Pheidole, among others [50].

Prey selection behavior in harvester ants Having identified the harvester ant, let us now turn to the following question: how does an ant choose its prey? Do they always choose those juicy spikes drawn in the representations of Aesop’s ant? The naïve unskilled observer may be sometimes disappointed when seeing a given ant carrying a nutritiousless stick. Simplest models of optimal foraging theory assume that the fitness of a forager increases with the rate of food intake, where food value is measured in calories or weight [51]. Increasing the rate of food intake involves a trade-off between a selective strategy, consisting in searching prey with higher energy content, and a more generalistic one, based in the reduction of the time costs of foraging. The expectation is that the more generalist a seed predator is, the more similar are the proportions of seed species in the environment and in the diet. According to literature, however, harvester ants tend to concentrate their diet on a small group of seed species, which are collected at a proportion differing from their availability in the environment [5, 14, 29, 50, 53]. At a first sight, this would confirm the selective behavior depicted at Aesop’s fable. However, if we analyse the seed traits selected by harvester ants, we find that, to a large extent, selection is explained by structural and morphological traits, such as size, shape, existence of appendages or number of seeds per propagule [3, 5, 13, 24, 38, 53, 54, 55, 56, 57, 58, 59]. Those traits seem to be related with time cost reduction in detecting and handling prey items, rather than with gains in energy content of the prey. This trend does challenge the fable’s depiction and matches with the occassional observations of nutritiousless sticks being carried by ants. Relatively small, spherical seeds, for example, get hidden and buried more easily than large and long propagules [64, 65] and are therefore less detectable. This means that a generalist ant that randomly searches for seeds regardless of its energy value will collect a higher proportion of relatively large and elongated seeds than what is present in the environment. Exactly this phenomenon has been observed in Mediterranean grasslands, where M. barbarus preferentially collects long propagules with dispersal appendages [5]. Moreover, the preference for prey items that are long and bear appendages may be explained in term of handling cost reductions, given that this kind of morphologies

Ant-seed interactions

make it easier for the ant to hold the items by the mandibles and to transport them to the ant nest. In this sense, harvester ants would rather limit time costs, what implies a tendency towards collect anything that can be easily detected and holded. We here describe this behaviour as voracious (a term already used, although not formally or systematically, in e.g. [66]), tracing a deliberate and potentially useful parallelism with roughage eater guilds already described in other animal groups such as ruminants [67] (see below). This voracious interpretation is supported by the facts that i) the balance between energy costs and benefits seems to be favourable for harvester ants even for very small prey [52, 62, 63], ii) environmental conditions and predation risk can greatly constrain ant activity when outside the nest (see below). These advantages contrast with the increase in the amount of low quality food and unpalatable material carried to the nest, which would force the colony to increase its internal activity. This increase, however, would be done under more favourable environmental conditions and while experiencing a lower predation risk. Prey selection in harvester ants is a complex subject anyway, and it is very dependent on the species and situations involved. Non-morphological traits, such as seed viability, nutritional content and chemical composition, have also been associated with prey selection in some species [46, 56, 60, 61]. In addition, prey selection for a given species is far to be constant, and can vary depending on the distance to the nest, [62], prey availability [28, 71], relative abundance of the different prey types [69, 72]; and presence of competing species [51, 73]. Due to the observational nature of many studes, an interesting field of research is to design specific controlled experimental designs to test for the hypothesis of prey selection introduced here. Space exploitation and seed harvesting by ants A prominent feature of harvester ant foraging activity is the complex network of trails which resemble those typically left by much bigger animals on the open landscape, such as those made by elephants on the African savanna. Is there an unknown cast of engineer ants paving the way of foragers? The answer is no. But what lies behind those trails is another important question about the provisioning strategies of harvester ants: how do these ants explore and exploit the space around them? This strategy is critical for minimizing foraging costs, mainly in terms of time invested in travel to and from food sources. The first constraint on time cost reduction is the position of the ant nest relative to food. However, food alone does not dictate nest placement. Other factors, such as soil properties, degree of solar exposure, and local vegetation

Francisco M. Azcárate & Pablo Manzano

structure can influence where nest are situated [22, 45, 74, 75]. Ant predators [76, 77] and interference amongst seed-eating ants [78, 79,] can also affect nest site selection. Consequently, nests are often located outside the most productive habitats, and many harvester ants use well-defined trunk trails to reach their foraging patches, thereby reducing the time needed for food search. The spatial distribution among foraging areas and trunk trails depends on seed availability, “topographic” factors (obstacles, bushes, excessively shadowed or insolated areas, etc.), the situation of the neighboring colonies (of the same or different species), and ant predation risk [30, 80, 81, 82, 83, 84]. Trunk trails of harvester ants have been compared with the branching and rooting systems in plants, due to their hierarchical character and their ability to respond to changes in resource density [84]. Differences among species may therefore be related with the trophic quality of their habitats. Construction of physical, permanent trails is often observed in more predictable habitats with a higher seed density, whereas poorer, more unpredictable habitats have less stable and less complex trails [30]. Even a given species can adapt the characteristics of its trunk trail system to the habitat it occupies, as does Messor barbarus, a species considered to be a permanent trunk trail builder. This species shows two different strategies: a guerrilla type (long and less branched trails) in systems with low seed density, and a phalanx type (trail networks that are much more complex and branched) in systems with abundant food [84]. Many harvester ants change from one foraging patch to another on a daily basis, especially in arid and unpredictable environments [71, 85]. Whether ants will forage on a given space or not depends primarily on the availability of food, although other factors can play a role [86]. Under conditions of thermal stress foragers may reduce the length of the foraging columns and/or look for alternative food sources, even if it means diverting from visible trails [87]. The selection of foraging patches usually involves patrollers marking one of the available trails (or a a sector of the nest mound which leads to the beginning of a foraging trail) with secretions from the ant’s Dufour’s gland [88]. Individual foragers can also show a certain degree of site fidelity, and return to the same site over and over within a day [89].

Temporal patterns of seed harvesting by ants Returning to the ant in Aesop’s fable, we can see that collecting food is not possible during certain times of year – indeed, this is why Aesop’s grasshopper starves. Is this true, even in countries with a relatively mild winter such as Aesop’s native Greece? Or does the laborious ant never stop

Ant-seed interactions

working, neither in cold nor in hot year times? The answer is that ant activity is indeed constrained by the environment. Many studies confirm that there are times when ant activity is not possible outside the nest [1, 22, 32, 44, 45, 46, 90, 91]. Microclimatic conditions in the soil and inside the nest, particularly temperature and humidity, are the main factors limiting harvesting periods [21, 33, 80, 92, 93, 94, 95, 96, 97]. Other abiotic variables such as sunlight, rainfall and wind intensity are also relevant for the activity of some species [44, 98, 99]. Generally speaking, the effect of microclimatic conditions comprises a range of conditions compatible with ant activity. In other words, if you are a small, dark ant you may get too dry or overheated very quickly in the middle of a clear, hot summer day, but not if it is cloudy, or if you simply wait for nightfall to go about your work. The intensity of activity and task allocation follow a more complex regulation, and biotic factors are needed to explain them [86]. In Pogonomyrmex barbatus, the return of successful foragers stimulates inactive foragers to leave the nest, much as humans at Boxing Day sales, and the rate at which successful foragers return to the nest depends on food availability [100]. Food availability is, in fact, the biotic factor most commonly related to harvester ant activity [1, 13, 101, 102]. However, other factors such as food satiation [103, 104], internal colony rhythms [23, 105], interactions with predators [106] and the physiological state or physical condition of workers [63, 95] have also been reported as contributing factors in patterns of ant activity. Temporal patterns of ant activity depend also on competing interactions with other colonies or species [102, 107, 108, 109, 110], with some remarkable cases. For example, in the Sonoran desert, workers of Aphaenogaster cockerelli fill the nest entrances of Pogonomyrmex barbatus colonies with sand, what delays the beginning of the activity period in this species [111]. Predation vs. Dyszoochory: Effects of Harvester ants on vegetation Aesopâ&#x20AC;&#x2122;s ant, in addition to being a tireless worker, is portrayed as neither losing nor dropping a single seed outside the nest. We have already seen that harvester ants do not work continuously under adverse conditions such as those imposed by the relentness summer sun, but what about Aesopâ&#x20AC;&#x2122;s claims of their foraging efficiency? Is seed mortality the only effect harvester ants have on plants? Harvester ants can be quite effective seed predators, and indeed can limit plant recruitment [12, 46, 49, 66, 112, 113]. However, seed predation is not a uniform process: not all plant species experience the same predation risk, and seed removal rates can vary both spatially and temporally. Some hypotheses

Francisco M. Azcárate & Pablo Manzano

about the effect of harvester ants on plant communities take this variability into account, and predict that ants negatively affect certain seed morphologies and life history strategies while indirectly favouring others [53, 58]. For example, it has been suggested that ants in Mediterranean grasslands limit the abundance of larger-seeded species while promoting small-seeded ones [3, 5, 31]. Unfortunately, this and other related hypotheses are difficult to test. The best experimental approach involves the use of ant-exclosures. Ants, however, will search relentlessly for (and usually find) weaknesses to overcome these exclosures, making the exclosures difficult to maintain effectively over time. On the other hand, the short-term use of exclosures has in some instances revealed that ants can influence the relative abundance and diversity of plants, as well as the distribution of seed sizes [8, 10, 11, 15, 18, 31, 78, 114]. However, some authors have alleged that, in the long term, effects of seed harvesting are negligible relative to other factors which shape plant communities, such as climate, grazing and ecological succession [115, 116]. There is increasing evidence pointing to the role of harvester ants as accidental seed dispersers, or agents of dyszoochory, and the ant’s status as mere predators is clearly being called into question [17, 25, 117]. The fate of harvested seeds ranges from the tragic end of being eaten, to the more positive fate of being abandoned at certain microsites where germination and subsequent survival are possible. Indeed, deposited seeds may even experience new benefits in their new location. Dyszoochory is, hence, one of the types of short-distance dispersal that prevents or limits density-dependent seed mortality [118, 119] by reducing intraspecific competition and predation risk, and in some cases by increasing the probability of arrival at favorable microsites. Ants can drop or mislay seeds in at least four places: foraging areas, foraging trails, chaff piles and granary chambers. The first option, where the ants abandon seeds in their own foraging areas [26], possibly by accident, results in short-distance dispersal that would extend the seed shadows. The second option, the abandonment of seeds on foraging trails, can be caused by several factors. M. barbarus presents load transfer between workers while transporting food to the nest, meaning that a prey item is passed on between workers building a chain until it reaches the nest. During this process and along with the closeness to the nest, load mass becomes increasingly correlated with the mass of the worker carrying it [59]. This explains how some prey items are abandoned as the selection process of matching the seed to the worker is “tuned”. For M. andrei, the presence of obstacles in or on the soil may be enough to cause the abandonment of bulky, cumbersome seeds [1]. Seed losses on trails may also happen due to accidents, as e.g. the

Ant-seed interactions

dropping of seeds from collective propagules during transport. In southeastern France, M. barbarus ants distribute small caches close to the trails that function as temporary stores, where some seeds could also be forgotten [26]. The amount of items left on trails likely depends on the species and specific circumstances involved, but as an illustration, it can reach up to 20% of the seeds transported by M. bouvieri [117]. This percentage is similar to other well studied dispersal phenomena, such as the survival of seeds after herbivore gut passage [120]. One suggestive but untested explanation would be the existence of critical dispersal rate thresholds for plant populations to survive. Once the prey reaches the nest, it must be cleaned prior to consumption. Here, the discarded parts are accumulated in characteristic chaff piles around the nest entrances. Due either to mistakes in processing or to further selectivity within the nest, those piles sometimes include a large number of viable seeds [35, 42, 17]. For example, the species richness found among the seeds at the middens of M. barbarus was higher than in the seed banks located in their immediate surroundings [32]. Finally, the seeds stored inside the nest’s chambers can retain their ability to germinate, especially if the storage chamber is near the soil’s surface [121]. Those seeds can be reincorporated into the surface seed bank through, for example, the digging activity of predators looking for alates during the ant’s reproductive phase [35]. One of the most interesting aspects of dyszoochory is the alteration of local microsite conditions often caused by the harvester ants themselves, which can be crucial to the performance of dispersed seeds. Workers can, for example, remove preexistent vegetation on trunk trails [30, 82, 83], thereby reducing future competition for the dispersed seeds at germination. Moreover, increased soil fertility caused by ant –mediated alteration of a soil’s physical, chemical and biological properties has been shown in a number of studies [12, 17, 32, 36, 37, 122, 123, 124, 125]. In summary, the overall influence of harvester ants on an ecosystem is a combination of the ant’s voracious efficiency (i.e., seed predation) and its incompetence (i.e., seed dispersal by dyszoochory). Those species of seed that are most intensively harvested are also most successfully dispersed [32], suggesting an interesting explanation of the stability of the interaction in natural or semi natural systems such as the pasturelands of the Mediterranean Basin. Ants may regulate the species composition of the pasture by causing a higher mortality among the most abundant or detectable (because of their size or shape) seeds. However, some of the collected species may be successful in colonizing former trails and chaff piles, to the point that harvester ant abundance and proportion of dyszoochorous species can be correlated

Francisco M. AzcĂĄrate & Pablo Manzano

through different successional stages [25]. Conversely, in less stable and more recent environments such as artificial pastures or agricultural systems, examples are found where harvester ants constituting a real problem due to the severe losses caused by a mere antagonist interaction with vegetation [48]. Systems suffering from invasions of exotic ants also show dysregulations in the predation-dyszoochory balance [126]. A suggestive area of research is how invasions by exotic grasses can boost local harvester ants, eventually driving plants severily impacted by ant predation to extinction [66].

â&#x20AC;&#x153;Gourmetâ&#x20AC;? ant-seed interactions Myrmecochory As we have seen, harvesters can be voracious all-you-can-eat guests that can ruin any banquet. Ants show, however, a wide variety of feeding strategies, and some ant guilds provide for the careful selection of prey for maximum energy content. Some ants eat insects, but among the seed-preying ants there is a certain degree of specialization towards high-quality food. This points out an interesting parallel with ruminants, which can be even more important in configuring plant communities [116]. They can be divided into generalist, low-quality food feeders or grazers [67], equivalent to voracious ants, and more selective feeders or browsers, which would correspond to gourmet ants. Myrmecochory is a good example of the latter. It consists of the dispersal of elaiosome-bearing seeds. An elaiosome is a structure rich in substances that are especially nutritious for ants, such as some fatty acids and nitrogen-rich aminoacids [127], present in plants from diverse phylogenetic lineages and originating from distinct parts of the embryo. To be effective, this manner of seed dispersal requires the ants to behave as true gourmets. They have to consume the elaiosome and reject the seed (which is usually bigger) before or after they taste the elaiosome (Fig. 2). Then, they must abandon it at a microsite appropriate for germination and suffering from a low predation risk [128]. Although myrmecochory may be underappreciated due to biogeographic biases, it appears principally in Australia [128, 129, 130, 131, 132] and South Africa (mainly in the Cape Region, 133, 134, 135], and, to a lesser extent, among herbs of Northern Hemisphere temperate forests [127, 136, 137, 138, 139, 140] and Mediterranean shrubs [141]. The infertility of Australian and South African soils has been seen as a factor favoring myrmecochory [142]. The "Nutrient-Poverty/Intense-Fire Theory" considers myrmecochory, together with other peculiarities of Australian flora, an evolutionary consequence of

Ant-seed interactions

Figure 2. â&#x20AC;&#x153;- Just the tipâ&#x20AC;?. In myrmecochory, ants reject and disperse the bulk of the seed, eating only the most nutritious part â&#x20AC;&#x201C; the elaiosome. Illustration by Clara Herreros Murueta-Goyena.

adaptations to nutrient poverty, compounded by intense fire that tends to occur as a result of nutrient poverty [132]. The repeated independent emergence of myrmecochory among multiple plant lineages means that it is an evolutionary stable strategy, and a very successful one. Its origin, however, is not clear. One possibility is that it has evolved from seed predation through dyszoochory [143]. Non-granivorous ants are, however, more often involved in myrmecochory, a fact that complicates this interpretation, and makes an independent evolution of both processes more probable. Myrmecochory requires changes in the seed rather than in the ant. Given that the fatty acids contained in the elaiosome are similar to the ones found in insects, non-granivorous ants feeding normally on insects would have been attracted to those seeds [131]. Regular granivorous ants would have been precluded to feed on myrmecochorous seeds because of their hard coats [144, 145] or their toxins [129]. Alternatively, myrmecochory could have appeared in seeds to escape from predation by rodents [127, 146, 147]. The underlying mechanism would be that ants quickly take the whole diaspore into the nest [148] so that they avoid overheating and/or predation (as discussed in the previous section). Once ants are inside their nest they can remove the elaiosome and discard the seed. Adaptations could have evolved in plants both for selecting seed dispersers and for manipulating the behavior of those dispersers, in order to

Francisco M. Azcรกrate & Pablo Manzano

increase the probability of a seed to be collected by them instead of by a seed predator [147]. Elaiosome size obviously plays a role in attracting ants [148], and likely represents a tradeoff between attractiveness and the energy demands of production. Potentially more important, however, is the detrimental effect an oversized seed can have on the ability of ants to remove and disperse it [127]. The custom of carrying a whole seed and eating just a small piece of it seems counterintuitive, so there must be some substantial benefits for ants. They do indeed obtain nutritional benefits from myrmecochory, but they are valuable in quality rather than in quantity. Elaiosomes contain essential nutrients and they are used to feed larvae. In Myrmica rubra, elaiosomes of Corydalis cava were found to be a far more attractive diet for the larvae than Bhatkar diet, an optimized artificial food source [139]. In Myrmica ruginoides, colonies fed with the elaiosomes of different Ulex species, larvae weighed 48 % more than non-elaiosome fed colonies [150]. The food supplement provided by Ulex elaiosomes was trivial in energy terms, given an ample diet, suggesting that these effects might be due to the presence of essential nutrients. Chemical analysis of Ulex elaiosomes showed the presence of four essential fatty acids and four essential sterols for ants. Complementarily, larvae accumulating more radio-label from elaiosomes tended to develop into gynes (virgin queens), whereas other female larvae developed into workers [151]. Elaiosomes can also differ in their quality, which can imply differences in seed dispersal success. In Mediterranean forests, the same guild of seed-dispersing ants was more prone to disperse Helleborus foetidus if the nutrient content of the elaiosomes was more attractive to them [140]. An interesting research question in this sense is therefore which chemical cues contribute to the identification and classification of elaiosomes by ants. Apart from the benefits for ants, myrmecochory must also be an evolutionary stable strategy for their counterparts to become established. Plants also get benefits associated with this kind of dispersal, which can be classified into four types. Firstly, myrmecochory can increase dispersal distances of seeds [147]. This form of dispersal is generally manifest at small scales [136, 152, 153], although relatively large distances have been recorded (e.g., 180 m for Iridomyrmex viridiaeneus and Acacia ligulata in New South Wales [154]). In this context, dispersed distance may be related to the worker size [155]. Often, long-distance dispersal involving ants operates in concert with other seed dispersal agents such as wind, vertebrates, etc. [138, 156, 157]. A possible benefit of long-distance seed dispersal is reduced density dependent seed mortality; however, this is not always the case [158]. A second possible benefit of seed dispersal is predator avoidance, already

Ant-seed interactions

observed in interactions at larger scales and with larger organisms [118]. Myrmecochory may reduce the overall levels of seed predation by breaking up aggregations of seeds and/or depositing dispersed seeds in safe sites, thereby allowing more seeds to reach the persistent soil seed bank [147, 158, 159]. Thirdly, through dispersal ants would likely reduce the relatedness among seeds [160], thereby reducing the levels of inbreeding that are detrimental to plant populations [161] if the optimal outcrossing distance is large enough [162, 163]. Fourthly, seeds could be transported to a safe and favorable microsite (directed dispersal [67], but see below). The interaction between plants and their formicid seed dispersers is by its very nature asymmetric, because dispersal to suitable locations is vital for plants [164] but not for the ants. Therefore, the evolutionary success of myrmecochory depends primarily on the benefits received by plants. Specifically, predator avoidance and arrival to favorable microsites can be seen as essential for seed survival; in fact, both factors are deeply interrelated because a favorable microsite can be partially defined as one with low predation risk. But whether or not these conditions are fulfilled depends highly on the ant species that collects the diaspore [137, 141, 147]. The elaiosome often attracts ants that will not conveniently disperse the seed [159, 165] as well as cheaters that will eat the elaiosome and leave the seed where they found it; such activity may nonetheless be limited by the ant species involved [149]. In the worst-case scenario for the plant, the elaiosome will attract harvesting ants that will also depredate the seed [128]. In sum, the success or failure of myrmecochory depends highly on the ant community and hence on the habitat type [147,166]. This may be the reason why myrmecochory is virtually absent in many ecosystems where harvester ants dominate, particularly open habitats in the Holarctic, and where the absence of rodents may also play a role [149]. Voracious, seed harvesting ants may leave no place for the highly discerning gourmet ants. Finally, myrmecochory has been observed to be one of the mutualisms disrupted by invasive species, especially Linepithema humile [167, 168, 169, 170] and Solenopsis invicta [171]. Diplochory and secondary predation To thoroughly review seed-ant interactions, we must look beyond those ants with a penchant for delicious food procured, albeit in small proportions, from elaiosomes. For example, some ants that stand out not because of what they eat, but because of the origin of their food. Many readers of Aesopâ&#x20AC;&#x2122;s fables would be very surprised â&#x20AC;&#x201C; and the very young driven to insatiable laughter â&#x20AC;&#x201C; if they realize where some of ants look for seeds: inside feces!

Francisco M. AzcĂĄrate & Pablo Manzano

However, before looking at this strange, and arguably distasteful, foraging behavior in detail, we should examine in broader terms the ecology of seed dispersal and its implications for plants. Earlier we discussed in general terms the benefits of dispersal. However, these benefits depend largely on the distance achieved by a given dispersal event, which is why we can distinguish between two broad types of dispersal â&#x20AC;&#x201C; coarse and fine dispersal. Long distance dispersal (LDD) could be described as the set of dispersal events that cross long distances but that do not usually deposit seeds in favorable microsites (as e.g. after wind dispersal or epizoochory) or that do not break the seed aggregation that causes densitydependent seed mortality (as e.g. after endozoochory). Such long distance dispersal can be alternatively described as coarse dispersal [172]. By contrast, fine dispersal events involve dispersal over much shorter distances, usually aided by small animals as vectors. Fine dispersal culminates in a much greater chance that the seeds will end up in favorable microsites. Thus, the quality of dispersal, from the plantâ&#x20AC;&#x2122;s perspective, must be viewed not only in terms of the number of seeds dispersed, but also the distances dispersed and the quality of the microsites in which seeds are deposited [173]. The fact that a seed needs the long-distance component as well as the microsite-directed component of dispersal to increase its establishment chances may be the reason why many seeds experience two phases of dispersal, a process known as diplochory. The first phase often consists of coarse dispersal followed by a second phase of fine dispersal [156, 157]. While reviewing dyszoochory and myrmecochory in ants we have reviewed the advantages of dispersal on a short scale, but the advantages of LDD are also striking and constitute a hot topic of research in the last decade [136, 174, 175]. Although affecting only a small proportion of the seeds produced by a given plant, LDD is very important in order to avoid inbreeding or extinction by changes in the near environment in a given plant population. One of the most closely studied mechanisms of two-phase dispersal is the combination of ballistic dispersal by the mother plant and a further phase by myrmecochory. This is a combination found not only in the most typical regions for myrmecochory, i.e. Australia and South Africa, but also in North America, Europe and Japan. Remarkably, seeds dispersed ballistically have usually smaller elaiosomes than seeds that are not dispersed ballistically [156]. This combination of dispersal modes does not achieve great dispersal distances, almost never exceeding 10 m. However, ballistic dispersal helps to put the diaspore outside the influence of the mother plant, hence protecting it against the attack of pre-dispersal seed predators such as insect larvae, rodents, birds, beetles, or harvester ants [113, 157]. Although ant-mediated seed dispersal would not increase the dispersal distance by much (indeed,

Ant-seed interactions

ants could even reduce dispersal distance depending on the direction they take after collecting a seed) their benefit would rely more on bringing them to favorable microsites where seeds are protected from desiccation and predators. The relatively frequent combination of a double ballisticmyrmecochory dispersal syndrome has proven to be an evolutionary stable strategy. Secondary dispersal can also be observed in combination with endozoochory, i.e. dispersal by animals of seeds contained in their food. This includes seeds contained in feces as well as spitted or otherwise discarded prior to gut passage. Interestingly, many plants adapted for this double-phase dispersal mode offer a reward for frugivores in the form of fruit pulp, as well as a reward for ants in the form of an elaiosome [157]. In other cases, the plant does not offer any reward for the ant and the seeds, which are secondarily dispersed by dyszoochory rather than by myrmecochory [156, 157]. There are also cases where ants behave as pure dispersers even if the seeds bear no elaiosome, because they feed on the pulp of fruits and leave the seed intact [60, 176]. The combination of endozoochory and myrmecochory offers benefits but also some risks for the seed. On the one hand, manure can offer nutrients for the seeds [177] and discourage seed predators [178], but on the other hand it can make seeds more detectable and facilitate their predation [179] and could add itself with the adhered pulp rests that facilitate the fungal attack observed in free seeds [180]. It is no wonder that it is difficult to observe a clear trend when reviewing different studies [158], and hence to deduce which benefits dispersal by ants can bring. There is, however, a clear effect of endozoochory in seed aggregation that, while in most cases is reduced if we compare it to the mother plant, still allows for the existence of density-dependent mortality that ant dispersal will further reduce. There is a remarkable absence of studies regarding diplochory processes where ants and large herbivores are involved, given the importance of the latter as dispersers [181]. Examples of elaiosome-bearing seeds that can be dispersed by ruminants can be found in the literature [120] but there may be a bias towards studies in the tropics because of the easier experimental manipulation, with bigger seeds [182]. This is indeed the case with ants that depredate seeds embedded in feces. There are some reports of this behavior in the tropics, with predation on small seeded species as a general pattern (see 158 for a review). However, this apparent selectivity may actually be a consequence of rodents preying on larger seeds, leaving only the smaller seeds for ants. In the handful of existing studies, predation on seeds in feces is performed by strongly social ants, confirming the tendency of those guilds for clumped resources [184]. The overall process consistently shows very high predation rates, ranging from

Francisco M. Azcárate & Pablo Manzano

50% to 90% of the embedded seeds, thus significantly altering seed shadows caused by endozoochory. This also confirms the effect that aggregation in feces has, increasing seed mortality compared to more diffuse patterns. Moreover, the process seems not to be restricted to the tropics. In Mediterranean grasslands, the main harvester ant of the system, M. barbarus, strongly predates on feces of the main herbivore in the system, i.e. domestic sheep [158]. Interestingly, in this case the ants do not change their usual diet by feeding on seeds embedded in dung, but they drastically use the feces as a resource because it is clumped. This only underscores the importance this kind of post-dispersal predation processes must have worldwide. Notably, the main actor in this last example is a typical Mediterranean harvester ant, which brings us back to our point of departure: Aesop’s ant. We would have never guessed where it sometimes gets its delicious food from.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

13. 14. 15. 16. 17. 18. 19. 20.

Hobbs, R. J. 1985, Oecologia 67, 519. Pol, P., and de Casenave, J.L. 2004, J. Ins. Beh., 17, 647. Detrain, C., and Pasteels, J. M. 2000, Sociobiology, 35, 35. Bestelmeyer, B.T., and Wiens, J.A. 2003, J. Arid. Environ. 53, 373. Azcárate, F.M., Arqueros, L., Sánchez, A., and Peco, B. 2005, Funct. Ecol., 19, 273. Pirk, G.I., De Casenave, J.L., Pol, R.G., Marone, L, and Milesi, F.A. 2009, Austral Ecol., 34, 908. López, F., Serrano, J.M., and Acosta, F.J. 1992, Deut. Entomol. Z. 39, 135. Brown, J. H.; Davidson, D. W., and Reichman, O. J. 1979, Am. Zool., 19, 1129. Brown, J. H.; Reichman, O. J., and Davidson, D. W. 1979, Ann. Rev. Ecol. Syst., 10, 201. Davidson, D. W.; Inouye, R. S., and Brown, J. H. 1984, Ecology 65, 1780. Davidson, D. W.; Samson, D. A., and Inouye, R. S. 1985, Ecology 66, 486. Beattie, A. J. 1989, Grassland Structure and Function: California Annual Grassland, L.F. Huenneke and H. Mooney (Eds.), Kluwer Academic Publishers, Dordrecht, 105 Crist, T. O., and MacMahon, J. A. 1992, Ecology 73, 1678. Crist, T. O., and Wiens, J. A. 1994, Oikos 69, 37. Brown, M. J. F., and Human, K. G. 1997, Oecologia 112, 237. Johnson, R. A. 2000, Sociobiology 36, 89. MacMahon, J. A., Mull, J. F., and Crist, T. O. 2000, Annu. Rev. Ecol. Syst. 31, 265. Anderson, C. J., and Macmahon, J. A. 2001, J. Arid. Environ., 49, 343. De Casenave, J. L.; Cueto, V. R., and Marone, L. 1998, Global Ecol. Biogeogr., 7, 197. Marone, L.; Rossi, B. E., and De Casenave, J. L. 1998, Funct. Ecol., 12, 640.

Ant-seed interactions

21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58.

Cerdá, X., and Retana, J. 1994, J. Appl. Entomol., 117, 268. Díaz, M. 1991, Insect. Soc., 38, 351. Díaz, M. 1992, Oecologia, 91, 561. Baroni Urbani, C. 1992, Ethol. Ecol. Evol., nº sp. 2, 73. Wolff, A., and Debussche, M. 1999, Oikos, 84, 443. Detrain, C., and Tasse, O. 2000, Naturwissenschaften, 87, 373. Lopez, F.; Acosta, F. J., and Serrano, J. M. 2000, Ecol. Res., 15, 449. Wilby, A., and Shachak, M. 2000, Oecologia, 125, 495. Wilby, A., Shachak, M., and Boeken, B. 2001, Oikos, 92, 436. Azcárate, F.M., and Peco, B. 2003, Insect. Soc., 50, 120. Azcárate, F.M., and Peco, B. 2006, J. Veg. Sci., 17, 353. Azcárate, F.M., and Peco, B. 2007, J. Veg. Sci., 18, 103. Azcárate, F.M., Kovacs, E., and Peco, B. 2007, J. Insect Behav., 20, 315. Dean, W. R. J., and Turner, J. S. 1991, J. Arid. Environ., 21, 59. Dean, W. R. J., and Yeaton, R. I. 1992, Afr. J. Ecol., 30, 335. Dean, W. R. J., and Yeaton, R. I. 1993, J. Arid. Environ. 25, 249. Dean, W. R. J. and Yeaton, R. I. 1993, Vegetatio 106, 21. Milton, S. J. and Dean, W. R. J. 1993, J. Arid. Environ., 24, 63. Marsh, A. C. 1985, S. Afr. J. Zool., 20, 197. Marsh, A. C. 1987, S. Afr. J. Zool., 22, 130. Capon, M. H., and O'Connor, T. G. 1990, S. Afr. J. Bot., 56, 11. Vorster, H., Hewitt, P. H., and van der Westhuizen, M. C. 1994, Afr. Entomol., 2, 175. Linzey, A. V., and Washok, K. A. 2000, Afr. Zool., 35, 295. Briese, D. T., and Macauley, B. J. 1980, Aust. J. Ecol., 5, 121. Briese, D. T. 1982, Ant-plant interactions in Australia, R.C. Buckley (Ed.), Junk, The Hague, 11. Andrew, M. H. 1986, Biotropica, 18, 344. Andersen, A. N. 1991, Ant-plant interactions, C.R. Huxley and F.D. Cutler (Eds.), Oxford University Press, UK, 493. Andersen, A.N. 1991, Applied Myrmecology: A World Perspective, R.K. Vander Meer, V. Jaffe and A. Cedeno (Eds.), Westview Press, Inc., Boulder, CO, 35. Ireland, C., and Andrew, M. H. 1995, Aust. J. Ecol., 20, 565. Andersen, A. N., Azcárate, F. M., and Cowie, I. D. 2000, J. Anim. Ecol., 69, 975. Pyke, G.H. 1984, Ann. Rev. Ecol. Syst., 15, 523. Fewell, J. H. 1988, Interindividual behavioural variability in social insects, R.L. Jeanne (Ed.), Westview Press, Boulder CO, 257. Willott, S.J., Compton, S.G., and Incoll, L. D. 2000, Oecologia, 125, 35. Pulliam, H. R., and Brand, M. R. 1975, Ecology, 56, 1158. Davison, E. A. 1982, Ant-plant interactions in Australia, R.C. Buckley (Ed.), Junk, The Hague, 1. Kelrick, M. I., Mac Mahon, J. A., Parmenter, R. R., and Sisson, D. V. 1986, Oecologia, 68, 327. Baroni-Urbani, C. and Nielsen, M. G. 1990, Physiol. Entomol., 15, 449. Schoning, C., Espadaler, X., Hensen, I., and Roces, F. 2004, J. Arid. Environ., 56, 43.

Francisco M. Azcárate & Pablo Manzano

59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74.

Reyes-Lopez, J. L. and Fernandez-Haeger, J. 2001, Insect. Soc., 48, 118. Pizo, M.A., Oliveira, P.S. 2000, Biotropica, 32, 851. Whitford, W. G., and Steinberger, Y. 2009, 53, 549. Reyes-López, J. L. 1987, Ecology 68, 1630. Nielsen, M. G., and Baroni-Urbani, C. 1990, Physiol. Entomol., 15, 441. Peart, M. H. 1984, J. Ecol., 72, 437. Thompson, K., Band, S. R., and Hodgson, J. G. 1993, Funct. Ecol., 7, 236. White, J.P., and Robertson, I.C. 2009, Ecoscience, 16, 508. Hofmann, R. R. 1989, Oecologia 78, 443. Fewell, J.H., Harrison, J.F. 1991, Behav. Ecol. Sociobiol., 28, 377. Reyes-López, J. L., and Fernández-Haeger, J. 2002, Sociobiology, 39, 1. Reyes-Lopez, J. L., and Fernandez-Haeger, J. 2002, Sociobiology, 39, 123. Rissing, S. W., and Wheeler, J. 1976, Pan-Pac Entomol., 52, 63. Risch, S. J., and Carroll, C. R. 1986, Ecology 67, 1319. Kaspari, M. 1996, Oecologia, 105, 397. Sudd, J. H., and Franks, N. R. 1987, The Behavioural Ecology of Ants, Blackie, Glasgow. Gordon, D. M. 1991, Am. Nat., 138, 379. Munger, J. C. 1984, Ecology, 65, 1077. Gotelli, N. J. 1996, Ecology, 77, 630. Brown, J. H., and Davidson, D. W. 1977, Science, 196, 880. Gordon, D.M, and Kulig, A.W. 1996, Ecology, 77, 2393. Crist, T. O., and MacMahon, J. A. 1991, Environ. Entomol., 20, 37. Gordon 1992, Behav. Ecol. Sociobiol., 31, 417. López, F., Acosta, F. J., and Serrano, J. M. 1993, Acta Oecol., 14, 405. López, F., Acosta, F. J., and Serrano, J. M. 1993, Oecologia, 93, 109. López, F., Acosta, F. J., and Serrano, J. M. 1994, J. Anim. Ecol., 63, 127. Went, F. W., Wheeler, J., Wheeler, C.G. 1972, Bioscience, 22, 82. Gordon, D. M. 1999, Ants at Work, Norton & Company, New York. Doblas-Miranda, E., and Reyes-López, J. 2008. Environ. Entomol., 37, 857. Greene, M.J., and Gordon, D.H. 2007, Am. Nat., 170, 943. Beverly, B. D., McLendon, H., Nacu, S., Holmes, S., and Gordon, D. M. 2010, Behav. Ecol., 20, 633. Cerdá, X., Bosch, J., Alsina, A., and Retana, J. 1988. Ann. Soc. Ent. Fr., 24, 69. Cros, S., Cerdá, X., and Retana, J. 1997, Ecoscience, 4, 269. Morrison, L.W., Kawazoe, E.A., Guerra, R. and Gilbert, L.E., 2000, Ecol. Entomol., 25, 433. Vogt, J.T., Smith, W.A., Grantham, R.A., and Wright, R.E. 2003, Environ. Entomol., 323, 447. Feener, D. H., and Lighton, J. R. B. 1991, Ecol. Entomol., 16, 183-191. Kaspari, M. 1993, Oecologia, 96, 500. Lighton, J.R.B., Quinlan, M.C., and Feener, D.H. 1994, Physiol. Entomol., 19, 325. Kaspari, M., and Weiser, M.D. 2000, Biotropica, 32, 703. Cerdá, X, and Retana, J. 1994, J. Appl. Ent., 117, 268.

75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98.

Ant-seed interactions

99. Wirth, R., and Leal, I.R. 2001, Ecoscience, 8, 450. 100. Gordon, D.M., Holmes, S., Nacu, S. 2008, Behav. Ecol., 19, 217. 101. Bernstein, R. A. 1979, J. Anim. Ecol., 48, 921. 102. Sanders, N.J., and Gordon, D.M. 2002, Insect. Soc., 49, 371. 103. Whitford, W. G., and Ettershank, G. 1975, Environ. Entomol., 4, 689. 104. Whitford, W. G. 1978, Ecology, 59, 185. 105. Houston, A., Schmid-Heupel, P., and Kacelnik, A. 1988. Am. Nat., 131, 107. 106. MacKay, W. P. 1982, Oecologia, 53, 406. 107. Brown, M. J. F., and Gordon, D. M. 2000, Behav. Ecol. Sociobiol., 47, 195. 108. Mehlhop, P. and Scott, N. J. jr. 1983, Ecol. Entomol., 8, 69. 109. Cerdรก, X., Retana, J. and Manzaneda, A. 1998, Oecologia, 117, 404. 110. Sanders, N. J. and Gordon, D. M. 2000, Oecologia, 125, 436. 111. Gordon, D. M. 1988, Oecologia, 75, 114. 112. Louda, S. M. 1989, Ecology of Soil Seed Banks, M.A. Leck, V.T. Parker and R.L. Simpson (Eds.), Academic Pres, Inc, New York, 25. 113. Albert, M. J., Escudero, A., and Iriondo, J. M. 2005, Acta Oecol. 28, 213. 114. Inouye, R. S., Byers, G. S., and Brown, J. H. 1980, Ecology, 61, 1344. 115. Wilby, A., and Shachak, M. 2004, Oikos, 106, 209. 116. Azcรกrate, F.M., Manzano, P, and Peco, B. 2010, Seed Sci. Res., 20, 179. 117. Retana, J., Picรณ, F. X., Rodrigo, A. 2004, Oikos, 105, 377. 118. Janzen, D. H. 1970, Am. Nat., 104, 501. 119. Vander Wall, S. B., Kuhn, K. M., and Beck, M. J. 2005, Ecology, 86, 801. 120. Manzano, P., Malo, J. E., and Peco, B. 2005, Seed Sci. Res., 15, 21. 121. Milton, S. J., and Dean, W. R. J. 1993, J. Arid. Environ., 24, 63. 122. Wagner, D., Brown, M. J. F., and Gordon, D. M. 1997, Oecologia, 112, 232. 123. Cammeraat, L. H., Willott, S. J., Compton, S. G. and Incoll, L. D. 2002, Geoderma, 105, 1. 124. Wagner, D., and Jones, J. B. 2006, Soil Biol. Biochem., 38, 2593. 125. Ginzburg, O., Whitford, W. G., and Steinberger, Y. 2008, Biol. Fert. Soils, 45, 165. 126. Oliveras, J., Bas, J. M., and Gรณmez, C. 2007, Vie et Milieu, 57, 79. 127. Mayer, V., ร lzant, S., and Fischer, R. C. 2005, Seed fate: predation, dispersal and seedling establishment, P.M. Forget, J.E. Lambert, P.E. Hulme, and S.B. Vander Wall (Eds.), CABI Publishing, Cambridge, 173. 128. Hughes, L., and Westoby, M. 1992, Ecology 73, 1285. 129. Majer, J. D., and Lamont, B. B. 1985, Aust. J. Bot., 33, 611. 130. Hughes, L., and Westoby, M. 1992, Ecology, 73, 1300. 131. Hughes, L., Westoby, M., and Jurado, E. 1994, Funct. Ecol., 8, 358. 132. Orians, G.H., Milewski, A.V. 2007, Biol. Rev., 82, 393. 133. Bond, W. J., and Slingsby, P. 1983, South Afr. J. Sci., 79, 231. 134. Bond, W. J., and Stock, W. D. 1989, Oecologia, 81, 412. 135. Bond, W. J., Yeaton, R. and Stock, W. D. 1991, Ant-Plant Interactions, C.R. Huxley, and D.F. Cutter (Eds.), Oxford University Press, 448. 136. Cain, M. L., Damman, H., and Muir, A. 1998, Ecol. Monogr., 68, 325. 137. Gorb, S. N., and Gorb, E. V. 1999, Oikos, 84, 110.

Francisco M. Azcárate & Pablo Manzano

138. Heinken, T. 2004, 170, 55. 139. Fischer, R. C., Olzant, S. M., Wanek, W., and Mayer, V. 2005, Insect. Soc., 52, 55. 140. Boulay, R., Coll-Toledano, J., and Cerdá, X. 2006, Chemoecology, 16, 1. 141. Espadaler, X., and Gómez, C. 1996, Ecography, 19, 7. 142. Westoby, M., French, K., Hughes, L., and Rodgerson, L. 1991, Aust. J. Ecol., 16, 445. 143. Levey, D. J., and Byrne, M. M. 1993, Ecology, 74, 1802. 144. Rodgerson, L. 1998, Ecology, 79, 1669. 145. Oliveras, J., Gómez, C., Bas, J. M., and Espadaler, X. 2008, Naturwissenschaften, 95, 501. 146. Midgley J., Anderson B., Bok A., and Fleming T. 2002, Evol. Ecol. Res., 4, 623. 147. Giladi, I. 2006, Oikos, 112, 481 148. Peters, M., Oberrath, R., and Bohning-Gaese, K. 2003, Flora, 198, 413. 149. Boulay, R., Carro, F., Soriguer, R.C. and Cerdá, X. 2009, Oecologia, 161, 529. 150. Gammans, N., Bullock, J. M., and Schonrogge, K. 2005, 146, 43. 151. Bono, J. M. and Heithaus, E. R. 2002, Insect. Soc., 49, 320. 152. Gómez, C. and Espadaler, X. 1998, Sociobiology, 32, 441. 153. Brunet, J. and Von Oheimb, G. 1998, J. Ecol., 86, 429 154. Whitney, K. D. 2002, Austral Ecol., 27, 589. 155. Ness, J. H., and Bronstein, I. L. 2004, Biol. Invasions, 6, 445. 156. Vander Wall, S. B. and Longland, W. S. 2004, Trends Ecol. Evol., 19, 155. 157. Vander Wall, S. B., and Longland, W. S. 2005, Seed Fate. Predation, Dispersal and Seedling Establishment, P.M Forget, J.E. Lambert, P.E. Hulme, and S. Vander Wall (Eds.), CABI Publishing, 297. 158. Manzano, P., Azcárate, F. M., Peco, B., Malo, J. E. 2010, Oikos, 119, 1537. 159. Auld, T. D., and Denham, A. J. 1999, Plant Ecol., 144, 201. 160. Kalisz, S., Hanzawa, F. M., Tonsor, S. J., Thiede, D. A., and Voigt, S. 1999, Ecology, 80, 2620. 161. Mix, C., Picó, F. X., van Groenendael, J. M., and Ouborg, N. J. 2006, Basic Appl. Ecol., 7, 59. 162. Price, M.V., and Waser, N.M. 1979, Nature, 277, 294 163. Waser, N.M. and Price, M.V. 1989, Evolution, 43, 1097. 164. Bond, W. J. 1995, Exctinction Rates, J.H. Lawton, R.M. May (Eds.), Oxford University Press, Oxford, 131. 165. Gómez, C., Espadaler, X., and Bas, J. M. 2005, Oecologia, 146, 244. 166. Guitián, P., Medrano, M., and Guitián, J. 2003, Plant Ecol., 169, 171. 167. Quilichini, A., and Debussche, M. 2000, Acta Oecol., 21, 303. 168. Christian, C. E. 2001, Nature, 413, 635. 169. Carney, S. E., Byerley, M. B., and Holway, D. A. 2003, Oecologia, 135, 576. 170. Gómez, C., and Oliveras, J. 2003, Acta Oecol., 24, 47. 171. Ness, J. H., Bronstein, J. L., Andersen, A. N., and Holland, J. N. 2004, Ecology, 85, 1244. 172. Wang, B. C., and Smith, T. B. 2002, Trends Ecol. Evol., 17, 379. 173. Schupp, E. W. 1993, Vegetatio, 107/108, 15. 174. Nathan, R. 2006, Science, 313, 786.

Ant-seed interactions

175. Nathan, R., Schurr, F.M., Spiegel, O., Steinitz, O., Trakhtenbrot, A., and Tsoar, A. 2008, Trends Ecol. Evol., 23, 638. 176. Traveset, A. 1994, Oikos, 71, 152. 177. Traveset, A., Bermejo, T., and Willson, M. 2001, Plant Ecol., 155, 29. 178. Martínez-Mota, R., Serio-Silva, J. C., and Rico-Gray, V. 2004, Biotropica, 36, 429. 179. Andresen, E. 1999, Biotropica 31, 145. 180. Leal, I. R., and Oliveira, P. S. 1998, Biotropica, 30, 170. 181. Malo, J. E., and Suárez, F. 1995, Oecologia, 104, 246. 182. Moles, A. T., Ackerly, D. D., Tweddle, J. C., Dickie, J. B., Smith, R., Leishman, M. R., Mayfield, M. M., Pitman, A., Wood, J. T., and Westoby, M. 2007, Global Ecol. Biogeogr., 16, 109. 183. Avgar, T., Giladi, I., and Nathan, R. 2008, Ecol. Lett., 11, 224. 184. Christianini, A.V., Mayhé-Nunes, A. J., Oliveira, P. S. 2007, J. Trop. Ecol., 23, 343.

Transworld Research Network 37/661 (2), Fort P.O. Trivandrum-695 023 Kerala, India

Predation in the Hymenoptera: An Evolutionary Perspective, 2011: 101-121 ISBN: 978-81-7895-530-8 Editor: Carlo Polidori

5. Individual prey specialization in wasps: Predator size is a weak predictor of taxonomic niche width and niche overlap Carlo Polidori1, Davide Santoro1, Josep Daniel Asís2 and José Tormos2 1

Dipartimento di Biologia, Sezione di Zoologia e Citologia, Università degli Studi di Milano via Celoria 26, 20133 Milano, Italy; 2Unidad de Zoología, Facultad de Biología Universidad de Salamanca 37071 Salamanca, Spain

Abstract. Intuitively, larger predators in a population may be favoured in subduing, handling and carrying to the nest a wider range of prey compared to smaller ones. In absence of any individual chemical or behavioural bias in prey choice, and independently from size-biased preference, it may be predicted: (1) that smaller wasps would have both narrower niches and lower niche overlap than larger wasps, (2) a positive correlation between individual niche width and niche overlap, and (3) a positive correlation between predator size and prey size. We tested these predictions using available data in literature on individual prey spectrum for 10 populations/generations of apoid wasps (Hymenoptera: Crabronidae). The analysis showed that individual specialization is widespread. Smaller individuals were more specialized than larger ones, as we hypothesized, in only one population, while the opposite was found in two cases; on the other side, niche width correlated (positively) with wasp size only in one population. In five out of the 10 examined cases predator size Correspondence/Reprint request: Dr. Carlo Polidori, Dipartimento di Biologia, Sezione di Zoologia e Citologia, Università degli Studi di Milano, via Celoria 26, 20133 Milano, Italy E-mail: carlo.polidori@unimi.it

102

Carlo Polidori et al.

correlated with prey size. Overall, the predictions were all true for just one population. In conclusion, different factors probably influence often more than predator size the dynamics of individual prey specialization in wasps. Circumstantial evidences suggest that individual-biased exploitation of different hunting areas, some behavioural traits of prey, and different ratios between prey and wasp size may partially account for the observed patterns.

Introduction Despite a lot of published models of predator-prey dynamics assume that conspecific individuals are identical and that they use the same resources [e.g. 1, 2], it is now clear that many apparently generalist species are composed of individual specialists that use small subsets of the populationâ&#x20AC;&#x2122;s niche [e.g. 3, 4, 5]. Such individual specialization may even exceed differences between conventional species [6, 7], and may have critical ecological and evolutionary implications [8]. For example, individual specialization has been thought to reduce intraspecific competition [9, 10, 11]. Individual specialization is generally related to constraints on an individualâ&#x20AC;&#x2122;s ability to efficiently exploit a wide variety of resources. Constraints generally arise from functional trade-offs in which consumers efficiently exploiting one type of resource are inefficient using another type of resource [8]. In central-place foragers, like predatory wasps, for example, the size of a predator may prevent the use of those prey taxa which are too large to be successfully subdued, handled, and carried to the nest [12]. However, most studies attempting to explain foraging behaviour in animals have assumed that there is no size variation among individuals within populations (for dissenting views, see [13, 14]), despite variation in individual body size can be large within populations [15]. Concerning Hymenoptera, only recently a central role of size dynamics was recognized in explaining host selection behaviour [e.g. 16]. Intuitively, larger individuals in a given population may be capable of using a broader range of prey species than smaller ones, which would be forced to hunt only small prey. Such pattern commonly occurs in other animal taxa and could be seen even through individualsâ&#x20AC;&#x2122; growth: for example, many marine fish predators undergo considerable increases in body size during their development, resulting in ontogenetic niche shift [17]. In wasps, such prediction is tentatively supported by recent studies. For example, in the digger wasp Cerceris arenaria L., a weevil-hunting specialist, larger females hunt both small and large weevils, while smaller females hunt only small ones (thus including in their diet a lower number of prey species) [18]. Such a study, however, based this suggestion on a

Individual prey specialization versus size in wasps

103

measure, prey species richness, which actually does not identify individual specialization [e.g. 11]. Moreover, larger C. arenaria females tend to prefer larger prey [18]; this positive correlation between the size of the predators and that of their prey has been repeatedly (but not always, see e.g. [19]) found also in various other species of wasps [e.g. 20, 21, 22, 23, 24, 25, 26], and seems to be ubiquitous throughout natural systems [e.g. 27, 28]. In wasps, even in monophagous species, larger females tend to hunt for larger individuals of the unique prey species [29]. Thus, if predator size shapes the prey spectrum at the individual level, one may expect intra-specific, size-dependent variation of the prey niche among individuals (in non-monophagous species). In this chapter, we contrasted available data on individual prey spectrum of selected species of apoid wasps (Hymenoptera: Crabronidae) with these predictions, starting from a simple model of size-niche relationship.

Assumptions and predictions of a simple size-niche model Consider a population with n wasps of different size (1/3 small, 1/3 medium-sized, and 1/3 large) hunt on the whole on m prey species (1/3 small, 1/3 medium-sized and 1/3 large). The individual selection may be mediated by wasp size at different degrees (from absence of prey size bias to strong preference of larger wasps for larger prey). As assumptions, encounter rate is identical for all females hunting on any of the prey species, and prey species have identical abundances and availabilities in the environment. In absence of any further chemical or behavioural (e.g. learning) bias in individual prey selection, the small wasps would hunt only the small prey (1/3m), mediumsized wasps both small and medium-sized prey (2/3m), and large wasps all the prey species, in proportions related to the degree of prey size preference (or due to constraints) (Fig. 1). Consider now two concepts related to individual niche: niche width and niche overlap. In the first case, specialists are individuals with foraging niches narrower than the populationâ&#x20AC;&#x2122;s niche, while in the second case specialists have niches that exhibit little overlap with the population niche [30]. These two variables related to niche do not have the same meaning. For example, the use of a rare dietary resource or having little dietary overlap with the population may not reflect niche width [31]. In our simple model, the following broadly used index, developed to estimate individual niche width [32] was used: Di =

1 â&#x2C6;&#x2018; pij2 j

104

Carlo Polidori et al.

where pij is the proportion of jth prey in the diet of ith individual; the higher is the value of Di, the wider is the niche. This predicts that the larger is the wasp size, the higher is Di. (Fig. 2a). Also, larger wasps are predicted to reduce such niche widths the stronger the prey size-bias selection (Fig. 2a). On the other hand, the following index, developed to estimate individual niche overlap [30] was used: PS i = 1 â&#x2C6;&#x2019; 0.5â&#x2C6;&#x2018; pij â&#x2C6;&#x2019; q j j

Figure 1. Schematic explanation of the model described in the text, using as example a hypothetical population composed of 3 females (in the picture, Stizus continuus (Hymenoptera: Crabronidae) of different size hunting overall 6 species of prey (in the picture, grasshopper prey of S. continuus) of 3 different size-classes. Each wasp hunted 100 prey individuals during the provisioning period, in proportions related to absence (percent numbers below the wasp pictures, valid for all the prey species), weak (percent numbers on the lines) or strong (percent numbers in brackets on the lines) size-biased positive selection. On the right of the prey pictures, the corresponding prey frequency in the diet of the population. See text for further explanations.

Individual prey specialization versus size in wasps

105

where pij describes the proportion of the jth resource category in individual iâ&#x20AC;&#x2122;s diet and qj is the proportion of the jth resource category in the populationâ&#x20AC;&#x2122;s niche; PSi varies from 0 (maximum individual specialization) to 1 (no individual specialization). This predicts higher PSi for larger wasps (Fig. 2c). Also, smaller wasps would be more specialized the stronger the prey sizebias selection (Fig. 2c).

Figure 2. Values of niche width (Di) (a), niche overlap (Psi) (b), their relationship (c), and predator-prey size relationship (d) resulting from the model of Fig. 1. In such simple model, the smallest female would have the narrowest niche width (lowest Di) (Fig. 2a) and it would be the more specialized (lowest PSi) (Fig. 2b). This would produce a strongly positive correlation between PSi and Di (Fig. 2c), which would be strongest in absence of individual size-biased selection. Also, an increase of individual size-biased selection would produce a stronger correlation between wasp size and prey size (Fig. 2d); this would decrease Di for larger females and decrease PSi for smaller females. See text for further explanations.

106

Carlo Polidori et al.

This situation will also produce automatically a positive correlation between PSi and Di, strongest in absence of individual size-biased selection (Fig. 2b), and a positive correlation between predator and prey size, even in absence of further size-biased selection (for medium and large wasps), (Fig. 2d). A first consequence of such simple model is that, in case females were all large enough to hunt for any prey species (i.e. very low variance in wasp size) and/or the prey have all the same exploitable size for all wasps, niche width and niche overlap would be very similar or even identical for all wasps (thus absence of individual specialization). A second consequence is that populations with lower variance in females’ size should have higher individual mean niche overlap (IS = mean PSi) when compared to populations composed of individuals with strong size variation. The population-level niche width (D) was calculated, as follows [32]:

D =

∑

1 q

2 j

(p and j as above)

Note that, contrary to IS, D can not be easily compared across populations, because its value is strongly dependent on prey richness [30, 32]. Hence, we compared among populations the ratio Di/D (referred in the following as “relative niche”), which represents how much the individual niche width is narrower than the population niche width. The four predictions of our model are thus summarized as follow: (1) smaller wasps would have narrower taxonomic prey niches (lower Di) than larger wasps, (2) smaller wasps would have lower taxonomic prey overlap (lower PSi) than larger wasps, (3) there would be a positive correlation between niche width and niche overlap per female wasp, and (4) there would be a positive correlation between wasp size and prey size.

Data sample and statistical analysis We selected from the literature those works that have information both on prey taxa used by different individuals of a given wasp species and a measure of size of individual wasps and their prey (body length, body weight, thorax width, or head width), in order to verify the four predictions listed above (Fig. 2). In some cases, such information was not explicitly available in these articles, but in these cases we had the raw data. Six works were selected, encompassing the following species: Cerceris arenaria [18], Cerceris rubida Jurine [24], Cerceris californica Cresson [20] Stizus continuus (Klug) [19, 33], Oxybelus lamellatus Olivier [34] and Bembix

Individual prey specialization versus size in wasps

107

merceti Parker [35]. We also used unpublished information on further samples of C. rubida (Polidori, unpublished data) and on one additional species (Bembix sinuata Panzer; AsĂs et al., unpublished data). All of these species, members of the family Crabronidae, share a number of important traits of their biology: they all provision subterranean nests with previously paralysed nymphs or adults of different insect taxa to feed their developing offspring. For two of these species, we obtained data for more than 1 generation and/or year; these data were analysed separately, because of the common fluctuations across time and space in the prey use by apoid wasps [33,36,37]. Thus, on the whole, data were available for 10 populations/ generations (Table 1). Because the indices used in the analysis may become quite inaccurate with very small sample sizes, only females associated with at least 4 prey items each were considered. The value of D indicates how much the niche is wide and the value of IS if prey spectra of individuals significantly overlap. To test the significance of the IS-values, we used the IndSpec1 software [30], which uses a nonparametric Monte Carlo procedure to generate replicate null diet matrices drawn from the population distribution, from which P values can be computed [30]. We used 10,000 replicates in Monte Carlo bootstrap simulations to obtain each P value. The calculated indices (Di and PSi) were plotted against the size of individual wasps and plotted one against the other, and predator size was plotted against prey size, to search for linear regressions. Pearson correlation test (for n>10; correlation coefficient: r) and Spearman correlation test (for n<10; correlation coefficient: Ď ) were employed to quantify the degree to which such variables are related. To avoid pseudo-replication problems, predator-prey size correlations were performed associating with each female the mean size of their prey. Wasp size variance of a population, calculated as the coefficient of variation (CV=SD/mean), was plotted against IS and Di/D across populations. Note that, for some populations, sample size for predator-prey size relationships does not always correspond to sample size used for analysis of taxonomic specialization (depending on how much individuals were associated to either prey spectrum or prey size). In the text and tables, mean values are reported Âą standard deviation.

Results and discussion In the Results section, species are grouped by their prey type (insect order). For each species, we give the results of the analyses and try to find out possible factors in the foraging biology that may account for the observed patterns.

Table 1. List of populations/generations selected from the literature (plus unpublished data) and used for the analysis, with information on sample sizes. * Sample sizes refers to the indicesâ&#x20AC;&#x2122; calculations and related analyses; for wasp-prey size relationships, in some cases sample was different (see text and Table 2).

108 Carlo Polidori et al.

Table 2. Prey species richness (R), niche overlap (IS), niche width (D), and average individual niche width (Di) for the 10 populations/generations used in the analysis, together with the results of linear correlation tests (in bold the significant tests). R=prey species richness. Size expressed as: 1body weight (mg), 2head width (mm), 3body length (mm). Niche overlap significance: *P<0.05, **P<0.01, ***P<0.001. Horizontal lines mark the three groups of populations/generations which hunt for three different insect orders (Orthoptera, Coleoptera, Diptera).

Individual prey specialization versus size in wasps 109

110

Carlo Polidori et al.

Orthoptera-hunting wasps Stizus continuus is a Mediterranean solitary wasp whose populations are bivoltine in Southern Europe (whole adult activity from early June to end of September) [38]; females dig multicellular nests in large aggregations, and hunt grasshoppers, mainly in the family Acrididae, to feed their immature offspring [39, 19]. The data used for this analysis come from three generations (the first one of 2005 and both those of 2007) at a single nesting site (Table 1). This was placed in the area of “Mallada Llarga”, located in “Dehesa del Saler”, part of the “Parque Natural de la Albufera” (Valencia, Spain). Two main microhabitats characterized the nesting site: a bared sandy soil, covered by a crust of salt (typical during the hottest months) and small and thick bushes composed by Salicornia ramosissima (J. Woods) and Sarcocornia fruticosa (L.). The individual prey specialization of this species varied through generations: diet overlap between individuals was significantly reduced and lower in the two generations of 2007 than in that of 2005 (Table 2). In this way, prey species richness was greatly lower (only 2 species hunted, with one accounting for >90% of prey specimens) in 2005 than in both 2007 generations, which showed similar values (although wasps of the two years hunted for different taxa) (Table 2). On the contrary, the niche width (D) and the average Di were more similar between the first generation of 2007 and 2005 than between the two generations of the same year; in particular, niche was widest in the second generation of 2007 (Table 2). Supporting our predictions, more specialized females (lower PSi) had narrower niches (lower Di) in the second generation of 2007. However, they had wider niches in 2005, disagreeing to predictions (Table 2). Also contrary to predictions, size of females did not correlate with individual niche overlap and niche width in both 2005 and the second generation of 2007, and the larger females had smaller diet overlap in the first generation of 2007 (Table 2). Moreover, the relationship between wasp and prey size varied through generations partially agreeing with our predictions, as it was significantly positive in both the two generations of 2007 (though in the first one the result was driven by an outlier, see [33]) (agreeing), but did not so in 2005 (disagreeing; Table 2 and Fig. 3). In this species, thus, size of females only partially, and sometimes not in the direction suggested by our model, seems to affect the degree of intraspecific diet variation. The observed patterns may be due to the factors that influence the overall prey preference at population level and probably also to the much lower number of prey species hunted in 2005 than in 2007. In 2005, S. continuus

Individual prey specialization versus size in wasps

111

primarily hunted for grasshoppers of the genus Heteracris, revealing, at least in the period of the study and in this area, this wasp to be almost monophagous. In contrast, grasshoppers of the genus Acrotylus were ignored by the wasps, in spite of their high abundance in the environment. Polidori et al. [19] found evidence that this bias occurred because Acrotylus is found more often on the soil and on grasses, while Heteracris is nearly only found on Sarcocornia bushes, which probably represents the habitat mostly exploited by the wasps for hunting. Very few females hunted for non-Heteracris prey, maybe because searching for prey on different microhabitats is not common. Also in 2007 S. continuus expressed selective predation, females hunting basically on species encountered at large bushes [33]. For example, the largely hunted Calliptamus (in the first generation) and Tropidopola (in the second generation) live mostly on bushes, whether the usually ignored Acrotylus and Sphingonotus lived preferentially on the ground or on short grass [33]. Once again, prey ecology largely accounts for the observed prey preferences. In the end of the 2007 summer, some prey species were excluded from the diet because they became too large to be carried in flight, so that variation in selectivity across wasp generations was probably weakly related to change in prey community composition [33]. Intraspecific niche variation in S. continuus seems thus a result of complex interactions between ecological prey preference and size constraints. In particular, a possible individual-based learning of hunting sites (bushes) would have stronger effect on shaping individual specialization than the female size. Moreover, in this wasp, the greater variance in size of the prey collected by larger females (that is, weak size-biased selection) may account for the absence of wasp size-prey size correlation in 2005; in this year, this would have been an additional factor that reduced the size-niche correlation, thus following our model. Interestingly, the ratio between prey and wasp size vary considerably among generations, being 0.47Âą0.23 in the first generation of 2007 and 0.88Âą0.24 in the second one (comparisons could not be done with 2005, because, in that year, only lengths of prey were available). This may force to a stronger individual prey specialization in the second generation and a stronger correlation between predator and prey size. At last, S. continuus females from both the first 2007 generation and the 2005 generation showed larger variation in size than those from the second generation of 2007 (Fig. 4). According to our model, we would expect an according variation in overlap among individuals (IS). This was not true: wasps, in fact, were not specialized in 2005, but exhibit strong specialization in 2007 (Table 2). A parallel situation was found for the relative niche width (Fig. 4). These results may be explained, at least in part, by distinct environmental abundances of potential prey in the two study years, roughly

112

Carlo Polidori et al.

reflected in the diet spectrum of the wasp [19, 33]. In 2005, the great variation in size among females could not reflect the great variation in individual diets, since the vast majority of potential prey was constituted by only one genus (Heteracris) and, possibly, by one single species (H. littoralis). On the other hand, in the first generation of 2007, the great variation in size among females could reflect the greater variation in individual diets. The discordance found in the second generation, however, could not be explained with this hypothesis.

Coleoptera-hunting wasps Cerceris arenaria is a Palearctic solitary digger wasp which nests in large and dense aggregations [40, 41]. It is a weevil-hunting specialist [37]. The studied nesting site was situated in the Regional Park of Adda Sud (Lombardy, northern Italy), which is mainly composed of cultivated fields (maize, soybean, and tomatoes). The nest aggregation of C. arenaria was located in a farm of a small town. At this site, females were active from the middle of June to the end of July [18]. Nine of the most abundant species recorded as prey for the marked females studied in 1999, belong to the genera Sitona and Otiorhynchus. Diet overlap between individuals was small, while niche width was around 3.6 for the population (D) and 2, on average, per individual (Di) (Table 2). There was no correlation between PSi and Di (Table 2). On the other hand, size of females negatively correlated with individual niche overlap, but did not with the niche width (Table 2). Thus, in this species, contrary to predictions of the model, the larger females were those more individually specialized (lower PSi). This parallels to the fact that the relationship between wasp and prey size was strong (Table 2 and Fig. 3). One possible explanation for this pattern may invoke the correlation between the size of prey and that of predator [18]. Such correlation is so strong that the larger wasps eventually discarded small prey species and prey mainly (or only) upon large prey, further increasing their specialization. Such situation resembles that found for S. continuus of the second generation of 2007, when larger wasps had lower diet overlap and predator-prey size relationship was very strong. Moreover, for 7 small wasps, all prey items were Sitona species. Such a genus was well represented in overall prey spectrum, so that, not being a rare resource, it does not increase specialization in such females. In addition, several evidences suggest that sometimes wasps exploit target hunting sites (e.g. trees) only for brief periods (from 1 day to 5 days), although some prey species were detected from wasps incoming to the nests for the whole study period (about 1.5 months) [18]. This may be a result

Individual prey specialization versus size in wasps

113

of temporary fidelity to particular, restricted hunting sites, and may be facilitated by the fact that Sitona and Otiorhynchus are associated with quite different plant hosts (basically grasses and trees/bushes, respectively). At last, the C. arenaria population showed the highest size variation among the study wasp populations (Fig. 4). Thus, according to our model, we expect low overlap among individuals. This holds true, with IS for C. arenaria population being the fourth lowest among the wasp populations (Table 2). Also the relative niche width (0.55) was moderately low (Fig. 4). Cerceris rubida is a small digger wasp confined to southern Europe, and its biology is remarkable in that it is the only European species of the genus known to be social [42]. Females dig nests in compact soils and fill them with previously paralysed small beetles as food for the larvae. This species is of interest to study diet preference dynamics because, in contrast to most Cerceris species, it is one of the few species that hunt for many beetle families (Chrysomelidae, Curculionidae, Nitidulidae and Phalacridae) [24]. Data for this species come from the Maremma Regional Park (Grosseto Province, Tuscany, Central Italy), from the middle of June to August (summer of 2005) [24]. In addition, unpublished data come from the same population studied in 2006. The Park is characterised by the Uccellina Mountains (a chain of hills parallel to the coast and covered by the thick Mediterranean maquis), Mediterranean pine-woods and extended cultivated fields (in particular maize, tomatoes, olives, and sunflowers). The nesting area of C. rubida consisted of a trail bounded by two cultivated fields (alfalfa Medicago sativa and wheat Triticum sp.). At this site, the wasp population hunted for phytophagous beetles mainly belonging to families Chrysomelidae, Curculionidae, and Phalacridae that were common around the wasp nesting site and that altogether represented almost the 95% of all the hunted beetles [24]. Diet overlap between individuals was small in both years, niche width (D) was about 1.6 and 4.6, and the average Di were about 1.3 and 2.3 in the two years (Table 2). There was no correlation between PSi and Di in any year (Table 2). Size of females neither correlated with the individual niche overlap nor with the niche width in any year (Table 2). The relationship between wasp and prey size was absent in 2006 but present in 2005 (Table 2 and Fig. 3). As to what found in S. continuus, and probably differently from C. arenaria, prey of C. rubida seemed to be hunted almost only in one specific environment (cultivated fields), thus suggesting that habitat accounted for most of the observed diet. However, inside these hunting sites, a further stronger selection appeared: beetle species differed markedly in their probability to be hunted, also after controlling for prey body size and prey body shape [24]. Chaetocnema scheffleri (Kutschera) (Chrysomelidae) had

114

Carlo Polidori et al.

the major risk to be hunted, followed by Chaetocnema tibialis (Illiger). Other chrysomelid beetles were not hunted, though very abundant in such areas. Thus, rather than an ecology-based selection (as in S. continuus), a taxonbased specialization at population-level is responsible for the observed pattern of prey preference, suggesting that in this species a restricted sensory window, that would exclude some beetle species from the diet, shapes prey choice. At last, C. rubida varied in size greatly in 2005 and moderately in 2006 (Fig. 4). According to our model, we thus expect lower overlap in 2005 and higher overlap in 2006. This seems to be true when comparing such values with those referred to other wasp populations (Table 2). A parallel situation was found for the relative niche width (Fig. 4). In particular, the very low overlap and the narrow relative niche in 2005 may be due to the same reason suggested for C. arenaria: larger females, which showed strong preference for larger prey, may discard smaller prey species; however, because, differently from C. arenaria, C. rubida females probably hunt all in the same hunting site, the segregation in the diet of small and large females would be less pronounced (i.e. PSi does not correlate negatively with size). Cerceris californica is a North American digger wasp nesting in dense aggregations [20]. The data analysed here come from a work carried out at a nesting site placed in the San Gabriel Mountains of Southern California. At that site, wasps provisioned their nests with buprestid beetles, mainly in the genus Acmaeodera. The authors noted that species of this genus are strongly associated with plants of the genera Adenostoma (Rosaceae) and Ceanothus (Rhamnaceae), so that they inferred that these plants are the main hunting sites for C. californica females. Thus, as in C. rubida and S. continuus, the ecology and feeding habits of prey species probably account for the populationâ&#x20AC;&#x2122;s diet preference. Diet overlap between individuals was small, while niche width (D) and the average Di were around 6.7 and 3.8, respectively (Table 2). A positive correlation appeared between PSi and Di (Table 2). In addition, size of females positively correlated both with individual niche overlap and with niche width (Table 2). Thus, in this species, according to predictions of the model, smaller females were those more individually specialized (lower PSi and lower Di). Also the relationship between wasp and prey size was positive and very strong (Table 2 and Fig. 3). At last, C. californica varied moderately in size in the studied population (Fig. 4). According to our model, we thus expect moderate overlap among individuals, and this seems to be true, giving that IS, although still very significant, was the sixth highest recorded (Table 2). Also the relative niche

Individual prey specialization versus size in wasps

115

Figure 3. Predator-prey size relationships for the 10 analysed populations. * body length for prey of C. rubida and S. continuus, pro-mesothorax junction width for prey of C. arenaria, elytra length for prey of C. californica, and head width for prey of B. sinuata; ** body length for C. californica and head width for all the other species. Trend lines are shown only for significant linear correlations. For population codes, see Table 1.

width was moderately high (Fig. 4). Therefore, this population almost perfectly agrees with all the predictions of our model.

Diptera-hunting wasps Oxybelus lamellatus is a solitary digger wasp distributed throughout the western part of the Mediterranean area. The data analysed here come from a study carried out at a nesting aggregation located in an area of compact sandy soil with little vegetation, surrounded by Arundo donax L., Juncus acutus L., J. maritimus Lamk., Phragmites communis Trin., and Thymelaea hirsuta (L.). This area was located some 100 m from the Mediterranean shoreline in the natural park called El Marjal del Moro, in the municipality of Sagunto (Valencia, Spain).

116

Carlo Polidori et al.

The prey consisted in males and females of Diptera belonging to 9 families, with Pollenia rudis (Fabricius) (Calliphoridae), Fannia scalaris (Fabricius) (Fannidae) and Neomyia comicina (Fabricius) (Muscidae) the most hunted species. Diet overlap between individuals was small and niche (D and Di) quite narrow (Table 2). Thus, despite O. lamellatus was considered opportunistic in regard to the type of prey [34], our analysis highlights that individual females are strongly specialized in prey choice. No correlation appeared between PSi and Di (Table 2), and size of females did not correlate with niche overlap and niche width (Table 2). Predator and prey sizes are not correlated (Table 2 and Fig. 3). Thus, in this species, female size does not seem to affect the intra-specific diet overlap. One of the possible reasons behind this pattern may consider the characteristic behaviour of prey carriage in this species: in fact, pedal carriage (prey hold with legs) was used for lighter prey, whereas abdominal carriage (prey impaled on the sting) was used for those of greater weight. For midsized prey, both types of carriage were observed. Thus, all individual wasps (regarding to their size, which, in any case, did not vary too much, see below) may have the possibility to carry to the nest prey of a wide range of sizes (prey weight/wasp weight from 0.6 to 2.4), shifting if this is particularly large from pedal to abdominal carriage. The individual niche width and the niche overlap, however, remain small due to some unknown individual behavioural bias in prey selection. At last, O. lamellatus was the species with lowest variation in size among the analysed populations (Fig. 4). According to our model, we thus expect high overlap among individuals (IS). This does not seem to be true, IS value being the lowest recorded among the wasp populations analysed (Table 2). The relative niche width was also low (Fig. 4). A study on the factors affecting the prey selection, including hunting sites, behavioural ecology of prey types, and chemical discrimination, are necessary to understand such patterns. Bembix merceti is a solitary Diptera-hunting wasp restricted to the Iberian Peninsula. The nests are unicellular and wasps form small aggregations. In a previous study, females of B. merceti were defined as opportunistic in prey selection, since whereas in 1993 nearly 90% of the prey were beeflies (Bombylidae), in 1994 the prey belonged mostly (50%) to Muscidae [43]. The data analysed here come from a study carried out in Soria (Central Spain) during summer of 2008, in a sandy area with sparse bushes in a landscape of cereal crops. Individuals of this species weakly differed in their

Individual prey specialization versus size in wasps

117

diets (P-value was just below the 0.05 level), while the relative niche width was moderate (Table 2, Fig. 4). PSi and Di were not correlated (Table 2), while size of females did not correlate with niche overlap and niche width (Table 2). The relationship between wasp and prey size was not significant (although close to significance) (Table 2 and Fig. 3). However, in this species, female size does not seem to affect greatly the intra-specific diet overlap. At last, B. merceti showed low variation in size (Fig. 4). According to our model, we thus expect high overlap among individuals. This seems to be true, IS showing weaker diet segregation with respect to all the other specialized populations (Table 2). One possible factor accounting for these patterns may be the relative small size of prey compared to wasp size. In fact, on average, prey weighed about 17% of the wasp weight (between 10% and 23%), and this may result in the possibility, for all the females, to hunt upon all the prey types in similar percentages. According to our model, if all wasps are large enough to subdue and carry all the prey types, intra-specific diet overlap will be greater and this seems, to some extent, to be the case for B. merceti. Bembix sinuata is a Mediterranean solitary wasp whose biology is little known. Published data are available only on the nest structure and prey [44]. This species nest in sandy, horizontal soil devoid of vegetation and prey upon flies of different families, such as Calliphoridae, Syrphidae and Muscidae [44]. The data analysed here come from an unpublished study carried out in Spain during 2007 summer, in the same area as previously described for B. merceti. Individuals of this species did not overlap their diet, while niche width (D and Di) was the widest recorded among the wasp populations reviewed here (Table 2). No correlation appeared between PSi and Di (Table 2), while size of females did not correlate with niche overlap and niche width (Table 2). As in the congeneric B. merceti, the relationship between wasp size and prey size was absent (Table 2 and Fig. 3). At last, B. sinuata showed moderate variation in size (Fig. 4). According to our model, we thus expect moderate overlap among individuals (IS). This does not seem to be true, IS being very low among the values obtained for the studied populations (Table 2). Also the relative niche width was quite reduced (Fig. 4). Interestingly, and maybe accounting for the observed difference between the two Bembix species, in B. sinuata the ratio between prey and wasp size was higher than in B. merceti, about 0.54 (between 0.48 and 0.57), i.e. in this case is unlikely that all the females may hunt all the

118

Carlo Polidori et al.

Figure 4. Mean niche overlap (IS=mean PSi) and mean relative niche width (Di/D) across the 10 analysed populations. Populations are ordered by increasing wasp size variance (CV=SD/mean, values showed above the bars). The CV values were all calculated from linear measurements (head width, thorax width, or body length), in order to make possible inter-specific comparisons. For population codes, see Table 1.

prey species. This would at least partially explain the lower intra-specific diet overlap in this species relative to B. merceti. Thus, in this species, female size does not seem to affect greatly the intra-specific diet overlap.

Across-populations comparisons and conclusions of the study This review shows that the degree of intra-specific diet overlap and the individual niche width are related to predator size in a variety of ways which are difficult to predict in absence of additional, detailed information on both the predator and prey populationsâ&#x20AC;&#x2122; biology. The four predictions of our model (positive correlation between niche overlap/niche width and size, positive relationship between the two indices and positive correlation between predator and prey size) were all accomplished by only one of the populations considered here (C. californica). In this case, as predicted by the model, smaller females weakly overlap their diet, have narrower niches and hunt for smaller prey indicating that predator size is a good predictor of the degree of individual specialization. Only the positive relationship between predator and prey size was found quite commonly throughout the study populations (in five populations). Interestingly, in two cases (S. continuus I and C. arenaria),

Individual prey specialization versus size in wasps

119

we found a negative correlation between size and niche overlap (larger wasps more specialized). Wasps eventually segregate their niches, disagreeing with the main assumption of our model, i.e. that larger wasps hunt for a wider spectrum of prey species than smaller ones. As a consequence of all these results, average individual specialization (IS and mean Di/D) in general did not correlate with the variance of femalesâ&#x20AC;&#x2122; size. On the contrary, populations having moderate size variance seem to be often also those with low individual specialization (Fig. 4). Maybe, in populations with high variance, larger wasps can eventually discard smaller prey species, as maybe occurs in C. arenaria, so that the average individual specialization increases. This situation is in accordance with the optimal foraging theory. On the other side, at least some populations with very small size variation (e.g. O. lamellatus, S. continuus II, C. californica) do not show, contrary to expectations, high overlap; maybe these wasps, to avoid competition, tend to segregate more their prey spectrum. Intra-specific competition has been shown to increase individual specialization in other animal taxa [e.g. 45, 46]. These findings somehow support what has been recently found for a fish, the threespine stickleback (Gasterosteus aculeatus L.): although the similarity in trophic morphology was correlated with dietary similarity between fish individuals, both body size and trophic morphology accounted for low variance in diet overlap [47]. Interestingly, the ratio between prey and wasp size seems to account, to some extent, for the observed differences among populations. In fact, although the ratio was calculated only if the weight was used as a measure for both predator and prey size (for a total of 4 populations), one notes that the two populations with higher prey/wasp weight ratio (S. continuus II: 0.88; O. lamellatus: 1.35) had very high individual specialization (S. continuus II: 0.34; O. lamellatus: 0.45). On the contrary, B. merceti and S. continuus I (0.17 and 0.47 of weight ratio, respectively) had lower specialization (S. continuus I: 0.64; B. merceti: 0.56). A weight ratio well below 1 may indicate that most of females are able to hunt all the (generally small) prey species. In our model, if all wasps could hunt all prey species, individual specialization would be drastically reduced. In conclusion, large females of apoid wasps rarely exploit their theoretical size-dependent advantage enlarging their prey spectrum, probably because different additional factors, such as individual-biased learning for certain prey taxa [e.g. 48], particular prey-subduing techniques [e.g. 49], particular hunting sites [e.g. 50], or even depending on competitionâ&#x20AC;&#x2122;s pressure [e.g. 45, 46], have more influence on the dynamics of prey specialization. In this sense, the niche axes have to be chosen appropriately when measuring niche variation, because, if functionally distinct resources

120

Carlo Polidori et al.

are pooled (e.g. simply according to taxa), one may underestimate individual specialization; conversely, high among-individuals variation may not be biologically significant if it is based on short-time sampling [30]. Thus, the use of several sampling schemes is necessary to better understand the predator-prey dynamics in wasps, as it (I) ensures temporal consistency in trophic spectrum variation; and (II) capture all possible behavioural and morphological traits of prey and predators. Nevertheless, predator size distribution and predator-prey size relationship still remain, despite in nonobvious ways, an important factor shaping resource use at the individual and population-level in these insects.

Acknowledgements Thanks are due to Riguel Feltrin Contente, which critically reviewed an early manuscript of this chapter, improving it. We are indebted to the Maremma Regional Park offices for the permit provided to study C. rubida biology in the protected territory.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.

Hutchinson, G.E. 1957, Cold Spring Harbor Symp. Quant. Biol., 22, 415. Pielou, E.C. 1975, Ecological diversity, Wiley, New York. Price, T. 1987, Ecology, 68, 1015. West, L. 1986, Ecology, 67, 798. Durell, S.E.A. le. V. dit. 2000, Biol. Rev., 75, 503. Werner, T.K., and Sherry, T.W. 1986, Proc. Natl. Acad. Sci. USA, 84, 5506. Ehlinger, T.J., and Wilson, D.S. 1988, Proc. Natl. Acad. Sci. USA, 85, 1878. Bolnick, D.I., Svanb채ck, R., Fordyce, J.A., Yang, L.H., Davis, J.M., Hulsey, C.D., and Forister, M.L. 2003, Am. Nat., 161, 1. Van Valen, L. 1965, Am. Nat., 99, 377. Roughgarden, J. 1972, Am. Nat., 106, 683. Svanb채ck, R., and Bolnick, D.I. 2007, Proc. R. Soc. Lond. B. Biol. Sci., 274, 839. Coelho, J.R. 2011, in: Predation in the Hymenoptera. An Evolutionary Perspective, C. Polidori, (Ed.), Research Signpost, India, Trivandrum. Mittelbach, G.G. 1981, Ecology, 62, 1370. Persson, L., and De Roos, A.M. 2003, Ecology, 84, 1129. Bolnick D.I., Svanb채ck R., Fordyce J.A., Yang L.H., Davis J.M., Hulsey C.D., and Forister M.L. Honek, A. 1993, Oikos, 66, 483. Henry, L.M., Ma, B.O., and Roitberg, B.D. 2009, Oecologia, 161. 433. Scharf F.S., Juanes F., and Rountree R.A. 2000. Mar. Ecol. Progr. Ser., 208, 229. Polidori, C., Boesi, R., Isola, F., and Andrietti, F. 2005, Eur. J. Entomol., 102, 801.

Individual prey specialization versus size in wasps

121

19. Polidori, C., Mendiola, P., Asís, J.D., Tormos, J., Garcia M.D., and Selfa, J. 2009, J. Nat. Hist., 43, 2985. 20. Linsley, E.G., and MacSwain, J.W. 1956, Ann. Entomol. Soc. Am., 49, 71. 21. Lin, N. 1979, J. Washington Acad. Sci., 68, 159. 22. Polidori, C., Boesi, R., Ruz, L., Montalva, J., and Andrietti, F. 2009, Stud. Neotrop. Fauna E., 44, 55. 23. Hastings, J.M., Holliday, C.W., and Coelho, J.R. 2008, Fla. Entomol., 91, 657. 24. Polidori, C., Gobbi, M., Chatenaud, L., Santoro, D., Montani, O., and Andrietti, F. 2010, Biol. J. Linn. Soc. 25. Alexander, B. 1985, J. Nat. Hist., 19, 1139. 26. Karsai, I., Somogyi, K., and Hardy, I.C.W., Biol. J. Linn. Soc., 87, 285. 27. Fisher, D.O., and Dickman, C.R. 1993, Ecology, 74, 1871. 28. Roger, C., Coderre, D., and Boivin, G. 2000, Entomol. Exp. Appl., 94, 3. 29. Strohm, E., and Marliani, A. 2002, Behav. Ecol., 13, 52. 30. Bolnick, D.I., Yang, L.H., Fordyce, J.A., Davis, J.M., and Svanbäck, R. 2002, Ecology, 83, 2936. 31. Sargeant, B.L. 2007, Oikos, 116, 1431. 32. Levins, R. 1968, Evolution in changing environments: some theoretical explorations, Princeton University Press, Princeton, New Jersey. 33. Santoro, D., Polidori, C., Asís, J.D., and Tormos, J. 2011, J. Animal. Ecol., doi: 10.1111/j.1365-2656.2011.01874.x 34. Tormos, J., Asís, J.D., Gayubo, S.F., Portillo, M., and Torres, F. 2000, Ann. Entomol. Soc. Am., 93, 326. 35. Asís, J.D., Baños-Picón, L., Tormos, J., Ballesteros, Y., Gayubo, S.F., Alonso, M. 2011, Behaviour, 148, 191. 36. Brockmann, H.J. 1985, J. Kansas Entomol. Soc., 58, 631. 37. Polidori, C., Boesi, R., Pesarini, C., Papadia, C., Federici, M., Bevacqua, S., and Andrietti, F. 2007, Zool. Stud., 46, 83. 38. Polidori, C., Mendiola, P., Asís, J.D., Tormos, J., Selfa, J., and Andrietti, F. 2008. Anim. Behav., 75, 1651. 39. Asís, J.D., Tormos, J., and Jiménez, R. 1988, Ent. News, 99, 199. 40. Field, J., and Foster, W.A. 1995, Anim. Behav., 50, 99. 41. Polidori, C., Casiraghi, M., Di Lorenzo, M., Valarani, B., and Andrietti, F. 2006, J. Ethol., 24, 155. 42. Polidori, C., Federici, M., Papadia, C., and Andrietti, F. 2006, It. J. Zool., 73, 55. 43. Asís, J.D., Tormos, J., and Gayubo, S.F. 2004, J. Nat. Hist., 38, 1799. 44. Asís, J.D., Gayubo, S.F., and Tormos, J. 1992, Deut. Entomol. Z., 39, 221. 45. Bolnick, D.I. 2001, Nature, 410, 466.a 46. Svänback, R., and Bolnick, D.I. 2007, Proc. R. Soc. Lond. B. Biol. Sci., 274, 839. 47. Bolnick, D.I., and Paull J.E. 2009, Evol. Ecol.Res., 11, 1217. 48. Punzo, F. 2005, J. N. Y. Entomol. Soc., 113, 222. 49. Weiss, M.R., Wilson, E.E., and Castellanos, I. 2004, Anim. Behav., 68, 45. 50. Ehlinger, T.J. 1989, Anim. Behav., 38, 643.

Transworld Research Network 37/661 (2), Fort P.O. Trivandrum-695 023 Kerala, India

Predation in the Hymenoptera: An Evolutionary Perspective, 2011: 123-198 ISBN: 978-81-7895-530-8 Editor: Carlo Polidori

6. The art of managing weapons: The stinging behaviour of solitary wasps in the eyes of past, present and future research Francesco Andrietti Dipartimento di Biologia, Sezione di Zoologia e Citologia, Università degli Studi di Milano via Celoria 26, 20133 Milano, Italy

Abstract. Studies on sting morphology and venom chemistry of predatory wasps showed impressive progress in the last two decades; by contrast, researches dealing with the behaviour associated with prey stinging are comparatively less numerous and much of what we know come from 19th and 20th centuries up to middle 80s’. Neverthless, in the last 10 years a new interest in this topic appeared, and a number of stimulating studies involving an integrative approach using histological, biochemical, ethological and neurological methods was carried out, although mainly concerning two species (Ampulex compressa and Liris niger). Here one makes a temporally organized review of what we know about stinging behaviour and prey manipulation of solitary predatory wasps. On the whole, more or less detailed information on stinging patterns is available for species belonging to all the major groups of solitary predatory wasps: Ampulicidae, Sphecidae, Crabronidae, Eumeninae and Pompilidae. Neverthless, some important lineages of wasps preying upon taxa which could elucidate the role of prey Correspondence/Reprint request: Dr. Francesco Andrietti, Dipartimento di Biologia, Sezione di Zoologia e Citologia, Università degli Studi di Milano, via Celoria 26, 20133 Milano, Italy E-mail: francesco.andrietti@unimi.it

124

Francesco Andrietti

neuroatonomy on the stinging behaviour were completely ignored. A detailed critical review of early observations (19th century) and comparisons with modern researches revealed that many of the hypotheses suggested by old classic studies were mostly confirmed using modern techniques of analyses. In particular, the “locomotory ganglia hypothesis”, which stresses the importance of distribution of prey ganglia on the spatial pattern of wasp stings, seems to probably unify the mechanism of stinging behaviour of predatory wasps. Nevertheless, exceptions to this rule emerged, and only new studies devoted to observe species neglected to date could confirm or reject it.

Introduction Belonging to the Aculeata, solitary wasps which prey upon other arthopods to feed their offspring (Apoidea and Vespoidea) possess a sting, i.e. a strongly modified ovipositor with no egg-laying function [1]. The ability to paralyse the prey by means of the sting is certainly one of the features of Aculeata that made possible the evolutionary success of this group of Hymenoptera. The sting apparatus of solitary wasps may be expected to have been under selection in relation with hunting behaviour, and modifications of the sting is expected to vary depending on the prey type and prey carrying [e.g. 2, 3, 4]. However, differently from research on the sting morphology and venom chemistry of the Aculeata, which showed impressive progress in the last two decades [e.g. 5, 6, 7], little is being currently done about the stinging behaviour of predatory wasps, in particular in relation to the position and frequency of stings on the prey body during the capture, although detailed data are now available for few species [8, 9, 10, 11, 12, 13, 14, 15, 16, 17]. Much of what we know about stinging behaviour of wasps from a comparative point of view date in practice before 1990, and no new reviews on this topic were carried out since 1986 [2]. However, the aim of the present study is by no mean that to give an exhaustive review of the ethology of the different ways how a solitary wasp may sting her prey. Such a work, at least up to the year 1986, has already been done by Steiner who, in the Tables I-V of his excellent paper, condenses all published information up that date regarding the stinging behaviour of “solitary wasps” (including some Terebrantia) [2]. Instead, the present work is focused to a few restricted points: 1) a rather detailed account of the oldest researches, confined within the 19th century; 2) a short exposition of different lines of researches developed since the last decades of the 20th century, and 3) a discussion regarding the “locomotor ganglia hypothesis”. According to it, the wasp directs a sting for each nerve centre involved in a locomotion, attack or

Stinging behaviour of solitary wasps

125

defence of the victim. This would imply that for most predatory wasps stings are directed to the throat (suboesophageal ganglion) and three thoracic segments of their prey.

Deep in the past: Studies on wasp stinging behaviour in the 19th century I will outline in this section four leading figures which operated within the last decades of the 19th century: Jean Henry Fabre, Charles Ferton, in Europe, and the Peckams, in the United States. Even if many other researchers working at the time could be considered, I believe that the four indicated above are those that better characterize this pioneering era of modern wasp ethology. Their researches will be examined in some detail, for the sake to show the variety of situations concerning the stinging patterns, the prey manipulation and the prey paralysis. Moreover, they may illustrate the first controversies regarding the fixity or variability of stinging behaviour, the relationship between stinging patterns and the location of prey nerve centers, the way how paralysis is induced, and other problems which will become matter of further analysis for the following century. For a more comprehensive understanding, in some cases the results of recent findings will be exposed, when they may either confirm or contrast those reported by the early literature.

Jean Henri Fabre (1823-1915) According to Fabre, the wasp stinging behaviour is strictly correlated to the nervous anatomy of the prey. In the case of Cerceris, for example, since the stinging position is located between the first and the second pair of legs, the thoracic ganglia should be concentrated there. This is the reason why, always according the Fabre, the type of prey hunted by this wasp is restricted to the coleoptera of the families Buprestidae and Curculionidae, which present thoracic ganglia fused or located very close together (Curculionidae) or partially fused (Buprestidae). Other coleoptera with fused thoracic ganglia like Scarabeidae, Histeridae or Scolytidae should be excluded due to the difference in habitat with respect to that of Cerceris or to their small size. Inside the two families, prey of different size may be captured according to the size of the wasp [18-I: pp. 77, 84-87]. In fact, Fabre did not know of any other family preyed by Cerceris wasps (except the case of a species which captures Halictidae: see [18-IV: p. 240]. In the same period when Fabre wrote the first volumes of his Souvenirs Entomologiques, Marchal [19] (an extract may be found in [20: pp. 200-210])

126

Francesco Andrietti

wrote a memory devoted to the behaviour of Cerceris rybyensis (Linnaeus); this wasp hunts different bee species in the family Halictidae which, as all other Hymenoptera, have a relatively non concentrated ventral cord [21-II: p. 1239]. He was able to study in detail the stinging method placing a wasp with a bee under a bell glass. It is worthwhile to report here some of his considerations, either for what strictly concerns the way of stinging or for the more general implications they have with regard to Fabreâ&#x20AC;&#x2122;s opinions At the moment of capture Cerceris seizes the bee brusquely, clasping the anterior part of the body with her mandibles. Her recurved abdomen darts the sting into the neck at the articulation of the head and thorax, the stroke being given vigorously as though it were of capital importance. For a few moments the two combatants roll on the ground; one or two quick strokes are given under the thorax, principally between the prothorax and the mesothorax, and the bee becomes motionless; then Cerceris holds her victim face to face with her, looks at it a few seconds, and turns it around so as to bring the neck opposite her mandibles. The bee being thus adjusted the wasp proceeds to dig at the nape of the neck, squeezing it for from two to four minutes...Marchal does not agree with Fabre in his belief that wasps are endowed not only with tools but with the method of using them, the gift being original, perfect from the beginning, not modified by past or future. The action of Cerceris does not imply any mysterious science. She runs the end of her abdomen along the under surface of the thorax of the bee and stings at the division of the segments â&#x20AC;&#x201C; that is at the points where the sting can enter. The order in which the strokes are given is very variable, and if the neck is protected by gum-arabic, so that it is impervious, the stings between the pro- and mesothorax give just the same result. All that is necessary is that the sting shall reach the line of nervous matter that runs along the ventral face of the thorax. As a matter of fact it does not touch the ganglia, but enters just half way between them. The distance is small and the poison quickly reaches the nervous centers. The order in which the stings are given is variable. We may consider than there are four classes: In class A. the sting is given: 1st, at the neck. 2nd, at the articulation of the prothorax and the mesothorax. 3rd, at the neck. In class B. it is given: 1st, at the neck.

Stinging behaviour of solitary wasps

127

2nd, at the articulation of the prothorax and the mesothorax. In class C. it is given: 1st, at the neck. 2nd, at the articulation of the prothorax and the mesothorax. 3rd, behind, toward the origin of the abdomen. In class D. it is very probably given in a considerable number of cases, either only in the articulation of the prothorax and the mesothorax, or only in the neck. Cerceris is far from using the exquisite method of malaxation followed by Sphex, as described by Fabre. She is on the contrary, quite brutal. She pricks and squeezes the neck of the victim and then lick off the juice that exudes. All the bees that have undergone this operation have the neck cut, some on the median line, some on both sides. The supposition that Cerceris proceeds differently with the bees which are destinated to feed the larvae, perhaps not malaxing them at all, or only delicately, proves to be incorrect, since half the bees that were taken from nests showed marks of being cut. The other half, upon which no sign of a wound was visible, had probably received lesions which resulted in death after a brief delay. Of five bees taken from nests where the eggs had not yet hatched...only one responded to stimulation of the electric current by flexing the anterior legs at the moment that it was applied, the others giving no reaction. Moreover of three taken from a nest not yet fully provisioned one gave no response to the electrical current (from the Marchalâ&#x20AC;&#x2122;s memoir extract given by [20: pp. 201-203]). However, half an hour after being stung, the bee shows a slight recovery whose extent depends upon the fact of having or not received malaxation (from the Marchalâ&#x20AC;&#x2122;s memoir extract given by [20: p. 205]). In conclusion, according to Marchal 1. The effect of the sting is to produce inhibition; this permits Cerceris easily to carry the bee into her hole. 2. Inhibition ceases, but the nervous system has been so injured that the insect does not recover. Still motion may persist for a long time if there has been no malaxation. 3. Malaxation by the effusion of blood which it produces, and also by the lesion of the nervous chain, strikes a fatal blow at voluntary movement, and leads to the suppression of animal life in about twenty-four hours (from the Marchalâ&#x20AC;&#x2122;s memoir extract given by [20: p. 207]).

128

Francesco Andrietti

Malaxation has an effect even on the preservation of the prey after they died (or at least cany movement was ceased): the malaxed bees dried up at once, while the others remained fresh for a longer period, maybe still maintaining a kind of vegetative life (from the Marchal’s memoir extract given by [20: pp. 207]). To validate his hypotheses Fabre performed a few experiments by inoculating a droplet of ammonia at prothorax junction behind the first pair of legs. The result was very variable according to the kind of insect: with Scarabeidae, Buprestidae and Curculionidae, which possess thoracic ganglia close to each other, a complete paralysis similar to that induced by the Cerceris was obtained and they remained in life for one or two months. In the case of insects with distant thoracic ganglia like Carabidae or Cerambycidae, instead, the effect was different and after some quick and uncoordinated convulsions recovered in a few hours or the day after [18-I: pp. 88-89, 91-92]. Ammophila shows a completely different pattern of stinging behaviour. Placed under a glass bell together with a caterpillar, an Ammophila hirsuta Scopoli stings it at the thorax level, from the third to the first segment, where it seems to be stung more insistently. After this first phase, the caterpillar is released and the wasp performs a very special behaviour which Fabre calls a “dance of triumph”, then seizes the caterpillar from the dorsal side and stings it again ventrally at every segment from the anterior to the posterior end, with exception of those already operated, progressively shifting behind the position where the prey is seized. The caterpillar is released, then seized again by Ammophila with her mandibles at the level of the beginning of first thoracic segment (close to the nervous centers of the brain) and chewed for many minutes: the wasp mandibles movements are sharp, but distant one from the other, as if the wasp was checking their effect. When this last operation is terminated, the mandibles of caterpillar, which after the stings were still in movement, become quiet. The provided scheme admits a few variations, as the reduction of the stings of the first phase to two or even one (at the first segment). In such case the missed segments are stung at the beginning of the second phase, and sometimes the first segments are stung twice, in both the first ands the second phase. It is frequent that the last two or three segments are missed and rarely the direction of the stings is reversed from the posterior to the anterior end. The chewing of the first segment (malaxation) may be omitted in case the caterpillar mandibles, which could interfere with the transport, remain motionless after the stings [18-IV: pp. 248-251]. Exactly the same stinging behaviour has been observed in A. apicalis Brullé, preying on a geometrid caterpillar - two stinging phases spaced by the “triumph” dance, malaxation, same number and order of stings - notwithstanding

Stinging behaviour of solitary wasps

129

the rather different external morphology of the caterpillars (like legs distribution). What makes Fabre to claim: “the anatomy of the game, rather than its form, controls the huntress' tactics” [18-IV: pp. 254-256]. A situation intermediate between that of Cerceris and that of Ammophila was depicted by Fabre in the chapter entitled “Three strokes of a dagger” [18-I: pp. 107-115]. It concerns Sphex flavipennis Fabricius which hunts young locusts. However there are some problems on the correct identification of this species: on the base of some drawings of Fabre, Berland [22] who examined Fabre’s material stated that it was Sphex maxillosus Fabricius (= Sphex rufocinctus Brullé), but Ferton excluded this possibility due to the kind of prey [23]. She places herself body to body with her adversary, but in a reverse position, seizes one of the bands at the end of the cricket’s abdomen and masters with her forefeet the convulsive efforts of its great hindthigs. At the same moment her intermediate feet squeeze the pantaing sides of the vanquished cricket, and her hind ones press like two levers on its face, causing the articulation of the neck to gape open. The Sphex then curves her abdomen vertically, so as to offer a convex surface impossible for the mandibles of the cricket to seize, and one beholds, not without emotion, the poisoned lancet plunge once into the victim’s neck, next into the jointing of the two front segments of the thorax... [18-I: pp. 108-109]. This first stinging is followed by a second one between the first and the second thoracic segment and a third one at the junction with the abdomen. In fact, in the nervous system of the cricket the three thoracic ganglia are separated: this should be the reason of the three stings [18-I: pp. 109-112]. Only slightly different is the stinging behaviour of Palmodes occitanicus (Lepeletier & Serville), called by Fabre “Sphex languedocien”, a related species hunting Ephippiger females with ripe eggs. The paralysis seems to be less pronounced that in the preceding case and this should be related, according to Fabre, to the fact that P. occitanicus, differently from S. flavipennis, hunts one single prey. In fact in both cases the egg is laid in a position protected from the movements of the prey (close to the base of one of the posterior legs or more laterally between the first and second pair of legs, respectively) to which is attached, but could be stressed by the movements of an other non completely paralysed prey previously brought into the cell. A first sting is made in the thorax, the second behind the neck, but apparently the sting is pushed toward the thoracic ganglia. If the Ephippiger hinds the transport by means of its tarsi or mandibles, the wasp

130

Francesco Andrietti

performs malaxation, chewing its brain centers with her mandibles inserted through the neck membrane. In such way a complete paralysis is induced which lasts a few hours, then one goes back to a less paralysed state [18-I: pp. 159, 172, 174-176, 178, 179, 183-184]. Among eumenine wasps, Odynerus nidulatur Saussure (= Symmorphus murarius (Linnaeus)) was found to turn the prey, a larva of leaf beetle (Chrysomelidae) close to pupation, ventral side up and to give one stings at the level of each of its thoracic ganglia. After stinging, Odynerus performs malaxation on the neck of the prey. In some cases the larva of the leaf beetle is stung in the terminal segments beginning from the most distal ones (i.e. in a reversed order with respect to the stings in the thoracic ganglia), and then the last three of them are chewed. In such way the liquid contained in the alimentary channel is pushed behind and is licked from the wasp. Then the prey is abandoned. However, this second behaviour is less regular than the standard one: sometimes the stings reach the thoracic ganglia, while in some other cases stings are absent and the wasp feeds on a non paralyzed prey [18IV: pp. 196, 204-207]. Even Ferton observed a different Odynerus species malaxating her prey, but without discarding it: from time to time Odynerus nobilis Saussure stopped her provisioning work and took out from the cell his whole content; then the prey were reviewed, malaxated to extract the juice which was licked from the wasp, and re-introduced in the nest. Malaxation was observed even in Odynerus alpestris Saussure, from the anterior to the posterior extremities of the preyed caterpillar [24]. Philanthus triangulum Fabricius equally feeds on the content of the crop of her prey, the honey bee worker. The prey is stung ventrally, between the head and the neck (prothorax). The sting reaches the cerebral ganglion inducing a real death, not only a locomotory paralysis. Next follows a special kind of manipulation, consisting in chewing and squeezing the bee neck and abdomen in a way which allows nectar contained in the crop to escape from mouth to be licked from the wasp. The rational of the behaviour to kill the prey, instead of paralyzing it, consists in the fact that in dead prey any muscular tonicity is abolished and the flowing back of the nectar made easier. This is the reason to sting below the neck instead than in the thorax. But there is an additional reason to empty the bee from its nectar, due to the rigorous carnivorous diet of wasp larvae which, after a first acceptance, soon disliked experimentally killed bees non previously emptied [18-IV: pp. 214-216, 221, 225, 231]. In accordance with this consideration, Fabre expected that any wasp hunting bees would behave in the same way [18-IV: p. 240]. One has seen that for C. rybyensis this prevision is only in part confirmed. Instead, for the other species of beewolves (Philanthus) preying upon bees, according to Evans & Oâ&#x20AC;&#x2122;Neill [25: p. 25] this behaviour has never been observed in any of

Stinging behaviour of solitary wasps

131

the North American species. However, Kurczewski and Miller [26] found that females of P. sanbornii Cresson, before introducing the hunted bees in the cell, probably cleaned them from the pollen they bored, sealing it at the end of a side passage. Regarding the other statements made by Fabre concerning P. triangulum, many of them did not receive confirmation. According to Rathmayer [27], the bee results to be stung behind the front coxae instead than under the neck and it is not killed. Moreover, provisiong is massive and non progressive and oviposition is made on the last instead that on the first prey, as supposed by Fabre [18-IV: pp. 228-230; 28]. The stinging method used by Scolia (Hymenoptera: Scoliidae) is of special interest since the prey is a larva of Cetonia or other scarabeid beetles for which one would expect a serial pattern of stings, as in the case of Odynerus or Ammophila. Instead, these larvae present a particular nerve cord which, as in the corresponding adult insects, shows a fused ganglionar organization since thoracic and ventral ganglia form an unique mass. In accord with this finding, Fabre foresaw that the stinging behaviour should consist in only one sting in the ganglionar mass. On the other hand additional stinging would be hard to realize in the difficult conditions of hunting under the ground [18-III: pp. 48-53]. To check what effectively happens during the underground hunt of Scolia is hardly feasible, but Fabre succeeded to study under a glass bell the case of S. bifasciata van der Lind. (= S. hirta (Scrk.)?) and Scolia (Colpa) interrupta Latreille, preying respectively on Cetonia ed Anoxia. The results of the observation agreed with the expected ones: the scarabeid larvae were effectively stung only once ventrally, as predicted on the base of the anatomical findings: between the first and the second thoracic segment for the first species; under the neck for the second one [18-IV: pp. 258, 260, 266]. The result is that of a very deep paralysis, â&#x20AC;&#x153;Never, since my remotest investigations, have I witnessed so profound a paralysisâ&#x20AC;? (18-III: p. 47). In fact, the unique stroke in the ganglionar mass paralyzes at the same time the legs and the whole animal. Even in the case of the mantis-killing Tachytes (probably Tachysphex jullianii Kohl, see [29: note 1 of the eds., p. 128], the anatomical structure of the prey nerve cord determines the stinging pattern of the wasp. In fact, in the mantis species the second and third locomotory ganglia are widely spaced from the first one, matching the long prothorax which divides the front legs from the four other ones. The first ganglion is the largest of the three, and is also that where the first stroke will be delivered in order to abolish any offensive movement of the front raptorial legs. Subsequently, the wasp retreats along the mantisâ&#x20AC;&#x2122; back and stings two closed points, corresponding to

132

Francesco Andrietti

the ganglia which innerve the second and third pairs of legs. What is more striking is the unsymmetrical pattern of the stings distribution which, unlike what observed in other cases, as in Ammophila, are not uniformly distributed along the mantis length, in conformity with the anatomy of the prey. A different prey, as a small grasshopper, even if accepted as a diet from Tachytes’larvae, will not be stung by the mother since, according to Fabre, she does not know the way to operate upon a different nervous structure [18-III: pp. 242, 254-261]. Lastly, for what concerns spider wasps, Fabre observed under a wiregauze cover the stinging behaviour of Cryptocheilus sexpunctatus Fabricius (called by Fabre Calicurgus scurra) in regard of Epeira fasciata Walck. Contrary to what expected according to the nervous organization of the spiders cord, consisting in only one ventral ganglion, the first blow is not aimed at the spider’s cephalotorax. In fact the wasp, after having overturned the spider on its back and mounted on its top, mastering its legs with her legs and hanging the cephalotorax with the mandibles, drives her sting into spider’s mouth “with minute precaution and marked persistency”. Immediately, the paralyzed poison-fangs close, leaving the spider harmless. A second sting made at the level of the cephalotorax-abdomen attachment paralyzes all eight legs at once. Even if the ganglion is located a little above this point, it should be reached due to inclination of the sting. The overall effect of the two stings is to paralyze both fangs and legs. The first ones will stay so for ever, while the legs movements, although in a quite uncoordinated manner, may recover after some time. The palpes, instead, will remain irritable and mobile [18-III: pp. 270, 276-281]. What is very surprising is the inability of the spider to defend itself efficiently by the wasp The Lycosa is soon seen. The Calicurgus approaches her without the least sign of fear, walks round her and appears to have the intention of seizing one of her legs. But at that moment the Tarantula rises almost vertically on her four hinder legs, with her four front legs lifted and outspread, ready for the counterstroke. The poison-fangs gape widely; a drop of venom moistens their tips. The very sight of them makes my flesh creep. In this terrible attitude, presenting her powerful thorax and the black velvet of her belly to the enemy, the Spider overawes the Pompilus, who suddenly turns tail and moves away. The Lycosa then closes her bundle of poisoned daggers and resumes her natural pose, standing on her eight legs; but, at the slightest attempt at aggression on the Wasp's part, she resumes her threatening position. She does more: suddenly she leaps and flings herself upon the

Stinging behaviour of solitary wasps

133

Calicurgus; swiftly she clasps her and nibbles at her with her fangs. Without wielding her sting in self-defence, the other disengages herself and merges unscathed from the angry encounter. Several times in succession I witness the attack; and nothing serious ever befalls the Wasp, who swiftly withdraws from the fray and appears to have received no hurt. She resumes her marching and countermarching no less boldly and swiftly than before. Is this Wasp invulnerable, that she thus escapes from the terrible fangs? Evidently not. A real bite would be fatal to her. Big, sturdily built Acridians succumb; how is it that she, with her delicate organism, does not! The Spider's daggers, therefore, make no more than an idle feint; their points do not enter the flesh of the tightclasped Wasp. If the strokes were real, I should see bleeding wounds, I should see the fangs close for a moment on the part seized; and with all my attention I cannot detect anything of the kind. Then are the fangs powerless to pierce the Wasp's integuments? Not so. I have seen them penetrate, with a crackling of broken armour, the corselet of the Acridians, which offers a far greater resistance. Once again, whence comes this strange immunity of the Calicurgus held between the legs and assailed by the daggers of the Tarantula? I do not know. Though in mortal peril from the enemy confronting her, the Lycosa threatens her with her fangs and cannot decide to bite, owing to a repugnance which I do not undertake to explain [18-IV: pp. 271-272]. The chapter entitled “Objections and rejoinders” [18-IV: pp. 287-302] was devoted to discuss the “discovery of the surgical methods” in wasp stinging behaviour and to reply to the many criticisms met from the entomological community. It is worthwhile to expose in some detail this discussion, which at great extent maintains its validity even nowadays. Firstly, Fabre examines the thesis of those who state that “the sting...is directed at one point rather than another because that is the only vulnerable point”, i.e. “accessible..., penetrable”. Such point may be supported in the case of the weevils and the buprestids preyed by Cerceris and stung behind the prothorax (even if there is a different spot, located under the throat, equally accessible but disregarded by the wasp). But “What are we to say of the Grey Worm and other caterpillars beloved of the Ammophilae? Here are victims accessible...everywhere with the same facility, excepting the top of the head. And of this infinity of points, which are equally penetrable, the Wasp selects ten, always the same, differing in no way from the rest, unless it be by the close proximity of the nerve-centres. What are we to say of the

134

Francesco Andrietti

Cetonia- and Anoxiae- larvae, which are always attacked in the first thoracic segment, after long and painful struggles, when the assailant can sting the grub freely at whatever point she chooses, since it is quite nake and offers no greater resistance to the lancet at one point that at another?” [18-IV: pp. 288-289]. Similar things may be said for the Sphex, whose prey are stubbed three times below the well defended thorax, neglecting the soft and bulky abdomen, and for the Philanthus “who takes no account either of the fissures beneath the abdominal plates or of the wide hiatus behind the corselet but plunges her weapon, at the base of the throat, through a gap of a fraction of a millimetre”. Analogous considerations may be made for the mantis-killer Tachytes and for the spider wasps, since if “there is one point about the Tarantula and the Epeira that is dangerous and difficult to attack it is certainly the mouth which bites with its two poisoned harpoons” [18-III: pp. 289-290]. Then Fabre argues against the supporters of the fact that the wasps’ stings may be directed thereabout at the nerve centres, not necessarily exactly where they are. To reject this possibility, it is necessary to consider the details of the stinging technique. In the case of Scolia interrupta, for example, in the preliminary conflict that precedes the stroke, the adversaries appear as “two rings interlocked not in the same plan but at right angles....Owing to her transversal position, the assailant is now free to aim her weapon in a slightly slanting direction, whether towards the head or towards the thorax, at the same point of entry in the larva’s throat”. However, the choice between the two different directions will not be uninfluential, since in the first case the sting will be driven into the cerebral ganglia what will cause the death of the prey: the way how the Philanthus kills her bee, by a slanting sting from below under the chin. With a slant towards the thorax, instead, the thoracic ganglion will be reached and the paralysis will follow: “A millimetre higher kills; a millimetre lower paralyses. On this tiny deviation the salvation of the Scolia race depends. You need not fear that the operator will make any mistake in this micrometrical performance: her sting always slants towards the thorax, although the opposite inclination is just as practicable and easy. What would be the outcome of a there or thereabouts under these conditions? Very often a corpse, a form of food fatal to the grub” [18-III: pp. 217, 290-292]. Additional support to his statements was found considering the case in which a S. bifasciata committed a slight mistake, a very rare thing: her sting entered a little laterally, about one millimetre, from the right spot between the first two thoracic segments, however always correctly on the median line. This small error of position is large enough to determine the paralysis only of the left side legs, while the others maintain their movement. Soon the paralysis gains even the right legs which become motionless. But, differently from those of the other side, they remain able to contract to any stimulation.

Stinging behaviour of solitary wasps

135

In this situation the wasp egg will be unsafe, easily crushed or detached from its position. However, “it is extremely unusual for the operator, no matter what her prey or her method, to make a slight mistake and sting merely somewhere near the requisite point. I see them all groping with the tip of the abdomen, sometimes seeking persistently, before unsheathing. They thrust only when the point beneath the sting is precisely that at which the wound will produce its full effect. The Two-banded Scolia in particular will struggle with the Cetonia-grub for half an hour at a time to enable herself to drive in the stiletto at the right spot.... The sting, by straying less than a millimetre, would leave the Scolia without progeny” [18-IV: pp. 292-294]. One has already said as the first stroke, which is given in the mouth, given by the pompilid wasp C. sexpunctatus on E. fasciata, causes the paralysis of the fangs but non of the palpi, very near to them. A similar result was observed in the case of the lycosid spider preyed by the spider wasp Cryptocheilus annulatus Fabricius. Fabre considered as more probable that fangs and palpi were innervated by the same ganglion, rather than from two different ones. In such case the different effect of the sting implies that the two thin (“as fine as a hair”) filaments innervating the fangs, and only them, must be injured directly. It is of special importance do not affect the cephalic centres, what will cause death instead of paralysis: “The Calicurgus [= pompilid wasp] has to reach them one after the other, to moisten them with her poison, possibly to transfix them, in any case to operate upon them in a very restricted manner; so that the diffusion of the virus may not involve the adjoining parts. The extreme delicacy of this surgery explains why the weapon remains in the mouth so long; the point of the sting is seeking and eventually finds the tiny fraction of a millimetre where the poison is to act. This is what we learn from the movements of the palpi close to the motionless fangs; they tell us that the Calicurgi are vivisectors of alarming accuracy.” On the contrary, if palpi and fangs were linked to two different ganglia, as it actually happens (palpi have their nervous centre in the subesophageal mass, while cheliceral ganglia are fused with the brain [21: pp. 884, 887]), the difficulty will be only a little less [18-IV: pp. 269, 271, 280, 294-295]. Fabre enumerates three different possibilities used by wasps to induce paralysis in her prey, which he is not able to discriminate: the ganglion directly injured; the venom drop deposed over the ganglion or in its immediate neighbourhood. Even if it is true that the effect of the venom may diffuse, as it has been mentioned above, the process takes some time, while the wasp egg requires absolute safety from the very first [18-IV: pp. 295-296].

136

Francesco Andrietti

A further example regarding the wasps stinging method concerns A. hirsuta: “the three thrusts in the caterpillar's thorax and especially the last, between the first and second pair of legs, are more prolonged than the stabs distributed among the abdominal ganglia. Everything justifies us in believing that, for these decisive inoculations, the sting seeks out the corresponding ganglion and acts only when it finds it under its point. On the abdomen this peculiar insistence ceases; the sting passes swiftly from one segment to another. For these segments, which are less dangerous, the Ammophila perhaps relies on the diffusion of her venom; in any case, the injections, though hastily administered, do not diverge from a close vicinity of the ganglia, for their field of action is very limited, as is proved by the number of inoculations necessary to induce complete torpor...”. The last point is proved by a caterpillar which has been removed by the Ammophila after it has received its first sting in the third thoracic segment. Only the legs of this segment remain paralysed, while the other maintain their usual mobility for two weeks, during which the caterpillar performs a normal life, walking, burrowing and retaining perfect liberty of action until it dies accidentally, not of its wound: during this time the effect of the venom is remained confined to the stung segment [18-IV: pp. 296-297].

Charles Ferton (1856-1921) Large part of the entomological community did not agree with Fabre’s opinions. The innate knowledge by part of the hunting wasp of the precise nervous structure of her prey seemed too difficult to be accepted. Among the many opponents was Charles Ferton who, however, maintained always in great consideration the opinions of “the Great Fabre” and always tried to conciliate, as far as possible, his own observations with those of the other, of whom he even wrote a short bibliography [29: Appendix]. The same may not be said for the angry intolerance of other entomologists as Rabaud and Picard who, in their preface to a collection of Ferton’s works, spoke of the “legends” spread by Fabre [29: Preface, p. VII]. Sphex. However, Ferton did not agree with the opinions of Fabre regarding the fact that the wasp stings should reach the thoracic ganglia of the prey, at least in some cases as that of the sphecid wasp Sphex subfuscatus Dahlbom (now Prionyx subfuscatus (Dahlbom)), which preys the orthopterous Calliptamus italicus (Linnaeus). Differently from Fabre’s description of Sphex flavipennis, Ferton did not report the “three strokes of a dagger” but only two: the first through the membrane which surrounds the anterior legs; the second at the origin of the middle or, more frequently, posterior legs. No attempt was observed, by part of the wasp, to wide the

Stinging behaviour of solitary wasps

137

articulation between the head and the first thoracic segment and to maintain the prey motionless (see above). But two other more important differences are to be taken into account. Ferton thought that the possible positions of the stings were necessarily restricted to a few points, where they had effectively been observed, since the sternum of the prey is protected from a chitinous plate which prevents any other possibility. However, Ferton did not attribute these differences to a mistake in Fabre’s observations, but only to an interspecific diversity between the two Sphex species. Moreover, it appeared improbable that the second sting could reach the thoracic posterior ganglion and it seemed more plausible that the paralysis of the rear legs was caused by the diffusion of the venom rather than by a direct lesion due to the sting of the nervous centre. After paralysis, the wasp licked the liquid flowing out from the prey’s mouth [30]. For what P. occitanicus is concerned, Ferton observed the wasp carrying to her nest a male of Tettigonia viridissima Linnaeus, contrary to Fabre who observed only females of Ephippiger, and was unable to make the wasp accept, through an experiment of substitution, a prey of the other sex. The disagreement is explained by Ferton assuming that the females prey are destinated to females new born wasps; male prey, instead, which have a smaller size, to males. Substitution may not be accepted by a wasp that has “decided” to lay an egg which will give rise to a female and for which a larger cell had been build [18-I: pp. 172-173; 31, 29: p. 120]. Ferton noted that the prey of P. occitanicus, as well as those of other Orthopteroids hunter wasps like S. maxillosus or Notogonia pompiliformis Panzer (the prey of the last species appeared to be very imperfectly paralyzed), often show mutilation of some legs. This possibly could be due to a process of autotomy, maybe consecutive to bites of the wasp [23, 32, 31, 29: note 1 of the eds., p. 121]. Spider wasps. In the case of spider wasps, the data of Ferton do not always match those of Fabre. He too observed “the spider to show, in regard to the assailant, an instinctive fear, at the most tossing about to take to flight” [33, 34]. In case of spiders leaving inside a shelter, they are in general dislodged before being attacked. This had already been observed by Fabre in the case of C. annulatus chasing a tarantula ...the Calicurgus, without the least fear, descends into the Tarantula's den and dislodges her. I imagine that things happen in the same fashion outside my cages. When expelled from her dwelling, the Spider is more timid and more vulnerable to attack. Moreover, while hampered by a narrow shaft, the operator would not wield her lancet with the precision called for by her designs. The bold irruption shows us once again, more plainly than the tussles on my

138

Francesco Andrietti

table, the Lycosa's reluctance to sink her fangs into her enemy's body. When the two are face to face at the bottom of the lair, then or never is the moment to have it out with the foe... Quick, my poor Lycosa, quick, a bite; and it's all up with your persecutor! But you refrain, I know not why, and your reluctance is the saving of the rash invader [18-IV: pp. 275-276]. Something similar had been observed a few years before, in the case of a spider of the genus Segestria chased by a Pompilus apicalis Van der Linden, which dragged the prey out from its silk tube before stinging it [18-II: pp. 230]. However, Ferton observed that in some cases the spiders were attacked inside their own shelter or in another where they were hidden [33]. A very interesting case is that showed by Pompilus vagans Costa (=Anospilus orbitalis (Costa)) which hunts the trap-door spider Nemesia in two different ways. Early in the season (in Corse), when the prey uses a single channel nest, the wasp enters inside, gives a light sting to the spider which induces a very short paralysis and lays its egg upon it. Later on in the season, when the Nemesia has added a second door to the nest which opens in a secondary channel, the wasp tactics is completely different: she enters the nest to dislodge the spider, which escapes from one or the other door but is soon reached and stung [34]. However, if in general the spider is powerless with regard to the pompilid which is its usual enemy, this may not be true with different spider wasps, as it was observed by Ferton in an experimental situation where a Gnaphosa alacris E. Simon was collocated close to a P. vagans: the wasp was attacked and killed. Instead, its usual prey Nemesia, even if much larger than the wasp, seems unable to fight with her [35]. For what concerns the stinging pattern, Ferton observed a wide range of situations: from random and ineffective stinging anywhere on the body to stinging in precise positions, as illustrated by Fabre for C. sexpunctatus (and suspected for C. annulatus), in Priocnemis affinis [33]. In addition, Pompilus effodiens Ferton, layd her egg on a Lycosid spider inside its own shelter, possibly without giving any sting to it [33, 34]. However, it is possible that when imprecise and random stings were observed, this was due to the impossibility to reach the right position. In fact, Priocnemis leucocelius Costa, at the beginning stung her Nemesia randomly and without any effect everywhere but eventually succeded to paralyze it giving her stroke in a point close to the mouth. A similar observation concerns a Pompilus republicanus Kohl which had firstly stung her prey, fallen on its back, on the dorsal side, without effect; subsequently in the right spot close to the mouth. This should be the â&#x20AC;&#x153;weak pointâ&#x20AC;? of any spider, as already illustrated by Fabre for

Stinging behaviour of solitary wasps

139

C. sexpunctatus and C. annulatus) (see above), and which most pompilids should try to reach. The effective result greatly depends on the spiders behaviour: some of them, when overtaken by the wasp, try to escape in any possible way, making it difficult to sting them with the due precision; others, instead, remain motionless, making the job easier. Ferton agrees with Fabre even for the presence of secondary strokes given more often under the thorax, after the first main stroke near the mouth [34, 36]. An interesting point regards the possibility, for pompilid wasps, to use the tip of the abdomen as a receptor organ. In some cases, as in Pompilus wesmaeli Thomson, Ferton observed the tip of the abdomen to explore without stopping the body of the preyed spider, possibly without unsheathing the sting, until it eventually reached the right position for the stroke, close to the mouth. And A. orbitalis, when has discovered the nest of her prey Nemesia, frequently inserts her abdomen in the nest opening, probably either to induce the spider to dislodge or to better localize it [36]. However, in some cases the stinging pattern is different, as it was seen in the case of an A. orbitalis which was trying to open a Nemesia single-channel nest. Offering to the wasp an already paralyzed Nemesia, she accepted the substitution giving to the spider, which was maintained with its axis perpendicular to that of the wasp, a first sting in the middle of the thorax. A second, and sometimes a third, stroke was given in the anterior and posterior part of the thorax, on a side. After the stings, the wasp performed malaxation of legs and thorax, close to the attachment with the abdomen [35]. A previous observer (Goureau, 1839, quoted by [33]) had already observed that some pompilid wasps cut the legs of their prey. Cutting or biting the legs of the spider, probably to make easier the transport either amputating or dislocating a few limbs, was confirmed by Ferton [33, 34, 23]. A special kind of mutilation was observed in Pompilus sexmaculatus Spinola (probably Aporinellus s. sexmaculatus (Spinola)), which depilates a small surface of the spider tegument where to attach the egg [37]. According to Ferton, the effect of the stings was very variable, and the prey may remain paralyzed or recover either slower or shortly after the closing of the nest. In some cases, after a few hours the spider was completely re-established, able to feed and to weave its web. This means that the wasp larvae may devour whether paralyzed or active spiders, which are prevented to escape only because confined inside a narrow cell. Among the spiders collected from 29 pompilid species, eight spider species were found re-established [34, 36]. The different effects may be due whether to the kind of sting or maybe the different quantity of inoculated venom: Nemesia badia Ausserer stung by A. orbitalis remains paralyzed for a short period; by Priocnemis leucocoelius Costa, instead, it recovers very unfrequently. But it

140

Francesco Andrietti

may depend even on the species of spider: the Lycosa, for example, seem to recover more easily than the other species [36]. Bembix. Among the many works devoted to sphecid wasps, one is exclusively concerned wih the flies hunters of the genus Bembix [38]. Fabre had already revealed a very interesting fact, quite unusual for the solitary wasps. It consisted in what will be later called “progressive provisioning”, to be distinguished from the fairly more common “mass provisioning” [18-I: pp. 264-267], a habit which the genus shares with social wasps. A part from this main result, which will be confirmed by all subsequent researches, Fabre stated that the diptera preyed by Bembix were killed by the wasp. Moreover, in a few cases the victim showed the head turned of 180 degrees around the neck, the abdomen open by a stroke of the mandibles, the wings rubbed, indicating an uncoordinated and very rapid struggle with a much larger and stronger enemy [18-I: pp. 278, 280]. Even if Fabre does not tell us the species of Bembix which was effectively concerned, this last finding results quite anomalous with respect to those of other observers. If Wesenberg-Lund (quoted by [38]) confirms, for Bembix rostrata (Gmelin), the fact that the food supply consists of killed and crushed flies, most other researchers maintained a different opinion, starting from Lepeletier who studied the same species (1841, quoted by [38]). Also Ferton found that, either in the case of B. rostrata or Bembix oculata Panzer, the largest number of hunted prey remained alive. Moreover, he never observed the brutalities on their bodies reported by Fabre or any indication that they were seized hastily and in a random fashion [38]. Later researches will not confirm Fabre’s and Wesenberg-Lund’s findings, since most flies were found paralyzed, only occasionally killed or crushed [39: pp. 112-113; 40: p. 13; 41: p. 356]. However, the paralysis increases gradually and in a few days kills the prey: “There can be no doubt that the flies do not remain in a fresh condition as long as do the prey of many other digger wasps” [40: p. 13]. What was already established by Fabre, who observed that flies withdrawn from the Bembix, differently from those taken from other wasps, soon became mouldy and rotted. However, this problem is overcome thanks to the progressive method of provisioning [18-I, pp. 275-277]. On the contrary, sometimes one has observed some flies to escape while the wasp was opening her nest or immediately after having been introduced [39: p. 111; 40: p. 73; 42]. In some cases (Bembix near capensis Lepeletier) this result is far from being rare, involving about the 30% of the hunted prey [42] and may not attributed to the breakdown of the venom stock, as supposed by Nielsen [39: p. 111], since it does not result correlated to the frequency of provisioning trips. The state of the prey may therefore result very variable: they may be killed, either immediately (as it happens for

Stinging behaviour of solitary wasps

141

Bembix texana Cresson or Bembix nubilipennis Cresson [41: p. 356; 43: p. 23, 32]) or, more frequently, after a short period. But it may even happen that the induced paralysis is very light (as in B. belfragei Cresson and B. hinei J. Parker [41: p. 356]) or completely absent. However, one may not exclude a possible modulation according to the environmental conditions or the different use of the prey, as stated by Tsuneki (1956, 1958, quoted by [41: p. 356]) who found, in B. niponica F. Smith, that the first prey (that used as pedestal for the egg) was killed, the subsequent others paralyzed. A different explanation (for B. near capensis) could be that of a “volutary” lack of stinging, in order to let escape the fly and so to mislead a satellite parasite present in the site [42]. According to Ferton, probably the wasp stings her prey in flight [38]; some species (as B. handlirschi) have developed particularly short wings, what make easier for them to follow the evolutions of flies flights [37]. However, it is very difficult to study the stinging method in a natural condition. What can be done is to induce a re-sting of the prey that has been withdrawn and then given back to the wasp. In such case B. rostrata may take off again to re-sting the fly in the air or may do the job on the ground or on a branch close to the nest. In the last two cases Ferton observed the wasp to seize the prey anteriorly, in a way to maintain it below and perpendicular to her own body. Then she bended her abdomen below the fly and stung it slowly and repeatedly just a little behind the mouth [38]. More recent findings regarding whether the position of the sting - close to the mouth or in the thorax [39: pp. 102-104; 40: pp. 12, 73] - or the fact that the in natural condition they capture flies either in flight or at rest [40: p. 12] - confirm at a large extent the observations of Ferton. Fabre maintained to have observed the Bembix to chew the head and the thorax of the prey, in order to give the “finishing stroke” needed to kill them [18-I: pp. 280-281]. Ferton disagreed even on this point, since he never observed anything like this. He reports sometimes malaxation, in order to feed on the prey internal juices. But in these cases the fly was brought into the nest only if still intact and living, otherwise it was abandoned [38]. Moreover, according to Ferton the quickness of the attack may not justify any imprecision in the stroke and the subsequent paralysis, as stated by Fabre [18-I: p. 280] and other authors, since in some cases, as in that of A. orbitalis, an extremely quick stroke induces a complete paralysis of the prey [38]. The fact that the Bembix prey remain fresh for a shorter period should be not due to an imperfect way of stinging, but to some reasons inherent to the flies themselves, maybe because they become dry more rapidly. This could explain why the large flies live more than the small ones [38].

142

Francesco Andrietti

Orthopteroids hunters. Ferton was very involved in studies regarding Orthopteroids hunters. For what regards Stizus faciatus (Fabricius), it is only said that the prey is stung below the thorax [44]; for Tachysphex mediterraneus Kohl, instead, a more precise description of the stinging pattern is given: a first sting between the two last pairs of legs; a second between the legs of the first pair; a third close to the throat [35]. Remark the difference from the description reported above from Fabre for the mantiskilling Tachytes. In Dolichurus haemorrhous A. Costa, a blattids hunter, the observed effect of the sting was very light, only able to reduce a part of the force of the prey which maintained its capability to run and to jump. Its vitality remained intact the first days after eclosion and the wasp larva ate a living and moving prey that, however, was unable to get rid of it [45]. Among the orthoptera-hunter wasps the most interesting observations concern the sphecid Sphex, already been long studied by Fabre. The method of stinging employed by the Sphex (=Prionyx) subfuscatus Dahlbom, restricted to only two strokes in the thorax of a female of Calliptamus italicus (Linnaeus), results more simple than that described by Fabre for S. flavipennis. One does not observe any stroke in the neck preceded by a widening of its articulation; neither any preliminary immobilization of the prey hindlegs. In the present case, the wasp rushed into the prey giving a first sting across the membrane which surrounds the anterior legs; then her abdomen moved to give a second and last sting at the origin of the second pairs of legs or, more commonly, at that of one of the legs of the third pair [30]. What is more important is the observation that “In spite of the great length of her sting, the Sphex probably does not reach with the second stroke the nervous centre of the two hind legs: it is by mean of the venom which it infuses, rather than by the lesion itself, that the legs paralysis is realized” [30]. Moreover, the position of the sting is determined by the fact that “the only vulnerable points are the soft membranes which surround the two anterior legs, and those of the articulations of the 4 posterior ones” [30]. The choice of the right target where to sting should not be attributed to an inborn knowledge of the internal anatomy of the prey, as supposed by Fabre [18-I: pp. 111-112], but to mere necessity. However, Ferton was far to doubt of Fabre’s findings, and ascribed the observed differences to interspecific diversity [30]. After having paralyzed her prey, S. subfuscatus female licks the liquids vomited from it. This operation seems to be useful either to the better conservation of the prey or to the feeding of the wasp [30]. Oxybelus. Offering an already paralyzed prey to an Oxybelus melancholicus Chevrier (= O. haemorrhoidalis Olivier) which is leaving her nest after having provisioned it, she hangs the new prey with her mandibles,

Stinging behaviour of solitary wasps

143

curves the abdomen and stings the fly below the thorax close to the throat. Then she straightens her body and without drawing out the sting carries in flight the fly impaled on it, keeping it with the hind legs. Probably the single sting in the thorax is the only one received by the prey. More than half of the flies collected from the wasp nest gave signs of life when stimulated. [23]. Eumenine wasps. Fabre had discovered that both Odynerus and Eumenes possess a “hanging egg” [18-II: p. 94]. He claimed to have obtained this result, confirmed by his own observation, by mean of a deductive way of reasoning. The paralysis was very incomplete, whether for the caterpillar or the larve beetles preyed by Odynerus or Eumenes: from a nest of E. amedei Lepeletier were extracted caterpillars half transformed in chrysalides. The wasp egg would have very easily been destroyed in the middle of the actively moving caterpillars which completely fill the cell [18-II: pp. 73, 76, 77, 82, 92]. Unless As I foresaw by my process of reasoning, the egg is slung from the ceiling of the cell. A very short thread fastens it to the top wall and lets it hang free in space. The first time that I saw this egg, quivering at the end of its thread at the least jerk and confirming by its oscillations the correctness of my theoretical views [18-II: pp. 93-94]. Re-examining the problem, Ferton confirmed the high vitality and mobility of the prey of the vespid wasps Odynerus [24, 23], Pterocheilus [31], Eumenes [23, 31]. But he could not agree with the Fabre’s opinion concerning the reason of the “hanging egg”. Firstly, because in six observations regarding one of the same species considered by Fabre, E. pomiformis Fabricius, only one case showed an egg maintained vertical; but it had been pushed against the wall of the cell by the mass of the stored prey. In the other five cases, the egg lay horizontal, resting above the prey layer and in contact with the cell ceiling. In all cases the egg was attached to its thread. Ferton reported that the different situations did not influence the egg eclosion and the initial feeding of the larva, not differently from what resulted in the case of an egg hanging freely in a cell from which part of the caterpillars were removed. The same result was obtained when the egg, always resting on the prey layer, was detached by its thread [30]. Similar findings were obtained for Odynerus and Pterocheilus: in both cases Ferton, differently from Fabre, was able to carry the content of the cell egg and prey – thrown without any caution in a box where the wasp larva regularly developed [18-II, pp. 94-95; 23, 31]. Moreover, the egg of Eumenes is not as frail as supposed by Fabre. At the contrary, one may manage it without many cautions, to let it fall or to

144

Francesco Andrietti

carry it detached by its thread without, in general, any damage [18-II, pp. 77; 30]. The fragility was an important reason invoked to explain the presence of a “hanging egg”, which otherwise would have been crushed or destroyed by the prey. The reason of the “hanging egg” should be found elsewhere: to escape humidity or, maybe, parasites of the wasp egg [23]. However, the incomplete paralysis could have an adaptative value and not merely be due to an imperfection of the stinging method. In fact, Ferton observed that in E. pomiformis the mobility of the caterpillars may be of help for introducing them in an already crowded and full cell [30].

George Williams (1845-1914) and Elizabeth Maria (1854-1940) Peckam One will review here a few findings concerning the stinging behaviour of North American solitary wasps which were revealed by the Peckam’s. Spider-wasps (Pompilidae) – Pompilus. The Peckam’s were unable to determine the method of stinging. The experiments of substitution of the prey failed, while in natural conditions the movements were too rapid to be observed. All they saw was a violent struggle with “both the combatants rolling over and over upon the ground”. After an instant it was over, the spider lying motionless on the ground. Before beginning the transport, the wasp checked her prey: “With the utmost circumspection she settled down upon the spider and made a prolonged and careful examination of the mouth parts” [20: pp. 131-132, 149-150]. However, the conditions of the prey stung by Pompilus quinquenotatus Say is very variable: they may be killed or very deeply paralyzed or alive and survive from a few days to more than forty [20: pp. 129, 136-137, 143]. This contradicts Fabre’s statements Fabre bases his very strongest arguments for the exactness of the method by which wasps sting their victims, upon the action of one of the Pompilidae, but certainly P. quinquenotatus can make no claim to nice workimanship, for if she occasionally stings in such a way that life is preserved for some time it seems to be a matter of chance rather than of skill. In one respect, however,...she is very successful...Even in the case of the spider that lived forty days the power of motion did not return, to any extent, during the first ten or twelve days, and before this time in the natural course of events, there would have been nothing left to move [20: p. 138]. Spiders collected by the Peckams’s from Pompilus biguttatus Fabricius confirm the wide range of possible situations.

Stinging behaviour of solitary wasps

145

Of three spiders stung by bigutattus, one was dead when taken from the wasp. The second, taken on July sixth, seemed to be quite dead until the eighth, when it gave a slight response to stimulation. From this time it improved, at first slowly and then rapidly, until on July fifteenth it drank water and moved all its legs without stimulation. On the eighteenth it began to walk, and by sixth of September it had entirely regained its health and was released. The third spiders, which was taken with the egg upon it, lived until it was destroyed by the larva [20: p. 140]. For other species one observed similar situations: spiders either dead or living for a few days (Pompilus calipterus Say); dead or alive from one to twenty days, but even a dead spider in sufficient good conditions to be used as food for eleven days (Pompilus marginatus Say); one dead spider (Pompilus interruptus Say). [20: pp. 144, 145-147, 149, 152, 153]. A big Lycosid spider captured by a Pompilus scelestus Cresson appeared immobile and dead after being stung, but after a few hours some quivering of tarsi, when stimulated, began to appear and one day later the prey was almost recovered. A few days later it fed on flies, was able to run and showed co-ordinated control of its movements. However it remained more sluggish than usual and permanently blind. A second case gave similar results: at first the spider taken from the nest was motionless and did not give any response to stimulation, but within twenty four hours the effect of the venom was at great extent passed off and in the third day the spider recovered its health and was released. According to these findings, the Peckamâ&#x20AC;&#x2122;s stated that the cephalotorax ganglionar mass should have been affected by the poison whose diffusion in the whole ganglion explained the complete paralysis. The main damage should not be structural, at exception of the region directly pierced by the sting that, in the first case, should have been responsible for permanent lesions at the optical centres. In fact, pushing a fine needle (six or seven times as thick as the sting of the wasp) through the ventral ganglion of a spider, one obtains a result which may not be comparable with that of the sting: a partial paralysis for a period of six days, ensued by the death [20: pp. 153, 160-163]. In some cases the Peckamâ&#x20AC;&#x2122;s observed the wasp biting the spider legs, which often appear to be cut off. For Pompilus fuscipennis St. Fargeau, this is a common habit, even if the behaviour is very discontinuous: four out of ten spiders had all their legs, the others being deprived of one or two of them. Biting at the spider legs is made at intervals, starting from the moment when the prey is captured, but becomes particularly frequent at the moment to introduce it into the nest. This operation is possibly related to the narrowness of the pompilid nest and cell which makes necessary for the spider â&#x20AC;&#x153;to be

146

Francesco Andrietti

shifted and turned and tugged”. It may not be necessary to remove the legs, but only to render the limbs more flexible, making it possible to drag the spider down. In fact, the legs of intact spiders were sometimes squeezed, showing the marks of the wasp mandibles [20: pp. 140-141, 143-144, 164]. Even Pompilus scelestus Cresson was observed biting and squeezing the spider legs [20: pp. 154, 155, 158]. During one of the frequent pauses made by the wasp while she was dragging her spider She seemed to be biting the legs, near the body, beginning with an anterior leg on one side and working backward and then repeating the operation on the other side. She went through this squeezing process again and again, and to us it looked as though she might be trying to force back the juices from the legs into the body preparatory to cutting them off, but after a time she would seize her prey and start on again [20: p. 154]. The high mobility frequently showed by the stored spiders after their recovery, should be one of the reasons of the narrowness of the cell: “The prey is buried alive in the fullest sense of the term but is wedged in so tightly that not the slightest movement is possible, and thus the egg is protected” (Peckam & Peckam, 1898, p. 163) [20: p. 163]. Spider legs amputation (see also above) was observed by many other researchers. Recently, Andrietti et al. [46] have confirmed the high variability of this habit in Anoplius infuscatus (Vander Linden), which may be alternatively present or absent either in individuals of the same species or in single individuals. In the same work is reported the high mobility of the stung spiders and the fact that in some cases one observes a certain difficulty in the introduction of the spider into the nest, what could justify the reason why legs may be cut. On the contrary in Episyron sp., where leg cutting is quite rare, no obstruction in spider introduction into the nest was ever observed [46]. Apart from Pompilus, the Peckam’s observed more briefly other spiderwasps genus obtaining similar results, either for what regards legs amputation (Agenia) or the condition of the prey: dead (Agenia, Salius) or alive and in general recovering from paralysis in a few days (Aporus) [20: pp. 54-57, 164-166]. In the case of Salius conicus Say, it was even possible to observe the method of stinging, consisting in seizing the spider by its head, to bend the abdomen around and below the prey and to sting in the middle of the ventral side of the cephalothorax. The spider collapsed almost immediately and the wasp repeated the operation a second time in the same manner. In a second observation, the spider was stung to one side of the middle of the ventral side of the thorax [20: pp. 53-54].

Stinging behaviour of solitary wasps

147

Astata. Is a genus of bug-hunters. The method of stinging of Astata bicolor Say consists in seizing the bug head in her mandibles, holding it dorsum up, and to curve her abdomen to sting it below the thorax, presumably in the thoracic ganglia. Malaxation is common, and in some cases it could substitute stinging since some bugs appear quite alive. But most of them appear to be dead or will die in a few days, and wasp larvae feed mainly on dead prey. Similarly quite dead were the bugs taken from Astata unicolor Say [20: pp. 92, 95-97]. Instead an European species, Astata picea A. Costa, according to Ferton stings her prey in the throat, and then performs malaxation to feed on the prey juices or maybe to complete the effect of the sting [23]. According to more recent findings, Astata prey are completely paralyzed [47: p. 212]. Cerceris. Cerceris clypeata Dahlborn hunts beetles (weevils) which stings a first time under the neck, and then behind the first pair of legs. The prey is picked up again and stung in the same manner four times in sequence, with intervals of five or six minutes between. For what concerns the condition of the stored beetles, most of them appear to be dead. The same may be said for the beetles hunted by C. deserta Say and C. nigrescens Smith, which mostly appear to be dead or quite dead [20: pp. 114, 116-117]. Sceliphron. To determine the stings pattern of the maud-dauber wasp Sceliphron, a hunting spider sphecid wasp, is quite hard. In his chapter entitled “Les Vivres du Pélopée” (“The provisions of Pelopaeus (= Sceliphron)” [18-IV: Ch. 2], Fabre clearly admitted the difficulties he met to discover the hunting method of Sceliphron. He could only observe …the wasp swoop down upon a spider, clasp it, and carry it away, almost without pausing in her flight. Other hunters alight on the ground, make their fastidious preparations sedately, and distribute the strokes of the sting with the calm deliberation which a delicate operation demands. This one darts down, seizes her victim and departs, something after the manner of Bembex. So rapid is the abduction that it may be presumed that the mandibles and sting are only used while the wasp is on the wing. This impetuous method, incompatible with learned surgery, explains to us, even better than the narrowness of her cells, the predilection of Paelopeus for small spiders. A larger and more powerful victim may put the wasp in danger. The faultiness of her art makes a weak victim necessary to her. We must suspect that a spider so hastily and carelessly taken, may be killed. As a matter of fact careful and repeated scrutiny of these cells, when the egg was not yet hatched, confirms these suspicions. There is never any trembling either of palpi or tarsi, and in about ten

148

Francesco Andrietti

days they decompose. This then is what is stored in the cells of Pelopaeus, spiders dead or nearly dead [18-IV: pp. 27-28] (free translation from [20: pp. 190-191]). A few years later Peckham & Peckham made intensive researches in this same topic. After many failures in the attempt to reproduce experimentally controlled situations [20: p. 181], they partially succeeded in observing direct attacks on the spider Epeira strix Hentz (= Larinioides cornutus (Clerck)) by part of the American species Pelopaeus coeruleus (= Chalybion caeruleum Linnaeus ?). In most cases, at the arrival of the wasp spiders were dislodged and dropped. Sometimes they were followed and were caught on the floor but more often the wasp left them escape continuing her searching on the wall, eventually succeeding in catching a victim. It appeared that the stinging was made in two steps: a first time, after being seized by the jaws and first legs of the wasp, the spider was stung under the abdomen or more commonly at the cephalotorax, underneath or at one side. The general impression was that this first sting was given at random anywhere, in order to induce a temporary quiet in the prey and to allow to perform the second step. It consisted in alighting upon a nearby object and to give a second sting, “either resting quietly or rolling the spider around and around” [20: pp. 182-183]. On a large sample, made mainly of P. coeruleus, but even of Sceliphron cementarium (Drury), which is a species now present in Europe too, the Peckham’s found that the 33 % of the spider were paralysed, the remainders being killed. The number of paralysed spiders was higher of that found by Fabre in France. Comparing their own data with those of different authors and related to different mud-dauber wasps, the Peckham’s found a variety of situations: some species killed almost all spiders, other more than half of them, others left them alive but in a helpless situation [20: pp. 176, 186-187]. Contrary to what happens for other ravishers like the spider wasps, the stinging method used by Sceliphron appears primitive and imprecise. Is the learned art of paralyzing practiced by Calicurgus upon the tarantula, which keeps it fresh for seven weeks, unknown here? Have we here to deal, not with a delicate operators who knows how to abolish movement without destroying life, but with a brutal worker who kills for the sake of rendering the victim immovable? Both the withered aspect and the rapid deterioration of the victims bear witness that this is true [18-IV: p. 28] (free translation from [20: pp. 191]). The way of consuming the prey is made in accord to this situation, namely the shortness of the period of Sceliphron larval life and the use of a large

Stinging behaviour of solitary wasps

149

number of relatively small spiders, which are devoured rapidly one after the other before decomposition can progress. On the contrary, as stated by Fabre, when a large prey is utilized, it is necessary to maintain them alive “and a special art must also be observed to eat it”, otherwise early decomposition will advance rapidly in the half eaten and disorganized body mass [18-IV: p. 28-29]. In any case Our study of the eating habits of these larvae has led us to the conclusion that they are not in the least fastidious as to whether the food is hard or soft, fresh or dry....On several occasions when playing nurse-mother to a number of growing larvae, which we kept in little glass saucers, where we had not provided a large enough food supply we made good the deficiency by adding a number of dead and dry spiders that we had had on hand for some three weeks [20: p. 189]. From a sample of 11 cells (Paelopeus coeruleus + S. caementarium), the Peckham’s found 75 spider killed and 84 alive, which mostly died within the first seven days from the moment when they were stored. Since dead spiders, even if drying up, remain in good condition for 10-14 days [20: pp. 186, 195-196], one sees that for the larvae there is not a big difference in the use of dead or living prey. The Peckham’s performed a certain number of experiments in order to determine the effect of the wasp venom. By mean of a S. caementarium, which was induced to sting a few spiders at the spinnerets, a point far from the central nervous system, they obtained the following results The spider was paralyzed at once...but in five minutes it recovered somewhat and was able to stagger about. This continued for ten minutes when the wasp was allowed to sting it again in the same place with the same immediate results as before but this time there was no recovery and in thirty minutes the spider was dead....All of these spiders and many others were killed by the general diffusion of the poison through the system and not by any wounding of a ganglion. When we allowed a considerable amount of venom to be injected death followed almost immediately but if a smaller amount was used the spider lived a day or more [20: pp. 194-195]. Trypoxylon. In Trypoxylon albopilosum W. Fox (now T. striatum Provancher) the stung spiders resulted either be killed at the moment of the capture or, if only paralyzed, they died in the nest from day to day. In

150

Francesco Andrietti

Trypoxylon rubrocinctum Packard (now T. collinum F. Smith ssp. rubrocinctum Packard) stung spiders were alive or, at a slightly minor extent, dead; among the living ones, none lived more than fifteen days, less than the best paralyzed prey of Sceliphron. Apparently, dead or alive spiders were equally usable by the wasp larva. A spider capture was observed in Trypoxylon bidentatum W. Fox: the wasp seized the spider by the top of the cephalothorax, curved her abdomen and stung it in the ventral side of the cephalothorax. The Peckam’s observed that, due to the large extension of the ventral ganglionic mass, “a thrust given in almost any part of ventral face of the cephalothorax, or even on either side of the anterior part of its edges, would reach the nervous center” [20: pp. 77, 78, 81-83, 85, 86]. Diodontus (D. americanus Packard). Aphid hunters. According to the Peckam’s there is no sting, the prey is only picked up and carried off. We found that when a wasp secured an aphis she flew with it to another leaf near by, alighting, this time, on the upper surface. She then passed it back from her mandibles to the second pair of legs, and holding it, with them, under the body, she proceeded to make use of the first pair in giving herself a through cleaning [20: p. 101]. The wasp, bringing the aphid forward, squeezes its neck delicately but repeatedly between her mandibles. In most cases the aphides were killed, but in other cases the operation was so light that they so little injured to be able to walk around as soon as they were released. However, only one, out from forty aphides drawn from nest, were alive [20: pp. 101, 105-106]. The whole process may be observed more conveniently, as many times as one wants, placing the wasp in a bottle together with a leaf upon which aphides are present [20: p. 101]. The tiny wasp would pounce upon an aphis, and holding it with the first legs would squeeze its neck gently between the mandibles, rolling it over and over. After a few moments she would pass the aphis back to the second pair of legs and rest for a short time, usually taking this opportunity to wash her face....In the open, Diodontus often alights on one leaf and malaxes her victim and then flies to another and another, repeating the process several time before she finally flies off to her nest [20: p. 102]. Sometimes the captured aphids served not as food for the larvae but for the wasp. In these cases there was no malaxation and the aphis was held in any position, sucked of all its juices and thrown away [20: p. 105].

Stinging behaviour of solitary wasps

151

The Peckam’s acknowledge that Fabre “is unquestionably the most accurate of observers”, but this does not mean that “all of his inferences must be accepted”. In particular they question the point, on which Fabre insisted so many times, regarding the way of stinging that should have been perfectly developed from the origin, remaining otherwise useless for the wasp. In fact the Peckam’s note that, to be true, “there is not a single species in which the sting is given with invariable accuracy...they scarcely sting twice alike”. This observed variability matches that regarding the state of the prey of the same wasp, some of which may result killed at once, others survive from one days to six weeks or eventually recover. This result seems to indicate to the Peckam’s that the primary target of the wasp is not that to paralyze her prey, but only that to subdue it. To this aim different methods may be employed: killing, paralyzing or simply capturing it. The habit to sting in the ganglia simply represents the most advanced degree of this process which has developed, when necessary, to overcome the difficulty to deal with a dangerous prey (as in case of spiders) or the difficulties inherent to its transport (as for grasshoppers). This different view clearly opens the possibility to explain the nowadays situation through a natural selection mechanism that Fabre’s perspective explicitly precluded [20: pp. 102, 222, 225-226]. Even the fact of stinging in the nervous centers is not a matter of experimental finding. It is simply an inference based on the observation that very often the prey are not killed but merely paralyzed, remaining motionless for weeks or even recovering. In fact, according to the Peckam’s, one third of the solitary wasps kill their prey. Moreover, they question the fact that the real target of the wasp should be that to induce paralysis instead of killing: in many cases the period required by the larva to feed is short enough to make it possible the use of either paralyzed or dead prey [20: pp. 222-224]. The “inference” about the anatomical position of the stings could not be resolved in more positive terms at the Peckams’s times and will remain object of experimental investigations for the subsequent century.

Recent findings I: Steiner’s review and Steiner’s work on Liris The works concerning wasps’ stinging behaviour written in the 20th century will be not summarized here, since they have already been reviewed by Steiner [2] until 1986, in a fairly exhaustive way (including a huge amount of bibliography). Instead, one will illustrate only a few specific points mainly related to the work of Steiner himself, whose studies represent the

152

Francesco Andrietti

most detailed and comprehensive contribute to the stinging behaviour of some selected species. Among the Sphecid wasps one may consider a first group of wasps hunting molecrickets (Larra: Larrini), cockroaches (Ampulicinae) and crickets (Liris: Larrini), which are all stung in the throat (together with other possible stings in different parts of the thorax). In the first case one observes a temporary partial paralysis followed by complete recovery (no deactivation); in the second case the paralysis is substituted by a condition of passivity and ineffective resistance which does not hinder the animal to feed (partial deactivation); in the third case, after recovery from paralysis (a few minutes), one observes permanent deactivation, without feeding: “The crickets have been transformed into passive “reflex machines”; they can stand on their feet and jump or walk if prodded but do not try to escape, feed or groom” [2: pp. 115-117, 122]. Cockroach hunters may be found even in the Sphecinae and in the Larrinae: in the tribe Sceliphronini one finds a group of cockroach hunters and even a few Tachysphex species (Larrini) make the same. However, for the prey of these hunters the situation is more variable than before, varying from complete to incomplete paralysis with subsequent recovery. The stings position is unknown [2: pp. 117-118]. Even some cricket-hunters may be found in the Sphecinae. Remember, for example, the case of the Fabre’s “Yellow-winged Sphex” (Sphex flavipennis), whose first sting, as reported above, is in the neck. However, the real cricket specialist is represented by the genus Chlorion, studied by Hingston (1925-1926, quoted by [2: p. 121]). She stings the thorax two to five times, in an irregular fashion which perhaps does not include venom inoculation, then the neck. After this last stroke the prey becomes completely motionless. This sequence of stings, according to the terminology employed by Steiner, should represent a fairly good representation of the C4SP pattern (four stings: a first sting in the neck, followed by a sting in each of the thorax segment), but in a reversed order. A few subsequent abdominal stings were described and a recovered but in some way weakened cricket three minutes later in the burrow [2: pp. 120-122]. The study of Steiner [48] of the stinging behaviour of Liris nigra (Van der Linden) (= Notogonia pompiliformis (Panzer)) is a true masterpiece, which constitutes a landmark in this kind of studies and deserves to be exposed in some detail. It was the result of many years of study on the stinging activity of this wasp made in laboratory on bred animals and of the analysis of thousands of stings [48: pp. 4-5]. A first point concerns the external localisation of the stings: “only the thoracic and cephalotorax regions of the cricket [larvae of Gryllulus

Stinging behaviour of solitary wasps

153

domesticus Linnaeus; L. nigra in general paralyzes immature only, since the adults, although attacked, are too large for this species] shows the marks due to the stings, and only their ventral or latero-ventral parts” [48: pp. 7, 9; 59]. And, more precisely: “All marks of stings are located on the soft and flexible intersegmentary membranes, excluding the chitinuous regions ...crowding, at a certain extent, in the folds, furrows of the soft membranes...or around the different sculptures or tegumental “accidents”” [48: pp. 9-10]. Steiner subdivided the whole “sting region” in twelve “sting areas”, located on the soft thoracic-cephalic membranes. Due to their symmetrical distribution with respect to the longitudinal axis, only six of them will be considered. Within each of these areas the density of stings is not uniform, but concentrated in the vicinity of a “characteristic point” around which the sting marks decrease according to a “centrifugal” gradient [48: pp. 10-11]. The stings dispersion around their characteristic points varies as a consequence of the shape and the accidents of the corresponding sting area: stretching in the direction of an elongated membrane, stopping against cuticular reliefs, following the membrane furrows or extending largely and freely around a region of membrane wide and uniform. It is particularly restricted when the characteristic point is situated in a part of the cricket cuticle better localized with respect to some morphological markings, as for examples sculptures, tubercules a.s.o., which allow the wasp to sting with improved precision. However, it depends even on the prey size, increasing with it and interesting even sectors which are unaffected in the smaller sizes. Two possible explanations have given to this fact. One, larger and more powerful prey may better react to the wasp aggression, reducing the degree of the stinging precision. In fact, stings dispersion decreases with subsequent strokes, since even prey resistance progressively decreases. In addition, with the largest crickets the wasp may experiment some difficulties to reach the median ventral line, so increasing the degree of lateral dispersion [48: pp. 15, 18, 20, 21]. Another possible source of variation should be attributed to individual differences: some particular individual wasps frequently sting the cricket more or less constantly in peripherical sectors which are not normally used [48: p. 21]. The second point regards the direction of the stings which, with a good approximation, is indicated by the orientation of the wasp abdominal extremity in a photographic frame. It varies mainly in the antero-posterior direction, while the fairly constant angle which it makes with the horizontal plane is small and may be disregarded. In each stinging area the sting maintains a direction sensibly constant, with a small dispersion around a value that may be considered as the “characteristic direction” of the area. It is

154

Francesco Andrietti

worthwhile to observe that analogously to what happened for the soft membranes, which were unable to explain the localization of the “characteristic points”, one may not invoke the presence of “postural reflexes” in order to obtain a quite automatical constancy in the direction of penetration of the dart in the stinging areas. Even if they certainly will play an important role in realizing and maintaining it, as a relief or any other accident in the tegument may better delimit the dispersion area around a “characteristic point”. In fact, the abdominal extremity of Liris assumes a direction very close to that characteristic of the area, in spite of the many different positions which the body of the wasp may assume during her fight with the cricket. Moreover, one finds that the sting direction may vary abruptly (still maintaining a bilateral symmetry) moving from one to an other of two close (which should imply similar postural reflexes) “characteristic points”, according to the different “characteristic directions” of the correspondent areas [48: p. 22, 23, 26]. For what concerns the internal localization of the stings, it is necessary to follow the “characteristic directions” starting from the “characteristic points”. One finds that all of them converge on a ganglion: the first two (Ia, Ib), beginning from the rear of the cricket, on the metathoracic one; the two intermediate (IIa, IIb) on the mesathoracic; the fifth (III) on the prothoracic; the sixth (IV) on the subesophageal one. The structure which governs the direction of the stings is therefore that of the four ganglia where they converge, rather than the six “stings areas” and correspondent “characteristic points”, which in fact mask the true four-element sub lying structure. According to this view, one observes only four kind of different paralysing effects according to the different stings: those in the area Ia or Ib, affect only the hind legs pair; those in IIa or IIb, the intermediate pair; that in III, the front legs; that of IV, the buccal parts. In every case the sting results in the immediate immobility of the respective appendices, and only of them. The complete paralysis of the cricket is therefore the result of the addition of elementary paralyses, separated in space and time [48: pp. 26-28]. Owing to the chitinuous shield which prevents the ganglia from being directly reached, it is necessary to sting them laterally. The only reason why the stings Ia and Ib, IIa and IIb converge on the same ganglia may be due to the presence of the bases of the two posterior pairs of legs, which divide the areas where the sting can be inserted [48: pp. 28-29]. For what concerns their temporal sequence, in “normal” conditions the wasp gives four paralysing stings, rarely more, still less frequently a minor number, according to the following ways: •

fundamental formula (Ia + III + IIb + IV) or formula a

Stinging behaviour of solitary wasps

The prey, pursued by the Liris, is boarded usually from behind; the wasp tries to get on the cricket back but, due to its violent opposition, rarely succeeds in doing so and what obtains is only to grasp at it, seizing a leg between her mandibles, in general the hindest one, with the additional help of her legs which may clasp, for example, those of the prey. The cricket, always running away, carries the sphecid behind itself (unless the frequent autotomy of the posterior leg has been able to free it from its persecutor), while the wasp tries to thread her abdominal extremity behind the base of the seized hind leg, on the soft articular membrane or sting area Ia (this position will be called posture Ia...); when she succeeds, she unsheathes her sting giving the first stroke...Immediately the posterior legs of the prey become motionless after a few last tremors...it is this first sting which may be exerted in the most variable positions, depending on the randomness of the fight with the cricket still in possession of its whole vitality; in spite of that, the direction of introduction of the sting, characteristic of the area Ia is almost invariably maintained by the abdominal extremity of the Liris... Notwithstanding the paralysis of the posterior legs, the cricket is still able to move or at least to toss about, using its anterior and intermediate unharmed legs. The sphecid (which has not to face with an opposition as strong as that of the beginning), is now able to rise on the back of her victim and places herself slantwise, at the same time lowering her abdomen toward the latero-ventral side of the cricket which is explored by mean of the abdominal extremity...which eventually stops when reaches a position between the bases of the first and second pairs of legs. Firstly, the abdominal extremity is pointed forward in the direction of the base of the foreleg, which receives a sting from behind (in the sting area III...): the two anterior legs are immediately immobilized, stopping each possible progression of the cricket which may nevertheless to shake its middle legs, still trying in vain to escape; then the abdominal tip of Liris pivots toward the rear of the cricket (while the rest of the abdomen remains in its position between the anterior and middle legs...), i.e. in the direction of the base of middle legs which receives then a sting from forward (on the sting area IIb), abolishing the movements of the middle legs; the cricket is still able to shake its head in every direction, opening and closing the mandibles, but a last sting in the neck in the area IV gets rid of this last possibility of movement....The immobility of the cricket is nearly total (some movements of the antennae, abdomen et cerci may persist) and lasts a few minutes, a time which is in general sufficient to the wasp to lay the egg and to end nidification. After this

155

156

Francesco Andrietti

delay the cricket recovers progressively the use of its different paralysed appendices. It may happen that the wasp skips one of the stings; in such case the prey maintains the use of the correspondent appendices... [48: pp. 32-34]. The fundamental formula includes therefore four stings but only three different postures, since during stings III and IIb the wasp maintains the same position with respect to the cricket body [48: p. 34]. It may be subject to some variations. Sometimes the wasp, as already said, may be able to get on the cricket back (what “typically” happens after the sting Ia, with an already weakened cricket). In such case she assumes a different posture (Ib, IIa): instead of placing her abdominal extremity behind the hind leg, she positions it between the middle and the posterior ones. The first sting is then given in the area Ib, before the posterior leg. However, even so both the posterior legs are paralyzed, due to the convergence of the “characteristic directions” of both stings Ia and Ib on the same ganglion. The remainder of the sting pattern will be the same of the previous one, and the new formula will be: •

•

Ib + III + IIb + IV [48: p. 35]. But still other variations are possible. For example, still starting from posture (Ib, IIa), a second sting is given in IIa, or sometimes simply sketched: Ib + IIa + III + IIb + IV. In such case the sting IIb will not produce any effect, since the middle legs have already been paralysed by IIa. But it may even happen, even if very rarely, that in the last variant the sting IIb is only sketched. Then one obtains again a four-stings formula (instead of the previous 5 stings sequence): Ib + IIa + III + IV [48: pp. 36-37].

In normal conditions holds the rule that every ganglion will receive one and only one sting, with the only exception of the variant of the five-stings pattern: Ib + IIa + III + IIb + IV. A second rule regards the postures, which give rise to a double stings when the wasp abdomen is positioned between two consecutive legs (“two-stings posture”). In the typical case represented by the fundamental formula both rules are respected. Instead, in the variants, one of the rule will be violated. Either that concerning the number of stings received by each ganglion or because a two-stings posture gives rise to one single sting [48: pp. 37, 39]. Until here one has considered a “normal” situation. But some difficulties may arise, which interrupt the “normal” development of the sequence, like

Stinging behaviour of solitary wasps

157

the fight with the prey or the presence of other individuals or mechanical obstacles. In such cases different solutions may be adopted. A.- Liris “skips” one given sting (rarely), passing directly to the subsequent one; then a) once the obstacle is removed, she may go back to the omitted sting (the most frequent case of the two); b) the omitted sting is definitively forgotten. B.- Liris stops (more common case). She does not give the subsequent stings trying obstinately to give the “blocked” sting. a) If the obstacle is removed after a certain delay: aa) Liris takes up again her paralysing activity, starting from the blocked sting; bb) Liris takes up again her paralysing activity, starting from the beginning (more rarely), so that a few sting will be given twice. b) If the obstacle is permanent or the area to be stung is artificially suppressed: aa) the vain efforts to execute the blocked sting vanish after a period which is in general rather long, the sting is omitted and the wasp goes on with her sequence; bb) the wasp abandons the “abnormal” cricket and begins a new complete stinging sequence on a new one; however, the preceding event will leave a trace, since it is rare that the wasp will perform in a typical way the sting that had been blocked: she will “insist” a long time in executing it or will give it many times, either consecutively or returning back to it after subsequent steps in the cycle [48: pp. 40-41]. In some cases, variations in the stinging sequence may be observed independently from external conditions, as at the beginning or end of the reproductive period, due to some internal factors or to undetermined or fortuitous reasons. It is even frequent to observe an inversion in the stings sequence, particularly of stings III and IIb, probably related to the facility to exchange them in the two-stings posture (III, IIb) [48: p. 41]. For what concerns the cricket, its “normal” behaviour in respect of Liris attack consists in a flight reaction. In the case of a different response, especially in the case of absence or weakness of flight or antagonistic reactions, the wasp may respond either with a stings sequence of the “normal” type, or with a paralysing sequence of a special kind (c-type, see above) or even shifting to a different section of her nesting cycle, for example prey transport or malaxation [48: pp. 42-43].

158

Francesco Andrietti

In much less frequent cases the cricket responds in a completely different way with a “freezing reflex”, consisting in a rigid posture in which the abdomen and the hind legs are straightened on high, while the other legs are strictly adherent to the body and the antennae completely folded back. This position is much more apt to avoid the attack of the wasp which in most cases, after a few turns around the cricket, went away without stinging it. Similar and still less frequent, is a specular reflex: the cricket assumes a position with the head, instead of the abdomen, straightened on high. Even in the few cases when this posture was observed, the Liris went away without giving any sting [48: pp. 44-45]. According to Steiner, in the “freezing” posture the cricket weakens the wasp stinging response. However, this effect should not be attributed to the immobility, since motionless crickets (paralyzed crickets or even exuviae) still elicit the stinging response or at least the beginning of it. One may not reduce the stimulus to an elementary external condition, like a smell or a specifical chemical substance, but one should think to a more complex factor, to a “significative stimulating situation”. One must observe that the kind of prey is very specific, since it was impossible to induce the wasp to sting insects others than crickets, even if closely related to them, unless they were imbued with the smell of G. domesticus or covered with its spoils. But, even so, one observed only a few attacks not followed by stings [48: pp. 45, 48]. However, also individual factors may play an important role. A few wasps, for example, seemed to be unable to unable to carry on a correct stinging sequence and inclined to rubber already paralysed cricket from other ones, developing a cleptoparasitism which is probably latent in all Liris, even in those (the most part) which master their paralysing activity [48: p. 46]. Even the size of the cricket is important, since the largest ones were able to repel the wasp’s attack. In this way one obtains an automatic control of the prey dimension, since too large prey are useless. In fact, if one prevents the cricket from defending itself, leaving it to be paralysed, the wasp will not be able to transport it to her nest [48: p. 46]. Steiner explored even the ontogenesis of the sting pattern, observing how it developed at the beginning of the hunting period. He found three forms of incomplete pattern (formulas of the kind b): Ia; Ia + III; Ia + III + IIb. These forms followed each other in this transitory period, eventually leading to the complete formula Ia + III + IIb + IV, matching the parallel increase of the hunting motivation of the wasp and of her sensibility to the presence of the cricket. The lacking stings may sometimes be sketched without unsheathing of the sting. However the stinging sequences, in spite of being shortened, show the same order of the normal one, which is simply interrupted before being completed. The wasp in general abandons the cricket subject to

Stinging behaviour of solitary wasps

159

incomplete stings pattern, not even malaxating the base of the anterior legs it as it is usual at the end of the normal case. In particularly good conditions the transitory period may be very short and difficult to detect or even completely absent, the wasp beginning immediately with the normal sequence [48: pp. 49-53]. If the sequences of the b-type, in spite of being incomplete in the number of stings, maintain the same structures of the a-type (typical), the same is not true for the c-type (atypical formulas): IIb; III; III + IIb; III + IIb + IV. In this case the basic sting, which is never lacking, is IIb (less commonly III), instead of Ia which, at the contrary, is the last to appear when the atypical sequence matches the typical one. The c-sequences appeared when the cricket did not show a “normal” reaction to the wasp, but a decreased one. This in general happened when it had already received the paralysing stings before. In fact the prey, a few minutes after being stung, progressively recovered its capability of moving and jumping. Even if, as already stated before, it will maintain a passive behaviour, able to respond to external stimulation but lacking of spontaneous movements as feeding (deactivation). In any case after a period of about one week the cricket resumed an increasing torpor which will lead it to die [48: pp. 54-56]. When a wasp, during the hunting period (characterized by the fact that she is not yet in possession of a prey) encounters a half recovered cricket (for example previously abandoned by the same or by another wasp), she may avoid to sting, maybe malaxating it if there is no response to the contact (what corresponds to the usual situation in the malaxation). Otherwise, she will begin to assume the posture corresponding to sting Ia; if the cricket does not respond vigorously to this contact, by opposition or by escaping, the wasp will leave this posture passing to (III-IIb), which will lead to sting IIb, that will result the first given to the cricket. Then she will provide subsequent stings in accord with the opposition of the cricket: no other sting, in case of weak reaction; or III + IIb, III + IIb + IV in case of increasing responses, however always lacking of the flight response. Only when this is present the sting Ia will be given, making the c-formula converge to the normal one [48: pp. 58-59, 62, 66]. C-sequences may even be observed in the subsequent phases when the wasp is already in possession of the cricket, as during the transport, when the prey oppose some resistance. This may happen if the prey has been badly paralyzed, or the transport phase is abnormally long or for accidental “exchanges” of crickets in different stages of their recovery period. But even due to factors independent from the cricket, as to obstacles to the transport due to the presence of roots or afterward an attempt of robbery of the prey by part of a different Liris individual. In these cases the wasp will equally make

160

Francesco Andrietti

use of the atypical c-sequences, with the difference that the preliminary Ia posture will be omitted, but with a lower level of response, what greatly increases the proportion of the stinging sequences reduced to only IIb or more rarely III. In other words, the threshold necessary to elicit the stinging response will rise as soon as one moves further away from the more sensible hunting phase. Then, at the closing phase, the wasp becomes completely insensible to the presence of the prey which is treated as mere building material [48: pp. 59-60]. According to Steiner, one may explain the wasp stinging response (which may be generalized to other kinds of responses of her biological cycle) in terms of a reaction to two different kinds of “situations”: that of “opposition to the manipulation of Liris on the cricket” and that of “subtraction of the prey = escape of the cricket”. In fact, in her stinging sequence the wasp does not make any distinction between active or passive movements of the prey. It is equally possible to observe the stinging reaction either in case of escape of the cricket or its accidental loss or when the observers try to take it (even using cricket recently dead). This may explain a fact well known to the observers, who often reported how wasps may be induced to sting their already paralyzed prey in many different circumstances [48: pp. 62-63]. In some cases only the stinging Ia is observed: this happens when the cricket performs only an escape reaction, independently from the opposing reaction, as it may happen when it has been paralyzed only in the first two pairs of legs, maintaining the movement of the posterior ones, the “jumping legs” which may enable it to jump away. To summarize the different possible stinging responses [48: pp. 63-64]: • • •

cricket escape + opposition opposition (only) escape (only)

= a-paralysis = c-paralysis (without Ia sting) = sting Ia (only or sometimes included in a complete sequence).

Trying to investigate the reasons of the different sting sequences, Steiner stressed the importance of the change in the internal factors which govern the different phases of the wasp biological cycle. For example, the Ia posture, which is characteristically at the beginning of the normal (a) stinging pattern, is only sketched or completely absent in the c-sequences (depending if the wasp is still or not within her hunting period) when she begins with the posture III, IIb. But this is even the posture assumed before egg laying, while the stinging areas IIb and III approximately match those of oviposition and malaxation. In particular, the “characteristic point” of the sting area IIb closely corresponds to that where the cephalic head of the egg will be

Stinging behaviour of solitary wasps

161

located. In addition, a non-escaping cricket is more similar, as a stimulus, to a an egg-laying than to a hunting situation. The postures Ia and III-IIb are spontaneously assumed in the two corresponding situation (hunting and egglaying), before any response by part of the cricket. Hence, they are unrelated to any external stimulus. Instead, the shift to the basic posture III-IIb in the c-stinging sequence should be related to a kind of “attraction” by part of the future phases of the biological cycle [48: pp. 67-69; 49]. Reciprocally, the sketched Ia posture in the c-sequences observed during the hunting phase, finds its explanation in the effect, still present at a certain extent, of the normal sequence scheme competing with the contemporary effect of the future oviposition phase. One reason to insist so much in this part of Steiner’s work regards his methodological implications. In nature it is rather difficult to observe hunting wasps in the moment when they catch the prey and sting it. What one does in many, if not in most cases, is to observe a re-stinging of the prey subtracted, and then given back, to the wasp in her way back to the nest. In general close to it, where it is easier to detect her. Or maybe substituting the prey with that subtracted to another wasp, or with an intact and artificially immobilized one to prevent its escape. The implicit assumption is obviously that the wasp stinging behaviour of the experimental situation will be the same of the natural one. The above analyses of Steiner show, on the contrary, that this may not be true, and the caution necessary in the use of re-stinging results. The last part of Steiner’s work examines the effects on the Liris stinging pattern exerted by some modifications operated in the cricket morphology. One will review them very briefly. Every specific act of Liris is in general preceded by an orientation phase. This holds for stinging as well as for other activities. For example, before paralysation the wasp will direct toward the posterior part of the cricket; before transport toward the head, to seize its antennae. The stimuli responsible to the orientation are in general different from those which produce the corresponding specific action. In spite of being strongly linked in natural conditions, the two kinds of stimuli may be experimentally separated: if one eliminates the portion of the cricket bearing a sting area, the orientation response will still be present without being followed by any stinging [48: pp. 83, 86]. Steiner determined a certain number of stimuli which govern the orientation phase, driving Liris to the first posture (Ia) which precedes the first sting: direction of prey movement; cricket shape, in particular its anterior-posterior polarity; influence of the prey posterior legs movement; influence of direct contact with the prey; influence of the dorsal-ventral polarity. The results obtained by a series of conflictual experimental

162

Francesco Andrietti

situations â&#x20AC;&#x201C; for example, moving the cricket in the direction of the tail the stimulus induced by the prey shape becomes contrary to that due to its movement â&#x20AC;&#x201C; allow to compare the relative importance of the different stimuli: cricket displacement direction < shape < movement of posterior legs, all of them being stimuli which act when the wasp is a at a certain distance from the prey. Instead, when she arrives in contact with the cricket the new stimulus due to the immediate tactile perception is strongest than the previous ones, and may modify or even invert their effects [48: pp. 87-89, 105-106]. Through a series of experiments of inversion and transposition of different segments of the cricket body, Steiner [48: pp. 90-107] showed that the wasp stings only if she finds the right sting area on the prey surface. If, for example, the sting areas III, IIb and IV are replaced by the ventral segment (where no sting area is present), no sting will be elicited. The same if the right sting area has been replaced by a different one. If, for example, the sting area IV is replaced by an additional area III, the stings Ia, IIb, III will be given normally; subsequently, the wasp will reach the position which, in normal situations, should be that of posture IV. But now her abdominal tip will meet sting area III, so that the sting will not be given. Considering now the case in which all sting areas are present, but one among them, for example area IV, has been displaced with respect to its natural position. After the first three stings the wasp will assume posture IV (with respect to sting area III), but no suitable sting area will be found. Then she will move the abdomen randomly on the cricket surface until its tip casually will contact sting area IV, and the sting in most cases will be delivered. One observes again the predominance of the tactile reaction with respect to the orientation stimuli (consisting in the posture assumed by the wasp in relation to the cricket regions stung before or by the shape or other directional indicators present on the prey surface): the lack of the tactile stimulus does not allow the sting which, on the contrary, may be elicited by the tactile stimulus alone, even in absence of any previous orientation [48: pp. 97, 100-102]. On the base of his experiments, Steiner excluded the possibility that the stinging behaviour, as any other behaviour of Liris biological cycle, may be explained uniquely through a chain of reflexes such that each of them acts automatically as stimulus for the subsequent one. Otherwise one could not explain, for example, the reason why the wasp does not sting a wrong sting area, provided that it is located in the right position with respect to the previous one. On the contrary, according to Steiner the wasp must possess a pre-existent scheme of the cricket which allows to sting only the right areas. This scheme does not involve necessarily the whole body of the prey, but only the region relevant to the concerned phase. Even in the case of a

Stinging behaviour of solitary wasps

163

c-stinging sequence which lacks of the sting in area IV, a decapited cricket works as well as an intact one: the wasp, after the last sting, may even begin its transport, provided that a pair of “antennae” are applicated to the decapited end [48: pp. 95, 110]. Considering the paralysing phase of Liris in the frame of whole her biological cycle, one observes that in many cases the lack of a characteristic area does not hinder the fulfilment of the corresponding act: a cricket deprived of antennae may be transported by a leg; deprived of anterior legs may be malaxated at the base of middle or even posterior legs; or the wasp may lay her egg if the prey is deprived of more anterior segments. However, the possibility of use substitution areas is not the same for the different activities of the wasp. In fact, it goes from a maximum for the transport to a minimum in the case of stinging, where no substitution is effectively possible (except a few rare and controversial cases). In the middle one finds oviposition and malaxation. The need to cross a threshold level explains the necessity of a certain delay and of the performance of void attempts, awaiting the moment when the degree of excitation is increased to a value such to allow the substitution. However, to any substitution area a certain value of threshold should be attributed in relation to the level of excitation of the wasp. In the case of the sting areas this threshold is so high that it will be never practically possible to reach it. The reason of this may well be due to their close relations with the nervous ganglia, what makes useless any other surface of the prey body [48: pp. 108-110]. The reason why the stings are given in the right position, from the beginning of the wasp adult life [48: p. 114], without any previous knowledge of the cricket body, remains very obscure. Even if one may not completely exclude that ...the larval life of Liris may have played a role in the formation of these structures of behaviour...Nevertheless in the case of the paralysing stings one may not reject the hypothesis of a direct detection of the nervous ganglia, what clearly would remove every innate and preformed character to the reaction; but this seems to us inconsistent with some data, as the attempt to give sting IV to decapitated crickets or the precision in the orienting postures before the sting, since the first paralysation (then without learning). If these different influences are ruled out, everything occurs to all appearance as if the “specific” behaviour of the adult Wasp, in relation to the cricket, was settled in function of a true “innate anatomical knowledge” (visual, tactile, etc.) of the cricket...” [48: p. 116].

164

Francesco Andrietti

The last words of this quotation are clearly reminescent of the Fabre’s concept of “an inborn knowledge of the internal anatomy of the prey” (see above).

Recent finding II: Steiner’s later works and 21st century’s contributions Steiner extended the kind of work made on Liris (not only L. nigra, but also Liris argentata (Palisot de Beavois) and Liris aequalis (W. Fox), which all show very similar sting patterns) to different genus of sphecid and vespid wasps, as Oxybelus, Prionyx, Tachysphex, Euodynerus, Podalonia, Larra [49, 50, 51, 52, 53, 2: p. 124]. In the following, one will review some of his results. Comparing the stinging patterns of different wasps to the correspondent prey nervous system structures, Steiner found additional support to what already stated in the case of Liris. In the flies, which are the prey of Oxybelus, the nervous system shows a remarkable level of concentration since all thoracic ganglia are fused in one single mass; moreover, many muscids lack of a distinct subesophageal ganglion, which is fused with the supraesophageal structures to form a single unit. Flies also lack of the powerful jaws present in the Orthoptera, which do not need to be neutralized. For these reasons, one understands that one single sting-pattern in the nervous thoracic mass may be sufficient, as one effectively observes. The fact that a different fly-hunter, Crabro latipes F. Smith, belonging to the same subfamily (Crabroninae) but to a different tribe, in addition to the thoracic sting sometimes delivers an occasional sting in the neck, could be interpreted as the presence of a “vestigial” sting in the way to be eliminated from the sequence [50, 54]. Even among species which share the same number of stings, one may observe a remarkable adaptation of the sting pattern to the ganglia position. This has been observed comparing the sting pattern relative to three Orthoptera-hunting wasps: Liris (hunting crickets), Tachysphex spp. and Prionyx parkeri Bohart and Menke (hunting short-horned grasshoppers). All of them show a C4SP pattern (see above), but with some differences in the order and position of the different stings: Liris and Tachysphex follow a posterior-anterior direction starting from the metathoracic segment. In contrast, Prionyx follows a reverse order, starting from the throat to terminate in the mesothoracic segment which will receive two stings. As in the case of Liris, the sting in the throat has an immediate local effect consisting in the elimination of any movement in the mouthparts of the prey and of its

Stinging behaviour of solitary wasps

165

defensive regurgitation, and a general deactivation effect which in crickets becomes apparent only after partial recovery from temporary paralysis: “If stinging is interrupted after the first sting both crickets [stung by Liris] and grasshoppers [stung by Prionyx] have lost one defence system (hind-legs vs. mouth-defences) but grasshoppers have also been “deactivated” and hind-leg defences have become depressed and inefficient, although not fully eliminated” [49]. The prey of all three wasp species show a subesophageal and three distinct thoracic ganglia, but with some differences, in the grasshoppers with respect to the crickets, due to a migration of the three thoracic ganglia toward the head, as it is shown by the correspondent increase in the length of the nerves roots of the different legs. According to Steiner, the basic stinging pattern remains unchanged even in fairly systematically distant wasps (Liris, Tachysphex = Larrinae; Prionyx = Sphecinae), since it is determined by the prey internal nervous structure which remains essentially the same, and should represent a typical situation of convergent evolution. In contrast, the order of the sting sequence or of the body segments involved may vary, depending on the divergent evolution of systematically distant wasps. However, one may not discard the possibility of second order adaptations. Liris is a frail wasp which needs to give a first stroke in the rear position, in order to master her cricket by paralysing its powerful hind legs. This is not the case for Prionyx, which overpowers her grasshopper “embracing” it, by mean of her strong and spinose legs, in an antiparallel posture, i.e. with the head directed toward the prey tail. In P. parkeri the sting in the mesothoracic segment (which receives two stings) rather than the metathoracic one, should be due to its minor distance from the third thoracic ganglion. In fact, one observes a parallel cephalization of both c.n.s’s and stinging patterns. This is shown even by an accurate examination of the stinging pattern of some common Tachysphex grasshopper-hunting species like Tachysphex tarsatus (Say) which show an intermediate situation, since the hindleg-paralysing sting is still on the third thoracic segment but in a position more frontal than in Liris (and in a non identified Tachysphex species it has shifted to the middle-legs as in Prionyx). In some cases the sting order may be modified under the influence of strong adaptation necessities, as in the mantis-killing Tachysphex species (see above) which give priority to the paralysis of the front raptorial legs, in contrast to grasshoppers-hunting Tachysphex species which firstly paralyze hind legs. A shift between long- and short-horned grasshoppers, and a parallel cephalization of both prey nervous systems and wasp stinging pattern, may be observed even within the subtribe Prionyxina, passing from Palmodes carbo Bohart and Menke to P. parkeri [49, 55].

166

Francesco Andrietti

The study of the stinging patterns of Prionyx gives the same results obtained for Liris: the analysis of the external positions and directions of the stings (including the abrupt change passing from the first sting directed toward the subesophageal ganglion, to the second one directed toward the first thoracic ganglion) makes it very plausible that their internal targets are the thoracic and the subesophageal ganglia. The presence of soft membranes is a necessary but not a sufficient condition to elicit the sting: for example, the last sting is located on a very reduced and inaccessible membrane when compared with the other ones; yet the total number of stings recorded in a certain number of cumulative observations is very similar for all of them. The thing is still more evident in the case of Oxybelus uniglumis (Linnaeus), which delivers only one sting, while the number of soft membranes is certainly much higher. In this case the sting, which is given at the base and rear of one foreleg, remains considerably constant both within and between Oxybelus species and even in C. latipes (which equally hunts on flies), showing that probably this point is primarily a stinging site widespread among Crabroninae, becoming only secondarily a transporting site as one observes in many Oxybelus species (Oxybelini) [50, 49]. The whole of Steiner’s works validates with strong evidence the exactness of Fabre’s opinion that the wasp stinging behaviour is strictly correlated to the nervous anatomy of the prey. In Steiner’s terminology it will be named locomotor ganglia hypothesis: “sting number and distribution are regulated only by ganglia involved in locomotion, escape and defence”. It closely matches not only the results relative to Liris, Tachysphex, Prionyx (and other wasps: Sphex ichneumoneus (Linnaeus), P. carbo, Isodontia sp.), where stings concern only thoracic ganglia (locomotion, escape through hind legs and defence system) and subesophageal ganglion (mouth defence, which may interfere with subsequent stinging or prey transport), but even those relative to O. uniglumis (only one sting directed toward the center of the single mass of fused thoracic ganglia that governs the whole locomotion of the fly, whose buccal apparatus is, on the other hand, completely harmless) and Podalonia luctuosa (F. Smith). The last species, differently from the others, deliver even abdominal stings. However this is only an apparent exception, since her prey is constituted of large caterpillars which use also abdominal segments for locomotion. The case of Pryonyx is a particular meaningful test that shows, due to cephalization of the prey neural system, as stings location is determined by the internal rather than the by the external anatomy of the prey (two stings on the same mesothoracic segment). What could not be excluded on the base of other cases for the one-to-one correspondence between internal and external features [50, 49, 51].

Stinging behaviour of solitary wasps

167

However, in some cases a reduction in the number of stings may be observed. Leaving apart the incomplete and atypical patterns as those observed in Liris (see above) but recorded even in Prionyx [49], this may be observed in the Eumenine wasp Euodynerus foraminatus (Saussure), whose prey are particularly small, weak and easy to overcome. They result very imperfectly paralysed as already observed by Fabre (see above), but without any loss of precision for the stings which simply skip some ganglia; however unable to escape, since too tightly packed to move anyway or owing to the â&#x20AC;&#x153;deactivatingâ&#x20AC;? effect of the throat sting, which may be of particular importance in case of temporary or (as in this case) very imperfect paralysis. An extreme case of reduction is observed when no sting is given, as in case of Microbembex, which preys dying or dead insects, or of the aphid-hunter Diodontus (see above) [50, 49, 51, 52]. A last point regards venom action. We remember as Ferton expressed doubts regarding the possibility that the venom could reach directly the ganglion, at least in the case of P. subfuscatus (see above). Steiner observed that the stinging pattern of P. subfuscatus is probably similar to that of P. parkeri. If so, Ferton, who described only two stings, has omitted the first one in the throat and probably an additional one in the thorax. Possible mistakes in number and positions of stings are very easy when one works without the help of video-photo techniques and without subsequent check of stings marks on the prey body. Alternatively, Ferton may have observed incomplete sting patterns in non optimal conditions. In any case, the lack of a one to one correspondence among the number of ganglia and those of the sting and the doubt that the sting could reach a ganglion located too far away should have been the main reasons to attribute the paralysis only to the diffusion of the venom. In fact, if the spread of the venom may play some role, it may not explain the instantly paralyzing effect, so important to prevent prey-escape. One should exclude also nervous peripherical effects, since otherwise the fourth sting given to P. parkeri, located at the base of the middle legs, would affect them instead then the rear pair [49]. An interesting problem regards the possible relations between stinging and carrying systems in Oxybelus. Already many early observers, starting from Fabre (quoted by Ashmead in [20: p. 73]; see also [56]) had reported that many species of this genus carry the prey impaled on the sting (abdominal transport), which in most cases is not taken out after the paralysing stroke. In some other cases, instead, the prey is brought with the legs (O. emarginatus Say: [57]; O. trispinosus Fabricius: [58] and Polidori et al., unpublished data). In still other species, however, one assists to both kinds of transport: the prey frequently is impaled on the sting only after reaching the nest (O. bipunctatus Olivier: [47: p. 366]). It should be of a

168

Francesco Andrietti

certain interest to determine the stinging position in species of Oxybelus using pedal transport, to check if it is the same of those which, as O. uniglumis, carry the fly on their sting. The method of stinging of the latter has been examined in some detail: the fly is stung on the ground, sometimes in flight; the wasp curls her abdomen around the anterior part of the prey close to the head, with her body axis quite transversal to that of the fly, slightly tilted forward; the sting is given in the ventral side, always in the same position behind a foreleg, right or left, while one wing of the prey is hold with the claws; after being stung, the abdomen unfolds (as already observed by Ferton, see above) and the fly changes its position from dorsal up to ventral up, while its longitudinal axis makes an open angle with that of the wasp; the stung side of the fly results uplifted with respect to the other one; instant and almost complete paralysis follows, all reactions disappearing within few days [57]. For what concerns the sting apparatus morphology, it has been found a reduction of the lancets length in two species showing abdominal transport (O. argentatus Curtis, O. haemorrhoidalis), so as to uncover the indented structures that could help transport on the sting (Polidori et al., unpublished data). One will examine only briefly two other works in which Steiner analyzed very accurately the different ways how the prey of L. nigra, P. parkeri, and some Tachysphex interact with the host wasps and try to avoid their attacks. Many other details may be found in the original works [59,55], In the case of Liris, one should distinguish between early and full hunting phases. In the first situation, to an increasing vigour of the wasp attacks correspond different gradual behavioural responses of the cricket (G. domesticus): raising abdomen, raising abdomen and body in a tilted position, body swaying. At this stage of the intensity scale threshold for escape is nearly reached, and one will observe: kicking with hind legs (less frequent than any other response), jumping and running away. Probability of subsequent pursuit is, however, still relatively low. In the full hunting phase, instead, responses lack of any â&#x20AC;&#x153;gradationâ&#x20AC;? either for what concerns the cricket or the wasp. The basic sequence, which involves detection by scent, sight or both, is now followed by attack (pouncing) of prey which escapes, by jumping or running, is pursued by the wasp and in general captured and stung. In contrast with the early hunting phase, the behaviour of the cricket is now unpredictable. It may jump away or stop suddenly, enter a burrow or change direction, what frequently disrupts wasp pursuit. Then the wasp stops instantly, standing upright in a particular head up posture, the fore legs maximally stretched, the vibrating antennae held in a special manner, making rapid movements of the body and often of the head in quick succession; or she may perform jerky zig-zag displacements. In this period of fast visual scanning of the surrounding, the wasp attacks any moving object. However, no

Stinging behaviour of solitary wasps

169

stinging of any insect other than crickets has ever been observed. Final capture of the prey results quite tumultuous and difficult. The wasp uses mandibles and legs to grasp parts of the struggling prey, very often one of the posterior legs (see above). Before being stung, the cricket may try to enter a burrow or to kick away the wasp with one of the hindlegs. Another possibility consists in the autotomy of the seized (hind) leg, observed frequently in captivity but seemingly widespread even in field conditions. Still one other situation arises when the pursued cricket suddenly stops on the top of a small obstacle assuming one of different possible â&#x20AC;&#x153;freezingâ&#x20AC;? postures. In most instances the wasp does not pounce on it; in case she does, in general no stinging or only abortive attempts of stinging follow (see above) [48: pp. 44-45; 59]. Many anti-predatory strategies have been recorded even in grasshoppers preyed by P. parkeri [55], like jumping and flying away, kicking, freezing, biting, assuming intimidating postures or by regurgitation of a repellent fluid. When the prey has already been seized by the wasp, it makes frantic efforts to kick or push away the enemy with the tarsi of its powerful hind legs, which appear to be precisely directed at the points grasped by the wasp. To prevent wasp initial sting posture, it may rise its long folded hind legs beyond the vertical, headwards. Instead, hind leg autotomy observed in crickets was never observed. If the wasp is able to overcome all difficulties, as she frequently does (even dashing at flying grasshoppers and stinging them in mid-air), she will deliver four successive stings, as described above [55]. In the case of P. parkeri and Tachysphex, a statistical analysis allowed only to assess a slight, non significant advantage in avoiding or reducing stings number in prey where defensive actions were recorded, even if some of them, like that consisting in jumping and flying away or perhaps freezing (that seems to be, however, less efficient in avoiding wasp attack than in the case of L. nigra) and mouth regurgitation, appear to be helpful. However, such investigations are made particularly difficult by the wide fluctuations in responsiveness of the hunting waps [55]. In a subsequent work, Steiner [51] was involved in the study of the stinging pattern of the Nearctic species P. luctuosa, one of the few sphecid wasps genus which, together with Ammophila, hunt caterpillars. In particular, Podalonia prey consist mainly in cutworms of the moth family Noctuidae, which burrow into the soil during daytime [51; 47: p. 143]. Even in this case, as in those examined before by the same author, data were obtained by offering the prey in cages held in laboratory, to overcome the difficulty to obtain them afield. This contrasts to natural conditions in which the caterpillars are found at the base of plants or within their subterranean nests. In spite of this unusual condition caterpillars were readily accepted by the wasps, and results were considered as valuable [51].

170

Francesco Andrietti

Figure 1. A Podalonia affinis (W. Kirby) stings a cutworm dug out from its subterranean shelter (from a video tape recorded sequence redrawn by S. Agostini): a-f, different positions assumed by the cutworm and the wasp during the fight; arrows indicate the direction of rotation of the cutworm around its longitudinal and transversal axes.

Defensive reactions of the cutworm consist in spraying of repelling fluid and vigorous coiling in alternation with flicking and twisting [51; see also 18-II: p. 25]. In Fig. 1 is represented in a half-schematic way a few (a-f) of the different positions taken from the cutworm and the wasp during the stinging phase, sketched from a video recorded sequence in natural conditions: a rare event which one had the chance to observe from the initial excavation of the prey (Andrietti, unpublished data). The hunting behaviour of the Palearctic species Podalonia hirsuta (Scopoli) had already been object of study by part of different authors, starting from Fabre (Ammophila hirsuta = P. hirsuta). In particular, Fabre firstly described digging behaviour during hunting, which is uncommon to observe in other sphecid wasps [18-II: p. 24; 51]. He also described the two sequences of 4 + 6 stings, in a forward and backward direction respectively, interrupted by the â&#x20AC;&#x153;triumphâ&#x20AC;? dance (see above and [51]). More recently (from 1966 to 1982: see [51] for detailed references) Fulcrand, Gervet and Truc made a careful investigation of the stinging pattern of this species.

Stinging behaviour of solitary wasps

171

The stinging pattern observed in P. luctuosa appears to be similar to that of the other examined species. It consists, at least in experimental controlled situations, in ten stings on the ventral side: a first sequence (type I) is delivered from T3 (third thoracic segment) to H (head segment), which matches the one described previously; a second sequence (type II) follows in a reverse direction, from A1 (first abdominal segment) to A6, where sting marks abruptly stop. In Steiner’s terminology the complete pattern will be represented by the formula C4SP + C6SP. In both cases the wasp faces the caterpillar head segment, proceeding in opposite directions [51]. However, it may happen that she mistakes the tail for the head of the cutworm, giving a few “erroneous” stings on the abdominal segments, either proceeding in the tail direction (type I “error”: [51]) or in the opposite one (type II “error”: [51]; see also [18-IV: p. 251]). These kinds of errors are quite common in caterpillar-hunters wasps [51; see also Fabre, above, concerning eumenine wasps hunting Chrysomelidae larvae]. In conclusion, the sting pattern of P. luctuosa reflects the caterpillar nervous organization which lacks of the any significant concentration. All segments are stung, from the head to the sixth thoracic one where the last fused ganglia are located [51]. In P. luctuosa sting areas are wider compared to the previous considered wasps (Liris, Prionyx, Tachysphex, Oxybelus) which sting sclerotized prey. In fact, no constraint due to the presence of hard membranes is present in caterpillars which are uniformly soft. However, characteristic points coordinates are well identified and they may be mapped by mean of a grid over imposed to the ventral surface of the cutworm. They appear to be located, in the thoracic and abdominal segments, respectively behind and before the closest ganglion. This matches the direction of the sting which, even if rather difficult to assess with precision, seems to be forwards in the C4SP sequence, backwards in the C6SP one, even if the change does not appear to be so abrupt and rigidly fixed as in the previously studied wasps. A part the higher number of stings, the Podalonia pattern is characterized by stings and sequences repetitions (2.5 repetitions on the average for the complete sequence) and, in general, by a larger amount of variability than the already mentioned wasp-prey systems. Final paralysis depends now more on the cumulative effect of numerous stings, each of ones appearing now to be less precisely localized and with mainly a local and segmental effect. Moreover, the decreasing response of the prey renders the last (abdominal) stings less important and less frequently repeated, in contrast with the other systems in which all stings are approximately equally important [51]. On the other hand, the relative scarcity of characteristic features increases the dimension of the stinging areas on the caterpillar segments and

172

Francesco Andrietti

the arising of possible sting errors. With the only exception of the head capsule which, correlatively, exhibits a highly restricted distribution of stings; conversely, they are more widespread on A1 and A2, devoid of appendages. The same variability is observed in the egg-position, in contrast to the greater site specificity reported for the other studied wasp-prey systems [51]. In other works specifically devoted to investigate the effect of the reactivity of the prey to the stinging pattern of P. hirsuta, Truc and Gervet [60, 61] found that: 1) in the normal (control) case the prey initially elicited mostly a type 1 attack (C4SP), with a trend to shift toward type 2 (2 â&#x20AC;&#x201C; 8 stings delivered on metathorax and abdomen) when it became progressively paralyzed; 2), in an experimental case made of permanently reactive caterpillars (since they were systematically replaced after the first attack), the onset of types 2 was delayed, even if they gradually increased in proportion, notwithstanding the state of the prey was stable; 3), in a third group of already paralyzed prey, the first attacks were made either of types 1 or 2 in equal proportion, with a tendency to alternate one to the other. These findings may be explained on the base of two tendencies, the first linked to the position in the stinging cycle sequence, the second to the state of the prey. If in case 1) they match together, in cases 2) and 3) they conflict giving rise to anomalous situations [61]. In Ammophila sabulosa (Linnaeus) the hunted prey were found to belong to the two caterpillars families Noctuidae and Geometridae, in roughly equal proportion. Differently from Podalonia species, these prey are not found in the soil but on the leaf of some trees, were they are traced in a peculiar way: wasps walk on the ground with the raised abdomen, searching very carefully for caterpillar faeces; once found them, they fly off exploring a conical three dimensional volume at increasing heights; if they find the prey, this is knocked onto the ground, where it is stung; otherwise, the wasps resume their exploring on the ground. Sting marks on the caterpillar body were found on the throat and in the first eight segment, leaving apart segments A6 and A7. In spite of the number of stung segments which are approximately the same (10 in P. luctuosa against 9 in A. sabulosa); the fact that the first stung segments (T3 or throat, respectively) are those more frequently stung; the increase followed by a decrease of sting frequencies from A1 to A5; the abrupt stop after A5 (A6 in P. luctuosa) which allows to maintain defecation of the caterpillar; and finally, the fact that the wasp faces the head of the caterpillar, one should note some differences in the sting patterns of the two wasps. The two different positions of the first sting could be related to different evasive manoeuvres of the prey: the cutworms hunted by P. luctuosa make vigorous coiling around a â&#x20AC;&#x153;hingeâ&#x20AC;? approximately located at the level of T3, which

Stinging behaviour of solitary wasps

173

should be firstly neutralized; if this would not happen for Ammophila prey (what should be checked), the sting pattern could begin at one end of the caterpillar and specifically at the throat, which contains higher centers which control mouthpieces movements. So the whole sting pattern proceeds regularly backwards and it has no need to be partitioned in two different sequences. Instead of a C4SP + C6SP one has now a single C9SP. After stinging, females sometimes used their mandibles to massage the preyâ&#x20AC;&#x2122;s body, maybe to permit an adequate spreading of poison since no feeding or fluid exuding from the caterpillar was observed. Since the mark of the sting is approximately midway between two consecutive ganglia, and no information is available regarding the sting direction, there is no way to decide about the possibility that ganglia may be reached or not by the sting tip. One has observed a (weak) positive correlation between stings number and prey size [62]. In an evolutionary perspective, the use of Lepidoptera larvae is probably a character derived from a previous situation of Orthoptera-hunters, as in most sphecine wasp. If a C4SP scheme or even a reduced one may be still sufficient for small and weak caterpillars packed in tight cells as in the case of eumenine wasps (see above and below), this is not the case for powerful caterpillars as those hunted by Podalonia and Ammophila. This could have brought to a secondary specialization with an additional C6SP pattern, and the discontinuity between the two sequences (in Podalonia but not in Ammophila, what should be verified) could be a sign of their different origin. In other hunting caterpillars solitary wasps, the number of stings appear to be reduced. The sting pattern of E. foraminatus, which feeds on small Lepidoptera larvae, was the object of a detailed study [52]. According to Steiner, in the Eumeninae the C4SP pattern is often reduced to the first and last step of the normal sequence (in the throat and the third thoracic ganglion: C2SP), sometimes up to four stings, instead than the regular 10. This modification should be due to the weakness of the Eumeninae prey, constituted mostly by lepidopterian caterpillars but even by larvae of weevils and chrysomelids beetles. Instead, T1 and T2 receive stings with a much lower frequency. A few additional stings may be found irregularly distributed in the abdominal segments which, in contrast with the regular thoracic ones, are irregularly scattered and not necessarily located close to the median line. However, the number of these irregular stings is low (three cases out of 23), and does not justify the claim of a fundamental imprecision and lack of constancy of this species, which fundamentally follows a 2-stings pattern rule. In fact stings are clearly clustered, not uniformly or randomly distributed. Stinging directions were not systematically investigated, due to difficulties of observation, but they were in general directed toward the

174

Francesco Andrietti

medial ventral line of the prey [52, 51, 2: pp. 129-130, 134]. Even Cooper (1953, quoted by [2]: p. 130) reported that Ancistrocerus antilope (Panzer) received three times more stings in the throat and the third thoracic ganglion than in all other areas combined. In contrast with the prey of Podalonia and Ammophila (which may show coiling and mouth regurgitation), eumenine wasps prey apparently lack of any mean of defence, except to suddenly drop the ground after having been extracted from within their rolled leaves. In all observed cases sting H always preceded sting T3, even if initially in a motionless prey either head or tail were investigated; this did not happen in moving prey, meaning that direction of escape, as in the case of Liris, is an important visual orienting clue. The wasp grasps the caterpillar dorsally with her mandibles facing its tail in an antiparallel posture, stings the throat area and then proceeds slightly tailwards to sting the metathoracic area; sometimes she repeats the same procedure even several times. After stinging is terminated, the wasp pulls or stretches the prey with the mandibles and/or compress and chews the tail or the head, presumably to lick some body fluid as already described by Fabre for a different eumenine wasp (see above). Differently from what observed in other wasps (see above), abortive or reduced sting patterns were rarely observed even in the case of using already stung prey, probably because they are quite as reactive as the non stung ones [52]. In any case many reports from the literature mention three thoracic stings in several Eumenines, as reported by Fabre for Odynerus (see above). So it is possible that in some cases the C2SP pattern may be substituted by a C3SP (Steiner 1983a, p. 19; 1986, Tav. VI, pp. 131-133), maybe even C4SP. In fact, Fabre underlines that O. nidulator (S. murarius), when giving the three stings in the thorax, “insists” particularly under the neck, what maybe could mean a first stroke in the throat (H), almost inseparable from T1 by direct observation alone [18-IV: p. 204; 52]. Recently, Budrienè and Budrys re-examined the sting patterns of a few Eumeninae in a series of interesting works [15, 16, 17]. Their results do not match completely those of Steiner, since the sting pattern appears to be much less stereotyped. In the case of Symmorphus allobrogus (Saussure), which hunts leaf beetles (Chrysomelidae) larvae, they found 6-9 stings (average 8.7) distributed over 4-6 (average 5.1) prey segments. The highest number of stings was given to the first five segments: the throat, the three thoracic and the first abdominal. Since the locomotion of the leaf beetle larvae is mainly due to thoracic legs, their CNS is weakly concentrated, and they use abdominal dorsal glands for defence, their sting pattern agrees with the “locomotor ganglia” hypothesis without the particular reduction in number reported by Steiner. In fact, there is a negative correlation between the

Stinging behaviour of solitary wasps

175

mobility of the prey and both the presence and the number of the stings in the three thoracic segments and the first two abdominal ones. Surprisingly, this same correlation becomes positive when made with respect to the throat. This last effect could be attributed, according to the authors, to the fact that the stings in this position, directed to the subesophageal ganglion, should produce a (transitory) deactivation of the prey (see above) which may reduce the number of subsequent stings. In such case, after recovery the prey will result more mobile [15]. In fact, this explanation does not match the absence of any correlation between the stings in the throat and those in the thoracic ganglia, the most important in controlling prey mobility. In S. allobrogus the effective C5SP pattern appears more flexible compared to that considered by Steiner. However, one should note that in this work the position of the stings was determined uniquely from the traces left on the body of the larva [15], without any control to establish the biological cycle of the wasp and her hunting condition in order to avoid incomplete or atypical sequences (see above). This may increases the variability of observed sting schemes [57]. Neither the effective presence of inactivation after stinging the throat has ever been really checked. On the other hand, when comparing the different patterns between E. foraminatus and S. allobrogus (C2SP and C5SP, respectively), one should also take into account the possible diversity in the organization of the CNS of the two kinds of prey (caterpillars and leaf beetles larvae). In a second work BudrienĂ¨ & Budrys [16] considered the stinging pattern in ten species of Eumeninae belonging to three different genera: Ancistrocerus, Symmorphus and Discoelius. The studied species of the first and third genus hunt caterpillars; those of Symmorphus hunt leaf beetles larvae, except S. debilitatus (Saussure) which hunts caterpillars. The statistical analysis revealed that in most cases the number of the sting traces (whose averages varied from a minimum of 3.9 in A. nigricornis (Curtis) to a maximum of 51 in S. murarius), and their distribution among the 13 body segments of the prey (including the throat) is a species specific characteristic. Moreover, in all the ten cases one found a very significant difference between the observed patterns of stings distribution and those expected on the base of a C4SP or a C2SP scheme. However, four caterpillar hunting species show a distribution close to C2SP, while the stinging scheme of the three species of Symmorphus which hunt Chrysomelidae larvae show a regular pattern consisting in stings in the throat, the three thoracic segments and the first abdominal one (C5SP). The distribution of stings on the prey may be used in phylogenetic studies as a comparative character. In fact, each of the three considered genus show some peculiarities with respect to the other ones: Ancistrocerus shows a lower stinging effort in the mesothorax; Discoelius, a

176

Francesco Andrietti

higher stinging effort in the fourth, third and fifth abdominal segments; Symmorphus, a higher stinging effort in the first abdominal segment. Since these differences were determined independently from the kind of prey, the authors find more plausible that the stinging pattern is a phylogenetically inherited feature rather than being mainly an adaptative character to the physiology and nervous anatomy of the prey as supposed by Steiner. A strict subordination to the taxonomy of the species seems to receive additional support from the fact that the difference between the stinging patterns of the less related caterpillar hunting genera Ancistrocerus and Discoelius is larger than that between the related genera (hunting different prey) Ancistrocerus and Symmorphus. In any case, adaptation to a particular kind of prey may play an important role as well: hunting leaf beetles wasps deliver more stings to the mesothorax and the first abdominal segments [2: p. 129; 16]. For what concerns a possible relation between the number of stings and the prey dimension, an analysis performed on eight of the above mentioned species has shown only a (weak) positive correlation, while the relative sting effort, i.e. the number of stings per prey weight unit, was negative with respect to the prey weight while the total stinging effort per offspring positively depended on the number of prey specimens stored in a brood cell [17]. Comparing the results relative to the stinging pattern of different eumenine wasps, one has the clear impression of a high level of variability even between closely related species [2: Table VI, pp. 131-133] or even intraspecifically [63, 64]. In some cases these differences may not be attributed to unreliable observations made in the field since, as in the case of Steiner’s and Budrienè & Budrys’s findings, they were both obtained by examining the sting traces left on the larva body.

Modern times and the future: Integrating neuroanatomy, electrophysiology, ethology and biochemistry In spite of the mentioned evidences, firstly presented by Fabre’s and subsequently reinforced by Steiner’s works, a definitive proof of the correctness of the “locomotor ganglia” hypothesis could arise only on an histological base. However, a simple histological investigation of the nervous system of paralyzed insects may lead to different results, as it appears from the works of Nielsen (1932) and Hartzell (1935) (quoted by [27]). Piek (1978, quoted in [2: p. 143 and Fig. 11, p. 144]) was probably the first to identify, by mean of an autoradiographic method, the site of the venom injection by part of a Mellinus arvensis Linnaeus inside the compound thoracic ganglion of a fly.

Stinging behaviour of solitary wasps

177

Gnatzy and Otto [65] used a different method, cutting a “window” into the metathoracic sternite of Acheta domesticus (Linnaeus) (one of the crickets hunted by L. nigra) and directly observing the penetration of the wasp sting into the metathoracic ganglion, without stopping in the hemolymph space. This result may be considered as an unequivocal proof that the venom, to act, must be inserted inside the ganglion. In fact, if it is applied artificially to the vicinity or surface of a thoracic ganglion, it fails to affect the CNS or the behaviour of the cricket: its molecules are too large to cross the perilemma membrane around the ganglion. Gnatzy and Otto were also able to correlate the venom injection to different electrophysiological and behavioural data. Extracellularly recorded spontaneous spike frequency from the connective between pro- and mesothoarax ganglia decreased continuously during stinging sequence, reaching its lowest level after it was completed. After 4 – 10 min interneuronal activity raised again slowly. This is the time necessary to recover from the initial total (legs, mouthparts and antennae) and transient paralysis. After this delay, the cricket is again able to respond to stimulations by reflex actions (intersegmental reflexes recover later and with more difficulties that intrasegmental ones), while the spike rate in the connective increases and, in about two hours, reaches the 80% of the original rate. The cricket, now become a “reflex machine” (see above), maintains this “deactivated” state for several days, becoming progressively worse for the reason that it does not feed any more: the motor behaviour is further reduced until the cricket eventually dies [65]. In a second set of experiments, recording was made at the level of the connective between SEG (subesophageal ganglion) and T1, while a burst of action potentials was elicited by mean of tactile stimulation. These were abolished after a sting in T1 or in SEG (by mean of a suitable apparatus which allows Liris to sting only at given positions of the cricket), to reappear a few minutes later. Moreover, stings in T3 and T1 only did not impair the reflexes of the middle legs. On the base of their results, Gnatzy and Otto concluded that the immediate action of the Liris venom paralysis likely consists in the inhibition of spike generations in neurons of the stung ganglia [65]. In subsequent works Gnatzy and co-workers established that, during total paralysis, venom blocks synaptic transmission and action potential generation. The first effect is probably localized at the presynaptic side, and is due, at least partially, to a block of inward calcium currents. The second is due to a block of voltage-gated sodium inward currents of the corresponding neurons. Even if the effect of the poison is restricted to the stung ganglion, it does not selectively affects leg motoneurons but any other neuron, like interand neurosecretory neurons. On the other hand, sensory information from

178

Francesco Andrietti

legs mechanoreceptors of the stung ganglion may be recorded during the whole period of total paralysis. For what concerns suppression of spontaneous behaviour, instead, it is necessary that the venom is injected inside the SEG [66, 67, 68]. The neuroethological researches of Gnatzy extended also to other aspects of the Liris-cricket interaction. They are quite interesting and merit to be examined, since they appear as a physiological correlate to the ethological investigations of Steiner [59]. Crickets (A. domesticus) respond to the approach of L. nigra with conspicuous behavioural reactions, like head-stand (raising abdomen), stilt-stand or defensive kicks, which seem to belong, at least partially, to the threatening postures repertoire of the males mating behaviour. Gnatzy has shown, through lesioning experiments, that defensive reactions (head-stand and even stilt-stand) are neither elicited by visual nor by substrate vibrations, but by the air movement generated by the predator approach. Reception is localised in the filiform hairs of the cerci. Neurograms were recorded from the axons of giant interneurons in the terminal abdominal ganglia of the crickets, which receive their inputs from the cercal filiform hairs and by adjacent campaniform sensilla. Results showed that they responded to air currents generated by flying wasp females when they were at a distance of 15-20 cm. Filiform hairs are especially sensitive in the range of velocities and wing beat frequencies of the flying wasps (50 cm/s and 150 Hz). Instead, if the wasp approaches the cricket from behind hunting â&#x20AC;&#x153;on footâ&#x20AC;?, no response is given up to the distance of 1-3 cm. These results may help to explain the method adopted by Liris, which never flies but only runs during hunting. In such situation, moreover, she holds her wings folded motionless over her back, in contrast to other solitary wasps as, for example, the Pompilidae. However, hunting females run very fast, at a speed not very lower than that of the flight. Even if this increases the intensity of the produced signals, it reduces the time available to the detected prey to react, and enables the wasp to explore a larger amount of territory [67]. Instead, Steiner [59] gives a different account of Liris hunting method The wasp walks slowly, head down with the antennae tapping the soil. Information picked up may be chemical or chemotactile. The wasp, like a hunting dog, follows with great precision the scent trails left by crickets and investigates dropping. In any case, prey recognition by part of Liris is due to visual (from a distance of up to 15 cm) and chemical cues, detected by mean of a kind of sensilla which are commonly found on the antennae of several insect orders: the sensilla basiconica. In the present case they are distributed in large

Stinging behaviour of solitary wasps

179

number only on the upper side of the six distal flagellomera of the females, only a few of them being present in male flagella. The prey is attacked only after their contact with the prey body. Only they are involved in prey recognition, as it is apparent in video analyses and in ablation experiments. Instead, the other types of contact chemosensitive hairs present in the antennal flagella of the females are unimportant. Their structure presents a perforated oval plate at the tip of the sensillum hair shaft, differing either from typical olfactory sensilla (whole hair shaft perforated) or typical contact chemoreceptors (only one pore). However, the need of contact with the prey supports the hypothesis that one is dealing with contact receptors [67]. An important role for prey recognition should be attributed to an attractive substance present on the prey cuticular surface, which may be transferred to filter paper or to another cricket species (Gryllus bimaculatus De Geer) normally not accepted as a prey. In crickets, contact chemical cues are used in sex recognition when acoustic communication is not possible, and may have been secondarily used by predators to localize their prey [67]. Recently, in a series of correlated papers [69, 70, 9, 12, 71, 72, 73, 11, 8, 13], Libersat and colleagues examined the site of injection of the venom, the electrophysiological and the pharmacological effects in the CNS of the common cockroach Periplaneta americana Linnaeus, when stung by the wasp Ampulex compressa (Fabricius). After grabbing the cockroach at the pronotum or at the base of the wing, a first brief sting (10-20 s) given into the first thoracic ganglion through the soft membrane between the front leg and the prothorax induces a few minutes of paralysis of the forelegs, which recover in a few minutes. Within this delay, the coackroach is unable to use its front legs to fight off the wasp, what facilitates the second and more important sting, much more precise and timeconsuming, in the head [72, 8, 74]. The authors made use of a combination of liquid scintillation and light microscopy autoradiography to investigate the precise localization of the venom. By mean of 14C radiolabeled amino acids injected in the wasp and subsequent measure of radioactivity in the stung (by the wasp) prey, they found that radioactivity was mostly localized in the first thoracic ganglion rather than in the other thoracic ganglia or non-neuronal tissue. To prove that the venom is injected directed in the first thoracic ganglion rather than to diffuse from outside into the nervous tissue, radiolabeled amino acids were experimentally injected into the thorax of the cockroach outside of its ventral nerve cord: most of the radioactive signal was recorded in the surrounding thoracic non-neuronal tissue, significantly less in the thoracic ganglia [72]. After the second sting, the radioactivity accumulates in the head ganglia of stung cockroaches (more in the brain than in the subesophageal ganglion),

180

Francesco Andrietti

remaining very low in the surrounding tissues. More precisely, radiolabeled venom was found posterior to the central complex and around the mushroom bodies of the brain and around the central part of the subesophageal ganglion. This seems to show that the wasp stings both into the SEG, which lies underneath the stinging site in the neck, and separately into the brain which lies 2 mm deep in the head capsule. Both of them are considered higher neuronal centers which send descending tonic signals to thoracic motor centers to modulate the beginning of locomotion, so providing their role in the motivation to initiate or maintain walking-related behaviours. According to Libersat, this precise stereotaxic injection should require the presence of sense organ on the sting tip. In fact, receptors have effectively been identified on the tip of the sting (or ovipositor) of different Terebrantia [75, 76, 77] and social wasps and bees [78], but in solitary wasps this information is known only for the sting sheaths [27, 67]. These last ones may allow detecting the precise position where to insert the sting on the cuticle surface, not inside the prey body. When labeled amino acids were experimentally injected in the head cavity, most of the radioactivity was found in the surrounding nonneuronal tissues and significantly less in the head ganglia. One may conclude that venom does not diffuse into the head ganglia from the stung site, but is directly inserted inside the brain by the sting tip [11, 8, 74]. The biochemical basis of the transient paralysis of the cockroach forelegs is probably to find in a central synaptic block due to the enhancement of chloride conductance by part of a GABA receptor channel present on the cell body of some (as the fast costal) thoracic motoneurons whose motor output is under cholinergic control. In fact it has been shown that Ampulex venom contains high level of GABA together with two agonistic substances – taurine and β-alanine – which have the similar effect to activate the conductance of a chloride channel and that additional to prolongate the action of GABA, probably inhibiting its reuptake from the synaptic cleft. The three amino substances present in Ampulex venom, together with GABA receptors, have been found even in the venom of Vespa and of the spider wasp Anoplius samariensis Pallas, but their physiological roles remain to be demonstrated [79, 7]. If the effect of the thoracic sting is to induce (1) transient paralysis of the front legs, that of the subsequent sting in the head produces (2) intense grooming, and (3) long-term hypokinesia [8]. The stung prey “first grooms extensively, after which it becomes sluggish and is not responsive to various stimuli. The wasp grabs one of the antennae of the cockroach, which follows docilely to a suitable oviposition location...” [11]. The ethology of host manipulation has been studied by Keasar et al. [80]. At the end of the grooming phase, the wasps cut the antennae of the prey (which are

Stinging behaviour of solitary wasps

181

constituted by females more readily than males). The cutting region is localized through sliding the mandibles several times over each antenna, and results close to where a maximal variation between consecutive segments length is perceived, which may be considered as a point of discontinuity. The wasp now feeds on the hemolymph from the cut end, which is located in a position which is optimal for an easy flow. Then the cockroach, kept by one of the stumps, is led to the nest with the cockroach proceeding forwards, the wasp backwards facing the prey. Learning does not appear to play any significant role in subsequent behavioural sequences, which appear to be primarily genetically determined [80, 74]. These unique effects (2-3) suggest that a high center of the insect CNS may be involved in the action of the venom, possibly the brain. Moreover, one has shown that the A. compressaâ&#x20AC;&#x2122;s venom has no effect on the cockroach neuromuscular junctions [8, 11]. The evoked grooming (2) lasts about 30 min nonstop following the sting and may not be evoked in any other way, neither stress nor attack or mechanical stimulation or venum injection in a location other than the head, apparently activating a specific neural network controlling all the components of a normal grooming behaviour which involves the coordinated movements of many different appendages (antennae, mouthparts, legs). The responsible of these effects, which are evocated only if venom is injected into the head, has been identified as a dopamine-like substance which could stimulate dopamine receptors present in the cockroach SEG, where a group of dopaminergic neurons (some with axons branching extensively in the SEG, others connecting to the brain or thorax) have been identified by mean of immunohistochemical studies. In fact, injection of DA or DA-receptor agonists induces excessive grooming, similar to venom-induced one, while injection of DA-receptor antagonist prior to sting markedly reduces it. However, a stung and hypokinetic cockroach does not show any grooming neither if stung a second time or injected by DA. This suggests the presence, in the venom, either of a DAlike substance inducing grooming and of a dopamine-blocking component which subsequently blocks it and could be even responsible for eliciting hypokinesia [8, 69, 74, 12, 71]. However, the primary function of the second sting is that to produce the long-lasting lethargic effect (3), which fully develops at the end of the extensive grooming, lasts for 2-5 weeks and may be defined as a â&#x20AC;&#x153;long lasting change in the threshold for initiation of various locomotory behavioursâ&#x20AC;?. Its action regards spontaneous activity and specific motor behaviours, leaving others like grooming, righting or flying unaffected. The most striking effect concerns the escape behaviour, which is no longer

182

Francesco Andrietti

elicited in stung cockroaches by wind stimuli directed to the cerci or tactile stimuli to the antennae or anal plates. Tactile (from cuticle or antennae) and wind sensory information all seem to converge on the same pool of thoracic interneurons (thoracic premotor circuitry) which, either directly or via other interneurons, excite the motoneurons involved in fast leg movements. Since experimental results have shown that the sting does not affect sensory descending or ascending (GI, giant interneurons) pathways, the authors propose that the ultimate effect of the venom is exerted at the level of the thoracic escape circuitry. Given that thoracic interneurons receive a comparable synaptic input from GI in control and stung animals, it is suggested that the venom acts at the level of descending neural inputs from the head ganglia, whose importance has been demonstrated [81] through their removal, which induces a reduction of leg movement and fast motoneurons activity as in stung cockroaches. Their effect may possibly be exerted through dopamine-like facilitating neuromodulatory neurons on the connections between pre-motor internerneurons and (fast) motoneurons of the thoracic circuitry or maybe directly on the motoneurons synapses. This appears to be a more parsimonious way to prevent escape, rather than to block all sensory inputs coming from antennae, cerci and cuticular surface. In addition, it explains the specificity of the inhibition, since stung animals maintain activity in slow motoneurons and do not produce rapid movements. However, fast motoneurons may be recruited during behaviours others than escape, as righting, swimming or flight. Recent studies have shown that in stung cockroaches (as well as in cockroaches lacking the SEG) the activity of identified neurons in the thorax (DUM neurons) which secrete octopamine (OA) and control the excitability of specific thoracic premotor neurons is compromised, probably through a modulation of calcium currents. This result appears presumably due to a removal of descending neuromodulatory inputs from the head ganglia. Since the postsynaptic activity on thoracic (OA secreting) DUM neurons of stung cockroaches was found comparable to that of control ones, the input of the descending pathways should not directly control these neurons, rather other interneurons which synapse on them. OA seems to play a role even at the level of cerebral ganglia, since injection of an OA-receptor antagonist into the brain of an unstung cockroach significantly reduces walking activity. On the contrary, octopamine partially restores walking in stung wasps. Clusters of OA-immunoreactive neurons were identified in the SEG; at least some of them provide dense innervation in the ellipsoid body of the central complex and in the protocerebral bridge, which are regions involved in the control of locomotion. In conclusion, according to the last

Stinging behaviour of solitary wasps

183

findings it appears plausible that the wasp manipulates octopaminergic circuits in cockroach cerebral ganglia that, in turn, influence octopaminergic circuits in the thoracic ganglia to induce hypokinesia [8, 74]. Many other works have been devoted to the pharmacology of the substances able to induce prey paralysis present in solitary wasps venom. They will not be reviewed here, since they are too far from the ethological focus of the present work. Some of them may be found in Piek [82] and in the references of more recent works [79, 7]. A last point, which is worthwhile to mention, concerns the state of the paralyzed prey. In particular two hypotheses may be considered: 1), paralysis is only a mean to immobilize the prey and/or to prevent escape, without any effect on its metabolic rate; 2), paralysis reduces the metabolic rate of the prey, so extending its life span at the interior of the nest. Probably the answer may vary according to different cases. In a rather old work Nielsen (1935, quoted by Roces and Gnatzy [83]) made an accurate study regarding the metabolic rates of caterpillars (Geometridae) and spiders (Epeira cornuta (Clerc)) paralysed by Ammophila campestris Latreille and Episyron rufipes Linnaeus. He did not find any difference in long term metabolic records between paralysed and food-deprived animals, suggesting mechanism 1). Instead Roces and Gnatzy [83], working on L. nigra, found that paralysis significantly reduces both the metabolic rate and the mortality of her prey, the house cricket (A. domesticus). Since the control sample was constituted by food and water deprived animals which were unable to move, the change in the metabolic rate of the hosts was due to a direct effect of the wasp venom, not to a decreased locomotory activity [83]. Analogously, Haspel et al. [10] found that the sting of A. compressa not only renders the cockroach prey helplessly submissive but also changes its metabolism, measured by a decreased oxygen consumption. This occurred even after pharmacologically induced paralysis or after severing the neck connectives. However, neither of these two groups of treated cockroaches survived more than six days, while 90% of stung cockroaches survived at least this period. In addition, cockroaches with severed neck connectives lost significantly more body mass, mainly due to dehydratation. One may conclude that the metabolic manipulation caused by the sting in the brain operates in a subtler way than simply removing descending inputs from the head ganglia, since it leaves some physiological processes, such as water retention, intact. Once again one finds a confirmation of what established by Fabre for the Orthoptera-hunting â&#x20AC;&#x153;Sphex languedocienâ&#x20AC;? (P. occitanicus), whose prey, Ephippiger, shows to live much longer when paralysed than when alive and without food

184

Francesco Andrietti

What seemed as though as it should be a cause of death was really a cause of life. However paradoxical it may seem at first sight, this result is exceedingly simple. When untouched, the insect exerts itself and consequently uses up its reserves. When paralysed, it has merely the feeble, internal movements which are inseparable from any organism; and its substance is economized in proportion to the weakness of the action displayed. In the first case, the animal machine is at work and wears itself out; in the second it is at rest and saves itself. There being no nourishment now to repair the waste, the moving insect spends its nutritive reserves in four days and dies; the motionless insect does not spend them and lives eighteen days. Life is a continuous dissolution, the physiologists tell us; and the Sphex’s victims give us the neatest possible demonstration of the fact [18-I: p. 187].

Prey selection and the “locomotory ganglia hypothesis”: A new frontier of investigation According to Steiner’s “locomotory ganglia hypothesis”, the stinging pattern of solitary wasps (number and position of stings) should match the nervous organization of the host. Under the hypothesis of a fixed stinging pattern for any given species, this would limit the range of the possible prey to those which share a common neuroanatomy. As a matter of fact, most solitary wasps hunt a restricted prey spectrum, in some cases only one single species (as P. triangulum) or only very related ones (as Liris). But this is not always true. Excluding extreme cases like Microbembex, which preys any sort of dead or dying insects but without stinging them, there are some genus which prey insects belonging to different taxa. One of them is represented by Cerceris, which may hunt beetles of different families but even halictids and other bees (see above). If in most cases the diet of one single Cerceris species is restricted to a single taxon (in general a family of beetles, like weevils or buprestids or chrysomelids), sometimes only to a single or few species (as Cerceris tuberculata (Villers), see [18-I: p. 67]), there are a few species more generalist. In a recent investigation which regards the prey spectrum of the beetle-hunting Cerceris rubida Jurine (Fig. 2), one has found that more than 95% of the hunted specimens belonged to three families: Chrysomelidae (50%; presence in the environment 39%), Curculionidae (32%; 17%), Phalacridae (13%; 7%). Among the other three hunted families, one found also a small amount of Scolytidae (1.3%; 0%) which, as the weevils, present fused thoracic ganglia

Stinging behaviour of solitary wasps

185

Figure 2. A female Cerceris rubida while stinging her chrysomelid prey. Drawing by Silvia Di Martino, from a picture taken on field by C. Polidori.

(see above) [84]. One could think to the poliphagy of C. rubida as representing a transition phase between the more common weevils (or buprestids) hunting species and the chrysomelids-hunting ones. Assuming, in absence of any data, that the stinging pattern of C. rubida consists of only one sting, as it is common for other beetles-hunting Cerceris species [2: Table V, pp. 113-114], it would be more suitable, on the whole, to paralyse weevils than chrysomelids. In fact, the first ones show in general a more advanced level of fusion of thoracic ganglia (second and third in quite all subfamilies, and partial fusion of first and second in a few of them) than the second ones [85, 86]. Effectively, the ratio given by (percentage of hunted prey)/(percentage of available prey) has been found higher in weevils (1.86) than in leaf beetles (1.27). In spite of her ‘generalist’ attitude with regard to different beetles subfamilies, C. rubida shows “taxonomical” specialization at the level of single prey species which may not be explained uniquely in terms of other well known factors responsible of prey selection as ecology, size, shape, age, sex, escape behaviour, etc., and, of course, availability [84]. To explore the possibility that this taxon-biased selection could be related to differences in prey nervous organization, one should dispose of precise neuroanatomical descriptions of all preyed species. Unfortunately, we may dispose only of data concerning the nerve cord structure of a few species of different leaf beetles and weevils subfamilies, not necessarily those hunted by C. rubida. For what concerns Chrysomelidae, the most frequently hunted genus is

186

Francesco Andrietti

Chaetocnema (5 species which, in total, represent the 47.6% of the total hunted specimens from all prey families; scarce among specimens collected in the environment); the second more frequent genus is Cryptocephalus (4.5%; 5.0%) [84]. The literature does not provide any data for Chaetocnema, even if in other species belonging to the same subfamily (Halticinae) one may find a moderate concentration of the thoracic ganglia, mainly of the meso- and the metathotacic ones which, in some cases, may be connate or fused [85]. However, to the same subfamily belongs Phyllotreta, a genus which results abundant in the environment (more than 40%) and completely absent in wasp hunting records [84]. In the case of Cryptocephalus, whose hunting frequency agrees with its environmental frequency [84], the investigated species show all thoracic ganglia contiguous with each other [85]. For weevils, the more hunted genus is Protapion (23.8% hunted; 15.8% in the environment) [84]. For the species of the same subfamily (Apioninae) which have been investigated for what regards the anatomy of the nerve cord, one reports fusion between the second and the third thoracic ganglia, and in some cases a partial fusion between the first and the second. However, these are features which are common to most weevil subfamilies [86]. The subfamily (Entiminae) to which belongs Sitona, a genus which is well represented in the environment (7.2%) but is very infrequently hunted (0.2%), does not show meaningful differences, since the investigated species show fusion of the second and third thoracic ganglia [84, 86]. On the base of the above considerations, it does not seem possible to attribute to the nerve cord morphology, at least at the level of the gross anatomical analysis considered here, a major role in determining the high discrimination in prey selection among species which are systematically very close (in some cases belonging to the same genus as in Chaetocnema: [84]), and probably share a similar nervous organization. However, a more reliable indication could be obtained only through a neuroanatomical study devoted to the prey specifically hunted by C. rubida.

Conclusions Looking back to the old literature outlined at the beginning of the present work, one sees how many of the problems which will be matter of discussion up to nowadays were already present from the beginning. This is true in particular for the controversial ideas regarding the way how stings induce paralysis in the prey. The initial position adopted by Fabre in the last decades of the 19th century fell soon in disfavour with respect to that of Ferton and followers, which remained dominant (even if not universally accepted) for large part of the 20th century. However, the first researches of Steiner on Liris

Stinging behaviour of solitary wasps

187

revitalized Fabre’s opinion, rebaptizing it under the term of “locomotory ganglia hypothesis” and were reconfirmed by his subsequent works. The arguments of Steiner appear quite convincing, at least in the case of Liris [48] for which the point and the direction of insertion of the sting have been accurately recorded. The same may be said for Oxybelus [50] and, even if at a minor extent, for Prionyx [49]. Still more convincing are the results of Piek for Musca domestica Linnaeus stung by the sphecid wasp M. arvensis, based on autoradiographic methods (see above); or those of Gnatzy and Otto for A. domesticus stung by L. nigra (see above); or those of Libersat and colleagues regarding the two stings received by P. americana by part of A. compressa, the first localized in the first thoracic ganglion, the second in two precise and separate locations of the head ganglia (brain and SEG) (see above). However, a part these few cases, anatomical evidence is lacking for the large majority of the predator-prey systems. On the other hand, observational data, in spite of their abundance [2: Tables I-V] are in general too imprecise and contradictory to allow unambiguous answers. Moreover, in many cases they were obtained in situations different from the “standard” ones, for example in re-stung cases, which may give rise to different responses (see above). It is probable that many, if not most of the old data concerning stings location, are affected by errors. This may explain the large differences in the results coming from different authors. Take, to consider only one case among the many possible ones, O. uniglumis: according to very old accounts (Sickmann, 1883, Chevalier, 1926, Adlerz, 1903, all quoted in [2: Tab. V, p. 106]) the unique sting is located in the neck. Instead, according to Steiner [50], and based on the marks of sting and 150 cases (no number of observations was indicated by previous authors, presumably very low or maybe only a single one), the sting is behind one foreleg base (see above). One has no doubt about the exactness of Steiner’s finding even if, without making use of the traces left by the sting, or at least some accurate video or photographical techniques, the posture assumed by the wasp during stinging may give the false impression that the prey is effectively stung in the neck [57: Figs. 1 and 4]. Among the other species (cfr. [2: Tav. V, p. 105]), O. bipunctatus has been reported to sting in the middle of the prey thorax [87] and O. haemorrhoidalis between the head and T1 [23]. Another discrepancy, which concerns the sting pattern in Prionyx, has already been discussed above. Apart the difficulty to localize the exact position of the sting, accidental elements may obscure the typical sting pattern, depending either on the state of the wasp in relation to her biological cycle, or on the state of the prey. This is one of the reasons why fortuitous field observations may give variable responses. To avoid this difficulty, one should make use of experimentally

188

Francesco Andrietti

controlled situations: prey (discarding those with deficient or absent responses) should be offered to wasps which are in the right phase of their nesting site [48: p. 8; 49], and a large number of data should be collected. If this method has the advantage to produce more regular results and to better identify the average typical sting behaviour, ruling out abnormal or irregular cases, it has also the difficulty to imply long-term studies, and the necessity to rear the wasps as a way to have a better control of their nesting cycle [57, 49, 51]. This is not always possible or easy to do. Moreover, in many cases the wasps will not always accept experimentally provided prey. Another point concerns the knowledge of the nervous anatomy of the prey, which often is scarce or not available. However, even among the wasps studied by Steiner according to his criteria of investigation (analysis of sting traces and, whenever possible, sting directions in reared wasps), there are some for which the “locomotory ganglia hypothesis” is far to be proved: in P. luctuosa there is only a vague match between the sting direction and the position of the ganglia (see above); still more uncertain is the situation for the eumenine wasps which, in addition, present a reduced number of stings which does not match a correspondent reduction in the number of ganglionic masses [52]. The same may be said for A. sabulosa, in which sting marks are about midway between consecutive ganglia (see above). A part these dubious situations, there is at least one case represented by the bee wolf P. triangulum in which the venom is injected in the hemolymph of the prey, a fact which was firstly acknowledged by Rathmayer [27]. He found that the sting delivered by the wasp was always given, either in experimental (prey and wasp maintained in small containers) or natural conditions, in the unsclerotized membrane on the ventral side of the bee behind the first pair of legs. One has already observed (see above) that this result does not match with those provided by Fabre. However, Tinbergen in a series of repeated observations confirmed Fabre’s findings, reporting that the bee was stung “under the chin” [88: p. 50, text and figure], at least in experimental situations. After the sting, which has a duration of about 30 s [27], the bee is turned ventral up and malaxated by the wasp which squeezes the bee’s abdomen, as observed by Fabre (see above) and confirmed by Rathmayer [27]. Of particular importance for the present discussion was the histological analysis performed by Rathmayer who, in ten out of twelve cases, found that the sting did not reach the first thoracic ganglion. In the other two cases the perilemma of the ganglion appeared to be damaged, but probably the wasp’s sting had broken off and so reached the ganglion. Additional considerations showed that the venom could not consist in a neurotoxin which blocks the

Stinging behaviour of solitary wasps

189

ganglionar centers, since it does not produce a sudden paralysis but takes effect gradually. In fact, the nerve cells of the different muscles of the same leg are positioned very close to each other in the same thoracic ganglia and should be reached by the venom at the same time. Yet observations proved that the coxae and femur were paralyzed when the tibia and basitarsus of the same leg were still movable [27]. One must conclude that the venom has a peripherical effect, firstly spreading around the puncture and then reaching the musculature by mean of the hemolymph. This explains the progressive paralysis of the leg segments. Muscles which control movements of the coxae and femora are directly bathed by the hemolimph, which is responsible for a quick transport of the poison, while the muscles of the tibiae are, instead, located within the femora. It takes a considerable time for the venom to reach them through the action of the accessory pulsating organs located at the basis of the leg. Still more to affect basitarsi muscles, which are located inside the tibiae [27]. Another point examined by Rathmayer concerns the site of the sting, to understand if its position behind the fore legs is made necessary for its closeness to the thoracic ganglia, as claimed by the â&#x20AC;&#x153;locomotory gangliaâ&#x20AC;? hypothesis, or only because it is one of the few sites which can be pierced by the sting. New stinging spots were predetermined, by experimentally cutting windows in the chitin in different positions on the bee body. The results showed that the delay in the onset of the paralysis varied according to the distance from the sting [27]. In conclusion, the venom does not affect the ganglia but produces paralysis when it reaches the hemolimph. Even in the case of a bee stung in the head, the effect is that to paralyze antennae and proboscis leaving other parts unaffected, without showing any disturb in the movements coordination as it would be in the case of any action on the CNS. A further proof came from the histology, which showed no changes in the brain, SEG and thoracic ganglia, except the first one in the immediate vicinity of the puncture in which alterations in the glial cells and neuropilem, not in the ganglion cells, were visible. Only 24 hours later these modifications spread out to the other thoracic ganglia, but nerve cells were almost never affected. Not until three days after, shrinkage in the nervous structures was noticeable. It is possible that after peripherical paralysis has set in, the venom disturbs the metabolic functions of the glial cells and only indirectly the neurons [27]. If the effective position of the sting has no more importance to induce the paralysis, but is only determined by the possibility to find a suitably soft membrane, it could vary at a certain extent. This could explain the diversity of reports concerning its effective location, rather than to attribute it to mistakes of very valuable observers.

190

Francesco Andrietti

The neurotoxins (philanthotoxins), reaching their sites of action represented by the neuromuscular junctions exert a diversity of effects on synaptic transmission processes, basically interfering presynaptically with the release of the excitatory transmitter and blocking the postsynaptic glutamate receptors [89, 8]. This allows to understand the reason why one single sting is sufficient to paralyse the bee, in spite of the fact that it does not own a fused thoracic gangliar mass as is the case for the flies, which equally receive only one sting by O. uniglumis (see above), or for spiders stung by pompilid wasps, which in general receive only one or few stings in the mouth and/or thorax/base of legs (see above and [2: Tab. IV, pp. 84-90]). The alternative explanation suggested by Steiner [49, 51] is so ruled out. It consisted in the possibility that in “higher” insects, as the bees, a single sting not too far from the SEG, whose functions of coordination and control are become particularly important, may be sufficient for a complete block of locomotion. In addition Philanthus venom lacks of specificity, since it exerts its effect in all insect orders, including closely related Sphecidae, to the extent to suggest its use as a new class of pesticides [27, 90]. One should stress the need for further researches, mainly on a histological level, to individuate the possible presence of wasp radiolabeled venom inside the ganglia of the host. The researches should be planned for the most important genus of solitary wasps as Cerceris, Bembix, Sceliphron, pompilid wasps, etc., not yet investigated in this way up to now and still awaiting for such an analysis, for which we dispose of not always reliable and often controversial data [2: Table V, pp. 94-114]. Take, for example, the case of Cerceris arenaria (Linnaeus). According to different authors and due to the extreme sclerotization of the weevil prey, stings may be delivered only between the head and the thorax, between T1 and T2 or between the thorax and the abdomen [50; 2: Table V, p. 113] (Fig. 3). In this case, however, we dispose even of a more reliable result given by Piek [82: p. 612 and Fig. 10 p. 613] who observed discolouration through the ventral side of the soft membrane between head and thorax of the weevil prey, providing a SEM picture of it. Even a search for possible receptors on the sting tip of solitary wasps should be of interest, to confirm the possibility to detect the right position where to inject the venom. In addition, the ethology of the stung prey should be re-examined to discover other cases of “manipulation of host behaviour”. Leaving apart the challenge that these findings pose to their explanations in terms of natural selection theory, they show that “some of the parasites know more about the brain than all neuroscientists combined” [74], so stressing the importance of their study. The phenomenon is not confined to wasps, appearing quite widespread among phylogenetically distant parasites like fungi, trematods,

Stinging behaviour of solitary wasps

191

nematods, Nematomorpha (Gordioidea), Acanthocephala, and is probably much more common than we actually know. In insects, this ability may be found in Terebrantia. The most striking situation has been described by Eberhard [91] and regards the ichneumonid wasp Hymenoepimecis, which induces a change in the orb web construction of her prey, a spider which, after recovering from the sting, resumes apparently normal activities and builds normal orb webs, while the wasp's egg hatches and the larva grows by sucking the prey's hemolymph. Instead, one or two weeks later the larva induces the spider to build a web especially designed to support the wasp cocoon. Its construction is highly stereotyped, consisting of many repetitions that are almost identical to the early stages of one subroutine of normal orb weaving, the other components of which are repressed [91]. A different case reported concerns the braconid wasp Perilitus coccinellae (Schrank), which preys on ladybirds. Emerging from the beetles, the parasitic larva weaves a cocoon around itself, entangling with some threads the prey (still alive) as well. In the case the cocoon is removed “the beetle runs around furiously in search of it; having found the cocoon, the beetle sits on it and tries again to get its legs entangled in the loose silken tissues surrounding it” [92: p. 102]. Among Aculeata, the number of stings and/or the complexity of hunting strategies are often correlated to the host/prey size and to its behaviour. In the case of pompilid wasps, for example, a first stroke in the mouth incapacitates the fangs of the spider, while Liris gives a first sting in the metathoracic segment to paralyze the strong cricket hind legs. When the prey is of small size or harmless, a single stroke may often be sufficient; otherwise, many of them are needed. In general, the wasps paralyze their prey to drag them into a burrow or a nest. This is not always the case: Larra anathema (Rossi), after having dislodged the prey (mole cricket) from its nest, induces a transient paralysis stinging it in the mesothorax, prothorax and below the throat, and ovoposit on it. Resuming from the short-term paralysis, the mole cricket will go back to its shelter with the wasp egg, eliminating the need of a long-term paralysis and sparing the wasp from building her own nest and to transport the prey into it. Also pompilid wasps hunting ctenizid spiders (Nemesia) oviposit in the prey’s nest (see above), without any need to build their own one. However, what marks the difference is the fact that Larra oviposits outside the mole cricket’s nest. This will leave the door open to superparasitism by part of the same or similar species of wasps, as effectively it has been observed in Larra bicolor Fabricius. However, in no instance did more than one larva per host survive to pupation. Superparasitism is rather common in Terebrantia and it has been reported also in primitive aculeate wasps as Tiphia femorata Fabricius, which equally does not build any nest

192

Francesco Andrietti

and, after oviposition, inspects the ventral aspect of the prey to discover strange eggs that, in the case, will be eaten. Instead, Larra “kneads” with the mouth the site of the prey where she will oviposit, probably to destroy any strange egg or larva. [92: pp. 119, 124, 125; 93; 74]. Among aculeate wasps which oviposit inside a nest from which the prey may not escape, superparasitism and consequent kneading behaviour are rare, even if non completely forgotten events, since they reappear in cases of intraspecific parasitism: for example, in a population of A. sabulosa were often observed wasps which opened the nest of a previous one to remove the egg from the caterpillar, replace it with their own, and then re-close the nest [94]. Instead, “host behaviour manipulation” is a quite rare phenomenon to observe, since its presence (together with the venom induced grooming behaviour) is restricted, as far as one knows, to A. compressa (see above) [74]. If the stings at the thoracic level induce paralysis in the corresponding limbs, those delivered in the head (throat) may give rise to a variety of situations which are quite complex to analyze, as one sees comparing the results from different wasps. In the case of different species of Larra, which adopt a C3SP (see above), but also a C2SP -scheme according to different species (and authors) [53], after a first temporary paralysis the wasp recovers in a few minutes and resumes normal activities until eaten by the developing host larva. Instead, in the case of Liris (C4SP) one has shown, by selective eliminations of the different stings, that only the last one in the throat is the necessary (and sufficient) condition to induce long-term hypokinesia (deactivation) in the cricket (that will not attempt to escape). The different action of the head sting in the two wasps is even shown by the fact that, according to Tsuneki [87], maxillary and labial palpi of the prey stung by Larra carbonaria erebus Smith could still move, in contrast to crickets stung by Liris wasp which seldom move these appendages. A similar permanent partial deactivation may be observed in Prionyx (C4SP in reverse order with respect to Liris) [87, 53]. Lastly, in Ampulex (C2SP), after the second sting (in the head) one observes grooming, hypokinesia and manipulated induced behaviour (see above). The diversity of responses due to the sting in the head should not wonder if one considers the neural complexity of the head central ganglia with respect to the thoracic ones. This means that the effect of the sting will change in relation to the specific head centers involved. We recall the precise stereotaxy of the venom injection of Ampulex in the brain centers and in the SEG (see above). It appears that even in some cases of behavioural manipulations induced by worms, there is a parasite migration in the head ganglia; in other cases an overproduction of particular proteins or amino acids has been measured in the

Stinging behaviour of solitary wasps

193

head or in the CNS of the host, probably induced by the parasite either directly or indirectly via a host genome response [74].

Acknowledgements The author is greatly indebted to Carlo Polidori and one anonymous referee for the many helpful suggestions he gave to improve the manuscript.

References 1. 2. 3. 4. 5.

8. 9.

10.

11. 12. 13.

Chapman, R.F. (1998). Insects: Structure and function. Cambridge. Cambridge Univ. Press, 770p. Steiner, A.L. (1986). Stinging behaviour of solitary wasps. In: Piek, T. (ed.), Venoms of the Hymenoptera, pp. 63-160. Academic Press, London. Evans, H.E., West-Eberhard, M.J. (1970). The wasps. University of Michigan Press, Ann Arbor. vi + 265 pp O'Neill, K.M. (2001). Solitary wasps: Behavior and natural history. Comstock Publishing Associates, New York, 406p. Packer, L. (2003). Comparative morphology of the skeletal parts of the sting apparatus of bees (Hymenoptera: Apoidea). Zoological Journal of the Linnean Society 138:1-38. Macalintal, E.A., Starr, C.K. (1996). Comparative morphology of the stinger in the social wasp genus Ropalidia (Hymenoptera: Vespidae). Memoirs of the Entomological Society of Washington (17):108-115. Hisada, M., Satake, H., Masuda, K., Aoyama, M., Murata, K., Shinada, T., Iwashita, T., Ohfune, Y., Nakajima, T. (2005). Molecular components and toxicity of the venom of the solitary wasps, Anoplius samariensis. Biochemical and Biophysical Research Communications 330, 1048-1054. Libersat F. (2003). Wasp uses venom cocktail to manipulate the beavior of its cockroach prey. J. Comp.Physiol. A 189, 497-508. Libersat, F., Haspel, G., Casagrand, J., Fouad, K. (1999). Localization of the site of effect of a waspâ&#x20AC;&#x2122;s venom in the cockroach escape circuitry. Journal of Comparative Physiology A 184, 333-345. Haspel, G., Gefen, E., Ar, A., Glusman, J.G., Libersat, F. (2005). Parasitoid wasp affects metabolism of cockroach host to favor food preservation for its offspring. J. Comp.Physiol. A 191 (6), 529-534. Haspel, G., Rosenberg, L.A., Libersat, F. (2003). Direct injection of venom by a predatory wasp into cockroach brain. J. Neurobiol. 56, 287-292. Weisel-Eichler, A., Haspel, G., Libersat, F. (1999). Venom of a parasitoid wasp induces prolonged grooming in the cockroach. J. Exp. Biol. 202, 957-964. Gal R, Rosenberg LA, Libersat, F. (2005). Parasitoid wasp uses a venom cocktail injected into the brain to manipulate the behavior and metabolism of its cockroach prey. Arch Insect Biochem Physiol 60: 198-208.

194

Francesco Andrietti

14. Gal R, Haspel G, Libersat, F. (2003). Wasp uses specialized sensors to probe and inject venom inside the brain of its cockroach brain. Neural Plasticity 10: 200 15. Budrienè, A., Budrys, E. (2004). Applicability of the “locomotor ganglia” hypothesis to the stinging behaviour of Symmorphus allobrogus, a predatory wasp hunting chrysomelid larvae. Acta Zoologica Lituanica 14(4), 23-33. 16. Budrienè, A., Budrys, E. (2004). Hunting behaviour of predatory wasps (Hymenoptera: Vespidae: Eumeninae): is the distribution of stinging effort phylogenetically inherited or dependent on the prey type? Ann. Soc. entomol. Fr. 40 (3-4), 259-268. 17. Budrienè, A., Budrys, E. (2005). Effect of prey size and number of prey specimens on stinging effort in predatory wasps. Acta Zoologica Lituanica 15(4), 330-340. 18. Fabre, J.H. (1879-1910). Souvenirs Entomologiques, Vols. I-X (Vols. I-IV : 1879, 1882, 1886, 1891). Delagrave, Paris. Pages references from l’Édition Définitive Illustrée (1914-1924). When not otherwise stated, quotations are taken from the partial English translation of Teixeira de Mattos contained in “Hunting wasps” (1916, reprinted by the University Press of the Pacific, 2002), “The Mason-Wasps” (1919, reprinted by the University Press of the Pacific, 2002) and “More hunting wasps” (1920, reprinted by Kessinger Publishing, 2004). 19. Marchal, P. (1887). Étude sur l’Instinct du Cerceris ornata. Archives de Zoologie Expérimentale et Générale, Deuxième Série, Tome Cinquième. 20. Peckam, G.W., Peckam, E.G. (1898). Instincts and habits of the solitary wasps. Madison, Wisconsin. 21. Bullock, T.H., Horridge, G.A. (1965). Structure and Function in the Nervous Systems of the Invertebrates, 2 vols., Freeman, San Francisco. 22. Berland, L. (1923). Notes sur les Hyménoptères fouisseurs de France. II, synonymie de quelques noms employés par J.-H. Fabre. Bull. Soc. Ent. France, 171-174. 23. Ferton, C. (1901). Notes détachées sur l’instinct des Hyménoptères mellifères et ravisseurs avec la description de quelques espèces. Annales de la Société Entomologique de France, 1re série, LXX, 83-148. 24. Ferton, C. (1896). Observations sur l’instinct de quelques Hyménoptères du genre Odynerus Latreille. Actes de la Société Linéenne de Bordeaux XLVIII, 1-14. 25. Evans, H.E., O’Neill, K.M. (1988). The natural history and behavior of North American beewolves. Cornell University Press, Ithaca, New York. 26. Kurczewski, F. E., Miller, R.C. (1983). Nesting behavior of Philanthus sanbornii in Florida (Hymenoptera: Sphecidae). Florida Entomologist 66(1), 199-206. 27. Rathmayer, W. (1962). Paralysis caused by the digger wasp Philanthus. Nature 196, 1148-1151. 28. Simonthomas, R.T., Veenendaal, R.L. (1978). Observations on the behaviour underground of Philanthus triangulum (Fabricius) (Hymenoptera, Sphecidae). Entomologische Berichten (Amst.) 38, 3-8. 29. Ferton, C. (1923). La vie des abeilles et des guêpes. Rabaud E., Picard, F. eds. Chiron, Paris.

Stinging behaviour of solitary wasps

195

30. Ferton, C. (1902). Notes détachées sur l’instinct des Hyménoptères mellifères et ravisseurs avec la description de quelques espèces. Annales de la Société Entomologique de France, 2e série, LXXI, 499-531. 31. Ferton, C. (1909). Notes détachées sur l’instinct des Hyménoptères mellifères et ravisseurs avec la description de quelques espèces. Annales de la Société Entomologique de France, 5e série, LXXVIII, 401-422. 32. Ferton, C. (1905). Notes détachées sur l’instinct des Hyménoptères mellifères et ravisseurs avec la description de quelques espèces. Annales de la Société Entomologique de France, 3e série, LXXIV, 56-104. 33. Ferton, C. (1891). Notes pour servir à l’histoire des Pompilides. Actes de la Société Linéenne de Bordeaux XLIV, 1-14. 34. Ferton, C. (1897). Nouvelles observations sur l’instinct des Pompilides (Hyménoptères). Actes de la Société Linéenne de Bordeaux LII, 1-34. 35. Ferton, C. (1908). Notes détachées sur l’instinct des Hyménoptères mellifères et ravisseurs avec la description de quelques espèces. Annales de la Société Entomologique de France, 4e série, LXXVII, 535-586. 36. Ferton, C. (1910). Notes détachées sur l’instinct des Hyménoptères mellifères et ravisseurs avec la description de quelques espèces. Annales de la Société Entomologique de France, 6e série, LXXIX, 145-178. 37. Ferton, C. (1911). Notes détachées sur l’instinct des Hyménoptères mellifères et ravisseurs avec la description de quelques espèces. Annales de la Société Entomologique de France, 7e série, LXXX, 351-412. 38. Ferton, C. (1899). Observations sur l’instinct des Bembex Fabr. (Hyménoptères). Actes de la Société Linnéenne de Bordeaux LIV, 1-15. 39. Nielsen, E.T. (1945). Moeurs des Bembex. Universitetets Zoologiske Museum, Copenhagen. 40. Evans, H.E. (1957). Comparative ethology of digger wasps of the genus Bembix. Comstock Publishing Associates, Ithaca, New York. 41. Evans, H.E. (1966). The comparative ethology and evolution of the sand wasps. Harvard University Press, Cambridge, Massachusetts. 42. Polidori, C., Ouadragou, M., Gadallah, N.S., Andrietti, F. (2009). Potential role of evasive flights and nest closures in an African sand wasp, Bembix sp. near capensis Lepeletier 1845 (Hymenoptera Crabronidae), against a parasitic satellite fly. Tropical Zoology 22 (1), 1-14. 43. Rau, P, Rau, N. (1918). Wasp studies afield. Princeton University Press, Princeton. Quotations from 1970 ed., Dover, New York. 44. Ferton, C. (1901). Les Hyménoptères de la Corse (Apiaires, Sphégides, Pompilides et Vespides). Sur les moeurs du Stizus fasciatus Fabr. Comptes rendus de l’Association Française pour l’Avancement des Sciences, Congrès d’Ajaccio, 1901. 45. Ferton, C. (1895). Sur les moeurs du Dolichurus haemorrhous Costa. Actes de la Société Linéenne de Bordeaux XLVII 1-7 46. Andrietti, F., Casiraghi, M., Martinoli, A., Polidori, C., Montresor, C. (2008). Nesting habits of two spider wasps : Anoplius infuscatus and Episyron sp. (Hymenoptera: Pompilidae), with a review of the literature. Annales de la Société entomologique de France, 44(1), 93-111.

196

Francesco Andrietti

47. Bohart, R.M., Menke, A.S. (1976). Sphecid wasps of the world. University of California Press, Berkeley. 48. Steiner, A.L. (1962). Etude du comportement prédateur d’un Hyménoptère Sphégien: Liris nigra V.d.L. (= Notogonia pompiliformis Panz.). Ann. Sci. Nat. Zool. Biol. Anim. 4, 1-126. 49. Steiner, A.L. (1981). Digger wasp predatory behavior (Hym., Sphecidae). IV. Comparative study of some distantly related Orthoptera – hunting wasps (Sphecinae vs. Larrinae), with emphasis on Prionyx parkeri (Sphecini). Z. Tierpsychol. 57, 305-339. 50. Steiner, A.L. (1979). Digger wasp predatory behavior (Hym., Sphecidae): fly hunting and capture by Oxybelus uniglumis (Crabroninae: Oxybelini); a case of extremely concentrated stinging pattern and prey nervous system. Can. J. Zool. 57, 953-962. 51. Steiner, A.L. (1983). Predatory behavior of digger wasps (Hymenoptera, Sphecidae) VI. Cutworm hunting and stinging by the ammophiline wasp Podalonia luctuosa (Smith). Melanderia 41, 1-16. 52. Steiner, A.L. (1983). Predatory behavior of digger wasps (Hymenoptera, Sphecidae) V. Stinging of caterpillars by Euodynerus foraminatus (Hymenoptera: Eumenidae). (1983). Biology of Behaviour 8, 11-26. 53. Steiner, A.L. (1984). Why can mole crickets stung by Larra wasps (Hymenoptera, Sphecidae: Larrinae) resume normal activities? The evolution of temporary paralysis and permanent deactivation of the prey. Journal of the Kansas Entomological Society 57(1), 152-154. 54. Hewitt, C. G. (1908). The structure, development, and bionomics of the housefly, Musca domestica, Linn. Part I. – The anatomy of the fly. Quarterly Journal of Microscopical Science, Vol s2-52, 395-448. 55. Steiner, A.L. (1981). Anti-predator strategies. II. Grasshoppers (Orthoptera, Acrididae) attacked by Prionyx parkeri and some Tachysphex wasps (Hymenoptera, Sphecinae and Larrinae): a descriptive study. Psyche J. Entomol. 88, 1-24. 56. Hartman, C. (1905). Observations on the habits of some solitary wasps of Texas. Bulletin of the University of Texas 65, 1-73. 57. Steiner, A.L. (1978). Evolution of prey-carrying mechanisms in digger wasps: possible role of a functional link between prey-paralyzing and carrying studied in Oxybelus uniglumis (Hymenoptera, Sphecidae, Crabroninae). Quaestiones Entomologicae 14, 393-409. 58. Bonelli, B. (1952). Osservazioni biologiche sul Mellinus arvensis L. e sull’Oxybelus trispinosus F. Boll. Inst. Entomol. Univ. Bologna 19(9), 137-143. 59. Steiner, A.L. (1968). Behavioral interactions between Liris nigra Van der Linden (Hymenoptera: Sphecidae) and Gryllulus domesticus L. (Orthoptera: Gryllidae). Psyche J. Entomol. 75, 256-273. 60. Truc, C., Gervet, J. (1974). Influence de la réactivité de la proie sur le comportement de piqûre chez un Sphégide chasseur de chenilles Noctuidae: l’Ammophile Podalonia hirsuta Scopoli - Hypothèses sur le mécanisme de l’enchaînement des actes au cours d’un complexe instinctif. Z. Tierpsychol. 34, 70-97.

Stinging behaviour of solitary wasps

197

61. Truc, C., Gervet, J. (1983). Influence of the reactions of the prey on the stinging patterns of digger wasps. Experientia 39 (11), 1320-1322. 62. Casiraghi, M., Martinoli, A., Bosco, T., Preatoni, D.G., Andrietti, F. (2001). Nest provisioning and stinging pattern in Ammophila sabulosa (Hymenoptera, Sphecidae): influence of prey size. Italian Journal of Zoology 68, 299-303. 63. Bonelli, B. (1976). Osservazioni eto-ecologiche sugli Imenotteri aculeati dell’Etiopia. VIII. Boll. Inst. Entomol. Univ. Bologna 33, 33-43. 64. Bonelli, B., Bullini, L., Cianchi, R. (1980). Paralyzing behaviour of the wasp Rynchium oculatum Scop. (Hymenoptera Eumenidae). Monitore Zoologico Italiano 14, 95-96. 65. Gnatzy, W., Otto, D. (1996). Digger wasp vs. cricket: application of the paralytic venom by the predator and changes in behavioural reactions of the prey after being stung. Naturwissenschaften 83, 467-470. 66. Ferber, M., Consoulas, C., Gnatzy, W. (1999). Digger wasp vs. cricket: immediate actions of the predator’s paralytic venom on the CNS of the prey. J. Neurobiol. 38 (3), 323-337. 67. Gnatzy, W. (2001). Digger wasp vs. cricket: (neuro-) biology of a predator-preyinteraction. Zoology, 103, 125-139. 68. Ferber, M., Hörner, M., Cepok, S., Gnatzy, W. (2001). Digger wasp vs. cricket: mechanisms underlying the total paralysis caused by the predator venom. J. Neurobiol. 47 (3), 207-222. 69. Fouad, K., Libersat, F., Rathmayer, W. (1994). The venom of the cockroachhunting wasp Ampulex compressa changes motor thresholds: a novel tool for studying the neural control of arousal? Zoology 98, 23-34. 70. Fouad, K., Libersat, F., Rathmayer, W. (1996). Neuromodulation of escape behavior of the cockroach Periplaneta americana by the venom of the parasitic wasp Ampulex compressa. J. Comp. Physiol. A 178, 91-100. 71. Weisel-Eichler, A., Libersat, F. (2002). Are monoaminergic systems involved in the lethargy induced by a parasitoid wasp in the cockroach prey? J. Comp. Physiol. A 188, 315-324. 72. Haspel, G., Libersat, F. (2003). Wasp venom blocks central cholinergic synapses to induce transient paralysis in cockroach prey. J. Neurobiol. 54, 628-637. 73. Gal R, Libersat, F. (2010). A Wasp Manipulates Neuronal Activity in the SubEsophageal Ganglion to Decrease the Drive for Walking in Its Cockroach Prey. PLoS ONE 5(4): e10019. 74. Libersat, F., Delago, A., Gal, R. (2009). Manipulation of host behavior by parasitic insects and insect parasites. Ann. Rev. Entomol. 54, 189-207. 75. Ganesalingam, V.K. (1972). Anatomy and histology of the sense organs of the ovipositor of the ichneumonid wasp, Devorgilla canescens. Journal of Insect Physiol. 18, 1857-1867. 76. Casiraghi, M., Andrietti, F., Bonasoro, F., Martinoli, A. (2001). A note on host detection by Buathra tarsoleuca (Schrank) (Hymenoptera: Ichneumonidae), a parasite of Ammophila sabulosa (L.) and Podalonia affinis (Kirby) (Hymenoptera: Sphecidae). Journal of Insect Behavior 14(3), 299-312. 77. Polidori, C., Federici, M., Mendiola, P., Selfa, J., Andrietti, F. (2010). Host detection and rate of parasitism by Acroricnus seductor (Hymenoptera:

198

78. 79.

80. 81. 82. 83. 84.

85. 86. 87. 88. 89. 90. 91. 92. 93.

94.

Francesco Andrietti

Ichneumonidae), a natural enemy of mud-dauber wasps (Hymenoptera: Sphecidae). Submitted to Animal Biology. Van Marle, J., Piek, T. (1986). Morphology of the venom apparatus. In: Piek, T. (ed.), Venoms of the Hymenoptera, pp. 17-44. Academic Press, London. Moore, E.L., Haspel, G., Libersat, F., Adams, M.E. (2006). Parasitoid wasp sting: a cocktail of GABA, taurine and β-alanine opens chloride channels for central synaptic block and transient paralysis of a cockroach host. J. Neurobiol. 66(8), 811-820. Keasar, T., Sheffer, N., Glusman, G., Libersat, F. (2006). Host-handling behavior: an innate component of foraging behavior in the parasitoid wasp Ampulex compressa. Ethology 112, 699-706. Schaefer, P., Ritzmann, R.E. (2001). Descending influences on escape behavior and motor pattern in the cockroach. J. Neurobiol. 49(1), 9-28. Piek, T. (1985). Insect venoms and toxins. In Comprehensive insect physiology biochemistry and pharmacology, Vol. 11, pp. 595-633. Kerkut, G.A., Gilbert, L.I. Eds. Pergamon Press, Oxford. Roces, F., Gnatzy, W. (1997). Reduced metabolic rate in crickets paralysed by a digger wasp. Naturwissenschaften 84, 362-366. Polidori, C., Gobbi, M., Chatenoud, L., Santoro, D., Montani, O., Andrietti, F. (2010). Taxon-biased diet preference in the ‘generalist’ beetle-hunting wasp Cerceris rubrida provides insights on the evolution of prey specialization in apoid wasps. Biological Journal of the Linnean Society 99(3), 544-558. Mann, J.S., Crowson, R.A. (1983). Phylogenetic significances of the ventral cord in the Chrysomeloidea (Coleoptera: Phytophaga). Systematic Entomology 8, 103119. Calder, A.A. (1989). The alimentary canal and nervous system of Curculionoidea (Coleoptera): gross morphology and systematic significance. Journal of Natural History 23, 1205-1265. Tsuneki, K. (1969). Gleanings on the bionomics of the East Asiatic non-social wasps (Hym.). I. Some species of Oxybelus (Sphecidae). Etizenia 38, 1-24. Tinbergen, N. (1984). Curious naturalists. University of Massachusetts Press, Massachusetts. Piek, T., Dunbar, S.J., Kits, K.S., Marle J., Wilgenburg, H. (1985). Philanthotoxins : a review of the diversity of actions on synaptic transmission. Pestic. Sci. 16, 488-494. Idriss, M. 1991. Philanthotoxin: a new model for an additional class of insecticides. J. King Saud Univ. 3, Agric. Sci. (1), 95-108. Eberhard, W.G. (2000). Spider manipulation by a wasp larva. Nature 406, 254-255. Malyshev, S.I. (1968). Genesis of the Hymenoptera. Methuen, London. Portman, S.L., Frank, J.H., McSorley, R., Leppla, N.C. (2009). Fecundity of Larra bicolor (Hymenoptera: Crabronidae) and its implications in parasitoid: host interaction with mole crickets (Orthoptera: Gryllotalpidae: Scapteriscus). Florida Entomologist 92(1), 58-63. Field, J. (1989). Intraspecific parasitism and nesting success in the solitary wasp Ammophila sabulosa. Behaviour 110(1-4), 23-46.

Transworld Research Network 37/661 (2), Fort P.O. Trivandrum-695 023 Kerala, India

Predation in the Hymenoptera: An Evolutionary Perspective, 2011: 199-216 ISBN: 978-81-7895-530-8 Editor: Carlo Polidori

7. The role of increased prey spectrum and reduced prey size in the evolution of sociality in Cerceris wasps Carlo Polidori Dipartimento di Biologia, Sezione di Zoologia e Citologia, UniversitĂ degli Studi di Milano via Celoria 26, 20133 Milano, Italy

Abstract. Two recent hypotheses have linked resource usage to the evolution of sociality in Hymenoptera. One hypothesis (Expanded Food Choice hypothesis) states that the spectrum width of food resources collected by adults for the brood positively correlates with the degree of vespid sociality. The other hypothesis (Reduced Food Size hypothesis) states that a reduction of size of resource exploited (from prey to pollen) favoured cooperation in Apoidea. I tested, using data available in the literature, if enlarging prey spectrum and/or reducing prey size are linked to social evolution in the genus Cerceris, which includes both solitary and primitively social beetle-hunting digger wasps. I found that social species of Cerceris hunt, on average, about twice (up to 6 times) the number of beetle families hunted by solitary species. Moreover, interestingly, the composition of beetle prey families seems to depend more on social organization than on phylogenetic relationships among wasp species. Social species are smaller than solitary ones. Despite the fact that the data did not allow the direct Correspondence/Reprint request: Dr. Carlo Polidori, Dipartimento di Biologia, Sezione di Zoologia e Citologia, UniversitĂ degli Studi di Milano, via Celoria 26, 20133 Milano, Italy E-mail: carlo.polidori@unimi.it

200

Carlo Polidori

comparison of the relative prey size between solitary and social species, a positive correlation between the number of prey families and the number of prey/cell indirectly suggests that social Cerceris tend to fill the brood cells with smaller prey. In support of this, I also found a marginal difference in the maximum number of prey/cell between the two groups. Thus, enlarging prey spectrum seems to parallel social evolution in Cerceris wasps, while new studies are necessary to confirm the preliminary suggestion that the relative prey size correlates with the social organization.

Theoretical background: Studying the ecological conditions that might favour social evolution One of the main problems related to the study of social evolution in Hymenoptera is to understand which are the adaptive advantages conferred by living in groups [1, 2, 3]. Understanding these advantages may help elucidate the selective evolutionary pressures which led to sociality. Genetic relatedness seems to have been important in promoting cooperative behaviour [4], as well, and in some cases most important, are ecological pressures; these include a better exploitation of finite resources such as nests (e.g. the digger wasp genus Cerceris) [5] and a better defence against natural enemies (e.g. allodopine bees and halictine bees, but also, again, Cerceris) [6, 2, 7]. Moreover, energetic efficiency is undoubtedly favoured by social cooperation [8]. So it seems that economic factors could be important in addition to genetic ones in the origin and maintenance of social behaviour [9]. One of these economic factors could be the exploitation of food for brood provisioning. Two recent hypotheses stress the ecological conditions that may have favoured social evolution, and both are related with food resource use. One hypothesis (Expanded Food Choice hypothesis, [10]) states that the spectrum width of food resources collected by adults for the brood positively correlates with the degree of sociality, and applies to vespid wasps (and it is strongly suggested in bees). The other hypothesis (here named Reduced Food Size hypothesis, [11]) states that a reduction of size of resource exploited (from prey to pollen) promoted cooperation in Apoidea (see below). Within a diverse (about 8000 species) group of Apoidea (Crabronidae) known to include both solitary and social species, such hypotheses have not been tested. This contrasts with the huge amount of information recorded in the last 100 years on their predatory habit and the choice of prey [reviews in 12, 13, 14, 15, 16]. In this paper I try to test the above-cited hypotheses through a comparative analysis of data available in literature on the genus Cerceris.

Prey use and sociality in Cerceris wasps

201

Links between sociality and prey taxonomic specialization: The Expanded Food Choice hypothesis Computer simulations suggest that social behaviour is unstable unless it provides important economic benefits and fitness gains to the individuals [17]. In this respect, we know that social behaviour favours energetic efficiency [8], but it might have many other advantages. Among the factors that may determine the evolution of castes in social insects, the distribution of food items in space and time and the size distribution of the items seem important [18]. Thus, the adaptive value of social species may increase by increasing the range of prey. This argument was used by Da Silva and Jaffe [10] to formulate an intriguing hypothesis called Expanded Food Choice, which states that, in a given taxon, social species should have a broader prey spectrum than solitary ones. The literature supports the view that a relationship between polyphagy and social behaviour exists. In vespid wasps (Vespidae), solitary Eumeninae hunt, on average, fewer orders of insects than social Polistinae and Vespinae [10]. In bees sensu stricto (i.e. pollinivorous Apoidea), members of eusocial Apinae and Meliponinae, for example, forage on a larger variety of flowering plants than solitary or communal species of, e.g., Andrenidae [19, 20, 21]. If this hypothesis is also valid for Cerceris wasps, one would predict that those solitary members of the genus hunt, on average, for a fewer number of beetle families than social species.

Links between sociality and prey size specialization: The Reduced Food Size hypothesis Brood provisioning by adult female hymenopterans might be expensive, so that cooperation may be promoted to save energy [22]. Strohm and Liebig [11] started with a comparison of several behaviouralecological traits between bees sensu strictu and apoid wasps to formulate hypotheses regarding the major distribution of eusocial species among the first group when compared to the latter. They first underlined that many components of the behaviour are very similar between these two groups. With the exception of parasitic species, all provision brood cells (some with mass provisioning, others with progressive provisioning), most species carry their larval provisions to a nest in flight, the nest types are also very similar, and both groups suffer from kleptoparasites and parasitoids [14, 19]. However, the authors noted that there is one important difference between apoid wasps and bees: the type of resources they use as food for the

202

Carlo Polidori

larvae (paralyzed arthropods versus pollen and nectar). The authors argued that this difference in the type of resource might have important consequences on the probability to evolve sociality. In particular, the size of the resource item used as larval provision is much larger in apoid wasps (relative to body mass) compared to bees. For example, forager bees can decide how much pollen and nectar to collect and transport to the nest, so that the load can be optimized with regard to carrying capacities. In apoid wasps, by contrast, the prey has to be transported to the nest as a whole. Thus, larger size of provisioning females would be promoted in apoid wasps, and this would contrast to the fact that, in social Hymenoptera, workers are often smaller than the foundresses [14, 22]. On the other side, in bees, all the individuals are capable of gathering pollen at a high rate. Thus, a reduction of food mass relative to body mass would favour cooperation (Reduced Food Size hypothesis). As an evidence for this hypothesis, the authors cited the case of the few highly social apoid wasps (Microstigmus, Spilomena [23, 24]). Consistent with their idea, these species hunt on very small prey. Not surprisingly, a similar trend can be seen in the vespid wasps. There are many species of (solitary) eumenines that provision brood cells with entire paralyzed arthropods; whereas the (eusocial) polistine, vespine and stenogastrine wasps chop their prey up before they transport it to the nest [19, 25]. If this hypothesis is valid for Cerceris wasps, one would predict that those social members of the genus hunt, on average, for smaller beetles (relative to wasp size) than solitary species.

Overview of Cerceris waspsâ&#x20AC;&#x2122; biology Apoid wasps comprise a paraphyletic group of mostly hunting Hymenoptera, and they form, together with the monophyletic group of the bees sensu strictu, the superfamily Apoidea [26]. The largest genus of apoid wasps is the genus Cerceris. This belongs of the family Crabronidae and consists of more than 900 known species from all continents and many large islands [12]. Nests of Cerceris wasps are dug in compact soils and are always multicellular, with the main burrow moderately (10-30 cm, in most species) to extremely deep (up to 1.3 m) [12, 27]. These underground brood cells are mass provisioned with only two orders of insects (only adult individuals are hunted): Hymenoptera (a few species) and Coleoptera (in most cases) [12]. Hymenoptera were suggested to be the primitive prey for the genus [28]. There were recorded 19 coleoptera families used as prey [reviewed in 28, 29,

Prey use and sociality in Cerceris wasps

203

30, 31], while Hymenoptera chosen as prey are mostly other Apoidea [28]. Size, availability and autoecology of prey seem to be important factors affecting prey selection at population and individual level [28, 30, 32, 33]. Although most of beetle-hunting species typically hunt for a single coleopteran family, cases of more than 1 used prey family are known [30, 31]. It has been suggested that this general resource partitioning (different prey families for different wasp species) arose to avoid competition for provisions, since many times different Cerceris species were observed nesting in the same restricted areas [e.g. 34, 35]. Within such areas, commonly females are engaged in intra-specific parasitism, usually nest usurpation [36, 37]. The genus Cerceris shows an interesting variety of social behaviours, including solitary and social species, although the level of sociality is still debated [35, 38, 39]. When nest sharing does occur, various kinds of female associations have been observed: young wasp-old wasp [40], motherdaughter and sister-sister [38, 41]. In C. antipodes, usually two related females share a nest with the possible benefit of an increased protection against ants, parasitoids and usurpation by conspecifics [7]. Due to the fact that studies on the social members of the genus are scarce [10 species reviewed in 35] and because division of labour, caste differentiation and reproductive skew were only marginally investigated to date [42, 43, 35, 44], nest-sharing Cerceris are still generally referred as communal [43, 44]). However, increasing evidence suggests that nest-sharing Cerceris possess a more complex social organization, and possibly eusociality (C. australias - [43], C. antipodes - [38, 41, 42], C. rubida - [39]). Most social Cerceris were observed in Australia, with very few reports from Southern Europe and the Americas (Fig. 1); all social Cerceris provide their brood with beetles [31, 30, 45].

Figure 1. Geographical distribution of the species of Cerceris considered in the analysis. Circles show the number of solitary and social species in each region.

204

Carlo Polidori

Data gathering and analysis A survey of the literature on prey selection and nesting biology was carried out for the genus Cerceris. For a valid selection of social species, I considered only the ones for which data were unambiguous. In particular, I scored as â&#x20AC;&#x153;socialâ&#x20AC;? the species which show: 1) more than the 50% of studied nests including >1 female, and/or 2) nests with > 1 female guarded by one of them which discriminate nestmates from non-nestmates (e.g. C. rubida), and/or variance in reproductive capability, high genetic relatedness and cooperation in brood care were recognized (e.g. C. antipodes, C. australis). For several species nest sharing is just a short-term nest co-provisioning [45, 46], so that those species were scored as solitary. On the whole, 12 social and 52 solitary species were selected for the analysis, covering 5 biogeographical regions (Table 1 and Fig. 1). All these species hunt for beetles, so that differently from Da Silva and Jaffe [10], which used the number of arthropod prey orders to define prey spectrum while testing the Expanded Food Choice hypothesis, I used the number of insect families as the degree of prey hunting specialization. The choice to use the family-level analysis of prey does not seem to be much hazardous: for example, in the vespid subfamily Eumeninae, social species were observed to hunt more lepidopteran families than solitary species [47]. In case of more than one population studied, the maximum number of beetle prey families recorded was used for the species. Data available on size of both wasps and prey were scarce: I obtained them for 2 social species and 22 solitary species. Thus, I could not compare directly this measure between the two groups. Instead, I considered two variables presumably linked with relative prey size: the maximum and mean number of prey per cell. In fact, if prey is small relative to wasps, more prey items should be necessary to completely feed a larva (placed in a single cell). In addition, a third variable also possibly related to sociality, the maximum number of cells per nest, was considered for some species. In fact, if more than 1 wasp is present in a single nest, one expects a higher number of cells constructed in a nest during the nesting cycle. One has to note that all these variables represent very rough surrogates of the relative size of prey, and that the variance is probably increased also by the fact that not always all the cells in a nest are dug during observerâ&#x20AC;&#x2122;s examination and the cells were not always fully provisioned during examination. However, such bias should be similar for both solitary and social species. In addition, the size of wasps did not always come from the studied populations, so that I had to use in many cases the wasp body lengths provided by taxonomic keys [48, 49, 50, 51, 52].

Table 1. List of solitary and social Cerceris species and related study areas used in the analysis.

Prey use and sociality in Cerceris wasps 205

Table 1. Continued

206 Carlo Polidori

Table 1. Continued

Prey use and sociality in Cerceris wasps

207

208

Carlo Polidori

Statistical analysis was carried out with non-parametric comparisons of medians (Mann-Whitney test), due to the characteristics of the data samples (discrete distribution and/or non-homogeneous variance). A cluster analysis was performed through the Wardâ&#x20AC;&#x2122;s method reporting Euclidean distance (dissimilarity) between pairs of species. The variable used to determine the clusters were the presence/absence of the beetle families used as prey for each species. This analysis produced a dendrogram and reported the dissimilarity value (truncation) which likely determines how many clusters best suit the data. Average numbers are given Âą standard deviation in the text.

Results Are Cerceris wasps in agreement with the Expanded Food Choice hypothesis? Solitary species hunt, on the whole, for 8 prey families, while social species for 11 prey families (Fig. 2). Comparing the prey spectrum of the 12 social species with that of the 52 solitary species, it resulted that social Cerceris used more prey families (range: 1-6, median: 2, mean: 2.4Âą1.5) than

Figure 2. Families of beetles used as prey by solitary (in the figure, as an example, C. arenaria) and by social (in the figure, C. rubida) species of Cerceris. Beetle families are grouped by superfamilies (on the right).

Prey use and sociality in Cerceris wasps

209

solitary ones (range: 1-2, median: 1, mean: 1.2Âą0.4) (Mann-Whitney test: U=505, P<0.0001). Therefore, the Expanded Food Choice hypothesis seems to be valid for the genus Cerceris. Interestingly, not only the number of prey families, but also the prey types differ somewhat between solitary and social species of the genus. In fact, in this sample, four beetle families appeared to be exclusively hunted by social species, while three were hunted only by solitary species (Fig. 2). No bias towards particular beetle superfamilies were detected (Fig. 2). This finding is mirrored in the cluster analysis; it revealed three clusters of species: one included most of the social species (58.3%), and was more distant to the other two, which comprised mostly solitary species (with the 16.6% and 25% of social species included) (Fig. 3). Notably, most of the social species (7) form a compact sub-cluster in one group, together with only two solitary species (C. clypeata clypeata and C. blakei), which are among the rare solitary species hunting for 2 beetle families. On the contrary, the few other social species were placed together with solitary groups probably because they represent the rare species hunting only one prey family. An effect of wasp phylogeny on prey composition seems unlikely. For example, C. antipodes, C. goddardi, C. australis and C. gilberti, all in the australisspecies group, were distributed in all the three clusters, while the species of the minuscula-group (C. minuscula, C. anthicivora and C. windorum) are placed in two of the clusters (Fig. 3). At last, two alternative hypotheses regarding variation in number of prey families may be excluded. In fact, such variation may be due either to the size of females (larger species may be advantaged in hunting a wider spectrum of prey) or to some geographical bias. Results suggest that both these possibilities are unlikely. First, the correlation between the body length of the females and the number of prey families was negative and not positive (Pearson correlation test: r=-0.36, n=39, P=0.026) (because social species were marginally smaller than solitary ones, see below); second, in Australia, i.e. the geographical region including the majority of social species, solitary species all hunt for single prey families (versus about 2.1 of social ones), while C. rubida, the only European social species, was also the only one hunting for more than 1 prey family (6 families) in the Paleartic region.

Are Cerceris wasps in agreement with the Reduced Food Size hypothesis? Wasp body length was 9.32Âą1.85 mm (range: 7.5-12.5 mm, n=9) for social species and 12.84Âą3.56 mm (range: 8.3-21 mm, n=30) for solitary ones,

210

Carlo Polidori

Figure 3. Dendrogram resulting from the cluster analysis, based on the presence/ absence of prey families in the Cerceris speciesâ&#x20AC;&#x2122; diet. Filled black circles indicate social species.

Prey use and sociality in Cerceris wasps

211

this difference being significant (Mann-Whitney test: U=47.5, P=0.002). The species which have the lowest wasp/prey size ratio were solitary: C. cubensis (0.23), C. callani (0.27), C. blackey (0.29) and C. latifrons latifrons (0.31). However, the social C. rubida also had a low ratio (0.32). This species was observed to fill each brood cell with up to 60 beetles [54, Polidori unpublished data] and to compact them in the cell in a way resembling the “collembolan balls” of Microstigmus comes Krombein [23] (Polidori et al. unpublished data). When mature larva pupates, it is covered with high numbers of small prey remains (mainly beetle elytrae) (Fig. 4). On the other extreme, two solitary species have size ratio close to 1 (C. fumipennis (0.95), C. bupresticida (1.08)), while the third highest was a social species (C. watlingensis (0.75)). The maximum number of prey/cell was marginally higher in social species (range: 9-60, median: 19.5, mean: 29±20, n=10) than in solitary ones (range: 6-51, median: 11, mean: 17.2±12.4, n=24) (Mann-Whitney test: U=172.5, P=0.047). One notes, also, that in about 30% of social species this value was above 50, while this occurred in only one solitary species. The mean number of prey/cell did not differ significantly between the two groups (social: range: 6.5-38.4, median: 11.8, mean: 17±12.2, n=10; solitary: range: 3.5-46, median: 8.5, mean: 13.5±11.4, n=22) (Mann-Whitney test: U=150, P=0.1). Interestingly, there is a correlation between the number of prey families and both the maximum (Pearson correlation test: r=0.51, n=37, P<0.001) and the mean (Pearson correlation test: r=0.57, n=32, P<0.001) number of prey/cell (Fig. 5). These results tentatively suggest indirectly that social species

Figure 4. A typical pupa of the social species, Cerceris rubida, covered with small prey remains.

212

Carlo Polidori

Figure 5. Relationships between the number of prey families and the mean (n=32 species) and the maximum (n=37 species) of prey/cell.

(which hunt more prey families) tend to reduce the relative prey size (i.e. they tend to fill cells with more prey items). The maximum number of cells/nest did not differ between the two groups of species (social: range: 12-231, median: 37, mean: 84.3Âą89.7; solitary ones: range: 5-31, median: 18, mean: 18.5Âą11.5) (Mann-Whitney test: U=47, n1=9, n2=6, P=0.08). However, very high number of cells/nest were recorded only for social species (C. australis (179), C. antipodes (196), C. goddardi (231)), while solitary species did not exceed 31 cells/nest (C. megacantha). This weak difference, thus, may be due to the small sample size, considering that minimum, maximum, median and mean values were all lower for solitary species. Thus, at the moment I can at least suggest that small prey size is associated with sociality in Cerceris wasps.

Conclusions This analysis supports the Expanded Food Choice hypothesis, i.e. a relationship between polyphagy and social behavior exists in the genus

Prey use and sociality in Cerceris wasps

213

Cerceris. Da Silva and Jaffe [10] stressed the importance to be cautious in assessing this hypothesis, because the relationship between sociality and polyphagy is not necessarily correlated in other taxa. However, they underlined how in bees sensu strictu a similar relationship is strongly suggested by a huge amount of data on pollen spectrum of both solitary and social species [e.g. 19, 20, 21]. I therefore add here new evidence that this pattern of sociality-resource spectrum relationship is possibly widespread in Hymenoptera. Also interestingly, the composition of beetle prey families seems to segregate species depending more on social organization than on phylogeny. This would suggest that a shift towards new beetle groups occurred during evolution. Prey shifting may be associated with social evolution: in the social genus Microstigmus, for example, prey belong to Collembola and Thysanoptera, which are poorly represented insects among prey of solitary Pemphredoninae, the subfamily it belongs (for which Homoptera is the common prey taxon) [23, 84]. With regard to the Reduced Food Size hypothesis, it can at least be suggested, with the available data, that prey size is negatively associated with social complexity. This relationship seems to be supported for the genus Cerceris when an indirect measure of relative prey size (maximum number of prey/cell) is compared between solitary and social species. In support of this suggestion, the highest values of the number of prey per cell (in many social species more than 50, up to 60) is not so far from what is found for a social species of Microstigmus (about 80), in particular in the light of the general much lower values reported for crabronid wasps [e.g. 13, 14]. However, new records on prey and wasp sizes are necessary to confirm this hypothesis for Cerceris wasps.

Acknowledgements I am indebted to Allan Hook, which kindly helped in collecting some information on Cerceris prey and revised an early version of the manuscript. Klaus Jaffe and Erhard Strohm kindly gave important suggestions on data analysis and discussion of the results.

References 1. 2. 3. 4.

Forbes, S.H., Adams, R.M.M., Bitney, C., and Kukuk P.F. 2002, J. Ins. Sci., 2, 22. Smith, A.R., Wcislo, W.T., and Oâ&#x20AC;&#x2122;Donnell, S. 2003, Behav. Ecol. Sociobiol., 54, 22. Dunn, T., and Richards, M.H. 2003, Behav. Ecol., 14, 417. Foster, R.F., Wenseleers, T., and Ratnieks, F.L.W. 2006, TREE, 21, 57.

214

5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24.

25. 26. 27. 28. 29. 30. 31. 32. 33.

Carlo Polidori

McCorquodale, D.B. 1989a, Ecol. Entomol., 14, 191. Schwartz, M.P., Bull, N.J., and Hogendoorn, K. 1998, Insect. Soc., 45, 349. McCorquodale, D.B. 1989b, J. Ins. Behav., 2, 267. Fonk, C., and Jaffe, K. 1996, Phys. Behav., 59, 713. Aratui, H.S., and Gadagkar, R. 1998, Oecologia, 117, 295. da Silva, R.E, and Jaffe, K. 2002, Sociobiology, 39, 25. Strohm, E., and Liebig, J. 2008, Ecology of Social Evolution, J. Korb and J. Heinze (Eds.), Heidelberg: Springer; 109. Bohart, R.M., and Menke, A.S. 1976, Sphecid Wasps of the World. A generic revision. University of California Press, Berkeley, Los Angeles, London. Evans, H.E., and O’Neill, K.M. 1988, The natural history and behavior of North American beewolves. Cornell University Press, Ithaca, NY. O’Neill, K.M. 2001, Solitary wasps: behaviour and natural history. Cornell University Press, Ithaca, NY. Iwata, K. 1976, Evolution of instinct. Comparative ethology of Hymenoptera, Amerind Publishing, New Delhi, India. Evans, H.E., and O'Neill, K.M. 2007, The sand wasps: natural history and behavior. Harvard University Press, Cambridge, Massachusetts. Jaffe, K. 2001, Acta Biotheoretica, 49, 29. Oster, G.F., and Wilson, O.W. 1978, Caste and ecology in the Social Insects. Princeton University Press, New Jersey. Michener, C.D. 1974, The social behavior of the bees. A comparative study. Harvard University Press, Cambridge, Massachusetts. Larkin, L.L., Neff, J.L., and Simpson, B.B. 2008, Apidologie, 39,133. Heard, T.A. 1999, Annu. Rev. Entomol., 44, 183. Ross, K.G., and Matthews, R.W. (Eds.) 1991, The Social Biology of Wasps. Cornell University Press, Ithaca, NY. Matthews, R.W. 1968, Psyche, 75, 23. Melo, G.A.R. 2000, Ecologia e comportamento de Insectos. Série Oecologia Brasiliensis, 8., R.P. Martins, T.M. Lewinsohn, and M.S. Barbeitos (eds.), PPGE-UFRJ, Rio de Janeiro, Brasil, 85. Raveret Richter M. 2000, Annu. Rev. Entomol., 45, 121. Melo, G.A.R. 1999, Scientific Papers, Natural History Museum, the University of Kansas, 14, 1. Giovanetti, M. 2005, Neotrop. Entomol., 34, 713. Gess, F.W. 1980, Annals of the Cape Provincial Museums. Nat. Hist., 13, 85. Polidori, C., Boesi, R., Pesarini, C., Papadia, C., Bevacqua, S., Federici, M., and Andrietti, F. 2007, Zool. Stud., 46, 83. Polidori, C., Gobbi, M., Chatenaud, L., Santoro, D., Montani, O., and Andrietti, F. 2010, Biol. J. Linn. Soc., in press. Evans, H.E., and Hook, A.W. 1986, J. Nat. Hist., 20, 1297. Polidori, C., Boesi, R., Isola, F., and Andrietti, F. 2005, Eur. J. Entomol., 102, 801. Polidori, C., Santoro, D., Asís, J.D., and Tormos, J. 2011, Predation in the Hymenoptera: An Evolutionary Perspective (C. Polidori, Ed.). Transworld Research Network publishing (in prep.).

Prey use and sociality in Cerceris wasps

215

34. Polidori, C., Casiraghi, M., Di Lorenzo, M., Valarani, B., and Andrietti, F. 2006, J. Ethol., 24, 155. 35. Polidori, C., Federici, M., Papadia, C., and Andrietti, F. 2006, It. J Zool., 73, 55. 36. Field, J., and Foster, W.A. 1995, Anim. Behav., 50, 99. 37. Polidori, C., and Andrietti, F. 2006, Sociobiology, 47, 455. 38. McCorquodale, D.B. 1989c, Insect Soc., 36, 42. 39. Boesi, R., and Polidori, C. 2011, Aggress. Behav., 37, doi: 10.1002/ab.20398. 40. Elliot, N.B, Shlotzhauer, T., Elliot, W.M. 1986, Ann. Entomol. Soc. Am., 79, 994. 41. McCorquodale, D.B. 1988, Behav. Ecol. Sociobiol., 23, 401. 42. McCorquodale, D.B. 1990, Ethol. Ecol. Evol., 2, 345. 43. Evans, H.E., and Hook, A.W. 1982, Aust. J. Zool., 30, 557. 44. Alcock, J. 1980, Aust. J. Entomol., 19, 223. 45. Hook, A. 1987, Sociobiology, 13, 93. 46. Evans, H.E., and Hook, A.W. 1986, Sociobiology, 11, 275. 47. Itino, T. 1992, Res. Pop. Ecol., 34, 203. 48. Pagliano, G., and Negrisolo, E. 2005, Fauna d’Italia, Vol. 40: Hymenoptera Sphecidae, Calderini, Bologna, Italy. 49. Bitsch, J., Barbier, Y., Gayubo, S.F., Schmid, K., and Ohl, M. 1997, Hyménoptères Sphecidae d'Europe occidentale, Fédération Française des Sociétés de. Sciences Naturelles, París, France. 50. Evans, H.E. 1982a, Trans. Am. Entomol. Soc., 107, 299. 51. Evans, H.E. 1988, Trans. Am. Entomol. Soc., 114, 1. 52. Scullen, H.A. 1972, Smithsonian Contrib. Zool., 110, 1. 53. Grandi, G. 1944, Mem. Acc. Sci. Ist. Bol., 1, 63. 54. Callan, E.McC. 1987, EOS - Rev. Esp. Entomol., 63, 101. 55. Evans, H.E. 1971, J. Kansas Entomol. Soc., 44, 500. 56. Scullen, H.A., and Wold, J.L. 1969, Ann. Entomol. Soc. Am., 62, 209. 57. Tsuneki, K. 1965, Etizenia, 9, 1. 58. Asís, J.D., Gayubo, S.F., and Tormos, J. 1991, Rev. Chil. Entomol., 19, 5. 59. Evans, H.E., and Rubink, W.L. 1978, Great Basin Naturalist, 38, 59. 60. Callan, E.McC. 1990, Entomologist, 109, 194. 61. Krombein, K.V. 1963, Bull. Brooklyn Entomol. Soc., 58, 72. 62. Fabbri, R. 1999, Quad. Studi Nat. Romagna, 12, 21. 63. Fabre, J.H. 2002, The Hunting Wasps. University Press of the Pacific, Honolulu, Hawaii. 64. Linsley, E.G., and MacSwain, J.W. 1956, Ann. Entomol. Soc. Am., 49, 71. 65. Genaro, J.A., and Sanchez, C.S. 1993, Carib. J. Sci., 29, 39. 66. Elliot, N.B, and Elliot, W.M. 1987, J. Kansas Entomol. Soc., 60, 397. 67. Kurczewski, F.E., and Miller, R.C. 1984, Flo. Entomol., 67, 146. 68. Krombein, K.V. 1964, Am. Mus. Novit., 2201, 1. 69. Alcock, J. 1974, J. Nat. Hist., 8, 645. 70. Marshall, S.A., Paiero, S.M., and Buck, M. 2005, Canadian Entomol., 137, 416. 71. Hook, A.W., and Evans, H.E. 1991, J. Kansas Entomol. Soc., 64, 257. 72. Evans, H.E. 1983, Entomol. News, 94, 45. 73. Byers, G.W. 1978, J. Kansas Entomol. Soc., 51, 818.

216

74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84.

Carlo Polidori

Asis, J.D., Tormos, J., and Gayubo, S.F. 1997, Ann. Soc. Entomol. France, 32. Evans, H.E. 1992, J. Kansas Entomol. Soc., 65, 91. Evans, H.E. 2000, J. Kansas Entomol. Soc., 73, 220. Evans, H.E, Matthews, R.W., Alcock, J., and Fritz, M.A. 1976, J. Kansas Entomol. Soc., 49, 126. Alexander, B.A., and Asís, J.D. 1997, J. Insect Behav., 10, 871. Lin, C.S. 1967, Proc. Ent. Soc. Wash., 69, 312. Salbert, H.A., and Elliot, N. 1979, Ann. Entomol. Soc. Am., 72, 591. Elliot, N.B., Elliot, W.M., and Salbert, P. 1981, Ann. Entomol. Soc. Am., 74, 127. Minkiewicz, R. 1933, Polsk. Pismo Entomol, 11, 98. Genaro, J.A. 2009, Solenodon, 8, 99. Asís, J.D. 2003, J. Insect Behav., 16, 49.

Transworld Research Network 37/661 (2), Fort P.O. Trivandrum-695 023 Kerala, India

Predation in the Hymenoptera: An Evolutionary Perspective, 2011: 217-245 ISBN: 978-81-7895-530-8 Editor: Carlo Polidori

8. Predators as prey: Top-down effects on predatory Hymenoptera Heike Feldhaar Behavioural Biology, University of Onsabr端ck, Barbarastrasse 11, D-49076, Osnabr端ck, Germany

Abstract. Predation and parasitism may incur high fitness losses to predatory Hymenoptera. Defensive traits have thus evolved for personal defence as well as defence of brood and resources. These traits range from morphological adaptations, chemical weaponry, nests build as fortresses, to behavioural defences performed by individuals or collectively including strategies involving recruitment. Predation may have direct effects by eliciting defensive behaviours and alterations in spatial and temporal foraging patterns or may affect Hymenoptera indirectly when dominance hierarchies of competing species are reversed in the presence of predators. In addition, predation may strongly influence social structure in Hymenoptera, facilitating aggregated breeding or sociality as well as the formation of morphological castes. Defensive strategies and foraging strategies may be interdependent in predatory Hymenoptera. The resources available to an individual or colony may constrain the kind of defences that can be realized. Behavioural strategies such as recruitment may also play a role in both contexts, foraging and defence. Here I present an overview of the most important predators and their impact on predatory Hymenoptera. I compare defensive strategies of solitary and social Hymenoptera. Potential trade-offs and constraints of defensive strategies are discussed. Correspondence/Reprint request: Dr. Heike Feldhaar, Behavioural Biology, University of Onsabr端ck Barbarastrasse 11, D-49076, Osnabr端ck, Germany. E-mail: feldhaar@biologie.uni-osnabrueck.de

218

Heike Feldhaar

Introduction In spite of the possession of potent weaponry that often comprises a sting, venom and strong mandibles, predatory Hymenoptera are themselves a favoured prey item for a wide variety of predators including other Hymenoptera as well as larger predators such as vertebrates. Impact of predation may vary greatly in dependence of the degree of sociality, life-stage that is preyed upon and the predator itself. In solitary Hymenoptera predation on either the adult itself or its brood will most often lead to a substantial reduction or total loss of fitness. In contrast, when only a few workers are lost from a large colony of ants or wasps due to a small predator, fitness losses may be minimal. However, as social Hymenoptera are a high quality food source due to the high local concentration of workers and their brood, they are an attractive food source for larger predators such as colonies of army ants, mammals, lizards or birds. These larger predators can again be detrimental to the whole colony. In addition, specialized parasitoids can incur high fitness losses by exploiting nests of social Hymenoptera over longer periods of time. High fitness losses due to predation or parasitism have resulted in the evolution of a suit of defence traits aimed either at personal defence or defence of the brood and resources collected to provision the brood. Defensive traits range from morphological traits such as spines or a rigid cuticle conferring passive defence [1,2,3], chemical traits such as venoms [1,4,5,6] or active defensive behaviours either by individuals or collective defence strategies in social species [1,2,3,7,8,9,10,11]. These active behaviours comprise recruitment behaviour in case of an attack [1,9,12,13], aggressive behaviours directed at a potential predator [1,2,14,15] as well as altered foraging strategies [16,17,18] or elaborate nest architecture to avoid predators [19,20,21,22]. Predators and parasitoids have also been a strong extrinsic selective pressure promoting nest aggregations and sociality and the development of castes in predatory Hymenoptera [23,24,25,26,27,28]. Aside from direct effects, predators may also mediate indirect negative effects on their prey. A change in foraging strategy in order to avoid predators may result in changes in competitiveness in the focal species, e.g. species dominance hierarchies within communities may be shifted [16,29,30]. In this chapter I first give a short overview over the range of predators and parasitoids and their impact on predatory Hymenoptera. Then defensive strategies of solitary and social Hymenoptera are compared and evolutionary trajectories selected for by predators discussed.

Effects of predators on predatory Hymenoptera

219

Overview over predators and parasitoids of predatory Hymenoptera Apart from microbial pathogens and internal macroparasites (e.g. nematodes), natural enemies of predatory Hymenoptera comprise predators, parasitoids and cleptoparasites [31]. Whereas predators immediately consume their prey, parasitoids develop within their host over a longer period of time before ultimately killing it [31]. Cleptoparasites affect the Hymenoptera indirectly by stealing prey items or placing their own brood amongst host brood [3]. In both cases the cleptoparasite will reduce the resources available for the host brood, in extreme cases so much so that the host brood will fail to develop. As predators and parasitoids are stronger selective agents due to the generally more severe damage they incur in comparison to cleptoparasites, I will concentrate on the former two and will for the sake of simplicity assume that defence mechanisms that have evolved to ward off predators and parasitoids may be efficient against cleptoparasites too. I will also exclude socially parasitic ants and wasps that may usurp colonies of congenerics as these topics have been covered in detail in several excellent recent reviews (e.g. [32,33,34,35]).

Predators Predators comprise a broad range of animal taxa with other arthropods and vertebrates being the most important. Predators of solitary species of wasps are often other arthropods specialized on solitary wasps such as flies, ants, other wasps or spiders [3]. Nonetheless, birds may incur a quite high mortality rate on solitary predatory Hymenoptera [3]. With an increase in group size from solitary to social Hymenoptera, longevity of a nest as well as the number of brood present in the nest increase. Thus, with increasing group size nests of social Hymenoptera become a more predictable and often highly apparent quality food resource for predators in comparison to solitary Hymenoptera. Therefore, social species are also threatened by larger predators such as lizards, mammals, or army ant colonies (which should be considered as large predators due to the biomass of such colonies) that have specialized on this more abundant prey, in addition to a broad range of arthropod predators and parasitoids. Depending on whether brood or workers are preyed upon, adaptations of the specialized predators vary. In contrast to the brood, palatability of adult workers is usually reduced due to venomous gland secretions and often a sting that can be used for personal defence.

220

Heike Feldhaar

Vertebrate predators Mammals Small mammals such as shrews, bats and mice as well as birds and lizards have been observed to prey on predatory Hymenoptera. As relative abundance of individuals of solitary species within a given habitat is often very low in comparison to social species, solitary species tend to be preyed upon by generalist vertebrate predators. In solitary species the impact of predation due to vertebrates is often not known and most observations of birds or lizards attacking solitary wasps are anecdotal [3], most probably owing to the difficulties in quantifying predation rate. In social species, vertebrate predation seems to have a much larger impact due to their high local abundance (in terms of individual workers and brood items) attracting generalist predators, such as bears or chimpanzees, or vertebrate predators that have specialized on social predatory Hymenoptera as prey items [36,37,38]. Larger generalistically feeding vertebrates such as bears may consume large numbers of workers and brood in times of shortage of other food sources, e.g. in spring [39]. Predation by vertebrates on whole nests can lead to high fitness losses in social Hymenoptera. In ground nesting species, the predators tend to open the nests (usually by digging) and can then either try to reach the brood or consume large numbers of workers that will pour out of the nest in order to defend it. The proportion of destroyed nests within a given habitat may be very high. Thus up to a third of all Formica mounds were found to be destroyed at least once by bears in central Sweden [39] and almost all of the 140 colonies of Iridomyrmex purpureus meat ants examined in a two year period showed signs of damage due to echidnas in a study site in Australia [40]. In the latter study however, impact of this high predation rate on colony survival and growth was not severe as most colonies were still found in the consecutive season. Nests in twigs can likewise be opened by small mammals such as squirrels, birds or monkeys and ants as well as brood may then be consumed [41]. Again, the percentage of trees harbouring ant colonies showing signs of damage by these predators may be as high as 70 percent [41] but the impact of predation on colony mortality is not known. Mammalian (including marsupial) predators specialized on social insects such as anteaters and aardvarks show only limited specialization on particular species [36,37], but may prefer species with higher palatability, e.g. with a lower content of formic acid or a less potent sting [36,39]. Rather, abundant species with large nests are preferred within a given habitat.

Effects of predators on predatory Hymenoptera

221

Birds Birds have been observed frequently to prey upon adult predatory Hymenoptera, which are either caught in flight or on the ground. Although data on predation rates are rather scarce, the widespread aposematic colour patterns of wasps [3,7,42] are generally assumed to be an adaptation against predation, especially by birds [3,43]. However, the impact of birds as predators is probably strongest when whole nests are attacked. Aerial nests made of carton of social wasps may be knocked down by birds who then feed on the brood that is no longer defended by the adult wasps [19,44,45,46]. This type of bird predation has been reported to be the most common cause for nest failure in Polistes exclamans, with the predation rates being nearly 100 percent in some years [26]. Following bird predation adult wasps build new nests, but they are often not successful as the adults will not survive long enough to be able to raise a new generation [26,45]. Especially primitively eusocial species whose nests comprise relatively few individuals only, do not seem to have effective strategies against bird predation [20]. In addition to aerial nests, twig nests may fall prey to birds that open twigs or thorns with their beaks [41,47]. Lizards and frogs Several frogs and lizards are especially adapted to prey upon predatory Hymenoptera. Specialized ant-eating lizards are known mostly from arid zones worldwide, such as the horned lizards of the genus Phrynosoma [48] of the North American deserts or the Australian thorny devil Moloch horridus [36]. Preference for certain species of ants as prey is on the one hand limited by snout size [36] and on the other hand by energy content of the prey items. Horned lizards show a preference for larger ants, irrespective of the mode of defence (chemical vs. sting), as predation on small workers may not be profitable enough to meet the required food intake [49]. Preferences of horned lizards are also dependent on foraging mode of the ants, with single foragers being preferred as prey in comparison to group foragers or foragers on trails that may vigorously mob lizards [50,51]. Predation-levels on ants due to lizards may be very high locally. Based on the daily number of ants consumed by horned lizards Whitford and Bryant [51] estimated that approximately 70% of the standing crop of harvester ants of the genus Pogonomyrmex may fall prey to horned lizards during a season. Horned lizards are highly adapted towards predation on harvester ants behaviourally as well as physiologically. Prey handling time is very short and ants are

222

Heike Feldhaar

swallowed quickly and incapacitated by a mucus-secretion in the pharynx [52]. In addition, Phrynosoma lizards have developed resistance towards the venom of harvester ants [6]. Wasps and ants may fall prey to frogs and toads feeding as generalists, which is demonstrated by remains of Hymenoptera in their stomach content and observations in the field [53,54]. Among others, several New World poison frogs (Dendrobatidae) have been shown to be specialized ant predators [54,55]. Interestingly, the aposematic species derive the alkaloids responsible for the toxicity of the frogsÂ´ skins from the venomous gland substances produced by their ant prey [55]. Impact of such predation on colonies of social insects has to my knowledge not been measured yet.

Arthropods Arthropod predators comprise a broad range of taxa such as ant-lions (Neuroptera) [56,57], robber flies (Asillidae), beetle larvae or spiders [3,58,59,60]. The prey items of robber flies have been shown to comprise up to 40% of solitary wasps, albeit of the non-predatory Tiphiidae [3]. Ant-lions and spiders are mostly sit-and-wait predators, catching single individuals at a time. Spiders have two different tactics of preying upon Hymenoptera. First, they may weave their nets in proximity of nest entrances and trap individual wasps and ants while entering or leave their nests [58,60] or ambush passing individuals [61,62], sometimes in proximity of the nest [63]. Secondly, a few of the numerous ant-mimicking spiders prey upon their ant model and use myrmecomorphy as a tactic to get close to their prey items unnoticed [59,64]. However, due to the relatively small size of these aforementioned arthropod predators, the impact in terms of fitness losses may be minimal in social predatory Hymenoptera but may be high in solitary species.

Intraguild predation by predatory Hymenoptera Intraguild predation plays a prominent role among predatory Hymenoptera, as ants and other wasp species are often viewed as the most devastating predators. Although predation on solitary wasps is often only reported anecdotally, predation rates on solitary and social species may be equally high [8,19,20,26,65,66,67,68,69,70]. Predation by ants and wasps seems to be highest in tropical regions [20,71], with ants together with birds being among the most important predators in the New World tropics and predatory wasps (especially hornets of the genus Vespa) becoming as important as ants in the Old World tropical regions [19,20].

Effects of predators on predatory Hymenoptera

223

The impact of ants as predators can be disastrous with all brood from a nest raided. Attacks on nests by scouting-and-recruiting ants can be averted by displacing or harassing individual scout ants. In contrast, mass foraging army ants arrive quickly in such large numbers at a nest [67] that defence attempts by wasps and other ants are usually futile [8,68,72,73,74]. In tropical regions army ants have been recognized as major predators of other ants as well as social wasps. Immature stages of social insects are the preferred prey items of most army ant species, whereas a few are more generalistic predators (overview in [75]). In wasps, usually only the brood falls prey to ants as the adult wasps are able to alight and escape, having to leave the brood unattended. In ants, brood as well as workers may be preyed upon, resulting in transient shifts in ant community structure after a raiding event [76,77]. In contrast to the ants, which are often smaller than their prey items and overwhelm their prey by numerical dominance, predatory wasps that have specialized on Hymenoptera as prey such as hornets (Vespa spp.) will usually attack as single individuals and are larger than their prey. Although recruitment during attack is only known for the giant hornet [78], the impact of hornets on the colonies of social wasps may be devastating [20,78]. Intraguild predation by solitary wasps does occur [3,79], albeit most species seem to be specialized on other arthropod prey items (including nonpredatory Hymenoptera such as bees) [3]. Specialization on ants as prey has evolved at least twice within the Crabronidae though [80,81,82]. The philantine Aphilantops frigidus, albeit being a solitary wasp, may incur quite high fitness costs to colonies of its prey species as it provisions its own brood exclusively with alate queens of the genus Formica caught during nuptial flights [81].

Parasitoids and cleptoparasites Insects, especially Diptera, Hymenoptera, Strepsiptera as well as Lepidoptera, comprise the most important parasitoids of solitary and social predatory Hymenoptera. Most often immature stages are attacked [3,31,83], albeit parasitoid flies of the families Conopidae and Phoridae are also known to attack adult workers during activities outside the nest [3,29,30]. Some may also develop in prey items collected for provisioning of the larvae. Parasitoids have evolved numerous strategies to overcome host defences and gain access to their hosts´ nests. By ovi- or larvipositing on foraging adults or prey items carried by them, parasitoids can ensure that their immature stages enter the host’s nests [84,85]. Parasitoids such as “satellite flies” (Miltogrammini and Sarcophagidae) may follow solitary wasps in order

224

Heike Feldhaar

to be able to oviposit them or prey items before the nest is entered [84,85,86]. Other parasitoids toss their eggs towards or into nest entrances – or what they perceive as nest entrances. Thus, some fly parasitoids have been observed to toss their eggs towards empty nests, glass vials or even eyelets of shoes (in [3]). Hymenopteran parasitoids may be able to directly oviposit into larvae or nest cells through the nest material or crevices with their ovipositor, especially Ichneumonid wasps [87]. Another strategy to enter the nest and gain access to the brood is by chemical mimicry. For example lepidopteran caterpillars of the genus Maculinea trick Myrmica ants into picking them up during foraging and carrying them into their nests by mimicking the smell of the ants´ brood [88]. Other parasitoids such as cuckoo wasps may be able to enter nests of their hosts undetected due to chemical mimicry of an adult digger wasp [89]. Parasitoids are often among the most disastrous predators as they may incur very high mortality rates in the brood or complete failure of nests, especially in solitary predatory Hymenoptera [3]. However, even in social Hymenoptera, infestation rates of workers or the brood of workers were found to be as high as 50 percent [90]. A general trend seems to be that solitary predatory Hymenoptera suffer highest mortality rate due to parasitoids. With increasing group size and closed (in contrast to open comb) or underground nests the brood is usually constantly cared for and parasitoids may need chemical mimicry to be socially integrated into a colony. Therefore parasitoid pressure seems to be less dramatic in terms of total fitness losses. In social species with large and less well accessible nests single adult workers may rather be attacked during foraging, such as ant workers being attacked by phorid flies. Cleptoparasites that either intercept prey-laden females before they are able to enter the nests or that enter the host nests in order steal prey there are a common threat to solitary predatory hymenoptera. Aside from other solitary wasps or flies [3,69,91] birds may steal prey from females before they can enter their nests. Small songbirds such as sparrows have been shown to be attracted to aggregations of digger wasps and to intercept about a third of all provisions [92]. Contrary to the long-held notion, birds that follow in the wake of army ant raids (“ant-following birgds”) have a negative impact on the army ants through cleptoparasitism as they may snatch about 30% of the ants´ daily leaf litter arthropod intake [93].

Defence strategies As predation and parasitism can incur high fitness costs, predatory Hymenoptera have evolved a range of defensive strategies comprising

Effects of predators on predatory Hymenoptera

225

morphological, chemical and behavioural adaptations including nest architecture. Selective pressures exerted by predators and parasites have been shown to be important forces driving the evolution of sociality as group defence strategies may be more effective in comparison to individual defences only. Sociality can arise either via the subsocial or the semisocial pathway [94]. Subsocial groups consist of mothers and their offspring, whereas sociality arises from groups of unrelated females along the semisocial pathway. An initial step along the route to subsociality is delayed nest provisioning of solitary species as this facilitates interactions between the mother and the developing larva. In contrast, aggregated nesting of solitary Hymenoptera can be the first step via the semisocial route as it facilitates interactions between nesting females [94]. Both initial steps â&#x20AC;&#x201C; delayed provisioning as well as nest aggregations â&#x20AC;&#x201C; are frequently found in solitary species of predatory Hymenoptera and can enhance resistance to parasites (see below) [23]. Groups or colonies of predatory Hymenoptera are able to apply collective defence strategies in addition to individual defences (e.g. [1]). With increasing group size the defensive effectiveness attained by collective defence strategies may exceed those of the participating individuals strongly. At the same time, collective defence mechanisms on the colony level may increasingly become the target of selection instead of individual defences. Thus, the collective defence mechanisms can be successful on the colony level but they may incur individual mortality.

Passive defences: Temporal and spatial predator avoidance Predators can be avoided either by structural traits such as very small body size, cryptic coloration, mimicry (MĂźllerian and Batesian) or by behavioural traits such as activity patterns that minimize overlap with potential predators in space and time. Thus, ecology and evolution of predatory Hymenoptera in general is strongly influenced not only by competitors but also by predators (see also below community level effects of predation) [2,29]. Times of foraging activity or nest building can be adjusted to activity times of the predators. It is generally assumed that the activity of parasitoids in comparison to their hosts is more strongly constrained by temperature due to the smaller body size of the parasitoids [24]. Thus, the sphecid Philanthus triangulum has been shown to attain temporal enemy-free space by shifting its activity towards the later afternoon hours when relative activity of its chrysidid parasitoid declined [18]. However, such a strategy may only be feasible when threatened by a specific parasitoid species and not a suit of

226

Heike Feldhaar

species that may differ in their biology. In addition, when parasitoids have several host species only those where the parasitoid has its highest impact may show activity shifts but not species that are threatened less. Thus, in contrast to P. triangulum, the activity pattern of the digger wasp Cerceris arenaria that is also parasitized by the chrysidid parasitoid was not shifted [95]. Hunt [2] describes the differences in activity patterns of two Myrmecocystus species (honey pot ants). One species forages nocturnally, the other species forages diurnally but only at ground temperatures that are high enough to kill most other ants (up to 50Â°C). Whereas this could be interpreted only in the light of two species feeding on similar resources occupying two distinct niches to avoid competition, Hunt points out that the night or extremely hot ground temperatures during the day provide enemy-free space from lizards [2]. The spatial distribution of nests if often hypothesized to be strongly influenced by parasite pressure. One explanation for aggregated nesting as commonly found in solitary Hymenoptera is enhanced nest defence and a negatively density-dependent parasitism or predation rate. Individuals within nest aggregations of solitary or primitively social species such as stenogastrine wasps could benefit from aggregating with conspecifics via passive defence mechanisms such as predator confusion and selfish herding or active mechanisms such as active group defences, and improved parasite detection [24]. However, results of studies have been ambiguous with respect to density-dependence of parasitism rate. A review of 14 field studies [24] showed that in the majority of cases parasitism rate was positively densitydependent (eight studies) in comparison to only two showing negative density-dependence [28,96], i.e. reduced parasitism rate with an increasing number of nests within an aggregation. In addition, two studies showed that in spite of the increase in the number of parasites present with an increase in nests within aggregations parasitism rate was density-independent [17,24,97] suggesting that the impact of parasitoids was reduced relative to the number present. The proximate mechanism facilitating negatively density-dependent parasitism within aggregations was not clear in most cases. Wcislo suggested that the parasitoids` efficiency may be reduced in high-density aggregations due to interference with other wasps or conspecific miltogrammine flies [28]. Stenogastrine hover wasps were shown to benefit from nesting in the centre of an aggregation which would be expected under selfish herding. Interestingly, whereas one study showed a selfish herd benefit against all ants as well as Vespa tropica and ichneumonid wasps [98] in a second species only the predation rate by ants was reduced in nests in the centre of aggregations [99]. As ants attack â&#x20AC;&#x153;on footâ&#x20AC;? they will always approach nest

Effects of predators on predatory Hymenoptera

227

aggregations from the periphery whereas flying predators and parasitoids can attack any nest first, irrespective of the position within the aggregation.

Passive defences: Warning coloration (and shape) Aposematic colour patterns advertising unpalatability or distastefulness of prey towards predators are widespread in solitary and social wasps. The (to us human observers) familiar yellow and black pattern is very common in sphecids and vespids, whereas pomilids, mutilids and some sphecids such as Ammophila have orange and black patterns [3]. This strategy of predator avoidance has been shown to be successful against vertebrate as well as insect predators such as dragonflies [100,101,102]. It is assumed that model species are those with high abundance and aggressive potential [42]. Thus, social Vespids are widespread models in tropical regions as they are among the most abundant wasps, highly aggressive and posses a potent sting. Such mimicry complexes can involve other social as well as solitary wasp species, often with less aggressive or solitary species resembling social Vespids [42]. As female but not male wasps possess a potent sting and sometimes different movement patterns dual mimicry systems are also known with a female engaging in Müllerian mimicry of one model and males engaging in Batesian mimicry of another model [3]. Thus, male Chirodamus pompilid wasps that spend most of the time in flight or in vegetation resemble workers of social Vespids that share this spatial niche whereas females hunting on the ground resemble other solitary ground-hunting wasps [103]. In contrast to predatory wasps warning coloration and Müllerian mimicry seem to be relatively uncommon in ants. Aposematic coloration is known from relatively few species only, such as the conspicuously bright yellow metaplural gland lobes of Crematogaster inflata and their Batesian mimics [104] or the brightly coloured gastral eyespots of several Cephalotes species [105]. The large number of myrmecomorphic insects that do not prey on their ant models (including some flightless solitary predatory wasps) [59] suggests that the shape of ants may be enough to signal unpalatability to predators. As ant workers move relatively slowly on foot and do not need to be identified in flight in comparison to alighted wasps, shape may be enough as a reliable cue. That shape may be an important cue to predators is shown by avoidance behaviour towards wasp-shaped artificial prey-items in comparison to flyshaped ones by dragonflies [100,101]. Predators such as frogs or chicks avoid palatable ants after coming into contact with unpalatable ant species [106]. Interestingly, female mutillid wasps – commonly known as velvet ants – are generally wingless and are often mistaken as ants due to their shape and way to move. In contrast to most ants, they are often far more conspicuous due to

228

Heike Feldhaar

their body being densely covered by hairs that are often brightly coloured in shades of yellow or red. Currently it is not known whether different mimicry systems exist between alate queens of ants and their workers. Reproductives are more vulnerable to predators as they often have less chemical weaponry and are not protected by workers. When nuptial flights occur at night alate reproductives may benefit from reduced visibility to predators. In synchronized mating flights, alate queens may occur en masse and may benefit from a dilution effect. But what about ant queens engaging in diurnal nuptial and unsynchronized mating flights? Queens of plant-associated Crematogaster (Decacrema) do not show synchronized mating flights and are brownish in colour. However one species of Crematogaster (Crematogaster) with diurnal mating flights shows a colour dimorphism between queens and workers. Queens have a brightly yellow coloured abdomen (at least part of the population â&#x20AC;&#x201C; this seems to be a colour polymorphism) but not the workers, whose body is always black, irrespective of the coloration of the queen within the colony (personal observation). Mimicry systems may evolve quite easily as exemplified by dual mimicry systems and intraspecific colour-polymorphisms in populations where two model species occur in different proportions along an elevational gradient [42]. Nonetheless, little is currently known about the evolvability of such systems. In addition, will solitary species usually mimic social species or can mimicry systems also be driven by solitary species as models?

Active defences: Individual defences Anybody that has ever been stung by a Vespa will probably suggest that the sting is the most potent weapon that aculeate Hymenoptera possess. In contrast, in solitary wasp species, the sting may not even penetrate through the human skin â&#x20AC;&#x201C; or that of other vertebrate predators [3]. Solitary predatory Hymenoptera often also have less potent venoms in contrast to social species as they need to paralyze their prey without killing it immediately. A functional sting is ancestral in the ants and still widespread among extant ant taxa, although it has been lost independently numerous times (e.g. within formicines and some myrmicines). Especially within the poneroid ants that often hunt solitarily or in small groups the sting is still used to catch prey however. Again, like in solitary wasps, such ants were shown to paralyze but not kill their prey in order to preserve it over longer periods of time, e.g. in the ponerine ants Harpegnathos or Leptogenys [107]. Therefore in these ants the venom may be constrained in toxicity like in solitary wasps. The sting can attain purely defensive functions in some ants as well as eusocial wasps when

Effects of predators on predatory Hymenoptera

229

it has no function in subduction or killing of prey items. In those species prey is then usually grabbed and killed with the mandibles and often masticated directly at the site of predation. Thus, venoms of eusocial species are often more potent and more painful (at least to a human) and – in conjunction with effective alarm communication – the sting becomes an effective defensive weapon against vertebrate predators not only for personal but also for nest defence [108]. In ant species that have lost the sting the glands associated with the sting apparatus may still produce venom used for chemical defence. Such chemical defences can be very potent and can repel as well as kill predators, especially other arthropods [5,109]. Venom spraying has also been reported from the social wasp species Parachartergus colobopterus [14]. Jeanne and Keeping [14] suggest that this means of defence may be more effective against small insect-gleaning birds that seem to be the main predators of the small and cryptic nests of this wasp species. Aside from the sting and/or chemical defences, morphological adaptations can protect individual predatory Hymenoptera. A very thick cuticle may not only prevent entomologists from inserting pins through the thoraces of mutillid wasps [3], but will aid in avoiding being damaged by other insects through biting or stinging. Ants foraging in vegetation are often spiny [1], and the large spines of some ants of the genus Polyrhachis are surely adaptations towards predation by birds and other vertebrates. A bird trying to swallow a worker of Polyrhachis bihamata with its nearly 5 mm long Y-shaped spines that remind of fishing-hooks may probably do so only once in its lifetime if at all. As cuticle formation as well as venom production require energy and often nitrogenous compounds it has been suggested that ants may face a trade-off between investment into morphological defences versus chemical defences [2,110]. However, as most ants are omnivores it is not clear yet whether this trade-off also applies to strictly predacious ants such as some ponerines or army ants as well as to predatory wasps. Individual defences may also include adapted patterns of movement in the presence of predators or parasitoids. Thus, individuals may try to escape detection by “playing dead” or freezing. This strategy is known from a number of ants but also solitary wasps [1,3]. As potential predators or parasitoids may require movement as a visual cue to detect their prey this may be very effective. Thus, instead of returning to the nest, solitary wasps may perch on the vegetation for some time after spotting a parasitoid [111]. On the other hand, escape movements are also a common individual defence tactic. When pursued by parasitic “satellite flies” on the way to the nest, sphecid wasps can perform evasive manoeuvres. Instead of approaching the

230

Heike Feldhaar

nest entrance they can land briefly and remain motionless for some time. During this “freeze stop” the pursuing fly`s attention may be drawn to something else [3]. Another strategy of sphecids is to make pursuit to the nest difficult for the flies by performing circuitous flights with erratic changes in direction [3,85]. Escape strategies of individual ants from predators may comprise surprising means such as ballistic propulsion by trap jaw ants or “directed aerial descent”. The closure of the mandibles in some trap jaw ants is among the fastest movements ever measured in the animal kingdom. Trap jaws are formidable weapons for predation, producing immense forces that can crush arthropods of similar size as the ant easily. Odontomachus trap jaw ants have been shown to also use the trap jaw mechanism as a means of escape: When the force of the mandible strike is directed against substrate the ants catapult themselves into the air, reaching about 10 cm height [112]. The ant Cephalotes atratus has evolved another means of escape: workers can jump off branches when threatened and can perform a directed descent, enabling them to reach nearby vegetation or the trunk of the tree. This directed flight enables these arboreal ants to circumvent having to land in unfamiliar understory or the leaf litter where they may fall prey to other ants [113].

Nest architecture and behaviour at the nest: Protection of the offspring Nests of predatory Hymenoptera serve two functions: they are an adaptive structure to buffer against the physical environment and they protect the offspring against a broad range of predators and parasitoids. More elaborate and stronger (deeper in the case of digger wasps) nests may protect offspring better from both environmental factors as well as predators. Nest architecture may be constrained however by available material and nesting sites as well as by energy expense during construction. Solitary predatory Hymenoptera need to leave their offspring unguarded during foraging. They have therefore evolved strategies to conceal their nest entrances such as dispersing the soil particles excavated from the burrow at some distance away from the nest entrance or levelling the mound before leaving to hunt [3]. The closure of the nest entrance when leaving for foraging has been shown to reduce parasitism rates [3,85]. Some wasps dig accessory blind burrows that are kept open and supposedly distract parasitoids from the real nest entrance [114]. Nest entrances are plugged more permanently after provisioning of the offspring is completed by groundas well as hole-nesting species [3].

Effects of predators on predatory Hymenoptera

231

As soon as more than one adult is present per nest the opportunity for nest guarding arises. The most primitive case of nest sharing would be communal behaviour, where a single nest entrance is used but all individuals build, provision and oviposit in their own cells [115]. Among wasps such nesting associations are known for Pompilidae, Sphecidae and Eumeninae [3]. Communal behaviour typically occurs only within clades in which there are no examples of caste-based societies [116]. Communal nests may benefit passively from the presence of more than one female due to the increased activity within the nest or around the nest entrance. On the other hand nests may be guarded actively. Studies quantifying the benefit of communal nesting are surprisingly scant. Polidori and coworkers showed that increased activity within the nest of a communally nesting bee decreased the time satellite flies spent inside the nest [117]. Active nest guarding in communally nesting Cerceris antipodes was shown to improve nest defence against mutillids. With an increase in group size the entrances were guarded more often and disturbance elicited more defensive reactions of the wasps [11]. Most ants and primitively social wasps (stenogastrines and majority of polistines) found colonies as a single queen or a small group of queens, i.e. independently without reliance on workers. Whereas ant queens are often sheltered quite well from predation by “locking themselves in” and performing claustral colony formation, primitively social wasps with their small nests consisting of an open comb suspended from a petiole are more easily accessible to predators and parasitoids. Again, like in solitary digger wasps, nest combs with larvae and pupae need to be left unattended during foraging by those wasps when colonies are still small, comprising a few individuals only. In independent-founding species the choice of nesting place can reduce predation rate by ants and birds. Stenogastrines and some independentfounding polistine wasps build their nests in dim recessed places, thus probably reducing the risk of being visually detected by vertebrate predators [20]. Most ants do not rely on visual cues during foraging and any wasp nest within the foraging range of an ant colony will eventually be found. An effective nest defence mechanism against ants is the application of glandular secretions (produced in the Van der Vecht´s gland in the sixth abdominal segment) with ant-repellent properties on the thin petiole that suspends the nest from the substrate [4,19,20,22,25]. The petiole also restricts access to the nest “on foot”. Thus, single scout ants arriving on the narrow petiole can be attacked individually and deterred by wing buzzing or grabbing the ant and letting it drop to the ground [20]. This behaviour may prevent scouting-andrecruiting ants from attacking a wasp nest but will not be effective when too many ants arrive at the nest at once such as foraging army ants.

232

Heike Feldhaar

In contrast, in swarm-founding wasps colonies are initiated by a queen (or several) that are accompanied by workers. Thus, nests never have to be left unattended when some individuals go foraging. The steady presence of workers at the nest seems to render ant-repellent substances unnecessary as the transition from independent- to swarm-founding is correlated with the loss of Van der VechtÂ´s gland where such substances are produced [25]. A transition in nest architecture from naked combs to combs covered by a nest envelope is also seen as an adaptation to parasitoids and predators [19,22,25,118]. Access to the nest is restricted to a small entrance by the envelope. This nest entrance can be guarded and defended more easily by a small number of workers than large open combs, thus being a functional equivalent to the shape of the petiole of the small suspended nests of independent-founding wasps [19]. Jeanne argues that ant-repellent substances could also be used on such entrances of enveloped nests but may be too costly to produce in the relatively larger amounts needed to cover an entrance area. Long tubular entrances into enveloped nests may be covered with ant repellents [19]. Greater security of nests does not only depend on nest architecture in terms of the geometry of nest architecture but also building material. Nests of Polybia emiciata are built with mud instead of wood pulp as is usual among Vespidae. Unlike congeners with â&#x20AC;&#x153;normalâ&#x20AC;? nest envelopes, wasps of this species retreated into the nest upon a disturbance mimicking a vertebrate attack rather than exit to attack the potential predator [119]. In contrast to wasps, nest architecture in ants has hardly been studied with respect to colony defence [1,120]. One conspicuous morphological adaptation occurs regularly among ants: the development of a specialized worker caste to seal nest entrances. Single workers with plug-like (phragmotic) heads can effectively block nest entrances (see below) [8]. Nest entrances can likewise be temporarily closed or hidden with soil particles or small twigs and will usually be defended by guards [1]. An adaptive strategy to counteract the negative effects of nest predation by vertebrate predators that may destroy whole nests is polydomy, i.e. the fragmentation of a colony into several physically distinct nests. Polydomy may be beneficial in several ways: First, not all the brood may be lost at once if it is distributed over several nests [26]. Second, in case of one nest being destroyed, workers can evacuate the brood and themselves to an intact nest quickly [121]. Third, with an increasing number of nests, the probability of depredation per single nest within a colony may be reduced and thus colony level impact of predation decreases [40].

Effects of predators on predatory Hymenoptera

233

Collective defences involving recruitment strategies Most social wasps as well as ants are capable of coordinated group defences upon detection of an enemy. A coordinated group defence requires some form of recruitment strategy to arouse and direct the attention of nestmates towards the threat. Attention can be aroused via tactile or acoustic means. In most species however, alarm pheromones induce and regulate alarm communication in defensive responses in ants and wasps alike [8,9,12,13]. Although communication via chemical and other signals plays a prominent role during foraging in numerous species or during swarmformation and emigration in swarm-founding species the semiochemicals used are usually different from those used in alarm recruitment. Alarm pheromones have probably evolved via a stage where venom released during stinging was used as a cue by nestmates, while only later release of certain compounds became a signal [12]. Even among social Vespidae the use of alarm pheromones does not seem to be universal. Thus, in swarm-founding polistines with small colonies alarm pheromones may have been lost again [12]. In some species of ants and wasps alarm signals can convey enemyspecific information or are elicited in an enemy-specific manner. Atta leafcutter ants only recruit massively when attacked by the army ant Nomamyrmex esenbeckii that is a specialized predator of Atta, but not when encountering other army ants [122]. Likewise, Pheidole dentata minor workers will recruit soldiers only when encountering ants of the genus Solenopsis [123]. Alarm communication may not only lead to aggressive defensive reactions but may also evoke a flight reaction and thus evacuation of nests. Thus, whereas larger colonies of Atta will fight against the army ant Nomamyrmex smaller colonies may immediately evacuate their nest in order to escape their subterranean predator [1]. Likewise, adult wasps use alarm communication to escape army ants. The polistine Protopolybia exigua binominata shows â&#x20AC;&#x153;group fanningâ&#x20AC;?, i.e. a synchronized fanning of several wasps that elicits a buzzing sound that alerts the colony, when an army ant worker approaches the nest. When a nest is evacuated, the brood needs to be left behind though [124,125].

Predation as a selective force on social structure: Group size and caste formation Predation has been recognized as an important ecological factor selecting for helping behaviour and the evolution of sociality in Hymenoptera. The

234

Heike Feldhaar

relatively long period of parental care needed by brood of Hymenoptera requires that an adult carer must survive over the whole developmental time of its brood in order to gain any reproductive fitness. When mortality is high, helpers at nests may be favoured by insurance-based advantages (“lifeinsurers”) [126]. One advantage can be assured fitness-returns, i.e. even when a helper dies it benefits from its investment of partly rearing offspring if other helpers survive to rear the offspring to adulthood [127,128]. A second advantage may be survivorship of the nest as a whole, i.e. the reduction of the probability that the nests fails due to the death of all adults. Chances that at least one or a few adults survive are supposedly higher with increasing group size [129]. Larger groups could also benefit from rebuilding nests faster after an attack by a predator [27]. With an increasing group size, division of labour in nest defence may evolve or (morphological) defensive castes. In many eusocial species of ants and wasps workers show age polytheism in division of labour with riskier tasks such as foraging and other “outdoor” activies being performed by older workers. Such older workers have also been shown to be more aggressive and engage more in colony defence [10,15,130,131,132]. Older workers tend to have depleted fat storages and are thus less valuable to the colonies. Porter and Jorgensen have therefore termed the defending and foraging workers “a disposable caste” [131]. Seeming self-sacrifice by such individuals also incurs fewer costs, e.g. by sting autotomy in some wasps [7,13] or “explosion” of the gaster to release chemical defence substances in the case of some Camponotus species [133] when facing a predator. There may actually be a twofold benefit in sacrificing older individuals with depleted storages: on the one hand the least valuable adult individuals within the colony are sacrificed, and on the other hand fat-depleted foragers are also less valuable to predators. Thus, if workers are preyed upon during foraging that merely are “a walking exoskeleton” with very little other nutritional value they may not be worth the effort for a predator. To date to my knowledge only general preferences of predators have been tested (i.e. preferred prey species) but no tests have been performed on within-species differences that may be correlated with nutritional value of the prey item. A morphologically distinct defensive caste of soldiers has evolved in some ant species. Soldiers are worker phenotypes that are specialized for their role in colony defence [1]. They occur in 30 ant genera from seven subfamilies (overview in [1]). Dornhaus and Powell [1] categorized soldiers into those specialized for active defence against vertebrates, active defence against arthropods and passive defence, e.g. by blocking nest entrances (see above). Those soldiers effective against vertebrates usually have mandibles that can pierce tissue, whereas those of soldiers effective against other

Effects of predators on predatory Hymenoptera

235

arthropods have mandibles with normal shape whose blades can close in order to clip of appendages or cut into integument. The large soldiers of army ants of the genus Eciton protect foraging trails, caches of prey as well as the queen during emigrations. These soldiers possess large mandibles with recurved tips that can easily penetrate through vertebrate skin. Due to their pointed tips however, they would be useless in capturing small arthropod prey items [134]. Soldiers of other ant species with less differentiated mandible shapes, such as large Atta or Pheidole workers, can dismantle arthropods [8,122]. A specialized defensive caste does not have to be large in comparison to the workers engaged in foraging or brood care. Thus, tiny â&#x20AC;&#x153;minimâ&#x20AC;? workers of some Atta species have been observed to hitch a ride on the leaf fragments that are carried back to the nest. Those hitchhikers do not only clean leaves from potentially detrimental fungi but can also be regarded as soldiers as they defend the larger worker against attacks from phorid flies that may attempt to oviposit onto the larger worker [135]. As already described above, phragmotic workers that possess a specialized head morphology used for plugging nest entrances can also be considered a soldier caste, albeit defence is passive in this case [8,136,137]. Nonetheless, this form of nest entrance blocking can be a highly efficient form of defence as a single large worker with a plug-shaped head may reduce the probability of nest failure considerably [136]. The interaction between natural variation in entrance sizes and the specialized head of the soldiers can, however, dramatically alter the effectiveness of soldiers [138]. Interestingly, specialized soldier castes have only evolved within ants but not among the Vespidae. A possible reason could be that the cost of producing and maintaining a soldier caste may be less in ants due to the lack of functional wings and wing muscles in workers in general.

Top-down effects on predatory Hymenoptera: Indirect effects Aside from direct negative effects such as loss of workers or brood, predators and parasitoids may also incur indirect negative effects on their prey. Species communities are structured by species composition and interference competition. For ant communities it has been shown that species coexistence is partially explained by a trade-off in resource discovery versus resource dominance [139,140]. Behaviourally dominant species are able to usurp and defend larger or more valuable resources, usually by recruiting colony members, and may thus often replace subdominant species from such resources who in turn are able to discover them first. In addition, physiological constraints such as temperature tolerance limit foraging ability [141].

236

Heike Feldhaar

The presence of specialist predators or parasitoids targeting only a subset of species within a community can lead to shifts in the dominance hierarchies due to altered foraging behaviour. Phorid flies are highly host-specific parasitoids. Phorids locate potential hosts by being attracted to nest sites or recruitment trails of the respective ant species [142] or alarm pheromones [143] and usually attack during foraging. Ants respond to the attack by fleeing from the resource or assuming a defensive body posture that impedes oviposition by the fly. Such behaviours may lead to a reduction in the ant´s interspecific competitive ability by reducing foraging efficiency [144,145,146] or changing the outcome of interference competition in favour of less dominant but unparasitized ant species [29,147,148]. Thus, on the community level the presence of such parasitoids may facilitate the coexistence of species by reducing the behavioural dominance of their focal host species [16,30]. Predator avoidance behaviour, such as abandonment of foraging trails of ants when attacked by phorids, may translate into adaptive foraging strategies over evolutionary timescales that is most often only interpreted with respect to foraging ecology but not defence. Thus, frequent attacks by phorids may either lead to temporal shifts in foraging activity (e.g. foraging predominantly at night) in order to gain enemy-free space or may select for abandoning the use of foraging trails altogether to reduce “visibility” to phorids. Similarly, as Hunt suggested, the presence of lizards that can be mobbed effectively by groups of ants but not single ants may select for foraging in groups or along trails instead of foraging as single workers [2]. Currently similar studies on indirect effects of parasitoids and predators on the community level are largely lacking for solitary as well as social predatory wasps.

Trade-offs and constraints in defence strategies Defence strategies of predatory Hymenoptera are currently studied with an emphasis on how they are mediated, whom they are directed against and how efficient they are. However, the suit of potential defence strategies available to an organism is intrinsically tied to –and constrained by - the foraging ecology and social organisation of predatory Hymenoptera (including the solitary state). An obvious trade-off is the potential investment into different defence strategies which may be limited by the amount of resources acquired by the focal individual or colony. Thus, ant species with potent chemical defences tend to lack strong morphological defences such as a thick cuticle and spines

Effects of predators on predatory Hymenoptera

237

[2]. In addition, predatory ants tend to have venoms that contain proteinaceous substances whereas the venoms of omnivorous ants tend to contain only very little nitrogen, which is most probably due to nitrogenlimitation [110]. Both, chemical defences as well as cuticle formation may require nitrogen and with a stronger limitation of nitrogen one of the potential defence strategies may have to be reduced. Thus dietary intake may feed back to the potential array of defences that can be utilized (overview in Fig. 1). The complexity of communication used in the foraging (or swarming) context may also limit the complexity of collective defence strategies employed by a species. Although a general alarm communication involving the alert of nestmates by cues (e.g. detection of venom compounds in the environment) rather than signals (i.e. substances evolved specifically for the purpose of alerting nestmates) is present in most social species, the evolution of more sophisticated forms of alarm communication may parallel the evolution of

Figure 1. Overview of the relation between predators and parasitoids and defence and foraging strategies of predatory hymenoptera. Predators and parasitoids effect predatory hymenoptera directly as a selection pressure on defensive strategies but may also have an indirect effect by altering spatial and/or temporal activity patterns. The resources available to the focal individual or colony may constrain the kind of defences that can be realized. Currently it is not known whether foraging and defence strategies evolved in concert or independently from each other. Thus, species recruiting to a food source may also show alarm recruitment, whereas species lacking recruitment during foraging may also show low levels of alarm recruitment in a defensive context.

238

Heike Feldhaar

more complex forms of communication in other contexts. I would thus predict that species without recruitment in the foraging context will not show collective defence strategies that involve recruitment. Ants may accordingly be classified by a â&#x20AC;&#x153;defence and foraging syndromeâ&#x20AC;?: the level of complexity shown in one is reflected in and constrained by the level of complexity in the other. Thus, have collective defence mechanisms evolved in concert with collective foraging mechanisms â&#x20AC;&#x201C; or are they usually uncoupled? In order to be able to exploit a large (or more permanent) high quality resource, an ant species benefits from being able to recruit nestmates to the resource. At the same time this resource is defended against other ants or foraging organisms (see also [1]). A similar suggestion has been made by Hunt with respect to the evolution of foraging patterns and vertebrate predators for ants [2]. Ant species lacking recruitment strategies and foraging singly will either resist predation individually by potent weaponry or will avoid predation by being small and cryptic. In contrast, ants with good recruitment in the foraging context will recruit nestmates (often including soldiers) when threatened. Recruitment of larger workers may also occur with increased quality of the resource [2]. As suggested by Dornhaus and Powell, comparative analyses incorporating information about phylogenetic relationships among taxa would be vital to shed light on the question whether foraging and defence strategies go hand in hand or evolve independently [1]. In contrast to ants, social wasps show only very little recruitment in the foraging context. Most social wasps forage opportunistically on prey items or nectar [149]. Unlike in ants, the coordinated group transport of large prey items back to the nest is not possible and monopolization of high quality resources against other wasps and ants may be difficult. Pay-off for social wasps may thus be higher by searching a larger area with individual foragers that may be able to find and exploit resources first [149]. In parallel to the lack of recruitment to food sources and lack of monopolization of resources, recruitment in a defensive context will only make sense in the vicinity of the nest to alert nestmates to the presence of a predator. Coordinated attacks aimed at potential predators at the nest may exist in wasps, albeit they have not been studied in detail [20]. Involvement of the sting in foraging when used for prey paralysation should constrict its functionality as a defensive weapon. Venom composition as well as the force applied when stinging are selected to paralyse but not kill the prey items. It may be possible that the venom contains components that may be harmful to aggressors such as vertebrates but may not interfere with foraging function or that sting force can be adjusted in dependence of function, but this has to my knowledge not been studied yet. Do social species with solitary foragers or small foraging groups only have a more potent venom when the venom is not used for prey paralysation any longer?

Effects of predators on predatory Hymenoptera

239

This could be tested with sister-group comparisons of closely related taxa with different foraging strategies. Another trade-off involves the production of signals for intraspecific communication and avoidance of parasitoids. Cues that are used by parasitoids to detect their hosts should underlie postive selection, i.e. they should become as variable as possible to hamper chemical mimicry by parasitoids as well as reduce detectability. However, when such cues are used as intraspecific signals, e.g. as sex pheromones or for recruitment and alarm communication they may be constrained to become more variable as this may increase noise in the signal and thus decrease their utility for communication.

Conclusion The vast number of different predators and parasitoids and the potentially high fitness losses they may incur has contributed greatly to the biology of predatory Hymenoptera. The idiosyncrasies of predatory ants and predatory wasps have probably led to a different emphasis in studies of those two taxa with respect of being preyed upon: Whereas the role of predators and parasitoids on the evolution of sociality and benefits of group size can be studied more easily in communal and primitively social wasps, the study of indirect effects on the community level is more feasible in ants, probably due to the reduction of foraging patterns to two dimensions. Nonetheless, a picture emerges that foraging strategies and defence strategies are generally closely linked (Fig. 1). This begins in solitary sphecids where mode of provisioning of the brood influences parasitism rate and ends with sophisticated recruitment systems of polymorphic ant species employed during foraging and in defence of the nest as well as resources. In order to understand defensive strategies of predatory Hymenoptera, a stronger integration of ecological and evolutionary aspects of defence and foraging strategies is needed. Comparative analyses, such as those performed by Smith and coworkers [25] on correlations between colony defence and social structure of eusocial wasps or the role of nesting ecology in development of a specialized soldier caste in Cephalotes ants [137] would be vital to this end. A stronger emphasize should also be on how foraging (and nutritional ecology) constrains defensive strategies and vice versa. Are coordinated foraging and defence strategies only two sides of the same medal or can they evolve independently? In addition, how are defensive means shaped by diet and dietary intake rate? For ants it is assumed that a higher carbohydrate to protein ratio is associated with higher activity levels in patrolling but limitations in morphological and chemical defence structures due to nitrogen-limitation. Is this a general trend that can be observed in other taxa such as wasps and bees as well?

240

Heike Feldhaar

In conclusion, the ecology and evolution of defensive strategies is strongly linked to nesting behaviour, social organization and foraging in predatory Hymenoptera. Although often hard to demonstrate conclusively, measuring the influence of predators and parasitoids on the evolutionary ecology of predatory Hymenoptera may make a key contribution to understanding their foraging ecology and structuring of species assemblages from populations to the community level.

Acknowledgements I would like to thank Carlo Polidori for the invitation to write this chapter and become part of this book. Jim Hunt and Scott Powell provided very helpful comments on an earlier version of this chapter. To the many researchers whose work on this topic was not cited in this chapter due to space limitation I would like to apologize. I am very grateful to Claudia Knobel for her understanding and endurance whenever the word “book chapter” is mentioned.

References 1.

Dornhaus A., and Powell S., 2010, Foraging and defence strategies. In: Lach L., Parr C.L., Abbott K.L. (Eds.), Ant ecology. Oxford, Oxford University Press. pp. 210-230. 2. Hunt J.H., 1983 Foraging and morphology in ants: the role of vertebrate predators as agents of natural selection. In: Jaisson P. (ed.), Social insects in the tropics. Paris, Presses de l'Université Paris XIII. pp. 83-104. 3. O'Neill K.M., 2000, Solitary wasps: Behavior and natural history. Ithaca, Cornell University Press. 4. Jeanne R.L., 1970, Science 168, 1465-1466. 5. Marlier J.F., Quinet Y., and de Biseau J.C., 2004, Behav. Processes 67, 427-440. 6. Schmidt P.J., Sherbrooke W.C., and Schmidt J.O., 1989, Copeia 3, 603-607. 7. Evans H.E., and West-Eberhard M.J., 1970, The wasps. Ann Arbor, University of Michigan Press. 8. Hölldobler B., and Wilson E.O., 1990, The ants. Cambridge, Harvard University Press. 9. Jeanne R.L., 1981, Behav. Ecol. Sociobiol. 9, 143-148. 10. Judd T.M., 1998, Insect. Soc. 45, 197-208. 11. McCorquodale D.B., 1989, J. Insect Behav. 2, 267-276. 12. Landolt P.J., Jeanne R.L., and Reed H.C., 1998, Chemical communication in social wasps. In: Vander Meer R.K., Breed M.D., Espelie K.E., and Winston M.L., (Eds.), Pheromone communication in social insects: Ants, bees, wasps, and termites. Boulder, Westview Press. pp. 216-235.

Effects of predators on predatory Hymenoptera

241

13. Sledge M.F., Dani F.R., Fortunato A., Maschwitz U., Clarke S.R., Francescato E., Hashim R., Morgan E.D., Jones G.R., and Turillazzi S., 1999, Physiol. Entomol. 24, 234-239. 14. Jeanne R.L., and Keeping M.G., 1995, J. Insect Behav. 8, 433-442. 15. Judd T.M., 2000, Anim. Behav. 60, 55-61. 16. LeBrun E.G., 2005, Oecologia 142, 643-652. 17. Rosenheim J.A., 1989, Behav. Ecol. Sociobiol. 25, 335-348. 18. Strohm E., Laurien-Kehnen C., and Bordon S., 2001, Oecologia 129, 50-57. 19. Jeanne R.L., 1975, Q. Rev. Biol. 50, 267-287. 20. Starr C.K., 1990, Holding the fort: Colony defense in some primitively social wasps. In: Evans D.L., and Schmidt J.O., (Eds.), Insect defenses: Adaptive mechanisms and strategies of prey and predators. Stony Brook: SUNY Press. pp. 421-463. 21. Wcislo W.T., 1996, J. Insect Behav. 9, 643-656. 22. Wenzel J.W., 1991 Evolution of nest architecture. In: Ross K.G., and Matthews R.W., (Eds.), The social biology of wasps. Ithaca, Cornell University Press. pp. 480-519. 23. Field J., and Brace S., 2004, Nature 428, 650-652. 24. Rosenheim J.A., 1990, Ann. Entomol. Soc. Am. 83, 277-286. 25. Smith A.R., O'Donnell S., and Jeanne R.L., 2001, Evol. Ecol. Res. 3, 331-344. 26. Strassmann J.E., 1981, Ecology 62, 1225-1233. 27. Strassmann J.E., Queller D.C., and Hughes C.R., 1988, Ecology 69, 1497-1505. 28. Wcislo W.T., 1984, Behav. Ecol. Sociobiol. 15, 157-160. 29. Feener D.H., 1981, Science 214, 815-817. 30. Feener D.H., 2000, Oikos 90, 79-88. 31. Schmid-Hempel P., 1998, Parasites in social insects. Princeton, Princeton University Press. 32. Brandt M., Foitzik S., Fischer-Blass B., and Heinze J., 2005, Biol. Rev. 80, 251267. 33. Cervo R., 2006, Ann. Zool. Fenn. 43, 531-549. 34. Lenoir A., D'Ettorre P., Errard C., and Hefetz A., 2001, Annu. Rev. Entomol. 46, 573-599. 35. Field J., 1992, Biol. Rev. 67, 79-126. 36. Abensperg-Traun M., and Steven D., 1997, Aust. J. Zool. 22, 9-17. 37. Redford K.H., 1985, J. Zool. 205, 559-572. 38. Jeanne R.L., 1970, J. Mammal. 51, 624-625. 39. Swenson J.E., Jansson A., Riig R., and Sandegren F., 1999, Can. J. Zool. 77, 551-561. 40. Van Wilgenburg E., and Elgar M.A., 2007, Biol. J. Linn. Soc. 92, 1-8. 41. Federle W., Leo A., Moog J., Azarae H.J., and Maschwitz U., 1999, Ecotropica 5, 35-43. 42. O'Donnell S., 1999, Biol. J. Linn. Soc. 66, 501-514. 43. Ihalainen E., Lindstrom L., Mappes J., and Puolakkainen S., 2008 Behav. Ecol. 19, 362-368. 44. Hermann H.R., 1984, Psyche 91, 51-66.

242

45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76.

77. 78. 79. 80. 81.

Heike Feldhaar

Litte M., 1977, Behav. Ecol. Sociobiol. 2, 229-246. Windsor D.M., 1976, Biotropica 8, 111-116. Janzen D.H., 1969, Condor 71, 240-256. Pianka E.R., and Parker W.S., 1975, Copeia 1, 141-162. Suarez A.V., Richmond J.Q., and Case T.J., 2000, Ecol. Appl. 10, 711-725. Rissing S.W., 1981, Ecology 62, 1031-1040. Whitford W.G., and Bryant M., 1979, Ecology 60, 686-694. Sherbrooke W.C., and Schwenk K., 2008, J. Exp. Zool. A Ecol. Genet. Physiol. 309A, 447-459. Ito F., Okaue M., and Ichikawa T., 2009, Myrmecol. News 12, 35-39. Toft C.A., 1981, J. Herpetol. 15, 139-144. Caldwell JP., 1996, J. Zool. 240, 75-101. Gotelli N.J., 1996, Ecology 77, 630-638. Griffiths D., 1980, J. Anim. Ecol. 49, 99-125. Hรถlldobler B., 1970, Psyche 77, 202-208. McIver J.D., and Stonedahl G., 1993, Annu. Rev. Entomol. 38, 351-379. Obin M.S., 1982, Psyche 89, 321-336. Huseynov E.F., Cross F.R., and Jackson R.R., 2005, J. Zool. 267, 159-165. Pekar S., Toft S., Hruskova M., and Mayntz D., 2008, Naturwissenschaften 95, 233-239. Porter S.D., and Eastmond D.A., 1982, J. Arachnol. 10, 275-277. Cushing P.E., 1997, Fla Entomol. 80, 165-193. Bouwma A.M., Howard K.J., and Jeanne R.L., 2007, Biotropica 39, 719-724. Cane J.H., and Miyamoto M.M., 1979, J. Kans. Entomol. Soc. 52, 667-672. Chadab R., and Rettenmeyer C.W., 1975, Science 188, 1124-1125. O'Donnell S., and Jeanne R.L., 1990, J. Trop. Ecol. 6, 507-509. Rosenheim J.A., 1987, Ann. Entomol. Soc. Am. 80, 739-749. Rosenheim J.A., 1988, J. Insect Behav. 1, 333-342. Jeanne R.L., 1979, Ecology 60, 1211-1224. London K.B., and Jeanne R.L., 2005, J. Kans. Entomol. Soc. 78, 134-141. Alexander B., 1986, J. Kans. Entomol. Soc. 59, 59-63. Powell S., and Clark E., 2004, Insect. Soc. 51, 342-351. Gotwald W.H., 1995, Army Ants: the biology of social predation. Ithaca, Cornell University Press. Franks N.R., 1982, Ecology and population regulation in the army ant Eciton burchelli. In: Leigh E.G., Rand A.S., and Windsor D.M., (Eds.), The ecology of a tropical forest: seasonal rhythms and long-term changes. Washington D.C., Smithsonian Institution Press. pp. 389-395. Hirosawa H., Higashi S., and Mohamed M., 2000, Insect. Soc. 47, 42-49. Ono M., Igarashi T., Ohno E., and Sasaki M., 1995, Nature 377, 334-336. Stubblefield J.W., Seger J., Wenzel J.W., and Heisler M.M., 1993, Philos. Trans. R. Soc. Lond. B Biol. Sci. 339, 397-423. Alexander B., 1985, J. Nat. Hist. 19, 1139-1154. Wheeler W.M., 1913, J. Anim. Behav. 3, 374-387.

Effects of predators on predatory Hymenoptera

243

82. Zettel H., Ljubomirov T., Steiner F.M., Schlick-Steiner B.C., Grabenweger G., and Wiesbauer H., 2004, Myrmecologische Nachrichten 6, 39-47. 83. Gadagkar R., 1991, Belonogaster, Mischocyttarus, Parapolybia and independentfounding Ropalidia. In: Ross K.G, and Matthews R.W., (Eds.), The Social Biology of Wasps. Ithaca, Cornell University Press. pp. 149-190. 84. Spofford M.G., and Kurczewski F.E., 1990, J. Nat. Hist. 24, 731-755. 85. Polidori C., Ouadragou M., Gadallah N.S., and Andrietti F., 2009, Trop. Zool. 22, 1-14. 86. McCorquodale D.B., 1986, Can. J. Zool. 64, 1620-1627. 87. Quicke D.L.J., 1997, Parasitic wasps. New York, Chapman and Hall. 88. Akino T., Knapp J.J., Thomas J.A., and Elmes G.W., 1999, Proc. R. Soc. Lond. B Biol. Sci. 266, 1419-1426. 89. Strohm E., Kroiss J., Herzner G., Laurien-Kehnen C., Boland W., Schreier P., and Schmitt T., 2008, Front. Zool. 5, 2. 90. Hughes D.P., Beani L., Turillazzi S., and Kathirithamby J., 2003, Insect. Soc. 50, 62-68. 91. Polidori C., Tormos J., Asis J.D., Mendiola P., and Andrietti F., 2006, Entomol. Fenn. 17, 405-413. 92. Benttinen J., and Preisser E., 2009, Can. Entomol. 141, 604-608. 93. Wrege P.H., Wikelski M., Mandel J.T., Rassweiler T., and Couzin I.D., 2005, Ecology 86, 555-559. 94. Wilson E.O., 1971, The insect societies. Cambridge, Belknap Press of Harvard University Press. 95. Polidori C., Bevacqua S., and Andrietti F., 2010, Acta Ethol. 13, 11-21. 96. Freeman B.E., 1981, J. Anim. Ecol. 50, 563-572. 97. Itino T., 1988, Res. Popul. Ecol. 30, 1-12. 98. Landi M., Coster-Longman C., and Turillazzi S., 2002, Ethol. Ecol. Evol. 14, 297-305. 99. Coster-Longman C., Landi M., and Turillazzi S., 2002, J. Insect Behav. 15; 331-350. 100. Kauppinen J., and Mappes J., 2003, Anim. Behav. 66, 505-511. 101. O'Donnell S., 1996, J. Insect Behav. 9, 159-162. 102. Schuler W., and Roper T.J., 1992, Adv. Study Behav. 21, 111-146. 103. Evans H.E., 1968, Psyche 75, 1-22. 104. Ito F., Hashim R., Huei Y.S., Kaufmann E., Akino T., and Billen J., 2004, Naturwissenschaften 91, 481-484. 105. de Andrade M.L., and Baroni-Urbani C., 1999, Diversity and adaptation in the ant genus Cephalotes, past and present. Stuttgarter Beiträge zur Naturkunde Serie B (Geologie und Paläntologie). 106. Taniguchi K., Maruyama M., Ichikawa T., and Ito F., 2005, Insect. Soc. 52, 320322. 107. Maschwitz U., Hahn M., and Schönegge P., 1979, Naturwissenschaften 66, 213214. 108. Kukuk P.F., Eickworth G.C., Raveret Richter M., Alexander B., Gibson R., Morse R.A., and Ratnieks F., 1989, Ann. Entomol. Soc. Am. 82, 1-5.

244

Heike Feldhaar

109. Heredia A., de Biseau J.C., and Quinet Y., 2005, Chemoecology 15, 235-242. 110. Davidson D.W., 2005, Oecologia 142, 221-231. 111. Rosenheim J.A., 1990, J. Insect Behav. 3, 241-250. 112. Patek S.N., Baio J.E., Fisher B.L., and Suarez A.V., 2006, Proc. Nat. Acad. Sci. USA 103, 12787-12792. 113. Yanoviak S.P., Dudley R., and Kaspari M., 2005, Nature 433: 624-626. 114. Evans H.E., 1966, Science 152, 465-471. 115. Michener C.D., 1974, The social behavior of the bees. Cambridge, The Belknap Press of Harvard University Press. 116. Wcislo W.T, and Tierney S.M., 2009, The evolution of communal behavior in bees and wasps: An alternative to eusociality. In: Gadau J., and Fewell J., (Eds.), Organization of insect societies: From genome to sociocomplexity. Cambrigde, Harvard University Press. pp. 148-169. 117. Polidori C., Scanni B., Scamoni E., Giovanetti M., Andrietti F., and Paxton R.J., 2005, J. Nat. Hist. 39, 2745-2758. 118. Jeanne R.L., 1991, The swarm-founding Polistinae. In: Ross K.G., and Matthews R.W., (Eds.), The social biology of wasps. Ithaca, Cornell University Press. pp. 191-231. 119. O'Donnell S., and Jeanne R.L., 2002, J. Insect Sci. 2, 3. 120. Bl端thgen N., and Feldhaar H., 2010, Food and shelter: How resources influence ant ecology. In: Lach L., Parr C.L., and Abbott K.L., (Eds.), Ant ecology. Oxford, Oxford University Press. pp. 115-136. 121. Droual R., 1984, Anim. Behav. 32, 1054-1058. 122. Powell S., and Clark E., 2004, Insect. Soc. 51, 342-351. 123. Wilson E.O., 1975, 190, 798-800. 124. Chadab R., 1979, Psyche 86, 115-124. 125. Young M.Y., 1979, J. Kans. Entomol. Soc. 52, 759-768. 126. Strassmann J.E., and Queller D.C., 2007, Proc. Nat. Acad. Sci. USA 104, 8619-8626. 127. Field J., Shreeves G., Sumner S., and Casiraghi M., 2000, Nature 404, 869-871. 128. Shreeves G., Cant M.A., Bolton A., and Field J., 2003, Proc. R. Soc. B Biol. Sci. 270, 1617-1622. 129. Queller D.C., 1994, Proc. R. Soc. Lond. B Biol. Sci. 256, 105-111. 130. Jeanne R.L., Williams N.M., and Yandell B.S., 1992, J. Insect Behav. 5, 211-227. 131. Porter S.D., and Jorgensen C.D., 1981, Behav. Ecol. Sociobiol. 9, 247-256. 132. Togni O.C., and Giannotti E., 2010, J. Insect Sci 10: 136. 133. Davidson D.W., Lessard J.P., Bernau C.R., and Cook S.C., 2007, Biotropica 39, 468-475. 134. Rettenmeyer C.W., 1963, University of Kansas Scientific Bulletin 44, 281-465. 135. Vieira-Neto E.H.M., Mundim F.M., and Vasconcelos H.L., 2006, Insect. Soc. 53, 326-332. 136. Hasegawa E., 1993, Behav. Ecol. Sociobiol. 33, 73-77. 137. Powell S., 2008, Funct. Ecol. 22, 902-911. 138. Powell S., 2009, J. Evol. Biol. 22, 1004-1013. 139. Davidson D.W., 1998, Ecol. Entomol. 23, 484-490.

Effects of predators on predatory Hymenoptera

245

140. Fellers J.H., 1987, Ecology 68, 1466-1478. 141. Cerda X., Retana J., and Cros S., 1997, J. Anim. Ecol. 66, 363-374. 142. Feener D.H., and Brown B.V., 1997, Annu. Rev. Entomol. 42, 73-97. 143. Witte V., Disney R.H.L., Weissflog A., and Maschwitz U., 2010, J. Nat. Hist. 44, 905-912. 144. Feener D.H., and Brown B.V., 1992, Ann. Entomol. Soc. Am. 85, 80-84. 145. Orr M.R., and Seike S.H., 1998, Oecologia 117, 420-425. 146. Orr M.R., Seike S.H., Benson W.W., and Gilbert L.E., 1995, Nature 373, 292-293. 147. LeBrun E.G., and Feener D.H., 2002, Oecologia 133, 599-607. 148. LeBrun E.G., and Feener D.H., 2007, J. Anim. Ecol. 76, 58-64. 149. Raveret Richter M., 2000, Annu. Rev. Entomol. 45, 121-150.