Spatial-temporal modeling vegetation patterns: Burning-grazing in páramo Los Nevados - Verweij 1995 by JPZ.

5:; L/, 5 oCJ8G

c.o.

v512-S e~ · .2_

. SPATIAL AND TEMPORAL MODELLING OF VEGETATION PATTERNS COl (Wlll I •

ltf.li'IERSIDAD NACIONAl DE Malurellt autltU\Q de CleOCI\>5

BIBLI01ECA

Burning and grazing in the paramo of Los Ne'{ados National Park, Colombi a

lTC Publication Number 30

Pita A. Verweij

SPATIAL AND TEMPORAL MODELLING OF VEGETATION PATTERNS

Burning and grazing in t-he paramo of Los Nevados National Park, Colombia

BIBLIOTE.CI\ ACADEMISCH PROEFSCHRIFT l

~i I 1lI ll -~

â&#x20AC;˘'.

ISBN 90-6!64-!09-8 T

' Pita A. ÂŠ 1995 Vcrweij, All rights reserved. No pan of this publication, apan from bibliographic data, brief quotations in critical reviews,. may be reproduced, re-recorded or published in any form including print, photocopy, microtilm, electlunic or electromagnetic record, without written permission.

ter verkrijging van de graad van doctor aan de Universiteit van Amsterdam, op gezag van de Rector Magnificus prof. dr. P. W_ M. de Meijer ten overstaan van een door het college van dekanen ingestelde commissie in het openbaar te verdedigen in de Aula der Universiteit (Oude Lutherse Kerk, ingang Singe! 411, hoek Spui) op woensdag 29 november 1995 te 10.30 uur

' i

door

. r Pita Adriana Verweij

Cover design : Taib el Ghazi Distribution : International Institute for Aerospace Survey and Eanh Sciences lTC, P.O. Box 6, 7500 AA Enschcde. The Netherlands

geboren te Benschop

.Âˇ.

.')

Promotor Prof. Dr. H. Hooghicmstra Co-promotores Dr. A.M. Cleef Prof. Dr. Ir. W. van Wijngaarden Other members of the committee Prof Dr. J. van Andel Prof. Dr. H. H. T. Prins Prof. Dr. J. Sevink Dr. F. Bouman

aan mijn ouders

â&#x201A;ŹJ

lTC

This study was carried out at the International Institute for Aerospace Survey and Earth Sciences (lTC), Enschede, the Netherlands and in Los Nevados National Park, Colombia; in cooperation with the Hugo de Vries-Laboratory, University of Amsterdam (The Netherlands Centre for Geo-ecolog ical Research, ICG) and the lnstituto Nacional de Recursos Renovables y del Ambiente (!NDERENA), Colombia.

·l

CONTENTS

Page

•' i'

Acknowledgements 1

INIRODUCI10N

1.1 1.2 1.3 2

2.1 2.2 2.3 2.4 2.5 2.6 3

3.6 4

I f

I-

Pita A. Verweij and Anne M. Schmidt 4.1 4.2 4.3 4.4 4.5-

i i

Introduction to paramo range management Estimation methods of forage intake Grazing behaviour Quality and quantity of forage Conclusions

EFFECTS OF FIRE AND GRAZING ON PLANT POPULATIONS

9.4

Introduction Methods of studying stem rosettes Management impact on stem rosette populations Methods of studying bunch grasses Management impact on bunch-grass populations

J l

l i

191

General remarks Potential of the modelling approach Grazing management Burning management Plant species diversity The natural resource managers

Appendix A Appendix B Appendix C

177

Introduction Monitoring fire events in GIS Reconstruction of fire history The role of fire in the distribution of grazing patterns Discussion and conclusions

CONCLUSIONS AND IMPLICATIONS FOR MANAGEMENT INCLUDING CONSERVATION OF PLANT DIVERSITY

9.1 9.2 9.3

151

Introduction Methodology Sources of variation in grazing intensity: regression analysis Grazing behaviour and terrain variables Forage availability Grazing management by man Spatial model Qf cattle distribution

BURNING AS A MANAGEMENT TOOL

9.5 9.6

Pita A. Verweij and Kasper Kok

5.1 5.2 5.3 5.4 5.5

SPATIAL MODEU..ING OF CATTLE DISTRIBUIION

8.1 8.2 8.3 8.4 8.5

127

Introduction Definition of functional plant groups Model formulation Model calibration Conclusions on vegetation development Simulating management strategies Limitations of the model

7.1 7.2 7.3 7.4 7.5 7.6 7.7

Methodology Definition of vegetation structure types Definition of floristic types Ordination analyses Vegetation map Conclusions

EXTENSIVE LIVESTOCK PRODUCTION IN THE PARAMO

SIMULATION MODEL OF VEC;ETATION DYNAMICS

6.1 6.2 6.3 6.4 6.5 .6.6 6.7

Climatic conditions Geology and geomorphology Soils and hydrology General land cover Land use: conflicting interests Human history and socio-economic setting

CHARACTERIZATION OF VEGETATION PATIERNS

3.1 3.2 3.3 3.4 3.5

The pararno ecosystem and the role of man Research objectives The modelling approach

THE PARAMO OF LOS 1\'EVADOS NATIONAL PARK

I !

Summary Resumen Samenvatting

Plant list Documentation of simulation model Vegetation table: attached to cover

209 215

225 228 231

ACKNOWLEDGEMENTS This dissertation would not have bo:cn completed without the coope ration and assistance of many persons and institutions. First of all; I am very grateful to both co-promotores, Dr. Antoine Cleef and Prof. Dr. Willem van Wijngaarden, for their valuable contributions in all stages of the research. Antoine Cleef has encouraged and supported the study of pararno vegetation dynamics ever since my MSc thesis in tropical ecology, which he also supervised. His extensive knowledge of the paramo vegetation and its ecological relationships has benefitted the research in many ways: Besides the organizational backing he provided, I also highly appreciate his friendship and moral support. r also received stimulating guidance from Willem van Wijngaarden. In particular, I learned much from his pragmatic approach to modelling and dealing with spatio-temporal variability. The discussions during fieldwork and the initial consiruction of the simulation model were inspiring. Furthermore, I also appreciate that Willem offered me many opportunities to gain experience, not only in research, but also in training and advisory services in different parts of the world. I thank Prof. Dr. Henry Hooghiemstra for his willingness to act as promotor. He fulfilled this task with dedication and has always followed the progress of the work with interest. I thank Prof_ Dr. Jelte van Andel, Dr. F. Bouman, Prof. Dr. Herbert Prins, Prof. Dr. Jan Sevink, and Prof. Dr. Ies Zonneveld for critical reading of the manuscript. I feel fortunate to have participated in the last Rural and Land Ecology Course guided by les Zonneveld, and I am sure that this dissertation, although not strictly adhering to the land unit concept, reflects much of his views and ideas. Preliminary conclusions were discussed with Prof. Dr. Thomas van der Hammen in the beautiful setting of fine a Santa Clara in Chia, Colombia. r thank him and his wife Anita for their hospitality. Pro.f. Dr. 't Manne~e of the Agricultural University Wageningen provided constructive comments on the article on which Chapter 4 is based. The in vitro digestibility analyses of plant material were carried out by his research group. Prof. Dr. Herman van Keulen (AUW) and- Dr. Peter Herman (NIOO, Centre for Estuarine and Coastal Ecology, Yerseke) gave advice in different stages of the modelling process. Henk Kloosterman (Rijkswaterstaat, Delft) provided many valuable ideas, based on his knowledge of the study area. At ITC, Dr. Carlos Valenzuela. Alejandro Bargagli, and Petra Budde contributed to theÂˇdevelopment of the GISbased method for geometric correction of aerial photo interpretations. I would like to thank my colleagues of the Vegetation and ~gricultural Sciences Division for their support. The restructuring of the PhD proposal benefitted from the discussions with Dr. Hein van Gils. The fieldwork data set and part of the analysis are the product of excellent teamwork. During two consecutive years, several graduates of the University of Amsterdam participated in the research through an MSc thesis. r am particularly grateful to Petra Budde, Bas Pels, Rien Beukema, Anne Schmidt, and Kasper Kok for their contributions to the work in a pleasant atmosphere. Maria . Isabel Valencia and Hernando Vergara also made enthusiastic contributions. I thank my fellow paramo-researcher Robert Hofstede for stimulating discussions, his critical review of several manuscripts, and his assistance in the observations on grazing behaviour of the cattle. In reply to what you stated in your interesting dissertation, Robert: I also look forward to writing a joint synthesis article!

, I

Acknowledgements

The National Parks Service of !NDERENA, Colombia, gave pcnnission to conduct research in Los Nevados National Park, including the collection of plant material, and provided accomodation in the park and transport facilities in the form of a four-\vheel drive and mules. Dr. Carlos Castaiio, lng. Gustavo Sanchez, and the park guards of Los Nevados were always willing to help whenever and wherever necessary. The lnstituto Geografico 'Agustin Codazzi' also provided logistic support, which was kindly arranged by Dr. Rob van Zuidam. In many occasions, staff of the Herbaria Nacional of the Instituto de Ciencias Naturales assisted in the identification of plant specimen. In particular the assistance by Dr. Orlando Rangel, Roberto Jaramillo, Jaime Aguirre, and Santiago Diaz was important. The floristic part of this study would never have been accomplished without the collaboration of many other botanic specialists, of the State University of Utrecht and elsewhere. La hospitalidad de doiia Clara y don Sergio Marquez se hizo famosa hasta en Holanda. A ellos, y a German, Nelson, Guido, y sus colegas del INDERENA estoy muy agradecida por el apoyo que brindaron a nuestro grupo de investigacion, ypor su amistad. A los habitantes de Ia region Otun-EI Bosque, les agradezco su amable cooperaci6n. Durante nuestra estadia en e1 clima frio del paramo, apreciamos mucho las numerosas tazas de chocolate caliente y agua de panela que recibimos en sus fincas, y otras muestras de calor humano. La infomtacion que obtuvimos a trav~s de entrevistas y su ayuda con las mediciones del crecimiento del ganado fueron indispensables. Especialmente a los senores Jaime, Juan, Jezus, Ariel, Erqin y sus familias quiero expresar mi sincera gratitud. La familia Gon:alez, la_fa.milia Mor_eno, Oscar Rincon, Kleoniki, Ernesto, Ing. Christina Escobar, e Ing. H1lda Chr1stma Salvatierra siempre me recibieron con mucho cariiio. Gracias a personas tan especiales como etlos, no me sentia en un pais extrai1o.

Acknowledgements

Vricndc:n en familieleden hebben stcrk meegelcefd en hct wordingsproct:s van dit bockjc op de vot:t gcl'olgd. Dil is YOor mij ecn bdangrijkc moti\'atic gcwccst. Janine, Martin, Astrid, Tincke, 13etske, Far, Marianne, Petra en \Vim: bedankt! Mijn collcga's Jan, Wietske en Joan wisten me op cruciale momenten een hart ond<!r de ric:m te steken. De nedcrlandse samenvatting werd verbeterd door Far Wevers. Tineke en Maarten ben ik erkcntelijk voor hun bijdrage in de rot van paranymf. Dit boek is opgedragen aan mijn ouders, Jan en Rina. Zij hi!bben mij niet aileen liefde voor de natuur bijgebracht, maar ook respekt voor het boerenbedrijf. Hun voortdurende steun heb ik altijd zcer gewaardeerd. Tot slot, maar zeker niet in de laatste plaats, bedank ik Talb: met kookkunst, grafische vormgeving en vooral vee! begrip heeft hij mij terzijde gestaan. Trub, zaindjeli, chokran!

The staff of the Cartography Division is acknowledged for providing advice and facilities during final map production. Ann Stewart supported the publication of this dissertation as part of the lTC Publication Series. The English text was corrected by Janice Collins; Maria and Guillermo Calder6n-Bastidas helped with the translation of the Spanish summary. Administrative and technical staff at lTC, of whom I mention Bert Riekerk, Benno Masselink. Gerard Leppink, Henk Scharrenborg, and Ronnie Geerdink, assisted in varrous ways. Secretarial support was given by Ceciel Wolters and Daniela Semeraro at lTC, and by Jody DosSantos and Erica Yssel at the University of Amsterdam. 1 am grateful to my employer, fTC, for financing large part of the research. The Hugo de Vries Laboratory of the University of Amsterdam contributed both financially and through the execution of the MSc research projects. Additional chemical analyses of plant material were carried out at the Hugo de Vries Laboratory by Rob Bregman and Annemarie Philip. I appreciate the financial support on behalf of the Treub Ma.atschappij, A~ster~am, to visi~ the UNESCO/MAB Seminar on utilization of tropical mountam ecosystems 111 Jujuy, Argentma.

The permission of Academic Press, London, to base Chapters 3, 4. and 5 onÂˇarticles published earlier in the book 'Paramo, a high mountain ecosystem under human influence' is gratefully acknowledged. Anonymous reviewers and the editors of this book, Dr. H. Balslev and Dr. J.L. Luteyn, provided many constructive comments. 11 10

INTRODUCfiON 1.1

1.2 1.3

The paramo ecosystem and the role of man Research objectives The modelling approach

1.1 .The paramo ecosystem and the role of man Whether one climbs Mount Kenya, Mount Wilhelm on New Guinea, or the Colombia.n Nevada del Rufz, above the natural forest line the landscape is strikingly similar. In environments of humid tropical mountains throughout the world, similar ecological co~ditions have led to similar ecosystems. In the elevation belt above the natural forest line and below the possible snow line, grasslands have developed with some conspicuous features: the dominant growth fonn are tussock or bunch grasses, often interspaced by giant rosette plants. This tropical alpine zone can be found between the Tropic of Cancer and the Tropic of Capricorn wherever the mountains exceed a critical minimum height. Locally, the system is referred to by different names: afroalpine zone (East Africa), tropical-alpine zone (Malaysia), puna and jalca (central Andes), zacatonal (Mexican and Guatemalan volcanoes), and paramo (northern Andes). Tropical alpine systems similar to the paramo are described by Fosberg (1959), Hedberg (1964, 1969, 1992), Coe (1967), Hnatiuk (1978), Smith (1986), Troll (1968), Ellenberg (1975), Lauer (1976), Smith & Young (1987), Almeida et al. (1994), and Islebe & Cleef (1995). Two major grassland biotic formations occur in the high Andes: the dry puna and the more humid paramo. The paramo ecosystem occurs from about 3000 m to about 4800 m elevation, primarily in Venezuela, Colombia, and Ecuador, with outliers in Costa Rica, Panama, and northern Peru. Yearly rainfall is in the range of 500 to 3000 mm. The climate is cool and humid, with daily temperatures ranging from -3°C to+ 12°C. Diurnal temperatures at ground level fluctuate drastically, from below oo to more than 25° in drier periods. Plants are adapted to rapid changes in insolation, intense ultra-violet radiation, low temperatures, and reduced water supply (Runde! & Smith, 1994). Descriptions of paramo vegetation were provided by Weber (1958) for Costa Rica, by Vareschi (1970), Monasterio (1980, 1986), and Smith (1981) for Venezuela, and by AcostaSolis (1984) and Ramsey (1993) for Ecuador. The general ecology of Colombian paramo vegetation was described by Cuatrecasas (1934, 1958, 1968), Cleef (1978, 1981), and Rangel (1991). Three elevation zones were distinguished within the paramo belt (Cuatrecasas, 1958). The subparamo is the lowest wne, a shrubby transition between forest and grass paramo. Upwards, the grass paramo or paramo proper is found, composed mainly of bunch grasslands. The uppennost zone is the superparamo, where plant growth is scattered, related to harsh climatic conditions. The present work focuses on the middle grass paramo from 3800 to 44.00 m.,

Introduction Chapter 1 On the basis of palynological evidence, Van der Hammen & Cleef (1986) concluded that natural paramo grasslands have occurred since Plio-Pleistocene times. As a result of recent human influence, the uppermost Andean forest has in many locations been replaced by grassland in which. paramo elements dominate.

The zonal plant communities of the grass paramo are characterized by bunch grasses of Ca/amagrosris and Feswca, small-ieaved shrubs, short grasses, ground rosette species, terrestrial bryophytes and lichens, and sedges (Cyperaceae). Giant rosette plants of Espeleria (Asteraceae) and Puya (Bromeliaceae) occur frequently. Azonally, cushion bogs occur, whereas a bamboo vegetation of Chusquea is common in the most humid paramo regions (Cleef, 1981; Bosman eta/., 1993; To! & Cleef, 1994).

A common management practice in paramo ecosystems of the northern Andes is the combination of an extensive grazing system with regular burning of the vegetation to remove dead biomass and stimulate the regeneration of more palatable young grass shoots (Verweij & Budde, 1992). Until recent years, the combined effects of grazing and burning on paramo vegetation received little scientific attention, except for some qualitative observations by . Grubb (1970). Research in the Colombian national park of Los Nevados has been focused on the combined impacts of burning and grazing (among others: Hofstede, 1995; the present dissertation).

Apart from natural variation, many high-mountain vegetations of the neotropics are characterized by variation due to human influences (Ellenberg, 1979; .Brush, 1982; Lauer, 1993). The following paragraphs describe recent land cover changes associated with different land use patterns. ·

The recent trend of upward expansion of the agricultural frontier into the lower paramo zones is due to high population pressure in the lowlands (Ferwerda, 1987; Hess, 1990; Verweij & Beekman, 1995). The emphasis is on the cultivation of potatoes and, to a minor extent, onions. Traditional land use systems of wheat and potato cultivation in the Venezuelan paramo were described by Sarmiento eta/. (1990, 1993). In both modem and more traditional systems, an intensification of the crop rotation system is observed. This is associated with shorter fallow periods, larger proportions of land under cultivation, and the lncreased use of fertilizers, herbicides, and pesticides (Hess, 1990; Verweij & Beekman, 1995). According to Ferwerda (1987), regeneration of paramo bunch grasslands after potato cultivation takes more than 70 years.

The palflmO has been subject to a process of degradation and conversion. The open vegetation of the paramo belt is expanding at the cost of the Andean forest. However, the actual intensification of land use leads to a gradual or sometimes abrupt transformation into man. made systems that in practice do not revert to the initial situation. Although floristic composition has changed to various degrees as a result of the use made by man, the pararno flora is still largely composed of native species. The occurrence of plant species in noncultivated pararno areas is determined in first instance by climate, hydrology and substrate, and secondarily by land use. Management strategies must therefore be based on the understanding of the ecological relationships between the plant community and other land attributes and how these are modified by management.

A parallel contemporary development is the artificial lowering of the forest line due to wood cutting and the subsequent establishment of pastures (Verweij & Beukema, 1992; Koker at., 1995). The wood is used as fuel and for fencing. Analogous to the cultivation practices, clearing the upper-Andean forest results in a drastic and permanent cover change: in this case into more intensively managed pastures with sown grasses. Market-oriented animal raising activities increasingly take place around the former natural forest-paramo ecotone.

The present work focuses on the evaluation of human impacts in the non-cultivated paramo, in relation to limiting conditions and potential for sustainable development. Potential for sustainable development The paramo is a fragile ecosystem. This results in a range of limiting conditions that should · be taken carefully into account before planning any development action. Afforestation attempts using introduced species such as monocultures of Pinus and Cupressus were often unsuccessful. As a result, I consider the following characteristics of the pararno ecosystem a fourfold potential for sustainable development: (l) its high biological diversity, (2) its importance for the regional hydrological system, (3) its suitability for specific agricultural activities, and (4) its suitability for specific forms of.tourism and recreation.

Under the influence of extensive grazing, the tussock grasslands of the paramo frequently develop into a mosaic pattern of spatially and temporally alternating vegetation succession stages. Destruction of plant cover by cattle grazing and trampling was reported by Grubb (1970), Cleef (1981), Verweij & Budde (1992), Verweij & Kok (1992), Perez (1992, 1993), Ramsey (1993), and Hofstede (1995). Man-made fires represent another form of disturbance. fn the literature, examples are given of the impacls of single fire events. Vegetation response to fire in the Costa Rican paramo was described by Janzen (1973), Chaverri et at. (1976), Williamson et at. (1986), and Horn (1988, 1989). A basis for comparison is furthermore provided by the post-fire succession on Mount Wilhelm, Papua New Guinea, as studied by Corlett (1987), and by a description of the recovery from fire of an alpine vegetation at Mount Kilimanjaro (Beck et at., 1986). The general tendency in these tropical alpine grasslands is a quick regeneration of the grass tussocks within a few years after fire, whereas the shrubs are characterized by a slow recovery taking several decades. Some ericaceous shrubs exhibit a greater regeneration capacity. It is debated to what extent the pararno ecosystem is natural or man-made as a product of fires in the past (Ellenberg, 1979; Vander Hammen & Cleef, 1986; Lregaard, 1992).

Biological diversity The paramo represents a chain of cool island-like habitats, which have repeatedly expanded and shrunk during the late Pliocene and Pleistocene. A number of studies deal with the phytogeography of the vascular flora of the Andean paramos·(Cuatrecasas, 1979; Cleef, 1978, ·1979, 1983; Cleef & Chaverri, 1992; Cleef & Rangel, 1984; Vander Hammen & Cleef 1983, 1986). The diversity of animal life has not been fully explored; an outline regarding the paramo fauna was presented by Stunn (1978). The.floristic elements of the paramo are mainly of neotropical, holarctic, and austral-subantarctic origin. .

'•

l 'i

lmroduction

Chapter 1 Many plant species are exclusively adapted to the cool paramo environm ent with its insular and dynamic character, resulting in a high degree of endemism. The paramo flora comprises between 3000 and 4000 species of vascular plants, with an estimated endemis m at species . level as high as 60% (Luteyn et al., 1992). The paramo flora has the highest degree of endemism and is among the richest found in the high mountains in the world (Cleef, 1983; Smith & Cleef, 1988). The high diversity and the biogeographical and evolutionary significance are intrinsic values of the paramo. These, together with a rapid rate of conversion into.man-made systems, qualify the pararno as a high priority area for conservation and research. The plants' potential medicinal uses seem important. Of the 77 medicinal plants used by the indigeno us people of the Cogui culture of the Sierra Nevada de Santa Marta, 56% grow in the paramo (Carbone, 1987: cited by Rangel, 1989).

Hydrology . The paramo has a crucial role in the regulation of water flow at local, reg1onal , and smaller 1 scales. At the level of the regional hydrological system, it is often assumed that human activities in the paramo have negatively affected the water quality and the stability of water supply, and have caused the silting of rivers and lowland lakes (Guhl! 1968; Luteyn, 1992). · Hofstede & Sevink (1995) found that in drier periods the undisturbed paramo has a larger water-storing capacity than burned or grazed sites. A dense vegetation layer of bunch grasses probably buffers fluctuations in soil moisture. A stable water supply to lower areas is of vital importance to the majority of the inhabita nts of Ecuador, Colombia, and Venezuela: they depend on the paramo streams as the principal source of drinking and irrigation v.:ater. Another related use is the operation of hydroelectric power plants. The importan ce at national and continental levels is indicated by the fact that most hydrological systems originate in the high Andes. Agriculture Jodha (1992) mentioned several aspects of mountain environments that determin e the potential for sustainable mountain agriculture, such as accessibility, fragility , marginal ity, and diversity. Fragility and marginality of paramo systems are related to low biomass production, and thus low regeneration rates and a low carrying capacity. Some land use practices are complementary to the economy of the adjacent lowlands and may have a sustaina ble basis (Sarmiento et al., 1993). Examples are the cultivation of tubers, cereals, and the cash crops onions and garlic. The paramo has an important potential as contribu tor to the genetic diversity of pre-Hispanic tubers, mainly potato (Solanum tuberosum), cubio (fropaeolum tuberosum), oca (Oxalis tuberosa}, and ullucu (Ullucu tuberosus). It is a suitable zone for growing seed potatoes, especially at the highest elevations where •. diseases hardly occur. In association with a variety of agricult ural producti on systems , there are specific forms of social organization characterized by complex labour relationships. These 1£ have hardly been studied, except in Ecuador and Peru (Schjellerup, 1989; Hess, 1990).

Tourism and recreation The scenic beauty and peacefulness of the paramo offer the potential for specific forms of recreation, such as hiking, fishing, and camping. Ecotourism in paramo areas is starting to be promoted by national and international organizations.

1.2

Research objectives

The aim of the present dissertation was to describe the effects of grazing and burning on paramo vegetation dynamics: A major hypothesis was that the actual vegetatio n pattern of the paratno proper is to a large extent explained by vaiiation in grazing and burning management. The following partial research objectives were formulated: to identify the key factors that explain the variation in grazing intensity and to model cattle distribution to reconstruct fire history by determining where, how frequently, and why fires occur to explain the present distribution of vegetation patterns in terms of natural variation and of variation due to management to replicate and predict vegetation development over several decades to draw conclusions on the implicat ions for management

1.3

The modelling approach

Models have been used in ecology to understand ecosystem properties. There are many different basic approaches to modelling. They can be contrasted as mechani stic (explaining actual processes) or empirical (based on observed correlations between controlli ng variables and processes); as deterministic (strict cause and· effect) or stochasti c (incorporating uncertainty); and static or dynamic. Dynamic models with stepwise time-bas ed calculations are called simulation models. They can be used as scientific tools and as management tools. The ideas underlying the use of models as scientific tools were described by J~rgenson (1994) , as an iterative developmeqt of a pattern, a stepwise procedure toward a better understanding of nature at system level. Carlson eta/. ( 1993) mentioned four different function s that can be performed by biophys ical research models: to organize and structure current knowledge on ~ a system of interest; to identify knowledge gaps to guide research ; to foster a multidisc iplinary approach to research; and finally to provide a cost-effective means to study the behaviour and interactions of complex systems . . Biophysical models can be used to predict responses of ecosystems to managem ent practices. These management models are used as tools in macro-scale planning, cost-ben efit analysis, and at operational level to select an optimal management strategy. Van Wijngaa rden (1985) and Starfield & Bleloch (1986) showed how models of herbivore populati on dynamics can be used as management tools at operational leveL Management models are often coupled to decision support systems (e.g. Bosch & Booysen, 1992). The models develope d in the present study are not true management models but rather research models. Their outcome s, however, do have important implications for management. · ·

- = - -__,_,;.__

4__ _ . ··-----4·-- -- - ---

Inrroducrioll Chapter I In the literature, two kinds of simulation model can be found that deal with plant-herbivore systems: discrete transition models (Swanzman & Singh, 1974; Redetzke & Van Dyne, 1976; Runkle, 1981; Usher, 1981; Lippe eta/., 1985; Turner, 1987; Pels & Verweij, 1992) and· continuous models (Smith & Williams, 1973; Noy-Meir, 1976, 1978; Innis, 1978; Fetcher, 19S!·' Walker et al., 1981; Pellew, 1983; Van Wijngaarden, 1985; Thalen eta/., 1987). . . . For transition models, a number of possible static states are defined at the level of single species (Usher, 1966; Runkle, 1981; Lippe et al., 1985) or at the scale of vegetation types' (Austin, 1980; Austin & Belbin, 1981; Usher, 1981). A group of models known as the JABOWA family of models has been used in many parts of the world to model forest growth on the basis of the cumulative growth of individual trees (Botkin et a/., 1973; Shugart & . West, 1977; Prentice, 1987; Nisbet & Botkin, 1993). In these discrete stochastic models, each state has -a certain probability of changing to another state. The transition probabilities are . kept constant in a Markov model (Van Hulst, 1979; Usher, 1981; Lippe eta/., 1985). Both terrain characteristics and the changing mosaic of vegetation patches that develops under grazing influence the grazing behaviour of canle. Taking into account the spatial and temporal · variation in grazing intensity, dynamic transition probabilities that incorporate these dimensions are preferred (Turner, 1987; Pels & Verweij, 1992). Transition models are also . applied to the simulation of succession in plant communities subject to fire (Noble & S1atyer, 1981~ ~ Transition probabilities between states of the landscape are often calculated from time series of remote sensing images. An example was provided by Johnston & Naiman (1990), who constructed a Markov model to describe the long-term landscape alteration by beaver (Castor , canadensis). When sufficient information is available on the processes that underlie the . observed changes, continuous modelling is preferred. Continuous simulation is applied in the · present work as it provides a better way to incorporate knowledge on correlations between : controlling variables and processes. French (1990) evaluated different types of continuous models according to scales of time and space. Relevant levels for the study of ·grazing systems are those of ecophysi?logy, biotic ·~ production, and the ecosystem. At the ecophysiologicallevel, Fetcher (1981) stud1ed the effect li of shoot removal on cold desert shrubs by simulating energy flows between plant organs. The . grazing model of Smith & Williams (1973), the ELM-Grassland Simulation Model (Innis, ·' 1978), the vegetation growth functions described by Noy-Meir (1978), and the SPUR model (Hanson et al., 1988) are examples at the level of biotic production. Walker eta/. (1981), :: Pellew (1983), Van Wijngaarden (1985), and Thalen et al. (1987) described the effects of ~ grazing on the interactions between different plant groups at the community or ecosystem : level. Forage quality and the food preference of herbivores are rarely taken into account • Forage quality was incorporated in the models described by Rice eta/. (1983) and Starfield ; & Bleloch (1986). Most ecological simulation models, whether discrete or continuous mathematics are used, are 1 based on a set of simplifying difference or differential equations. Steady-state assumptions ,'• are common and spatial heterogeneity is generally not considered. System components and .• processes are often spatially Jumped together, thus referring to an idealized ecosystem. In this "' way, the effect of certain management decisions can be completely masked. ~ ·~

When the state of one spatial unit ~ffects the future state of another, spatial modelling is required (Hunsaker er al., 1993). In paramo gr.tsslands, a mosaic of vegetation patches with ·different degrees of utilizmion develops under the i ntluenc~ of burning and grazing. The development of ~dch patch is influenced by its present state and by the degree of utilization of surrounding patches. Therefore, spatial modelling was included in the analysis of management effects on the paramo grasslands. · Maps and geographic information systems (GIS) can provide us with not only important information on vegetation patterns in time (D. Miller, 1994), but also information on the spatial variation of management variables themselves in relation to land attributes and management decisions. The spread of fire or other disturbances across heterogeneous landscapes was analyzed using a simple landscape model of cell arrays (Green, 1989; Turner et at., 1989). An overview of current spatial mode.lling.applied tR eca!Dgical syilems was provided by Hunsaker er al. (1993). Applications of GlS to landscape ecology were reviewed by Haines-Younger al. (1993). Examples of GIS application s for conservation were presented in another recent book (R.I. Miller, 1994). Most authors who worked on a coupling of environmental models and GIS reported on limitations that must be overcome to facilitate integration (Livingstone & Raper, 1994). Most demonstrations of spatial aspects in ecological models display independent effects as if they were integrated in space, but there is no spatial interaction between cells (Hunsaker et at., 1993). Simulation models based on ~ low number of interacting cells or polygons do exist (Sklar er al., 1985; Boumans & Sklar, 1990). GIS has often been used as only an output display device, and not as an integral part of dynamic simulation (Nisbet & Botkin, 1993). In other cases, use is made of GIS modelling capabilities, the models being incorporated as GIS subroutines. However, !.O date GIS-driven modelling has been limited to specific GIS capabilities such as map overlayin·g; it is not yet suitable for complex models. The approach chosen here was not to try to overcome the practical limitations to coupling spatial and temporal models. Appropriate software packages and interfaces will soon become available and increasing hardware_capacity will favour linkage by permitting more complex calculations. However, appropriate models and concepts have not been fully developed. This is seen as a major constraint to integration. Much depends on a proper definition of the spatial units and a proper understanding of their interactions. Only then will a true integration of the spatial dimension into ecological modelling occur. In the present work, spatial distributions and vegetation response in time were modelled separately, meanwhile identifying relevant common variables and parameters that will facilitate future integration. By means of a more qualitative assessment, the outcomes of both types of model were compared to check on their validity and relevanc~ for management. A major problem that needed to be addressed was how to deal with spatio-temporal variability. Data from tropical mountain environments is noisy and is characterized by a large amount of variation due to differences in elevation and corresponding temperature, slope, aspect, drainage conditions and other factors. Su~marizing, the natural spatial variation of tropical mountain ecosystems is high. As described above, management causes changes in · vegetation structure and composition in most paramo areas. Therefore, the time component :is also important in the characterization of vegetation patterns. 19

Inrrodaction Chapter 1

Chapters 7 and 8 show the development of spatial models of the management regime in relation to relevant land attributes and management decisions. By means of multiple regression analysis, the key factors explaining the variation in grazing intensity are identified. This leads to a spatial model of cattle distribution: it predicts where and with what intensity livestock graze (Chapter 7). The fire history of part.of the study area is reconstru cted. The resulting GIS model gives insight into where and how frequently burning occurs. Following the description of these research models and their outcomes, the implications for managem ent are discussed in the final Chapter 9. A separate section is dedicated to the analysis and discussion of the overall effects of management on plant species diversity.

Wiue (1994) considered that the dimensionality and the noisy character of the data need to be reduced for the purpose of quantitative analysis, by either smoothing technique s or data~ compression. In the present work, the following strategy was developed to handle spatial and temporal variability. Where possible, original data were used instead· of classified or smoothed data.' The natural variation within the data set was f1rst reduced, step by step. In all samples, management was quantified as well as possible by estimating the time since f1re occurrenc~~ and by measuring grazing intensity. Classification and ordination techniques were useful tools in the process of 'zooming in' until a data set remained in which management explained most of the variation . The subsequent analysis of point data provided a clear picture of the main impacts of management on the vegetation. ·,

LITERATURE CITED

The models developed in this study in fact represent spatial and temporal inter- and : extrapolations of the main trends. Time series of vegetation development after f~re formed the basis for the construction of an empirica l, continuous simulation model at ecosystem level. . It is a deterministic model in the sense that it describes the development of determine d plant , groups in response to management. ~ For the spatial models of land attributes ahd management regime, extrapolation of the . f' established correlations was based on visual interpretation of a time series of aerial ; photographs. The relationship between photo image characteristics and the objects in reality l is of key importance: it determines to what extent both spatial and temporal dimensio ns can be incorporated in the models. GIS was used to integrate data from different informati on sources, to interpolate data, to analyze temporal change, and to study the relationsh ip between ·t . m'"•gomoot '"' l"d 'llrib•tt•. •

I l

Outline In this chapter, Chapter 1, an introduction is ·given to the research problems, objective s, and the modelling approach. Chapter 2 presents the biophysical and socio-economic setting of the study are<~, the paramo ecosystem of Los Nevados National Park in the Colombian Cordillera Central. Actual land use patterns and land use history are discussed. The vegetatio n patterns are characterized in Chapter 3 by means of classification, ordination, and mapping techniques.

The major impacts of burning and grazing on vegetation structure and composit ion are identified. Chapter 4 provides a description of the extensive livestock system from the perspective of the cattle. Grazing behaviour, intake, and secondary production are analyzed. The resulting information on forage quality and forage preference of cattle is an essential input for the models described in Chapters 6 and 7. Chapter 5 ·also provides important elements for model construction. The effects of burning and grazing on the two principal plant groups of the zona·! grassland paramo are quantifided. A model offfthe hum~·infl fubencedh , i dynamics of stem rosette populations is presented an the process o ragmentauon 1 o unc 1 grasses is documented. The building blocks for the conceptual development of the simulation model described in Chapter 6 are then identified. Responses of specific plant groups to fire occurrence and grazing intensity are simulated. Some of the parameters are known from previous chapters; others are estimated by a calibration procedure.

Acosta-Solis, M. 1984. Los pararnos andinos del Ecuador. Publicaciones Cientffica s MAS., Quito. Almeida, L., Cleef, A.M., Herrera, A., Velazquez, A. & Luna, E. 1994. El zacatonal alpino del Voldn Popocatepetl, Mexico, y su posici6n en las montaiias tropicales de America. Phytocoenologia 22(3): 391-436. Austin, M.P. 1980. An exploratory analysis of grassland dynamics: an example of a lawn succession. Vegetatio 43:.87-94. Austin, M.P. & Belbin, L. 1981. An analysis of succession along an environmental gradient using data from a lawn. Vegetatio 46: 19-30. Beck, E., Scheibe, R. & Schulze, E.-D. 1986. Recovery from f~re : observations in the alpine vegetation of western Mt. Kilimanjaro (Tanzania). Phytocoenologia 14(1): 55-77. Bosch, O.J.H. & Booysen, 1. 1992. An integrative approach to rangeland condition and capability assessment. Journal of Range Management 45: 116-122. Bosman, A.F., Vander Molen, P.C., Young, R. & Cleef, A.M. 1993. Ecology of a paramo cushion mire. Journal of Vegetation Science 4: 633-640. Botkin, D.B., Janak, J.F. & WJi!!is, J.R. 1973. Some ecological consequences of a computer model of forest growth. Journal of Ecology 60: 849-872. Boumans, R.M.J. & Sklar, F.H. 1990. A polygon-based spatial (PBS) model for simulating landscape change. Ecological Modellin g. Brush, S.B. 1982. The natural and human environment of the central Andes. Mountain Research and Development 2(1): 19-38. Carbono, E. 1987. Estudio etnobotanico entre los Cogui de Ia Sierra Nevada de Santa Marta. Tesis de Magister en Sistematica, Instituto de Ciencias Naturales, Universid ad Nacional de Colombia, Bogota. Carlson, D.H., Thurow, T.L. & Jones, C.A. 1993. Biophysical simulation models as a foundation of decision support systems. Pp. 37-67 in: Stuth, J.W. & Lyons, B.G. (eds.), Decision support systems for the management of grazing lands. Man and the Biosphere Series 2. UNESCO/The Parthenon Publishing Group, Paris. Chaverri, A., Vaughan, C. & Poveda, L.J. 1976. Informe de la gira efectuada a! Macizo de Chirrip6 a rafz del fuego ocurrido en marzo de 1976. Revista de Costa Rica II: 243279.

lncroduction Chapter I Cl f AM 1978 Characteristics of nco-tropical paramo vegetati on and its subantarctic ee' ' : ' Troll c & Lauer w (eds) Geoecological relations between the re1auons. 1n: , . ' . ., . haf southern temperate zone and the tropical mountains. Erdwissensc tI'IChe Forschung ·, 11 : 365-390. ar n · Cleef, A.M. 1979. The phytogeographical position of the neo_trop. ical vascular p -~o ora with special reference to the Colombian Cordillera Onent~. Pp. 175-184 m. Larsen, K & Holm-Nielsen, L.B. (eds.}, Tropical botany. AcademiC Press, London. . Cleef, A~M. 1981. The vegetation of the paramos of the Colomb ian Cordillera Oriental. _ Dissertationes Botanicae 61. J. Cramer, Vaduz. 320 PP· Cleef, A.M. 1983. Fitogeografia y composici6n de Ia flo~a vascula r de los paramo~ de Ia . Cordillera Oriental Colombiana (Estudio comparauvo con otras alta: montanas del , tr6pico). Revista de Ia Academia Colombiana de Ciencias Exactas, Ffsicas y Naturales · 15(58): 23-29. . Cleef, A.M. & Chaverri, P. 1992. Phytogeography of the pararno flora of Cor~llera. de Talamanca, Costa Rica. Pp. 45-60 in: Balslev, H. ~ Luteyn, J.L. (eds.}, Paramo. an Andean ecosystem under human influence. Academtc Press, London . . Cleef AM & Rangel, 0. 1984. La vegetaci6n del p~ramo del noroest e de Ia Sterra N~vada ' d~ Santa Mana. Pp. 203-266 in: Van der Hammen, T. & Rufz, P.M. (eds.), Studies on Tropical Andean Ecosystems 2. J. Cramer, Berlin. . h Coe, M.J. 1967. The ecology of the alpine zone of Mt. Kenya. Dr. W. Junk Pubbshers, T e ,_ Hague. 136 pp. · . · 19· ' Corlett, R.T. 1987. Post-fire succession on Mt. Wilhelm, Papua New Gumea. 8 Iotropic a . 157-160. · d 1M N · I Cuatrecasas, J. 1934. Observaciones geobot~icas en Colo~bia. TrabaJO S e useo ac10na de Ciencias Naturales Series.Botanicas, 27. Madrid. 114 P_P. . , Cuatrecasas, J. 1958 · Aspectos de Ia vegetac i6n natural de Colombia. Revista de Ia Academia 1 221 264 Colombiana de Ciencias Exactas, Fisicas y Naturales 10(40): - · c atrecasas J 1968 Paramo vegetation and its life forms. In: Troll, C. (ed.), Geoecology of , u the ~o-untain~us regions of the tropical Americas. Colloquium Geogra phicum 9: 163- 1 · 186 · Comparaci6n fitoge~grafica de paramos entre varias Cordilleras. Pp. 89Cuatrecasas, 1- 1979· iCe d 99 in: Salgado-Labouriau, M.L. (ed.), El medio-ambiente plllamo . ntro e Estudios Avanzados, Caracas. . . h Ellenberg, H. 1975. Vegetationsstufen in perhumiden his peraride n BereJchen der tropiSC en Anden. Phytocoenologia 2(3/4): 368-387. . . Ellenberg, H. 1979. Man's influence on tropical mountain ecosyst ems m South Amenca. Journal of Ecology 67: 401-416. , . Ferwerda, W. 1987. The influence of potato cultivation on the natural bunchgrass paramo m ·I the Colombian Cordillera Oriental. MSc. thesis, Internal repon no. 220, Hugo de Vries-Laboratory, University of Amsterdam. 83 pp. _ . Fetcher, N. 1981. Effects of grazing on cold desert shrubs: a snl)ulat iOn model based on relative growth rate. Ecological Modelling 13: 49-86. .. . Fosberg, F.R. 1959. Upper limits of vegetation on Mauna Loa, Hawan . ~cology 40. 237-266 . ~ French, N.R. 1990. The utility of models in the study of mounta m development and transformation. Mountain Research and Development 10: _141- 149. .. Green, D.G. 1989. Simulated effects of fire, dispersal and spal!al pattern on compeuuon within forest mosaics. Vegetatio 82: 139-153.

Grubb, PJ. 1970. The impact of man on the Cerro Antisana, Ecuado r. Journal of Applied Ecology 7(2): 7P-8P. Guhl, E. 1968. Los paramos circundantes de Ia Sabana de Bogota; ~u ecologfa y su importancia para el regimen hidro16gico de Ia misma. Colloquium Geographicum 9: 195-212. Haines- Young, R., Green, D.R. & Cousins, S.H. (eds.) 1993. Landsc ape ecology and GIS. Taylor & Francis, London. 288 pp. Hanson, J.D., Skiles, J.W. & Panon, W.J. 1988. A multi-species model for rangeland plant communities. Ecological Modelling 44: 89-123. Hedberg, 0. 1964. Features 'of afroalpine plant ecology. Acta Phytoge ographica Suecica 49: 1-144. Hedberg, 0. 1969. Growth rates of the East African giant Senecio s. Nature-22: 163-164. Hedberg, 0. 1992. Afroalpine vegetation compared to paramo: converg ent adaptations and · divergent differentiation. Pp. 15-31 in: Balslev, H. & Luteyn, J.L. (eds.), Paramo: an Andean ecosystem under human influence. Academic Press, London . Hess, C.G. 1990. "Moving up - moving down": agro-pastoral land-use patterns in the Ecuadorian paramos. Mountain Research·and Development 10(4): 333-342. Hnatiuk, R.J. 1978. The growth of tussock grasses on an equatorial high mountain and on two subantarctic islands. Erdwissenschaftliche Forschu ng 11 : 159-190 . Hofstede, R.G.M. 1995. Effects of burning and grazing on a Colomb ian paramo ecosystem. PhD dissertation, University of Amsterdam. 199 pp. Hofstede, R.G.M. & Sevink, J. 1995. Water and nutrient storage and input:output budgets in burned, grazed and undisturbed paramo grasslands. Pp. 121-147 in: Hofstede, R.G.M., Effects of burning and grazing on a Colombian paramo ecosystem. PhD dissertation, University of Amsterdam. Hom, S.P. 1988. Effect of burning on a montane mire _in the Cordill era de Talamanca, Costa Rica. Brenesia 30: 81-92. Hom, S.P. 1989. Postfrre vegetation development in the Costa Rican paramos. Madroiio 36(2): 93- 114. Hunsaker, C.T., Nisbet, R.A., Lam, D., Browder, J.A., Baker, W}, Turner, M.G. & Botkin, D. B. 1993. Spatial models of ecological systems and processes: the role of GIS. Pp. 248-264 in: Goodchild, M., Parks, B. & Steyaen, L. (eds.), Geogra phic information systems and environmental modeling. Oxford University Press, New York. Innis, G.S. (ed.) 1978. Grassland simulation model. Ecological Studies 26. Springer-Verlag, New York. Islebe, G.A. & Cleef, A.M. 1995. Alpine plant communities of Guatem ala. Flora 190: 79-87. Janzen, D.H. 1973. Rate of regeneration after a tropical high elevatio n fire. Biotropica 5: 117122. ' . Jodha, N.S. 1992. Mountain perspective and sustainability: a framew ork for development - strategies. Pp. 41-82 in: Jodha, N.S., Banskota, M. & Panap, T. (eds.), Sustainanable mountain agriculture, perspectives and issues. Volume I. Oxford & ffiH Publishing Co. Pvt. Ltd:, New Delhi. Johnston, C.A. & Naiman , R.J. 1990. Aquatic patch creation in relation to beaver population trends. Ecology 71: 1617-1621. J~rgensen, S.E. 1994. Fundamentals of ecological modelling (2nd edition). Developments in environmental modelling, 19. Elsevier, Amsterdam. 628 pp.

lmroduction Chapter 1 Kok, K., Verweij, P.A. & Beukema, H. 1995. Effects of cutting and grazing on Andean for .line vegetation. In: Churchill, S.P., Balslev, H., Forero, E. & Luteyn, J.L. (eds. Biodiversity and conservation of neotropical montane forests. Scientifi c Pub!., . New York Botanical Garden. Llf!gaard, S. 1992. Influence of ftre in the grass paramo vegetation of Ecuador . Pp. 151-11 in: Balslev, H. & Luteyn, J.L. (eds.), Paramo: an Andean ecosystem under hu influence. Academic Press, London. Lauer, W. 1976. Zur hygrischen Hohenstufung tropischer Gebirge. Biogeog raphica 7: 16~. 182. Lauer, W. 1993. Human development and environment in the Andes: a geoecologic_ overview. Mountain Research and Development 13(2): 157-166. Lippe, E., De Smidt, J.T. & Glenn-Lewin, D.C. 1985. Markov models and success1on: a tes from a heathland in The Netherlands. Journal of Ecology 73: 775-791. , Livingston, D. & Raper, J. 1994. Modelling environmental systems with GIS: theoretic barriers to progress. Pp. 229-239 in: Worboys, M.F. (ed.), Innovations in GIS; selecte& papers from the ftrst national conference on GIS research UK. Taylor & Francis, London. Luteyn, J.L. 1992. Paramos: Why study them? Pp. 1-14 in: Balslev, H. & Luteyn, J.L. (eds.), Paramo:' an Andean ecosystem under human influence. Academic Press, London. Luteyn, J.L., Cleef, A.M. & Rangel, 0. 1992. Plant diversity in paramo: towards a checklist of paramo plants and a generic flora. Pp. 71 -84 in: Balslev, H. & Luteyn, J.L. (eds.), Paramo : an Andean ecosystem under human influence. Academic Press, London. . Miller, D. 1994. Coupling of process-based vegetation models to GIS and knowledge-baseq systems with reference to vegetation change. Pp. 241-250 in: Worboys, M.F. (ed.), Innovations in GIS; selected papers from the first national conference on GIS research UK. Taylor & Francis, London. Miller, R.I. (ed.) 1994. M~pping the diversity of nature. Chapman & Hall, London. 218 pp. Monasterio, M. (ed.) 1980. Estudios ecol6gicos en los paramos andinos. Universidad de los Andes, Merida, Venezuela. Monasterio, M. 1986. Adaptive strateg.ies of Espeleria in the Andean desert paramo. Pp. 49· 80 in: Vuil!eum.ier, F. & Monasterio, M. (eds.), High altitude tropical biogeog raphy. Oxford Umverslly Press, New York. ·. Nisbet, R.A. & Botkin, D.B. 1993. Integrating a forest growth model with a geographic information system. Pp. 265-269 in: Goodchild, M., Parks, B. & Steyaert , L.. (ed~.), Geographic information systems and environmental modeling. Oxford Umverslly . Press, New York. .. . Noble, I.R. & Slatyer, R.O. 1981. Concepts and models of succession in vascular plant communities subject to recurrent fire. Pp. 311-335 in: Gill, A.M., Groves, R.H. & Noble, l.R. (eds.), Fire and the Australian biota. Australian Academy of Science, Canberra. Noy-Meir, I. 1976. Rotational grazing in a continuously growing pasture: a simple model. Agricultural Systems 1: 87-112. Noy-Meir, L 1978. Stability in simple grazing models: effects of explicit functions. Journal of Theoretical Biology 71: 347-380. Pels, B. & Verweij , P.A. 1992. Burning and grazing in a bunchgrass paramo ecosystem: vegetation dynamics described by a transition model. Pp. 243-2~3 in: Balslev, H. ~ Luteyn, J.L. (eds.), Paramo: an Andean ecosystem under human tnfluenc e. Academtc Press, London.

I 1

1 1

1·

. 1

Pellew, R.A.P. 1983. The impacts of elephant, giraffe and fire upon the Acacia toni/is woodlands of the Screnget i. African Journal of Ecology 21 : 41-74. Perez, F.L. 1992. The ecological impact of catth: on ~:au b~:cm Andean rosettes in a high Venezuelan paramo. Mountain Research and Development 12(1): 29-46. perez, F.L. 1993. Turf destruction by cattle in the high equatorial Andes. Mountain Research and Development 13(1): 107-110. · Prentice, I.C. 1987. Description and simulation of tree-layer composi tion and size distributions in a primaeval Picea-Pinus forest. Vegetatio 69: 147-156. Ramsey, P.M. 1993. The paramo vegetation of Ecuador: the community ecology, dynamics and productivity of tropical grasslands in the Andes. PhD dissenation, Universi ty of Wales. 274 pp. Rangel, 0 . 1989. Caracteristicas bioecol6gicas y problem<\tica de manejo de Ia regi6n . paramuna de Colomb ia. Suelos Ecuatoriales 19(1): 11-18. .. Rangel, 0. 1991. Vegetaci6n y ambiente en tres gra~ientes montaiiosos de Colombta. PhD dissertation, University of Amsterdam. 349 pp. Redetzke, K.A. & Van Dyne, G.M. 1976. A matrix model of a rangelan d grazing system. Journal of Range Management 29: 425-430. Rice, R.W., MacNeil, M.D., Jenkins, T.G. & Koong, L.J. 1983. Simulation of the herbage/animal interface of grazing lands. Pp. 475-488 in: Lauenro th, W.K., Skogerboe, G. V. & Flug, M. (eds.), Analysis of ecological sy_stems : state of the art in ecological modelling. Elsevier, Amsterdam. Runde!, P.W. & Smith, A.P. (eds.) 1994. Tropical alpine environments: plant form and function. Cambridge University Press, Cambridge. 376 pp. Runkle , J.R. 1981. Gap regeneration in some old-growth forests of the eastern United States. Ecology 62: 1041-1051. Sarmiento, L., Monasterio, M. & Mantilla, M. 1990. Succession, regenera tion and stability in high Andean ecosystems and agro-ecosystems: the rest-fallow strategy in the Paramo de Gavidia, Venezuela. Pp. 151-157 in: Winiger, M., Wiesma nn, U. & Rhecher, J.R. (eds.), Mount Kenya area: differentiation .and dynamics of a tropical mountain ecosystem. Geogr. Bern, African Study Series A8. Sarmiento, L., Monasterio, M. & Mantilla, M. 1993. Ecological bases, sustainability, and current trends in traditional agriculture in the Venezuelan high Andes. Mountain Research and Development 13(2): 167-176. Schjellerup, I. 1989. Children of the stones. The Royal Danish Academ y of Sciences and Letters' Publication No. 7, Copenhagen. 96 pp. Shugart, H.H. & West, D.C. 1977. Development of an Appalachian deciduo us forest model and its applications to assessment of the impact of the chestnut blight. Journal of Environmental Management 5: 161-179. Sklar, F.H., Costan~. R. & Day, J.W. 1985. Dynamic spatial simulation modeling of coastal wetland habitat S!Jccession. Ecological Modelling 29: 261-281. , Smith, A.P. 1981. Growth and Population Dynamics of Espeleria (Compo sitae) in the Venezuelan Andes. Smithsonian Contributions to Botany 48: 1-45. Smith, A.P. & Young, T.P. 1987. Tropical alpine plant ecology. Annual Review of Ecology and Systematics 18: 137-158. · . ·Smith, J.M.D. 1986. Origins and history of the Malesian high mountai n flora. Pp. 469-477 in: Vuilleum ier, F. & Monasterio, M. (eds.), High altitude tropical biogeog raphy. Oxford University Press, New York.

Introduction Chapter I Smith, J.M.D. & Cleef, A.M. 1988. Composition and origin of the world's tropicalpine fl Journal of Biogeography 15: 631-645. Smith, R.C.G. & Williams, W.A. 1973. Model development for a deferred-grazing syst Journal of Range Management 26: 454-460. Starfield, A.M. & Bleloch, A.L. 1986. Building models for conservation and wil management. Macmillan Publishing Company, New York. Sturm, H. 1978. Zur Oekologie del andinen Paramoregion. Biogeographica 14. The Ha 121 pp. . Swartzman, G.L. & Singh, J.S. 1974. A dynamic programming appr9ach to optimal gr · strategies using a succession model for a tropical grassland. Journal of Appl Ecology 11: 537-548. · ' Thalen, D.C.P., Poorrer, H., Lotz, L.A.P. & Oosterveld, P. 1987. Modelling the struc changes in vegetation under different grazing regim~s. Pp. 167-183 in: Van Andel, Bakker, J.P. & Snaydon, R.W. (eds.), Disturbance in grasslands: causes, effects. processes. Dr. W. Junk Pub!., Dordrecht. Tol, G.J. & Cleef, A.M. 1994. Above-ground biomass structure of a Chusquea tessel bamboo paramo, Chingaza National Park, Cordillera Oriental, Colombia. Vegeta 115: 29-39. ·Troll, C. (ed.) 1968. Geoecology of the mountainous regions of the tropical Americ Colloquium Geographicum 9. Bonn. Turner, M.G. 1987. Spatial simulation of landscape changes in Georgia: a comparison three transition models. Landscape Ecology 1: 29-36. Turner, M.G., Gardner, R.H., Dale, V.H. & O'Neill, R.V. 1989. Predicting the spread ~ disturbance across heterogeneous landscapes. Oikos 55(1): 121-129. Usher, M.B. 1966. A matrix approach to the management of renewable resources, wi special reference to selection forests. Journal of Applied Ecology 3: 355-367. Usher, M.B. 1981. Modelling ecological succession with particular reference to Markovian models. Vegetatio 46: ll-1 8. _ Vander Hammen, T. & Cleef, A.M. 1983. Datos para Ia historia de Ia flora andina. Revista Chilena de Historia Natural 56: 97-107. Vander Hammen, T. & Cleef, A.M. 1986. Development of the high Andean pararno flora and vegetation. Pp 153-201 in: Vuilleumier, F. & Monasterio, M. (eds.), High altitudt tropical biogeography. Oxford University Press, New York. . Van Hulst, R. 1979. On the dynamics of vegetation: Markov chain as models of succession. Vegetatio 40: 3-14. Van Wijngaarden, W. 1985. Elephants-trees-grass-grdzers. Relationships between climateA soils, vegetation and large herbivores in a semi-arid savanna ecosystem (Tsavq; Ke11ya). lTC Publication 4, Enschede. 159 pp. ~ Vareschi, V. 1970. Flora de los paramos de Venezuela. Ediciones Universidad de los Andes; Merida. 429 PPVerweij, P.A. & Beekman, A.M. 1995. Elementos para el manejo del paramo colombiano en relaci6n a pastoreo, quema y cultivo de papa. Pp. 101-109 in: Rabey, M.A. (ed.), ~f uso de recursos naturales en las montaiias: tradici6n y transfom1aci6n. UNESCO/MAB, Montevideo. , Ver\veij, P.A. & Beukema, H. 1992. Aspects of human influence on upper-Andean forest li~e vegetation. Pp. 171-175 in: Balslev, H. & Luteyn, J.L. (eds.), Paramo: an Andean' ecosysiem under human influence. Academic Press; London. ·•

Verweij, P.A. & Budd.:, P.E. 1992. Burning and grazing gradients in the p:iramo of Parque Los Nevados, Colombia: initial ordination analyses. Pp. 177-195 in: Balslcv, H. & Lutcyn, J.L. (cds.J, Paramo: an Andean ecosystem under human influence. Academic Press, London. Verweij, P.A. & Kok, K. 1992. Effects of fire and grazing on Espeletia hartwegiana populations in the paramo of Parque Los Nevados, Colombia. Pp. 216-229 in: Balslev, H. & Luteyn, J.L. (eds.), Paramo: an Andean ecosystem under human influence. Academic Press, London. Walker, B.H., Ludwig, D., Holling, C.S. & Peterman, ltM. 1981. Stability of semi-arid savanna grazing systems. Journal of Ecology 69: 473-498. Weber, H. 1958. Die Paramos von Costa Rica und ihre pflanzensoziologische Verkettung mit den Hochandcn Siidarnerikas. Abhandlungen der Mathematisch-naturwissenschafdichen Klasse 3: 120-194. Akademie der Wissenschaften und der Literatur in Mainz, Wiesbaden. · Williamson, G.B., Schatz, G.E., Avlarado, A., Redhead, C.S., Starn, A.C. & Sterner, R.W. 1986. Effects of repeated fires on tropical paramo vegetation. Tropical Ecology 27: 62-69. Witte, H.J.L. 1994. Present and past vegetation and climate in the Northern Andes (Cordillera Central, Colombia): a quantitative approach. PhD dissertation, University of Amsterdam. 269 pp.

TilE PARAMO Of LOS NEVi\ DOS NATIONAL PARK

~ -!

( "lim.clit -:om!itiun.; 0<'<Jitc;;y ;md ~ ~onwrphulc><!Y S<>ils and hydrology G~ncr.t l land cover Lmd usc: conllicting imcrcsts I Iuman history and socio·cconom ic selling

', 2.3 2A 2.5 ~ .h

·' In th is chapter. tht: biophysical and socioeconomic setting is described. The present study was ~·:urkJ ou t in Los .' \cvados National Park ('Parquc Nacional Natural Los Nevados'), located .., · in th ~ Ruiz-Tolima Massif in tht! Cordillera Central of the Colombian Andes (4°35'- 4°60'N, 75"11!'-7:1.30' \V). The first systematic inventory of the region including the study area was don~ in the fonn of an elevation transecttravt:r\ing the Cordillera Central, the Parque Los \ cvado.; ·i"ranscct cTPi\. at 4'"50'1\J. The main exp~ditions took place in 1980 within the fra mework of the ECQA~DES project rSwdie.\ on Tropical Andean Ecosystems I Estudios de Ecosistcmas Tropandinos). This research programme is a joint effort of Colombian and Dutch n::search institutes. Its main objective is to study the stntcture and functioning of the c:cusystems of the Colombian Andes. Sofar, four transects traversing the three cordilleras at .t::"krent latitudes have been rcal i7Ccl ! Fi~1re 2.1). ,\ long th~ TP~ transect, multidiS\:iplinary teams of Colombian and Dutch researchers studied of climate, geology, geomorphology, palynology, pedology, plant and animal taxonomy, and vegetation ecology. About every 100 m, observations were made and meteorological stations installed. On both the east and west mountain flanks, the rransect ~:ompri~~s: humid tropical lowlands. wet montane fores ts, cool paramo grasslands, and lastly th~ p~riglacial ant.! glacial environments. The results of the TPN and of additional studies '-'-ef~ published in the series 'Studies on Tropical Andean Ecosystems' (Vander Hammen et ui .. 19X.1: 19XlJ; Van der Hammen & Dos Santos, I995a; 1995b), and give a detailed d~ s<:ription of the biophysical setting of the study area. ;~spet:ts

The inve ntory of the first phase has been indispensable for obtaining characterizations of the principal ecosystems and of the zonation of \·egetation and its long-tenn history. The present dissefiation is part of the second phase of ECOANDES, which focuses on the stu~y of ecosystem processes and human impacL The northwestern part of Los Nevados National Park, when: the full range of grazing intensities and fire regimes are represented, was selected as the study area. The following sites and their surroundings were included: Lorna Bonita, Laguna de la Leona, Baga Seca, Laguna del Otun, Valles del Condor, Mosquito, and Paramillo de Santa Rosa. They are indicated in Figure 2.2, which shows the location of the .. ' tudy area.

. -----

- - -·-· --

C-11a-rpt-"

:----------~

·• The param o of Los Nevados National Park

2.I

,T~·--------~72_·

r--- --7ra·_________7T6_·________ 12 '1- --+ --- -- -- -

Climatic conoifions

____ ____

•1 .

l!l

Elevation ranges from 3800 to 4300 m in the study area. The lapse rate of average monthly temperature is approxima tely 0.65 •ctlOO m (Thou ret, 1983; Florez, 1986b). Stable soil temperature in the study area is 7.3 to 7.5 •c on avera ge (Salomons, 1989), which is generally accepted as being equal to the mean annual tempe rature. A mean air temperature of s•c was reported by Witte (1994) for 1983. Throughout the year, temperature is more or less constant. However, differences between day and night temperatures are very pronounced (Oster, 1979). The diurn al oscillations are greatest in the drier seasons, when the soil surfac e cools down considerably during the night due to the absence of cloud cover. Frost occurs daily from 4000 m upward (Witte, 1994), and oc~:asionally at lower elevations in drier periods. The frost regime sets an upper limit to the cultivation of potatoes and other tubers.

The area is subject to the influences of the intertr opical convergence zone (ITCZ). The ..northeastern trade winds have a continental chara cter, whereas the southwestern air masses come from the Pacific Ocean and are very humid . As the impact of the latter is expected to be more pronounced on the west flank, this side of the cordillera was identified as more humid than the east side (Perez, 1983). This theory is supported by data on vegetation zonation on both flanks (Kloostennan eta/. 1995; Salamanca eta/. , 1995), but is contradicted by Witte's climatological model (1994). In any case, the inter-Andean valley surrounding Laguna del Otun is climatically drier than the exterior slopes due to the interception of precipitation by the old volcallo Param illo de Santa Rosa, west of the lake (Kuhry et a/., 1983; Witte, 1994). Mean annual precipitation is approximately 1300 mm. The shifting position of the ITCZ detennines seasonal rainfall patterns. When the ITCZ passes over the area, a period of higher precipitation and cloud iness prevails, while a more distant position results in a drier season. · Thus, precipitation is characterized by a bimod al distribution, like in other tropical regions close to the equator. The relatively more humid seasons are from March to May and from June to August. The seasonal bimodality appears to be weake r above 3500 m (Witte, 1995). The available rainfall, calculated as the sum of precipitatio n after subtracting potential evapotranspiration using Thornthwaite's formula (1948), is about 500 mm. Acco rding to Witte (1994), approximately 35% of availa ble precip itation falls in the dry seasons. With the . absence of distinct colder and warmer periods, this leads to the conclusion that seasonality is not pronounced in the study area. Figure 2.3 shows a climatic diagram for the meteorological station of Otun (4000 m), based on data for the year 1983. Rainfall had a weak bimodal distri bution, with peaks in April and Dece mber. In the early morning, the paramo is generally cloud -free. Durin g the day, the clouds ascend from the montane forests to the paramo belt, frequ ently giving rise to drizzle and rain in the afternoon. Hardly any quantitative data on cloud iness and drizzle are available. Vegetation .. physiognomy indicates, however, that interception of atmospheric moisture represents an :~important factor (Cleef, 1981). Figure 2.1

Locati on of the transects of the ECOANDES researc h programme.

·~

.·

··--

--- -- ~-----

The pc1ramu uf Los Nevados Nmional Park

T oC

Prec. mm. -500

T-: 5

-400

T•: 6

-300 -200

P- : 4

P•: 146 Py: 929

-1 00

- eo - 60

20 -

- 40

1,0 -

0 -1.0-

- 20-~~~~-7~ ~~~~.

Jrmomjjeson Laguna del Otun 4000

Figure 2.3

2.2

Walter diagram for the meteorological statio n of Onin at 4000 m, based on data of 1983 (Witte, 1994).

Geology and geomorphology

In the high Andean moun tains, glacial and volcanic phenomena are associated . Both volcanism and Quaternary glaciations gave rise to conspicuous geomorphic features. A number of high volcanoes were formed durin g the Plio-Pleistocene on top of a metamorp hic basement of Palaeozoic I Mesozoic age. The Quaternary glaciations left their trace s in th~ form of lateral moraines, end moiaines, U-shaped valleys, glacial depressions, and cirques. Glacial features are present throughout the paramo, although less distinct near the fores t line. The area was shaped particularly by the last glaciation, 'Otun tardio', between 11,00 0 and 10,000 BP (Thouret & Vander Hammen, 1983 ).

E 0 0

0 0

"'., "'..,.

E 0 0

"' "'..,. ., "' v

0 0 N

~. I

0 0

:!!

"'>-c ::>

(ij

D0 •

....0

0. '

e"' c: .g,. "'0 '' ..J <

Extrusive volca nic processes resulted in the formation of lava fields, block lava, and pyroclastic flows. According to Herd (198 2), pyroclastic eruptions in the Rufz-Toli ma area appear to have occurred once every 250 to 500 years on average. Between 7400 and 6000 BP, a volcanic eruption produced the mixed flows of lava-lahar or the block lava flows of La Leona and Totarito (Thouret & Van der Hammen, 1983). Another violent eruption occurred around 5400 BP. One of the resulting lava flows, from the direction of Nevado de Santa Isabe l, obstructed the river Rfo Otun as a natural dam, thus forming a lake, Laguna del Otun. Layers of ash, lapilli, and other pyroclastic material produced by frequent explosive erup tions have covered the area with a thick sequence (2 m on average) of tephra and buried soils. Eruptions in ± 7500, 6100, and 3600 to 3000 BP resulted in ftres that destroyed extensive parts of the vegetation cover. The last erup tion, of Cerro Bravo, deposited a lapilli layer that is observed in soil profiles above 3600 m.

Chapter 2

The volcano Nevado del Ruiz (5400 m) is still active. The dramatic eruption in 1985 in mud flows that caused the death of more than 25,000 people of the villages A Chinchina (Thouret et al., 1989).

The other volcanoes are Cerro Bravo (4000 m), El Cisne (4600 m), Nevado de Santa (5100 m), Paramillo de Santa Rosa (4600 m), El Quindio (4700 m), and Nevada del (5200 m). Above 4700 m, an icecap is present nowadays. A terrain map was prepared using aerial photo interpretation and the existing u11 ,. ,.u~••• map (Villota, 1984). The photo interpretation features were directly digi georeferencing was done in a GIS. Detailed procedures are described in Chapter 8, 8.2. The terrain map is pro>entod in .figure 2.4 on page 157. Compared with physiographic map (I.e.), the present map includes more detail regarding different classes, and the distinction of glacial depressions and valleys. These units were because they affect the distribution of cattle and fire patterns.

3km

Using a digital terrain model or DTM (Chapter 8, Section 8.2), a slope gradient calculated according to standard procedures in JLWIS (ILWIS Manual, 1993). Pixel correspond to the average slope percentage of that site. The slope gradient map was according to slope steepness classes given by Van Zuidam (1985). The resulting slope map is presented in Figure 2.5 at page 160. The major terrain units such as block lava, glaciated lava fields, and lateral moraines recognized on the orthophoto of Figure 2.6, which was superimposed on the DTM. spots correspond with burned areas.

2.3

Soils and hydrology

The tephra layers deposited by volcanic eruptions form a rather uniform parent material pedogenesis. The volcanic ash products led to the development of andosols (Andepts, characterized by a strong accumulation of humus and a quick insolubilization of metallic complexes (Duchaufour, 1976; Thouret, 1989). They have a low supply of low pH, a low available P content, and are rich in organic material. Characteristic of andosols is the high affinity of active AI and Fe components with phosphate (Mizota & •· ·, . .·'-';/: ·. Reeuwijk, 1989). The cool and wet climate favours the fonnatfon of amorphous "''""""'....' . '---~-~· . .. ·,...,. mainly allophane. The allophane not only fixes phosphate, but also fonns aggregates with . ...... organic matter, thus protecting it from decomposition. This explains the strong accumulation. Andosols have a high water retention capacity. The aggregate stability is and in combination with a high porosity and high infiltration rates, this soil type is corlstderC(IID one of the most resistant to erosion. Drying of andosols may result in an irrl"vl"r·~ihieR crystallization of the allophanes (!meson & Vis, 1983). It is not cenain whether ,Figure 2.6 ORTHOPHOTO of 1959, including the volcano El Quindio, which was desiccation can be produced by paramo fires, for which temperatures of up to 420 •c superimposed on the digital terrain model. Geometric correction of the scanned aerial been recorded at ground level (Ramsey, 1993). Photograph (on top) was done in a GIS. The black patches correspond with burned areas.

·.J~;~~~~} r

.. 'f

The paramo of Los Nevados National Park

Chapter 2 Most paramo soils are inceptisols, only slightly developed and relative ly young. Bcs elevation gradient that detennines the distribution of soils at macro-l evel, there are some factors that influence soil genesis locally. Most important is humidit y. Slope stee plays an important role: the more differentiated soils are associat ed with gentle g~neral , the position of the soils is related to variation in the hydrological regime toposequence (Thouret, 1989). On moderate slopes, the soils are well drained. umwn~.~,~m. the toposequence, the soils coverin g gentle slopes or glacial depress ions are poorly to hydromorphic. In the paramo proper, Cryandepts (andosols) are the zonal soil Histosols (hydromorphic soils) characterize the very wet places, e.g. with cushion Finally, aspect is important at least on a small scale. The contrast between the atmiOSJ>heJricaiiU more humid west side of the cordillera and the drier east side is reflected in the hydromorphic characteristics of the soil profiles on the west flank. It is uncertain aspect also causes larger-scale variation. Thouret (1989) described the zonal soils of the paramo proper as young andosols, vitric and desaturated, classified as lithic and dystric Cryandepts. The same_author the following characteristics of the study area: these soils are rather deep (70 to 90 em) show a profile with Ao, Ah, and C horizon s. The organic Ah horizon is thick, with the type of humus; and goes down to a depth of about 55 em. Down to 65 to 70 em, a lapilli mixed with ashes is present, with an approximate age of only 3000 to 3600 (Salomons, 1989). This layer may act as a barrier to water infiltrat ion as a continuous · pan is frequently found on top of it. Downwards, buried soils and tephra layers are The upper watersheds of some important rivers are located in Los Nevados National The study area is drained by Rf.Q Otun, an affluent of Rfo Cauca.

2.4

General land cover

Compared with that of the Cordillera Oriental, the paramo flora of the Cordillera Central relatively poor in species . According to Van der Hammen & Cleef (1 986), this may explained by the (Piio-)Pieistocene volcanic origin of the high mounta in region, conreSJ>Onlltng,J to a younge r age of the high Andean biota in comparison to the Cordillera Oriental. major uplift phase of the Cordillera Oriental occurred around 12 to 9 Ma ago (Vander 1989), while the uplift of the Cordillera Central took place between 10 and 4 Ma (Kroonenberg eta/., 1990). The most recent lavas of the Manizales Formation were dated at 3 to 4 Ma (F16rez, I Thus, species might have become extinct as a result of volcanic activity. development of the paramo of El Rufz after the volcanic eruption of 1985 was monitored documented by Salamanca (1991 ). Zonal paramo vegetation of the Cordill em Central was described by Cuatrecasas (1934, I 1968), Cleef (1981), Cleef eta/. (1983), Salomons (1989), Salama nca (19Y I), and "~l~m~n.r.a eta/. (1995). A landscape-guided vegetation survey of a large part of Los Nevados Park was carried out in 1980/1981 (Kloosterman eta/. , 1995).

In Los Ncvados National Park, tht: following elt:vatio n zones (from low to high) were A11dcan forest, high Anuc;m lim:~t. pararno proper, supcrparamo, and the nival t;.:!t (Cicd' eta/., 1983). Fluctuations in pulaco-climatc caused the repeated up- and downward ,hift of the vegetation zones. Nowadays, the uppt:r elevation limit of the Andean forest is located between 3750 and 3800 m. recogn i z~d:

The study area is located mainly in the elevation belt of the paramo proper, also called the ·grass pararno' or 'pa!amo propiamcnte dicho', sensu Cuatrecasas (1958). Four principal \'egetation structure types are present in the paramo proper of Los Nevados National Park: bunch grasslands cushion bogs 'dwarf forests' or shrublands · short matted grasslands or forblands .. These are described briefly below. nrmch grasslands The zonal bunch-grass vegetation is dominated by tussocks of Calama groscis recta, C. effusa, .tnd Festuca species, accompanied by stem rosettes of Espelet ia hartwegiana ssp. ccncroandina. The zonation within the parJmo proper is determined by. the presence of the ..:iffcrcnt tussock species, occunring either almost excl usively, or in combinations. The different communities occunring in the study an:a were described by Salamanca eta/. (1995).

Two important associations dominate the paramo proper. One is the Ca/andrino acaulis Calamagrostiewm reccae. This means C. recrae is the dominant bupch grass, whereas stem rosettes may or may not occur. This bunch-gr.tss vegetat ion occurs in the upper part of the paramo proper; it extends from 4150 to 4400 m. The other association is the Calamagrostierum effuso - rectae, with an upper elevation limit of about 4200 m. Diagnostic i' the presence of the combination of Espe/etia harrwe~:iana, Catamagrostis effusa, and Cilslilleja jissifolia. In addition , the Espeletiu hartwegianae - Calama grostietum effusae is a common association on the external humid slopes of the Rufz Tolima Massif. Diagnostic . is 1he absence of Cerastium subspicamm, Luzrrla racemosa, and Senecio formosus. Tussocks of Feswca are apparen tly related to more humid places. They are also frequen tly prominent in cushion bogs.

Cushion bogs · On valley floors, in glacial depressions, and in concave parts of slopes, cushion bogs occur. This vegetation is well-known in the high tropical Andes, general ly in places with a high water table. The dominant species is either Plantago rigida or Distich(a muscoides. They probably play a major role in the regional hydrological system (Cleef, 1978, 1981; Bosman ec a/., 1993). . Dwarf forests a11d shrublands This structure type includes the so-called dwarf forests of Aciach ne acicularis - Esca/lonia myni/loides. The Escallonia trees which usually grow in the Andean forest extend into the · paramo and show a stunted growth. 17

..... The paramo of Los Nevados Nati11na/ P<ITk Chap£er 2 In fact, their growth form rdther resembles that of a shmb. They are identified as low forests, which is confirmed by palynological evidence: In addition , there are shrublands that belong to the Baccharido tricuneatae - Hypericetum Jaricifolii.

Sllort matted grasslands or forblands Three replacement COII)mu nities urtder intensive grazing have been recognized: Aciachnetum acicularis Vareschi 1953, a xerophy tic association of dry plains, distributed throughout the northern Andean paramo; secondly, the Agrostio Lachemilletum orbiculatae as described by Cleef (1981 ), which occurs on slightly humid grounds. Agrostis haenkeana is another important short grass species. u~<•l'.uu>IIIC Taraxacum, Poa annua, and Trifolium cf. amabile. Finally, Salamanca (1991) and Sa et a/. (1995) found another community on the borders of Laguna del Otu Muhlenbergietum cleefii. These are the most humid meadows with the shonest plant

· Examples include the endangered spc~:tacled bear ('oso de anteojo s'J Tremarctos omatus, the '."- tapir Tapims pinchaq11e, tht: feline species Felis concolor and Felis tigrina pardinllidts, the dwarf squirrel Microsciums pucherani salentis, the eagle (';iguila re<~l') Oroacws isiduri, the ~ndemic humming bird ·O:qpogon querinii ~1uebel i and various duck spccies Anus spp.. Of :\·the plant species, the endemic Senecio isabelis an_d the national tree ('palma de cera') · -Cero:cyloll quindiuense are worth mentioning. Populations of the last species have been included by th~ last extension of the park boundary.

.···~·~ince its establishment, the park has been managed by INDERENA. The law of 1959 prohibit

s all agricultuml activities, hunting, and fishing. Access to the area has been limited to one unpaved road. For this reason, tourist numbers have been relative ly low. To11rism has been officially restricted since the eruption of the Rufz volcano in 1985. Fishing of trout' .., ·•) introduced into Laguna del Otun is allowed, providing one of the main recreational activities. · _··. The use of the nature reserve for research and education has been limited compared with • ': ·'10 \ver elevations. As is the case in other national parks in Colombia, official permission is • :·;' required for ihe collection of plant and animal material. ., ~.

2.5

Land use: conflicting interests

Under Article 13 of Law 2' of 17 January 1959, the Colombian 'nevados' or moumains and their surroundings were declared national parks. The institution in implementing this law was INDERENA, the national institute for th~ management of resources and the environment. In March 1973, Los Nevados Nauona l Park was apparently with a total area of 38,000 ha. Canographic revision later revealed that the defined by the park boundaries was in reality 58,300 ha (INDER ENA, 1985). afterwards, the nationa l park was extended to approximately 900 2 km • It falls under jurisdiction of 10 different municipalities, corresponding to four differen t depanments: Quind{o, Risaralda, and Tolima. With the creation of the national park in 1973, the following objectiv es were defined for area: (1) (2)

(3) (4)

to preserve a representative sample of the flora and fauna of the region of the Tolima volcanic massif; to protect endangered species of flora and fauna of the superpa ramo, paramo, upper-Andean forest; · to- provide opportunities for research , education, and recreation that are with the conservation of resources and cause no deterioration; and to protect the hydrological systems of the upper watersheds of impona nt rivers in depanments of Caldas, Risaralda, Quindfo, and Tolima.

These goals largely coincide with the potential of the paramo for sustainable del•elc>prTielll mentioned in Chapter 1. However, agricultural activities are clearly excluded. Despite a lack of information on population sizes, it is certain that important animal have their habitat in the national park.

<. Laws and regulations seem to have overlooked or ignored the fact that when the national park was created, the area was already inhabited by farmers. The main agricultural practices are ·extensive livestock production and, up to ± 3900 m, the cultivat ion of potatoes and other .~crops. Extensive grazing in the bunch-grass p~amo is often pmctise d in combination with the use of tire to stimulate the regrowth of fresh shoots of higher forage quality. Figure 2.7 shows the effect of ftre on the general aspect of the bunch grass vegetati on. Fuelwood collection, mainly for cooking purposes, represents another important land use (Verweij & Beukema, 1992; Kok eta/., 1995). An average family of seven persons uses about 20 kg of fuel wood a day, corresponding to a cleared area of 1 ha a year. .In 1985, 29% of the national park was private propeny, 11% was owned by government institutions, e.g. by !NOERENA, and the remaining 60% was state property (Alvarado et al., 1985). The national park management aims to buy the agricultural land. It is seen as the best way to restore the natural state of the ecosystems accordi ng to the conservation objectives mentioned above. The main preoccupation is that the paramo system under agricultural practices may lose part of its water regulating function. For cities such as Pereira and Manizales, this function of the upper watersheds is important in relation to the periodic occurrence of droughts and floods, and for the provision of drinking water. However, at the time this investigation began, knowledge of the grazing and burning regime and its effects on vegetation dynamics was scarce. Park management and people involved in the management of other pararno areas experienced this as a serious problem. Recently, a research project was finished that provided information on the effects of ~urning and grazing on ecological processes, including aspects of biomass, nutrient status of vegetation and soil, and hydrology (Rossenaar & Hofstede, 1992; Hofstede & Witte, 1993; Hofstede, 1995; Hofstede era/., 1995; Hofstede & Rossenaar, 1995). Another research project, of which the final results are presented in this dissertation, focused on the impacts of management on vegetation development and landscape ecology (Schmidt & Verweij , 1992; Pels & Verweij, 1992; Verweij & Beukema, 1992; Verweij & Budde, 1992; Verweij & Kok, 1992; Koker a/., 1995; Verweij & Beekman, 1995; Verweij et al., 1995). It is the first time that the management regime has been analyzed quantitatively. 39

The paramo of Los Nevados Na1ional Park Another probkm faced by the park management is coordinating acti vi ti~s with the relevant institutions of four difkrcnt dcpanmcnts. The actual development of tourism, for instance. is promoted at departmental level and is accompanied by the con~truction of roads that often kad up to the park boundaries. This process will defini tely have an impact on the national park and conflicts with the objective of conservation. At the same time, the regional management corpo rations of the .four departments tCorpo raciones Regionales) a're increa singly interested in an effective protection of the upper waters heds in order to avoid the adverse effects of land use on the regional hydrological svstem and on the related supply of drinking water. It has been stated several times that the i~tcres ts of a relatively small group of individual fanner s should not outweigh the intere sts . of the numerous inhabitants of urban areas at the foot of the mountains. Still, it cannot be denied that the paramo farmers have certain land use or property rights, as shown ~Y the history of land occupance described below.

2.6

Human history and socioeconomic setting

It is difficultÂˇ to trace when farmers first settled in the arc~ now constituting Los Nevados \ational Park. The paramo was most probably not inhabited before the Spanish Conquest. unl ike the puna and jalca (Ellenberg, 1979; Schjel lerup, 1992). Around 2000 BP, some remarkable changes can be seen in the palynological diagrams of El Bosque, which is located dose to the natural tree line (Kuhry er at., 1983). In particular the pollen curve of the tree Polylepis shows a strong and sudden decrease. This might be due to a volcanic catastrophe (although no other evidence has been found) or to wood extraction by indigenous tribes in this area.

Historic documents mention that the surrounding area was inhabited by people of different indigenous cultures. The Quimbayas lived in the region of Quindfo, the Picaraes in NE Caldas, and Tolima was the territory of both the P.anch es and the Pijaos. They conducted religious ceremonies in the paramo, around glacial lakes that were considered sacred places. A few old ,trails date from times before the Colony and served as trade routes connecting different departments. After prolonged battles, most of the fertile lowlands of the inter-Andean valleys were confiscated by the Spaniards. The indige nous people were either killed or forced to move upwards into the Andean and upper Andea n zones. At the beginning of the 19th century, a reserve had been demarcated up to the Âˇparamo of Quindfo to compensate the indigenous people for the 'loss' of their land. Aroun d 1824, a Spanish family rented the paramos of Rufz and Santa Isabel, along with aÂˇherd of abandoned cattle, originally from the lowlands of the Magdalena valley (Fabo, 1926). Anoth er repon on the presence of freeroaming cattle stems from a scientific expedition in 1843 (Londono, !936). At this time, the paramo was apparently still uninhabited. In the upper watershed of Rio Otun, a group of appro ximately 25 farming families live in the vereda of El Bosque. Most of these families live on land owned by one of the first colonists, who settled there around 1920. ~~ is estimated that most permanent agricultural practices in the study area have a history of 60 to 80 years. 41

The paramo of Los Nevados National Park Chapter 2

During the civil war 'La Violencia' (1948-1958), many of the present farmers came <:rea as refugees. According to a census carried out by !NDERENA in 1985, onr.rmrim ••i 7_35 persons lived in the area of Los Nevados National Park at that time. Since then, population has probably remained stable. Agriculture has always been ma,rginaL Nowadays, a caretaker system prev~ils. colonists moved out of the area as soon as a small capital allowed them to settle in the Andean valleys. Their farms or 'fincas' are administrated by caretakers who live there their families. They are the real farmers who are responsible for day-to-day ma11ag,en'ie Incomes are generally below the national minimum level. It sometimes consists of a s~ary supplemented by the sale of cheese. The incomes of other fanners are exclusively based on cheese production and, occasionally, on the sale of their own However, most of the herd belong to the tinea owners. The construction of new fincas the extension of agricultural land are prohibited. There are no health-care facilities; one primary school for the children living in the vereda of El Bosque. Farmers ..,· ua'""''); national park do not have access to credit facilities. This is one of the measures m"'•n•''~~• discourage the 'colonists' froma permanent stay. Even in these marginal conditions, farming families said they intended to stay, for the simple reason that no better '""'•''',.. are available.

LITERATURE CITED

Alvarado, C.A., Cardozo, J.E., Cuestas, E. & Rojas, S.M. 1985. Estudio y an~lisis predial. 27-46 in: Plan de Manejo Parque Nacional Naiural Los Nevados. 11'1-'r:.•=~··~~·• Ministerio de Agricultura, Bogota. Bosman, A.F., Vander Molen, P.C., Young, R. & Cleef, A.M. 1993. Vegetation ecology a paramo cushion mire. Journal of Vegetation Science 4: 633-640. Cieef, A.M. 1978. Characteristics of neotropical par_amo vegetation and its subant<lfCU~ relations. Pp. 365-397 in: Troll, C. & Lauer, W. (eds.), Geoecological .........vu••~ between the southern temperate zone and the tropical Erdwissenschaftlichen Forschung 11. Franz Steiner Verlag, Wiesbaden. Cleef, A.M. 1981. The vegetation of the paramos of the Colombian Cordillera Dissertationes Botanicae 61. J. Cramer, Vaduz. 320 pp. Cleef, A.M., Rangel, J.O. & Salamanca, S. 1983. Reconocimientode Ia vegetaci6n en Ia alta del Transecto Parque Los Nevados. Pp. 150-173 in: Vander Hammen, T., A. & Pinto, P. (eds.), Studies on Tropical Andean Ecosystems I. 1. Cramer, Cuatrecasas, J. 1934. Observaciones geobotanicas en Colombia. Trabajos del Museo Na,;wn.at.. de Ciencias Naturales Series Botanicas 27: 1-44. Madrid. Cuatrecasas, J. 1958. Aspectos de Ia vegetaci6n natural de Colombia. Revista de Ia Ac;ademJ!,• Co!ombiana de Ciencias Exactas, Ffsicas y Naturales 10: 40. Cuatrecasas, J. 1968. Paramo vegetation and its life forms. In: Geo-ecology mountainous regions of the tropical Americas. Colloquium Gcographicum 9: 163-1 Duchaufour, Ph. 1976. Atlas ecologique des sols du monde. Masson, Paris, 178 pp. Ellenberg, H. 1979. Man's influence on tropical mountain ecosystems in South Am,,.nc:a-. Journal of Ecology 67: 401-416.

!'abo. M. d~:. I926. Historia de Ia ciudad de Manizales. Torno I, p. 41. fltirl'l . .-\. I 'JX(Ja. Gcomorfologia del <irca :\lanizaks-Chinchina, Cordilkra Central, Colombia. - PhD dis~wation, University of Amsterdam. 15X pp. fior.:z, A. !Y86b. Relaci6n altitudinal de Ia temperatura del suelo y del aire en los Andes centrales de Colombia. Colombia Geognifica 12(5): 5-38. J-krd, D.G. 1982. Glacial and volcanic geology of the Ruiz - Tolima volcanic complex, Cordillera Central, Colombia. Publicaciones Geol6gicas Especiales de1 1ngeominas 8: 1-48. Hofstede, R.G.M. 1995. Effects of burning and grazing on a Colombian paramo ecosystem. PhD dissertation, University of Amsterdam, 199 pp. Hofstede, R.G.M., Mondragon, M.X. & Rocha, C.M. 1995. Biomass of grazed, burned, and undisturbed paramo grasslands, Colombia. I. Above ground vegetation. Arctic and Alpine Research 27: 1-12. . Hofstede, R.G.M. & Rossenaar, A.J.G.A. 1995. Biomass of grazed, burned, and undisturbed p;iramo grasslands, Colombia. II. Root mass and above ground:below ground ratio. Arctic and Alpine Research 27: 13-18. l lof\ tede, R.G.M. & Witte, H.J.L. 1993. An evaluation of the use of the dry-weight-rank and the comparative yield biomass estimation methods in paramo ecosystem research. Caldasia 17(2): 11-14. II.WIS 1.4 User's Manual 1993. lTC, Enschede. ln~cson. A.C. & Vis, M. 1983. Los procesos de erosion bajo bosque en suelos de cenizas volcanicas. Pp. 88-112 in: Van tkr Hammen, J.. Perez, A & Pinto, P. (eds.), Studies on Tropical Andean Ecosystems I. J. Cramer, Vaduz. l\'DERENA. 1985. Plan de Manejo Parque Nacional Natural Los Nevados. Ministerio de Agricultura, Bogota. Kloosterman, E.H., Cleef, A.M. & SalamiUJCa, S. )995. Vegetation map of the Parque Nacional Naturdl Los Nevados (Central Cordillera, Colombia). In: Vander Hammen, T. & Dos Santos, A.G. (eds.), Studies on Tropical Andean Ecosystems 5. J. Cramer, Berlin. Kok, K., Vc!rweij, P.A. & Beukema, H. 1995. Effects of cutting and grazing on Andean forest line vegetation. In: Churchill, S.P., B alsl~v. H., Forero, E. & Luteyn, J.L. (eds:), Biodiversity and Conservation of l'ieotropical Montane Forests. Scientific Pub!., The New York Botanical Garden. Kroonenberg, S.B., Bakker, J.G.M. & Van der Wiel, A.M. 1990. Late Cenozoic uplift and paleogeography of the Colombian Andes: constraints on the development of highAndean biota. Geologie en Mijnbouw 69: 279-290. Kuhry, P., Salomons, J.B., Riezebos, P.A. & Vander Hammen, T. 1983. Paleoecologia de los ultimos 6000 aiios en el area de !a laguna de Otun-EI Bosque. Pp. 227-261 in: Van der Hammen, T., Perez, A. & Pinto, P. (eds.), Studies on Tropical Andean Ecosystems !. J. Cramer, Vaduz. Londono, L. 1936. Manizales 1936. Contribuci6n al estudio de su historia hasta el septuagesimo quinto aniversario de su fundaci6n. p. 132. :Vlizota, C. & Van Reeuwijk, L.P. 1989. Clay mineralogy and chemistry of soils fom1ed in volcanic material in diverse climatic regions. Soil Monograph 2, ISRIC, Wageningen. Oster, R. 1979. Las precipitaciones en Colombia. Colombia Geografica 6(2): 8-147. Pels, B. & Verweij, P.A. 1992. Burning and grazing in a Colombian bunchgrass paramoecosystem: A transition model. Pp. 243-263 in: Balslev, H. & Luteyn, J.L. (eds.), Paramo: an Andean ecosystem under human influence. Academic Press, London. 43

The parumn of Los Nevados National Park

Chapter 2

J.:r Hammen, T. & Dos Santos, A.G. (cds.) 1995b. Studies on Tropical Andean Ecosystems 5. 1. Cramer, fkrlin. \':1n do;; r Hammen, T., P~rcz, A. & Pimo, P. (t:Js.) 1983. La Cordillera Central Colombiana, Transct:to Parque Los Nevados (inuoducci6n y datos iniciales). Studies on Tropical Andean Ecosystems I. J. Cramer, Vaduz. Van der Wiel, A.M. 1989. Uplift of the Prcca mbri:m Gar..:6n Massif (Eastern Cordillera of the Colombian Andes) in relation to fluvia l and volcaniclastic sediments in the adjacent Neiva Basin. Abstracts 28th Internationa l Geological Congress Washington 3: 505. Van Zuidam, R.A. (cd.) 1985. Aeria l photo -interpretation in terrain analys is and geomorphologic mapping. Smits Pub!., The Hagu e. 442 pp. Vaweij, P.A. & Beekman, A.M. 1995. Elementos para el manejo del pararno colombiano en relaci6n a pastoreo, quema y ~ultivo de papa. Pp. 101-109 in: Rabey, M.A. (ed.}, El uso de los recursos natura les en las monta iias: tradicion y transformaci6n. UNESCO/MAB, Montevideo. . Vawcij, P.A. & Beukema, H. 1992. Aspects of human influence on upper-Andean forest line vegetation. Pp. 171-175 in: Balslev, H. & Lutey n, J.L. (eds.), Paramo: an ~ndean _ecosystem under human influence. Academic Press, London. Vaweij, P.A. & Budde, P.E. 1992. Burni ng and grazing gradients in the paramo of Parque Los Nevados, Colombia: initial ordination analy ses. Pp. 177-195 in: Balslev, H. & Luteyn, J.L. (cds.}, Paramo: an Andean ecosystem under human influence. Academic Press, London. Vaweij, P.A. & Kok, K. 1992. Effects of fire and grazing on Espeletia harnvegiana populations in the paramo of Parque Los Nevad os, Colombia. Pp. 216-229 in: Balslev, H. & Luteyn, J.L. (eds.), Paramo: an Andean ecosystem under human influence. Academic Press, London. Verweij, P.A., Kok, K. & Budde, P.E. 1995. Aspec tos de Ia transfonnaci6n del paramo por el hombre. Vander Hammen, T. & DosSantos , J. (eds.), Studies on Tropical Andean Ecosystems 5. 1. Cramer, Berlin. Villota, H. 1984. Canograffa de Ia fisiografia y erosi6n de las cuencas de los Rfos Otun y Consota - Barbas, Departamento de Risaralda. CIAF, Bogota. 136 pp. ' Witte, H.J.L. 1994. Present and past vegetation and climate in the Northern Andes (Cordillera Central, Colombia): a quantitative approach. PhD dissertation, University of Amsterdam. 269 pp. Witte, H.J.L. 1995. Seasonal and altitudinal distribution of precipitation, tempe rature and humidity in the Parque los Nevados Transect (Cord illera Central, Colombia). In: Van der Hammen, T. & DosSantos, A.G. (eds.), Studi es on Tropical Andean Ecosystems 4. J. Cramer, Berlin. \';111

Perez, A. 1983. Algunos aspectos del clima. Pp. 38-47 in: Vander Hammen, T., Perez, & Pinto, P. (eds.); Studies on Tropical Ande an Ecosystems I. J. Cramer, Vaduz. ~ Ramsey, P.M. 1993. The paramo vegetation of Ecuador: the community ecology, dynam~ ~.~~ and productivity of tropical grasslands in the Andes. PhD dissertation, University & it Wales. 274 pp. · Rossenaar, A.J.G.A. & Hofstede, R.G.M. 1992. · Effects of burnin g and grazing on ~ ~~ ~. biomass in the paramo ecosystem. Pp. 211-213 in: Balslev, H. & Luteyn, J.L. (eds.X • · Paramo: an Andean ecosystem under human infl~e nce . Ac~de~ic Press, Lo.ndon. · Salamanca, S. 1991: The vege.tation of the param o an.d I!S dyna~cs m~he volc~mc ~assit 1 Ruiz - Tohma (Cordillera Central, Colombia). PhD dissenauon, Umverstty ~· .: Amsterdam. 122 pp. Salamanca, S., Cleef, A.M. & Rangel, J.O. 1995. The paramo vegetation of the Ruiz- Toli .. massif. Jn: Van der Hammen, T. & Dos Santo s, A. G. (eds.), Studies on Tropic . Andean Ecosystems 5. J. Cramer, Berlin. Salomons, J.B. 1989. Paleoecology of volc~ nic soils in the Colombian Central Cordillera..Pp~ 15-215 in: Vander Hammen, T., Diaz, S. & Alvar ez, V.J. (eds.), StudieS on Troptcal Andean Ecosystems 3. J. Cramer, Berlin. " Schjellerup, I. 1992. Pre-Columbian field system s and vegetation in the jalca of northeastern! Peru. Pp. 137-150 in: Balslev, H. & Luteyn, J.L. (eds.), Paramo: an Andean ecosystem, under human influence. Academic Press, Londo n. · Schmidt, A.M. & Vcrweij, P.A. 1992. Estimates of forage intake and secondary production" . in extensive livestock systems in the paramo of Parque Los Nevado~. Colombia. Pp.~ . 197-210 in: Balslev, H. & Luteyn, J.L. (eds.) , Paramo: an Andean ecosystem under human influence. Academic Press, London. Thomthwaite. C.W. 1948. An approach towar d a rational classification of climate. The · Geographical Review 38: 55-94. New York. Thouret, J.C. 1983. La temperatura de los suelos : temperatura estabilizada en profundidad y ' correlaciones tennicas y pluviometricas. Pp. 142-1 49 in: Vander Hammen, T., Perez, A. & Pinto, P. (eds.), Studies on Tropical Ande an Ecosystems I. J. Cramer, Vaduz. ~. Thouret, J.C. 1989. Suelos de Ia Cordillera Centr al, Transecto Parque Los Nevados. Pp. 293- , 441 in: Van der Hammen, T., Diaz, S. & Alvar ez, V.J. (eds.), Studies on Tropical·. Andean Ecosystems 3. 1. Cramer, Berlin. Thouret, J.C. & Van der Hammen, T. 1983. La secuencia holocenica y tardiglacial en el Parque Los Nevados. Pp. 262-276 in: Van der Hammen, T., Perez, A. & Pinto, P. (eds.), Studies on Tropical Andean Ecosystems l. J. Cramer, Vaduz. Thou ret, J.C. , Gourgaud, A., Vatin-Perignon, N. & Calvache, M.L. 1989. The eruption of the Nevado del Rufz on the 13th of November 1985. Pp. 217-256 in: Vander Hammen, T., Diaz. S. & Alvarez, V.J. (eds.), Studies on Tropical Andean Ecosystems 3. J. Cramer, Berlin. Vander Hammen, T. & Cleef, A.M. 1986. Deve lopment of the high Andean paramo flora and vegetation. Pp. 153-201 in: Vuilleumier, F. & Monasterio, M. (eds.), High altitude tropical biogcogmphy. Oxford University Press, Oxford. Van dcr Hammen, T., Diaz, S. & Alvarez, V.J. (eds.) 1989. La Cordillera Central Colombiana, Transecto Parque Los Nevados (Segunda parte). Studies on Tropical Andean Ecosystems 3. 1. Cramer, Berlin. Van der Hammen, T. & Dos Santos, A.G. (eds.) 1995a. Studies on Tropical Andean Ecosystems 4. J. Cramer, Berlin.

f ·:

·l

·I

45 44

.........

CHARACTERIZATION OF VEGETATION PATTERNS

3.1 3.2 3J 3.-t 3.5 3. 6

Methodology Definition of vegetation structure types Definition of noristic types Ordination analyses Vegetation map Conclusions

The objective _of the present chapter is to characterize the vegetation patterns occurring in the st~tdy area in relation to environmental and management factors. The spatial and temporal dimensions of these patterns are only briefly treated here; they are dealt with in more !ietail in Chapters 6 and 7. The analysis focuses on those vegerarions of !he grass paramo that are pote_ntially or actually subjecl to degradation. Classification, ordinarion and mapping t~chniques are used to portray !he vegetarion patterns. The relarionshi p berween vegelation patterns, burning, and grazing management is quanrified.

· · 3. 1

Methodology

3.1.1 'Retrospective monitoring' Sire selecrion was carried out with a dual purpose in mind: to cover both natural and humaninfluenced variation of !he vegetation. A stratified random sampling approach was followed. The vegelation map of Los Nevados National Park at scale 1: 50 000 by Kloosterman eta/. (1995) provides an excellent strarificarion of the study area according to the natural variation of vegetarion attributes. A series of aerial photograp hs of different years, dating from 1955 ro 1989, was iised 10 select sample sites represenrarive of cenain changes in vegetation strucrure. These changes in structure were expected to be related to management variables. The study area is covered by aerial photographs of the years 1955, 1959, 1966, 1970, 1975, 1978, 1983, 1985, 1987, and 1989. The scales range from 1: 11 000 to 1: 33 000. For many sites, burning history could be reconstructed by identifying clear black spots on the photographs. Three to five years after a fire had occurred, these spots were still visible but had then turned dark grey, contrasting only slightly with the grey-toned natural vegetation . Another change was recognized, representing the development of light-toned shon grass ·communities, and was found to be related to the influence of grazing. Chapter 8 elaborates on the reconstruction of fire history, using a GIS in combination wilh the same time series of aerial photographs.

Pan of this chapter has been published as: Verweij, P.A. & Budde, P.E. 1992. Burning and grazing gradients in the p~ramo of Parque Los Nevados, Colombia: initial ordination analyses. P~ramo: an Andean ecosystem under human influence. © Academic Press, London.

,·

Characterization of vegetation patterns Chapter 3 3.1.2 Data collection The fieldwork that constitutes the basis of this chapter was carried out from March to 1989. The Colombian Geogra~hical Institute 'Agustin Codazzi' flew a series of · photographs at scale I: 30 000 m the same year. In the field, integrated releves following the landscape ecological survey methodology described by Zonneveld ( 1979· . Zonneveld eta/. (1979), and KUchler & Zonneveld (1988). Plot size ranged £rom 5 ; 10 * 10 m, de~nding on t~e occurrence of Espe/etia harrwegiana or shrubs (Cleef 1983). The cho1ce of plot me was supported by species-area curves (Kent & Coker, Te~n ":as described by elevation, landform, slope, slope length, aspect, micro ·, and · rehef. S~tl profiles were described after augering down to depths of 120 em, each honzon the corresponding depth, pH, texture, colour, and presence of rc:o~s, o: lapill~. In almost all soil samples, a clear 5 to 10 em thick layer of lapilli dtstmgutshed, m general at depths of 60 to 90 em. Because this layer impedes the of roots and water, its upper boundary was taken as the effective soil depth. Erosion were included by estimating the average size (cm2) and cover (%) of bare soil spots.

Aspects of vegetstion structure included in the relevcs were cover, height, dominant of the vegetation strata distinguished, and total real cover. Stem rosette, tussock, ground-covering layers were recognized, sometimes with an additional shrub layer. For bunc? ~rasses (~ussocks) and stem rosettes, a few additional parameters were recorded, as mtmmum hetght, mean height, maximum height, density, number of dead stem l~ngth of Espelcria dead leaf column, tussock diameter, and percentage of dead . biOmass. Floristic composition was recorded by estimating species cover according to a Braun-Blanquet s..·ale (Mueller·Dombois & Ellenberg, 1974). Above 5% cover, the nPri'Pnt,,.,; was estimated, wher~as _b~low this value an abundance rating was given (r= rare, p= few, · abun_dant, m= many IndiVIduals). The growth form of each species was also noted. v .....uu·wn spectes were colkcted and dried for later identification in the Herbaria Nacional in A species checkli;t for the area was used for this purpose (Rangel eta/., 1983). The number of C<)W droppings (or faecal pads) per unit of surface area was taken as · indi~ation of gra~in~ intensity. Grazing impact was described by noting the abundance grazmg traces wtthm the vegetation (including traces of scraping by cattle on the roseu~s), the occurrence of cow paths, terracettes, and other trampling impacts such as spots 1nd~ced by cow hoofs. On slopes under low grazing pressure, trampling impact reflected m the <'.:cur:rence ~f (occasional) cow -paths. With increasing trampling impac~ terracenes develor as mterlacmg networks of low but long tracks following the contour lines (Selby, 1993). Thi~ process of micro-terracing is attributed to soil creep and small landslides at slope a~gl~s t<Xl great for stability (Gary eta/., 1972; Allaby, 1994). The spacing terracettes IS 1ntlu~nced by the size of the animals and the slope angle (Howard & 1987). Cattle _trea<ling can develop terracettes of several centimeters to 1 m in height and an average Width of I m (own observation).

Additional information on land mimagemcm was provided by local farmers and park guards. Indications of the number of years elapsed since last burning were especially instrumental where no appropriate coverage by ac:rial photographs was available. The height of the dead . Jdf column of regencmting stem rost:ues was also taken as an estimate of the number of · years elapsed since burning (Verweij & Kok, 1992). The method is documented in Chapter

5. Vegetation classification · ,\s a first'Step, vegetation structure was classified using TWINSPAN (Hill, 1979). This ·, clv.stcring program con~tructs an ordered two-way table from a sites-by-species matrix. Instead of plant species, structure variables were entered. The structure variables concerned represent '' height and cover of plant groups or vegetation strata. Stem rosette and tussock height were rransfonned to the same scale. The number of cut levels was nine. The values of the cut levels corresponded to 0%. p, m, 15%, 25%, 35%, 45%, 55%, and 75%, respectively, on the _ scale of cover and abundance. A weight of 0.5 was given to the structure variables tussock . -.. height, Espeletia height and·Espeletia juvenile cover. Tussock cover was assigned a weight .' of 1.5. The other variables, with a standard weight of 1.0, were cover of the strata of . Espeletia, shrubs (including Escallonia myrril/oides), tall grasses, (tall) forbs, short grasses, ground rosettes. cushion bogs, occasional trees, and bare soil percentage. The information from the first five cut levels was used to prepare the ftrst crude maoix, in which clusters of samples with a similar vegetation structure are distinguished. Subsequently, a few samples were reorganized manually by shifting them to other clusters where, according to the cover ranges of the diagnostic strata of these cl.usters, they fitted better. In this way, a key to the classification of vegetation structure was prepared as a hierarchical, dichotomous structure, where at each level discrete boundaries indicate the structure group to which a_certain releve belongs. As a second step, a floristic vegetation table was prepared using TWINSPAN. This rime, species data with corresponding cover values were entered. The cut levels.applied were the same as in the vegetation structure classification. Up to six division levels were used to distinguish the clusters representing different vegetation types. Another TWINSPAN table was produced for the data set of relatively dry bunch-grass communities. A summary in the form of a bar diagram has been published (Verweij & Budde, 1992).

3.1.4 Ordination analyses The aim of the ordination analyses was to describe a gradient of changes in the vegetation structure and floristic composition of the paramo tussock grasslands as a function of grazing intensity and fire history. To achieve this, it was necessary to ftrst eliminate the non-relevant variation in the vegetation parameters, i.e. the main natural variation and variation related to · management factors other than grazing or burning. 49

Characterization of vegerarion patterns Chapter 3 Initially, the full-d:na set was taken as the starting point for analyses using the CA program (Ter Braak, 1987a). All environmental and management variables were u·•~atut(Jid'!.' the set of environmental variables to be related to vegetation parameters. Three samples could be excluded from further analyses: those samples relating to the ...~.~~·..,;' ·3S described by Cleef era/. (1983), the azonal cushion bogs, and the regeneration potato cultivation which occurs in the lower areas. The relationship between excluded and the factors ·explaining the 'undesired' variation was confirmed by the ordination An iterative process of excluding irrelevant samples and subsequently analyzing the led to a final smaller data set where only management factors accounted for the variation in vegetation characteristics. Only samples containing stem rosettes were in order to have a consistent indication of the number of years elapsed since burning age). This final data set corresponds to the zonal bunch-grass vegetation of the paramo and its derived succession stages.

u is based on a unimodal response curve. CA gave a clear separation of the species. The icng:h of the first a:ds-wa~-3~8 standard deviation units, which indicated that the use of a un imodal response curve was more appropriate (Jongman et ul., 1987). On the basis of the CA. the best explanatory variabks were selected. Subsequently, direct gradient analyses were carried out with the selected variabl~s. using the technique of canonical correspondence analysis (CCA). CCA constructs ordination axes directly out of linear combinations of the environmental variables, thus distinguishing it from CA. The separate influence of burning and grazing on species composition still could not be seen very clearly in the prepared biplots due to the large number of species and management v3ri3bles. Therefore, three biplots were produced with CCA in which the first axis was kept . constrained for only one variable and the other axes were unconstrained. In each analysis, the constrained axis was constructed out of one determined variable. These variables were the number of cow droppings, trampling impact, and flre age. By this method, species reaction to one variable was revealed.

Vegetation structure The variation in the cover of vegetation structure layers among the samples of the final set was analyzed using a linear response model. Principal components analysis (PCA) ordination technique that constructs the theoretical variable that minimizes the total sum of squares after fitting straight lines to the cover data. PCA was carried out to best explanatory variables out of the 20 available. The natural logarithm (In) of variables was also included as a possibl~ variable. From the intra-set correl~uions;j~ environmental variables with the axes, the following variables were selected: • the In of cow droppings (n/100 m2), as an estimate of grazing intensity • trampling impact, on an ordinal scale of 0.0 to 2.0 with intervals of 0.5 • the In of regrowth of Espeleria leaf column, as an estimate of the number of elapsed since burning or fire age (em) • % live, the percentage of living tussock biomass • slope, in degrees Subsequently, an overall redundancy analysis (RDA) was performed to relate ve~:etaltl.®l structure to these five active variables. RDA is the canonical form of PCA and selects combinations of environmental variables to construct the ordination axes, minimizing the residual sum of squares. The resulting ordination clusters are described. The significatJce of the first axis was tested with a Mome Carlo permutation test, a standard of CANOCO. After 99 random permutations, eigenvalues did not change (P < 0.01). As it proved difficult to analyze the influences of grazing and burning separately, detailed RDA ordinations were carried out to describe the influence solely of grazing on structure variables. For this purpose, samples were divided into three different fire age Ordinations were carried out for each fire age class.

3.1.5 Preparation of vegelalion map As the existing vegetation map at scale 1: 50 000 (Kloosterman et a/., 1995) provides in~ufficient detail for the purpose of this study and pays little attention to human-induced

variation in the vegetation, a new map was produced for the sample area alone. However, this new vegetation map is also based on .the findings of Kloosterman et at. (I.e.). The delineation of vegetation units was based mainly on aerial photographs of 1989, with some additional photographs of 1983 and 1985. The scales ranged from 1: 25 000 to 1: 30 000. There were no problems_ with cloud cover, unlike in the satellite images then available for the study area. Photo image characteristics such as tone, texture, stereoheight, linear features, and pattern were instrumental in defining preliminary units. Each releve (marked on the aerial photographs) had been classified according to structure and floristics (Sections 3.2 and 3.3), which provided the basic information for defining final vegetation units that are homogeneous as regards a combination of structure and floristic composition. The means of geographical interpolation of classified point data are further documented in the results (Section 3.5). By using the classified field data of 1989 for reinterpretation, the final vegetation units reflect the situation of 1989. The interpretation features were digitized d1rectly from the aerial photographs into a GIS. The procedure for geometric correction and the connection of the' resulting vector maps is described in Chapter 8. The intermediate polygon map was rasterized to obtain the final vegetation map with a pixel s.ize of 10 * 10

3.2 Species composition For species occurring more than twice in the data set, both direct and indirect analyses were carried out, based on their cover values in each sample. First, a linear model was tested against a unimodel response model by using PCA and correSI>Oni:Jenfil analysis (CA), respectively. CA is a technique that constructs theoretical variables that explain the variation·in species scores;

Definition of vegetation structure types

Eleven vegetation structure types are distinguished in the TWINSPAN table (see Table 3.1). A dichotomous key summarizes the final classification rules, and serves as a decision support tool to evaluate the structure of paramo vegetation. 51

:: 0.. )(' ~

(IQ

O!l~ 0 .,

:::1 0..~

a 6.

C<r.! :::1 "'l 0..~

a !a. "' "' "' ..=

0;:;

"' =

;:;

"'~

c:r-o

8~ "' 3-·

0.., ~g-

< Q. .., c c

;;;r

~§ 0

..,

~·s (")

"' 0

g ..,~ 0 c .., 0

g -

(') 0

"' VI VI "' c;· <fl :::1'

~g' •

" 0-.o y.

~::r

~ ~ "' ~ "'

,..,

'ric

Cl)o

::r<

0::t 0..,_

::,.

c;;i"

1/\ ~

"'- .cfl

=,.,

~ .., "' 3 "'

:::1

0 :::1 !;"~-;:::::

:::1 "' c.::::

c:r-;

~ § 0 g..

OCI

•

CJQ

::raga

Cl)

~~ g

V> 0 OC'O .., ;:;_

"'l

-i ~ -

-0

o.,< r.!

~ lA ~ =r~;_

g (i == :-:-" '"' ; 1/\ ~ ~

w .. u.

~ cfl Q. ~

v.•

<fl..., ..

r,)

. g en c:r

"'l

a-~~ ii0' 6. ; a-a·· c:r"'

~ -

;;;r

"' 0 II\ ::I

-~ v...,

"' a='c

::1 0.

% 0

0..

~ 8 C.<

.., n ...,

c:r~

~ :::; ~ <> 0

E - · "'

b ~

;:;

..,n~ "'

::;0

ill

1/\ c:r-

;

s-- '"' s·.

en "'

- v. Vlcfl

c:r"'

"' "'

"' < 1/\ ~ c -

~ ~ 0

.., "' 1/\ 8

0 "' 0

::;

"'"o

"' 0 "' 0< ::> 0.. .., 1/\ N

N< VIC>

Cl)

~ ~

'J> "' :.ll. <> cr-

"' '-< a--· c""

Cl,.:jO" "' c "' "' n::s ~ ::r

"' .., "'l:2. 0 < <"'CJ Q ~ 1\ ~

:::1

::r· Q. -·en :r Dr ?'o: :::

::'~ ~

~~a

ira fJ. < c c:r ::>

~§g.

:::1

~ ~

0..

~0"' < ::r

g-~2

0. 1/\

~ t; ~

~ "' ("')

0 <

.., "' 0<

-·

~ "' v. :.\:lVl JJ 0

O'w i3-'-"

"' cfl

1/\ .

._,to v.~

@ ~ 0 0... "'

("')

II\~

:::1

~~g 0..

:r2 8 c:r<

8..6 '3 d-o

"' !a

II\

..,;;i

:.ll.., ? 1/\

"' ~

gj ..,

9. ~

(JC

...

(')

a .a~

~0~ a u: s

~~,., ~ ::r.. ~ -· en 0 "' 0 ..,

~ ~ c;;· c.v. w

v.o- o

cfl=~

Ln::t

g~~

0 c:r~VI

0.. ~

cod~

~ ~ ~ 0. ~

gj "'

Struc ture varia ble 2 Espa leti4 adul ts 3 Espe letia heigh t 5 tusso cks 6 tusso ck heigh t 8 Espe letia juv. 4 shrub s

12 cushi on bogs

13 bare soil 1 trees 7 ta l l grass es 10 short grass es 11 groun d rose ttes

9 forbs

Struc ture code

Samp le numb er Struc ture varia ble 2 Espe letia adul ts 3 Espe Jeeia heigh t 5 tusso cks 6 tU5SO Ck heigh t 8 Espe latia juv. 4 shrub s

12 c u shion bogs 13 bare soil 1 trees 7 tal.l gra.s 3es

10 short grass es 11 qroun d rose ttes

9 forbs

Table 3. 1

(,1)

1/\

••

"' :::s :ro

::IO!j

O!l

"'~ "'~ ~ c: 1/\ ::! 0

~:3

a"'

...,.,

og~

- ""' 0

N~8

~("')

~E.~

•c,Sioc

o:;. Q .., V

OV>" O

a to~

0 0 -""

VI~

ao c: "' ~.

(1<>

~~O!l

g;@~

"' """'

VlO<

~0~ ......

a5~ .,nw

oau ; ~5

g::r o

Q .....

e:t?

:::> c ao

0 -

::n :::l

[.e: :::1

c:r::r ~ "' cn "

()Q~cn

~ O!l ..,

0 O.&J

""'

~·

~1(

-· E;

n ..,

1!o:: (,I)

r:r (,f;

~ .._,

0 "' -"' ~5

"'0

"' OCI

so~ a5.

8.2~ (YST g

Vt::s <i

0..

-8. 5 > g_

0 0

~ oc;· en ::: :r 0

&g ~~~ "' C> "'

.B~ c

~ r 3 ~ ~ ""1

6'3 8

::!1,..

::> "' 0\ P..3 u.

"':r 0 <> <fl •

o~gg

. g' ~

0 (') c."' ::>

5' ~ ~ ~-

0 ::>'<

s· g-

~'ifl =•

..,

('}

l"'

~..,~

l»

0 gj! ~

~ c "' "'

s.a ;

••

OO o-

Cl)

iiU'l~ O!l -

<> -

"' ;;l ::rO'Q §~

~ C) JP =r - · C )

Cl)

(')

::;cn o

Struc ture

v;·

::s

...,

ta)

a2 "' "0=0"'c:r "' < 0 "' .., < ., 0

~~00 ~ ='"' (") 1/\ 2 3

~[~

"'~OCI ~ o

::>

O..;;;r

..,

0..

Samp le numbe r

a?l: r. :><"0

0 CJQ

en ::J ::r·~

ac ..g_ ~ ~~

("')

~ ~ ~ C.orl

i'fc:r~ - n 9"

O!l :.:;-

;::;

I:X1

<flc. ::r

5· ~ ;;, :=. 2 c

= ::::1 '< ~ O!l" .,a "~c.

<> "" ~ o~

tv~§ '-" :::1

•

.fl. ::r "' "' ::r

f!~

?.-

o-('")

\NO~ U\0>= ~~.

'< 0"

~ ::l

;r:~

"'0~.§!. ~ • Q.

::r c:r- "'

~t;;g.

<»0"

-~(lQ

g, 5 ., R· c5 ~

-· "" v;· = -= a., -= ~ ii ~ ~~

:r:r o-

11\~-l

"'=

(5l~ g-3 g-

E - c Si!-: :s

til(/)

g~~

:::l

-a

~ <iQ" ~

C.t;o

0 "' 0 0\~3

~3~

CJI

0..~

:::1

-00

0 cflo.. C . ""'3 Qs~

N::l? .-

OCI

sg, ::

..,

~"0 2:

..,ac

~ r;

- · V'l tn~;:

(JQ

::r

::I~

3 ~

r.fl .., ~8 <

=ooj;3

Vlcn o cfl g "0

I:X1o

e:o" Ci " 2;

g·

sa:r..,3

c:rn oo

~·

§~2:

c;!l' §. c

::>

(')

... !2..

OCI

(')

!':!

sssss sbss ss bssss ss sssss ssbbb 58118 92238 9 24121 02 27078 21104sssbb bsbs b3bbb bbbbb bbbb ssbbb bbbb sssssbbbbbb bbbss bsbbb

657 8 40069 98 17071 84 37740 15624 70211 6340 18332 11043 4562 54004 55560 17723 21445 55602 56578 7483 12 62392 72915 24 43613 2 8540 07842 68601 50054 32142 21173 ---- ---- -- ----3 4- -2323 3323 2221 111-1 14 21123 231 1 312 -2322 13223 23233 23232 ----- ------ ----7 9- -8889 94488 231-- - 2123 56543 32454 -- 3723 2-- 88888 89888 59595 6-5- 98599 59573 5 765 -5745 75666 79455 48376 654-76667 67677 - 79 467- - 66686 87776 89898 9888 76667 777 76 667 6 54565 34455 54544 44454 55543 -4747 - - -- - - - -- --1-- 11 1221 2-122 58786 7868 77675 46777 6567 74365 553 55345 33 4544 33437 33336 4435 35 33333 43333 - 5 444 5 45 32223 3222211--1 11-1- 11---221- -- 112 -22 2- 11122 2-2 1211 -112 11--2 44556 67877 6 - - - - - - - - - - - 2111 2 1212 21222 23211 1212 -2221 12112 12112 21111 111-2 2-11 21212 1 - - - -- - - -1112 1114 2---- -- 32 33133 3332 11241 2 2 3 34344 64336 3543- 4 5 4554 54 55665 66666 56566 49944 75 ----- - - - -12- --- - - - - 1 - - - - - - - 11-1 ----- --1 2-133-3 1232 24 23534 33 22221 ----- ---------- ---- ---- 12222 22112 223- 22223 22212 2134 33343 43454 2 57599 99 24454 42544 64332 44433 3637 54245 35533 4334 45556 64531 22223 33123 33331 14422 54444 34223 3 62222 22 34332 22233 44656 43325 5634 34336 34666 5 32323 3342 22332 33433 5332 13322 32222 22352 22424 32332 22322

·-- ---- ---- -- --- --- --- - ---- ---- ---

•

sbbbb bsbbs sssbs soso1 ossbb bbbs 40436 43034 30435 48820 42111 1067 sssss bss sobb bssbb ssbss sbbss ss 1 59336 34451 94070 67802 99453 852 33311 377 33325 99235 46150 06595 4 62042 539 78776 31885 883 05909 506 -2132 2---- ---43 12-- 3--14 22212 -757 75--- ----7 99 9--8- -5764 944 244- ---- 2324 1---- ----- ----- -656- ---- 697 79-- ----66655 44446 75324 55653 55554 -112 ---- 1131- ---2 31-----56756 36357 66568 68777 755775434 1--- 6656 4 5 4---- - 646- ----- 879--5-- --11121 111-- ---2- -2-12 --12 21212 21223 12213 13323 22311 -1123 1212 232- --- 2 - 2 1 - - - - - 1 1 - - - - - - ---- -- - - - ----- --- ----1 21122 --- 22131 1223- 111- 1---- -33344 44443 122 3 23 3323- 12322 331 - - ---1 2 - 1 -- - -- - - - - - - - 131-- 1-1 1321 -3 12411 3--2-3 1112 2--1 ----1 ----- 1---1 ----- 1---2 1------ --12- 3- 22-44 ----- 21 34 23422 33435 53133 21125 36536 4558 88688~a8 55155 33414 23433 76455 58675 58878 86785 85557 7663 43224 53677 45 88989 89989 89999 99988 1 32323 42222 32322 22333 22332 3232 22222 232 22222 22222 22333 22243 8 3

----------------------------- --------

•

b1ss s sssss ss sssss sssss bsbsb 20666 76666 55 35896 6 94 79 59296 61085 16421 32 51897 34902 96976 12-- ----- --88--------2- - - --6-- ----- --- -- -- - 3-12 2-11 1------- ----1 221-1 21132 55 - 3 -- - 77667 98867 57 242-3 21112 22 0 5865 44434 43 33244 24364 44

----- --------- ----4332- - 9888 ----- - - -- - - 3132 --11- 221 3 - - -1 2211 44 4224 - - ---3 --2--4 43-1 44231 122-4 8753 43254 5 96886 68895

2---4-1---3---1-l-1----18865 23246 65455 223 33

Vegetation struct ure table based ori clusre ring by TWIN SPAN . Struct ure codes: a= shrubland or shrubby grassland, c= cushi bunch grassland, e= sparsely vegetated land, on bog, d= dense f= (tall) forbland, g= shon grassland, h= medium-dense bunch grassland, m= mixed o= open bunch grassland, s= short matte grassland, d herbland, t= tall grassland. The cover codes are trans'Iated as follows: 1= r or p, 2= a 15%, 4= 15-25%, 5= 25-35%, 6= 35·45 %, or m, 3= 5· 7= 45-55 %, 8= 55-75%, 9= >75%.

... Characterization of vegetation patterns

3.3

~no_

>10o/·C~

Definition of floristic types

The first phytosociological classifications of the paramo vegetation were published by Cuatrecasas (e.g. 1958, 1968). Cleef et at. (1983) gave a firs t approximation of the classification and eleva tiona I zonation of the vegetation of Los Nevados National Park. The vegetation zonation of the study area was described most recemly by Salamanca et at. (1995). In the following description, the vegetation types resulting from cluster analysis are presented and related as much as possible to the existing vegetation classificatioo of the area. The floris tic TWINSPAN gradient described below runs from high to low elevations, and from intact to the most degraded vegetations.

Cushion bog

Tall g-assland

Senecio tatijlorus - Calamagroslis ligulata superparamo vegetation (n = 3)

The first floristic type on the gradient prepared using TWINSPAN is represented by three samples. These we£ taken merely as a reference in the higher superparamo, between 4400 and 4570 m. The differences with the paramo proper communities described below are clearly recognized. Above 4300 m, the vegetation becomes more sparse. Calamagrostis recta, C. effusa and Espeletia harcwegiana are virtually absent. Senecio canescens, Cerastium floccosum, Draba Ira/Iii, Tri.set11111 andinwn, Agrosti.s foliata, Festuca doliclwphyl/a, Lucilia kunthiana, Erigeron chionophilus and Catamagrostis ligutata are diagnostic. Bromus tanatus is common, especially as a colonizing species in tall grasslands on recent volcanic sand deposits.

W.ec ·Gill?- bunch grassland

This type corresponds to the association of Senecioni latiflori · Calamagrostietum ligulatae described by Salamanca et al. (1995). Safar, patches of this association have been reported between 4400 and 4500 m. Vegetation cover is sparse, and corresponds to either sparsely vegetated land (e) or tall grassland (t). Scattered Senecio ground rosettes, small Werneria hwnWs cushions, and tufted grasses of Agrostis foliata and Calamagrosti.s ligulata are conspicuous features. In the vegetation table, the floristic proximity to the Plantago rigida cushion bogs is apparent.

Short grassland

Shoct herbland I

>30o/. FO [ f ] G n o <20% BUT

fcmtand

m ~}11!:-:J

Soils are shallow and not differentiated. They consist of loose, colluvial material of coarsegrained volcanic sand. Rocks were encountered at 20 to 40 em depth in the profile. Diurnal freezing and thawing are characteristic of this subnival environment. The continuous retreat of the glaciers allows subsequent colonization by ·this sparse and species-poor community. No grazing influence was recorded.

g-e:sstand

Genlialulla dasyantha • Plantago rigida cushion bog community (n

Figure 3. 1

Dichotomous selwk'n key to the classification of vegetation structure. Qu<mtitatnt criteria are based on CO\'er of characteristic plant groups. BUT= bunch CUS= cushion plants, FOR= tall forbs, ROS= ground rosette layer, SHO= grasses, SHR = shrull layer, TAG= tall grasses. 54

= 16)

The cushion bogs are represented by l 6 samples. With the exception of one releve, all samples represent the Gentianello dasyamhae - Plantaginetum rigidae, subassociation brometosum lanatae.

Chara,·terization of vegetation pattems

Chapter 3 The dor..:.:am species are the cushions of Plantago rigida and Werneri a crassa, the shrub Di:.:~ ri:;rr.a emperrifolium, the shrub Hypericum lancioides, the sedges Carex OOilpiGtfldl• and Unc.''_jj meridensis, the tussock grass Festuca sublimis, and the short grass coarcrarc.. Cu;hion bogs of Plamago rigida have also been reported for the Ecuador ·Godi!y. 1978; Acosta·Solfs, 1984), for Peru (Gutte, 1985), for Bolivia Menhofe:. I%lJ, and for the paramos of Venezuela (Monasterio, 1980). Diagnosti; species are Muhlenbergia cleefii, Pedicularis incurva , and Niphogeton Taxa ch<:.-.:..::te..'istic of these cushion bogs are Colllla mexica110, Laclzemilla marnd.oni,aril.'ll Disrerig,.,..; empeirifolium, Oritrophium iimnophilum, Ranunculus peruviamlS, vaccinioi.:ts, Gentianella dasyantha, and Calamagrostis macrophylla.

Seneci() j:"'1UJS!.lS, \'aleriana planraginea, Brom!.lS lanarus, a~d £/aphog lossum mathewsii diagnosri; :o: the subassociation bromerosum lanarae (Salamanca et a/., 1995). One (s98) rep:-:;em.s a transition between types A and B. It resembl es the variant "'·'·ur•·•n.;oo mriScoide~ of th= IVernerierum humilis Salamanca et al. !995. This cushion bog a>:>I.JI.I'"u'JIIJII occurs at '-.uund 4200 m onJ:old wet valley floors, where the water table is near the (Salaman.::!. 1991; Salamanca eta/., 1995).

Table 3.2

TWINSPAN noristic vegetation table of the upper watershed of Rfo Ouln, Los Ncvados National Park, Colombia: this table is presented in Appendix C. TI1c total number of taxa is 203. Legend of codes: Species cover and abundance: 1= r or p, 2= a or m, 3= 5-15%, 4= 15-25%, 5= 25· 35%, 6= 35-45%, 7= 45-55%, 8= 55-75%, 9= >75%. Elevation: I= 3600·3700 m, 2= -3800 m, 3= -3900 m, 4= -4000 m, 5= -4100 m, 6=4200 m, 7= -4300 m, 8= -4400 m, 9= >4400 m. Terrain: B=block lava, C=colluvial slope, D=glacial depression, E=end moraine, F=lacustrine flat or cirque, G=glaciated lava field, L=latcral moraine Slope: 1= 0-10", 2= 11·20", 3= 21-30",4= 31-40", 5= 41-50°. Mottling: ordinal scale from I (a few small mottles) to 6 (many big mottles. in soil profile).

Grazing intensity: 0:: 0 cowdr./50m 2, 1= 1·2 cowdr./50m2, 2= 3-8 cowdr./5 2 0m , 3= 9· 20 cowdr./50m1 , 4= >20 cowdr./50m1. Fire age: code corresponds with the number of years elapsed since last burning, while 9= >8 years. Structure: code corresponds with that of the structure classification (Table 3.1).

The domi;..;.11t \·egetation structure is that ·o f the cushion bog. Some samples are classified forbland ~ ~-· under the influence of grazing. Especially where many hoof-marks or cow are regist=J. the tendency is a transformation of the cushion bogs into (shrubby) Excessive ;:-ainage of the cushion bogs by grazing and tramplin g impact seems to be cause of rb:;e sites having dried out. In the intact cushion bogs, grazing has occasionally recorded. •i th low intensities (0-4 cow dr./50 m2). Ground cover is complete. In forblands. ::owever, up to 15 cow dr./50 m2 and 5% cover of erosion spots have recorded. One sampi: (s391 corresponds to a cushion bog burned 30 years ago, which had rl'r·nvo~rP.il to a large !:<:tent Shrubs still show traces of burning. The limited length of both the branche s a::.;! the extensions of the old ones indicated, that growth rates are extremely Based on 3) years of post-fire regeneration, a growth rate of I em per year was derived

Escal/()nic >';yrri!loides.

Cushion b::-;; are associated with valley bottoms, glacial depressi ons or concave parts of lateral mo~ines, and with depressions in glaciated lava fields. The average slope is 7.6•, with a rJnge of(! to 15°. All samples have a groundwater table at a depth of 10 to 150 em. Half the sampb z;e permanently waterlogged, with a water table at I 0 to 50 em, and consequently . do not sh0-.:. signs of mottling. In the other samples, mottling is very conspicuous, indicating a tluctuati~.f wat~r table. Lapilli layers are either frequent or absent through out the profile. It can be ~~oned that pyroclastic material was deposited by wind action and might not have reached so:-:e concave parts of the slopes where cushion bogs are located. It could also be ' n:l:ued w ~ ,·ariable rate of accumulation of organic material in the profile. The elevation r.tnge of L'l! samples was from 3920 to 4140 m (4062 m on average ). When the samples not locatec in flat terrain, the gentle slopes are exposed predom inantly to the west or northwe st Two san1ples have a more eastern aspect, and also showed a distinct vegetation structure (< :!.1d f).

Chapter 3 C

Clraracterization of vegetation parrerns

Espeletia hartwegiana - Calamagrostis effusa bunch grassland

A general tendency is visible when structure is compared with average fire age. The more open the vegetation, the lower fire age. In t h~ case of dt!nse bunch grasslands that were recently burned (three to tivc years before), thl!re is no actual grazing influence, which facilitates a quick recovery of the vegetation.

(n = 21)

This bunch grassland community is most similar to the association Espeletiewm - Ca/amagrostiettun effusae Salamanca era/. 1995. The dominant species ar~ the bunch Calamagrostis effusa, the stem rosene Espeletia harrwegiana ssp. centroandma, ~nd the Carex tristicha. Accompanying species are the bunch grasses Calamagros11s recta festuca sublimis, and the short grass Calamagrostis coarctara. According to Salamanca et al. (1995), the absence of Cerastirtm s~ws~icatum: racemosa, Valeriano planragin"ea, Senecio formosus, and Aphanacus Jamesomana . diagnostic. According to the present findings, Baccharis genisre_ll~ides, Azorell~ . Festuca andicola, and Gnaplza/ium graveolens can be added to th1s hst. Charactensuc that are shared with the other bunch-grass communities of the study area (types D and Lycopodium clavatum, Castilleja fissifolia, Hypericum /aricifolium, an.d Aa colt7mllianaH Formerly, this association was described as pure C. effusa stands, exclud1~g the. . C. recta. However, C. recta is present in most cases. It can be concluded that II IS absence of C. recfa that is diagnostic, but rather the presence of certain cushion-bog that are scarce or absent in the other bunch-grass communities. Examples include A 7 ''rpr,rm.

In order to compare grazing intensities, the number of cow droppings was converted to a figure on a yearly basis for the sites most recently burned, i.e. less than one year since burning. The combined fire and grazing gradient is briefly characterized as follows: Dense bunch grasslands with an average fire age of 7.2 years (two unburned samples not counted): average grazing intensity is 2.5 cowd.r./50 m2• Cow paths and terracettes are absent. (n = 8). Medium-dense bunch grasslands with an average fire age of 3.8 years: grazing intensity is 3.0 cowdr./50 m2• Some cow paths and hoof-marks are present. {n = 3) . . .. Open bunch grasslands with an average fire age of 0.5 years: grazing intensity is 5.8 cowdr./50 m2• Some cow paths are present and, occasionally, the formation of terraceues is observed. (n = 8).

Two samples correspond to a mixed grassland after recent burning and are subject to high grazing intensity. These samples occur on more gentle slopes without terracettes (5 and IS").

aretioides, Lupinus rolimensis, Huperzia cruenta, Carex bonplandii, Ca;ex tristicha, Lachemilla hispidu/a. Almost exclusive species are Cortaderia spp. and vulcanica. The Espeletietum harrwegianae - Calamagrosriecum effusae community is associated humic and poorly differentiated andosols, and with the desaturated andic rankers d.'e!.;cri bed\.l by Thouret ( 1983, 1989). Sometimes, the water table is present just beneath th~ lap1lh In some cases, the lapilli layer is· also soaked, or has a ferrolayer on top. Monhng values high: on average 3.1 ± 1.4 on a scale frotn 0 to 6. Mottles have bright red and orange and are often big. This is a sign of a fluctuating water table.

Calamagrostis effusa- C. recta (var. typicum) bunch grassland n

=35

Of all bunch grasslands, this community is associated with the ~ore humid. places. coincides with the observation that a number of species are shared wllh the cush1on bo?s. slopes east of Laguna del Otun, the eastern part of Lorna Bonita, and the surroundmgs Laguna Negra and Laguna Ia Leona are covered by this vegetation, and th~s form a ~ore less continuous belt. Kloosterman era/. (1995) concluded that this commumty (to wh1ch assigned code t7a) is associated with coarse-grained ashes. If the more hum.id conditions of floristic type C are indeed related to the occurren~e of ~oarse-grat~ed these have a wider extension than had been assumed for the semt-deta1led vegetauon

The Calamagrostis effusa - C. recta (var. typicum) community is largely similar to type E. Both types D and E are variants of the same associaiion Calamagrosriecum effuso ·-rectae Salamanca et al. 1995. According to Salamanca et a/. (1995), the combined presence of Espe/etia hartwegiana, CastiJieja fissifolia, and Calamagrostis efjusa is diagnostic. Characteristic but not exclusive taxa are Cerastitun subspicarum, Luzula racemosa, and Senecio formostlS. Valeriano p/anraginea and Aphanacris jamesoniana are also mentioned as characteristic species, but in the present study area these are rarely observed beyond the range· of the cushion bogs. However, some other species appear as additional characteristic species of this association, though they may occur more abundantly above and below the paramo proper. These are Myrrhidendron glaucescens, Gnaphalium graveolens, Elaphoglossum mathewsii, and Arenaria spp.

Elevation ranges from 3980 to 4220 m, with an average of 4108 ± 69 m. A~erage steepness is 24.3 ± 8.5•. It can be concluded that this community represent~ the h1ghest grassland belt of the paramo proper of the study area, with zonal vegeta~ons_ on moder;atelJ'I to poorly drained slopes. The predominant aspects are N, NW, and W, whtch ts related to macro-topography of the study area.

The typicum variant is distinguished from the Esc-a/Ionia myrti/loides variant (type E) by the presence and absence of several taxa. Azorella multifida, Lachemilla hispidula, and Lupinus tolimensis are differentiating species. The combined occurrence of Carex pichinchensis, Oritrophium peruvianun1 and Baccharis rupicola is diagnostic. Fesruca sublimis generally covers more than 5%. Escallonia myrtillodes and Gentiane/la dasyantha hardly occur, whereas Geranium multiparricum, Aciachne acicularis, Poa subspicata, and Grammitis moniliformis are completely absent in the typicum variant.

Three main vegetation structure types are related to type C: dense, medium-dense, and bunch grasslands (d, h, and o, respectively). There is one deviating sample of a • burned mixed grassland with high grazing intensity, located on the most gentle slope (5 ).

This bunch-grass community occurs within an elevation range of 3930 to 4120 m, the average being 4014 ± 62 m. This coincides with the elevation range of 3900-4200 m reported for the association to which it belongs.

·l

Characterization of vegetation patterns

Chapter 3 In the Otlin study area, it is restricted to the lateral moraine of Lorna Bonita and the block lava flows . Due to the orientation of the lateral moraine and the slopes at the the block lavas, the.nonh- or south-facing slopes are dominant. Average slope ·--· · .,.... 24.9 ± 7.1•, which is comparable to the average inclina tion of the other variant, type C.. Soils are well drained, non-differentiated andosols with an A-C profile. !(present, mottles few and small, with a brown ish-red to red colour, though less bright than the ones nrc·11rr;ft':.'llll in the soil profiles underneath floristic type C. The average mottle score is 0.5 ± Occasionally, a wet lapilli layer or ferrolayer is presen t. A trend similar to that observed in type C is visible in the response of structure to combined burning and grazing regime. Predominan t structure types are also bunch gra:ssl:anii!IU of variou s densities (d, h, and o). A few mixed grassla nds occur on the more gentle The gradient of predominant structure types in relatio n to mamigement is as follows: Dense bunch grasslands with an average fire age of 3.2 years (one unburned not counted): average grazing intensity is 0.8 cowdr ./50 m2• Some cow paths present. (n =5). Medium-dense bunch grasslands with an average fire age of 1.7 years (two o"''''""'';:~• not burned): grazing intensity is 1.7 cowdr./50 m2 • Cow paths are often presen t trampling impact in the fOIJII of hoof-marks and terrace ttes was observed. (n = 7). Open bunch grasslands with an average fire age of 2.0 years: grazing intensity is cowdr./50 m2 on a yearly basis. Cow paths are presen t but numbers vary, hoof-marks and terracettes occur frequently. (n = 16). Mixed grasslands with an average fue age of 1.6 years: average grazing intensity 6.2 cowdr./50 m2• Trampling impact (cow paths and terracettes) is pronounced. (n =6).

Calnmagrostis effusa - C. recta (var. Escallonia myrtil loides) bunch

n = 26

grassland

Diagnostic for the Escallonia myrtilloides variant of Ca/amagrostietum effuso - rectae is the combined presence of Calamagrostis effusa , C. recta, Espeletia hartwegiana, the dwarf tree £sea/Ionia myrtilloides and/or the shrub Hypericwn /ancioides, the forb Gentianella dasyantha, and the creeping species Satureja nubige na. Characteristic species that are generally absent from the typicum variant, are Geran ium multipartitwn, Huperzia cruenta, Trisetum irazuense, Hieracium tolimense, Azore/la crenata, Stachys elliptica, Aciachne acicu/aris, and Poa subspicata. Shon grasses and shon forbs are more abundant than in the l)picum variant. The dwarf shrub Pernettya prostr ata and the ground-covering species Lupinus microphyllus are also more prominent. Other differences are indicated above. The variant of type E is related to the extrazonal Aciac hne acicularis - Escal/onia myrtilloides dwarf forests reponed by Salamanca et a/. (1995 ). The stunted Escallonia forests were sampled at 3910 and 3930 m in the vicinity of Lagun a del Otun. However, Escallonia cover is much higher (60% or more) and bunch grass cover much lower (around 10%) in comparison with the zonal bunch grassland type documented here. The elevation range of this bunch -grass community is simila r to that of typeD, from 3945 to 4155 m with an average elevation of 4005 ± 71 m.

Block lavas arc the domin an t terrain type (13 out of 26 samples). Where block lavas underlie this vegetation type, soils are generally well draint:d. Mottles are scarce, the average mottle score being 0.3. In the case of lateral morain es nnd glacia ted lava fields, soils arc imperfectly drained, which is reflected in an average mottle score of 2.4. In l(Jsampl es out of the 26, rocks or big stones wer~ encountered in the soil profile. Where rocks are present, the average soil depth is 72 em. The presence of nearby rock outcrops was recorded on many occasions, even for samples not located in block lava flows. Withi n this community, Escallonia myrtilloides is strongly associ ated with the block lava~ . A twofold explanation is possible for this association. On one hand, the bl~k lavas provide a ran~e ?f micro-habitats, some of them especially favourable to the establtshment of woody species m . terms of shelter, humidity, and radiation balance. During the night, big dark rocks slowly release the energy captured during the day, resulting in less extreme cooling of the surface. · ·Roots of the dwarf trees and shrubs may reach the subsurface water running across the block Java substrate. On the other hand, human intervention may have led to a dramatic reduction in woody cover in the more accessible areas of latera l morai nes and glaciated lava fields. £sea/Ionia myrrilloides, in particular, is suitable for fuelwood purposes, as this dwarf tree fonns thick branches. Which explanation is more valid is hard to say, but a combination of both factors seems likely. Average slope steepness is 21.6 ± 12.9°, slightly lower and more variable than _the slope , steepness values of types C and D. The high propo rtion of block lava samples w1th gentle _ siopes accounts for this difference. Apan from the block lava flows, the average slope does not deviate from that for the other bunch-grass comm unities. The vegeta tion structure gradient in relation to mana gement is as follows: . Dense bunch grasslands with an average fire age of 6.3 years (one sample not burned): average grazing intensity is 0.7 cowdr./50 m2• Cow paths are generally absent. (n =6). Medium-dense bunch grasslands with an average fire age of 7.1 years (one sample not burned): not significantly different from the dense bunch grasslands. However, grazing influence is more pronounced, average grazing intens ity being 5.2 cowdr./50 m2• Co":' paths are sometimes present. (n = 4). • Mixed grasslands with an average fire age of 4.4 years (two samples unburned): grazing intensity is 6.7 cowdr./50 m2 on average. Some cow paths are present. (n =.I 0) . Open bunch grasslands with an average fue age of 3.3 years (four out of e1ght samples unburned): grazing intensity is 9A cowdr./50 2 m on a yearly basis. Cow paths, hoof-marks and terracettes are frequent (n = 6). F

Aciachne acicularis - Calamagrostis coarctata short grassland . n =23

This community c~rresponds to the association of Aciac hnetwn acicularis initially described by Vareschi (1953) for the pararnos of Venezuela. Cleef (1981) also described it for the Cordillera Oriental, and Salamanca et a/. (1995) define d two subassociations for the RuizTolima volcanic massif. The grassland community of type F belongs to the subassociation /upinerosum microphylli. Differentiating taxa for the subass ociation are Lupinus microphyl/us, Lachemil/a orbiculata, Agrostis haenkeana, and Hypoc hoeris sessiliflora. 61

Characterization of vegetation pauerns

Chapter 3 The dominance of Calamagrosris coarctata is especially remarkable; this short grass at least 10%. The other dominant species are the short grass Aciachn e acicularis and creeping species Lachemil/a orbiculata. The short grass Fesruca breviari stara is "'~""''<"" but does not occur frequently. Calamagrosris tussocks and Espeletia stem rosettes common, whereas the bunch grass Festuca sublimis is almost absent. The vegetation slructure types are shon grassland (g), mixed grassland (m), and shrubby gra:sslrul fu,

(a).

The elevation range is from 3950 to 4180 m; average elevation is 4084 ± 61 m. The Ar.inrJI,~~l:l

acicularis - Calamagrostis coarctata community is developed in the northern northwestern sections of the study area, from Paramillo de Santa Rosa to Laguna del and is associated with less-dissected terrain. The average slope steepnes s is 12.4 ± 10.1•. I

All slopes with an inclination of more than 5• have an east or southeas t aspect. remarkable phenomenon can be explained by the favourable radiation balance of east-e>wosea,'··• slopes, which receive sunshine in the morning when clouds are not yet intercepting radiation. Therefore, it is assumed that at a given elevation east-exposed slopes are warmer than slopes with a west aspect. For grasslands on Mount Wilhelm , New was demonstrated that vegetation on east- and west-facing slopes was related to climatic differences Smith (1977). "It is not surprising that Esca/lonia, in particular, is abundan t on the slightly warmer east/southeast slopes. Calamagrostis coarctat a and acicularis might as well have a preference for warmer places. Klooster man et al. ( mentioned the occurrence of A. acicularis on dry, convex slopes in the basin of Laguna On1n. The abundance of mottles in the soil profile is variable. The average mottle score is 2.7 ± 1.9, which is moderate. Where the Escallonia myrtilloides variant occurs, mottles are generally · abundant (mottle score 3.4). Where Hypericum spp. and Ecallonia are absent or have low cover values, mottle score is lower: 1.4 on average. It can thus be conclude d that both woody . species tend to be associated not only with warmer places but also with the more humid sites. The pronounced fire regime of types C, D, and E is absent. There is one sample with a relativel y recent fire age of 5.8 years. Four other samples have a fire age between 20 and 34 years, but no evidence of burning was recorded for the rest. One sample classified as short matted herbland was burned 31 years ago, meaning that the transitio n from burnable to unbumable vegetation is possible within that time span. Only one dead Ca/amagrostis recta bunch was recorded for this releve. Grazing intensity is highly variable. The gradient of structure related to increasing mean grazing intensity is as follows: Short matted herblands: 2.7 cowdr./50 m2 on gentle slopes. (n =3). Shrubby grasslands: 10.0 cowdr./50 m2 on moderate slopes. (n = 5). Short grasslands: 23.4 cowdr./50 m2 on flat parts (slope up to 8°). (n = 7). Mixed grasslands: 25.9 cowdr./50 m2, variable slope steepnes s. (n = 8). Cow paths are sometimes visible, especially on slopes. Trampling impact in the form of foot· marks is occasionally pronounced Despite the high grazing intensities, the amount of bare soil is low. 62

Ground cover is almost complete, the highest amount of bare soil recorded being 12%. . Terracettes arc almost absent. From the data on floristics and structure, it can be concluded that this commun ity is most probably derived from type E, the Escal/onia myrtilloides variant of Cala/7Ul grostieuun ef!uso . rectae. The vegetation table demonstrates the floristic pro~timity as type F is adjacent to type E on the gradient. Species shared by both types, thus supporti ng this hypothesis, are

Escallonia myrtilloides, Gentianella dasyantlra, Aciachne acicularis, Poa subspicata, Potentilla lreterosepala, and Geranium multipartitum. They also have in common the higher abundance of LupimlS microplzyllus, Satureja nubigena, and Bacchar is genistelloides. The Aciaclme acicularis - Calamagrostis coarctata short grasslands include releves where Escallonia myrtil/oides reaches higher cover values, occasionally liP to 40%. This indicates the affinity of type F with the Aciaclme acicularis - Esca/lonia myrtilloi des dwarf forests, parallel to type E. Both commun ities can be regarded as relicts of the dwarf forests.

- Paleoecological studies demonstrate that around 5000 to 6000 years BP the forest line was located a couple of hundred meters higher than it is today (Van der Hammen , 1974, 1979; Kuhry et a/., 1983; Melief, 1985; .Salomons, 1986). This implies that the area comprising types E and F used to be under forest cover. It is hypothesized that with a decrease in temperature, the Escallonia dwarf forests gradually evolved into bunch grasslands where Escallonia is still present as an accompanying species with a shrubby growth form (type E). The relict vegetation only survived in relatively warmer and sheltered conditions, e.g. close to Laguna del Otun. In other areas, replacement by types C and D took place. On the more gentle slopes, grazing history, possibly in combination with fire occurrence, resulted in a transformation of bunch grassland E into the Aciachne short grasslands (type F). The mixed grasslands of types E and F are consider ed the 'stepping stone' in the rransitio n from E to F.

Agrostis ltaenkeana - Lachemi/la orbiculata short grassland n = 13

This community coincides with the Agrostio breviculmis - Lachemilletum orbicularae Cleef 198 1. Cleef (1981) mentioned that Agrostis haenkeana is usually far more common than A. brevicu/mis, which is even absent here. The ground-covering species Lachemi l/a orbiculata dominates these herbaceous short grasslands. Taraxacum cf. o!ficinal e, Poa anmw, and Trifolium cf. amabile are diagnostic for the study area. Dwarf shrubs and bryophytes occasionally reach high cover percentages. The dominant short grasses are Agrostis haenkeana and Festuca andicola . The abundance of the latter is especially diagnostic. Baccharis genistelloides and Hespero me/es heterophylla are characteristic but not exclusive species. Besides Taraxacum cf.. officinal e and Trifolium cf. amabile, taxa that are shared with grasslands subject to stronger human influence (types H and.! below) are: Veronica serpyllifolia, Cerastium arvense, Trisetum irazuense, Acaena ova/ifolia, Anrhoxantum odoratum, Poa pratensis, and ]uncus bufonius. The species Azorella multifida, Plantago /inearis , and Lachemilla hispidu/a are shared with type D, and this affinity implies that type G is most probably derived from type D. 63

Characterization of vegetation palterns Chapter 3 The association was described as a replacement community in heavily grazed bun grasslands and deforested areas near the high Ancfean forest line. The uppermost forests ha~ been part!~ re~oved thro~_ghout the area to give way to extensive livestock production ano potato ~ulnvanon (VerweiJ & Beukema, 1992; Kok et al., 1995). The elevation range oftJi ,1gros~IS W:enkeana- Lachemilla orbiculata short grassland is from 3845 to 4050 m. Averagt eleva:10n IS 3938 ± 6~ m, which is significantly lower than that of the short grassland~ descnbed _above. Slope steepness shows similar vaJues, mean slope being 16.3 ± 9.7". ty~e F, _this ~hort ~assland is associated with lateral moraines and glaciated lava fields, aiij' wuh soils w11h an Intermediate mottle score of 2.0. Aspect is variable.

Lif

The dominant structure types are short matted herbland (s) and mixed grassland (m). Ollt". sample represents a humid shrubby herbland derived from a degraded cushion bog, anothlf a forbland of Rtunex acetosella and Lachemilla orbiculata. Both sites were used in the pal for potato cultivation. Only one sample was burned 20 years ago. The actual vegetation of: the burn~d sample corresponds to a short matted herbland. This shows that the rapidr degeneration of burnable bunch grassland into short matted vegetation is possible within ~ time span of 20 years. For the other releves, no fire history was detected. It can be concluded that the present vegetation structures no longer aJlow the spread of fire. Fuel load·is very low;' and, if bunch grasses with dead material are present at aJI (in the case of mixed grasslands) they are spaced widely apart. · ·~

Short matted herblands are grazed .with an average intensity of 9.6 cowdt./50 m2• Even higher · !ntensities (30, 80 coW<ir./50 rn2) were recorded for the mixed grasslands. Trampling impact IS strong, and the formation of terracettes is especiaJiy pronounced on slopes. 'I

,. I

Poa spp. - Lachemilla orbiculata grassland

The elevation range is from 3700 to 3950 m, average elevation being 3887 ± 74 m. This type of grassland 1s common on lateral moraines and glaciated lava fields, with an average slope of 22.4 ± 10.0". Soils are well drained, which is reflected in the low avemge mottle score (0.8). The dominant vegetation struciUre type is short matted herbland (s). Two samples represent a tall _gra~sland (t) and_ one a short grassland (g). There is no evidence of fire. Grazing intenSHy IS htghly vanable, from 0 to 60 cowdr./50 m2• Cow paths, hoof-marks, and terracettes are common features. Incidental sheep grazing was recorded. These grasslands are located not far from the fincas and are used for rotational grazing. At the fincas of Potosi and El Bosque (not included in the vegetation map), the grasslands are generally fenced. Acaena ovalifolia and other undesired shrubs are often eradicated from the grasslands by the farmers a practice called 'desmatonar' (see also Verweij & Beukema, 1922).

Poa spp. - Lacltemilla orbiculata var. Rumex acetosella grassland n

=19

The floristic composition of the Poa spp. - Lachemilla orbiculata grasslands is described above. The Rumex acetosella variant differs in severaJ aspects. Bunch grasses are completely absent. Species still abundant in type H but absent here are: Cerastium subspicatum, Oreomyrrhis andicola, Sisyrinchium trinerve, Trisetum spicatum, and Lupinus microphyl/us. Diagnostic of this variant are Solanum tuberosum as a remnant of potato cultivation, the abundance of R1unex acetosel/a, and the presence of Lolium multiflorum, Sherardia arvensis, Viola sp., Tragopogon sp., and Halimolobos hispidula. Anthoxantum odoratum sometimes reaches hi~h cover va_lues, especially in those samples representing regeneration stages after potato culnvauon. Thts was aJso reported by Ferwerda (1987) for old field succession in the Cordillera Oriental.

n:: 10

The Poa spp. - Lachemilia orbiculata grasslands (types H and I) are characterized by the diagnostic species Poa prarensis, Poa annua, Plantago australis, and the dwarf shrub Acaend .ovalifolia. Lachemilla orbiculara, Anthoxantum odoratum, and Trifolium cf. amabile are also dominant. ' Characteristic are Carex pichinchensis, Taraxacum cf. officinale, Veronica serpyllifolia, · Cerastium arvense, Plantago linearis, Galium corymboswn, Nasella pubiflora, Hesperome/es hererophylla, and Trisetum irazuense. · This floristic type is considered a transition between the paramo proper bunch grasslands and the anthropogenic pastures. Bunch grasses are rare, and some typical bunch grassland species have completely disappeared, such as Hypericum lancioides, Baccharis caespitosa. Senecio formosus, Jamesonia goudotii, Senecio repens, Erigeron ecuadoriensis, Geranium multipartitum, and Castilleja fissifolia. Most species are still native paramo plants, but introduced species are aJso gaining importance. The appearance of a number of exotic weeds well-known in temperate pastures and agricultural fields (Ferwerda, 1987) is striking. Sown grasses such as Dacryiis glomera/a and Lolium perenne are occasionally present in type H, although they only become significant in type I.

Whether a ~oup_ of species-poor samples should be regarded as a separate floristic type is open .to discussiOn. These samples occur at the end of the gradient prepared with _TWI!'<SPAN. A few of_them resemble the Muhlenbergietum cleefii described by Cleef (1981), but would then _be ~n _Impoverished stage of these short grasslands. Given the present low nu~ber of spectes, It IS hard to dtaw any defi nite conclusions. For practicaJ reasons, it is dectded ~o treat them, together with the other heavily disturbed samples, as one provisionaJ community. Compared with the natural bunch grasslands, this type ~f grassland corresponds to the most degra~ed ~egetauon. It occurs in a broad elevation range, from 3600 to 4030 m. Average elev~uon IS 3830 ± 142 m. The terrain units to which this vegetation belongs include glaciated lava field_s, colluvial slopes, lacustrine flats and cirque bottoms, and occasionally lateraJ or end morames. The average slope is 12.6 ± 13.0". Drainage conditions are variable. The average mottle score is 1.1, but soils cultivated in the past tend to be more humid. The samples with a history of potato cultivation have a predominantly west aspect. Macrotopography could be an imponant factor. On the other hand, it can be reasoned that the sensitivity of potato crops to frost damage plays a role.

I. I·

65 64

Chapter 3

Characterization of vegetation pallerns

According to the fanners, rapid heating of the soil surface causes the aboveground to wilt, leading to frost damage, especially in the morning. This would make the slopes more suitable for cultivation. The following vegetation structure types occur: Tall forblands (f) and sparsely vegetated land (e) are both associated with grazing intensities of about 5 cowdr./50 m2• These samples comprise few attractive to grazers, and vegetation cover in general is poor. Rumex abundant in forblands regenerating from potato cultivation, is considered an species by the farmers. The sparsely vegetated land occurs near drinking fincas, a result of strong trampling impact. (n = 5; n = 2). Tall grasslands also used to be under potatoes, but are now under higher pressure (11.7 cowdr./50 m2 on average). (n = 9). 2 The shon matted herblands are intensively grazed (up to 3 cowdr. per m !) n•. '""'u"'' the in located are samples These taxa). 10 or (8 exrremely species-poor and cirque surrounding Laguna del Otun, showing closed mats of LUl:ne~IJUII i orbiculata with a few accompanying species. It is thus demonstrated that terrain Lachemil/a orbiculata..survives best under high grazing pressure. (n =

COIIDRUI

lt!:CROKIJI

>•U

~chaclc

"' •

•dwc~

, he iqhtCa!!u

..:'"'

bU< >140

1 • .d .

Terracettes are most pronounced in association with this floristic type. This is not to grazing activities, but often also a result of plowing to prepare the land for cultivation. After the harvest, sowing of grasses such as Lolium spp. is a common 1

3.4

£ro1lon \

Ordination analyses totTY81 I

Vegetation strocture A biplot of the RDA ordination is given in Figure 3.2. The clusters were drawn to insight into the variation of relevant variables. They were drawn around small .Samples along or just in between arrows of management variables, and some distance the origin. Table 3.3 presents the average values of the structure variables and. management variables of the distinguished clusters of Figure 3.2. Fire age and """~""""'" living tussock biomass·(% live) are best correlated to the first axis, whereas the cow droppings (In) and slope steepness are best correlated to the second axis. The output shows the following figures for eigenvalues and species-environment -.v11 "~'" """' 1st axis 2nd axis

A= 0.23 A= 0.14

R = 0.85 R = 0.77

The fi rst two axes account for 87% of the variance in the weighted averages of the with respect to the environmental variables. This percentage is calculated as in components analysis by taking 100 x (A 1 + A2)/(A. 1 + ... + AP) (Ter Braak, !987). In biplot of Figure 3.2, scores of structure variables, sample scores according to these variables, and biplot scores of environmental values are plotted. The first axis describes a decrease in tussock height and diameter and an increase in erosion (bare soil) are correlated to recent burning.

Figure 3.2

SLOPE . 1 '· 4 •

Biplot of the overdU structure analysis resulting from RDA ordination. Structure variables: total tussock cover, Calamagrostis effusa heigh!. C. ejfusa diameter, C. coarctata cover, Aciachne acicu/aris cover, Lachemilla orbiculata cover, and percentage cover of erosion spots. Management variables: regrowth of Espeleria (In) as fire age variable, cow droppings (In), trampling impact, percentage live tussock biomass, and slope. Scale of the diagram: I unit in the plot corresponds to 1 s.d. unit for the sample scores, 3.45 s.d. for the structure variables, and 0.29 s.d. for the management variables.

The estimates of fire age presented here differ from those reported by Verweij & Budde ( 1992), being based on the growth rates for Espeletia cited in the literature; i.e. 2 to 3 em per year instead of the exceptionally high growth rates reported in Chapter 5. The second axis represents the positive correlation between a decrease in tussock cover 'and an increase in matted herb cover on one hand, and a higher.number of cow droppings on the more gentle . · · · . slopes on the other.

Characterizatio n of vegetation palter/IS

Chapter 3

Variable

l (n=18)

Total tussock cover(%) Calamagrostis effusa height (em) Calamagrostis effusa diameter (em) Short grass cover(%) Matted forb cover (%) Lachemilla orbiculata cover (%) Erosion spots (%) Espeletia regrowth (em) Fire age, if burned (yr) Cow droppings (n/50 m1) Live tussock biomass(%) Slope (') Trampling impact (ordinal scale from 0 to 2) Table 3.3

63 i 14 44 ± 8 16 ± 5

29 i 8 18 ± 7 4± 2

3±2 20 ± 7 0± 0 7± 5

7± 3 19 ± 10 0i 0 30±9

59 5.8

1.2

I i

1st axis 2nd axis

U (n=20)

36 ± 13 28 ± 5 0.3 ± 0.4

R =0.88 R =0.87

The percentage of ·variance accounted for by the first two axes equals 84%. The environmental variables best correlated to the first axis are tramplin g and cow droppings (ln), which means that the first axis is a grazing gradient. The biplot of Figure 3.3 shows three clear groups close to the edges of the diagram . Corresponding value.s of structure and environmental variables are presented in Table 3.4.

12 • Ero•1on '

4± 4 73 ± 19

1 a.d.

25±7 0.9 ± 0.6 • totTuee

Average values and standard deviations of structure variables and maJnag.emi:Ji!c.l variables of clusters l-111 in the overall RDA structure analysis .

Cluster I (Figure 3.2) corresponds to samples with extensive tussock cover (63%) and values for tussock height and diameter. Matted herbs are not importa nt, erosion is low. trampling impact and the number of cow droppings are close to zero, and regrowth Espe/etia indicates that fire age is 5.8 years on average, or that no fire has occurred the lifetime of the present stem rosette population. This cluster was thus identified as the disturbed. Cluster II represents recently burned samples with an average regrowth of 12 which corresponds to a fire age of ± 1.2 years. It is therefore not surprising that values tussock cover, height, and diameter are low, or that a high percenta ge of green biomass is present (73%). The bare soil percentage is important (30%). Finally, cluster l1I is characterized by extensive cover of maned grasses and forbs, tussocks are almost absent, with a cover value of 5%. If tussock s are present, the size of bunches is rather large, which indicates that cattle apparently prefer to graze on the grasses or young bunch grasses. The number of cow droppin gs indicates a high pressure, and such samples occur on the more level terrains. The few stem rosettes I<:<u"""'"'~· possess tall dead leaf columns, corresponding to a long regeneration time since burning or np~ burning at all. The absence of sufficient fuel probably prevents these sites from being burned. Of the detailed RDA ordinations carried out separately for each fire age class, one exampl~ is presented. Figure 3.3 shows the ordination diagram of sites burned at least 2.7 years ago. (regrowth stem rosenes 40 to 300 em). As a consequence of this separate analysis, the sample number became smaller, but tendencies are nevertheless clear. Eigenvalues and species-: environment correlations are:

A.= 0.27 A. = 0.13

1 • . d.

t d1am.Ceffu

>d1

1 hoightC effu

1 •. d .

Figure 3.3

•

Lachorb l

calacoar

Biplot of an RDA structure analysis of samples that were not burned, or had a fue age of more than 2. 7 years. Structure variables: total tussock cover, Ca/ama grostis effusa height, C. (/[usa diameter, C. coarctata cover, Aciachne acicularis cover, l..achemil/a orbiculata cover, and erosion percentage. Management variable s: regrowth of Espeletia (In) as fire age variable, cow droppings (In), trampling impact, percentage live tussock biomass , and slope. Scale of the diagram : 1 unit i~ the plot corresponds to 1 s.d. unit for sample scores, 2.76 s.d. for the structure variables, and 0.39 s.d. for the management variables. ·

COW DROPPINGS

Chapter 3

Variable

I (n=8)

Total tussock cover (%) Ca/amagrostis ejfusa height (em) Ca/amagrostis effusa diameter (em) Shon grass cover (%) Maned foro cover (%) Lachemil/a orbiculata rover(%) Erosion spots (%) Espeleria regrowth (em) Fire age, if burned (yr) Cow droppings (n/50 m2) Live tussock biomass (%) Slope(") Trampling impact (ordinal scale from 0 to 2) Table 3.4

l l·

63 ± 10 50± 9 24 ± 6 4± I 22 ± 4 0± 0 4±4 145

-;:::~::~~~~:-T~:::~::-;~~=~- z.:~ -

II (n=S)

Myrrhidendron glaucescens Werneria crassa Cortaderia nitida Hieracium tolimense Nertera granadensis Galium corymbosum Jamesonia goudotii

21 ± 16 21 ± 12 7± 3 18 ± 10 35± 6 I± 1 20 ± 11

8.9

96 5.1

I± 2 33 ± 12

9± 9 58± 12

27 ± 11 0.1 ± 0.2

25 ± 13 1.6 ± 0.4

Poa subspicata Aphanactis piloselloides ; Ca/amagrostis coarctata ! l.Achemilla orbiculata Festuca andicota Geranium mullipartirum l Veronica serpylltfo/ia Aciachne acicularis Ranunculus peruvianus

1.5 1.5 1.3 1.3

1.3 1.3 1.2

1,.'

Cortaderia biftda Myrrhidendron glaucescens Werneria crassa Huperzia cruenta Jamesonia goudotii Lupinus iolimensis Lachemilla hispidula Hypericum lancioides Hieracium tolimense Bromus lanatus Castilleja [tSsijolia

_(~~~:;··

·-~;~~~;~~-~actio~~----···----·--......,~:;;;....-..

4.0 3.6 3.1 2.8 2.7 2.2 1.9 1.8 1.8

.;':, i

I ;

1.8 1.7

1 ! i.: :

1 .'

j L

a h'

1.'

;;

3.4

3.1 3.1 3.0 2.3 2.2 1.9 1.9 1.1 1.7

1.6 1.5 1.5 1.4 1.4 1.3

-~E~?=~n--~~~;~-------<~;- Aphanactis piloselloides Agrosris haenkeana Poa subspicara Hieracium tolimense Escallonia myrtilloides Hypericum lancioides Geranium sibbaldioides Lachemil/a orbiculata

Species composition

~~~":r~a/:~:na

Gentianella dasyantha Geranium multipartirum

Table 3.5

3.2 3. 1 2.9 2.9 2.4 2.2 2.1 2.0 1.9

~:~

1.6 1.5

.' ,~ :

I :, : ='!.!

Carex pichinchensis Disterignta empetrifolium Oritrophium peruvianum Arell(lria serpens Azore/la multiftda Nertera granadensis Lupinus tolimensis Cortaderia bifuia Gnaphalium antennarioides

2.4 2.0

· 1.8 1.8 I. 7

1.6

1.5

1.4 1.2

'.'':,:1

Results_ of th.ree CANOCO species analyses in which the first axis was kept constramed WJth one or two management variables. For each species, distance from the origin is listed in standard deviation units. 71

Arenaria serpens Poa annua Veronica serpyllifolia Lachemilla mandoniana Geranium columbianum Rumex acetosel/a Agrostis tolucensis Gamochaeta purpurea Agrostis haenkeana Arenaria sp. 1 Ranunculus praemarsus Luzu/a racemosa l.Achemil/a holosericea Bi<iens triplinervia LGupinus microphyllus entiana sedijo/ia

FIRE AGE and % LIVE

It can be concluded that grazing on slopes leads to a decrease in bunch-grass cover, ----------·"' and height, and simultaneously to an increase in the percentage of bare soil and tramplmg:J impact. On the valley bottoms and on flat terrain, grazing is related to the development closed mat of short grasses and ground-covering forbs.

Three separate analyses were performed in which only one variable determined the constrained axis. Three joint representations of species scores and biplot scores . environmental variables were prepared, analyzing response to number of cow droppings•. trampling impact, and Espe/etia regrowth combined with percentage of live tussock biomass. Eigenvalues of the first axes were 0.21 , 0.09, and 0.!3, respectively. The · species-environment correlations for each of these first axes were 0.78 (R for cow~,...,.., ...,.,. 0.83 (R for trampling impact), and 0.68 (R for Espe/etia regrowth combined with of live tussock biomass), respectively.

2.3 2.1 1.8 1.5 1.2

. . . . ._. ___,_. . -r.

TRAMPLING IMPACT ·--;~~·;~·~~-~~~~~~:

Average values and standard deviations of structure variables and mrunag,eme._~ variables of RDA ordination groups in a structure analysis of samples Espeleria regrowth of >40 em. This shows the influence of grazing on sites that not burned, or had a fire age of more than 2.7 years.

Cluster I is the relatively undisturbed situation on moderate slopes, similar to the ftrst of the overall ordination (Figure 3.2). If clusters II and III are compared, the effect of steepness becomes clear. Both groups are more intensively grazed, which on the slopes of cluster II clearly leads to a more pronounced trampling impact and more bare cover than on the gentle slopes and flat terrain of cluster III.

3.8 3.6 3.3 2.5

VEGETATION

Chapter 3

••'·'

Sparsely vegetate(,~ superparamo (e.g. Senecio /ariflorus - Calamagrostis Bunch grassland of Ca/andrinia acau/is- Calamagrostis recta, upper paramo Bunch grassland of Cafandrinia acaulis - Calamagrostis recta, upper paramo Cushion bog and shrubland of Gemiane/la dasyantha - Plantago rigida Idem, but in combination with a .mixed grassland of Aciaclme acicu/,ario

~r ·

Calamagrostis coarctata

Dense bunch grassland of Espeleria hartwegiana - Calamagrostis effusa Dense bunch grassland of .Ca/amagrosris effu.sa - C. recta (var. typicum) D¢nse bunch grassland of Ca/amagrostis ejfusa - C. recta (var.

mynilloides)

:\!edium-dense bunch grassland of Espeletia harrwegiana- Calamagrostis effusa

grassland :\!edium-dense bunch grassland of Calamagrostis effusa - C. recta (var. :\ledium-dense bunch grassland of Ca{amagrostis effusa - C. recta (var.

myrtilloides)-

$:-:

s: e!

t.~':l

a= aF W.

, Open bunch grassland of Espe/etia hartwegiana - Calamagrostis ejfusa Open bunch grassland of Calamagrostis effusa - C. recta (var. typicum) Open bunch grassland of Calamagrostis effusa- C. recta (var. Escal/onia my.ruuu1~rm1 :\!ixed grassland of Espeletia hartwegiana - Calamagrostis effusa :\lixed grassland of Calamagrostis e/fifSa - C. recta (var. rypicum) :\lixed grassland of CalamagroStis effusa - C. recta (var. Escallonia m•••rtill'nir,rpr) :\lixed grassland of Aciachne acicularis - Calamagrostis coarctata :\!ixed grassland of Agrostis haenkeana - Lachemilla orbiculata Short grassland of Aciachne acicularis :- Calamagrostis coarctata Short maned herbland of Calamagrostis effitSa- C. recta (var. Escal/onia mYirtil/oidi!S)1JI Short maned herbland of Agrostis haenkeana - Lachemil/a orbiculata Short matted herbland of P_pa spp. - Lachemilla orbiculaca Short maued herbland of Poa spp. - Lachemil/a orbicu/aca (var. Rumex acl!tosella)ll Sparsely vegetated land of Poa spp. - Lachemilla orbiculaca (var. Rumex ace!tos,efla)~l Tall grassland of Poa spp. - Lachemilla orbiculata and forbland (id., var. acetosella) Shrubland of Pencacalia vernicosa Shrubby grassland of Calamagrostis e/fitSa - c. recta (var. Escal/onia mvrll/ilflMII'.n'~-11 Shrubby grassland of Aciachne acicularis - Calamagrostis coarctata

Water body, lake

Legend of the vegetation map of 'the upper watershed of Rfo Otun, Los National Park, Colombia

)'_

3km

CJ D D

• •

s UP2

[}TI

dC dO dE hC hD

•

D D

B B/mF

G D

me mD ~.., mE

UP1

Ill

rz:.-..:c 3.6

hE oC

D mG D .gF

~ ~. ~.

_;j,;

sE sG sH sl el tH/fl

a aE aF

Fig~re 3.4 VEG ETATION MAP Qf the upper watershed of Rio Otun, Los Nevados National Pa1 aenaf photographs of 1983, 198!;i, and1989 (scales 1: 25,000 to 1: 30,000) and field data of 1! overlaid.

7'5

VEGETATION

D D D

--g

D D D

,.. D D D

s UP1 UP2 B

8/m F dC dD dE hC hD hE oC oD oE

â&#x20AC;¢

3km

me mD ' mE mF mG D . gF ~ sE sG D sH D sl D el D tH/fl -a ~ aE aF

5J D CJ

CZJ

Figu re 3.4 VEG ETA TION MAP of aeri al phot ogra phs of 1983 , 1985 the uppe r wate rshe d of Rio Otun , Los Nev ados Nati onal Pari<, Co lomb , and1 989 (sca les 1: 25,0 00 to 1: 30,0 00) and field data of 1989. Terr ia. Base d on ain segm ents a re over laid.

Characterization of vegetation patterns Species located at a distance of at least 1.2 units of standard deviation from the origin (distance as projected on the first axis) were regarded as showing a positive or negative reaction to the management variables concerned. Results are summarized in Table 3.5, listing the species that show either a positive or negative reaction to grazing intensity, trampling impact, and fire age. Some species, such as Corraderia bifida, Jamesonia goudotii, Myrrhidendron g!attcescens, and Âˇ. Werneria crassa demonstrate a negative response to both trampling impact and the number of cow droppings as measures of grazing intensity. There is a group of species consisting of Lachemil/a orbiculata, Taraxacum cf. officinale, Festaca andicola, Poa subspicata, and Aphanactis pilosel/oides that show a positive reaction to grazing intensity, but a neutral or negative reaction to trampling impact. Arenaria serpyllifolia and Poa annua are two examples of species reacting positively to trampling impact only.

Carex tristicha, Lupinus tolimensis, and Corraderia bifida are species related to recent burning. These species are also related to more humid conditions, as indicated by earlier TWINSPAN and CANOCO analyses. Aciachne acicularis only occurs at unburned sites or at places that were burned a long time. ago. Woody species showing a positive reaction to a long regeneration time since burning are Escallonia myrtilloides and Hypericum /anciodes. Other shrub species such as Hypericum laricifolium, Pentacalia vaccinioides, and Pernetrya prostrara were found to be related to the average fire age.

3.5

Vegetation map

The aim of the vegetation map was to represent vegetation units that are homogenous as regards their combination of floristic composition and structure. The relationships between structure and floristics are described above (Section 3.3). They are summarized in the cross matrix of Table 3.7. The clear diagonal structure of this matrix indicates an important trend: the floristic gradient in the direction of more intervened or degraded vegetations corresponds to a process of degradation in the vegetation structure; i.e. the opening up of the bunch grasslands and their subsequent replacement by other structure types. To define the content of each land cover unit, point data were extrapolated using different knowledge-based rules on correlations derived from the classification of the releve data. Knowledge-based rules were formulated for the following kinds of (spatial) relationships: (1)

(, Colombia. Based on 89. Terrain segments are

The relationship between image characteristics of the aerial photographs and vegetation attributes, using stereo height: for the distinction of stem rosettes, shrubs, and dwarf trees; tone: black spots for recent burning (patches of open bunch grassland with presence of ashes), dark grey for cushion bogs, medium grey for bunch grasslands, light grey for short matted herblands, white for short grasslands, etc.;

r==

. Cl)

CI)CI)Ol

(")NCO

COO CO N\O<o'

ll)

0\C0\00

<J)

Q('t')<qllf)

.0.0"'"' .0

co N

f"'-COlf')C"")

('f")\OLOO\

.0.0"'

Cl)

.... ....0)

ON M<r

(2) (3)

<J)

.O.OOl Ill

(") ...

N.-<\0 0\ 0 ....... .0N CI)CI)(J) CI)CI) "' "'"'"'

If) Cl)

11'1 M

Characterization of vegetation patterns

\l) Cl)

"'.0 CI)CI) "' CI)CI)CI) "'"' "'"'"'

If) Cl)

,....

O<r.-< \ll\Or-

If)

.Qfll

Cl)

r""''-'> lf'l.-l

NU"''OCJ'\

.O.QfiiCI)

(4) Oh--tll)N O <r

r<N

lf'lO'\<q~f'

co.-<r- N0\00\ M O.-< N M('l")('l')f""- .-<

CI)O)CI)CI) 1/)CI)

f""iN<Q"If)

r..

.QCI)(J)CI) (ll(llCI)

......... . NM

..,..-< oco

CI)CI)

.Q<Il<ll

Mlfl ON

N .-<

.-<<o'

II) I/)

.Ofll

MM\0 .-<MOO

.QCilfll

.-<N .-<CO

ON riN

Q("')q>f'""'i

U) "'

"'"" .,.N Ill

.OIIl

.QCI)

Nlfl Mlfl

.....

t:l

qOf""-CO

C""-\Oqtt""-

,_.or--r--

co M

"' "'

Homogeneous units representing a single vegetation community were delineated wherever possible, accepting a level of impurity up to ± 10%. Sometimes, complex units had to be defined, consisting of two individually unmappable elements. The final vegetation map (pixel size 10 * 10 m) contains 30 different units, and is presented in Figure 3.4 and Table 3.6 (correspond ing legend) on pages 72 and 73. The scale is reduced only for reasons of presentation .

.0.01/lCil

01/)tn\OM

Mlfl (")\0

Ill

.0.0

WO~r-1\0

MrtMCON

r-\ON r-ft/')(\IC''Q'(t"') OM\0 ONU"ltn\Oqo O<r

CX)f'lf')O

.OCil

,Q.Q.Q.Q

(I) (/)(1)

CX)'O' riO

Nl'"-

riM ,Q.Q

OMU")LOO

II'IM 0\0

.0.0.0.0"'

.0.0

"""'

.......

QU"')OCJ)

""''CT'NC"'-

... 0\

.O.OOlfll

N(t') r'\0

"'"'

.O.O.Ofll

0\ ....

.0 11)11)11) .Q.Q

N'Cl"qof""-

Ill

.-<Ill CO "'"'"' CI)II)Ul

<X)

co Ill

·c:"' 0 c:

....

r-moco

.-<M0\0\

11'1

(I) (I) (f) (f)

0\00'\1.0

.-<

NNCOCO

11'10\ MO\

"'.0

Ill .0 IJl!JlCI)

·c: co

CI) CI)

.....

0 .-<

~:;; ~-.,

Ill

.1:

3.6

"'.0

.0.0.0 .0.0 .0.0.0 Ill .0.0

.0 !Jll/)1/l .0

If) I()

0\0\

.0.0

.Q.Q

1/)CI)CI)

qoor....... co

"'"' .-<.-< .01/l 0\ll

.... M

<rNM r<Mr-

OM~r"-

texture: coarse for shrublands, medium coarse for bunch grasslands, fine textured for mixed grasslands, smooth for short maued vegetations; linear features: for terraccnes related to more open bunch grasslands on slopes; pattern: field pauem in case of regeneration stages after potato cultivation and burning pattern. The relationship between structure and floris.tics as shown in the diagram ofTable 3.7. The restriction of floristic types at macro-scale to certain geographical areas as determined by a combination of elevation, grain size of volcanic ashes, major terrain units, and macro-topography: e.g. the occurrence of the Espeletia hartwegiana Calamagrostis effusa vegetation (type C) in the upper paramo proper, on coarsegrained ashes with poorly developed humic andosols. Type H occurs mainly outside the area mapped, in the regions of El Bosque and Potos!. Other geographical relationships are documented in the description of the f101;istic types. . The relationship between vegetation communities and environmental variables at .meso-scale. An example is the occurrence of shrublands along drainages, on scree slopes, and in the transition from bunch grassland to cushion bogs. Other established associations are between cushion bogs and the concave parts of the terrain (valley bottoms, glacial depressiO!l$), between short matted vegetations and flat parts or gentle slopes, and between ·bunch grasslands and slopes.

Ill

Conclusio ns

For the characterization of vegetation patterns, a logical sequence of techniques was used. TWINSPAN ·analyses were instrumental in organizing data on both vegetation structure and floristic composition in a coherent way. The ecological relationships of the main vegetation communities have been evaluated in this chapter. This is an indispensable· basis for the analysis of variance due to human impact. For each community, the relationship with burning. and grazing variables has been quantified. The floristic clusters resulting from 1WINSPAN analysis reflect a combined elevation and human impact gradient. Information on floris tic composition facilitates the reconstruction of vegetation development-from intact communities into secondary types, as certain species remain present. In this way, it is concluded that floristic types F and G are derived from types E and D, respectively. Multivariate ordination analyses (CANOCO) helped to further unravel the interrelationships and tendencies. Variance due to elevation, excessive humidity, and potato cultivation was eliminated by an iterative filtering process. The ordinations are useful in confirming the relationship of the floristic 1WINSPAN gradient to certain degradation processes, with more quantitative evidence of the separate contributions by burning and grazing variables. 77

Chapter 3 However, ordinations do not prove direct causal relationships. In this sense, the results need to be interpreted with the utmost care. The c~oss ma~~ of vegetation structure versus floristics is another useful tool. It shows ~otent.tal transition~ between vegetation types. The direction of change is determined by

mtenmy and durauon of the management regime. If the information in the cross combined with map data on the geographical areas to which the communities are it becomes clear what kind of transitions are realistic. Transitions among types dC, hC, and mC are common. The same holds ITUe for transitions between any other pair of typ~s .corresponding to the same floristic type. The other type of transition, i.e. flonsttc types, generally requires more time. Examples are the possible transitions from F, and f~om D to G. Transitions among types A, B, C, D, and E are unlikely. The de~adauon stage under high grazing pressure is type I (el for drier sites, sl for moist whtch seems almost irreversible. Although floristic types H and I are usualiy products gradual degradation processes, they can also find their origin in the disruptive event of · cultivation.

II I

i '

Characterization of vegetation patterns :'.-·' >This is the first time that the influence of management variables on paramo vegetation has · been quantified separately from natural variance. Ramsey (1993) also used multivariate _analysis techniques for paralllo vegetation in Ecuador, as did Witte (1994) for paramo ,·, vegetation in Colombia. Both authors succeeded in the characterization of the main factors detemnining the natural variation in the paramo vegetation studied. However, data on human ."· influence were still scarce or lacking, which impeded conclusions in this respect. For an ·· .appropriate evaluation of human impact on paromo vegetation, it is fundartJental that , , , management factors are quantified as fully as possible. The present work provides basic r,· elements for such a study in general. ·

+ An aspect requiring further attention is the rate of the observed changes. For modelling purposes, the key variables that are assumed to steer the replacement of one community by : •: another can be selected on the basis of the ordinations described. The speed of these processes requires a more detailed an;~lysis. A start was made with the development of a transition .model for a small pilot area (Pels and Verweij, 1992). A first continuous simulation model describing the influence of burning and grazing on paramo vegetation is presentect in Chapter , 6. Based on the present chapter, the building blocks for modelling are identified: plant groups with a similar behaviour with respect to ecological and management variables. Examples of -these so·called functional plant groups include bunch grasses, stem rosettes, shrubs, tall forbs, short grasses, and matted forbs.

The gen~ral tendency can be summarized as fQilows: with an increasing impact of and grazmg, the structure of the bunch grasslands becomes more open. Fire events te'?poral ch~ges in vegetation structure over the relatively short time span of a few Wtthout the mfluence of grazing, the vegetation is able to recover and returns to the structure of dense bunch grassland, the floristic composition remaining largely unc:hatnged~tl Under prolon~ed grazing pressure, however, short matted grasses and forbs increase ~adual transformation into other communities occurs. Changes in slnlcture are of trnp~rtance in relation to the water holding capacity of both vegetation and soil. Hofstede . -LITERATURE CITED s.evm.k (1995) concluded that water retention capacity is reduced in grazed and sltua~ons, with an increased risk of drying especially in the drier seasons. Acosta-SoHs, M.A.S. 1984. Los paramos andinos del Ecuador. Publicaciones Cientificas Processes on slopes differ from those in flat areas. Grazing on slopes leads to a decrease in M.A.S., Editorial ENA, Quito, 222 pp. bu.nch-grass cover, diameter, and height, and to a parallel increase in the percentage of Allaby, M. (ed.) 1994. The concise Oxford dictionary of ecology. Oxford University Press, ~otl and trannpling impact. This was also found by Hofstede (1995), who analyzed the changes Oxford. 415 pp. m Slnlcture .and biomass in detail. The chance of transition from one floristic type to another Beck, E., Scheibe, R. & Schulze, E.-D. 1986. Recovery from fire: observations in the alpine seems less iikely on slopes than on more level terrain. On the valley bottoms and on flat parts vegetation of western Mt. Kilimanjaro (Tanzania). Phytocoenologia 14(1): 55-77. or gentle slopes, higher grazing intensities were recorded, which are associated with the Cleef, A.M:. 1981. The vegetation of the paramos of the Colombian Cordillera Oriental. development of a closed mat of short grasses and ground-covering forbs. Dissertationes Botanicae 61. J. Cramer, Vaduz, 320 pp. Cleef, A.M., Rangel, 0. & Salamanca, S. 1983. Reconocim iento de Ia vegetaci6n en Ia parte Another ~nteresting aspect is the influence of fire frequency on vegetation dynamics. As alta del Transecto Parque Los Nevados. Pp. 150-173 in: Vander Hammen, T., Pinto, repo:red m the present study, woody species require a long regeneration period, if they P. & Perez, A. (eds.), Studies on Tropical Andean Ecosystems I. J. Cramer, Vaduz. sumve a.t all. This was also documented by Janzen (1973) and Hom (1989) for post-fire Cuanecasas, J. 1958. Aspectos de Ia vegetaci6n natural de Colombia. Revista de Ia Academia reg~nerauon o~ paramo vegetation in Costa Rica, and by Beck et al. (1986) for _the Colombiana de Ciencias Exactas, Flsicas y Naturales 10(40): 221-264. tropical alpme vegetauon of Mount Kilimanjaro. In a study of the mortality and regeneration of · Cuanecasas, J. 1968... PcU:amo vegetation and its life forms. 1n: Geoecology of the woody perennials in a Costa Rican paramo vegetation, Williamson el al. (1986) concluded mountainous regions of the tropical Americas. Colloquium Geographicum 9: 163-186. that frequent fires may maintain a dominance by graminoids (mainly bamboo) at the expense. Ferwerda, W. 1987. The influence of potato cultivation on the natural bunchgrass paramo in of woody .species. In the case of the paramo of Los Nevados, it seems likely that ftre the Colombian Cordillera Oriental. MSc. thesis, Internal Report no. 220, Hugo de f~que?cy mfluences the interaction among woody species, stem rosettes, and tussock grasses. Vries-Laboratory, University of Amsterdam. Frre .htst~ry is analyzed in more detail in Chapter 8. Some effects of management on plant Gary, M., McAfee, R. & Wolf, C.L. (eds.) 1972. Glossary of geology. Amefican Geological spec1es dtversity are discussed in Chapter 9. Institute, Washington D.C., 857 pp. . Godley, E.G. 1978. Cushion bogs. Erdwissenschaftliche Forschung 11: 141-148. 78 79

Chapter 3 Guue, P. 1985. Beirrag zur Kenntnis zentralperuanischer Pflanzengesellschaften II. hochandinen Moore und ihre Kontaktgesellschaften. Feddes Repertorium 91 (5-6): 336. Hill, ·M.O. 1979. TWINS PAN - a Fortran program for arranging multivariate data in ordered two-way table by classification of individuals and attributes. University Press, Ithaca NY. Hofstede, R.G.M. 1995. Effects of burning and grazing on a Colombian paramo ecclsysterrl'<ll!!l. PhD dissertation, University of Amsterdam. 199 pp. Hofstede, R.G.M. & Sevink, J. 1995. Water and nutrient storage and input:output budgets burned, grazed and undisturbed paramo grasslands. Pp. 121·147 in: Hofstede, ·~·'"'·••t;,...,:1 .•,1 Effects of burning and grazing on a Colombian paramo ecosystem. PhD <lts!;ert!ttiotlmt!l University of Amsterdam. Hom, S.P. 1989. Postfire vegetation development in the Costa Rican paramos. MaOro,fi61lilU 36(2): 93-114. Howard, J.K. & Higgins, C.H. 1987. Dimensions of grazing-step terraceues and significance. Pp. 545-567 in: Gardiner, V. (ed.), International geomorphology 1986: proceedings of the lst international COIJ..ference on geomorphology, part II. John Wiley~ Chichester. Janzen, D.H. 1973. Rate ofregeneration after a tropical high elevation fire. Biotropica 5: 1 122. Jongman, R.H.G., Ter Braak, C.J.F. & Van Tongeren, O.F.R. 1987. Data analysis com_munity and landscape ecology. Pudoc, Wageningen, 299 pp. Kent, M. & Coker, P. 1994. Vegetation description and analysis, a practical approach. John · Wiley, Chichester. 363 pp. Kloosterman, E.H., Cleef, A.M. & Salamanca, S. 1995. Vegetation map of the Parqu(: Nacional Natural Los Nevados (Central Cordillera, Colombia). in: Vander Hammen, T. & Dos Santos, A.G. (eds.), Studies on Tropical Andean Ecosystems 5. J. Cramer, Berlin. Kok, K., Verweij, P.A. & Beukema, H. 1995. Effects of cutting and grazing on Andean forest . line vegetation. In: Churchill, S.P., Balslev, H., Forero, E. & Luteyn, J.L. (eds.), Biodiversity and conservation of neotropical montane forests. Scientific Pub!., The New York Botanical Garden. Kuchler, A.W. & Zonneveld, I.S. 1988. Vegetation mapping. Handbook of vegetation science, vol. 10. Kluwer Academic Publishers, Dordrecht. 635 pp. Kuhry, P., Salomons, J.B., Riezebos, P.A. & Vander Hammen, T. 1983. Paleoecologia de los ultimos 6000 aiios en el area de Ia Laguna de Orun- El Bosque. Pp. 227-261 in: Van . der Hammen, T., Pinto, P. & Perez, A. (eds.), Studies on Tropical Andean Ecosystems !. J. Cr.uner, Vaduz. Melief, A.B.M. 1985. Late Quaternary paleoecology of the Parque Nacional Natural los Nevados (Cordillera Central), and Sumapaz (Cordillera Oriental) areas, Colombia. Pp. 1-162 in: Vander Hammen, T. (ed.), The Quaternary of Colombia 12. University of Amsterdam. Monasterio, !11. 1980. Las fonnaciones vegetales de los paramos de Venezuela. Pp. 94-158 in: Monasterio, M. (ed.), Estudios ecol6gicos en los paramos andinos. Ediciones de Ia Uni\'ersidad de los Andes, Merida. · Mueller-Dombois, D. & Ellenberg, H. 1974. Aims and methods of vegetation ecology. John Wiley, New York, 547 pp.

Characterization of vegetation patterns ·Pels, B. & Verweij, P.A., 1992. Burning and grazing in a Colombian bunchgrass paramoecosystem: a tmnsition model. Pp. 243·263 in: Balslev, H. & Luteyn, J.L. (eds.), Paramo: an Ande~n ecosystem under human influence. Academic Press, LondQn. · Ramsey, P.M. 1993. The paramo vegetation of Ecuador: the community ecology, dynamics and productivity of tropical grasslands in the·Andes. PhD dissertation, University of Wales, 274 pp. Rangel, 0., Dfaz, S., Jaramillo, R. & Salamanca, S. 1983. Lista del material herborizado en el Transecto del Parque Los Nevados (Pteridophyta-Spennatophyta). Pp. 174-205 in: Van der Hammen, T., Pinto, P. & Perez, A. (eds.), Studies on Tropical Andean Ecosystems I. J. Cramer, Vaduz. Salamanca, S. 1991. The vegetation of the paramo and its dynamics in the volcanic massif Ruiz - Tolima (Cordillera Central, Colombia). PhD dissertation, University of Amsterdam, 122 pp. Salamanca, S., Cleef, A..M. & Rangel, J.O. 1995 . The paramo vegetation of the Ruiz- Tolima massif. In: Van der Hammen, T. & Dos Santos, A.G. (eds.), Studies on Tropical Andean Ecosystems 5. J. Cramer, Berlin. · Salomons, J.B. 1986. Paleoecology of volcanic soils in the Colombian Central Cordillera (Parque Nacional Natural de los Nevados). Dissertationes Botanicae 95. J. Cramer, Berlin, 212 pp. Seibert, P. & Menhofer, X. 1991. Die Vegetation des Wohngebietes der Kallawaya und des Hochlandes von Ulla-Ulla in den bolivianischen Anden. Phytocoenologia 20(2): 145276. Selby, M.J. 1993. Hillslope materials and processes (2nd ed.). Oxford University Press, Oxford, 451 pp. Smith, J.M.B. 1977. Vegetation and microclimate of east- and west-facing slopes in the grasslands of Mt Wilhelm, ·Papua New Guinea. Journal of Ecology 65: 39-53. Ter Braak, C.J.F. 1987a. CANONO - a FORTRAN program for canonical community ordination by [partial] [detrented] [canonical] correspondence analysis, principal components analysis and redundancy analysis (version 2.1) ITI-TNO, Wageningen, 95 . pp. Ter Braak, C.J.F. 1987b. The analysis of vegetation-environment relationships by canonical correspondence analysis. Vegetatio 69: 69-77. Thouret, lC. 1983. Observaciones geopedol6gicas a lo largo del Transecto Parque los Nevados. Pp. 113·141 in: Vander Hammen, T., Pinto, P. & Perez, A. (eds.), Studies on Tropical Andean Ecosystems 1. J. Cramer, Vaduz. _ Thouret, J.C. 1989. Suelos de Ia Cordillera Central, Transecto Parque Los Nevados. Pp. 293441 in: Van der Hammen, T., Diaz, S. & Alvarez, V.J. (eds.), Studies on Tropical Andean Ecosystems 3. J. Cramer, Berlin. Van der Hammen, T. 1974. The Pleistocene changes of vegetation and climate in tropical . South America. Journal of Biogeography 1: 3-26. Van der Hammen, T. 1979. History of the flora, vegetation and climate in the Colombian Cordillera Oriental during the last five million years. Pp. 25-34 in: Larsen, K. & Holm-Nielsen, LB. (eds.), Tropical botany. Academic Press, London. Vareschi, V. 1953. Sabre las superficies de asimilaci6n de sociedades vegetales de cordilleras tropicales y extratropicales. Boletfn de Ia Sociedad Venezolana de Ciencias Naturales 14: 121-173. Caracas.

Chapter 3 Verweij, P.A. & Beukema, H. 1992. Aspects of human influence on upper-Andean forest vegetation. Pp. 171- 175 in: Balslev, H. & Luteyn, J.L. (eds.), Paramo: an ecosystem under human influence. Acad~:mic Press, London. Verweij, P.A. & Budde, P.E. 1992. Burning and grazing gradients in the paramo of Los Nevados, Colombia: initial ordination analyses. Pp. 177-195 in: Balslev, H. & Luteyn, J.L. (eds.), Paramo: an Andean ecosystem under human influence. Academic · Press, London. Verweij, P.A, & Kok, K. 1992. Effects of ftre and grazing on Espeletia populations in the paramo of Parque Los Nevados, Colombia. Pp. 216-229 in: H. & Luteyn, J.L. (eds.), P<iramo: an Andean ecosystem under human mu"""""'· Academic Press, London. · Williamson, G.B., Schatz, G.E., Avlarado, A., Redhead, C.S., Starn, A.C. & Sterner, R. 1986. Effects of repeated fires on tropical paramo vegetation. Tropical Ecology 27: 62-69. Witte, H.J.L. 1994. Present and past vegetation and climate in the Northern Andes (Cordillera Central, Colombia): a quantitative approach. PhD dissertation, University of Amsterdam, 269 pp. Zonneveld, I.S. 1979. Land evaluation and land(scape) science. lTC textbook of photo intel]lretation .VII(4), 2nd ed. ITC, Enschede, 134 pp. Zonneveld, I.S. 1995. Land ecology, an introduction to landscape ecology as a base for land evaluation, land management and coll5ervation. SPB Academic Publishing, Amsterdam, 199 pp. Zonneveld, I.S., Van Gils, H.A.M.J. & Thalen, D.C.P. 1979. Aspects of the 1TC approach to vegetation survey. Documents Phytosociologiques IV, pp. 1029-1063. Lille, France.

EXTENSIVE LIVESTOCK PRODUCTION IN THE PARAMO

Pita A. Verweij and Anne M. Schmidt 4.1 4.2 ~.3

.9-4 4.5

Introduction to paramo range management Estimation methods of forage intake Grazing behaviour Quality and quantity of forage Conclusions

· , The objective of the present chapter is to evaluate grazing managemen t from the perspeclive of the livestock producers and livestock of Los Nevados National Park. Analysis of grazing management in this case includes the study of the grazing behaviour of canle. It is the first time that the forage consumption and the production of cattle have been quantified for a paramo ecosystem. The feed intake of cattle in terms of quantity and diet composition, and secondary production were estimated. The study of grazing behaviour provides important · ·. 'elements for the construction of the spatial model of cattle distribution presented in Chapter 1. The effects of management on vegetation and micro-relief are better understood when management is ~analyzed independen tly of the impacts on the ecosystem.

4.1

Introduction to paramo range management

In many Andean paramo ecosystems, grazing by domestic animals such as cows, mules, horses, and sheep (Pastrana et al., 1991a, 199lb) plays an important role in determining vegetation structure and composition. Few estimates of the consumption and secondary production by cattle at high elevations are known from the literature. In Colombia, extensive livestock production systems have been studied predominantly in lowland areas and lower mountain belts up to elevations of about 3200 m (Diaz, 1985; Koeslag, 1985). At these elevations, cultivation of introduced grass species of higher forage quality is possible, whereas in the extreme climatic conditions of the paramo, this is not feasible judging by farmers' experiments in the study area. Grasses are sown only after potato cultivation; these activities are, however, mainly restricted to farms outside the mapped area. The diet of the grazing animals of the paramos therefore consists of a selection of species from the (semi)natural vegetation. The characteristics ·o f the vegetations ' concerned and an overall grazing gradient are described by Verweij & Budde (1992) and in Chapter 3. Perennial bunch grasses are an important component of this vegetation, and a major ingredient in the feed of ruminants.

Part of this chapter has been published as Schmidt, AM. & Verweij, P.A. 1992. Forage intake and secondary production in extensive livestock systems in paramo. Paramo: an Andean ecosystem under human innuence. © Academic Press, London. 82

Chap1er 4

E.nensive livestock production

The bunch grasses, mostly of the genera Calamagroslis and Feswca, are known to be low in nitrogen content (Beekman & Verweij, 1987; Hofstede, 1995). In view of these funrlannP.n:ril~ differences in forage quality, extreme care has to be taken when extrapolating data from and more intensively managed grazing systems to the traditional livestock system of the paramo.

Th' brings about heart problems, especially when th.: animals already suffer from ma:~utrition, diarrhoea, or infection caused by endopar~sites. T~c on ly way to cur~ animals suffering from papcra is to move them in time to lower elevations. The second tmpona~t of death is accidents resulting from the rough topography, e.g. when cattle get stuck 111 :~i~ages or deep block-lava holes. More data on aspects of animal husbandry are found in Schmidt (1992).

The following description of range management is based on interviews with farme~. cattle are crossbreeds between Normando and Red Poll. Both are dual-purpose breeds, · means the cattle are raised for both milk and meat production. The emphasis is, however, beef production. As Normando bulls are always crossed with the mixed breed, the possesses mainly Normando characteristics; this facilitates comparisons with data in literature. Range management involves decisions on the grazing system, grazing management, husbandry. control of woody pla!'lts, and burning. The paramo grazing syst~m is chal!actenzed by continuous grazing. There al!e, however, some characteristics of deferred grazing in a remote areas, where the vegetation is allowed a period of rest to regenerate and to acc:unlul:ate: fuel. The animals are broughi into these areas in times of forage scarcity. Othenvise, roam freely over e~tensive al!eas controlled by individual f3i!mers.

....

The main variable of grazing· management is the stocking rate, defined as the actual number_ of animals on a specified area at a specific time (FAO, 1991). In order to avoid damage, stocking rates should be determined by the grazing capacity of the rangeland. Grazing capacity is defined by Breman & Ridder (1 991) as the number of animals that can be fed per surface unit in order to reach a cenain production level while preserving the production capacity of the pastures. Stocking rates applied by the farmers vary from 0.04 to 0.15 · animals/ha (see Chapter 7). According to the f3i!mers, during long periods of drought forage becomes scarce. The present stocking rates are probably not far above or below the overall grazing capacity. Overgrazing is observed in some areas, especially close to the fincas and ~ some water points. The issue of grazing capacity is further addressed in the final chapter. Animal husbandry includes administration, milking and reproduction practices, and the control and treatment of diseast:s. Written accounts or production figures are seldom kept by the farme!'$. although the cheese production is sometimes registered. In order to control the herds, the cattle are counted every day. The fa1!n1ers go through the grazing areas and round up the , · cattle. The lactating cows and their calves are locked up in separate enclosures, every '.. afternoon between 3 and 4 p.m. This ensures that the calves do not comsume the entire milk production ..Milking is done by hand once a day between 6 and 9 a.m. About a qumer of the milk production is left for the calves, the rest is for own consumption or processed into cheese. Every farmer has one or two pure Normando bulls for reproduction purposes. The bulls are kept away from the heifers. The only reproduction control takes the fo1111 of castrating young bulls at a maximum age of 15 months. At the age of two to three years, the bulls arc sold, as are non-productive or infertile cows. The average life span of the cows is estimated at 10 to 12 years. The principal natural cause of mortality is known by the farmers as 'papera' or 'mal de altura', which means 'elevation disease'. The main problem is the stress attributable to the high-mountain environment, with its low oxygen pressure. X4

The prevailing careta.ker syst~m has many implic~tions ~o.r manageme.~t. ~e absentee. owner lives far away from his herd and therefore cruc1al deciSions ~n vaccmauon and selling al!e often delayed, resulting in the spread of diseases, and lower pnces (F~O,_ I 991). The owners h0 live in the inter-Andean valleys use their herds mainly as capital mvestment and al!e · w~ n not interested in sustainable rangeland management. The level of material inputs is low, O•te . I' . d A th . · the amount of purchased dmgs, lickstones, feed concentrate, etc., bemg 1m11e . s e mam income is generated by milk production · sometimes supplemented by pay~e~ts m cas_h or food, or a small percentage of the offspring - the caretaker. fanners _try to l11~1t expen~uure on material inputs to a minimum. Milking and cheese-makmg pracuces requtre a relattvely high labour input. Woody species, especially Acaena ova/ifo/ia, are eradicated manually in the more intensive~y grazed pastures close to the fincas. The bunch-grass vegetation is burned e~ery fe w years m order to stimulate the growth of young grass shoots of higher fora ge qualny. ~ead tussoc_k leaves hinderlhe arowth of new leaves as less light penetmtes the canopy. A detatled analysts of the use of fire"'as a management tool and a reconstruction of fire history for part of the study area are prese nted in Chapter 8.

4.2

Estimation methods of forage intake

In the . heterogeneous p3i!amo bunch grasslands, the composition of the animal diet as ' influenced by selection is an important pammeter in the estimation o_f the amount of forage consumed ·(Mannetje, 1974). The ratio between the energy (consrdered to be eq~al to digestible dry matter) and the protein content of the feed determine~ to what e~tent.t~~ mtake requirements can be fulfilled (Breman & De Ridder, 1991). I~ thts study, ?•g~sublltt~ and crude protein or nitrogen content are used to evaluate the.qualtty of cattle di~t m relauon to total intake. Differences in the forage quality of plant species are related to ammal preferen_ce (Van Dyne et a/., 1980). Besides forage quality, selection_ is conditioned by the spaual distribution and relative abundance of the preferred fracuon of the forage (Arnold & Dudzinski, 1978). Total forage intake was estimated according to three different methods: • field observations on bite counts per time unit; • measurement of the ·daily fecal excretion; and • indirect estimations based on requirements for growth, milk production, and reproduction. The grazing model in Figure 4. 1 shows the different ways of al!riving at an intake assessment.

., Chap1er 4

Extensive livestock produclion

Grazing behaviour Free-ranging animals were observed in order to note their grazing and resting rhythm, and relative abundance of differen t plant species composing their diet. The grazing hPfoovin .... A''·' six cows was followed during an average observation period of eight hours. Bite counts unit of rime were made and bite sizes estimated. At intervals of 15 minutes, the n bites · per cow were recorded for five minutes. The plant groups to which the corresponded were also noted, i.e. bunch grasses, short grasses, or ground-covering (mainly Lachemma orbiculata). To enable a quantitative comparison of the intake of cows, they were observed in areas with similar forage, mainly bunch grasses and short (floristic type D, Chapter 3).

but dEIS>lY

aUC:eprotein diCCSttl>llty ~ICSO'ow.>

dead / alive

FECES

Bite sizes were simulated by two observers harvesting the amounts approximately corJSUJme<Eiiltl of each plant group in one bite. These artificial bites were harvested at bite depth (Ungar Noy-Meir, 1988), with a cylindric shape for bunch grasses and an elliptic shape for grasses. The bite frequency of each animal was calculated over the entire active period observed. The number of bites was then extrapolated to 24 hours using this frequency. The total dry-matter intake was assessed by multiplying bite size by bite frequency and grazing time (Forbes, 1986).

;i.

.·. i' /

Selection and forage quality By interviewing local fanners, additional information was collected about cattle preference for certain plant species. For the most abundant species in the cattle's diet, digestibility was analyzed using the modified Van Soest method (Van Soest, 1982) of in vitro organic matter digestibility. This method simulates animal digestion, using rumen fluid. Of the same plant . samples, crude protein content was determined according to the Weende analysis based on nitrogen content. Nitrogen content was determined using a Carlo Erba 1106-Elemental Analyzer. Ivlev's electivity index is a quantitative measure of food selection (lvlev, 1961; Jacobs, 1974); this is defined as the relative difference between the fraction of a given forage type in the animal diet (r) and the fraction of the same forage in the vegetation (p). A differentiation into five plant groups was made as follows: green tussock leaves, dead tussock leaves, shon grasses, ground-covering species together with forbs, and shrubs. At this level of detail, the groups could be clearly distinguished during field observations of grazing behaviour. Biomass data of 39 plots of I m2 were used to assess the contribution of the different plant groups to the total aboveground phytomass, excluding litter and dead leaf bases. lvlev's electivity index £ , where E = (r-p)/(r+p), was detem1ined for each of the five plant groups, and varied from -I to 0 for negative selection and from 0 to +I for positive selection. It was determined whether selection corresponds to differences in forage quality among the groups.

Fecal excretion Four adult cows wen: kept in an enclosure for 7, 14, 24, and. 24 hours, respectively. Feces produced within each p~riod were collected and weighed.

86 I

CALVES

Figure 4.1

Model of the grazing system, showing the .factors detennining cattle intake, (derived from Mannetje & Ebcrsohn, 1980).

An equation used in calculating forage intake, based on fecal excretion and dry-matter digestibility of the forage, is given by Van Dyne eta/. (1980) as: -F = 100

* E/(100- D)

Equation 4.1

where D =digestibility of dry matter(%); E =fecal excretion, dry weight (kg/day); F = forage intake, dry weight (kg/day).

Nutritional requirements . A third method for estimating in@<e is the summation of requirements for the maintenance and secondary production of cattle; maintenance is a function of body weight. As derived from Barrett & Larkin (1974), maintenance requirements correspond to 33 g digestible dry matter and 0.46 g nitrogen per unit of metabolic (body) weight, which is equal to body weight to the power 0.75. . Live-weight gain was estimated for different age classes by measuring body weight changes of 50 cows within a period of four months. Heart ginh was used to assess live-weight (Vos & Vas, 1967). 87

Chapter 4 On a modern farm where a balance for weighing cows was available, these mP"""••~·-~" we~ calibrat~d experimentally by eswblishing the relationship between heart girth and wetght. In thts way, an average Jive-weight gain per day- was detcm1ined. The expressi~g dry~matter intake for growth as a function of Jive-weight gain is given by & de Rtdder (1991~. Energy content of the feed is assumed to be the limiting factor gro,v.th. Dry-matter mtake used for growth is then: D:-.11g = LWG

* 12.1/(D * 18.4 * 0.49)

where DMig = dry-matter intake used for growth (kg); LWG =live-weight gain {kg/day); 12. energy content of I kg live-weight gain (MJ) for an average adult cow (Balch eta/., 1980); 18.4 energy content of dry matter (MJ/kg) (Barret & Larkin, 1974); 0.49 =transformation efficiency to energy (Barret & Larkin, 1974)_ Live weight-gain was measured under different management conditions and compared the_figures for s.tandard conditions in Colombia at elevations below 3000 m (Geoffray, 198 whtch are constdered to be more favourable. The management conditions differ in tenns forage quality, reproduction system, and environmental stress. As discussed above, the of cattle in the paramo consists of a selection of the natural vegetation. At Finca N"''"'''n..t•r~ •~• (3600 m), adjacent to Los Nevados National Park, improved grass species such Amho.:~ mum odorarum, Dactylis gfomerata, Lofium peremze, and Phafaris sp. are culti In ~dd_mo~, con~en~ate~ are used as supplementary feed. Another important difference is arufic~al msemma1ton IS used at Finca Normandia to control reprOduction. At the elevauon of the paramo however, environmental stress is more severe than at Normand~a- Besides maintenance and growth, milk production and reproduction also require extra nutnents. According to ~arr~t & Larkin (1974), energy and nitrogen requirements per kg milk of 4% fat are ~32 g dtgesublc dry matter and 8.4 g nitrogen respectively. Daily milk production was detennmed_for seven ~dndomly chosen cows at different stages of lactation. Dry-matter intake used for ~mlk pmducuon was then calculated. In a similar way, dry matter intake used for reproduction was assessed based on energy and nitrogen requirements. In the last two to three months ~f-ge~tation, 27 g digestible dry matter and 0.43 g nitrogen are required per unit of metabolic wetght (Barret & Larkin, 1974). !he reproductive cycle was studied in order to define the time fraction during which the _ mt~e of an av~rage _cov: from the herd increases due to lactation or reproduction -· requz:ements. By mtervtewmg local fanners, information was collected about management pracuc~s ~~d herd stru.ctu_res. Questions were asked about age at first calving, calving interval, cow vtab1 l1ty, calf vtabthty, and lactation period. Calving percentage for the herd was calculated from herd composition, by dividing the total number of calves born in one year by the total number of breeding females (Wagenaar & Konrrohr, 1986). Environmental effects on voluntary intake occur at temperatures below !SOC (NRC, 1981; Fox, 19~6); an adju~tmcnt of +3% in dry-matter intake between temperatures of 5 to 15"C is · · ap~r~pnate. Mannetje ~ Eb~rsohn (1989) indicate an adjustment of about +20% for grazing ~cttv~ty. Bec~use of rehef dtfferences and the fact that cattle walk relatively long distances m thts cxtens1ve area, extra energy is required. 88

Extensive livestock production Hafez & Dyer (1 969) report a daily energy requirement of 79 kcal/ 100 kg/mile for walking and 207 kc:~VI OO kg/1000 ft extra for ascent All corrections were applied in the final calculation of forage in take.

Secondary production A cow productivity index was computed as the product of cow viability (%) • live-weight gain (kg) +cow viability (%) +calving percentage(%) • ca lf viability (%) • calf weight at 1 year (kg)+ cow viability (%)*calving percentage(%) * lactation milked out yield (kg)/9 (Trail & Gregory, 1981). This productivity index was expressed in kg per cow per year and in kg per 100 kg of adult cow maintained per year to provide a basis for comparison with ·other extensive livestock systems.

4.3

Grazing behaviour

From field observations in the study area and interviews with the local fanners, it can be concluded that the cattle graze from 5 or 6 a.m. until 8 or 9 p.m. A mean active grazing time of 60% was found, which included rest periods of less than 15 minutes. This implies that cattle graze approximately nine hours per day, with four or five rest periods amounting to six hours. In general, cows graze from four to nine hours per day, with more time spent grazing ·when on rangelands than when on dense pastures (Van Dyne et af., 1980). The cattle of the paramo apparently have to make a considerable effort to meet their nutritional requirements by grazing for long periods over large areas, which confinns the extensive nature of the grazing system. the simulated bite sizes were assessed at 1.06 ± 0.44 g dry w.:ight (± s.c., n = 20) for short grasses and 1.29 ± 0.68 g (± s.e., n = 50) for bunch grasses. Results of intake estimations based on. bite counts are presented in Table 4.1. According to this method, total intake was estimated at 13.8 kg dry matter per day.

4.4

Quality and quantity of forage

From observations of grazing behaviour, it was concluded that most of the cattle diet consists , of short grasses (with sedges included) of mainly Calamagrostis coarctuta and Carex tristicha. Bunch grasses of Cafamagrostis spp. and Fesruca spp. and the ground-covering species Lachemilla orbiculata are important components. Of the forbs, the dominant species are Trifolium cf_amabile, Rumex acerosella, Castilleja[!Ssifolia, and Bartsia pedicularioides, but in comparison with the grasses they play a minor role in the cattle's diet Hardly any shrubs are consumed. Remarkably, cattle sqmetimes eat the inflorescence of the stem rosette Espeletia hartwegiana.

Chapter 4

Cow

Oite rate (!/min.) bunch grasses

I 2 3 4

12. 1 6.4 7.3 6.4

Extensive livestock production

Average intake short grasses

bunch grasses

13.5 17.8 15.7 10.4

8.4 4.4 5.1 4.5

Total intake (kg/day)

shon grasses 7.8 10.2 9.0 6.0

In Table 4.3, the quality per distinguished plant group is given in terms of digestibility and nitrogen content. The average nitrogen content and in vitro digestibility of the total diet were ' determined at 1.2% (N) and 52% (D), respectively. Quality per plant group is compared with Mev's electivity index as a quantitative measure of selection. lvlev's electivity index appears to correspond well with differences in forage quality: This is shown in Figure 4.2, where relative differences from the average digestibility and nitrogen content are plotted against ·Ivlev's electivity index. A linear relationship is suggested.

Plant group Table 4.1

Intake estimates based on bite counts during the active grazing period of canle in · paramo of Los Nevados National Park. Grazing period 9 hrs; bite size oT bunch and short grasses 1.29 g and 1.06 g dry weight, respectively.

Table 4.2 shows the relative contributions of different plant groups to the dry-matter of cattle. Our observations confirmed that the- cattle select leaf rather than stem and E'"'"'~•• rather than dead material as was reponed by Mannetje (1974) and Mannetje & Ebersohn ( 1980). Gr~en tussock leaves have a significantly higher nitrogen content and digestibility than dead ~atenal. The same holds true for short grasses, which make up only a small pan of the.. total b10mas~ (9%) and are preferred to bunch grasses. ·

Plant group

% of diet

bunci1 grasses

30 ± 5

shon grasses

40 ± 5

ground-covering species

20 ± 5

forbs

10 ± 5

shrubs

stem rosette flowers

Table 4.2

Dominant species Calamagrostis effusa green Calamagrostis effusa dead Calamagrostis recta green Ca/amagrosris recta dead Festuca sublimis green Festuca sublimis dead Calamagrostis coarctara Carex tristicha .4grostis tolucensis Lachemilla orbiwlata Lupinus microphyllus Saturcja nubigena Rumex accwsclla Trifolium cf. amabi/e Castilleja [cssifolia Bartsia pedicu/arioides Baccharis genistelloides £sea/Ionia myrtilloidcs Espeleria hartweRiOIIO

D(%)

N(%)

33.7 ± 1.1 20.7± 3.9 29.9 ± 2.3 19.2 ± 1.3 39.9 ± 6.6 20.7 ± 9.5 62.8 52.8 71.1 61.8 59.7 62.8

0.8 ± 0.14 0.2± 0.1 2 0.9 ± 0.05 0.3± 0.10 0.8 ± 0.08 0.3 ± 0.09 0.9 1.8

79.2 68.5 62.6 66.7 54.0 45.8

Fraction of vegetation p

E =(r·p)/(r+p)

(%)

(g)

bunch grasses green dead

30 21 9

29.4 33.4 20. 1

0.7 0.8 0.3

82 25 57

602 ± 85 184 ± 26 419 ±59

-0.47 -0.09 -0.73

shon grasses

58.8

1.4

64 ± 51

+{).63

61.7

1.3

48 ± 34

+{).62

forbs

70.1

1.9

shrubs

60.3

1.1

15 ± 16

total diet

100

52.0

1.2

Table 4.3

Comparison of forage qualiry and forage preference using lvlev's electivity index E (lvlev, 1961 ), defined as the relative difference between the fraction (r) of a given forage type in the animal's diet and the fraction (p) of the same forage in the vegetation [£ = (r·p)/r+p)). Quality is expressed as in vitro digestibility (D) and nitrogen content (N) of consumed plant groups (weighted averages according to species preference) and of total cattle diet.

1.2 2.2 1.3

1.6 3.1 . 1.5 1.5

0.8 1.4 1.6

Botanical composition of cattle diet in the paramo of Los Ncvados National Parle D = in vitro digestibility, N = nitrogen content.

Fraction of D diet r

Fecal excretion Mean daily excretion of feces was assessed at 4.9 (± 0.9) kg dry matter (± s.e., n = 4). The average digestibility of the consumed dry matter was estimated at 52%. Using equation 4.1, the following estimate of forage intake could be derived: 10.1 ± 1.8 kg dry matter per day. Nutritional requirements and secondary production As mentioned above, forage intake is highly dependent on the ratio of en~rgy to nitrogen content of the feed: The critical level of crude protein, below which voluntary intake of dJy · matter by beef cattle is depressed, is 7% (FAO, 1991). In this study, nitrogen content was found to be the limiting factor solely for milk production. 91

Chapter4

Exte11sive livestock productio11 Management conditions Weight (kg) at birth 1 year 3.5 years adult (6 years) Growth (kg per month) 0- 1 year I - 3.5 years 3.5 - 6 years 3.5 - 12 years

favourable

moderate

paramo

38-40 190 490 532

38-40 165 387 459

30-40 165

12.5 8.9 1.4

10.5 7.4 2.4 1.2

10.5 6.5 1.9 1.2

360 417

Weight and growth per age class of Normando cows under different management conditions: standard favourable conditions in Colombia (below 3000 m; Geoffray, 1981), moderate conditions at Finca Normandfa (3600 m), and the more extreme paramo (± 4000 m).

0.2

· lvlev's electivity index .A

Digestibility (%)

• . Nitrogen (%)

Figure 4.2 Relationship between the forage quality of plant groups as relative difference from mean forage quality of the total diet and lvlev 's electivity index as a measure of selection. Relative forage quality difference from the mean is calculated as (%N of group - 1.2% N)/1.2% N, or as (%D of plant group - 52% D)/52% D. BUO = grasses, dead biomass; BUY= bunch grasses, green biomass; BUT= bunch grasses, ROS = ground rosette layer; SHO = shon grasses.

In the case of all other feed requirements such as maintenance, live-weight gain, reproduction, energy 'was found to be limiting. The results of measurements of live-weight gain are presented in Table 4.4. In the pararno at Finca Normandfa, growth is apparently most reduced in the age class 1 to 3.5 years, continues longer. This indicates that cattle reach maturity later than under more conditions, which is confirmed by the age of the animals at first calving, i.e. 3.5 years against three years under more favourable management. In the paramo, lower growth throughout result in lower adult weights. The average body weight of an adult cow determined at 417 kg and its mean live-weight gain 1.2 kg per month. Dry-matter intake for growth was determined at 0.1 kg per day. 92

According to local farmers, the mean calving interval is 12 months, unsuccessful gestations not taken into account. Suc~ssfu l calving percentages of 78%, 59%, and 56%, respectively, 'were calculated for the three. different herds. Mean daily milk production of paramo cows is assessed at 5 kg milk per cow, of which I kg is consumed by the calf. The lactation period lasts seven months. The lactation yield within this period is assessed at about ll 00 kg milk. Taking into account a mean calving percentage of 64%, an average cow from the herd . requires 0.6 kg dry matter per day for reproduction and 1.3 kg per day for milk production. Dry-matter intake used for maintenance is assessed at 5.9 kg per day. Summing maintenance and production requirements, dry-matter intake is estimated at 7.9 kg per day. Besides normal grazing activity, the cattle were observed to walk about 5 km extra per day with an estimated iiscent of 50 m. After adjustments for walking, grazing activity, and environmental stress, a total dry-matter intake of 11.4 kg per day was obtained for adult cows (live-weight 417 kg). Acow productivity Index of 144 kg per adult female per year was calculated, which is equal to 35 kg per 100 kg live-weight per year.

4.5 ')

Conclusions

r.•

Of the three final estimates of daily dry-matter intake by cattle, the one derived by the first method (bite counts) is probably an over-estimation. The figure of 13.8 kg dry matter per day !s above the 2 to 3% of live body weight generally taken as· a rule of thumb to roughly · estimate daily intake .. Simulated bite size is highly variable; nevertheless, it is a critical parameter in the intake calculations based on bite counts. The simplicity of this method is due . to using continuously variable parameters as single means or totals (Hodgson, 1982).

Extensive livestock production

Chapter 4 However, the method is useful for deriving information on the plant composition and of the cattle diet. This infom1ation is an essential input other intake estimll!cs, and related to selection processes and changes in the natura l vegetation due to grazing.

' LITERATURE CITED

. Arnold, G.W. & Dudzinski, M.L. 1978. Ethology of free-ranging domestic animals. (Developments in animal and veterinary sciences No. 2). Elsevier, Amsterdam, 198 pp. ·. . vBalch , C.C., Armstrong, D.G. & Greenhalgh, J.F.D. 1980. The nutrient requirements of ruminant livestock: technical review by an Agricultural Research Council working party. Agricultural Research Council, Commonwealth Agricu ltural Bureaux, Slough, 351 pp. .', Barren, M.A. & Larkin, P.J. I 974. Milk and beef produc tion in the tropics . Oxford University Press, London, 245 pp. · · Beekman, A.M. & Verweij, P.A. 1987. Structure and nutrient status of a paramo bunchgrass vegetation in relation to soil and climate. MSc. thesis. Internal Report No. 233, Hugo de Vries-Laboratory, University of Amsterdam. ·Breman, H. & De Ridde r, N. 1991. Manuel sur les patura It appears that if the forage quality of a plant group ges des pays saheliennes. ACCfis higher relative to the averag . CTA-Karthala, Paris. quality, the animal preference for this plant group, expressed as lvlev's electivity · Dfaz, T.E. 1985. Alime ntaci6n de vacas lecheras. Pp. increases in a linear way. Increasing height, bulk density 17-24 in: Koeslag, J.H. & Urbina, N. , and cover are mentioned by (eds.), Producci6n de leche, zonas de ladera fiia. ICA-C & Noy-Meir (1988) as ways to maximize intake rate. CH, Pasta. Despite their lower height, lower ·. -FAO 1991. Guidelines: land evaluation for extens density, and lower cover, the short grasses and ground ive grazing_ FAG Soils Bulletin No. 58. -covering species are preferred FAO, Rome, 158 pp. . bunch grasses. This suggests that forage selection in the studied paramo bunch grd.~Sliinus,lft Forbes , T.D.A. 1986. Quantitating forage availabili ty. Pp. 143-!5 controlled by quality factors rather than the spatial organi 8 in: Owens, F.N. (ed.), Feed zation of herbage. intake of beef cattle. Oklahoma State University, Stillw ater. Fox, D.G. 1986. Voluntary intake of beef cattle. Pp. An initial transition model, descri bing vegetation dynam 193-207 in: Owens, F.N. (ed.), Feed ics influenced by grazing and intake of beef cattle. Okl~homa State University, Stillwa in a small pilot area, was developed by Pels & Verwe ter. ij (1992). Basic data on Geoffray, H. 1981. El normando colombiano en cifras. behaviour were, however, still lacking. The present study Norma ndo Colombiano 1(3): 25-38. provides building blocks for Asociaci6n Colombiana de Criadores de Ganado Norrna spatial and simulation models of the grazing system. Especi ndo, Santafe de Bogota. ally the data on diet con~positicmU Hafez, E.S.E. & Dyer, I.A. 1969. Animal growth and and grazing preference for certain plant groups were nutrition. Lea & Febiger, Philadelphia. essential for developing the Hodgson, J. 1982. Influence of sward characteristics on presen ted in Chapters 6 and 7. diet selection and herbage intake. Pp. 153-166 in: Hacker, J.B., (ed.), Nutritional limits to animal production from pastures. Commonwealth Agricultural Bureaux, Farnham Royal. Indices of cow productivity are given by Trail & Gregory (198 1) for extensive Hofstede, R.G.M., 1995. Effects of burning and grazin systems in the Kenyan highlands at elevations of 1800 g on a Colombian paramo ecosystem. to 2200 m. Climate there is <f'nn~-•<n~.,,. PhD dissertation, University of Amsterdam, 199 pp. (yearly rainfall 610 to 680 mm) and the breed concerned is Sahiwal cattle and its crosses Ivlev, V.S. 1961. Experimental ecology of the feeding Bos taurus and the indige nous Bos indicus. Cow of fishes. Yale University Press, New productivity indices of six different Haven, 302 pp. · range from 32 to 68 kg per 100 kg Jive-weight per year. Our estimate of cow productivity Jacobs, J. 1974. Quantitative measurement of food selecti 35 kg per 100 kg per year thus corresponds to the lower on: a modification of the forage ratio end of the bracket. and lvlev's electivity index. Oecologia (Berlin) 14: 413-41 7. Koeslag, J.H. 1985. Produ cci6n de leche en zonas de l'he extensive nature of the livestock production system ladera fria. Pp. 1-8 in: Koeslag, J.H. & is also illustrated by the low Urbina, N. (eds.), Producci6n de leche, zonas de ladera of cattle diet and the high grazing effort in terms of grazing frfa. ICA-CCH, Pasta. time and walking distance. · Manne tje, L. 't 1974. Relations between pasture attribu appare ntly invest a lot of energy in obtain ing enough tes and Jiveweight gains on a forage of sufficiently high protein subtropical pasture. Proceedings 12th International Grassl energy conten t. The low nitroge n content is a limitingiacto and Congress,-Moscow-3: 299r in-milk-pi·odu c.llol~i-matntf:nan<CtM -- 304. growth, and reproduction are limited by the amoun ' t of energy consumed. At these Manne tje, L. 't & Ebersohn, J.P. 1980. Relations betwe elevations, it is not surprising that the harsh paramo en sward characteristics and animal environment sets limits not only production. Tropical Grasslands 14: 273-280. human activities, but also to the -performance of Norma ndo cattle. Manne tje, L. 't & Ebersohn, J.P. 1989. Tropical grassland improvement. Wageningen . Agricultural University, Dept. of Field Crops and Grassl and Science, Wageningen, 39 pp. Minson, D.J. & McDonald, C.K. 1987. Estimating forage intake from the growth of beef cattle. Tropical Grasslands 21: 116-122. ·NRC. 1981. Effect of environment on nutrient require ments of domestic animals. National Academy of Sciences, Washington D.C., 152 pp. The other outcomes are of a similar order of magni tude (10.1 and 11.4 kg day·') considered to be more reliable. The result of the intake estimate based on ma:intenar1c1 produ~tion requirements can be partly validat ed u~ing the general intake equation for (Minson & McDonald, 1987). The prediction of dry-ma tter intake for a steer of 417 weight and a growth of 0.04 kg day·' is 7 kg day·', with a coefficient of variation of According to our calculations, an average cow of the paramo would need 7.2 kg dry per day for maintenance, growth, and normal grazing activit y alone, which is within the predicted by Minson & McDonald (I.e.).

Chapter 4 Pastrana, R., McDowell, L.R., Conrad, J.H. & Wilkinson, N.S. 199la. Macromi neral oLshcep in the paramo region of Colombia. Small Ruminant Research 5: 9-21. Pastrana, R., McDowell, L.R., Conrad, J.H. & Wilkinson, N.S. 199lb. Mineral status of in the pa[amO region of Colombia. Small Ruminam Research 5: 23-34. Pels, B. & Verweij, P.A. 1992. Burning and grazing in a Colombian bunchgra ss ~cosystem: a transition model. Pp. 243-263 in: Balslev, H. & Luteyn, J.L. (eds.), an Andean ecosystem under human influence. Academic Press, London. Schmidt, A.M. 1992. Study of extensive livestock systems in the paramo of Parque Los Nevados, Colombia. MSc. thesis. Internal report, Hugo de University of Amsterdam. Schmidt, A.M. & Verweij, P.A. 1992. Forage intake and secondary productio n in '"''""'''VP. livestock systems in paramo. Pp. 197-210 ilt: Balslev, H. & Luteyn, J.L. (eds.), an Andean ecosystem under human influence. Academic Press, London. Trail, J.C.M. & Gregory, K.E. 1981. Sahiwal cattle: an evaluation of their contribution to milk and beef .production in Africa. ILCA Monograph 3. ILCA, Ababa, 128 pp. Ungar, E.D. & Noy-Meir, L 1988. Herbage intake in relation to availabili ty and structure: Grazing processes and optimal foraging. Journal of Applied Ecology 25: I 1062. Van Dyne, G.M., Brockington, N.R., Szocs, Z., Duek, J. & Ribic, C.A. 1980. Large subsystem. Pp. 269-537 in: Breymeyer, A.I. & Van Dyne, G.M. (eds.), Grassland analysis and man. Cambridgc:, University Press, Cambridge. Van Soest, P.J. 1982. Nutritional ecology of the ruminant. 0. & B. Books Inc., Corvallis, pp. Verweij, P.A. & Budde, P.E. 1992. Burning and grazing gradients in the paramo of Los Nevados, Colombia: initial ordination analyses. Pp. 177-195 i11: Balslev, Luteyn, J.L. (eds.), Paramo: an Andean ecosystem under human influence. ACltdeJmtc Press, London. Vos, M.P.M_ & Vos, H. 1967. Estimation of the live weight of cattle from measurements. Diergeneeskunde 92: 1073-1081. Wagenaar, K.T. & Kontrohr, E. 1986. Appraisal of the ILCA cattle herd dynamics using data from pastoral systems in Mali and Kenya. Modelling of extensive I production systems. ~p . 231-247 in: De Ridder, N., Van Keulen, H., Seligman , !\.G. & Neate, P.J.H. (eds.), Proceedings of the ILCNARO/CABO Workshop, 5-9 Febr. 1985, Aro, Bet Dagan, Israel. ILCA, Addis Ababa.

EFFECfS OF FIRE AND GRAZING ON PLANT POPULATIONS

Pita A. Verweij and Kasper Kok 5.1

5.2

5.3 5.4 5.5

Introduction· Methods of studying stem rosettes Management impact on stem rosette populations Methods of studying bunch grasses Management impact on bunch-grass populations

The objective of the present chapter is to describe the response of the dominant plant'groups of the natural bunch-grass paramo to burning and grazing gradients. The populatio ns studied correspond to the stem rosette Espeletia lumwegiana ssp. centroandina Cuatrec. and the two most common bunch grasses, CalafTUJgrostis effusa (H.B.K.) Steud. and C. recta (H.B.K.) Steud.

5.1

Introduction

5.1.1 Stem rosettes In the

mont~e

landscapes of the Andean paramo, giant caulescent rosettes of Espe/etia features. Natural selection favours longevity; 10 companson wuh recently established plants, taller and thus older individuals are more successfu l in maintaining a favourable water balance in the growing tissues, which increases their chance of survival (Smith, 1981; Monasterio, 1986). The leaf rosette of the adult is · situated at a 'safe ~eigfl t' above ground level, where fluctuations between day and night temperatures are greatest. On the other hand, the water-storing pith tissues in the center of the stem are protected by dead leaves. Frost damage is thus reduced by the retention of dead leaves on the stems of giant rosettes. Experimental removal of these leaves caused higher mortality during the dry season in Espeletia schultzii (Smith, 1979; Goldstein & Meinzer, 198_3). Furthermore, the densely packed bases of dead leaves probably serve as protection agamst ftres. ~Asteracea~) and :elated genera are characteristic

Before the .creation ~f Los Nevados National Park in 1973, Espeletia plants were occasiol]ally harvested m the neighbourhood of the fincas. Stems were used as fuel for cooking and heating, and as insulation material in the walls of farmers' huts. Extensive collecting in combination with intensive trampling by cattle might have Jed to the present scarcity or absence of Espeletia hartwegiana in the vicinity of formerly inhabited places.

Pan of this chapter has been published as Verweij, P.A. & Kok, K. 1992. Effecis of fJ.re and grazing on Espeletia hortwegiana populations. Pmuno: an Andean ecosystem WJder human influence. © Academic Press, London.

96 97

Chapter 5 Little has been reported sofar on the effects of fire and grazing on Espele1ia populations many paramos of-the Colombian Andes, the bunch-grass vegetation is purposely local fanners to stimulate forage regrowth for cattle. In a Venezuelan paramo, Smith found that as much as 55% of adult £. sclzultzii plants could be killed in a dry season Perez (I 992) observed a decrease of 37% in rosette cover of Coespeletia timotensis as a of browsing by cattle. The impact of grazing and burning practices on the following aspects of C.SJ7ell!lin.~:J hartwegiana population dynamics was investigated: (I) growth of individuals of different classes; (2) demographical aspects, such as age-specific mortality rates and survival juveniles; and (3) population density.

Espeletia growth patterns as affected by fire and grazing were characterized in terms mathematical functions. It was hypothesized that burning and/or grazing causes incrf'A•"'~''"' mortality, particularly among adult individuals. Although cows rarely eat from the rosettes, their trampling may exert significant influences. The reproductive patterns of the Espeletia populations have only been taken into accoun~ indirectly. As demonstrated by Monasterio (1986) for the Venezuelan paramo, a massive seci! ·: production occurs at irregular time intervals, yet sufficiently to replenish the soil seed bank. An important portion of the seeds is also retained on the inflorescence stalks. As long as the population density does not decrease dramatically, we can assume that availability of viable ·. seeds is not as limiting for regeneration from seedlings as is the availability of space, light, . and nutrients.

5.1.2 Bunch grasses _ The second important plant group in the ecosystem is the bunch or tussock grasses. The standing dead biomass of tussock grasses, like that of the stem rosettes, provides protection from the extreme climatic conditions of the paramo. Regarding the effects of grazing and burning on bunch grasses, literature does not provide ari ,· unambiguous picture. In general, burning is believed to have a stimulating effect on production and total biomass. Abrams eta/. (1986), Hadley & Kieckhefer (1963), and Mentis . & Tainton (1984) report shon-term increases in production. In South African Themeda · triandra grasslands, reduced yields were only recorded during the season immediately , following a fire (Tainton & Mentis, 1984). Some authors indicate that production is at pre· bum levels after 5 to I 0 years (Collins, 1987; Allen & Partridge, 1988). McNaughton ( 1985 ), . on the other hand, did not measure a significant production increase and Allen & Partridge (1988) and Van Groen (1987) even report decreased productivity. Conclusions on the effects of fire cannot be generalized as the influence of abiotic factors is crucial (Malanson, 1984). ·.

Effects on plam populations Furthermore, the last author gives nine reasons for th is sti mulation, of which the most important are: (I) higher photosynthetic rates; (2) older tissues producing well below the maximum photosynthetic rate are removed; (3) increased light intensities; and (4) nutrient input from urine. What seems decisive is the state of degradation of the investigated sites: overgrazed plots produce less, whereas moderately grazed plots show increased production levels (Titlyanova eta/., 1988; Kelting, 1954). Besides actual grazing intensity and fire age, burning and grazing history are therefore important determinants of the state of the tussock grasses.

. . The effects of burning and grazing on the tussock grasses arc to some extent similar. The aspects in common are the (partial) removal of plant biomass and the enrichment owing to an increase in available nutrients. Hulbert (1969) and Schiatterer & Tisdale (1969) state that the principal effect of both fire and grazing (without making a distinction between the two) is that litter is removed. Collins (1987) and Tainton (1982), on the other hand, concluded that total removal of biomass by burning or clipping is not comparable to partial removal of biomass by grazing. · In paramo ecosystems, fire practically always occurs in combination with grazing (Van Groen, 1987; Verweij & Budde, 1992), thus complicating the analysis of the separate effects.

Tussock fragmentation The death of a tussock by fire or grazing is a rare event, as the tussock growth form provides the meristem with optimal protection (Van Groen, 1987; Lozano & Schnetter, 1975; Hofstede . et a/., 1995a). Instead, the major form of human impact on paramo bunch grass is the phenomenon of fragmentation of the tussock base (Verweij et al., 1995). The central part of the tussock base remains bare and several smaller tussocks develop at the edges. These isolated tillers are defined here as 'sub-tussocks'. If tlie old tussock base is further damaged by trampling impact, the sub-tussocks may continue growing as new independent tussocks. Van Groen (1987) indicates· some general patterns of fire and grazing that jointly cause fragmen_tation, whereas according to Rosen (1982) cow trampling alone causes lichens to fragment. Evidence of fragmentation in grazed vegetation types was reported by Riney (1963), Sala et a/. (1986), Titlyanova et a/. (1988), and Sundriyal & Joshi (1990). Without the influence of grazing, tussock density also increases after fire (Allen & Partridge, 1988). The fundamental mechanism that underlies fragmentation is probably the removal of the dominance of the apex, which normally prevents lateral tillering (Tainton, 1982). Though it is considered an important process, rio quantitative data concerning the fragmentation of paramo tussocks were available. The following specific aspects of vegetation structure and population structure were investigated: (1) changes in tussoc~ size; (2) biomass; <!nd (3) productivity. The effects of burning and grazing on the tussock grasses Calamagrostis effusa and C. recta were investigated. FestLtca sublimis is limited to slightly different sites with a higher humidity and was excluded from the analysis.

Reports on grazing effects agree even less. Sundriyal (1989), Sundriyal & Joshi (1990), and Karunaichamy & Paliwal (1 989) report production decreases of 20 to 75%. McNaughton (1979, 1984, 1985), on the contrary, measured a production increase of 85 to 100%.

Chaprer 5

5.2

Methods of studying stem rosettes

Site descriptions

The selection of sites was based on the initia l grazing and burning gradient determined Verweij & Budde (1992). Ten experimental sites were chosen at elevations of 4000 to m. An attempt was made to select sites with similar topographical conditions, i.e. slopes. However, in reality, heavily grazed situations where Espeletia populations still only exist on flat valley bottoms and the sligh tly undulating ridge tops of lateral Furthermore, grazing influence is common at burned sites, especially after recent fires the vegetation is most attractive to the cattle . Fire age was determined by evaluating photographs of different dates (10 series from 1955 to 1989, see Chapter 4) and interviews with farmers and park guards. Detai ls of the study sites are given in Table S.l. sites correspond to the Calamagrostiemm effuso - rectae association. 5.2.1

E-<

-d Ill

"'- ~ l:Q~ ~

Cll'-

,:,/,

,....

~ ~ ~ loo

a-<">

~~~ ("\.-< '<!'

..... ....

Ott'\ .........

It'\

.... It'\ ....

:g~~

OOM -Or- -

\Q ....

a_.c

ltY VI

......

M'D MO -M

_.._

gr-11'\ o-ooo-or--

OO<"l

- - M C'1M

:::i~

~~~88

'0"'

V'lO it'\O V'l

'0"'

MM M'-O M

E-<8

Growth measurements

Growth of Espeletia hartwegiana was deter mined in three different ways: by "''"" 0 "'""" annual height increments, by establishing the leaf expansion in one year, and by ev<Uucttin2~ the height increment of regenerating stem rosett e populations after ftre events of a known The annual height increment was measured in four different plots under different macnaj~ell'l ent'ÂŁ 1 conditions (Undist. 3, Fire 2, Fire 6, and Grazed 1). In each, 30 to 50 individuals measured, starting from a place chosen at random. The same parameters as those describing population s~cture were recorded. In addition, the dead leaf columns of the plant s were marked during the initial survey with a horizontal line of red paint close to the rosett e. Height up to this red line was recorded as a fiXed reference in order to avoid variation in height due to soil erosion or irregular groun d surfaces. These measurements were taken in April 1989 and repeated twice, about 1 and years later. To monitor leaf expansion, the same indiv idual s were followed by marking the leaf tips both the youngest visible and the oldest living leaves during the initial survey. After one year, the number of newly deve loped leaves were counted. In addition, the total number of leave s present in the rosette were counted. Finally, where the age of the fire was know n, an average growth rate was derived from the regrowth of the dead leaf column of burne d stem rosettes. 5.2.2

....,::1>

-; 0

Population parameters

Plot size was approximately 300m \ so at least ISO adult stem rosettes were covered per site. An adult was defined as an individual taller than 30 em. Individuals exceeding 50 em were observed to flower. 100

....l.... l;:E

loo

E-<~

......

~c.

<"'

::E~

....l.... l...l.... liXl

...l...l

OOV 'l

Olt '\00 8 N("f") ('I"')M

~('I

~ZCI)CI) '

WU.I

"'-('I

iii â&#x20AC;˘\:

::E::E~~...l

00... .1

~~~

CI)CI) CI)

CI)CI)

Chapter 5

Effects on plant populations

The following parameters were recorded for each Espeletia adult encountered: total from ground level to the highest leaf tip, height of dead leaf column from lower to point of auachment of the dead leaves, height of bare stem due to burning, traces of by caule, and phenological stage, i.e. vegetative or ~owcring. In juveniles, only measured. Furthermore, it was noted whether the stem roseues occurred in groups, and the group size. If the distance to their nearest neighbour was less than 5 em, juveniles . considered to be part of a group, whereas the critical distance for adults was defined em.

·.. Demography . . .. For each study site, life tables were filled in with the num~rs. of. Espeleua md1v1duals per . h · ht class complete with values of height-specific monahty md1ces (~,) calculated from .:_ . ;;gpopulati~n structures. Tables produced by this method are known as static or vertical life tables. '

Mortality rates Monality rates for adults were determined in two different ways. First, all dead indi within the experimental plots were marked during the initial survey. After one year, number of individuals that had died and their corresponding height were recorded. way, mortality per height class was obtained. In gra~ed situations, however, the tr~ crmPnto'''~ of dead specimens caused by the trampling of cows made it impossible to original height.

!or

Besides, it was not always clear when an individual had died, and low numbers of dead per height class interfered even more with the collection of reliable data using this method. Therefore, estimates of mortality were based mainly on population •m•N•,,.,.• combining the decrease in the number of individuals per heighi class with the growth rates.

5.3

The survival of juveniles was monitored at site Fire Ia, which had been burned exactly year before. From the analysis of the results, it appeared that for Espeletia individuals 30 is the critical height below which they mostly do not survive a·flre. Furthermore, muuv1umus below this height showed more fluctuations in density, due to variations in the rates recruitment, growth, and mortality. Therefore, the upper limit of 30 em seems to be y·ustlltll:j~'il for the definition of juveniles, and they were analyzed separa~ ly.

Regression analysis Due to the fact that populations were not normall y distributed, simple analysis of v"' '"'"~".i:n was not possible (Sokal & Rohlf, 1981). Thus, an alternative method had to be used in to compare the size and age distributions of Espeletia harrwegiana populations in the ""'r~-·-' ' "" management situations. Differe nt regressions between density and stem rosette height weie compared in order to obtain the best lit. The In-transformation of density per height class' showed the best correlation with. height. In the same way, a correlat ion betwee n regrowth afte.r. · fire and the initial height at the time of fire was establis hed for recently burned populations. Tests to see whether the obtained regression coefficients differed significantly from zero (Sakal & Rohlf, 1981), and at what significance level, were carried out to determine whether these relationships were different from the undisturbed situatio n. Subsequently, the mean of annual height increments was taken for the different recently burned plots, and in this way . a general fire-influenced growth equation was derived.

D namic or horizontal life tables, in which a c9hort of seedling s is followed taking censuses a/intervals, are less subject to error (Silvertown, 1987). Obviou sly, it would be rath~r · practical in the case of long-lived plants like Espeletia to make a census throughout thetr :~etime. The assumption of a constant yearly input of see.dlings to the ~ero-age clas~, however; enables the treatment of a static life table as though 11 was a dynam1~ one. ~or th1s purpose, an average delay time per height class was calculated all pd~~dlaeduobnsdu~mg. the obtained growth equations. Subsequently, height-specific mortalll y was . tvt Y e ay time to obtain mortality rates (per time unit). A dynamic life table represe nting each ~ana~ement situation was calculated. Based on these theoretical monality figur~s and start~ng w1th the same number of individuals in the initial adult height class, a thcorctt c.al populauon ~tructure was calculated. These values were compared with data from the field m order to vahdate the described model.

Management impact -on stem rosette populations

5.3.1 Growth According to annual height increment measurements in undisturbed situations, a growth rate of 8.8 ± 3.5 em per year (mean ± s.e., n =80) was recorded f?r all individ ual~ over 3~ c~ regardless of their age. Annual height inc~ments reponed fo~ g1ant rosett~ spec1es are w1th1~ a range of 1 to 5 em (Hedberg, 1969; Sm1th, 1981; Monaste~o, 1986~ Smuh & Young.' 198?, Monasterio & Lamotte, 1989; Cavelier et al., 1992). There ts no straightforward relauons~tp between growth rate and elevation. The high figures obtained in t~is study, co~pared wlth those for similar ecosyStems, are not yet fully understood. A poss1bl e explanauon could be the input ·of extra nutrients originating from the volcanic ash produc ts generated by the frequent eruptions of the Rufz-Tolima mountain chain. Growth rates of burned individuals varied significantly. In contras t to the undisturbed situation, the growth rate, as.based qn regrowth measure~ent~, related linear!~ to the initial stem height at the time of fire. Regressions are plotted m F1gure 5. ~. A we1ghte? annual average regrowth was derived for burned Espeletia individuals in the five .study s11es. The mean (r,) is 0.050 ± 0.009 and the mean (c) is 7.3 ± 1.2 (n = 5), whtch leads to the followin~ growth equation for burned individuals: .'

Y"' = 0.05

* Xr + 7.3

Equation 5 .I

where Y., = annual regrowth (em) and Xr = stem height at time of fire (em).

102

103

Chapter 5

Effects on plant populations

survival and growth of juve11iles In the recently burnt situations in Fire Ia and Fire lb, a growth rate of 26 and 25 em per year, respectively, was determined for juveniles established after the fire. InT-ire 2, the growth rate of newly recruited juveniles was 20 em per year and in Undist. I a rate of 9 em per year was found. Standard errors in these estimates are high, but do not indicate the real variation in the growth rates as many juveniles have been recruited within less than one year. Juvenile gro~th appears to be stimulated immediately aft~r burning. This effect decreases with time, until the growth of juveniles stabilizes at a similar rate to that of adults in undisturbed situations. Smith (1981) found a juvenile growth rate of 0.3 to 2.0 em per year in a natural situation, which also corresponds to adult growth rates.

E' .s c

~Q) c

·c

-Ol Q)

Not only do juvenile growth rates vary considerably in the different management situations, but the same holds for the number of juveniles surviving. Juvenile density immediately after fire is low: one year after fire, 27 and 75 individuals per 100 m2 were recorded in comparison io 151 newly recruited juveniles in the undisturbed situation. On the Fire 2 site, a number of 4.5 individuals per 100 m2 was found, reflecting a possible trampling impact. However, . juvenile densities are rather variable here related to local micro-climate and -relief.

Leaf production 80

100

120

The mean number of leaves produced per year is 118 ± 28 (n = 80), showing a high variation. An increase in both the number of leaves present and leaf production with stem height was observed. The average number of 1eaves per rosette is 70 and the mean rosette turnover time is 0.59 ± 0.07 years (n = 80). Monasterio (1986) found higher 1\Jmover periods of one to two . years for different Espe/etia species at 4200 m elevation. The only species known from the literature with a similar rapid rosette turnover of 0.56 years is £. schultzii at 3600 m (Smith, 1981). Effects of fire and grazing on leaf production :-vere not investigated.

140

.Xf (em)

•

Figure 5.1

Fire 1

Fire2

)t(

Fire 6

Block lava

5.3.2 Population parameters

Rcgrc~sion line~ of regeneration height (em) of Espe/etia hanwegiana as a function of their stem height at the time of burning. Y = r,, • ~ + c.

Height of burnt Espeletia individuals is then given by: X, = (I + 0.05

* t) * Xr + 7.3 * t

where t = time artcr burning (yr).

Fr~m equations 5.1 and 5.2, the general equation of height growth after fire was derived: dx

0.05 *X,

7.3 + - - -- I + 0.05 * t I + 0.05 * t

Concerning reproduction, none of the individuals defined as juveniles was observed in flower. Nor did adults up to a height of 90 em flower in undisturbed or grazed situations. In the case of recently burned plots, however, this height class had entered the reproductive phase. The percentage of flowering individuals in the other height classes was 4.3 ± 2.0% (n = 8, height classes of 40 em) in undisturbed and grazed situations. An increase (35 to 64%) in the flowering percentage was noted in the two years following a fire. Even two flowering cycles per year instead of one were recognized. Nevertheless, among different sites and within one population, the reproductive pattern could vary considerably, which agrees with reports by Smith (1981), Monasterio & Lamotte (1989), and Sobrevila (1989) on the Venezuelan paramo.

The consta~Jt term of 7.3 em annual regrowth in equation 5.1 does not vary significantly from th_e determmed natural growth rate of 8.8 em per year. This implies that Espeleria growth is stimulated by fire with a certain increment on top of the natural growth rate represented by the first term of equation 5.1. '

In Figure 5.2, plant size disttibutions from undisturbed and burned sites are compared. In Table 5.2, some population parameters are presented for all study sites. The observed decrease in the number of individuals as a function of height (Figure 5.2) was analyzed by means of regressions. This height-dependent mortality proved to be constant over all height classes with high coefficients of regression (R2 > 0.9). In order to be able to compare the mortality rate for the different management situations, the varying growth rates have to be taken into account.

104

105

Equation 5.3

Eftem on plant populations

250 0 0 0 ,...

.......

200

'iii c QJ 0

N' E 0 0 0

150

-T""

.s z. 'iii

J·

j £ill Undist. 1-3 • Figure 5.2

Fire 1

Plant size distribution of Espeleria hartwegiana populations in burned and undisturbed · situations. A. Juveniles (0-30 em). B. Adults (over 30 em). '

Mean adult height is smaller (96 to 104 em) in recentl y burned situations compared wit~ undisturbed conditions (!53 em). The presence of big rocks in the irregular terrain of the. block lava, which was burned 15 years before, explain s the lower adult density in comparison · with other sites. On the other hand, they provide shelter and thus a better chance of survival for the adults. This is expressed in a lower mortality per height class, resulting in a greater · · average height (190 em). The burned plots are charac terized by lower adult and juvenile. densities (Table 5.2). In the most intensively grazed plot, adult density is higher , although the height of most of the individuals is below 100 em. The highest mortality per height interval was found here, reflecting a high grazing pressu re on the population. It appeared that Espeletia exhibits a special survival strategy. High percentages of both adults and juveniles (75 and 96%, respec tively) occur in groups. The juveniles are genera lly not connected to the adult individuals. ·

Occurring in group

(n/100 m1)

(em)

{%)

M, {%J

(per 5 em)

(em)

ad.

juv.

max X age

39.0 ± 9.9 30.7 ± 5.6 48.3 ± 9.2 33.3 ± 5.6 36.3 ± 9.1 31.2 ± 4.1 0

440 220 260 310 465 325 225

juv.

ad.

Undist. Fire I Fire 2 Fire 6 Block lava Grazed I Grazed 2

91 44 39 49 30 81 ISS

323 59

153.7 100.8 95.8 103.9 190.1 156.3 78.4

9.9 11.1 25.3 ± 2.9 8.9 57.1 28. 1 ± 8.6 11.4 67.5 31.2 ± 8.8 6.6 10.7 28.1 ± 8.!1 16.4 7.9 9.4 ± 0.& 25.6 68.5 13.1 ± 2.0 74.8 95.9 41.9 ± 9.8

43 56 44 172

juv.

Age or maximurn

(per 20 em)

ad.

Table 5.2

Mean X

Site

100

(!)

Density

{years)

51.5 19.4 20.9 24.6 43.2 40.2 23.9

Main characteristics of Espeleria hartwegiana populations in different management situations. X =total height; M. =average hdght-dcpcndcnt m<?_rta lity per size ela~s, me~± s:e.; age of maximum= age of the tallest individual measured; ad. =adult; JUV. = JUVerulc.

This phenomenon of clustering may be a product of trampl ing i~pac~, while ~t the same time providing a certain protection agains t it. Those individ uals st~din_g m the _nud~e of a group are expected to be less subject to scraping and trampl ing, which IS especially 1mpo~ant ~or juveniles. A similar tendency, although less pronounced, can be obser:ved _under lo~-mtens1ty grazing. Smith (1981) demonstrated a positive correla tion between climauc stress m ~e fonn of seasonality and decreasing nearest-neighbour distan ce. This confirms our observations on a clustered distribution of stem rosettes under unfavo urable conditions. Table 5.2 shows a remarkable clustering effect of juveni les, mainly around adults in recently burned situations. This might be t:xplained by two different conditions. First, the greater amoun t of bare soil available, due to the absence of bunch -grass cover, is important for rapid coloni zation close to the parent individual (barichorous dispersal system). Second, the possib~e presen ce of higher nutrient concentrations immed iately around the stem of Espefella individuals which was demonstrated by Garay et at. (1983), may favour the observed clustering. 'It is assumed that trampling does not to play a role in detennining the clustered distribution, as the recently burned plots are not grazed . The number of dead individuals per height class encou ntered in the plots divided by the number of living individuals could be used as a valida tion for the mortality rates de_riv~~ from population structures. A constant mortality rate of 13.0 ± 2.8% (n = 10) of adult mdiv1duals within each 20 em height class was determined for the unburned plots, and ~or. t~e burned plots a range of 7.6 to 25% was found. The higher rates correspond to taller mdiVJduals. Finally, Table 5.2 demonstrates that the maximum age that can be reached varies consid~rably among the different populations. In stable populations, this maximum age can be con~1dered a maximum turnover time. The tallest individuals in the undisturbed plots reach an estimated age of more than 50 years, whereas the maximum turnov er in recently burned plots is reduced by more than half.

106 107

Chapter 5 Demography A dynamic life table, representing each management situation, is given in Table 5.3. values of these thcorcticat population structures arc compared with data from the field. demonstrated, a re~sonable fit of predicted values with the real values is obtaint:d. conclusion can be drawn that, compar~:d with undisturbed plots, the recently burned sites 1- 6') are characterized by higher mortality rates. Mortality increase s with stern rosette "'"l!nll.l'u.'.l However, !ncrcased growth rates of the taller individuals compensate for this in such a that the height-dependent mortality M, remains constant throughout the life span of Espeletia plants.

.,_ .... .

· - 00

t.<.N

Mortality patterns in the block lava site (fue age 15 years) and in the site of grazing are of the same order of magnitude as in the undisturbed situation . In Grazed 2, constant morrality rate of 21% per height class of 20 em was determin ed; this is to that of the tallest recently burned individuals and almost twice the natural mortality rate.

z'C'

Q'\['\Olf )"'::tV'lC ""tt"'t-.. ::tN-.-

5.3.3 Concluding remarks

The effects of fire and grazing on the stem rosette populations are summar ized in the scheme in Figure 5.3. ' ·

.,_

.... .

· - 00

t.<.N

The fact that stem rosettes arc not consumed, except occasionally the inflorescence, is· apparently no guarantee of their survival. Grazing by cattle has a major impact on the adult mortality of Espe/etia hartwegiana. Increased clustering of individuals, while interpreted as a result of trampling impact, at the same time provides protection against trampling and scraping by cows. Reports on afroalpine giant rosettes show that predatio n · by hyrax on Lobelia telekii (Young, 1985) and by elephants on Senecio keniodendron or Dendrosenecio (Mulkey eta/., 1984) also affected the larger individuals more. The main impacts of frre on Espeletia populations can be summar ized as follows: most juveniles (i.e. over 80%) are killed immediately after a fire, though this height class is quickly replaced by individuals benefitting from the extra amount of space, light, and nutrients available. Whether fire activates··the soil seed bank or stimulates the ripening of Espeletia · seeds is not known. Adults are subject to increased mortality because of fire damage. Silvertown (1983) mentioned the possible occuirence of peaks in age structures by the truncation of certain age classes. In this case, truncation of older age classes due to fire does indeed 'occur, whereas truncation of the younger age classes is partly compensated for by a more rapid replacement. On those parts of the laterdl moraines where a fire frequency of once every 5 to I0 years is common, stem rosettes are able ro survive despite a significant decrease in their average height and the higher adult mortality mentioned above. The increased growth rates and lower maximum turnover period arc an effective response to the fire regime. These mechanisms are clearly different from those listed by Silvertown (1983); populations may be cushioned from local disturbances by the immigration of plants from unaffected areas or by changes in fecundity or mortality in the populati on itself.

108

§ z

@~~~t--'<I"N--0

'2. C!C!~~~~C!~~C!

~ NN0ic:iMN(::i;ic::iN

z'C' g

0'\0\00V 'IV")\0\0 V'IV'III" lV)'I:t'C ""l(""")N

000000 ::00000 000000 000000 000000

~ ~ ~~~~~~~~~~~~~~

z'C'

N~ ~\01.1"1~~\()~~~NN~ ----

::: ::: ::: ::: ::: ::: :

~ --~---- ---~ -~---~--

.·.

Chapter 5

Effects on plant populations

Apparently, E. Jumwegiana possesses certain unique fire-adaptive ttaits which increase capacity of populations !Q.S.Ul'V ive. However, there are definitely limits to the stress be endured, especially in the case of combined burning and high-in tensity grazing.

· 5.4 Methods of studying bunch-grass populations S.4.J Population parameters

NATURAL CONDITIONS (under constant climate)

ESPELETIA population

The two parameters best reflecting the age of a Ca/amagrostis elfusa or C. recta tussock are height and diameter. All possible parameters describing the life stage of a rossock change with .P·age, but they are also affected' by grazing and burning.

constant growth rate constant monality

0-f '

FIRE

AND/OR

r--- ---- -.-- -----, growth rate

monality rate

density

GRAZING

adults

juveniles

aduhs

juveniles

stimulated (heightdependent)

stimulated

unaffected (?)

increa~ed

(heightdependent) decrease

average height

unaffected (?)

clustered distribution

absent

Fourteen different sites were selected, following the same procedu re as described for the stem rosettes. Ten of these sites had been selected for the study of Espelet ia population structun::s (see Table 5.1); the four additional locations without stem rosettes are described in Table5.4. All sites correspond to the Calamagrostietwn effuso - rectae associa tion.

Site

Fire age (yr)

Grazing intensity

Aspect

_Slope (0)

Terrain rorm

Tussock cover

Bare soil (%)

(%)

initially high

s1r0ng decrease

increased, _around adult stems

increased (constant)

increase <100 em decrease ~ 100 em

slight decrease

undisturbed Undist. 4

Undist 5

burned Fire 1-2

1.5

Fire4

Table 5.4

decrease almost exclusively present

Figure 5.3 Schematic representation of the general effects of fire and grazing hartwegiana populations.

SE 3

2 10

<5 <5

33 21

GL LM

15 35

20 35

Additional site descriptions for the study on Ca/amagrostis effusa and C. recta populations. Grazing intensity: 1 =very low; 2 = low; 3 = medium. GL = glaciated lava field; LM = lateral moraine.

A plot was selected in each of the 14 sites a plot was selected and the following chantcteristiCs were measured or recorded for a total of 200 individu als,: (1) total height, measured from ground level to highest leaf tip; (2) diameter at ground level, both of the old (fragmented) tussock base and, if present, of separate tillers or 'sub-tussocks'; (3) proportion of dead bunch-grass cover il'l relation to total bunch-g rass cover; (4) number of sub-tussocks per fragmented tussock, counted separate ly when the nearestneighbour distance was more 'than 3 em; and . . (5) - level of fragmentation, divided into four classes: not, slightly, moderately, and highly fragmented. When traces of the old tussock base were still visible between the fragmented new tussocks, ·they were considered part of one tussock. This method implies that at sites with more trampling impact, and thus a more rapid disappearance of old tussock bases, the average · diameter was lower.

110

lll

CJwpcer 5 For the population struciUres of the bunch grasses it was impossi ble to convert he~gh~ diameter classes into ~ge classes. Average height and diameter were used to charactenze population stnJctures as nonnully distributed. Tus~ock .volume and the ~ea~ive ratio used in combination as basic parameters in the esumauon of forage ava1labJll!y (see & Verweij 1992) and of fuel load.

Diameter distribution The 14 sites were divided into five different groups according to their diameter w,,,.,u,uu01 The student's t-test (Sakal & Rohlf, 1981) was used to determi ne significant differences between the mean diameters. If the frequency was not nomtally distributed, shape of the distribution was also taken into account.

Effects on plane populations - However, clipping the vegetation means simulating grazing pressur e. Second, removal of litter , influences micro-climatic conditions and, therefore, production (Wiegert & Evans 1964). In this study, itwa~ attempted to measure production in an alternat ive way. Fourteen tussocks of both species were selected in seven different sites, covering the complete range of tussock volumes. Leaf production and leaf mortality was monitored for four months, from May to August 1990. Every selected tussock was fenced in such a way that rabbits and cattle were unable to reach the leaves. All litter within the fence was collecte d before and after the experiment and dry-weighed. All the green leaf tips were marked with water-resistant red paint. Four months later, the whole tussock was clipped ut ground level. After division into different groups, the plant material was dried and weighed in the same way as described ~~ -

5.4.2 Biomass A panly destructive method was used to measure biomass. This method was deri~ed digital picture processing as described by Roebertsen et al. (1988). A number of shdes taken from the vegetation as vertical profiles against the neutral background of a 2 screen. They were taken from ground level, each one representing an area of 1 * 1 m to a horizontal depth of 1 m. ,One representative plot per site was harvested as a Nonnally, the grey-tone values of rasterized slides are related to the bi~mass However, linear regression correlations proved to be higher if the volume tnc data photographed tussocks were related to the biomass figures instead of th~ tussock. measured with a digital planimeter. Tussock volumes were treated as cyhnders. Shde were taken at nine different sites, covering all important manage ment situations. The of slides taken per site was eight, following the recommendatio n of Lutz &·.Vader ( who used eight samples aS' the optimal number in a clipping experiment on ~a~o v.'....~.. ,.~, grasses. Harvesting was done by clipping the vegetation at ground level and dtvtdmg ll five groups: (I) living tussock grass leaves, (2) dead tussock grass leaves, (3) forbs, (4) grasses, and (5) shrubs. Furthennore, (6) litter and (7) bases of tussock leaves were After division into different groups, the plant material was dried in an oven at 70° C for hours, and dry weights were subsequently detennined.

Four different categories of leaves were present: (I) green leaves without a red mark, equal to newly produced leaves; (2) green leaves with a red mark; (3) dead leaves without a red mark; and (4) dead leaves with a red mark, i.e. senescence over the preceding four months. Categories 2, 3, and 4, together with the litterfall, represent the biomass at the beginni11g of the experiment. The sum of categories 1 to 4 gives the biomas s at the end. The assumption is that the turnover time of Ca/amagrostis leaves is more than four months. Four plots that were monitored had been clipped to ground level one year before. In this way, annual biomass , production of disturbed plots could be compared with that of the four-month period. Lutz & Vader (1987) found no significant differences in production betwee n the drier and the more humid periods; this was confinned by Hofstede (1995) for the study area. Data were therefore extrapolated to annual figures.

· 5.5 Management impact on bunch-grass populations 5.5.1 Population parameters In Table 5.5, the mean values of population parameters are presented for different management situations. The site descriptions arc given in Tables 5.1 and 5.4.

5.4.3 Productivity Production can be measured in several ways following the general equation given by Perkins et al. ( 1978): P = (B2-B 1) + L + G where p =production. B2-Bl =change in biomass. L =litter fall, G =loss due to grazing. The change in biomass can be estimated by harvesting and comparing di~ferent. p~ots ~~ · different moments in time. In paramo bunch grasslands, howeve r, the spaual v.ru:~uo? of biomass is usually large in comparison with the temporal variatio n. Another poss1bthty ~~ \o, exclude grazing and to measure change in biomass and litter fall after clipping the vegetauon and removing the litter at t = 0 (Cardozo & Schneller, 1975).

· Diameter and height For the tussock base, a mean·diameter of 20 to 25 em was found in undisturbed situations and under low-intensity grazing. In recently burned situations, diamete r decreased to 16to 20 em. In the burned and grazed plor of Fire 2, mean diameter was only 12 em. Under intensive grazing (Grazed 2), mean diameter was surprisingly larger than in natural situations. Lateral tillering might be increased as a result of trampling impact. A similar tendency was observed regarding height. At undisturbed sites, a mean height of 37 to 61 ern was found, with maximum values up to 130 em. [n recently burned situations, mean height was reduced to 12 to 24 em. When grazed as wel,l (Fire 2), a further reduction to 8 em was noted. In grazed situations, mean height is slightly reduced . ··

11 2

113

Chaprer 5 Site

..:-.

Density Mean height (n m·2) (em)

Effecrs on plant poptdarions

Mean diameter (em)

Fragmentation

(%)

Undist. l Undist. 2 Uhdist. 3 Undist. 4 Undist. 5

3.2 2.7 1.3 3.3 1.6

41.0 ± 2.5.1 37.1 ± 18.8 4.5.8 ± 22.4 60..5 ± 20.6 45.7 ± 2Q.8

20.8 ± 14.4 24.6 ± 13.3 2.5.0 ± 1.5.1 25.8 ± 12.3 19.9 ± 10.9

13.2 26.7 20.1 11.5 15.2

Fire Ia Fire lb Fire 1-2 Fire 2 Fire 4' Fire 6 Block lava

5.7 6.1 3.6 16.7 13.3 8.0 5.4

11.9 ± 4.2 13.1 ± 5.5 23.5 ± 9.8 8.3 ± 2.6 21.6 ± 9.1 37.7 ± 18.8 44.6 ± 25.5

17.0 ± 7.8 17.6 ± 7.3 26.2 ± 13.1 11.8 ± 6.8 16.2 ± 8.5 22.7 ± 13.9 20.8 ± 14.8

96.0 9-2.6 80.1 75.3 55.5 28.5 19.1

Grazed 1 Grazed 2

0.4 1.0

29.7 ± 8.8 34.2 ± 26.9

25.0 ± 9.9 34.1 ± 16."1

27.4 59.8

In recently burned situations where fire had destroyed all dead biomass one to four years before, the percentage of dead biomass was 17 to 35%. In grazed situatio ns, the_dead fraction , . wliS slightly lower. This is explained by the stimulating effect of grazing on production, and IJy the fact that many leaves are consumed before they reach the stage of senescence.

Diameter distribution ·The following five groups were fonned according to diameter distribu tion: .:_.1 Undisturbed or recovered vegetation: Undist. 1, Undist. 5, Block Lava, and Fm 6. . II Zero grazing at sites with a grazing history, or low-intensity grazing : Undist. 2, Undist. 3, Undist. 4, Grazed 1, and Fire 1-2. Recentl '?· III y burned sites, occasionally grazed: Fire Ia, Fire Ib, and Fire 4. '· -: IV Intensively grazed vegetation: Grazed 2. v Recently burned vegetation under intensive grazing: Fire 2.

500

Table 5.5 Maid -characteristics of Calamagrosr,is tussock populations in management situations. Fragmentation (%} =percentage of tussock s with more sub-tussocks. Mean diameter and mean height reflect the short-term impacts of burning and grazing. effects are visible immediately after fire, and rapidly recover at zero grazing. Four to six after a fire, mean height and diameter had returned to pre-bum levels.

Density and fragmentation Undisturbed sites showed a tussock density of 1.3 to 3.3 m·•. At grazed sites, density slightly reduced, while at sites with a fire history tussock density was increased (3.6 to m·2). The highest figures of tussock density corresponded to the lowest figures of diameter - at sites Fire 2 and Fire 4. These numerous small tussocks were products fragmentation. The percentage of fragme nted tussocks varied between II to 27% undisturbed plots. The fragmentation percentage is 93 to 96% one year after a lire, decreasing to 56% four years after a fire. High grazing pressure also stimulates fragmentation , (~ce the figure of 60% for fragmented tus'socks in Grazed 2).

E 0 0

300

1--

.,....

-.s ~

'iii 200

0 100-1- -----l f-

2.5

7.5 12.5 17.5 22.5 27.5 32.5 37.5 42.5 47.5 52.5 >55

Diameter (em)

ntation. Fragmentation and of the tussock s recover more slowly than height and diameter of the tussocks; the regeneration time is estimated at more than six years.

114

·~ ·· Height, diameter, and density are strongly related to fragme

Dead biomass In natural conditions on slopes, the percentage of dead biomass was 61 undisturbed site in concave terrain (Undist. 4) had a slightly lower value of 51%.

400

group I -

group Ill

group v

_Figure 5.4 Distributions of tussock-base diameter of Calomagrostis ejfusa and C. recta of group l (undisturbed), group li,I (recently burned) and group V (recentl populations y burned and grazed).

115

Effects on plane populations

Chapter 5

·c. A high grazing pressure combined with a grazing In Figures 5.4 and 5.5, the average frequency distributions of the differen.t groups history and no burning (group IV) was are lated to the-almost complete disappearance of small tussocks (diameter <IS_ Groups m to V display the different reactions of the tussock-base dta~etcr em). Density t~ fire grazing. Recent fire without grazing (group III) was related to a decrease m · :creased without the development of new recmits: in fact, the smallest-dia~eter maxtmum class is mean diameter, but since hardly any grazing took place, recovery towards the un<ltsturbedlifit.l . missing entirely. -Tussocks with a diameter . s~aller than 25 c_m that survtv_e are o~ten fragmented due to cow trampling and consumption, an4 are subJect to strong situation was rapid. Recent fire in combination with grazing (group V) was mterspeeific more The tussock gra sses still persisted and were numerous, but with strongly reduced ' competition with the shon grdsses. di Tussock-base diameters over 30 em were rare. The larger tussocks (diameter >45 em) are present in relatively high numbers . Larger tussocks with a higher proportion of dead biomass enjoy a cenain advantage: due The distinction between groups I ~d JI is based mainly on the considerably to their low higher in the smaller-diameter classes in group I. The lower densities of juvenile tussocks palatllbility, they are rarely consumed (Schmidt & Verweij 1992). The presence of palatable in sholt grasses and forbs and the decreased intraspecific competition make this II are probably caused by a higher cover of ground-covering species (due management to grazing i~ past), leaving less space for germination. In the case of Fire 1-2, the scarcity of. small tussocks~m -regime favourable for larger tussocks. However, due to the lack of recruitment of juveniles, · in the long term the tussocks disappear from the system if grazing is sustained, can also be attributed to the additional effect of fire. This mechanism demonstr as could be ates ~mma•mc:s with the behaviour of Espeletia juveniles. The distribution of group I is compara observed throughout the study area. ble to found by Sala et al. (1986), though that study concerned a species with a smaller diamelte~J~I ·,oi~eter distribution reacts in different ways to grazing and burning. It can be conclude and a greater mean height. d that the distribution of tussock-base diameters, in particular, provides more long-term information ·on the state of the tussock populations in relation to management regime and management history. The process of tussock-fragmentation due to burning and grazing is shown 80~-------------------------------------~ in Figure

5.6.

5.5.2 Biomass 60

N' E 0 0

T"" ...._

.s c '(j)

The relationship between the volume of the tussocks as calculated from the slides and .biomass is:

•· ~

B =c * V c = 15.09 g dm'3

Equation 55

where B =biomass of standing crop (g m·1 ); V = tussock volume (dm3 m·1 ); r"" = 15.09 ± 0.77;

· n 9; R2 = 0.95.

(!)

ol---~~~WJ~~~~~~~~~~~

~~~~

12.5 17.5 22.5 27.5 32.5 37.5 42.5 47.5 52.5 >55

Diameter (em)

group 1

group II

~ group IV

Figure 5.5 Distributions of tussock-base diameter of Calamagrostis ef[usa and C. recta populations of group I (undisturbed), group II (low-intensity grazing), and group IV (intensive .· grazing). Note the difference in vertical scale in comparison with Figure 5.4. 116

Volume calculation using the data of 14 harvested tussocks provided a similar relationship. though with a lower r , • Volume or surface calculations based on slides 1 are a good alt~rnative to entirely destructive methods of estimatin g the aboveground biomass of tussock grasslands. The method requires little time in the field and provides a good basis for : extrapolation. When more time is available, biomass can be calculated by measuring the . height and diameter of the individual tussocks in the field. . The heterogeneity in biomass of the paramo grass vegetation is considerable, ·even within one site. In Undist 1 and Fire 6, the plots with the highest mean biomass, values ranged from 2 2 0.376 kg m· to 1.423 kg m· • Coefficients of variation of undisturbed and lightly grazed situations were 21 to 43%. Lutz & Vader (1987) measured comparable coefficie nts of variation of 20 to 44% for undisturbed sites. Variation at sites with a grazing history was muc·h higher, due to the fact that tussock grasses were absent in some of the e_ ight 1 * 1 m2 plots. In natural situations, tussock biomass repeatedly reaches values of more ~han 1 kg m·2, • mean biomass being 0.94 ± 0.26 kg m'2 0

117

Chaprer 5

Effects on plant populations Modt:rate grazing intensities inl1uence biomass to a limited extent; mean biomass remains at a high level. Higtrgrazing intensities with or withou t burning re!iulted in a strongly reduced tussock bioma ss: 35 ± 12 g m 1 in Fire 2 and 29 ± 39 g rn·2 in Grazed 2. The effect of a single fire event without grazing is very lim ited, as recovery can be rapid regarding the maximum value of 0.574 kg m·2 of a recovering plot located in an exclosure of Fire 2. The biomass values of sites rire 4 (0.49 ± 0.10 kg m·2 ) and Fire 6 (0.77 ± 0.28 kg m·2 ) are similar to moderately grazed sites without burning.

1' I Recovery I I

5.5.3 Productivity

Grazing

It appeared that the change in bioma ss of the painte d tussocks is roo small to measure in a four-month period. No significant net accumulation of green biomass was measu·red, the .- differe nce between newly produced leaves and senescence being either slightly positive or slightly negative. The production of new tussock leaves over 120 days ranged from 1.1 g for a tussock of Grazed 2 to 20.4 g for a big Fire 6 tussock. The amou nt of living biomass present accoun ts for a large part of the variation, as more living leaves represent more photosynthetic active tissue (McNaughton, 1985). The following relationship between green biomass and production was found:

P = 14.4

Equation 5.6 where GB =green biomass at t = 0 (g); P =production per tussock (g./120 d); R1 =0.72; r, = 14.4 ± 2.7; n = 14. GB of measured tussocks was 15-300 g. 1

-1'

I Recovery

I I

* log(GB) - 13.6

Grazng

For tussocks with agreen biomass of 15 g to 300 g, the production range was 3.3 to 22 g /120 d. The productivityyer tussock is, in the first place, determined by the factor of present green biomass. Management impact is related to a decrease in tussock volume and biomass, which causes a lower production of individual tussoc ks.

Production per rrl The additional figures for the production of green biomass in 12 months did not differ significantly from the fo ur-month period, and these yearly production figures were therefore taken into account. Based on tussock size and number, figures were translated into a production estimate per m2 of 208 ± 128 g m·2 yr' 1 (n = ll) for undisturbed plots. In a similar way, a production of 361 ± 26 g m· 2 yr' 1 (n = 12) was calculated for recently burned or clipped plots. The ranges of production values were 117 to 318 and 347 to 394 g m2 yr'1 , respectivt:ly. The mean production pt!r m2 of undist urbed sites proved to be significantl y lower (P <0.05) than the production of grazed and recent ly burned sites.

Figure 5.6 The process of tussock fragmentation under the influence of burning and grazing. The · final stag_e is a short matted vegetation of grasses and forbs.

Due to the fact that tussocks are largely fragmented in recently burned situations, the mean amount of green biomass per tussock is lower than in undisturbed situations. At first sight, a lower production would be expected. However, it can be deduced from equation 5.6 that several small tussocks produce more than one large tussock with an equal total amount of green biomass. The lower total green biomass per surface unit is appar~ntly less important than the positive effect of the change in structure after burning. The decreased height and · smaller fraction of dead biomass allow more light to reach the active photosynthetic tissues . 119

Effects on plant populations

Chapter 5

0.9 0.8

@' 0.7

E ..._ Ol

c. (/) (/)

0.6 0.5

0.4

m 0.3 0.2

0.1

0 10

Fire age (yr)

•

Fragmentation (% +

Recovery after fire . . . . ce fire ages are cswblishctl with the aid of acn;1l photographs and plots wuh different fire S10 . . b I . I . g pressure -': es are available. th.: recovery after fire at sites wnhout su sequent ug 1 grn~n _ag •~ quantified. The plots with lire ages of l, U, 4, and 6 years were used, complemented cai~Undist. 1 to represent total recovery. The resuhs are_ presented in Fi~ure 5.7, in whic_h , ~ undisturbed situation was assumed to be equal to complete recovery alter 10 years. Thts _t e mption is based on extrapolation of the data on reccntly burned plots and takes into . as~~unt the characteristics of the recovered block lava vegetation with a fire age of 15 years. - ~~rtherrnore, a staning biomass of zero and a fragmentation of 100% immediately after the fire were assumed. . The recovery pauerns of fragmentation, height, and dead biomass show ~imilari ties with a - sigmoidal cu':"~· m~st recovery taking plac~ between one and five years after fire. ~h~ increment in hvmg btomass, on the contrary, 1s continuous t~roughout th~ years. The grazmg : influence of site Fire 4 is expressed in a lower mean hetght and a htgher percentage of fragmented tussocks than expected. - The linear-increase in living tussock biomass enabled the calculation of_ a mean fi~ure of green biomass accumulation, 25.4 g m'2 yr'': This low figure explams w_hy btomass accumulation could not be measured over a penod of 120 days. Net accumulation of green biomass is constant, though production increases with an increasing green biomass (equation 5.6). This can be explained by a parallel increase in the rate of senescence and by the changes in bunch-grass structure described above.

5.5.4 Concluding remarks Figure 5.7 Recovery of Calamagrostis effusa and C. recta populations after a lire without grazing pressure (based on the plots of Fire Ia, Fire lb, Fire 1-2, Fire 4, Fire 6, and Undist. 1). The input of an extra amount of available nutrients (Hofstede, 1995) could be a imponant reason for the higher production of grazed and recently burned plots, although this seems only to play a role in the first year after a fire.

Turnover time The calculated production range can be used to calculate the turnover time of an average tussock or leaf by dividing green biomass by daily production: TT = GB I [(14.4

* log(GB) - 13.6)/120]

where TT =turnover time (day); GB =green tussock biomass (g) According to equation 5.7, the production figures were translated into a turnover time of 1.5 years for the smallest tussocks (GB = 15 g) and 4.5 years for the biggest (GB = 300 g).

120

Table 5.6 demonstrates that the values of total tussock biomass and dead biomass fraction in this study are consistent with those of Hofstede er at. (1995b) and with values for other paramos in the Cordillera Oriental of Colombia as regards C: effus_a (Lutz & Vader, 1987; Beekman &·Verweij, 1987; Van Groen, 1987). Hnanuk (1978) mvesugated tussock grasslands (Deschampsia klossii) on the equatorial high mountain of Mount Wilhelm, New Guinea, with study sites at elevations of 3200 to 4350 m. Mean height, density, and the percentage of dead leaves all demonstrate a great similarity with our fig ures. Hofs tede et al. ( l995a) estimated production based on leaf elongation rates, and repon a similar production of 198 g m'2 yr' for undisturbed conditions. As data are based on , projected tussock cover, the figures for disturbed situat:ons with a more open cover cannot be generalized. Production values reponed for undisturbed paramo grasslands generally range from 0:11 to 0.36 kg m·2 yr' (Lutz & Vader, 1987; Beekman & Verweij, 1987; Hofstede et a/., l995a; this study). A similar range is reported for Mount Wilhelm, New Guinea, where Hnatiuk (1978) measured a production of 0.13 to 0.44 kg m'2 yr1 in undisturbed sites. An aboveground production of 0.20 to 0.26 kg m·2 yr'1 is reported for the alpine grasslands of the Himalaya (Sundriyal, 1989; Sundriyal & Yoshi, 1990). In comparison, the American prairie and the tropical lowland grasslands have higher production levels, often more than 1 kg m·2 yr·• (Hadley & Kieckhefer 1963; Singh & Yadava, 1974; Abrams et al., 1986). In tropical lowlands, production generally increases with rainfall (McNaughton, 1985), whereas production in the paramo is primarily temperature-limited. 121

Chapter 5 Tussock density

Effects on plant populations

%Dead biomass

Total biomass (kg m· 2)

Produt:tion (kg rn·2 yr' ')

84-89

2.3-3.4

0.13 -0.44

0.8-1.3

0.1\-0.36

0.4-0.5

0.22

Van Groen, 1987 id., diswrbed

70 30

1.0-1.2 0.1-0.4

Los Nevados N.P. Hofstede, 1995 id., disturbed

68-73 59

0.51 0.07-0.15

0.20

60-80 30-60

0.6-1.2 0.05-0.3

0.12-0.32 0.33-0.40

Reference

(m-l)

New Guinea Hnatiuk, 1978

2-7

Lag . Verde, Colombia Lutz & Vader, 1987 Beekman & Verweij, 1987

11.5

The common nmnagcment practice of burning and subsequent grazing leads to sustained fragmented tuss<x:ks, with a low heig ht, diameter, and biomass. This is also the stage in which tussocks are mnst vulnerable and the disappearance of sub-tus~ocks is poss ible. Production, however, is increased . When grazing activities stop, complete recovery is possible. When, on the other hand, gmzing does not ceas e, short matted herbs that arc bette r adapted to cattle grazing gradually occupy more spac e. If fire docs not occur at this stag e, a matted herb vegetation may develop. Recovery is inhibited as tussock recruits rarely brea k this dominance. The shon matted gra~s Aciachne 'acic ularis is a major determinant. This species is common in sites Grazed I, Grazed 2, Undist. 3, and Undist. 5. Its resistance to flre is low (Verweij & Budde 1992). Furthermore, its limited zooc horous dispersal ~ystem limits expansion into other areas. Tussocks that have disappeared through grazing impact are not repl aced by recruitment due to the physical barrier imposed by the densely growing shon grasses. This explains the large -extensions of short matted gras slands of Aciachne acicu/aris - Cala magrostis coarctata (type gF of the vegetation map, Figu re 3.4) in non-burned areas, despite the often low grazing intensities. This vegetatio n is merely a product of grazing history, maintaine d by a low actual grazing pressure.

Verweij & Kok; 1995 id., disturbed Table 5.6

1.3-3.3 3.6-16.7

LITERATURE CITED

Comparison of main characteristics of bunch grasses of different tropical ecosystems.

. Abrams, M.D., Knapp, A.K. & Hulbert, L.C. 1986. A ten year reco rd of aboveground · biomass in a Kansas tallgrass prairie: effects of fire and topographic position . American Journal of Botany 73 (10): 1509-151 5. Allen, R.B. & Partridge, T.R. 1988 . Effects of spring and autumn·fires Com lete recovery of biomass- and on the composition of Chionoch/oa rigida tussock gras . P_dal curve and is estimated tussock size after fire to pre-bum levels sland, New Zealand. Vegetatio 76: to take place in about 10 years. Red 37-44. Beek man , A.M. & Verweij, P.A. 1987. Structure uced inr_•~rsiJec·ifiC:,:r.t ~~:;~tition and 'a more open structure facil and nutrient status of a pacamo bunc itate a rapid recovery of the tussock vege hgrass tatio n in relat ion to soil and climate. MSc. thes gra .:s, f he s stem is not being grazed. Van is. Interaal Report 233, Hugo de Groen (1987) reported a tota~ reco Vrie s-La bora tory, Univ t ersit very yb d y of Amsterdam. ~~ o Cardozo, H. & Schnetter, M-L. 1975 years, ase on both diameter and heig ht. In the case of a Ch10nochloa ngulil lu~:>l )l;''i.."l . Estudios ecol6gicos en el Paramo I d in New Zealand at an elev-a· eriod of de Cruz Verde, nine years Colombia. Ill. La biomasa de tres grass~ b Allen & Partridge (1988).tion of 668 m, a recovery p asociaciones vegetales y Ia prod Fire was reponed to stimulatt! the uctividad de Cala niagrostis effusa (H.B.K.) Steud. grm~th of ~~;~~~~s a~d may lead to increased genninati y Paepalanthus columbiensis Ruh comparaci6n con Ia concentraci6n on and the establishment of seedhngs l. en de chlorofila. Caldasia 11(54): 69-8 1965a; 1965b). 3. · Cavelier, J., Machado, J.L., Vale ncia , D., Montoya, J., Laignelet, A., Hurt ado, A., Varela, A. & Mejia, C. 1992. Leaf demography and growth rates of Espe/etia barc The response of Calamagrostis tusso · d · g can be summarized layana Cuatr. cks to burnmg an graz m. (Compositae), a caulescent ros~tte in a Colombian pacamo. Biotropica . . follows. The level of fragmentation 24: 52-~3. Collins, S.L. 1987. Interaction of distu .and tussock diameter are considered. rbances in tallgrass prairie: a field expe the st'"'"'''"r:u• ararneters The effect of a fire is riment. Ecology invariably the complete fragmentatio ~o... 68(5): 1243-1250. n o p ks. Fra entation consists of a chai .Garay, I:, Sarmiento-M., L. & Mon n of dependent proces.ses:. a tusso_ck base does asterio, M. 1983. Le paramo dese ~~~~~crate enrrely, it Splits up by fonn rtique: elements biogenes, peuplements des microarth ropodes et strategies de survie de Ia become new independent small tusso ing lateral tillers. ~~ g~azm~ ISdSUSt.~ln~~· t~~es'"e~~n;:kn~ vegetation. Pp. cks. The number an t us t e e~sl 127-134 in Lebrun, Ph., Andre, H., De Medts, A., Gregoire-W., C. & Wau .Y increase . Thi's fragmentation is temp thy, G. (eds.), orary and recovery ca~ _take ~lace ~lth Proc eedi ngs of the Vlll Inte rnational Colloquium of Soil Zoo ~n a~ou t. . d bove The effect of moderate logy , as menuone Lou graz vain ing -la-Neuve, inten Aug a sltle . 30-Sept. 2, 1982. Dieu-Brichart, Ottig s Without fire IS mmor. · slightly fragment, but . nies-Louvain-la-Neuve. · h' the total biomass of the standwg Goldstein, G. & Meinzer, F.C. crop remains 'gh. 1983. Influence of insulating dead leaves and lower temperatures on water balance in an Andean giant rosette plant. Plan t, Cell & Environment 6: 649-656. 122

123

Chapter 5 Hadley, E.B. & Kieckhefer, B.J. 1963. Productivity of two prairie grasses in relatio n to fire frequency. Ecology44{ 2): 389-395. . . . . 1969. Growth rates of the East Afncan g1 ant SeneCIOS. Nature 22: 163-164. Hedb erg, 0 . . I h" h Hnatiuk, R.J . 1978. 1he growth of tussock grasse . d t s on an -\~~ sub-antarctic islands. Pp 159-190 in: Troll, C. &cq~a-tona .:Vg (~ou)nt;~n a~. ~...aucr, . e s. .' cocco1 g~ca relations between the southern temperate zone and the tropical mountams. Erdwissenschaftliche Forschung 11, Steine r-Verl ag, W1esbaden: 1995 Effect s of burning and grazing on a Colom b1an paramo Hofstede, R.G·M· · PhD dissertation, University of Amsterdam, 199 pp. . . Hofstede, R.G.M., Chilito, E.J. & Sandoval, E.M. 1 99~a. ~egeta~1ve structu re, rmcroc and leaf growth of a paramo tussock grass spec1e s m und1sturbe~, burned an~ conditions. Pp. 21-38 in: Hofstede, R.G.M. 1995. Effects_of burnmg and graz111g on Colombian paramo ecosystem. PhD dissertation, Univers~ty of Amsterdam. Hofstede, R.G.M., Mondrag6n, M.X. & Rocha, C.M. 1995b. Biomass of gra~ed , bume_d, undisturbed p<hamo grasslands, Colombia. I. Above ground vegeta uon. Arcnc Alpine Research 27: 1- 12. .. . Hulbert, L.C. 1969. Fire and litter effects in undist · urbed bluestem prame m Kansas. Ecology. 50(5): 874-877. Karunaichamy, K.S .T.K. & Paliwal, K. 1989. Prima . f ry productivity and transfer dynarrucs o grazing lands at Madurai, southern India. Tropic al Ecotog_y_30(1 ): 111-117. .. · R w 1954 Effects of moderate grazing on the composmon and plant produwon of KeIung, . . . a native tall-grass prairie in central Oklahoma. Ecolo 207 gy 35(2): 200- · ' Lozano-C., G. & Schnetter, R. 1975. E~tudios eco16~icos en el Paramo de Cruz Verde Colombia. II. Las comunidades vegetales. Calda sta 11(54 ~: 53-68 . . . . Lutz, R. & Vader, P. 1987. Biomass, productivity , and nutnent status ~~ a Colombian bunch grass paramo. MSc. thesis. Internal report 229, Hugo de Vnes-Laboratory, Unive rsity of Amsterdam. _ . . .. . Malan son, G.P. 1984. Fire history and patterns of Venturan subassoctatlons of Cahtomtan coastal sage scrub. Vegetatio 57: 121 -128. . . Mark, A.F. 1965a. Flowering, seeding, and seedh ng establishment of narrow-leaved snow tussock, Chionochloa rigida. New Zealand Jou_m al of Botany 3: 180-193. Mark, A.F. J965b. Effec ts of managemen t practi ces on narrow-leaved snow tussock, Chionochloa rigida. New Zealana Journal of BOtan y 3: 300-3 19. . .. . McNaughton, S.J. 1979. Grazing as an optimization process: grass-ungul ate relauon~htp s m the Serengeti. American Naturalist 113: 691-7 . McNaughton, S.J. 1984. Grazing lawns: anima 03. ls in herds, plant fonn, and coevoluuon. American Naturalist 124: 863-886. • . . McNaughton, S.J. 1985. Ecology of a grazin . g ecosystem: the Serengeu. Ecolog 1cal Monographs 55(3): 259-294. . . Menti s, M.T. & Tainton, N.M. 1984. The effect of fire on forage produ ~uon and quaht y. P~245-254 in: Booysen, p_ de v. & Tainton, N.M. (eds.), Ecologtcal.eff~cts of fire m South African ecosystems. Ecological Studies 48, Springer-Verlag, B: rhn. Monasterio, M. 1986. Adaptive strategies of Espel etia in t~e Andea~ dese_rt paramo. Pp. 49-80 in: Vuilleumier, F. & Monasterio, M. (eds.), High altitud e troptcal b10geogn1phy. Oxford University Press, New York. Monasterio, M. & Lamo tte, M. 1989. Les popul . ations d' Espe/e~ia timotensis dans le param o desertique des Andes du Venezuela. Rev. Ecol. (Terre Vte) 44: 201-227.

124

Effects on plant populations Mulkey, S.S., Smith, A.P. & Young, T.P. 1984. Predation by elephants of Senecio .ke11iodendron in the alpine zone of Mount Kenya . Biotropica 16: 246-248. Perez, F.l .. 1992. The ecological impact of cattle on caulesccnt Andean rosettes in a high Venezuelan paramo. Mountain Research and Devel opment 12( 1): 29-46. Perkins, D.F., Jones, V., Millar , R.O. & Neep, P. 1978. Primary production, mineral nutrients and litter decomposition in the grassland ecosystem. Pp. 304-331 in: Heal, O.W. & Perkins, D.F. (eds.), Production ecology of British moors and montane grasslands. Ecological Studie s 27, Springer-Verlag, Berlin. Riney, T. 1963. A rapid field technique and its applic ation in descri bing conservation status and trends in semi-pastoral areas. African Soils 8(2): 159-258. ~oebertse~, H., Heil, G.W. & Bobbink, R. 1988. Digital piciure processing: a new method to analys e vegetation structure. Acta Botanica Neerla ndica 37(2): 187-192. R_osen, E. 1982. Vegetation development and sheep grazing in limestone grasslands of South 61and, Sweden. Acta Phytogeographica Suecica 72: 157-172. Sala, O.E., Oesterheld, M., Le6n, R.J.C. & Sorian o, A. 1986. Grazi ng effects upon plant · community structure in subhumid grasslands of Argentina. Vegetatio 67: 27-32. Schlatterer, E.F. & Tisdale, E.W. 1969. Effects of litter of Artemisia, Chrysothamnus, and . Tortu/a on germination and growth of threeperenn ial grasses. Ecology 50(5): 869-873. Schmidt, A.M. & Verweij, P.A. 1992. Forage intake and secondary production in extensive livestock systems in paramo. Pp. 197-210 in: Balsle v, H. & Luteyn, J.L. (eds.), Paramo: an Andean ecosystem under human influence. Acade mic Press, London. Silvertown, J.W. 1987. Introduction to plant popul ation ecology (2nd edit.). Longman Scientific & Technical, Harlow, and John Wiley, New York. 229 pp. ·. Singh , J.S. & Yadava, P.S. 1974. Seasonal variat ion in compo sition, plant biomass and net primary productivity of a tropical grassland at Kurukshetra, India. Ecological Monographs 44: 351-376. Smith , A.P. 1979. The function of dead leaves in Espei etia schul~zii (Compositae), an Andean giant rosette plant. Biotropica 11: 43-47. Smith, A.P. 1981. Growth and population dynam ics of Espeletia (Compositae) in the Venezuelan Andes. Smithsonian Contributio ns to Botany--48: 1-45. Smith , A.P. & Young, T.P. 1987. Tropical alpine plant ecology. Annual Review of Ecology and Systematics 18: 137-158. Sobrevila, C. 1989. Effects of pollen donors on seed formation in Espeletia schultzii (Compositae) populations at differe nt altitudes. Plant Systematics and Evolution 166: 45-67. Sakal, R.R. & Rohlf, F.J. 1981. Biometry (2nd edit.). W.H. Freeman & Company, New York, ~w . . Sundriyal, R.C. 1989. Efficiency of energy captur e by an alpine grassland at Tungnath (Garhwal, Himalaya). Tropical Ecology 30(1): 65-68 . . Sund[iyal, R.C. & Joshi A.P. 1990. Effect of grazin g on standing crop, productivity, and efficiency of energy capture in an alpine grassland ecosystem at Tungnath (Garhwal, Himalaya). Tropical Ecology 31(2): 84-97. Tainton, N.M. 1982. Response of the humid subtro pical grasslands of South Africa to defoliation. Pp. 405-4 14 in: Huntley, B.J. & Walke r, B.H. (eds.), Ecology of tropical savannas. Ecological Studies 42, Springer-Verlag, Berlin. Tainton, N.M. & Mentis, M.T. 1984. Fire in grassl and. Pp. 115-147 in: Booysen, P. de V. & Tainton, N.M. (eds.), Ecological effects of fire in South African ecosystems. Ecological -Studies 48, Springer-Verlag, Berlin . 125

·-·,

Chapter 5 .. ch G & van der Maarel, E. 1988 1ttlyanova, Ad., RusO .. l;nd ln relation to grazing intensity . Biomass structure of lin:estone p~~soo . Acta Phytogeographica Suectca 7~: • · 125-13!. 1987 Effects of fires on struc mrc , com posi tion ,_ and nutrient status in . Van Groenh, . pa;amos near Bog ota (Colombi<~). MSc. thesis, Inte bunc grass rnal report, Hugo . Vries-Laboratory, Um.vcrslly Amsterdam. . . .. p A & Budde, P.E. 1992of . Verw:'J:. . . . . , al . . Pp . Burning and grazmg gradtcnts m paramo mtual ordtnauon an yses. . 177-195 in: Balslev, H. & Luteyn, J.L. (eds.), Para " mo: ~ Andean ecosystem u_nder hum . . '! ~ lnfluence. Academ~c Press, London .. A & K k K. 1992 Effects .. of fire and grazmg on Espelena VerweiJ, Pl. : Ppo popu auons. · Zl5-229 i~·· Balslev H. & Luteyn, J.L. (eds.}, Paramo: an ' . os stem under human influence. Aca Lo d ~ YA K k K & Budde p E 1995 demtc Press, n on. . . Asp ecto VerweiJ, P. ·• o • · s de Ia tranformact6n del paramo.por ' ' hombre: In: V~ der Ham'men St , T. & _Dos Santos, A.G. (e ds) . . • udies on Ande~ Ecosystems 4. J. Cram er, Berl m. . . G & E . F C 1964 Primary production and W1egert R the disappearance of vans, · · · ve,get~ti~n on an old field in sout heas tern M~chigan. ~cology 45(1) 49 63 Young, T.P. 1985. Lobelia te/ekii herb ivory, mortahty, ~d size at reprodu :. : . . t' with cuon. vana ton _,_.:· growth rate. Ecology 66: 1879-83. ··

6 6.1

6.2 6.3 6.4

6.5 6.6 6.7

SIMULATION MODEL OF VEGE TATION DYNAMICS Introduction Definition of functional plant groups Model formulation Model calil:iration Conclusions on vegetation development Simulating management strategies Limitations of the model

This chapter describes an abstracti on of reality: the behaviour of spec ies and groups of plants as a function of management vari ables in the form of a model. Diff erent approaches to ecological modelling are discusse d in Chapter l. In ftrSt inst~ce, the model at ecosystem level is used to analyze, describe, and replicate the development of para mo grassland under human influence. Also, it may be used to simulate the development of vege tation composition towards .desired management objective s.

6.1

Introduction

M~agement objectives range from · the vegetation that offe rs the greatest potential for (sustainable) animal production to pure conservation. The maintenance of biological diversity is another possible goal of increasi ng relevance. The management tool s are the ftre and grazing regimes that c~ be applied. The options for management are the ~ges of vegetation structure and floristic composition that are possible on each tract of land. Functional plant groups were defined to serve as buil ding blocks for the modeL These are groups of plants that show similar responses to environm ental and management variables. Finally, the above elements are combined to evaluate some management strategies, dem onstrating how the · model can be used as a tool to pred ict the possible course of vegetatio n development and so support management decisions. Con clusions on the implications of the model outcomes for management are discussed in the-f inal chapter, Chapter 9. Three main phases c~ be recogniz ed in the building of a simulatio n model, with the following related activities (adapted from Swartzman & Kaluzny, 1987 ): (1) Conceptualization: includes the definition of model objectives, state variables, boundaries, ~d subsystems, the construction of a conceptual diag ram, and the development of equations ' (2) Implementation: includes paramet er estimation, sensitivity analys'is, (3) and calib Evaluation: includes the evaluatio n of simulation experiments, valid ration ation or corroboration, ~d the definition of the limitations of the model. The SENECA simulation environm ent was used; th\s is a software package designed to develop models of time-dependent processes and, in particular, ecosyste m models (De Hoop er a/., 1992). It is based on continuo us mathematics ~d facilitates the creation of different compartments.

126

127

Chapter 6 Simulation model of vegelation dynamics

Model objecti~es . . . The first conceptual step in model development is setung the model ObJectives. These formulated as follows: • to structure data on the paramo ecosystem and its components in an integrated ·to replicate paramo vegetation response to fire and grazing • to predict the effects of management on vegetation development The model is constructed for zonal grass paramo vegetation only under the influence burning and grazing (floristic types C to I, Chapter 3).

6.2

Definition of functional plant groups

Graminae that occur with low covers in the natural vegetation to so~e e~tent overshado-:ved by the buncb grasses, i>Ut main!; growtng m the spaces m between. The main species are Calamagrostis coarclata, Agros1is haenkeana, Trisetum spicatum, and Festuca andicola. The maximum height reached by short grasses is about 20 em, while the average height is 5 to 10 em without inflorescences. They may increase under grazing, forming a closed mat with the shon forbs or the ground rosette layer, but are also vulnerable to adverse trampling impact

·I Ground rosette layer:

In order to define appropriate state variables, it has to be realized that it is impos~ible · describe the reactions of all species to all combination s of possible impacts. Plant species similar behaviour need to be grouped together. The basic assumption underlying the is that plants compete for space, light, and scarce nutrient_s. In view of this, strat~gies ground surface occupance were used as a summarizing criterion to group plant speci~S. ~e outcomes of the TWINSPAN and CANOCO analyses (Chapter 3) were used to u1~'uu~;Ul~u the main groups. The resulting functional plant groups consist of a number of plant sptcies. that show a similar response to specific management and environmental variables. The. functional plant groups are also homogeneous according to forage quality aspects (Ch?pter 4, Table 4.2). This concept of functional plant groups is similar to the theory of 'funcuon~ types' described by Smith & Huston (1989) and Huston (1994), which can be used to gam . insight into the regulation of species diversity in natural conditions.

Forbs:

include all herbaceous species not belonging to the ground rosette layer. Mature plants are taller than 5 em. Their basal area is low (2% or less), hence the influence on the establishment and growth of other species is insignificant

In relation to burning and grazing in zonal-bunch grass paramo, the following functional plant' groups were recognized, to a large extent though not necessarily coinciding with srructure layers present in the vegetation: Bunch grasses:

synonymous with tussock grasses. The main species are Calamagrostis effusa, C. recta and Festuca sublimis. They can reach a height of about 1.2 m and a diameter of 60 em.

Stem rosettes:

of the species Espeletia harrwegiana. This funct~onal pl~nt group includes seedlings, juveniles, and adult specimens with stem. The number of seedlings is highly variable, but hardly · influences total stem rosette cover. Therefore, all age classes of stem rosettes were lumped together.

Shrubs:

all woody species except those dwarf shrubs that have a prostrate growth form. Examples of upright-growing shrubs are Escallonia myrtilloides, Baccharis spp., and Hypericum spp. Growth rates are very low, probably due to the harsh climatic conditions that limit the growth of woody species in panicular.

i 6.3

comprises ground rosettes (e.g. Geranium spp., Senecio repens, Ranunculus praemorsus, Be/loa longifolia, Gnapha/ium Spp.), procumbent or creeping species (Lachemil/a orbiculata and other Lachemi/la spp., Lupinus microphyllus), and some dwarf shrubs. The short forbs ($ 5 em) are species that adapt rapidJy to changing conditions. Cover is highly variable. Dwarf shrubs, such Pernettya prostra1a, Dislerigma empetrifo/ium, and Muehlenbeckia volcanica exhibit a reaction to environmental changes similar to that of creeping species and are therefore included. seecies included in this group show the typical responses of colonizing or opportunist species.

Model fonnulation

' Each vegetation structure type is defined by ranges of cover values for each structure layer , (see C~apter 3). When cenain plant groups, by increasing or decreasing pass a critical level, a transtuon from one structure type to an other occurs. Time is needed under a certain : man~~ement regime tha.t 'pushes' the vegetation towards degradation or recovery to effect a transtu~n. from. one fl~nsnc type to another. Structure may change rapidly, whereas species con:posllton sull remams stable over several decades. As soon as certain plant groups start to dtsap.pear, h~wever, re-establishment is less probable and some types of degradation appear almost IITeveTSlble. The processes are shown in the conceptual diagram of the vegetati~n/herbivore interactions

present~d in .Figure 6.1. ~e simulation model is primarily related to structure ch;tnges in the

veget.anon, I.e. changes m the cover of the functional gioups defined above. For each funcuonal plant group, a separate submodel was defined.

128

1 29

Simulation model of vegetation dynamics

Clwprer 6

Another state variabk which is related to vegetation structure, representing not the functional plant groups but rather-their ;~bscnce, is the percentage of bare soil surface. The reason to include the development of bare soil in the mudd is that it is related to erosion and run-off processes. An additional submodel was created to simulate the development of bare soil . percentage. Two other submodels describe the variation in grazing intensity and the occurrence of fire, which are the two forcing variables. To simulate the effect of different grazing intensities, four different compartments were created, each subject to different grazing · intensity.

Fire OCCUIT8009

FUEL

These model compartments can be considered idealized vegetation patches with different degrees of utilization. The forcing variable grazing intensity was in this way incorpora ted into the model. One external forcing variable remained: the occurrence of fire. This depends on the ·fuel load of the vegetation and management decisions or accidents that set the veget<jtion on fire. Before presenting any details on the grazing and burning functions and auxiliary variables , model construction with regard to the functional plant groups is described. The complete documentation for the model is given in Appendix B, including the submodel ' equations and a list of all state variables, variables, and parameters used.

,.-·-·-·-·-· -· ··-·-·-·-·-·-·"'\ j i i i

BUNCH GRASSES

DE'AD / GREEN

'-::======:::!

ij r : i

STEM ROSETTES

I I~==S=H=R=UBS==:::::

Hierarchy of fu nctional plant groups The responses of the functional plant groups to increasing fire age and to grazing intensity, as well as the interactions among these groups, were in the first instance plotted in spreadsheet. Certain plant groups are clearly dominant. The maximum cover of stem rosettes, for example, is limited by the actual bunch-grass cover. This is shown in Figure 6.2. Actual . stem rosene cover is inversely related to actual bunch grass cover. One could deduce, of course, that the opposite is true: bunch-grass growth would be limited by stem rosettes. An argument against this is that 1dult stem rosette cover never exceeds 20%. Their root system is extensive neither in the horizonta l nor in the vertical direction, so it is physically improbable that they would limit bunch-grass growth.

SHORT GRASSES

l:== ===~ GROUND ROSETTE

LAYER

FORBS

L.'---.--. - - ---..J

By analyzing in this way the relationships between the maximum cover of functiona l groups ' and the actual cover of others, a hierarchical sequence of plant groups was constructed, showing which groups are dominant. This hierarchy is shown in Figure 6.3. The combined actual covers of bunch grasses, stem rosettes, and short grasses determine the maximum space left for the ground rosette species. The amount of bare soil is lowest in the hierarchy, as it is limited by the weighted cover sum of all plant groups.

VEGETATION ATTRACTIVENESS

( Figure 6.1

SLOPE

)

WALKING DISTANCE

Conceptual diagram of the vegetation/herbivore interactions. Cattle distribution is · modelled in Chapter 7. Other components of the diagram refer to the simulation model. Prod.= production rate; cons.= consumption rate; rectangular boxes represent state variables; rounded boxes represent auxiliary variables. 130

Time unit The time unit on which frrst modelling efforts were based was a year. It became clear, especially when comparing field data concerning recent frres, that the course of vegetation recovery needed to be monitored using narrower time intervals. Fortunately, it was concluded that there are no significant seasonal differences in biomass production (see Chapter 5 and Hofstede et al., 1995). It could therefore be safely assumed that a model with a basic time unit of one month would similarly approximate vegetation responses. An advantag e was that parameter estimation could be executed with greater precision, as field data with fire ages expressed in fractions of years were available. The Euler integratio n method, which is suitable for modelling discontinuous events such as the occurrence of fire, was applied. 131.

Simulation model of vegetation dynamics

Chapter 6

. CJ.

SHRUBS

Figure 6.2

Cover of adult stem rosettes compared indicates maKimum stem rosette cover.

General form of equations Exponential growth of the functional plant group s is only valid in certain time intervals. ~ · Sooner or later, space and nutrients become limitin g. Growth rate decreases with increasing ' plant cover-until growth equals zero. The conce pt of carrying capacity, K (regarding a · · functional plant group), is then the maximum at which cover stabilizes when the growth rate reaches zero. The simplest assumption is that growt h rate decreases linearly with cover. This leads to the classical Verhulst-Pearllogistic growt h equation (J0rgensen, 1986): · dN/dt = r * N * (K-N)/K in which r = the relative growth rate of N, K = the carrying capacity. This growth equation is generally applied to anima l populations (e.g. Van Wijngaarden, 1985).

. . ·- .. 132. . ~

..,~ .

Figure 6.3

Hierarchy of functional plant groups. ,

..,, ?;>:R!"li"'l'l) f'U:!o r 1Al DE COLO MOl4 !:·::_;~~ "' • .:.;rh;!·.u l~::tur oloa

}·;;_,;y

·r-:c-:.

It was applied to vegetation struc!ure layers by Heil & Bobbink (19?3). For the fa~ous _fo~est succession model of Hubbard Brook (Botkin ec a/., 1972), a crowdmg factor was m a mrular way defined as the fraction of the maximum possib le basal area that is actually covered_by trees (Swartzman & Kaluzny, 1987). Translated into SENECA statements, the rate equauon is as follows: -FPGPROD = FPGRGR * FPGAC

* (FPGMC - FPGAC) I FPGMC

in which FPGPROD = the increa se of cover of functional plant group FPG due to production, FPGRGR =relative growth rate of the cover of FPG, FPGAC = actual cover of FPG, FPGMC = maximwn cover of. FPG. The logistic growth equation appeared to descr ibe the growth curves of plant cover o~ly partly. Especially during the first few years a~ter a fire, a_ll _cover v~u.es show an exp~ostve increase instead of the slow initial growth that ts charactensnc of logtsnc growth. Despite the soil surface being almost completely bare after a fire event, new sprouts quickly appear.

133

Chapter 6 This phenomenon proves that the remnant tussock bases, the root system, and the bank together provide enough-reserves for regrowth, the remaining biomass being a limiting factor to post-fire production. Besides, the increased availability of nutrients, and space are probably responsible for stimulating production. Therefore, first four years after fire the relative growth mte was multiplied by the stimulus: Fire stimulus = FSFPG

* (I

+ ln(49/(FIREAGE+ I)))

in which FSFPG =fire stimulation factor close to l which is specific for each functional plant FIREAGE = time after fire occurrence in months. This explains the post-fire regenemtion of the functional plant groups in a far With time, the effect of fire stimulation on the growth rate decreases. As a result of biomass reserves become exhausted, so the rate of recovery goes down. After four growth stabilizes and the equations without fire stimulus apply. Grazing pressure also stimulates the production of certain plant groups. In the chapter, higher production rates were measured for bunch grasses in grazed grazing stimulus' of production was derived empirically for most of the functional groups. For the inedium-tall forbs, no positive effect of grazing on the growth rates found. The following equation describes the grazing stimulus by which the relative rate is multiplied: GSFPG

= GS *(I+ ln(GRI+l))

GSFPG = grazing stimulus of production of plant group FPG with value I or in a range from 1.25 . 1.4, GS =grazing stimulation factor "Specific for FPG, GRI =grazing intensity in cowdr./50 m2• Under the influence of grazing, some plant groups decrease in cover, others increase. depends on the balance between the positive effects of the grazing stimulus on production the negative effects of trampling and co.nsumption. In this case, the effects of COI1Sumptioni-6J and trampling on cover are modelled, not the quantity of biomass removed or suppresseq by grazing. The consumption /trampling impacts on the cover values of individual plant groups increase with grazing intensity and also depend on the actual cover itself. The effect · · grazing on cover is therefore given by the rate equation: FPGCON = FPGGF

* FPGAC * GRI(t)

in which FPGCON = consumption and/or trampling effect on cover of FPG. FPGGF =FPG-spccific_ grazing factor. After every integration time step, the change in each state variable is added to its derivative. For the development of any functional plant group in time, the general state equation becomes: FPGAC(t + ll) = FPGAC(t) + FPGPROD - FPGCON

134

Simulation model of vegetation dynamics in which FPGAC = actual cover of FPG, FPGPROD = cover increase of plant group FPG due to production, FPGCON =cover decrease of FPG due to grazing.

Bunch grasses

·The bunch grasses deserve special auention. Their contributions to forage supply, to fuel accumulation and to the water retention capacity of the system are important. Moreover, bunch grasses are the dominant plant group, which implies that actual bunch-grass cover determines the maximwn cover of an other plant groups to a certain extent During first modelling auempts, it was noted that the development of the bunch ·grasses was ' not properly described by the above general equations. The process of fragmentation and the · ' possibility of local extinction of the bunch grasses especially required an additional vari.able that would keep track of the reserves of the bunch-grass subsystem. To this purpose, the ' ·.development of the tussock base percentage effectively producing tillers was evaluated. The state variable TB was created as a measure of the reserves present in the tussock base. After 1 a fire, these reserves are deple ted in the course of the rapid regrowth of fresh leaves. The ~ 1turning point comes about four years after a fire, when the growth rate stabilizes at the prefire level. Then, reallocation of carbohydrates to the tussock bases generates a net accumulation of reserves. Payton & Mark (1979) also found that for. a period of four years after early-season burning · leaf elongation increased for Chionoch/oa rig ida snow tussocks in New Zealand. With regard '· to these processes, a linear decrease in TB was defined during the four years after burning, · followed by a slow linear increase in the absence of grazing. If grazing occurs, trampling impact and reserve depletion may result in a linear d~crease in the tussock bases. In order to describe growth, senescence, and litter fall, the additional state variables of gree n . and dead bunch-grass cover were defined. The life spans or turnover times of green and of dead material determine the rates of senescence and litter fall, respectively. The rates of senescence, litter fall, and possible consumption control the state variable of dead bunch-grass · cover. The corresponding equations are as follows: BUYSEN = (1/LSBUY) * BUYAC BUOLIT = (I!LSBUO) * BUOAC BUOAC(t + ll) = BUOAC(t) + BUYSEN - BUOLIT - BUOCON

(rate) (rate) (state)

in which BUYAC = actual green bunch-grass cover, BUOAC = actual dead bunch-grass cover, BUYSEN = rate of senescence, BUOLii = rate of liner fall, DBUOAC = derivative of dead bunchgrass cover, BUOCON =cover decrease of dead bunch grasses due to grazing. Consumption of green and dead bunch-grass cover were defined in the usual way. Production of green bunch-grass cover is simulated by applying the relative growth rate to the tussock base instead of to the actual green bunch-grass cover. The maximum cover of the green bunch-grass leaves is limited directly by the actual cover of the dead bunch-grass leaves. Total bunch-grass cover is equal to the sum of green and dead bunch cover. The dead fraction of tussock biomass is also calculated.

135

Chapter 6

Fire occurrence The occurrence of fire is modelled as follows. After a fire, a fuel load gradually buildS' the vegetation. As long as the amount of burnable material remains below a critical . vegetation will not bum. Releve data from sites that were burned twice were derive empirically the conditions for fi re occurrence. Whether a fire will spread or not,, of the vegetation is set on fue, depends on three variables: (1)

(2) (3)

the percentage of dead tussock biomass which has to exceed 55 to 60%; if it below this threshold, the vegetation is too green and holds too much water tQ the cover of bunch grasses has to be more than 50%, otherwise the bunches ·. widely spaced and the fire cannot spread from one bunch to another; and the bases of stem rosettes, which also contribu te (although to a lesser extent) potential spread of a fire as long as the dead leaves are not removed from the

Besides the physical potential for burning or fuel load, a fire e<vent depends on the the vegetation is set on fire. This can be due to a decision made by a farmer or by accident, or by lightning (rare). For this purpose, a burning function was defined, toggles the possibility of burning. The conditions for a fire to occur are spe~ified as

FUEL = BUOAC + 0.7 * ESPAC BURN =FI(t) * FUEL in which FUEL= fuel load, BUOAC =dead bunch-grass cover, ESPAC = stem rosette cover, BURN:!:r':1 = burning potential, FI(t) = forcing function tllat toggles the possibility of burning. Data on sites that had been burned more than once indicated that the minimum time before a new fire occurs is 2.5 to 3.5 years. A threshold value for the minimum fuel required for the occurrence of fire, was derived in this way. When the fuel load as "A'•n~• "'' above, exceeds a value of 32, a fire occurs on the condition that the burning function is to 1. Fire age is then set to zero. This subsequently causes a strong reductio n in plant in each submodcl. The reduction factors corresponding to the various functional plant um'"n~;k.J were derived from releve data on recently burned sites. Immediately after a fire, cover generally reduced to values below 2%._·

Grazing intensity Grazing intensity is calculated in the function of forage availability. Chapter 7 (equation 7.1) gives further details concerning the followin g measure of forage availabi lity, the vegetation . . attractiveness for grazing: • VA

= 2*BUYAC + BUOAC .+ 4*SHOAC + 2*ROSAC + FORAC

in which VA= vegetation attractiveness, BUYAC =green bunch-grass cover, BUOAC =dead bunchgrass cover, SHOAC =short grass cover, ROSAC =cover of the ground rosette layer, FORAC = (tall) forb cover. Both forage quantity and quality aspects are included in this indicatio n of forage availability. In Chapter 7, it is demonstrated by means of multiple regression analysis that of all factors included vegetation attractiveness is the one that best explains variation in grazing intensity. 136

Simulation fTUJdel of vegetation dyllamics Grazing intensity (In cowdr./50m2) was plotted against vegetation attractiveness. The relationship of average grazing intensity to vegetation attractiveness was established in this way. Tht: at:tual grazing intensity of compartment (i) can be higher or lower than the average grazing intensity, dependi ng on a use factor that determi nes the degree of utilization of the ·· vegetation patch. The corresponding expressions are: . · GRIA y = e (o.oos•VA+0.98) GRI(i) = UF * GRIA V · in which GR!AV = average grazing intensity, VA = vegetation attractive ness, GRI(i) = grazing ·intensity of companmcnt (i), UF = use factor. ·

6.4

Model calibration

Ecological simulation models are examples of models describing complex systems that are a priori poorly underst<>D9. Uncertainty in data from field observat ions may further mask system-controlling mechanisms. To reduce the uncertainty in the model output, one can optimize parameter values by calibration. In SENECA, a quantita tive measure for the agreeme nt between model results and field data acts as an object function that has to be minimized. The quantitative measure calculated is the 'goodness of fit'. Ranges of initial parameter values make up a parameter vector space. The dimensionality of this parameter vector space is equal to the numb~r of parameters in the model. The distributions of these ranges (uniform , uniform Latin Hypercube, or· normal distribution) are also defined. When k parameters of a model have to be estimated, a much larger number of parameters (m) is randomly drawn and stored in a so-called 'vase'. Subsequently, each of these m parameters is evaluated by running the model with this paramet er vector and storing the corresponding goodness of fit. The larger the number of parameters stored in the vase, the more thoroug h the search for a global minimum will be, thus avoiding local minima of the object function . After the ininal examina tion of the k-dimensional vector parameter space, new parameter vectors are drawn by a controlled random search method. The centroid is calculated for a set of n parameters (n < m). From the remainder of the parameter vectors in the vase (m - n), a vector is drawn anandom and reflected in the centroid. If the resulting new parameter vector fits in the initial parameter ranges, the new vector replaces the worst vector stored in the vase on· the condition that it gives a better fit. The final result is a set of new parameter .ranges and the parameter values that give the best model performance. The calibration method described above is especially suitable for multivariate parameter estimation of non-linear simulation models. A high degree of uncertainty and structural model errors due to the aggregation of processes does not allow a stochastical error approach. For the purpose of calibration, grazing intensity was fixed for the four differen t compartments. · In .that way, model behaviour could be compared with the field data. Four different time series of vegetation development were extracted from the point data set. 137

Chaprer6 These time series included releves of different grazin g intensity classes: the ranges for four compartments were 0, I to 2, 3 to 9, 10 to 20 cowdr./50m2, respectively. Besides intensity, fire age and the cover of the defined functio nal plant groups were also known each point. Time series were constructed according to fire age: at t=O, fire age is also Data smoothing was applied by calculating threedata running averages of the time series. Following the hierarchy of functional plant group s (Figure 6.3), a sequence of "'"'·•u•auon was executed. First, the unknown parameters of the bunch-grass submodel were This was an iterative procedure, leading to modif ications of the bunch-grass equations improvement of the submodel structure, each time followed by calibrations. Not satisfactory behaviour had been obtained for the bunch-grass submodel, could ~.;cu. wrauo proceed for the next submodel in hierarchy, the stem rosettes. The advantage of ~u11m·v•t1,,n. the model into four compartments that represent vegetation patches under different intensities is that the four different time series conne cted to these patches are all included the calibration. · The maximum cover of each plant group was in the first instance obtained from the data According to the .hierarchy of Figure 6.3, functi ons were derived for each plant group, describing how maximum cover depends on the actual cover of others. The factors by which: each plant group had to be multiplied to estimate their limiting effect on the {jevelopment of. cover of plant groups lower in the hierarchy were later calibrated. For bunch grasses: maximum cover was fixed after calibration. The values of some important parameters, after calibra tion, are listed below in Table 6.1; the rest are listed in Appendix B. Relative growth rates are in the range of 0.004 to O.l6/month.'1" The cover increase of stem rosettes and shrubs is slowest, the bunch grasses occupy an intermediate position, and the cover of shan grasses, ground rosettes, and (tall-) forbs increases the most rapidly.

State variable

MAX

relative growth rate (1/month)

grazing factor

fire stimulus

maximum limitation cover% of max. cover by tussocks

0.053

ESP SHR SHO ROS FOR BAR

0.004 0.012 0.160 0.080 0.095

Table 6.1

These outcomes confirm our expectations on the behav iour of the different plant groups. The functional plant groups with the highest grazing factor values are the forbs, the ground rosette layer, and the short grasses. Tit is means ihat under the same grazing intensity, the greatest effects of reduction in cover are observed in these plant groups. This is consistent with the results of Table 4.3, which indicates a hig.her forage preference for these groups, probably in relation to a higher forage quality. The frre stimulu's of cover increase by stem rosettes is highest in comparison with other plant groups, whereas the cover increase of short grasses is not stimulated by ~re occurrenc~. The figures for the maximum cover per plant group and the extent to whtch these maxtmum values are limited by bunch grass cover are also realist ic.

Sensitivity analysis

The outcomes of sensitivity analyses on the perfor mance of the separate submodels were difficult to interpret The wide probability ranges of cover values indicated were due on the one hand to the abrupt cover changes occurring during the growth peak after burning, and on the other probably to the influence of noise still present in the vegetation data set. The assumption that grazing intensity was constant for the time series is a simplification of reality. Yet, some conclusions regarding sensitive param eters can be drawn by observing the parameter changes during subsequent calibrations and by varying the parameters manually. Ii was noticed that for plant groups that are high in the hierarchy (bunch grasses, stem · rosettes, shrubs), the relative growth rate and the grazing factor are most sensitive. For the other plant groups, parameters limiting their maxim um cover are most sensitive. The shape of the tussock-base curves is crucial to the developmen t of all plant groups.

6.5

RGR

BUY BUO

Simulation model of vegeration dynamics

* w·l

1.2 1.9 0.2 1.5 8.5 9.0 16.0

1.12 3.00 1.55 0.00 1.17 1.15

80 80 36 18 37 100 100 100

0.35 0.08 0.51 0.99 0.70 0.70

Calibration results for a number of imponant parameters, corresponding to different submodcls and state variables. All other parameter values arc listed in Appendix B. 138

Conclusions on vegetation development

Some of the·.curves resulting from calibration are presen ted in Figures 6.4 to 6.11. Each graph shows the post-fire development of a functional plant group for SO years under the influence of the four grazing intensities. Corresponding observ ations are included, and the time axis is indicated in months. A clear tendency is the decrea se in the bunch grasses and shrubs under increasing grazing pressure. Stem rosette cover increases .at intermediate grazin g pressure (up to 10 cowdr./S0m2), but declines under more intensive grazing. Immediatel y after a fire, species belonging to the ground rosette layer quickly colonize the open space . They also grow over the tussock bases until the regrowing grass shoots overshadow them. As other plant groups recover gradually, the cover of the ground rosette layer decreases. The short grasses show a similar tendency, though less pronounced. Regeneration of short grasse s after frre is a slower process.

139

Chapter 6

Simulation model of vegetation dynamics

BUTAC(1)

SHAAC{1.j.

x-BUTAC(2)

x-SHAAC{2)

·--

BUTAC(3)

SHAAC(3)

. ..........

6········ ..

~.~

BUTAC(4) .

SHAAC(4)

.·~

•

.....

·~~. .. .... "

····· ···)!············

0 0

Figure 6.4

120

240

360

480

The post-fire development of bunch-grass cover (BUT) under different intensiti es. I= 0 cowdr./50m2; 2= 1-2 cowdr./50m 2; 3= 3-9 cowdr./5 2 0m ; 4= 2 cowdr./50m • X-axis: fire age in months. 30

120

600

240

·--

SHOAC(l) 6

)( - -

.....

.... ····· ....······ .................................. ·~

·······

ESPAC(2)

600

+--

480

The post-fire development of sllrub cover (SHR) under different grazing intensities. I= 0 cowdr./50m2; 2= 1-2 cowdr./50m2; 3= 3-9 cowdr./50m 2; 4= 10-20 cowdr./50m2• X-axis: fire age in months.

Figure 6.6

ESPAC(1)

360

___

;;- - - -

ESPAC(3)

x-SHOAC(2)

SHOAC(3)

15 A •••·•·•··•

...........

ESPAC(4)

SHOAC(4)

10 X

.......

················

Ot--~-,--,--,--,-~ ·-.-~ ··_···_ ··~···~~~ -·~···~·-~··~-·~ ·· ·~·-~ ·· 0 120 240 360 480 600

Figure 6.5

The post-fire development of stem roselle cover (ESP) under different grazing intensities. I= 0 eowdr./50m 2; 2= 1-2· cowdr./50m 2; 3= 3-9 cowdr./5 2 0m ; 4= J0-20 2 cowdr./50m . X-axis: fire age in months. -140 •. ..r: •·

. Figure 6.7

120

240

360

480

600

The post-fire development of shon grass cover (SHO) under different grazing intensities. I= 0 cowdr./50m 2; 2= 1-2 cowdr./50m2; 3= 3-9 cowdr./5 2 0m ; 4= 10-20 cowdr./50m2• X-axis: fire age in months.

141

Chapter 6

Simulation 17U)del of vegetation dynamics

·--

" ........................................".................. ······ ·········

·--

ROSAC(1)

BARAC(1) 70

x--

x- -

ROSAC(2)

BARAC(2)

·--

ROSAC(3)

BARAC(3)

........... ROSAC(4)

. ..........

BARAC(4)

20 X

... •A.'!. ..••, .. •••.•••. +

240

Figure 6.8

360

480

600

The post-fire development of lhe groilnd rosette layer (ROS) under different intensities. I= 0 eowdr./50m2; 2= 1-2 cowdr./50m2; 3= 3-9 cowdr./50m2; 4= I cowdr./50m2 . X-axis: fire age in months.

·--

FORAC(l)

x -FORAC(2)

·--

FORAC(3) 15

A·········· FORAC(4)

Figure 6.9

120

240

360

480

600

Figure 6.10

120

240

360

600

The post-lire development of bare soil percentage (BAR) under different grazing intensities. I= 0 cowl:lr./50m1; 2= 1-2 cowdr./50m2; 3= 3-9 cowdr./50m2; 4= 10-20 2 cowdr./50m • X-axis: fi re age in months.

The cover values of both_short grasses and the ground rosette layer increase with grazing influence. Forb cover varies immediately after a fire, with values above and below normal, but then stabilizes at about 7 to 10%. This does not vary much with differences in grazing intensity. Bare soil percentage is substant ially increased during the fust few years after ftre: Immediately after burning, higher cover values were observed for stem rosettes in the most intensively grazed situation th.an predicted by the model. The occurren ce of younger individuals without a stem is highly variable and might be explaine d by mechanisms other than d10se affecting adults (see Chapter 5). The smallest individuals might be included in the ground rosette layer, or cover development of this age class could be modelled separately. However, for the other grazing intensities a reasonable fit was obtained . Calculation of the dead fraction of tussock biomass showed that grazing intensity limits the maximum dead percentage. This means that the more grazed the bunch grasses are, the slower senescence occurs and the greener they remain. This has implications for the bumability of the vegetation. Above a detennined level of grazing intensity , fuel loading or the accumulation of burnable material over time is reduced to an extent that inhibits burning.

The post-fire development of forb cover (FOR) under different grazing intensitie s. ... I= 0 cowdr./50m2; 2= J-2 cowdr./50m2; 3= 3-9 cowdr./50m2; 4= 10·20 cowdr./5 2 0m • X-ax.is: fi re age in months. 142

480

143

'··

Chapter 6

6.6

Simulation model of vegetation dynamics

Simulating management strategies

In the case of frequent burning (Figure 6.16), the bare soil percentage is almost continuously above 10% if the flre recurrence interval is three years. Only when fire frequency declines after.IO years due to depletion of the tussock bases does the bare soil percentage decrease to values below 5%. If burning is not repeated, it takes 1.5 to 5 years of post-fJre recovery before the bare soil has stabilized at pre-fire levels (10% or less). The time required for -recovery of the protective cover of the soil surface depends on grazing intensity and the tussock-base percentage.

Simulation experiments were executed investigate the effects of different strategies on vegetation development. The burning function and the use factor grazing intensity were the management tools that were varied in different Results of the simulation experiments are presented in Table 6.2 and in Figures 6. 11

Grazing Figures 6.11 to 6.15 show the post-fire development of the main functional plant the absence of repeated burning and under different grazing intensities. From the model it can be observed that degradation may occur within a few decades continuously high grazing pressure. Four times the average grazing intensity results · disappearance of the bunch grasses with_in 50 years.

FI UF

The grazing intensity below which the bunch grasses remain in the system, in the repeated burning, is 3.1 cowdr./50m2. (use factor UF = 0.4). At the average grazing (UF =1), the bunch grasses slowly decline, which means that most grazed vegetation are actually overutilized in relation to this plant group.

Burning Figures 6.11 and 6.12 show that initial tussock-base percentage determines the of regeneration to a large extent. If the tussock bases are intact (about 30% basal recovery of total tussock cover to 80% occurs within about three years. If the fragmented, after a few years recovery is limited by the slow growth of the tussock might take more than 10 years. In the latter case, species of the ground rosette layer longer from the available space, light, and nutrients. Payton & Mark (1979) regeneration period of 14 years before the biomass of Chionochloa tussocks approached of unburned plants. The magnitude of growth stimulation after fire increased with the · between fires (Mark, 1965). A similar tendency is-observed for the simulated ae•veuJpnJe~t·,~l of paramo grasslands (see Figure 6.16; FI = 1, UF = 0). A comparison of Figures 6.16 and 6.17 shows that the depletion of the bunch grasse§ accelerated by frequent fires in combination with grazing. Whether the possibility of occurrence is at fixed intervals of 10 years or whenever sufficient fuel has accumulated not make much difference, as can be seen in Table 6.2 (FI = 0. 1 or FI = 1). Above a certain gtazing intensity (UF > 0.4), burning is inhibited by insufficient accumulation. The dead percentage of tussock cover and total tussock cover are \frastically reduced. If burning is in principle possible at any moment in time, the burning frequency depends primarily on the state of degradation of the· bunches as '""·~··••n by the tussock bases. The short gmsses and ground rosene layer are especially favoured when the tussock decrease. In the_ case of burning and light-intensity grazing (Figure 6.16), the short reach a cover of 20% and the ground roselle layer about 55%.

·•>~

0 0

TB1=0

TB BUT DEP ESP SHR SHO ROS FOR BAR

30 80 67 7 8 0.1 0.1 10 9

GRI

Table 6.2 ·

0 0.4 17

0 1 13

0 4 10

0.1 0 30

0.1 0.4 17

1 0 30

1 0.4 17

13 43 61 15 6 9 37 5

3 9 56 9 7 24 58 4 6

0.1 0.2 n.a. 0.1 0.1 16 71 9 9

11 32 65 10 6 17 47 0.1 1

5 20 65 15 6 21 52 3 5

12 34 65 9 7 16 46 0.1 1

5 17 61 19 7 22 54 I 5

7 3.1

7.9

3.1

Simulation results after 50 years of vegetation development under different management regimes. Fl refers to the burning function; FI = 0: fire does not occur, Fl = 0.1: fi re may occur once a decade, A = I: fire may occur any time, whenever sufficient fuel has accumulated. UF = usc factor on which grazing intensity depends; DEP =percentage of dead bunch-grass cover; GRI =grazing intensity in cowdrJ50rn2 •

Validation · A real validation or corroboration procedure cannot be carried out for this simulation model due to the unavailability of an independent data set. What can be noticed is that the curves as predicted by the model fit the distribution of the observations reasonably well. In this case, the iterative calibration procedure resulted in a better structured model and an improved goodness of fit. The same model structure was applied for all functional plant groups, whereas the compartments with different grazing intensities provided a fourfold repetition of the same response model. The calibration results agree with existing knowledge on a number of ecological processes. Other findings that concur with the model outcomes are presented in Chapters 7 and 8. An example is the conclusion that, under moderate to high grazing intensities, burning of the vegetation is not possible due to a permanently low fuel load. This is confirmed by the spatial model of frre history of Chapter 8, which shows that burning occurs at intermediate distances from the fincas, where grazing intensity is lower. r

144

145

~·

., Chapter 6

Simularion model of vegetarian dynamics

BUTAC(2)

70 ESPAC(2)

SHRAC(2)

40 SHOAC(2)

20 ROSAC(2)

BARAC(2) 120 ~

Figure 6.11

240

360

480

600

Simulated vegetation development for 50 years; Fl= 0; UF= 0; TB,.o = 30.

Figure 6. 13

120

360

240

480

600

Simulated vegetation development for 50 years; A= 0; UF= QA; TB,.o = 17. J

BUTAC(3)

70 ESPAC(3)

ESPAC(1)

~----

SHRAC(3)

SHRAC(l )

30l~~ 20 10

SHOAC(l)

SHOAC(3)

~ £\/·············································································

t~ /:

ROSAC(3)

ROSAC(l )

~-t~:.~---~~::-:····:·· -~

·····...

1 -...~-=----=.-..=.-..=.-:.-...-...-•..-..-....-=...==...=..=....==...==...=...

BARAC(3)

BARAC(1) 120

Figure 6.12

240

360

480

600

Simulated vegetation development for 50 years; Fl= O; UF= O; TB..o = 22.

146

Figure 6. 14

120

360

240

480

600

Simulated vegetation development for 50 years; FI= 0; UF= I; TB,.o= 13. 147

Chapter 6

Simulmion model of vegetation dynamics

BUTAC{2)

ESPAC(2)

SHRAC(2)

SHOAC{2)

.......................................................................................

. . . . . . - . - . · · - · - . ... . . . 0 - · - · - · · . . . . . . . . . . . . . . . . . . . . . . . . ............. - · ...... -

......................

·············

················ ROSAC{2)

- · - · .. .

BARAC(4) 0

Figure 6.15

120

240

360

480

Simulated vegetation development for 50 years; Fl= 0; UF= 4; TB,.o= 10. 80

6.7

240

360

480

600

Simulated vegetation development for 50 fears; Fl= I; UF= 0.4; TB,.o = 17.

Limitations of the model

ESPAC(1)

SHRAC(1)

40 SHOAC(1)

201!\l;

;

~~\~. . .:~...·V\ ....~:{'·····- ·-...

i ,\ /

\,/ /

r. ::

ROSAC(1)

-----=

0 0

Figure 6.16

Figure 6.17

120

BUTAC(1)

BARAC{2)

600

120

240

360

480

BARAC(1)

600

Simulated vegetation development for 50 years; Fl= I; UF= 0; TB,.()= 30. 148

A model of vegetation development for paramo grasslands was developed as described above. Certain restrictions to its application have to be taken into account. The climatic and edaphic conditions are fixed for this simulation model. It is applicable to other paramo areas, to afroalpine ecosystems, or to any other bunch grassland sysiem related to volcanic soils without a pronounced seasonality. Relative growth rates and equations describing the maximum cover of plant groups probably have to be adapted, which can be done by calibrations. The functional plant groups as such are not expected to vary much, whereas the species of which they are composed are highly variable. The model is actually restricted to the management regimes for which data were available. At grazing intensities much higher than the observed situations, prediction of vegetation development is not reliable. Degradation of vegetation cover under high grazing intensities (> 40 cowdr./50 m2) was observed locally around water points, but is not included in the model. For other types of management such as potato cultivation and sowing of grasses, the model could be expanded. Again, the building blocks of the functional plant group remain more or less the same, except for the additional group of the tall grasses. Limitations regarding temporal aspects are as follows.

149

Chapter6 The long-tenn effects of grazing are difficult to estimate, as grazing probably has not constant over time.and-Cannot be derived from remote sensing characteristics. question whether the intensively grazed situalions are actually stable or not answered. Likewise, the effect of repeated fires could not be fully be incorporated model as data on older fires were scarce. In particular, the development of tussock bases time needs more detailed investigation, as this is an important component of the many regards. Although for practical reasons the time unit of the model is a results of the model should be interpreted on an annual basis. Species migration in relation to the local extinction of species is not taken into dispersal of stem rosette seedlings is limited, which increases the risk of local After a gradual process of fragmentation, the bunch grasses may also disappear from an a mat of short grasses and forbs forming a physical barrier to the re-establishment tussocks. All other spatial interactions related to grazing behaviour are dealt with in the chapter.

LITERATURE CITED

Botkin, D.B., Janak, J.F. & Wallis, J.R. 1972. Some ecological consequences of a model of forest growth. J~umal of Ecology 60: 849-872. · De Hoop, B.J., Herman, P.J., Scholten, H. & Soetaert, K. 1992. SENECA 2.0: simulation environment for ecological application. Netherlands Institute of Centre for Estuarine Ecology, Yerseke. 224 pp. Heil, G.W. & Bobbink, ~- 1993. "Calluna", a simulation model for evaluation of impacts atmospheric nitrogen deposition on dry heathlands. Ecological Modelling 68: 161-1 Hofstede, R.G.M., Chilito, E.J. & Sandoval, E.M. 1995. Vegetative structure, mi·<:roc:Iimate: and leaf growth of a paramo tussock grass species in undisturbed, burned and conditions. Vegetatio 119: 53-65. Huston, M.A. I994. Biologicaldiversity: the coexistence of species on changing !a[tds<:aJ)l~s. Cambridge University 'Press, Cambridge, 681 pp. J~rgensen, S.E. 1986. Fundamentals of ecological modelling. Developments in Emrironm<~nta Modelling 9, Elsevier, Amsterdam. Mark, A.F. 1965. Effects of management practices on narrow-leaved snow Chionochloa rigida. New Zealand Journal of Botany 3: 300-319. Payton, 1.1. & Mark, A.F. 1979. Long-term effects of burning on growth, flowering, and carbohydrate reserves in narrow-leaved snow tussock. New Zealand Journal of 17: 43-54. Smith, T.M. & Huston, M.A. 1989. A theory of the spatial and temporal dynamics of plant communities. Vegetatio 83: 49-69. Swanzman, G.L. & Kaluzny, S.P. 1987. Ecological simulation primer. Biological Resource Management Series, Macmillan Pub!., New York. 370 pp. · Van Wijngaarden, W. 1985. Elephants - Trees - Grass - Grazers. Relationships between climate, soils, vegetation and large herbivores in a semi-arid savanna ecosystems (Tsavo, Kenya). lTC Publication 4, Enschede. 159 pp.

!50

SPATIAL MODELLING OF CATILE DISTRJBUTlON

7.1 7.2 7.3 7.4 7.5 7.6 7.7

This chapter deals with the question of how grazing intensity at a certain location can be predicted. There are different ways to quantify the value for the mean of a variable within a spatial unit. The areal mean can be taken as the value of a representative site within that unit. However experienced a researcher may be, bias is easily introduced if this approach is followed. Another option is to consider the land and management attributes as spatially correlated variables and to use spatial classification of relevant land attributes as the basis for estimating grazing intensity. Procedures· of spatial classification were discussed by Webster & Oliver (1990). Grazing and biophysical variables are apparently interrelated. In order to unravel the complex interaction between grazers and the grazed system, it is important to select the main attributes that determine the variation in gr!ll:ing behaviour and .to quantify the separate influence of each. If a large part of the variance in grazing intensity can be explained in this way by biophysical attributes that can be maf.ped, the spatial variability of grazing intensity itself can also be modelled. This leads us to the objective of the present chapter: to develop a spatial model of cattle distribution.

7.1

Introduction

The effects of grazing on conservation aspects such as landscape patchiness and plant species diversity are discussed in the final chapter, Chapter 9. In the present chapter, grazing patterns are analyzed from the perspective of the cattle. In previous chapters, it is shown that many different ecological units occur in the northwestern part of Los Nevados National Park. The differences between these units can be attributed largely to vegetation, terrain, and soil properties. Chapter 3 describes how the grazing regime alters vegetation structure and composition, whereas ·micro-relief and top soil are also affected to some extent. Grazing behaviour is influenced by the distribution of the land units with their corresponding biophysical attributes: the vegetation/herbivore interaction is determined mainly by the nonrandom grazing behaviour of the herbivores. Grazing strategy is directed at obtaining an optimal quality and quantity of grazed forage in relation to the grazing effort. In a comprehensive review, Arnold & Dudzinski (1978) concluded that at least four factors affect grazing behaviour. These· are green forage density, animal grazing time, grazing bite rate, and grazing bite size. At a low green forage availability, grazing time and bite rate of cattle increase. 151

Chapter 7 5:-,bbs (l9B, 1974) observed that not only bite frequency but also bite size is affected by fo:age density and maturity. By adapting grazing time, bite rare, and bite size, the animals er.:.i?it a flexible response to variation in forage availability in order to meet their feeding ~:.nre~cms. ~eve~heless, there appears to be a limit to the annount of effort cattle expend . lf. grazutg. In srtuanons of sparse forage availability, the animal~ may spend more than nine ~~-to~ ma~~um. of 1_2) hours a day ~azing (Stobbs, 1974). Forage availability, including IL .paual drstnbunon, rs therefore constdered the key variable on which the other variables of grazing behaviour depend. Wi:hin an area available for grazing, animals exhibit preference for certain sites and for ce~.ain plant species or groups. At low stocking rates, the cattle are able to graze according to prefere_nce at the cost of a low grazing effort. Consequently, centers of intensive grazing ~ establtshed where forage is abundant, of higher quality, or more accessible, while other Sitts are underutilized, fonning small islands of more or Jess intact vegetation. This patchiness wa; also reponed for sheep grazing (Bakker eta/., 1983b). At high stocking rates, however, the more preferred vegetation or locations become depleted. Animals are then forced to co::sume the less desirable vegetation or go to less desirable sites. With time, a more unifonn pat:!m of use is then achieved.. Besides the natural spatial variability of the resources, the extent of patchiness is thus strongly related to stocking density over time.

The spatial structure of the grazed landscape is characterized .by the number, location, size, and shape of the different patches. Each patch with .its geographical attributes is related to a number of biophysical characteristics- that determine the number of animals that can be pennanently supponed by that patch: the carrying capacity. No:" what are the variables that control both the selection of patches and the intensity with w~ch they are grazed, given a.cenain stocking density? Rice et al. (1983) listed the following v~~bl~~ governing the selection of grazing locations in rangelands: herbage density, water ~V<U!abtlity, relief, slope, elevation, exposure, natural and artificial barriers, herd social Interactions, prior experience, and climate. Some of these variables can already be excluded from fUI1her analysis of spatial relationships. Herd social interactions, prior experience, and climate, for exannple, are assumed to be stable and ~an be considered as the general setting of the study area, not related to specific locauo?s. Water availability is not a limiting factor in paranno systems, at least not so ct:amancally as in many African semi-arid rangelands. The grazing areas penaining to different farmers are separated from each other by some natural barriers such as block lavas with steep slopes, steep ridges, and glacial lakes. There are hardly any fences and the few present are the only artificial barriers to grazing. To some extent, farmers guide the animals to prevent _them invading the grazing areas of others. They gather pa11 of the herd for milking, and somenmes direct the cattle to recently burnt places. Summarizing, the key variables determining cattle distribution in paramo ecosystems can be grouped into three broad classes: (1) terrain variables affecting grazing effon • accessibility of terrain • slope

152

Spatial modelling of cattle distribution (2) forage availability • forage quantity • forage quality (3) management variables • stocking density • travel distance, e.g. in relation to milking practices • limitation of the grazing area • use of fire

7.2

Methodology

The variables mentioned above need to be quantified in order to simlllate cattle distribution over the study area. How to separate slope effects from the general accessibility of terrain? The problem is that cenain terrain types are associated with certain ranges of slope values and/or certain types of vegetation. · It is clear that separating the influences of the different variables is problematic. The cattle diet was studied independently of variation across sites (Chapter 5). The outcomes were used in the following way. Data on the grazing intensity of each releve were first combined with the sum of the cover values per plant group. The correlation found was weak. More satisfactory results could be obtained when each functional plant group was assigned a weight or preference factor. The assignment of preference values according to the outcomes of the observations on grazing behaviour resulted in the bes.t correlation. Quality and composition of diet as presented in Table 4.3 thus fanned the basis of the following equation: VA = 2*BUY + BUO + 4*SHO + 2*ROS + FOR in which:

Eqtu~tion

7.1

VA = Vegetation attractiveness (%) BUY= Green bunch-grass cover (%) BUO =Dead bunch-grass cover (%) SHO = Short grass cover (%) ROS =Cover of ground rosette layer (%) FOR = Tall forb cover (%)

By defining this estimate of vegetation attractiveness, both forage quantity and quality aspects are included. Funhermore, it allowed an analysis of the influence of forage availability independent of site and management variables. In the study area, distance travelled by cattle depends mostly on milking practices and on the fact that farmers prefer to have the cattle close to their fincas so they can keep an eye on them. Travel distanee is related to two variables: distance to ftnca and elevation. These had been recorded in every releve. The grazing areas pertaining to the different farmers and to INDERENA were delineated on aerial photographs. The interpretations were digitized directly from the photographs, and the resulting vector maps were geometrically corrected following the same procedure as described in Chapter 8, Section 8.2. 153

Chapter 7 Spatial modelling of carrie disrriburion

Stocking density was then calculated as the total numb er of cattle of one fanner , divided by the grazing area corresponding to that fanner. ·.

Code

Variab le

Units

ELEV DIST

Elevation Distance to finca Fire age ln(Fire age) Grazing intensity In(Grazing intensity) Stocking density Jn(Stocking density) Slope percentage Vegetation attractiveness Block lava, nominal Colluvial slope, nominal Glacial depression, nominal Glaciated Java field, nominal Lateral moraine, nominal Valley bottom, nominal Terrain preference

Table 7.1

variables examined for the regressiorrmodel.

' The Spearman rank correlation coefficients give insigh t in the interrelationships between the different variables, independent of the scale of measu rement. Whelf data are not bivariate normally distributed, the calculation of coefficients of rank correlation is a common method to test the significance of association between two variables (Sokal & Rohlf, 198\ ).

7.2.2

Preparation of aUribute maps

After selecting the most appropriate variables for furthe r inspection, stepwise regression was executed several times, using the full data set based on releves. The default alpha to enter or remove a predictor from the model (0.15) was used. Interactive stepwise multiple regression (both forward and backward stepping) instead of the automatic procedure gave the same results. During each trial, at least one terrain variab le was made passive, since the occurrence of a certain terrain unit is correlated to the absenc e of all others. Relative preference values were established for each terrain unit by taking the corresponding regression coefficients resulting from subsequent analyses. A general terrain preference variable was then composed as the summation of the individual preference values multiplied by the nominal terrain variables. The composition of this terrain preference variable is presented in Table 7.5 of the results.

VA:

7.2.1 ·Statistical analysis of point observations A simple parametric model was developed, based on multiple regression between point observations on cow droppings and the values of bioph ysical and management variables for the same locations. An important assumption is that the samples are representative of the full range of grazing intensities and their frequency of occurrence. The statistical analysis was carried out using the SYSTAT package (Wilkinson eta/., 1992). . The examined variables are presented in Table 7 .1. The terrain variables were initially included in the analysis as nominal variables, with values 0 or I. Instead of a regression analysis per group of similar terrain units or for single units, it was preferred to include them in the overall analysis. Only then can the relative prefer ence values for terrain be quantified as the influence of the general accessibility of the type of terrain, independent of variance due · to differences in slope or vegetation attractiveness. First, correlation matrices were constructed to evalua te the significance of the variables considered and their covariance. Pearson and Spearman correlation coefficients were calculated using the SYSTAT package. From the Pearso n correlation coefficients; it could be derived which transformation of variabies gives the highest correlation with grazin g intensity. Elevation and distance to finca (DIST), for example, were also tested as squared standardized values and as a joint variable composed of the (weig hted) sum of both. Although during the . : first attempts, slope measured in degrees gave a slightly better correlation with grazing intensity, slope percentage (SLOPE) was selected for practical reasons, as it can be easily calculated from the digital terrain model.

FILA LNFI GRI LNG SD LNSD SLOPE VA B

c I)

G L

m yr

droppings/50m2 A.U./km2 % (%)

. appeared to make a significant contribution Attnbute maps were prepare d of all. variab.les that . . . to the explanation of the variance m graztng mtens lly.

SLOPE:

SD:

Ma of ·vegetation attractiveness. For every unit of the. vegetation map in Fi ~re 3.4 (Chapter 3), an average vegetation anracuveness v_alue was cafculated based on the point data of the releves. The onl~ excepnon wa~ a small patch of tall grassland after potato cultivatio~, for ~_Vhtch an assumpt~on had to be made as the observations· on cattle diet dtd not tnclude consumpnon of tall grasses. Slope map. Using the digital terrain model, a slope gradient .map w: t dard procedures (ILWIS user s manu calculated in ILWIS accord.mg to san • f th "te A 1993). Pixel values cortespond to _the ~verage slope percentage o at st . classified slope map is presented tn Ftgure 2.5 (page 160). Map of stocking density. This map is derive~ by grazing areas. Ln-transformed values of stocking cl~ssi~:t~~2~a~e~ denstty . assigned to each grazing area.

154

155

Chapter 7 TP:

Map of terrain preference. Terrain preference values were assigned toeach unit of the terrain map in Figure 2.4 (page 157) according to the outcomes of the regression analysis. As grazing influe~ce was hardly observed in the superparamo, terrain preference was set equal to that for cushion bogs. Even the so-called 'blue grass' fields of the lower superparamo (Salamanca et al., 1995) are to a large extent inaccessible in the study area. Water bodies given the arbitrary value of -1.

DIST:

Distance map. This map is based on a raster map showin g the positions of the six fincas and on the map of grazing areas. A distanc e in meters was calculated for each pixel, relative to the geographical position of the finca to which each grazing area belongs. The formula applied is: Dist where Dist dx dy p

/(dx

+ dyl)

TERR AIN

= distance to tinea (m) = distance in S-W di rcc~on, no. of map columns = distance in N-W direction, no. of map rows

= pixel size, in this case 10 m

This procedure was repeated for each grazing area; the-pa rtial distance maps per grazing area were joined to obtain the final distanc e map DiST. All maps were prepared with the same geomenic transfo rmation and cover the same area, coinciding with the area represented in both vegetation and terrain maps. Pixel size is 10 * I0 m. The maps were classified for presentation purpos es. Âˇ 7.2.3

Map of cattle distribution

The procedure of creating a grazing intensity map is illustra ted in the flow diagram of Figure 7.1. Overall grazing preference values were calculated by filling in the values of the different attribute maps in the regression equation. Original, non-cl assified atnibute maps were used for these calculations; the intermediate results were In-tran sformed grazing preferences. A new map was created representing linear grazing preference values per pixel. Water bodies were given an arbitrary terrain preference value slightly lower than that for the glacial depressions, and a vegetation attractiveness value of zero. Calibration of grazing preference was executed for the lowest values: calcula ted grazing preference for water bodies already approximated zero but was set to zero manua lly. Value s for superparamo w~rc also slightly lowered. These units were obviously undersample d but, by fixing the lowest values at near zero or zero, the reference level was satisfied. These grazing preference values were aggregated per grazin g area. In order to predict grazing intensity per pixel, the grazing preference value of each pixel was compared with the sum of grazing preferences per grazing area according to the following equation:

156

3 km

Superparamo GlaCiated lava field Glaciated lava field, steep slopes Glacial depression

f#ÂĽi1 Lateral moraine I==:J Lateral moraine, moderate slopes c=J Lateral moraine, steep slopes

1-,::. =~I

End moraine

Lacustrine flat I cirque bottom

CJ Valley

Block lava

. . Colluvial slope

Block lava, steep slopes

CJ Lake

Figure 2.4 TERRAIN MAP of the upper water Park, Colombia. Based on aerial photograph shed of Rio Otun, Los Nevados National s of 1983-1989 (scal es 1: 25,000 â&#x20AC;˘ 1: 30,00 0) the physiograph ic map of Villota (1984).

. •co.

u;;

•twt

Clu:Jpter 7

PGR1 1 = where

Spatial modelling of cattle distribution

HERD • GP 1 ---.!__ --

7.3

:Ei>l •.o (GP 1 * A 1 )

PGRI 1 =~redic~ed grazing i~tensity of pixel (J) (A.U./kml) HERD = lierd s1ze of a gTaZing area consisting of n patches or pixels (number of cattle, A.U.) A1 = area over which gr.uing intensity is calculated in. this case pixel (i) (100 m2) • ' GP 1 = grazing preference value for pixel (i)

The results of the correlation analyses are presented in Tables 7.2 and 7.3.

ELEV DIST

~e .~h?le procedure. is

nothing _more than the calculation of a key to probable s!O unon. m ~e ~ng area. Pomt data are used to build up a spatial model by means "':htch. gr8Zlng mtenstty is predicted for each spatial unit In this case, the spatial units ptxel~, however, for other types of spatial units the procedure of predicting grazing remams the same. ·

VEGETATION

ClASSIFJCATON

Sources of variation in grazing intensity: regression analysis

VEGETATION

ELEV DIST FILA SD GRI LNFl LNG LNSD SLOPE TP VA

1.000 0.595 1.000 0.021 -0.227 -0.382 -0.075 -0.025 ·0.203 0.042 -0.236 -0.175 -0.290 -0.359 -0.073 O.o35 0.010 0.146 ·0.164 ·0.131 -0.178

ALA SD

GRI

LNFI LNG

1.000 ·0.277 1.000 0.183 -0.001 1.000 0.904 ·0.245 0.185 0.158 0.129 0.698 ·0.293 0.878 0.003 -0.249 0.070 -0.264 0.080 -0.158 0.145 0.262 -0.058 0.274

1.000 0.199 ·0.274 -0.263 0.042 0.349

LNSD SLOPE TP

1.000 0.169 1.000 ·0.308 0.089 1.000 0.270 -0.069 0.102 1.000 0.466 ·0.102 -0.168 0.179

1.000

ATTRACTIVENESS

Table 7.2

TERRAIN

CLASSIFJCATON

PREFERENCE

OTM

Pearson correlation matrix.

TERRAIN

SLOPE FlR'ICTON

MUlTIPlE REGRESS ON EQUATON

GRAZING

SLOPE

PREFERENCE

From the Pearson correlation matrix, a selection of variables to be included in the regression an<!lysis was made. In Table 7.2, it can be seen that, of the variable s representing l~al grazing intensit}(, the In-transformation of grazing intensity (LNG) gave the highest correlauon coefficients with other variables. Analogous to this, the In-transformatio n was also selected for the variables fire age (LNFI) and stocking density (LNSD).

DISTANCE TOFJNCA

STOCKING DENSITY HERD SIZE PER GRAZING AREA

GRAZING INTENSITY

Figure 7.1

ELEV Dist LNFI LNG

CALJBRATON

Flow diagram representi?g the P~ure 10 create a map of cattle distributiorL Every box rep~ents a ~ap. D1ffereru attribute maps are classified or used as input in map calculations to denve .the allribute maps required as input in the regression equation. The ~erd of each grazJng area is then distributed over the area according to calibrate d grazmg preferences.

LNSD SLOPE TP

ELEV 1.000 DIST 0.494 1.000 LNFI -0.057 ·0.320 1.000 LNG ·0.155 -0.239 0.151 1.000 LNSD -0.327 -0.062 ·0.167 0.192 1.000 SLOPE -0.066 -0.057 -0.284 ·0.267 0.095 1.000 TP 0.142 ·0.240 0.102 0.265 -0.145 0.055 1.000 VA -0.127 ·-0.277 0.294 0.492 0.057 -0.131 0.149 1.000

Table 7.3

Matrix of Spearman rank correlation coefficients.

158 159

TERRAIN PREFEREN CE

, ......

0 2

â&#x20AC;¢

0-2% - 1.00

0.00

- 0.98

c::J

0.32

55- 140 %

0.85

> 140 %

. . 2 -7 %

20 - 55 %

D D

7-13%

~ 13-20%

3 km

Figure 2.5 MAP OF SLOPE CLASSES, based on the DTM . Terra i ~ segments are overlaid.

Figure 7.2 TER RAIN PREFERENCE MAP. Preference values (Table 7.5) were assigned to the terra in map (Figure 2.4). Terrain segments are overlaid.

Spatial modelling of ca11fe distributi on

Chapter 7

According to Webster & Oliver (199 0), in routine surveys, where data arc: recorded for variety of purposes, strong correlatio ns are generally not the case. Published values correlation coefficients between soil properties lie mainly in the range 0.3 to -0.3 and a few exceed 0.5. A practical considera tion is the fact that, if two variables are _strongly correlated, only one needs to be reco rded. In this light, some correl~rion ~oeffic1ents ~ surprisingly high. Vegetation attractive ness shows the strongest correlauon w1th LNG (0.4.66 and 0.492, Pearson and Spearman coef ficients, respectively). All variables taken into account in the regression analysis have correlatio n coefficients higher th~ 0.1~, which. m:ans that single contributions of these variables to the explanation of vanance m are s1gr uficant .. , Summarizing the stepwise multiple regression was executed with the varia bles (ELEV), dis~ce to finca (DIST), ln(fire age) (L~), ln(stoc~ng density) (LNSD), percentage (SLO ), terrain preference (fP) , and vegetauon attracuveness <'!A ). Th~ results presented in Table 7.4. Elevation (ELE V) and fire age_(LNFl) ar_e not mcluded 1~ th.e . regression equation: their contributions to the explananon of vanan~e v.:ere. not Significant. The reasons for this can be derived from Tables 7.2 and 7.3. Elevanon IS h1ghly correlated with distance to tinea, and the latter is a better predictor of grazing intensity. Parallel to this! fire age shows a high correlation with vege grazing intensity. A,more detailed discu tation attractiveness,: but is~ P?O":r predictor. of ssion on the role of fue m the dismbutto n of grazmg patterns follows in Chapter 8, Section 8.4.

Variable

VA • 10·2 SLOPE LNSD

0.466 0.52 1 0.571 0.619 0.640

0.217 0.271 0.327 0.383 0.410

DlST • 10"2 Variable

Coefficient

s.e.

VA • 10·2 SLOPE LNSD

0.655 -0.015 0.197 0.626 -0.013 0.844

0.122 0.003 0.056 0.194 0.005 0.323

5.372 -4.241 3.525 3.235 -2.468 2.616

Mean square

DIST • ·10..2 Constant

R 2 change

0.217 0.054 0.056 0.056 0.027 p (2 tail)

0.000 0.000 0.001 0.002 O.GI5 0.010

ANALYSIS OF VARIANCE Sources

Sum of squares

Regression Residual

92.640 133.408

136

Table 7.4

18.528 0.981

18.888

Results of the multiple regression analysis. Dependent variable is ln(grazing intensity). n=142. R2 change is the increase in expla ined variance contributed by each predictor . 162

Hence, the final regression equation base d on point observations of grazing inten sity has the following form: LNG = 0.00655*VA -0.0 lS*SLOPE +0.1 97•LNSD -H>.626•TP -0.00013•DIST -HJ.8 44 EqUiltion 7.4 An explanation of the abbreviations of variable names is given in Table 7.1.

7.4

Grazing behaviour and terrain variable s

Of all variables included in the regre ssion equation (equation 7.4), the cont ribution of SLOPE is the one that varies most. With a mini mum slope value of 0 % and a maximum of 440 %, the contribution to the regression equa tion ranges from -6.6 to 0. Although vegetation attractiveness explains the larger pan of the variance in grazing intensity, its range is narrower: from 0 to 2.1. The explanati on is that extreme slope values do not occur frequently: the area occupied by valley bot~oms, other level areas, and very steep slop es is small. This is demonstrated in the classified slop e map· of Figure 2.5, page 160. Cattle tend to choose relatively easy walking routes to meet their goals. Besi des the slope variable, the general accessibility of the terrain strongly influences cattle distributi on. This can be seen from the values of terrain preference that resulted from the regre ssion analysis presented in Table 7.5. The resulting terrain preference map is presented in Figure 7.2 on page 161.

The most obvious example is formed by the glacial depressions associated with cushion bogs: due to the irregular topography of hum mocks and water-filled hollows, these units are hardly visited for grazing purposes (TP= -0.98 ). Still, quite a number of palatable plants occur,, including tussock grasses with a high proportion of green biomass. Somethin g similar holds true for the block lava flows. With their rough and irregular topography and the presence of large and small rock blocks interming led with deep holes, these are dangerou s areas where cattle may .break limbs. Farmers some times lose cattle through this type of accid ent, when the injured animal is not found in time. Cattle tend to avoid the block lavas, except for the smoother parts and some walking route s towards more attractive terrain. Only in periods of forage scarcity are the block lavas visit ed more frequently. The most accessible terrain units are the smooth valley bottoms, colluvial slopes, and lacustrine flats (TP= 0.85). During the analyses, the samples corresponding to these units were combined, as their relation to grazing intensity was very similar. The glaciated lava fields and lateral moraines occupy an intermediate position on the scale of accessibility. The glaciated lava fields were kept passive as a nom inal variable during all regressions; their value therefore corresponds to the zero preference reference level. As expected, the smooth slopes of the lateral moraines have a higher accessibi lity than the more irregular glaciated lava fields. A factor of local importance, which has not been incorporated in the mod el sofar, is the distance to sites of special attractiveness to the cattle.

163

VEG ETATI ON ATTRACTIVENESS

DiSTANCE TO FINCA

•

3km

0- 500 m

150 . 200

0 . 100

200- 250

100 - 150

500- 1000 m 1000- 1500 m 1500 · 2000 m 2000-2500 m

250- 350

2500- 3000 m 3000- 3500 m

Figure. 7.3 MAP OF V~GETATION ATTRACTIVENESS CLASSES. Values of vegetation attractiveness for waz1ng (Table 7.6) were ass igned to the vegetation map (Figure 3.4). Segments of terram umts are overlaid.

3 km

....c=J c=J CJ c=J c=J

3500 - 4000 m 4000- 4500 m 4500 - 5000 m 5000- 5500 m 5500- 6000 m 6000- 6500 m

MAP OF DISTANCE TO FINCA of the six grazing areas, distance classes of

Chapter 7

Terrain unit

Terrain preference

Block Java Colluvial slope Lacustrine flat I cirque bottom Lateral moraine Glacial depression Glaciated Java field Valley Table 7.5

Spatial modelling of cattle distribution

- 0.41 + 0.85 + 0.85 + 0.32 -0.98 0.00 + 0.85

Preference values per terrain unit, outcomes of multiple regression

analys~.

Examples include some drinking places: the immediate surroundings of a few small glacial lakes of easy access show higher densities of cow droppings. Other places more frequented by cattle are the slightly concave tops of lateral moraines, which are used as wind-sheltered resting places. The influence of travel distance is discussed in Section 7.7.

Floristic composition

Dense bunch grassland (d)

154 16

147 8

160 25

Mediumdense bunch grassland (h) Open bunch grassland (o)

133

119 28

152 20

151 39

165

Mixed grassland(m)

206 32

164 36

208 29

Structure Cushion bog (c)

Shrubland(a)

7.5

133

As short grasses have the highest preference in comparison with all other functional groups, it is not surprising that the short grasslands gF and gH show the highest VA values. Type gH occurs in the lower zone after potato cultivation. In the vegetation map in Figure 3.4, gF covers an extensive area. Next to the short grasslands, both mixed grasslands and short herblands are most attractive to cattle. Shrublands of floristic types F and G are characterized by high covers of short grasses and forbs, whereas shrubs do not contribute to VA. Their VA value therefore resembles the mixed grasslands and ~hort herblands ofF and G, respectively. There is a marked general tendency within the diagonal matrix of floristics versus structure (Table 7.6): in both the horizontal and vertical directions, VA reaches an optimum approximately in the middle. Going from left to right and from top to bottom in the matrix, grazing impact increases (see Chapter 3). It can be concluded that, with increasing grazing intensity, the cattle transform the natural vegetation (dC, dD and dE) into vegetation types with a higher forage availability.

166

250 43

Short grassland (g)

312 50

Short matted herbland (s)

249

Tall grassland (t)

Table 7.6

211 68 220 310 245 45

n.a.

Forbland (f) Sparsely -vegetated land (e)

G -

230 18

Forage availability

For each unit of the vegetation map (Figure 3.4), a vegetation attractiveness value was calculated from the cover values per functional plant group using equation 7 .1. This measure of forage availability is listed in Table 7.6 for each occurring combination of floristic type with vegetation structure. From Table 7.6, it is clear that the vegetation attractiveness (VA) of type C decreases when the vegetation opens up (from dC to hC to oC). There is only . limited replacement by short grasses and forbs. In types D and E, a tQlnsition is noticeable, towards an open bunch grassland with a higher cover of short matted species and therefore a higher VA.

98 19

202 15

n.a.

n.a. ;

187 37

213 40

175 109

108 23

Values of vegetation attractiveness (VA) for each vegetation type as a combination of vegetation composition (columns) and structure (rows). Staridanl error is indicated immediately below in italics, when sufficient data are available.

167

PREDICTED GRAZING INTENSITY

FIRE AGE

(Chapter 8)

3km

fill OA.U./ha h::41;] 0.000: 0.025 A.U./ha

Ia 0.025 - 0.050 A.U./ha

0.125-0.150 A.U./ha

H,¥NJ 0.150-0. 175 A.U./ha lllj 0.175 - 0 .200 A.U./ha

~·~~~ 0.050- 0.075 A. U./ha

k~ 0.200- 0.225 A.U./ha

0.075 - 0. 100 A.U./ha

~ 0.225- 0.250 A.U./ha

(:=J

0. 100-0. 125 A.U./ha

Ell

> 0.250 A. U./ha

Figure 7.5 MAP OF CATTLE DISTRIBUTION, indicating classes of predicted grazing intensity. Segments of terrain units are overlaid.

0 - 2 years

6- 10 years

> 10 years

Ill

2 - 4 years

l \'j~)

4- 6 years

No fire detected on AP lime series

Figure 8.2 FIRE AGE MAP, indicating the time before 1989 that a certain area was burned or that no fire was detected on the AP time series of 1955 - 1989. Segments of terrain units are overlaid.

Chapter 7

Spatial modelling of cattle distribution

Forage quality, especially, increases; forage quantity rather decreases. Subsequently, trampling impact- inGreases further and quantities of yielded forage are continuously forage availability measured as VA decreases. This process of increase and suclse<tue1it~i~.l decrease in VA was als.o described by Pels & Verweij (1992). The ·VA values of Table 7.6 were assigned to the vegetation map. The resulting map · vegetation attractiveness is presented in Figure 7.3 on page 164. In the northwestern part the study area; vegetation units show contrasting VA values. Units with short and m1'J/,.,EI;,<;R grasslands are intermingled with cushion bogs of low vegetation attractiveness. ImrnediateiY.1~ east and south of the Laguna del Otl1n is the area owned by the National Park Service INDERENA. This area shows low VA values corresponding to the vegetations. Weighted averages of VA (included in Table 7.7) do not differ much among the areas. It can be concluded that the overall forage potential in all six areas is similar, the variance.in VA among spatial units may be higher or lower. The high correlation of VA with the observed grazing intensity and its important contribution

to R2 of the multiple regression (0.22) both indicate that this estimate of forage availability

is effective.

The associated stocking densities per grazing area are indicated in Table 7.7. Even in area Ill, which is designated as a pure nature conservation area (owned by the National Parks Service), a few cows are present. The grazing areas of land users II, IV, and VI have higher stocking densities. The short maned grasslands and herblands are concentrated in their grazing areas. The state of degradation of the vegetation patches in relation to grazing intensity will be evaluated in more detail in the last chapter.

Grazing area Managed by

Ill

farmer

INDERENA

farmer

fanner

Herd size

'90

Stocking density (A.U./kmz)

4.0

-9.2

0.2

14.8

6.6

8.6

Mean vegetation attractiveness

189

146

136

1.75

136

138

Table 7.7

7.6

farmer

Herd size and stocking density figures for the six grazing areas represented in the map of Figure 7.4 (page 165). Data refer to the situation by the end of 1990.

Grazing management by man

In the introduction to this chapter, several relevant management variables were identified: stocking density, travel distance, limitation of the grazing area, and the use of fue. The role of fire in the distribution of grazing patterns is evaluated in the next paragraph. · Both stocking density and distance to finca proved to be significant predictors of grazing intensity. In Figure 7.4 (page 165), the variable distance to finca is represented by means of distance classes of 500 m, calcul~ed per grazing area. That this was the fifth significant predictor is related to the fact that farmers tend to keep their herds close to the farms in order to have some control over the animals. Partly, this is due to milking and vaccination practices. · On the ether hand, reasons of safety also play an important role. In remote areas, cattle are sometimes stolen. The likelihood that an injured animal is not found also increases with distance. Besides grazing management by man, preference by cattle probably also influences the distribution pattern in relation to the position of the fincas. Elevation is related to distance . to finca (Pearson correlation 0.595}, which implies a preference for the lower areas where temperatures are also more favourable. To what extent these variables detennine travel distance is not known, but it has been clearly demonstrated that distance to finca is inversely related to grazing intensity. The overall forage potential in the six grazing areas is similar. This is expressed by mean vegetation attractiveness values that range from 136 to 189 (Table 7.7). However, stocking density varies between 0.19 and 14.8 A.U./km2• Stocking density is of course very much . related to the limitations of the grazing area. 170

7.8

Spatial model of cattle distribution

<·

Figure 7.5 (page 168) shows the resulting map of predicted cattle distribution, classified into 12 classes on a linear scale. This spatial model should be interpreted as a prediction of grazing intensity at meso-scale. The variance in cattle distribution over short distances (< 50 m) due to the presence of dripking or resting places, for example, is not taken into account. Local higher concentrations of cow droppings, which are not accounted for in this spatial model, are mostly related to resting and drinking activities. The map of cattle distribution models grazing behaviour rather than occupancy. In the map of grazing intensity, it can t>e noticed how relatively small differences in overall stocking density affect cattle distribution. Grazing areas II, IV and VI have higher stocking densities. Grazing intensities exceeding 12.5 A.U./km2 are associated with these areas and do not occur elsewhere. The occurring short matted grasslands and herblands are also concentrated in these grazing areas. The southern part of the study area, which has a marked fue history, shows intermediate to relatively high grazing intensities. The state of degradation of the vegetation patches in relation to stocking density will- be evaluated in more detaiJ·in the last chapter. The highest grazing intensity value represented in the map is 34.1 A.U./km2• The average grazing intensity for the entire area studied, is 3.3 A.U./km2 • These figures seem very low. Data on stocking densities in paramo ecosystems have not been encountered in the literature. However, a basis for comparison is provided by the study of Bradford eta/. (1987) on bunch grasslands in the moist puna belt of central Peru. 171

(Chapter 8) FIRE FREQUENCY

Sparia/ modelling of cattle distribution Elevation range-and precipitation values are comparable to those of the study area. On glaciated mountain valley slopes, a system of rotation grazing is pra<;tised. The recommen ded stocking rate is 3 ewes/ha/yr for sheep and alpaca. When convened to FAO livestock units, the equivalent is about 0.3 L.U./ha or A.U./ha (FA9. 1991). Hence, the favourable grazing conditions of the gentle puna slopes can be compared to the most heavily grazed paramo sites. The puna, however, has been grazed by domestic livestock for over 10,000 years (Engel, 1976). No quantitative data are available to validate the presented model. However, researcher s who studied other aspects in the area, such as cattle feeding behaviour, con finned the general tendencies in cattle distribution predicted by the above model. This can be accepted as a positive qualitative validation of the model. It can be concluded from the above that a spatial model of cattle distribution for this type of mountainous area can be based on dropping counts. A systematic sampling procedure by means of transects with direct observations of cattle numbers would be very time-consu ming. In other areas of the world where visibility is limited, dropping. counts have also been applied successfully. An example is the census of forest elephants (Shon, 1983; Jachmann & Bell, 1984; Merz, 1986; Barnes eta!., 1991). In the case of sheep grazing on heathlands, a measure of comparative grazing intensity was calculated based on the number of droppings , after establishing a clearcut correlation between sheep and dropping counts (Bakker eta/., 1983a). In this approach, it has been shown how a selected combination of statistical techniques is applied for the purpose of spatial modelling: interpolation techniques as standard procedures in GIS, e.g. in the calculation of slope and distance_ values per pixel; stratification of the region into classes according to the type of vegetation and physiography; classificat ion of point data.and the subsequent extrapolation of terrain and vegetation Âˇ preference values by assi.gning these values to the corresponding spatial units; and multiple regression analysis as the basis for extrapolation of grazing intensity values per pixel (or per spatial unit).

Zero to very low (no fire detected on AP)

Low

3 km

(1 fire observed)

High

(3 fi res observed)

Very high (4 or 5 fires observed)

Medium (2 fires observed)

s.3 MAP OF FIRE FREQUENCY. The AP series covers about 30 years, within the from 1951 to 1989. Segments of terrain units are overlaid.

In grazing studies, cattle preferences for particular vegetation communities are usually detennined by comparing the actual with the expected number of sightings. Expected values are calculated on the assumption that if cattle utilize the vegetation communities randomly for grazing, the frequency of sighting per vegetation unit must correspond to the proportion of the study area covered by that vegetation unit. An example is described by Van Rees & Hutson (1983). The disadvantage of calculating crude preference values is, however, that little _ is understood about the mechanisms that underlie these preferences. This implies that neither spatial variability of relevant land attributes nor temporal variability of the vegetation as a response to grazing or fire management are considered. As a starting point for spatia-tem poral modelling, these aspects need to be taken into account. Funhermore, preference factors for vegetation units do not provide infonnation about the contributions to the observed preference by variables other than forage availability, such as slope. Therefore, predictions on the response of grazing behaviour to a changing environment should be based on a set of detailed, independent preference values related to variables that explain the variance in grazing intensity as much as possible.

173

Chapter 7

LITERATURE CITED Arnold, G.W. & Dudzinski, M.L. 1978. Ethology of free-ranging domestic Developments in animal and veterinary sciences No. 2. Elsevier, Amsterdam. Bakker, J.P., De Bie, S., Dallinga, J.H., Tjaden, P. & De Vries, Y. 1983a. Sheep-grazing aS a management tool for heathland conservation and regeneration in the Netherlands. . · "'- · Journal of Applied Ecology 20: 541-560. Bakker, J.P., DeLeeuw, J. & VanWieren, S.E. 1983b. Micro-patterns in grassland created and sustained by sheep-grazing. Vegetatio 55: 153-161. Barnes, R.F.W. , Barnes, K.L., Alers, M.P.T. & Blom, A. 1991. Man detennines distribution of elephants in the rain forests of nonheastern Gabon. African Ecology 29: 54-63. Bradford, P.W.• Bryant, F.C. & Fraga, V.B. 1987. An evaluation of range condition on range site in the Andes of Central Peru. Journal of Range Management 40(1): Engel, F.A. 1976. An ancient world preserved: relics and records of prehistory in the Crown Publishers, New York. FAO, 1991. Guidelines: land evaluation for extensive grazing. FAO SGils Bulletin No. FAO, Rome. ILWIS 1.4 User's Manual 1993. ITC, Enschede. Jachmann, H. & Bell, R.H.V. 1984. The use of elephant droppings in assessing nunnt>eJ:S; occupance, and age structure: a refinement of the method. African Journal of 22: 127-141. Merz, G. 1986. Counting elephants (Loxodonta africana cyclotis) in tropical rain forests particular reference to the Tai National Park. Ivory Coast. African Journal of 24: 61-68. Pels, B. & Verweij, P.A. 1992. Burning and grazing in a Colombian bunchgrass ecosystem: a transition model. Pp. 243-263 in: Balslev, H. & Luteyn, J.L. (eds.), Paramo: an Andean ecosystem under human influence. Academic Press, London. ,.. Rice, R.W., MacNeil, M.D., Jenkins, T.G. & Koong, L.J. 1983. Simulation of the herbage/animal interface of grazing lands. Pp. 475-488 in: Lauenroth, W.K, Skogerboe, G.V. & Flug, M: (eds.), Analysis of ecological systems: state-of-the-art ecological modelling. Deve1opmems in Ecological Modelling 5, Elsevier, Amsterdam Salamanca, S., Cleef, A.M. & Rangel, J.O. l 995. The paramo vegetation of the Ruiz- Tolima massif. In: Van der Hammen, T. & Dos Santos, A. G. (eds.), Studies on Tropical · Andean Ecosystems 5. I. Cranner, Berlin. Short, J.C. 1983. Density and seasonal movements of the forest elephant (Loxodonta africana cyclotis Matschie) in Bia National Park, Ghana. African Journal of Ecology 21: 175184. Sakal, R.R. & Rohlf, F.J. 1981. Biometry (2nd edit.). W.H. Freeman & Company, New York, 859 pp. Stobbs, T.H. 1973. The effect of plant structure on the intake of tropical pastures.!. Variation in the bite size of grazing cattle. Australian Journal of Agricultural Research 24: 809819. Stobbs, T.H. 1974. Components of grazing behavior of dairy cows on some tropical temperate pastures. Proceedings of the Australian Society of Animal Production 299-302.

Spatial modelling of cattle distribution '/~n Rc~s, H. & ~utson , G.D. 1983. The behaviour of free-ranging cattle on an alpine ran e g In Austraha. Journal of Range Management 36(6): 740-743. W~b\ter, R. & Oliver, M.A. 1990. Statistical methods in soil and land resource su ~ecken,. P.H.T., Burrough, P.~ .• G_oodcliild. M.F. & Switzer, P. (eds.), Sp~~~ , .. . mformano~ systems. Oxford Umversuy Press, New York. 316 pp. -'·i!lk1nson,_ ~·· H11l, _M., Welna, J. & Birkenbeuel, G.K. 1992. SYSTAT for Windows: stausucs, vemon 5 ed. Evanston, IL: SYSTAT, Inc.

174 175

BURNING AS A MANAGEMENT TOOL

8.1 8.2 8.3 8.4 8.5

Introduction Monitoring fire events in GIS Reconstruction of fire history The role of fire in the distribution of grazing pattc ms Discussion and conclusions

Fire is common in most tropical alpine areas throughout the world. If fire is used as a management tool, it generally occurs at intervals of several years (Cleef, 1981; Smith, 1981). However, no quantitative data exist on fire frequency in tropical alpine ecosystems (Smith & Young, 1987). In this chapter, fire history of pan of the study area is reconstructed:

8.1

Introduction

Research in fire ecology has concentrated on succession after single fires. Longer-te rm analysis is needed to evaluate the relationsh ips between burning regime and landscape heterogeneity. Spatial variations in biophysical conditions have hardly been taken into account in fire models. Recently, efforts have been made to incorporate spatial variation in fire hazard models (Chuvieco & Congalton, 1989) and in fue spread models (Green, 1989; Davis & Burrows, 1994). Fire history maps over longer periods (of a few decades or more) that incorporate spatial variation can provide insight into the factors detennining fue occurrenc e. A palynological sequence from a site near the Laguna del Otun was studied; the sequence corresponds to the past 10,500 years (Salomons, 1989). Three different levels of charcoal were found, dating from 2400 to 7500 BP. During that periOd, the area was probably covered by Andean dwarf forest, and the charcoal pieces were products of fires caused by volcanic activity or lightning. Although little is known concerning fue frequency over the past 2400 years (during which time the study area has been covered by grass paramo vegetation), it can be safely assumed that natural fire hazard is low. Fires due to lightning or volcanic activity occur rarely, probably less than once in 1000 years. Cleef (1981) reponed that he had never _ observed any during two years of intensive fieldwork in pararnos. Horn (1989) also concluded that, for the Costa Rican paramos, there are as yet no documented cases of fues caused by lightning. She mentioned carelessly tossed matches and cigarettes, arson, and airplane crashes among the other possible sources of recent paramo fires. Although farmers who live in the study area sometimes accuse tourists of setting frre to the vegetation, they themselve s are considered the main culprits. The potential fire frequency of an area is detennined by the amount of rainfall and its distribution, and the availability of fuel, which is equal to the amount of burnable biomass remaining after consumption by the cattle (Mannetje & Ebersohn, 1989). Dry seasons are not pronounced for this paramo ecosystem and therefore burning can be practised, in principle, throughout the year. However, farmers make use of irregular periods of extended drought to set the vegetation on fire. 177

Chapter 8 Burning is an illegal activity within the boundaries of the national park. For this reason, was difficult to obtain reliable informati on related to the location and frequency of from the farmers. The following sections deal with a practical solution to this probl em.

Burning as a managemenl tool

. .,

8.2

Monitoring fire events in GIS

Recent fire history can be studied using satellite imagery (Chuvieco & Congalton , Minnich, 1983). Generally,longer-term information at a higher spatial rosolution is more cheaply by time series of aeria l photographs. However, scale difference s, displacement, and other distortions limit ed the use of aerial photographs for time analysis. Interpretations of aerial photogra phs used to be manually transferred to rnnnO Ton~. ;,;., base maps by means of an optical panto graph or a zoom transfer stereoscope, often leading to unacceptable inaccuracy when monitorin g large-scale changes. Nowadays, the cons truction of fue history maps is facilitated by the use of GIS facilities. This section deals with the development of a GIS method for monitoring frre even ts observed Âˇ on a time series of aerial photographs (AP) covering a time span of 35 years . The areas of Los Nevados National Park known as Baga Seca and Lorna Bonita, and the surro undings of Laguna La Leona(Âą 10 km2) were selec ted for this purpose. From a general inspe ction of the aerial photographs, it appeared that fires are more or less resoicted to this part of the study area. The aerial photographs used show a high level of detail, with scales ranging from 1: 12,000 to 1: 33,000. The size of burned patch es is sometimes rather small, up to 100 * 100m. The required accuracy of the projection of the vectors was estimated at a minimum of 10 to 20 m. The main difficulty encountered is transforming vector maps prepared from AP interpretations of various years in such a way that they all fit into one common coordinate system, thus making the precise spati al overlaying of vector information- poss ible. Several problems are related to this: Tilt, shifts, rotations, and scale difference s of the APs shou ld be corrected. In mountainous terrain, at least eight control points per photograph are requ ired to calculate accurate parameters for geor eferencing and transforming the vector maps. The topographical map at scale 1: 25,00 0, however, has a low level of detail due to the lack of mappable topographical featu res, which is related to the nature of the terrain. The small number of drainage intersections, lakes, and farmhouses appe aring on the map provide insufficient contr ol points for an accurate transformation of detailed vector information. Even when sufficient control points could have been localized on both the AP and the topographical map, relief displacement was expected to result in considerable shifts in the coordinates of the transformed vecto rs in comparison with their real coordina tes.

178

Figure 8.1

Reconstruction of fire history using a GIS and a time series of aerial photographs. This flow diagram illustrates the sequence of steps to prepare a fire age map and a fire frequency map of the southern part of the study area.

In order to solve the problems mentioned above, geometric correction was performed in a GIS according to the following procedure (see Figure 8.1). Aerial photographs of the years 1955, 1959, 1966, 1970, 1975, 1978, 1983, 1985 , 1987, and 1989 that cover the study area were interpreted Effective areas were caref ully drawn for each photograph ~o. tha.t the interpretations of adjacent photos woul d match. Three ftre age classes. were disnn gutshed: recently burned patches (black), patches with a longer time since burnmg (~ ~ey), .and unburned areas. Interpretations of the effec tive area in each photograph were diglt tzed mto separate segment files in photo coordinate s. A digital terrain model (DTM) is used as the height information map.f~r. geome~c correction. To this purpose, contour lines in a range of 3350 to 4700 m were digtUzed at mter vals of 50 m from topographical maps nos. 225-Ill-B (1974) and 225-N-A (1966), scale 1:25 ,000. ~fter rasterization and interpolation, the resul ting DTM contains height values for every ptxel. 179

Chapter 8

Burning as a management tool

Detailed procedures for creating a DTM are specified in the ll..WIS manual (1993). One ~erial photograph of !959 at scale 1: 33,000 was scanned, as this photo overvt~w of the area of mte~st and shows good contrast. Scanning was done

resoluuon of I 67 dots/inch, whtch corresponds to a pixel size of 0.15 mm on the 5 m on the ground..For the geometric correction of the raster image of the p~otograph, a resamphng procedure (affine transformation) including a correction for displacement was developed (Bargagli, 1991) using the information of the DTM. This orthophoto served as a raster base map with ground coordinates to which the interpretations of larger scale photographs were projected. Control points were extracted the orthophoto by ~reen digitizing. For other smaller-scale photos, control points were from the topographtcal map. By means of a monoplotting program (Bargagli , 1 segment maps of burned patches were transformed into ground coordinates. These are at present incorporated in ILWIS (1994). . · The orientation param~ters th~t dete~in~ the rot.ation matrix for the AP segment maps . calculated. By excludmg pomts with htgh restduals, the rotation matrix was •m~ uvc<L However, if the points with high residuals were important in the overall distributi 11 on of control points, they were not excluded. Vectors of adjacent effective areas were "v'''"'"·'w manual editing. After polygonization, the frre maps of the IOdifferent YC!ars were ra~tP.ri7Pti Patches burned several years before ·were sometimes confused with cushion bogs, as both have the same dark grey tone. The terrain map of Figure 2.4 (Chapter 2) enabled calibration of the .maps produced of bume~ areas for each year. The terrain map had been prepared accordmg to the p~ocedure descnbed above, by extracting control points from the orthophoto and ~e top.ographtcal ll_lap. Further details on the terrain map are given in Chapter 2. By supenmposmg the terram map on the frre maps, cushion bogs that had been classified as burned patches were discarded as such. GIS facilities of spatial modelling were also used to calculate the fin-;u maps of frre age and fire frequency, based on the separate fire maps of different years. Analysis of the time series revealed that burned patches are visible on aerial photographs for a period of three to five years. By overlaying frre maps of subsequent years, those burned spots that appeared . repe~tedly an~ were the results of the same fire event were deleted in the most recent maps. I~ thts way, btt maps of recent frre occurrence were created. By overlayin g the 10 maps of dtffe~nt years, a map of fire age was calculated, indicating the number of years elapsed since bummg, the year o~ reference being 1989. Summation of the ten bit maps of recent fire occurrence resulted m a fire frequency map. Fire frequency per major terrain unit was derived fro~ the b~med.CU:a over the entire AP series of nine complete coverage s. Burned area per maJOr terram umt 1s calculated according to the following equation: .

E;. • .n (fire frequency * area 1 ) Burned area%= 100 * - -1- = ------9 * E i=t...n (area 1)

Fire frequency was calculated as burned area divided by four(± 1), taking into account that burned spotnemain visible on APs for three to five years.

8.3

Reconstruction of fire history

8.3.1 Results 1 Fire history ·was reconstructed for a total area of 27.5 km1• Of this area, at least 7.1. km (16%) was burned during the last four decades (1951 to ~989). Th!s is a mi?imum esnmate as there are some gaps in temporal and spatial coverage m the senes of aenal photogra phs. Several patches were burned more than once. For the area studied, the average coverage of burned spots was 4.9%. Assuming that the darker tone of burned areas on AP lasts three to five .years, the overall fire frequency is estimated at 1.0 to 1.6 times per 100 years. For the northern part of the study area (not included in the analysis below), the average fire frequency is estimated at less than 0.1 time per 100 years.

The synthesis maps showing the fire history of the southern part of the stu~y area a:e presented in Figures 8.2 and 8.3 (pages 169 and 172). The fire age classes of the map m figure 8.2 are determined by the in~i:rvals between the last AP coverages (1989, 1987,1985, 1983, and 1978). Extensive areas that were recently burned are located far away from any tinea, e.g. surrounding the lake Laguna de Ia Leona From the fire frequency map of Figure 8.3, it appears that the same areas have been burned with a high O:Cque~cy; som~tim~ s even m?re than once a decade. A high fire frequency does not necessanly comctde wtth htgher ~mg intensities: the remote areas with a high frre frequency are merely used as stock for Urnes of forage scarcity. Close to the farms, the absence of a sufficiently high fuel load i~ pro~ably the main constraint to fire occurrence. Neither is the distribution of burned patches over the terrain units homogeneous. The lateral moraines, especially, are burned more frequently than other major terrain units. Coverage of burned patches and fire frequency of the major terrain units are presented in Table 8.1. The block lava and superparamo are burned less, probably because they are less accessible to both cattle and humans, and the presence of rocks, bare soil, or sparsely vegetated land leads to the discontinuity of fire. Regression of .terrain preference values as a predictor of frre frequency per terrain unit (fab!e 8.1) gives a correlation of 0.73 (Rl = 0.53). This means that farmers tend to bum those temun units that are preferred by cattle for grazing purposes.

Equation 8.1

where area, = the area of patch i out of n patches belonging to the same terrain unit

180

181

Chapter 8 Major terr ain unit

Burning as a management tool

Burned area (%)

Block lava Colluvial slope Glacial depression Glaciated lava field Lateral moraine Superparamo Valley Table 8.1

11------

1.6 4.4

0.0 5.6 10.6

4.9

Fire frequency per major terrain unit. The number of complete AP nine. Burned area (%)was calculated acco rding to equation 8.1. Fire •-~.. then derived, taking into account that burn ed spots are visible on APs for three years.

A - • •: ' " '

•

5 4

8.3.2 Evaluation of the method

An important indication of the accu racy obtained in the correction of the scanned imag~ · the subsequent transfonnation of segm ents is the root mean sums of squares directions (RMSx, RMSy, and RMS in the X xy). For the correction of the scanned image, an of 12 m was obtained. The topograp hical map coincides with the features of the scanned image. The transference of the vector informati on proved to be quite sensitive to the even distrib,utic>n and the accurate localization of cont rol points. For the calculation of the trailsfc>rmllltic>n parameters, 8 to 15 control points per AP were included. An RMSxy of 5 to ~ 5 m obtained for the transfonnation of the individual interpretations. In the final projection of . vectors, only small gaps or overlaps of segments of adjacent effec tive area s appeared. The differences were zero to two pixels, or 0 to 20 m. · The extraction of control points for each AP makes the procedure rather time consuming. Despite this limitation, it is worthwh ile to implement the developed meth od in other studies using aerial photographs, especially in those requiring a high accuracy, such as monitoring studies. The- main advantage is that intermediate transferring steps leading to higher inaccuracy are avoided. The extra ction of control points directly from a smaller-scale orthophoto is useful for large-scale APs in case the topographical map does not provide sufficient detail. ,'

An interesting question is whether burn ing practices have decreased since the creation of the . national park in. 1973. Fire occurren ce over time was therefore evaluated. In Figure 8.4, the area burned per AP coverage and fire occurrence in terms of the area burned annually are compared with the average fire frequ ency for the area analyzed.

182

----,

0.1

8.3.3 Fire occurrence in time

--------

/ /

/... . . .

"'··· · · ·.~·.....

........

........................ ./ · ··-L. ~.·~-/~-------~-~,~--~--.~~-=·==--~..~.•~.---..-._-,_ -_-___-..-...-.7/~------~-~-·-~·-.-,.~~--~.//~-.·~-... ol-----~r-----~~--~~~-

1955

1960

1965 .

1970

--:,9~7~5----~,~9~ao~--~19ess~s----~1~990

"* Burned area (%) ·-•·- Fire occ. ("(o/yr) Figure 8.4

Mean fire freq.

Burned area(%) and fire occurrence(% 1 yf ) in time for the s~uthem.p~ of the study area. Average fire frequency (1.2/100 yr) for the san1e area 1s also tndicated for entire period. Aerial photographs were available from 1955, 1959, 1966, _1970, the 1975, l978, 1983, 1985. 1987, and 1989, gene rally 1~en in J:muary. lnlerpolauon betw een observalions to·illuslrate the general tend ency IS tenlauve.

Interpolation between the years of AP coverage is only tentative, as the time span.":~~7n two subsequent coverages is sometim es fiv~ or six years. As .the burned spot s are V1Sl e or ·three to five years, some burning even ts m1ght have been m~ssed. In the graph of Figure 8.4, it can be seen that the change in m~nagement policy with X:g~ to burning has not resulted in a decr ease in fire occurrence smce _1973. Two peaks ~:S coverage of burned spots are observed : one in 1959 and the other m 1978 . The ye~ I and 1989 show another modest incre ase. After each peak, fire occurre.nce appe~ to. op. a~ can be inferred that, with a large exte nsion of area burn~. the potenual for burnmg .m are d attractive to cattle is also decrease d during the penod needed for regeneratton an accumulation of fuel load.

183

Chapter 8 Climatic data for the study area and its immediate surroundings are scanty, and even existe nt for the p.eriod.before 1980. Thus, it canno t be concluded from local data years of increased burning activity were generally warmer or had longer dry spells. '-UUiatili~ll data from the nearby H!MAT station of Las Brisas at 4150 m showed only that 1986 1987 were the wannest years during the period of registration from 1981 to 1987. The years 1958 and 1978 correspond to the low phase of the Southern Oscillation and called El Niiio years (Swetnam & Betancourt, 1990). These authors reported a cor:retaltiOri~ between fire oc<:urrence (in Arizona and New Mexico) and extreme phases of the "'""~.,_, .. Oscillation. Only the 1982 to 1983 El Nino episo de, the most severe of this century, show a substantial increase in fire activity in the United States. The same can be cmlc!u<led:;;;._a for the burned area in the paramo.

8.4

The role of fire in the distribution of grazing patte rns

Fire age The role of fire in the distribution of grazing patter ns is seen mainly via a temporary · in vegetation attr~ctiveness. This appeared from the multiple regression analysis, in which influence of fire age on the variation in gr:azin g intensity was evaluated. As a variable, fire age proved not to be a significant predictor of grazing intensity. As was · in the Spearman rank correlation matrix cif Table 7.3 of the previous chapter, fire age correlated with several other variables: with distan ce (·0.320), slope (·0.284), and ve~!eta.tion attractiveness (0.294). Now what do these correl ations mean? At medium to large distances from the fincas, fires are more frequent than nearby. This can be explained as follows. Where grazing intens ity is low, a high fuel load is present. Litter accumulates, cover by dead bunch-grass material increases. A function describing the critical minimum fuel load could be derived from the releve data, which was calibrated by the simulation model described in Chapter 6. The total cover of bunch grasses should exceed 50% and the percentage of dead material should be more than 55% to allow frre occurrence. If the · tussocks remain below these values, al!hough the spread of fire through the vegetation might not be totally impossible, in practice fire occur :rence has never been recorded in cases of smaller fuel loads. In the immediate surrounding s of the fincas, tussocks have become extinct,· or so fragQlented that fire potential approximat es zero. Especially in drier years, forage· resources near the fincas are depleted. Thus, fire is considered a management tool for stimulating the extension of grazing into areas at medium to large distances from the fincas. Fire age shows another negative correlation with slope. Steeper slopes are associated with a shorter time since fire. Once more, this phenomeno n can be explained by an evaluation of fuel ·: . load. In the valleys and on gentle slopes, bunch grasslands have generally been replaced by short matted vegetations which cannot be set on fire anymore. The map of fire age in Figure 8.2 was overla id with the map of vegetation attractiveness (classified map presented in Figure 7.3, page 164) and with the vegetation map. Weighted · averages of vegetation attractiveness and contri butions of vegetation structure classes were calculated for the total vegetation cover of each fire age class. 184

Burning as a management tool The results are presented in Table 8.2. From this table, it is inferred that vegetation attractiveness is not linearly related to fire age. VA increases for four to six years after fire, and then tends to decrease again. At what level VA stabilizes depends, of course, also on grazing intensity. This tendency is confirmed by the releve data, which show a peak in grazing intensity a few years after fire. From Table 8.2, it can be seen that a certain propo rtion of the vegetation (about 37%) returns to its original structure of dense grass land in two to six years' time after a frre. Dense grassland covers about 66% of the area in non-b urned situations. The vegetation structure of the other patches remains altered, probably due to a continued grazing pressure. The cover of open tussock grasslands decreases rapidly, but stabilizes at 13%, which is higher than in non-burned situations. The cover percentages of 'secondary' vegetation strUctures such as open bunch grass lands, medium-dense bunch grasslands, mixed grasslands, and -shon herblands are each lower than 10% in non-burned situations. More than six years after a f~re, each of these structures contributes more than 10% to the post·ftre regeneration. Fire age (yr)

0·2

2·4

4·6

no fire on AP

Vegetation attractiveness (VA)

131.4

143.0

171.7

157.9

145.4

Structure type % dense bunch grassland (d) 5 medium-dense bunch grassland (h) 1 open bunch grassland (o) 91 mixed grassland (m) 2 short herbland (s) 1 sparsely vegetated land (e) 0 other (c, I, a, t) 0 Table 8.2

34 20 32

39 5 22 12 22 0 0

37 22 13 16 11 1 0

66 9 4 10

3 0 0

1 4

Average vegetation attractiveness value 01 A) per fire age class, and relative contributions (%) of structure types to total cover per fire age class. Other structure types of non-burned vegetations are cushion bogs, tall grasslands, shrublands, and tall forblands.

After fire occur:rence, a single subclimax veget ation apparently does not exist, but rather a dynamic balance of vegetation structures among which transitions are possible. In ~uppon of the development of a discrete transition mode l for part of the study area, matnces were constructed that also indicate various possibilities of transition among different structure types (Pels & Verweij, 1992). The direction of structure changes, towards recovery or not, and ~e rate of the changes are probably determ ined by both management history and actual grazmg pressure. For the present work, the results of Table 8.2 can be considered a rough validation of the outcomes of the continuous simulation model presented in Chapter 6. Although not very detailed, the table provides a picture that is analogous to the findin gs of the simulation model. 185

Chapter8 Fire frequency As described in previous chapters, the regeneration process is constrained by frequent The reserves of the tussock bases decrea~e, the tussocks are subject to fragment ation, replacement by short matted species occurs. The map of fire frequency was combined thar of vegetation attractiveness to· evaluate the general ~endency. The resulting Table shows that only in the case of low to medium fire frequencies (once or twice every decades), is there an overall increase in vegetation attractiveness. If fire frequency is (once every 10 years or more), the average vegetation attractiveness goes down, and is lower than that of the natural vegetation. Hofstede (1995) also found that bu.ming lead to an increase in production or digestibility. Fanners who bum the paramo vegetation justify this by stating that forage ""u'"' IJu1;1v, increased. This is true for low fire frequencies. However, it is concluded that, vegetation is burned frequently, the farmers' justification is rrot valid for a time span than a few years. In the long term, f1re frequencies of more than once per decade anrlt>.J>r·'•-"':iltl cause degradation of the forage resources.

Fire frequency

Fire occurrence (per 30 y)

rero/very low low medium high very high Table 8.3

8.5

1 2 3 4 or 5

Average vegetation attractiveness 145.5 155.4 150.9 145.4 128.3

Average vegetation attractiveness value (VA) per fire frequency class. Times of fire occurrence correspond to the number of times that recent fire had been recorded on aerial photographs since 1955, covering an actual total lime span of 30 years. ·

Discussion and conclusions

The methodology described for monitoring fire events in a GIS, based on a time series of aerial photographs, proved to be an effective and accurate method of reconstru cting the fir( history of the study area. For those areas affected by recurrent fires, maps of frre age and frre frequency were produced. Recurring fires are considered an evolutionary force, because they promote the local _ extinction of species and the isolation of others (Patterson & Backman, 1988). Selection for fire-adapted species is a function of the intensity, frequency, and pattern of fires. Fire · temperatures in herbaceous fuels are relatively low, generally in the range of 100 to 400 •c ·_.: (Bailey & Anderson, 1980). This was confrrmed by Ramsey (1993) for grass paramo f1res in · Ecuador. The other two variables, frequency and pattern of f~res, are documen ted in this : · chapter.

Burning as a management tool Fire pat/ems Regarding fire patterns, several tendencies are remarkable. Besides the requirem ent of a minimum fuel load, it seems to be a matter of IJ'adition or personal preferenc e whether burning is practised or not The grazing areas in the southern part with a f1re history are managed by three different fanners. The northernmost grazing areas that are rarely or never burned correspond to the land managed by two other farmers and by the National Parks Service of INDERENA. The park wardens advocate the abolition of the use of fire, and actively try to convince the farmers not to bum. It might be. that farmers who live near the area owned by the National Parks Service have followed this advice, whereas others who live further away are less sensitive to the attempted control of burning practices. Second, if fanners have a tradition of burning the vegetation deliberately, they do so in tenain types that are most accessible, such as the lateral moraines, and at those places where sufficient fuel load is available. Deferred grazing in rem01e areas is sometime s related to previous burning, which is meant to generate forage of higher quality to be consumed in periods of scarcity. In the param~ ecosystem, burning represents an additional source of vegetation heterogeneity, which increases the concentration of cattle at recently burnt places and enhances the patchiness of the graZi!d landscape. Especially on slopes at medium to large distances ~m the fincas, burning play_s a role by enhancing meso-pattern. About 40 to 45% of the vegetanon that would correspond to a dense bunch grassland under non-burned conditions is maintained in a more open structure, even long after burning. These open and mediumdense bunch grasslands, mixed grasslands, and short matted herblands are sustained by grazing influence and are generally characterized by a higher forage availability.

Fire frequency A high f1re frequency influences vegetation dynamics on a time scale of several decades or more. It may stimulate some plant groups at the expense of others. Frequent fireS may cause important nutrient losses by run-off and volatilization, especially of nitrogen and sulphur (Hofstede, 1995). These losses slow down the regeneration processes, particula rly in the long term. If repeated burning is followed by intensive grazing, the vigour of perennial ·.grasses is reduced, leading to range deterioration (FAO, l 99l). lndeed, the tussock grasses in frequently burned areas were observed to be in very poor condition, with a strongly decreased tussock volume. Likewise, Espeleria seemed to be severely affected. The effects of burning and grazing on the bunch-grass and stem rosette populations are described in detail in Chapter 5. Noble & Slatyer (1981) defined the plant attributes vital for analyzing the response of vegetation to disturbance. These concepts were applied to f1re modelling, e.g. by Kes~el (1979). Two different categories of mechanism were recognired that allow plants to pemst through a local disturbance. The first of these relates to species that lack the ability to_sprout · after fire and thus depend on regeneration from seed and other propagules on the stte; the second includes species with a'vegetative regeneration from surviving plant parts. In paramo ecosystems, the dominant plant species belong to the latter category, as they resprout after a fire. This holds true for the stem rosettes, bunch grasses, shrubs, and probably .most of the ground-covering species. Autecological studies are needed to understand the response of other plant species; knowledge is also lacking on the content and dynamics of the soil seed bank.

186 187

CJwprer 8

Therefore, the effect of fue fre caus al relationships. . . quency on flon.snc diversity cannot be evaluated Floristic composition probably chan r.. the fire history maps, the map ac~uesalove . v lm~ spans of a few decades on!y a small proportion of burnof e. ed ve ~getauon, and Table 8.2, it can orbemor \OUIIICIUCII'n. probably were the original, preremain in the vegetation for a Ionfire v~et::~~t has c~an_ged ~ regards the floristic time gFl ?n.s. This II~~hes. that the characteristic structure. Based on the releve da~ h . f~nsuc composmon IS less dyn and discussed in Chapter 9 Sec amic than t" ' t9e5e ects of fue on plant dive rsity are further ' !On •. Because of high relative growth herbaceous rin sprouters over woodrates and. tille . . . ( g in grarnmo . 1ds, h1gh fire regenerating paramo vegetation inY spec1es Janzen 1973· Kru 1984)frequencies Cos ta Rica w·l r • frequent fires may determine a dom ' ger, . In a study inance ' I la.ms?n eta/. ( 1986) indeed ..u,,t~.;utnl'll A fire frequency of once a decade rna be by &:am mol ds at the expense of woody ecosystem. In the case of th burned areas is indeed low~ ~ar y f sufficient to preclude dominance by shrubs in ;mo o ~s Nevados,_ ~e :over by shrubs in rrec~ueJntl•··~ characteristic and common (type E vegdeF)tanon com~umnes m whi d de fi ch y.toody species eca s, ues were absent or negs an are restncted to . . most natural vegetations The on! ligible. Wood areas where dunn g ~e past . E species are, however, not dom · Yexcepuon areysom mant in the tree sea/Ionia, which prob e patc he · h h" ably . . s wit a lgh cover of the dwa in possible combination with gracould n h n . I ot ave perststed tn the case affects the cover by woody sp of a high fire . ~ng.b •_can be concluded that, in dominance. mos t areas, fire frequenc ecles, ut not to the extent that it determines their eventualy

Post

-fi~ s~cture differences either disa ear dense wlthm two to six years, orare sustJ:ed the . . grasslands growmg ?ave a low fuel load, the observed from open to, ..: dro in fire y grazmg. As the alte red vegetation structures • IS t~ be expected: the potential area cames sufficient fuel load is then f~r burnil~curre~ce after a ye-;r of high burning activity.. ' "d that IS both attracuve for cattle graz cons1 era bl ygredu ing and •;, ced. H?rn (1991) reported a minimum Rica. This is the period of post fire recurr . . -fire w~ce mt~r val of SIX years for the paramo subsequent fire. In the study area a ; fi reqmred to generate enough fuel in Costa ~ecorded in some cases. It does n~t msea~rt: ya atl~ recurren~e int~rval of 3.0 to 3.5 toyearcarr m three years. The minimum reco s was e "od t e vegetauon Will have recovere is estimated at about 10 years (Ch d completely a vte7spe~ for the bunch-grass paramo of the study area a strongly reduced bunch-grass co~e model described in Chapter 6 Mor r Jh· c:~uently-burned vegetation was observed to have ' I d so agre~d with the outc omes of the simulation . eover, a ecre re at ase tn forag 'I b·t· be I . edd to fire frequencies of h e aval a mor re~ogmze that, in day-to-day practicee t an once a decade D · Ih'ny was observed to quick post-fire regeneration of pala , farmers will often . espl!e t IS conclusion, it is tabl ~ the vegetation resources. choose the shon term perspective: e orage at the cost of long -term impoverishment of

b b~nch

Fire fr~uency has not changed in areas that have bee b change In managem ent policy with . the creation of the n _urned smc ~ the fifties, in spite of a to hold true for the stocking rate s applied . nauonal park m 1973. The same seems 188

Burning as a management tool Both fire frequency and overall grazing intensity deternnine the long-term balance among patches of high vegetation attractiv eness, degraded patches with a low forage availability, and all other situations in between thes e two extremes, including the und isturbed vegetations.

LITERATURE CITED

Bailey, A.W. & Anderson, M.l . 1980. Fire temperatures in grass, shrub and communities of central Alberta. Journal of Range Management 33(1 aspen forest ~argagli, A. 1991. Screen digitizing, geo ): 37-40. Landsat and normal aerial photogr metric corrections and monoplotting for SP0T, aphs on the ILWlS environment. Internal report, ITC , Ens che de, 29 pp. Cleef, A.M. 1981. The vegetatio n of the paramos of the Colomb ian Cordillera Oriental . Dis sertationes Botanicae 61. J. Cra mer, Vaduz, 320 pp. Chuvieco, A.E. & Congalton, R.G . 1988. Mapping and inventory of forest ftres from digital processing of TM data. Geocarto lnternational4: 41-53. Chuvieco, A.E. & Congalton, R.G . 1989 information systems to forest haza. Application of remote sensing and geographic rd mapping . Remote Sensing of Environment 29: 1~1~. Davis, F.W. & Burrows, D.A. 1994. Spatial simulation of fireregime in Mediterraneanclimate landscapes. Pp. 117-139 fire in Mediterranean·type ecosin: Moreno, J.M. & Oechel, W.C. (eds.), The role of ystems. Ecological Studies 107 , Springer-Verlag, Berlin. FAO. 199 1. Guidelines: land evaluation for grazing. FAO Soils Bulletin No. 58. FAO , Rom e. 158 pp. Green, D.G. 1989. Simulated effe ct of fire, dispersal, and spatial pattern on competition within fore st mos . Vegetatio 82: 139-153. Hofstede, R.G.M. 1995. aics Effects of burning and grazing on Colombian paramo ecosystem. PhD dissertation, University of Am sterdam, 199 pp. Horn, S.J. 1989. Postftre vegetatio n development in the Costa Rican pararnos. Madroiio 36(2): 93-1 14. Hom, S.P. 199 1. Fire history and ftre ecology in the Costa Rican paramos Nodvin, S.C. & Waldrop, T.A. (eds .), Fire and the environmen . Pp. 289-296 in: t: ecological and cultural perspectives. U.S. Dept. of Agri~ulture, Asheville NC. ILWIS 1.4 User's Manual. . ITC, Enschede. ILWlS 1.41. 1994. Supplem1993 ent to 1.4 User's Manual. ITC, Ens Janzen, D.H. 1973. Rate of rege neration after a tropical high elevchede. ation fire. Biotropica 5: 117- uz .: Kessel, S.R. 1979. Gradient mod eling: resource and ftre managem ent. Springer Series on Environmental Management 1, Spr inger-Verlag, New York. 432 pp. Kruger, F.J. 1984. Effects of fire on vegetation structure and dyn amics. Pp. 219-243 in: Booysen, P. de V. & Tain ton, N.M . (eds.), Ecological effects of ftre in South African ecosystems. Ecological Mannetje, L. 't & Ebersohn, Studies 48, Springer-Verlag, Berlin. J.P. 198 Agricultural University, Dept. of 9. Tropical grassland improvement. Wageningen Field Crops and Grassland Science; Wageningen, 39 pp.

extensiv~

189

Chapter 8 Minnich, R.A. 1983. Fire mosill'cs . · m southern Cal1'< · Sctence 219: I287-129l •Orma and northern Baja Noble, I.R. & Slatyer R 0 198.1 Th . · ' · · · e usc of vual attribut . · In P1ant communities sub ·ect . es to predtct al Patters~n, W.A. III & Backman, ~A.E~ol~~urr~nt dtstur~ance, ~egetatiosuccession 43: 5-21. .. . m: Huntley, B. & Webb T UI . Fue and ~sease history of forests. Pp . Science 7, Kluwer Acad~mi~ p ~~~), Vegetation History. Handbook of . Pels, B. & Verweii p A 1992 B .u IS ers, Dordrecht. · ·• · urnmg · · ecosystem: "'a transition mod 1 Pp and 2 grazm~ m a Colombian bunchgrass PMa!no: an Andean ecos ste e , . 43-263 m: Balslev, H. & Luteyn Ramsey, P.M. 1993. The pMam~ m un.der human influence. Academic Press • d . . of Ecuador· th · • an producttvl!y of tropicalvegetauon gras 1 ds . · e commumty ecology Wales, 274 pp. s an 10 the Andes. PhD dissertation, ' Salomons, J.B. 1989. Paleoecolo . 15-"215 in: Vander Ham!Y of;olc~lc soils in the Colombian Central '-"'u .. , ..... . Andean Ecosystems 3. J. ~ ., D~azl•.s. & Alvarez, .V.J. (eds.), Studies on Smub, A.P. 1981 G er, er m. .• . row!h and population d . . . . Venezuelan Andes. Smilhsonian Co . yn~mtcs of Espeletia (Compositae) of Smith, A.P. & Young T p 1987 T . ntnbunons to Botany 48: 1-45 . .• · · . · 1 · and Systeniatics 18: 137-158rop1cal atpme pant ecology. Annual Review of . Swetnam TW & B · • . . etancourt, J.L 1990 . wm· southwestern United States. Science 24tt~-~7~~~;; Oscillation relations in . . Iamson, G.B., Schatz, G.E., Avlarado, A., Redhe . 1986. Effects of repeated fues on tr . al ad, C.S., Starn, A. C. & Sterner R W 62-69. optc Paramo vegetation. Tropical Ecolo~y ; 7

CONCLUSIONS AND IMPLICATIONS FOR MANAGEMENT INCLUDING CONSERVATION OF PLANT DIVERSITY General rem·arks Potential of !he modelling approach Grazing management Burning management Plitnt sp:cics diversity The natural resource managers

There is no single optimal management strategy for the paramo. The management ~!Tategy depends on !he management objectives of the paramo area concerned. Objectives often include the conservation of biological diversity, the conservation of a stable hydrologic al system,. recreation, agricultural production, or a combination of several objectives. In the process of modelling described in the previous chapters, conclusions were drawn that have serious implications for management of the paramo grasslands. These implications are '' discussed in this final chapter. The potential of !he modelling approach followed is also briefly dealt wilh. A separate section is dedicated to the analysis and discussion of !he overall effects of management on plant diversity. As conservation is such an important manageme nt objective for the study area, diversity a~cts require special attention. The chapter concludes with some considerations concerning the people and institutions responsible for the management of the paramo natural resources.

9. 1

General remarks

A rough translation of the number of cow droppings into grazing intensity expressed in animai units per hectare (A.U ./ha) is possible. Eight experimental plots were monitored for three months to establish the turnov~r of marked cow droppings. For the stable plots, where the input of new cow droppings was close to the rate of disappearance, turnover was calculated at about one year (± 10%). Excretion of an average cow is about 10 cowdr./day. One cowdr./50m2 corresponds then to about 0.055 A.U./ha on a yearly basis. This field measure of grazing intensity is a practical indication for those involved in the management of paramo grasslands. It can be easily adapted for other regions by estimating the turnover time of cow droppings. Inspection of the vegetation map reveals !hat, of the total area occupied by zonal bunch grasslands and successional stages, 50% is notably affected by management impact. This means that the original dense bunch grassland structure has changed into other, more open vegetation structures. The initial research hypothesis, that the actual vegetation pattern of the paramo proper is to a large extent explained by variation in grazing and burning manageme nt, is therefore accepted. The figure of 50% affected area is high, especially considering the national park status of the study area. 190

191

··.

Chapter 9 The actual upper limit for livestock grazing here is about 4500 m, and for cultivation 3900

The tussock bases provide crucial information -on management impact on the grasslands. Diameter distribution, in particular, is a good indicator of the fragmentation as defined and described in Chapter 5. Fire-induced fragmentation of tussocks into isolated tillers is temporary in the absence of disturbance. In comb.ination wi~h grazing ~r under exclusive grazing influence, the fragmentation of . · grasses ts, respecuvely, sustamed or induced by the impacts of trampling and consumption, The tus~ock base becomes more an? more fragmented. This results in numerous tussocks with •. small dtameters. Further degradation results in lower densities and possibly in the · · disappearance of bunch grasses. ' ' ·· The height disoibution of stem rosettes also reflects management impacts. Under the influ~nc ·' of grazing and burning, the turnover or life span of stem rosettes decreases. This can be see~ ~rom the larger ~umbers ?f small individuals, whereas the tallest individuals are subject to . mcreased m:>nahty, ~r dtsappear. Patches of high stem roselle density of tall individuals deserve ~pec_tal att~?tton, as these have become scarce phenomena. They are maintained by low grazmg mtensllles (less than 0.16 A.U./ha) which keep the bunch grasses at a low cover. A de~se mat of ~ho~ _herbs and ~e effect of shading by the stem rosettes probably also ~ontnbut~ to mamtammg the domm~nce of the latter. Repeated burning also favours an mcreas_e ~~ ~tern rose tte cover, but at the same time strongly increases the mortality of the tallest mdtvtduals, and consequently the risk of local extinction. On the basis of the tussock and stem rosette population structures, it can be easily evaluated which are the more intact vegetations. On grazed slope~, the development of terracettes and the increase in bare soil, possiblyenhanced by bu:nmg, probably have important consequences for the water retention capacity. Hofstede ~ Se~mk (I ~95) concluded that the water retention capacity of disturbed vegetations · decre~ses m drier_penods. The effect of soil compaction on run-off has not yet been fully exammed. Assummg that the bunch-grass layer is the best water regulator compared with the o~her plant gr?ups (~ofstede, I995), the model can be used to predict its development under · dtfferent grazmg regtmes. Some conclusions in this respect are presented in Section 9.3. The ~egradation_ of cushion bogs due to trampling impact was also described (Chapter 3). Grazmg results m a gradual transformation of the cushion bogs into mixed grasslands with some isolated remnants of ~ushion plants, or into open shrubland vegetations. Impacts on hydr~logy need to be quant~fied, but what was qualitatively observed was that the 'sponge' funcllon of degradat_ed cushton bogs is largely lost: water is no longer retained up to ground surface level. Espectally on sl?pes.' it is worthwhile to consider fencing them off, to prevent cattle from entenng. Water pomts m gently sloping terrain such as the top of lateral moraines are associated with less deterioration.

Conclusions and implications for ma11agement

,. 9.2

Potential of the modelling approach

The spatial and temporal resolution of aerial photographs proved to be ~ighly suitable ~or the reconstruction of recent fire history. Limitations and potentials of satelltte remote sensmg for tht: study of fire distribution and frequency were discussed by Justice et at. (1993). In this ~tudy, use of satellite images was even impossible due to considerable cloud coverage. :ire scar analysis as done for Espeleria regrowth to determine the number of years <:lapsed smce burning is complementary to the applied remote sensing approach. According to a recent report by a group of fire specialists, the potential of combining fire scar analysis with remote sensi ng information has yet to be fully examined (Malingreau, 1993). The vegetation and fire history maps in combination with the simulation model provide spatially referenced data on fuel structure, and information on fuel loading over time and orr burning practices. These· are the basic data requirements for a general fire infonnation system, according to the same group report (Malingreau, I.e.). It was generally assumed that a model of vegetation development in terms of cover per stratum or plant group cou ld not serve as a basis for evaluating grazing capacity. Therefore, most authors choose an approach in which biomass is modelled instead of plant cover (e.g. Van Duivenboden, 1993). Biomass estimation is time-consuming , especially when different plant groups of varying forage quality are taken into account. By means of the present study, it is shown that the weighted sum of plant cover can also provide a measure of forage ·availability which is correlated to grazing intensity. The analysis of the composition of cattle diet is the key to establish the response of grazing behaviour to variation in vegetation structure in terms of cover of the functional plant groups. Especially in spatially heteroaeneous grasslands with large forage quality differences between plant groups, it seems more ;ppropriate to model plant cover instead of biomass. For the description of veget~tion response to burning and grazing, the proper identification of functional plant groups wuh a similar reaction to variation in management is crucial.

A comparison between the outcomes of the developed spatial and temporal models is facilitated by the variables they have in common. These are: cover (mnges) of functional plant groups and bare soil percentage vegetation attractiveness as a measure of forage availability grazing intensity fire occurrence (nominal variable) the number of years elapsed since burning ('fire age') · In the map of cattle disoibution in Figure 7.5 (page 168), those units with a p~edicted grazing intensity of more than 0.16 A.U./ha are highlighted by their orange and reddish colours. The outcomes of the simulation model indicate that these units are overutilized with respect to the bunch-grass layer, which is supported by visual impressions in the field. If the _acc~racy _of prediction of grazing intensity is taken into account, it might well be that the grazmg mt~nstt~ belongs in reality to a lower or higher class in the map. Therefore, to be on the safe stde,_It is recommended that the possible degradation processes occurring in yellow-coloured umts (0. I0 - 0. I25 A. U./ha) be carefully examined as well.

192 193

Chapter 9 The map of fire frequency shows that the overall fire frequency corresponds well with simulation experin:ents. At higher grazing intensities, fire occurrence is inhibited by a fuel load. Indeed, 111 the map of grazing intensity it can be seen that areas with a intensity exceeding 0.18 A.U./ha are not burned in practice. This corresponds well with outcomes o~ simulation e~pcriment s. The areas with the highest fire hazard, especially lateral morames and glaciated lava fields, have a grazing intensity ranging from 0.075 0.175 A.U./ha. The areas that are both attractive to grazers and have a sufficient fuel load prone to burning. The minimum fuel load required for fire occurrence is associated mainly with dense grasslands, and occasionally with medium-dense bunch grasslands with a total bunch of around 50%. This explains the absence of frre in the northwestern section of the study are From the map of vegetation attractiveness (Figure 7.3), it can be derived that burnable areas have a .VA valu~ ranging from 100 to 200. Values below 100 correspond to superparamo vegeta_uon, cushwn bogs, and recently burned vegetations with a low fuel load. Vegetat!o~ attractiveness values above 200 belong to mixed grasslands and short matted vegetations a low fuel load.

Strateg~es ~f defe~ grazing can also be evaluated using the simulation model. If th; vegetanon IS able to recover during a period of zero grazing, the fire路 return interval is expected.to be shorter. ~he surroundings of Laguna de Ia Leona are characterized by high frre frequencies, where graZing is deferred. . 'l The basis foi evaluating different potential management strategies is established with the development o_f both types of model. However, the potential of this approach can be further explore~. Spauo-temporal representations of vegetation development could be elaborated if appropnate software becomes available to enable coupling of the simulation model and GIS. In Clfapter I, hov.:ever, it is argued that concepts and suitable models have not been fully' . : d~velop~d, and th1s fo~s a major constraint to spatia-temporal modelling. In the present 路 d1sserta11on, an attempt 1s made to overcome some of these problems related to concepts and partial models. Within. a gTazing area of one owner, a series of vegetation patches was distinguished as

~cologt~all y homogeneous units with the same managemen t history. Grazing implies spatial

mteracuon amo~g. these patches: the degree of utilization of one patch is influenced by all other patches wttht~ the same grazing area. The distribution of patches also depends on frre occurrence. Vegetatton development could be simulated for each patch. Based on the weighted 路 sum of the cover of fu nctional plant groups, a new measure of forage availability could be calculat.ed .afte~ one year of vegetation development under a cenain grazing intensity. A new ~attle dtstnbut~on could then be calculated within the spatial model. This could give insight tnto the potential development of the paramo landscape over several decades. -

Conclusions and impUcatiOitS for management

9.3

Grazing management

Low grazing in tensities are important to maintain the diversity or heterogeneity. of the landscape in tenns of patchiness. Exclusion of livestock favours the development of a uniform 'blanket' of dense tussock grasslands, relieved by isolated patches of cushion bogs and shrublands. Patchiness that results from the spatial heterogenei ty of grazing intensity is reponed to increase total species richness by increasing between-habitat (beta) diversity (Bakker et a/., 1983; Milchunas et a/., 1988). Moderate grazing intensities are known to increase community level (alpha) diversity of plants (Naveh & Whittaker, 1979; West, 1993). In the puna grasslands of Peru, plant species diversity was indeed highest at intermediate grazing intensities (Wilcox er a/., 1987). There are indications that the species richness of the paramo grasslands shows a similar trend. This is demonstrated in Section 9.5. Heterogeneity of the landscape in th~ form of patchiness also maintains the diversity of fauna by creating refuges of reduced competition or predation. Richly variable environments can support a greater number of animal species than more homogeneous ones (McLaughlin & Roughgarden, 1993). In order to establish the grazing capacity of paramo grasslands, several considerations have to be taken into account. Decisive are the objective(s) that should be met: Is the principal objective optimal livestock production? This means maximal production without depleting the forage resources in the long term. A relatively high grazing intensity is then permitted at the cost of degradation of the bunch-grass layer. Grazing intensities up to 0.8 A.U./ha can be supponed, and on flat pans local grazing intensity may even exceed I A.U./ha. Is optimal protection of the hydrological system the main goal? Maintenance of the bunch-grass layer then plays an important role. Fire should be suppressed as much as possible. Low grazing iniensities can be supported by the bunch grasses without irreversible damage. The adverse impact of trampling in the form of soil compaction should be further investigated, as it affects water retention capacity. Grazing intensities below 0.16 A.U./ha allow maintenance of the bunch grasses in the absence of burning. Zero grazing is expected to maintain the protective bunch-grass layer best. Should landscape patchiness be favoured to maintain habitat diversity and related species diversity of both flora and fauna? The mosaic structure of the vegetation of different grazing areas can be evaluated and compared with the actual stocking densities in order to choose an appropriate grazing strategy. ijabitat suitability of the patches can be analyzed for different species. 路 Should the present situation be maintained? The actual average grazing intensity in utilized vegetation patches is 0.37 A.U./ha. According.to outcomes of the simulation model, this situation is not stable but tends towards a gradual degradation and the subsequent disappearance of the bunch grasses. If the present situation is to be maintained, stocking densities of the grazing areas (except the one managed by INDERENA) need to be reduced.

194

195

Chapter 9 . Is the objective pure nature conservation? Th. . I' h uon of the ecosystem in a state as close as ossibl~ t0 ~s lm~ l_es t ~ preser;a liS ongmal, nawral state: dense p h grasslands and cushion bog s. In sue · a landscape B 1 , os aun/S IS an undesired species and zero grazing is the best strategy. . Very few figures for carrying capacity are known fr e. On~ wa~ presented the_ p~a~o. grasslands of Ecuador: 0. 14 A. U./ha (MAom the lueratur1977; cued 111 Hess, 1 TOM, G/ORS . . fo figure above the to Th1s IS smular at the condition of maintenance of the bunch grasses of 0 16 A U /h Th r grazmg capacuy ds of Lo N dos National Park ar(!. · · a. e paramo . grasslan · s eva _ expected to be slightly more humid in com Ec~ador. INDERENA (1985j of thos~ wu? p::: acit ca g carryin for figures also reported wnh the farmers. According ":s lntervle to their information, the carrying ca:acit:~f b ~n A.U./ha, of Ulc:Jze,ni/1.,> 0.39 IS d grasslan unc f d /h U A 78 0 meadow orbiculata fi pastures I 6 A.U.lh a. Th· . . a, an o sown . A U /h ·~e lgure of 0.39 . . . a for bunch grasslands indeed corres d . ed actual average grazing intensity of utilized patches of 0.37 AU ~oni s _wuh the calculat I S clear that the management goal of the t .. a . k livestoc l maxima simply is farmers Although this strategy can be perfectly understood in view of their struggle r.o;rsu bu~non. ly leads to the degradau.on of it invariab SIStence . cover of the bunch gra . Thi h . ' d t e protecnve 5 s a verse effect zs clearly noticeable, especially s es. close to the fincas.

· 9.4

Buming management

. Fire favours the occurrence of discrete homo ene sharp boundaries. The size of ..the burned areas has implications f~r the !sul~~s patches wuh e. Another report by structur pe landsca g . a group of fire researchers evaluated th level (Binkley, 1993). em ecosyst at fires ~f ts edlmpac I may fires small that ed It was conclud .. ea to mosaics of patches of dI·f~erent composmon · and structure, whereas large fires tend to homo enize Ia is it area, study the regards As . s ndscape . h hg ~oncluded that fire increases patchiness lly at especia , grazmg b~ ed mamtam ~~~e;c7) (Cha fincas the mtermec!iate distances from IS generally followed by graz~ng i~ the burned patch is accessible a~d no; t;,e occurrence s tend to bum those Farmer remote. b ed ~ are terram units that results also show y cattle for grazing. Close to the fincas, the fuel load is too small to allow burnin:.re err As the cattle range freely they are attracted b d forage Y th~ burned patches of increase . availability. If the burned p'atches are sma 11, oven!ra . . zm!! occurs · Act ua1grazmg y is . mtensn ~ ~ h apparentIy difficult 10 regulate Meanwh .1 t e _tussocks are especially vulnerable to tmmplincro impact during the first four y.ears . I eb, ummg f The risk of smce permanent ragmentation or · comp Iete disappearance of the tussock s 1.s t hen .mcrease d (Chapter 5). . . mcrease of bare sm·1 su rface IS . percemaoe The d (> 10%) d · " . v unng ~he penod I .5 to 5 years .· smce burning, depending on the regenerat' influence. c~pacny of the vegetation and·grazing Erosion is thus favoured by fire • and SOll_ol nl osses f are expected 1 . burning. ter G . a o occur J_ust razmg can mcrease erosional loss, especiall on slo remains surface sOJI the_ of part as pes, Y ). r bare due to trampling impact (Chapte 3

196

Conclt~~iOilS

and implications for manageme11r

to be relative ly low and Erodibility of volcanic ~sh soils in natural conditions was reported have not as yet been losses soil to decrease with elevation (!meson & Vis, 19B2), but actual estimated for paramo .:cosystems. erosion can be the result Surface drying of andosols decreases cohesion, and severe water slope instability seems a , erosion surface with ison compar In (Warkentin & MaeLia, 1980). slumps, and soil creep. es, f!lajor problem. Mass movement can occur in the form of landslid n than that of the cohesio lower a in results surface When exposed, drying of soil at the erracing can micro-t of form the in creep Soil I.e.). Maeda, & ntin undried subsoil (Wark.: practices, grazing and burning by soil therefore be panly attributed to the increase in bare of number and size the and angle, slope ons, conditi apan from the influence of climatic grazing animals. of bare soil, there are other In addition to the effects on the bunch-grass layer and the amount by volatilization can sulphur and adverse impacts of fire on the ecosystem. Losses of nitrogen (Hofstede, 1995). so'il the in s nutrient of amount ble be important in relation to the extracta lization: especially immobi to due ion vegetat the to able unavail become can Nutrients ne-like constituents and Alphosphorus is easily fixated by phosphate adsorption to allopha In addition, substantial 1995). Sevink, & e Hofsted 1985; 1980; humus complexes (Wada, tive transport or convec by sites quantities of non-volatile materia l can be lost to burned losses, the nutrient these with ison compar In s. particle zed surface wash-off of ash and carboni Most of 1993). eta/., t (Menau magnitude of nutrient loss due to leaching is generally small t sites, recipien to input nutrient extra an nts represe the material lost to burned sites on slopes . bottoms valley and terrain the mostly concave parts of causes local degradation of the Summarizing, burning on slopes in combination with grazing development of terracettes bunch-grass layer, increase in the percentage of bare soil surface, t fires probably result Frequen s. nutrient of ility of compacted soil, and decreased local availab in Section 9.5: It has ed discuss and d analyze is which y, diversit in a decrease in plant species , impossible, practice in is, cy to be taken into account that control on the actual frequen to allow full tly frequen too applied is fire , practice In whether the use of fire is legal or not. y. bunch-grass recover a high burning frequency was In the deferred grazing system close to Laguna de Ia Leona, d, although the benefit is observed. Forage reserves for drier periods are thus increase for a comparatively short burned are areas ve Extensi relatively low in terms of grazed yield. cattle were provided with the or lower, were s densitie stocking If . scarcity period of forage would not be necessary. damage this , supplementary feed during a period of forage scarcity hile to change this: fire is As the use of fire is based largely on tradition, it may be worthw burn regularly. Generally, others areas, grazing their not necessary. Some farmers do not burn eness for grazing was attractiv ion Vegetat . seasons drier in except le, enough forage is availab decade (Chapter 8). a once than observed to decrease in cases of burning frequencies of more s. An example was the With the suppression of fire, the risk of the rare large frre increase Rica (Chaverri et al., Costa of 6 Chirrip extensive paramo fire of 1976 in the Macizo de animals can be used grazing pe, landsca the of nization 1976). In order to avoid the homoge 1993). y, to maintain patchiness and thus to reduce fire hazard (Binkle 197

,, Chapter 9

9.5

Plant species diversity

Conclusions and implications for management . and .mtraduced weeds and of other nati vc VPT along the floristic gradient Numbers of nauve • d' fF 9I TWlNSPAN an: graphically prcsentc!I in the stack:d. bc.r l3gram o tgure . . It is that total species diversity is similar for flonsuc types . . h.orbic:ulata var. Rumex acerose a flon~uc type I (t c poa spp. - ThLochemil/artion of Introduced VPT increases gradually along spectes numbers seem to drop. e propo . · levation (Chapter the floristic gradient according to increasing human tmpact and de~~a~mfo~owed by a slight 3) The native weeds show a slight increase along the sa_me gra b~en : 'th the reduced de~rease This can be clearly observed in Figure 9.I, m com t~atton WI numbers. of other native VPT at the most disturbed end of the gradtent.

~brained b

concl~ded

The overall effects of management strategies on conserYation aspects also include the of management on plant diversity. This is analyzed in the present section. As a plant species diversity, the number of vascular plant taxa (VPT) per releve was Most taxa are identified to species level. In the footer of the vegetation table (Table toral number of VPT is indicated for each releve. · For each floristic type, the average species richness was calculated according to this · A distinction was made between introduced and native paramo plant species. expen knowledge and the listing of tlie species in the 'Flora Europaea' (Tutin er 1980) guided the distinction of the introduced species. Besides introduced grass agricultural weeds from temperate latitudes are also frequently encountered in grazed vegetation, such as Rumex acetosella, Taraxacum cf. offtcinale, and Cerastium Twenty introduced taxa were recorded. Another group of species is that of the native weeds: those plant taxa that occur vegetation under natural conditions at a low cover and abundance, but increase as a •._.,~~·· •·• a high rate or' a high intensity of disturbance. These native.-weeds (including the Andean crop Solanum tuberosum) are sometimes more abundant in lower zones, as is -·-.-.._.,. for Acaena ovalifolia and Veronica serpyllifiJlia. Recognition of the native weeds was out independently of this study by the paramo expert Dr. A.M. Cleef. The plant Appendix A includes a column that indicates to which of the three groups a determined belongs.

a,/ograH~s~t~lr) ~~

Q; D

E ::J c

The average numbers of both introduced taxa and native weeds were calculated per type and expressed in absolute numbers and proportions relative to the total number of Results are presented in Table 9.1.

Floristic type

c D E

F G H I

Table 9.1

Number of reieves 16 21 35 26 23 13 10 19

Total number ofVPT (± s.e.)

Number of Proportion Number of introduced introduced native weeds weeds weeds (± s.e.) (%) (± s.c.)

32.8 ± 7.3 31.0 ± 8.5 37.7 ± 5.5 37.5 ± 7.7 29.8 ± 8.1 39.0 ± 7.3 30.3 ± 7.6 19.9 ± 6.6

0.5 ± 1.1 0.2 ± 0.4 1.4 ± 0.8 1.3 ± 1.2 1.5 ± 0.6 3.6 ± 1.5 5.3 ± 1.6 5.9 ± 2.3

1.5 0.8 3.7 3.5 5.0 9.3 17.5 29.6

co Q;

Proportion native weeds

Floristic types

(%)

5.6 ± 2.8 6.5 ± 3.0 10.8 ± 2.4 11.7 ± 2.9 10.5 ± 2.9 14.4 ± 1.9 10.5 ± 2.8 6.3 ± 1.7

Numbers of introduced and native weeds compared with total numbers of VPT for each floristic type of the grass paramo belt. VPT =Vascular plant taxa.

~ Introduced taxa ~ Native weeds

Figure 9.1

Other native taxa

Species diversity along the floristic gradient obtained by TWIN~P~ ~l~ter :ruysis. The average numbers of VPT per floristic_ type are plotted; dtsungutshing tween native paramo species and introduced spectes.

H stan (1994) described the following theory on the invasion of exotic species. E~osyst~ms

w~h low productivity and a high degree of endemism ~e easi~ ~~e~ ~~ :~~~~:e;t:~ F nate!

the same conditions that allow the evo1unon an s 0 .. ~ aly, of the invading species under natural condinons. spectes so prevent the do~;nance ....... 19H

199

Chaprer 9 Changes in the disturbanc e regime, howeve r, can resu lt in major changes in composition and a reduction in species diversity. For the strongly disturbed increase in introduced species (up to 30%), !he decrease in native species, and the in plant species diversity are strjking. The sin1atio ns of hig h dist urbance intensity are patches under hig h-intens ity grazing at sha n distances from fincas or water points, gentle sloping tops of lateral moraines or at valley bottom s, and regeneration stages potato cultivation with or wit hout sown grasses. -

Conclus iolls alld implication s for

•

Effects of fire Variation in plant diversity along a gradient of years elap sed since burning was sut,seQi analyzed. The numbers of uei VPT of medium-dense to den se bunch grasslands do not sign ific antly between the flor istic types C, D, and E (Table 9.1 ). For these floristic type numbers of VPT per releve s, and of moving averages are plotted in Figure 9.2. From !he time series data of Figure 9.2, it can be deri ved !hat it takes approximately half year before the total number of species richness of the studied VPT exceeds 38.5 , which represents the average level paramo bunch grasslands. Thi s means that after half a year pos t-fire regeneration, most vascular species are again obs ervable in the vegetation. During the recovery phase of bunch-grass cover (Chapter 5), species diversi ty is vari eight years since burning. At able approximately the same time since burning, variation in spe diversity among sites is probab ci~s ly related to grazing influence. With time, the overall is an increase in the averag e number of VPT until five years after burning. The lncr availability of light and nutrien ea!ied'•~:1 ts following a fire (Hofstede, 1995) allows !he germination growth of a number of shan-g and rowing species. Shon grasses, ground rosettes, and other sho forb species colonize the ope n spaces between the tussock n bases. Five to eight years pos species numbers seem to dro t-fire, p slightly. When the bunch gra sses attain a full cover, som these species are probably e of outshaded. Subseq uently, the average number of VPT incr slowly but steadily. This is eases probably related to the re-e stablishment of a number of sensitive species. According fireto Tab le 3.5 (Chapter 3), a num ber of tall forbs and procumben forbs of the ground rosette laye t r are positively related to incr easing number of years elap since burning or the absenc sed e of fire. Alth ough the above tendenc y is not clearly articulated, it can be concluded that it take than I0 years before the ave s more rage number of VPT has stab ilized at pre-fire levels. Thi s that fire frequency may infl implies uence the num ber of VPT pres ent in the vegetation . Average numbers of VPT are pre sented for different fire frequencies in Table 9.2 , both all structural. vegetation type for s together and for dense and medium-dense bunch grassla (types d and h). If all structur nds al vegetation types are conside red, a weak tendency is see n decreasing -species diversity of with more frequent fires. The se differences are however statisticall y significant. For not only structural vegetation type s d and h only, the decreas species diversity is more pro e in nounced wit h a significant diff erence between non-burned rele and releves burned twice (P ves = (J.Ol ).

(ij

-s0

~ 30

-- ~

•

0,__ Q)

.D.

E ;j

Figure 9.2

running ave rage

lands (floristic types C. D. and Species richness of pararno bun E) along a ch gradient of the time elapsed s~nc ~ fi age) Species richness is expressed as the total number of VPT of smge bu7m~( If~ of fi~e-data running averages. le re ev an

Floristic type

Vegetation structure types

C, D,E

d, h, o, m, e

Number of fire events in± 30 yr

·o

d, h

Number of re!eves

Number of VPT {± s.e.)

16 45 8

38.6 ± 6.9 36.7± 6.1 35.5 ± 4.3

21 4

38-S ± 3.1 36.4 ± 6.2 32.0 ± 2.2

l 2

C, D, E

Fire age (yr) ,., single releve data

Table 9.2

200

/1Ulnagemenr

2 d as the num ber of fire events Species diversity and fi~ .frequen observed in 30 years. Only flonsuc ~~cy expre~~ E were analyzed (in the case. of.fire C, D.. ) Th number of VPT per releve n s year elapsed Stnc occurrence. -;>V. IS given e burrung . e _ for different fire frequenaes.

201

Chapter 9 In situations of low productivity, low spec ies diversity is found in relation to the life when fire •rcctuet cv 1 cycle of most plants (Huston, 1994). If bu 1 irreversibly in relation nch-grass cover. to the deple tion of tus sock=base reserves by it is ex pe ct¢ that frequent fire s the provide a less divers more uniform vegetation structure and e habitat to other org related rruc:ro-clitioai.O anisms. Effects of grazi11g The same kind of an alysis was carried ou t for species diversity response of grass an along a grazing d forb diversity to increased disturbanc unimodal, or decreasi e can be ng (Huston, 1979). W oody plant diversity increasing disturbanc is expected to e in the paramo, wh ich is near the physi in relation to elevatio ological lim it !o plant n. Figure 9.3 shows the vu.riation in the tot grazing gradient in the al number of VPT grass paramo. Floris tic types H and I are as the impacts of po excluded from this tato cultivation and the sowing of grasse variation in species s are additional numbers. From Figure 9.3, it is concluded that the va riation in VPT numb fitting did not lead ers is high. Therefore, to satisfactory result s. However, a kind visually. In compari of optimum can be son with zero grazin g, species numbers moderate grazing an tend d to decrease under higher grazing intensit to increase slightly native VPT appears ies. The maximum to be about 0.16 f..U./ha. The dominant tussoc k grasses exclude a numb moderately grazed sit uations, additional sho er of species in· the absence of grazing.' In rt herb species benefit increased availability from the open space of light, moisture, an and d nutrients. In more Lachemil/a orbiculata intensively grazed ~m• and several short gra au'"""• ,,-. _,. ss species become do minant. For the paramo ecosys tem, the resulting ten dency of an optimum grazing intensities is species diversity at mo not clearly articulated derate . Ac diversity, low-produc tive grasslands, grazin cording to Milchunas er a/. ( 1988), in hig hg is expected to pro diversity than in pro duce a smaller increa ductive sites. se in The paramo grasslan ds of the study are a are average species richn ess of dense to mediu characterized by a high plant diversity. The m-dense bunch grassl higher than the few ands of 38.5 ± 3.1 VP figures reported for other tropical mountai T is slopes of the Rulz n ecosystems. For the volcano, 20 to 32 spe exter cies were recorded 1991), and for the hig in seven releves (Sala nal h mountain vegetation manca, of Mexico, the zacato recorded in 16 relev es at the volcano Po nal, 4 to 16 species pocatepetl (Almeida were eta/., 1994). It is concluded tha t recognition of fun ctional plant groups important insight int in paramo grasslands o how the vegetation provide reacts to disturbance, analysis of the effec ts of management on and that a more de tailed plant diversity would analysis of species ric therefore benefit fro hness per functional m the plant group.

202

Conclusions and imp licario11S for manage ment

60 )(

)(

Ill

)(

X X X X X

)(

a )(

.... (I)

..0

E 30 ::J

)(

X XX

)(

-~.,J: · o

. ~X~ X

)(

)(~

)(

X )(

(!)

)(

)( )(

)(

X X

~X )(

Ol Ill

)( )(

)(

.... 20 (!)

)(

ol_------r-------~2---

---~3------~4-------ss

Figure 9.3

9.6

~-----es

ln(Grazing intensit y} . Number of VPT along a g~m~

d. L Ploned releve s correspond to floristic types

en . xpressed as ln(cow s.c.o. E, F. and G. Grazmg mtgraenst_tty dr./50m2). ts e

The natural resource managers

: . x ~ted to bene •s natur The management o~ fit from the the p~t has beeal resources lS e pe · al n stated already ~a nagement t. the opum rmaazi ng and de pe nds snate . .. ed or will conunue m i:t o pra burninngt . f more than a ,ew management strategJffi accou ees . .th rnm years. A schemewt as u ary of the long-tenn e strategies . resented in Tab19 on the paramo ecosystem IS e ·3· P . . of possible ma For the successful . it is of eruct·a1 nagement im ecos stem, . ortance that the plementau_on . olved in the manageme strategieS, different parues mv nt of the paramo tmp y agree on the strategy to be followed.

considgyer~~~npsa:~!!y~~;~~~::~:~t:re:~~ible

~n (ar~-)s~~co~~:.t~:~:-i:;:~

t~take

c~:~~ ~~:~~~~a:agement

203

Chap1er 9 Management

Zero grazing, no burning

Bunch grasses

Espeletia populations

Conclusions and implications for management Actual burning frequency -

Short herbs Protective cover of the soil surface

As long as farmers continue living in the national park, their participation in the development and implementation of possible new management strategies is an important condition. farmers' interests have been neglected too long. Sofar, it has been attempted to prevent negative management impacts only by discouraging fire practices and by limiting credit . facilities for investments. This has not had the desired effect: fire frequency has not ... :".diminished since the creation of the national park (Chapter 7). A more promising approach -" could be the design and implementation of a management plan that considers farmers' · interests, provides them with a better perspective, and at the same time meets consel"{ation purposes. Training and the education of the younger generation should be mentioned in this context as instruments to raise awareness concerning conservation issues. A better education would give the younger people other perspectives for the future, so that they would not necessarily depend on agricultural activities within the park boundaries.

Plant diversity - introduced species - native species

Forage availability

+I-

Neither has the potential role of the land owners been fully recognized. The land owners who usually live in the inter· Andean valleys are affected by the consequences of the degradation of the paramo grasslands in the form of irregular supply of drinking water and water scarcity in drier periods. They can play a speciflc role in management decisions concerning herd size and the use of fire.

+ +

Landscape heterogeneity

Table 9.3

Actual grazing intensity

+(-)

Overall long-term effects of mana erne

++ .

ccos~stc:n. Actual burning imcnsfty: 2n::~~t~gJcs on different aspects of the paramo grazmg intensity: 0.37 A. U/lr . r h . . .. 1. 00 Y~ for the lateral moraines; actual , . + :positive effect or incrca~/: .'gnetg;agt~mgf'mtensJty: < 0.16 A.U./ha; . · ·

1ve e .ect or decrease.

In the case of Los Nevados National p k I . .place, the fanners or caretakers andarth~ le ;:~ural resource manag.ers include, in the first guards who arc Involved in the daily, ~ management of the natural resources At institutions play a role as they c· .. tlanot er level, the absentee owners and a number of · . an In uence manage ment dCCJSions • . by means of rules, regulations, laws, and subsidies. ... If t~e aim is to lower stocking densities co requtred. This can include incentives to dec~e 7e~s~~IOn for bo~h farmers and owners is managed or owned by them for conse . ase er SIZe or set aside part of the grazing area fb ti rvauon, or a combin 'd . a on o oth. (Eco)tourism could provt e an extra source of income if th ~ · sueh as accommodation provtded servICes 'armers e · · tounst gwdes or mules £,or hl're (Verwe1J.. era/ 1995) K · • ' . . eepmg a few mules could provide ., them Wtth a better income than is the c 10 compensate the farmers for moving t~s~a~~prese~t. Another option is to buy the land and very costly solution. Furthermore th . outside the park boundaries. This.. would be a . · · • e Importance of ·.low grazmg · tntensutes for the . . mamtenance of landscape heterogeneity a d n specieS dtversHy should be considered.

Interinstitutional and interdepartmental coordination of possible actions in relation to management is necessary. Due to the multiple purposes that should be served, it is obvious that many different institutions are involved: besides the National Parks Service, which is passing into the control of the newly created Colombian Ministry of Environment, and the Corporaciones Regionales, other institutions are responsible for, among other things, the provision of drinking water (Empresas Ptlblicas), the provision of hydroelectric power, the _ development of tourism, and the possible constructjon of roads. Many paramo areas are intersected by administrative boundaries. The area of Los Nevados falls under the jurisdiction of four different Corporaciones Regionales, one for each department. These Corporaciones are responsible for the management of natural resources at departmental level, which task seems· to partly overlap that of the National Parks Service. Non-governmental organizations are important in relation to nature conservation and ecotourism. A framework for inierinstitutional action is indispensable for the successful implementation of any possible management plan.

LITERATURE CITED Almeida, L., Cleef, A.M., Herrera, A., Velazquez, A & Luna, E. 1994. El zacatonal alpino del Volcan Popocatepetl, Mexico, y su posici6n en las montaiias ttopicales de America. Phytocoenologia 22(3): 391-436. Bakker, J.P., DeLeeuw, J. & VanWieren, S.E. 1983. Micro-patterns in grassland vegetation created and sustained by sheep-grazing. Vegetatio 55: 153-161.

204 205

Chapter 9 Binkley, D. (rapporteur) 1993. Gro up report: impacts offires on ecos ystems. Pp. Crutzen, P.J. & Goldammer, J.G. (eds.), Fire in the environment: the atmospheric, and climatic importa nce of vegetation fires. Dahlem Wor kshop Environmental Sciences Reports , Environmental Sciences Researc h Report John Wiley, Chichester. Chaverri, A., Vaughan, C. & Pov eda, L.J. 1976. lnfonne de Ia gira efectuada al Chirrip6 a ralz del fuego ocurrido en marzo de 1976. Revista de Cos ta Rica 279. Hess, C.G. 1990. "Moving up • moving down": agro-pastoral land-use patterns Ecuadorian paramos. Mountain Research and Development 10(4 Hofstede, R.G.M. 1995. Effects ): 333-342. of burning and grazing on a Col ombian paramo PhD dissertation, University of Amsterdam, 199 pp. Hofstede, R.G.M. & Sevink, J. 1995 . Water and nutrient storage and input:output burned, grazed and undisturbed paramo grasslands. Pp. 121-147 in: Hofstede, Effects of burning and grazing on a Colombian paramo ecosyste m. PhD russentatic University of Amsterdam. Huston, M.A. 1979. A general hypothesis of species diversity. Am erican Naturalist 11 101. Hus ton, M.A. 1994. Biological diversity, the ct>existence of spec ies on changing lan•dscape Cambridge University Press, Cam bridge, 681 pp. '1meson, A. C. & Vis, M. 1982. Factors influencing the erodibility of soils in natural and natural ecosystems at different altitudes in the ·eentral Cordille ra of Colombia. Geornorph. N.F. 44: 91-105. INDERENA.. l985. Plan de rnan ·· · ejo Parque Nacional Natural los Nevados. Ministerio Agricultura, Bogota. Justice, C.O., Malingreau, J.-P . & Setzer, A.W. 1993. Satellite remote sensing of fires: potentials and limitations. Pp. 77-8 8 in: Crutzen, P.J. & Goldammer , J.G. (eds.), Fif1: _ in the environment: the ecologic al, atmospheric, and climatic imp ortance of vegetation fires. Dahlem Workshop Reports , Environmental Sciences Reports , Environmental Sciences Research Report ES 13. John Wiley, Chichester. MAG/ORSTOM. · 1977. Mapas de distribuci6n de Ia poblaci6n pecuaria en el Ecuador. Compiled by Cevallos, E. & Port ais, M., Quito. Malingreau ,. J.-C. (rapporteur) 199 3. Group report: quantification of fire characteristics from local to global scales. Pp. 329-343 in: Crutzen, P.J. & Goldammer, J.G. (eds.), Fire ~·. in the environment: the ecologic al, atmospheric, and climatic imp ortance of vegetation fires . Dahlem Workshop Reports , .Environmental Sciences Reports , Environmental ,. Sciences Research Report ES 13. John Wiley, Chichester. McLaughlin, J.F. & Roughgarde n, J. 1993. Species interactions in space. Pp. 89-98 in: Ricklefs, R.E. & Schluter, D. (eds.), Species diversity in ecol ogical communities: historical and geographical pers pectives. The University of Chic ago Press, Chicago. Menaut, J.-C., Abbadie, L. & Vito usek, P.M. 1993. Nutrient and orga nic matter dynamics in tropical ecosystems. Pp. 215-231 in: Crutzen, P.J. & Goldammer, J.G. (eds.), Fire in the environment: the ecological, atmospheric, and climatic importa nce of vegetation fires. Dahlem Workshop Reports , Environmental Sciences Reports , Environmental Sciences Research Report ES 13. John Wiley, Chichester. Milchunas, D.G., Sala , O.E. & Lauenroth, W.K. 1988. A general ized model of the effects of grazing by large herbivores on grassland community structure. American Naturalist 132: 87-106. 206

• Conclusions and implicari<ms for management Naveh z. & Whittaker, R.H. 1979. Structural and 11oristic uivc rsity of _shru , woodlands in northern lsmd and othc r ~kd iterr:~nean arc;t~. V:egetau ~lands and o 41. ~7 1 · 1~. Salamanca, S. 1991. llu: vegetatio n of the par;uno ami its dynamiC s m :he volc~mc ~asst~ Ruiz . Tolima (Cordillera Cen tral, Colombia). PhD dtssenauon , Umversuy o Amsterdam, 122 pp. T.G . et a/. · 1964 -198 u · · 0. Flora Europaea, ·vol. 1-5. Cam Tutin, bndge mverstty Press, Cambridge. Van Dui venboden, N. 1993. Grazing · · ·d · · as . a case study in the north-weste a tool for rangeland managc~ent m semlart ~;:~ rn coastal zone of Egypt. Agn cult ure, Ecosyste Environment 43: 309-324. v .. p A Kok . K & Budde P.E. 1995. Aspecto s de la tmnsfonnac 16n del ar erweiJ~I ho~bre. .Van der H;mmen , T. & Dos Santos, J. (eds.), Stud P amo JX>r ies on Troptcal Andean Ecosystems, Volume 5. J. Cramer, ~erlin. Wada, K. 1980. Mineralogical char . . . acteristics of And1sols. Pp. 87-107 m. Theng, B.K.G.~ed.), Soils with variable charge. Offs et Publications, Palmcrston N~rth, New Zealan · Wada, K. 1985. The disiinctive properties of A~dosols. Pp. 174229 w: Stewart, B.A. (ed.), Advances in soil science, volume 2. Spnnger·Verlag, New Yo~k Warkentin, B.P. & Maeda, T. 198 . 0. Physical and mechanical charact .. ensucs of And 1~ols: Pp. 281-301 in: Theng, B.K.G. (ed. ), Soils with variable charge. Off set Pubhcauons, Palmerston North, Zealand. West, N.E. 1993. BiodiveNew . rsity of rangelands. Invited synt f hesiS paper. Journal o Range Management 46: 2-13. · .. WI.lCOX, B· p·• Bryant' F C & Belaun V. 1987. An evaluation of rang e condti!On on one range · ' site in the Andes· of Central Peru . Journal of Range Management 40(1)• 41 45 · · ·

in:

APPENDIX A

PLANT LIST of the upper watershed of the Otlin river,

Los Nevados National Park, Colombia no.- Genus or species name

Family

12 Aa colombiana Schltr. Orch idaceae I Acaena elongala L. Rosaceae 2 AcaeM ovalifolia Ruiz et Pav. Rosaceae 100 Achyrocline alata (Kunth) D.C. 1 Asteraceac 4 Aciachne aciculari.s La:gaard Poaceae 224 Agropyron attenuatum (KuE.!!J) Roe_f!h & Sch_!llt Poaceae 6 Agrostis alba auct. cf. Poaceae 7 Agrostis araucana Phil. Poaceae 8 Agrostls breviculmis Hitchc. Poaceae 9 Agrostis foliara Hook. f. Poaceae 10 Agrosrls haenkeana Hitchc. Poaceae II Agrostis tolucensis Kunth Poaceae 13 Anrhoxantum odoratum L. Poaceae 14 Aphanactis piloselloides Cuatrec. Asteraceae 15 Arenaria serpens H.B.K. Caryophyllaceae 16 Arenaria spec. I Caryophyllaceae 17 Arenaria spec. 2 Caryophyllaceae 18 Azorella aretioides (Spreng.) D.C. Apiaceae 19 Azorel/a crena/a (Ruiz et Pav.) Pers. Apiaceae 20 Azorel/a multljida (Ruiz et Pav .) Pers. Apiaceae 21 Baccharis caespitosa (Ruiz et Pav.) Pers. var. alpina (H.B.K.) Cuatrec. Asteraceae 22 Baccharis caespirosalgenistelloides hybrid sp. Asteraceae 23 Baccharis caldasiana Cuatrec. Asteraceae 24 Baccharis genistelloides (Lam.) Pers. Asteraceae 25 Bacchari.s revoluta H.B.K. Asteraceae 26 Baccharis rupicola H.B.K. Asteraceae 27 Baccharis tricuneata (L.f.) Pers. Asteraceae 28 Bans/a pedicularoides Benth. Scrophulariaceae 29 Be/loa longifolia (Cuatrec. & Arist.) Sagast. & Dill. Asteraceae 30 Bidens triplinervia H.B.K. Asteraceae 31 Bomarea spec. Alstroemeriaceae 32 Bromus lanatus Kunth Poaceae 33 Buddleja sp. Loganiaceae 35 Calamagrosris coarctala (Kunth) Steud. Poaceae 36 Calamagrostis ejJusa (H.B.K.) Steud. Poaceae 37 Calamagrostis heterophylla cf. (Wedd.) Pilg. Poaceae 38 Calamagrosti.s intermedia (c. Pres!) Steud. Poaceae 39 Calamagrosti.s jamesonii Steud. Poaceae 40 Calamagrostis ligulata (H.B.K.) Hitchc. Poaceae

Introduced /native weed

nltive w~.:d native wâ&#x20AC;˘'<~~ native w~l!\1 introduced native w~.:d native w~ed introduced native weed

native we.:d native weed native we.:d

native we.:d native weed

Achyrocline a/ara comprises plant specimen initially recognized as Gnaphalium lanuginosum and Gnaphalium pellitum.

209

.: ,

Plant list 41 Calamagrostis macrophylla (Pilg.) Pi!g. Poaceae 42 Ca/amagrostis planifolia (H.B.K.) Trin. Poaceae 34 Ca!amagrostis recfjJ (H.B.K.) Trin. ex Steud. Poaceae 43 Calamagrosris spec. Poaceae 44路 Calceolaria mlcrobefaria Kraenzl. Scrophulariaceae 45 Ca/ceolaria perjoliata L.f. Scrophulariaceae 46 Calceo/aria spec. Scrophulariaceae 47 Cardamine spec. 1 Brassicaceae 48 Cardamine spec. 2 Brassicaceae 49 Carex bonplandil Kunth Cyperaceae SO Carex microglochin Wahlenb. Cyperaceae 51 Carex pichinchensis H.B.K. Cyperaceae 5 Carex spec. Cyperaceae 52 Carex tristicha Spruce ex Boott Cyperaceae 53 Castilleja fissifolia L.f. Scrophulariaceae 54 Castilleja pumila (Ben~.) Wedd. Scrophulariaceae SS Cerastium arvense L. 路路 Caryophyllaceae 56 Cerastiumjloccoswn Benth. Caryophyllaceae 57 Cerastiwn subspicatwn Wedd. Caryophyllaceae 58 CalobanJhus quitensis (Bartl.) H.B.K. var.1 Caryophyllaceae 59 Calobanthus quitensis (Bartl.) H.B.K. var.2 Caryophyllaceae 60 Calf'J1A uliginosa (Benth.) Cuatrec. Asteraceae 61 Cartaderia bijida Pilger Poaceae .62 Canaderia nirida (H.B.K.) Pilger Poaceae 63 Cotu/a mexica/Ul (D.C.) Cabrera Asteraceae 64 Dacrylis glomera/a L. Poaceae 65 Diplostephiwn revolurum Blake Asteraceae 66 Diplostephiwn schultzii Wedd. Asteraceae 68 Diplostephiwn sp. Asteraceae _ 69 Disterigma empetrifolium (H.B.K.) Drude Ericaceae 70 Distich/a muscoides Nees et Meyen Juncaceae 71 Draba hallii Hook. f. Brassicaceae 72 Draba pachythyrsa Tc. et Planchon Brassicaceae 73 Elaphoglosswn lindenii (Bory ex H e) Moore Aspleniaceae 74 Elaphoglosswn mathewsii (F&) Moore Aspleniaceae 75 Elatine cf. ecuadoriensis Molan E1atinaceae 76 Eleocharis stenocarpa Svenson Cyperaceae 77 Epilobium denticulatum Ruiz et Pav. Onagraceae 78 Erigeron chionophilus \Yedd. Asteraceae 79 Erigeron ecuadoriensis Hieron. Asteraceae 80 Eryngium humile Cav. Apiaceae 81 Esca/lonia mynil/oides L.f. Grossulariaceae 82 Espeletia hartwegiana Cuatrec. ssp. centro-arulina Cuatrec. Asteraceae 84 Festuca andicola H.B.K. Poaceae 85 Festuca breviaristata Pilger Poaceae 86 Festuca do!ichophyl/a Persl. Poaceae 87 Festuca procera H.B.K. Poaceae 88 Festuca rubra L. cf. Poaceae 89 Festuca sublimis Pilger Poaceae 90 Gaiadendron punctatum (Ruiz et Pav.) G.Don. Loranthaceae

native weed native weed

native weed introduced

native weed

introduced

native weed native weed

introduced

P/aru list

Rubiaceae 178 Galium corymboswn Ruiz et Pav. Rubiaceae 179 Galium hypocarpium (L.) End!. ex Griseb. Asteraceae 98 Gamochaeta purpurea (L.) Cabrera Ericaceae 91 Gaultheria spec. Gentianaceae 92 GenJiana sedifolia H.B.K. Geatianaceae 93 GenJianella dasyanlha (Gilg.) Fabris Geraniaceae 94 Geranium co/umblanum R. Knuth Geraniaceae 95 Geranium multipanilum Benth. Geraniaceae 96 Geranium sibbaldioides Benth. Asteraceae 97 GIUlphalium anlennarioides D.C. Asteraceae 99 GIUlphalium cf. graveolens H.B.K. Asteraceae 102 Gnapha/ium sp. 218 Grammitis monilifonnis (Lag. ex Sw.) Proctor Polypodiaceae Asteraceae 103 Gynoxys to/imensis Cuatrec. Gentianaceae 104 Halenia spec. Brassicaceae 105 Halimoiobos hispidula (D.C.) O.E.Schulz 107 Hesperomeles heterophylia (Ruiz et Pav.) Hook. Rosaceae Rosaceae 106 Hesperomeles lanuginosa Ruiz et Pav. Asteraceae 108 Hieracium avi/ae H.B.K. Asteraceae 109 Hieracium tolimense Cuatrec. Poaceae 110 Ho!cus lanatus L. Lycopodiaceae 111 Hupenla cruenta (Spring) Rothm. Apiaceae 112 Hydrocotyle bonplandii A. Rich. Apiaceae 113 Hydrocotyle sp. Hypericaceae 114 Hypericum /ancioides Cuatrecasas Hypericaceae 115 Hypericum laricifolium Juss. ssp. laricifoliwn Asteraceae 116 Hypochoeris sessilijlora H.B.K. Adiantaceae 117 Jamesonia goudotii (Hieron.) C.Chr. 路 Juncaceae 118 Juncus bufonius L. Rosaceae 119 Lachemilla andina (L.M.Perry) Rothm. Rosaceae 120 Lachemilla galioides (Benth.) Rothm. Rosaceae 121 Lachemilla hispidula (L.M.Perry) Rothm. Rosaceae 122 Lachemilla holosericea (L.M.Perry) Rothm. Rosaceae 123 Lachemil/a killipii (Rothm.) Rothm. Rosaceae 124 Lachemilla mandoniana (Wedd.) Rothm. Rosaceae 125 Lachemilla orbiculata (Ruiz et Pav.) Rydb. Rosaceae 127 Lachemilla pectinata (H.B.K.) Rothm. Rosaceae 220 Lachemilla spec. Rosaceae 222 Lachemilla tanacetifolia Rothm. Poaceae 128 Lolium multijlorum Lamk. Poaceae 129 Loliwn perenne L. Poaceae 130 Loliwn temulenJum L. cf. Asteraceae 131 Lucilia kunthiana (D.C.) Zardini Fabaceae 132 Lupinus alopecuroides Desr. Fabaceae 133 Lupinus microphyllus Desr. Fabaceae 134 Lupinus sp. Fabaceae 135 Lupinus tolimensis C.P. Smith Juncaceae 136 Luzula gigantea Desv. Juncaceae 137 Luzula racemosa Desv. Juncaceae 139 Luzula vu/canica Liebm. cf. Lycopodiaceae 140 Lycopodium clavatum L.

211 210

native weed

native weed native weed

.introduced

introduced

native weed

introduced introduced introduced native weed native weed

·' ·

Plant list 141 142 143 144 145 147

Miconia salicifolia (Bonpl. ex Naudin) Naud in Mannina spec. Mom ia meridensis Friedr. Muehlenbeclda vulcanica (Benth.) Endl. Muh/enbergia cleejil Lregaard Myrrhidendron glaucescens (Benth.) Coult.

Plant list Melastomataceae Polygalaceae Portulaccaceae Polygonaceae Poaceae

et Rose Apiaceae 148 Nasella pubijlora (frin. & Rupr.) E. Desv. Poace ae 149 Nenera granadensis (Mutis ex L.f.) Druce Rubiaceae 150 Niphogeton dissecta (Benth.) Macbr. Apiaceae 151 Niphogeton lingula (Wedd.) Math. et Canst. Apiaceae 152 Ophioglossum crotalophoroides Oseja Walt. Ophi oglossaceae 153 Oreobolus clee.fi/ L .f. Mora Cyperaceae 154 Oreomyrrhis andicola (Kunth) Hook . f. Apiaceae 155 Oritrophium limnophUum (Sch. Bip.) Cuatrec. ssp. murisianum (Cuatrec.) Cuatrec. Asteraceae 156 Orirrophium peruvianum (Lam .) Cuatr ec. ·Asteraceae 157 Oxalis spec. 1 Oxal idaceae 158 Oxalis spec. 2 Oxali daceae 159 Paspalum bonp/andianum Fliigge Poaceae 160 Pedicularis incurva Benth . Scrophulariaceae 191 Pemacalia reissiana (l·lieron.) Cuatr ec. Asteraceae 161 Pemacalia vaccinioides (H.B .K.) Cuatr ecasas Astera~eae 162 Peperomia hartwegiana Miq. Piperaceae 163 Peperomia spec. Piperaceae 164 Pernettya prosrrata (Cav.) D. C. Erica ceae 165 Plantago australis Lam. Plantaginaceae 166 Plamago linearis H.B.K. Plantaginaceae 167 Plamago rigida H.B.K. Plantaginaceae 168 Poa annua L. Poaceae 169 Poa pratensis L. Poaceae 170 Poa subspicata (Pres!) Kunth Poaceae 171 Poa trivialis L. Poaceae 221 Poa spec. Poaceae 172 Polypogon elongatus H.B.K. Poaceae 173 Potemilla heterosepala Fritsch Rosaceae 174 Ranunculus flagellifonnis Smith Ranunculaceae 175 Ranunculus peruvianus Pers. Ranunculaceae 176 Ranunculus praemorsus H.B. K. ex D.C. Ranunculaceae 177 Ranunculus spec. · Ranunculaceae 181 Rhynchospora spec. Cype raceae 182 Ribes leptostachum Benth. Saxifragaceae 183 Ribes spec. Saxifragaceae 184 Rubus sp. Rosaceae 185 Rumex acetoselfa L. Polygonaceae 186 Rumex crispus L. Polygonaceae 3 Saracha quitensis (Hoo le.) Miers Solanaceae 187 Satureja nub/gena (Kunth) Briq. Lami aceae 188 Senecio canescens (H.B .K.) Cuatrec. Asteraceae 189 Seneciofonnosus H.B.K. Asteraceae 190 Senecio latijlorus Wedd. Asteraceae

212

native weed

native weed native weed

192 193 194 195 196 197 198 199 225 200 201 202 203 223 204 205 206 207 208

209 210 211 212 213 214 215 216

Asteraceae Asteraceae

Senecio repens D·C. Senecio sp. Sherard/a arvensis L. Sibthorpia repens (Mutis ex L.) Kuntze Sisyrinchium convolutum Nocca Slsyrlnchium trinerve Baker Solanum columbianum DunaJ cf. Solanum ruberosum L. Solanum sp. Stachys elliptica Ku~ Stellaria cuspidata Wllld. ex Sch. Taraxacwn cf. officlnale Weber 1ibouchina spec. Tragopogon spec. . . Trifolium cf. amabrle H.B.K. Trisetum spec. . Trisetum irazuense (Kun tze) ~Jtchc. Trisetum spicatum (L.) K. Richter Uncinia meridensis Steyerm. Uncinia spec. . Valeriana plantagmea H.B.K. Vallea stipularis Mutis ex L.f. Veronica serpy/lifolia L. Viola spec. Weinmannia spec. Werneria crassa Blalce ssp. crassa Werneria humilis H.B.K.

Rubiace a~

Scrophulariacue Iridaceae Iridaceae Solanaceae Solanaceae Solanaceae Lamiaceae Caryophyllaceae Asteraceae Melastomataceae Asteraceae Fabaceae Poaceae Poaceae Poaceae Cyperaceae Cyperaceae valerianaceae Elaeocarpaceae Scrophulariaceae Violaceae Cunoniaceae Asteraceae Asteraceae

introduced introduced introduced introduced introduced native weed

introduced introduced native weed native weed

213

introduced

native crop native weed introduced· introduced native weed native weed

native weed native weed

APPENDIX B

Documentation simulation model Unit

State variables of the model BARAC BUOAC BUYAC BUTAC DEPAC ESPAC FIREAGE FORAC GRI ROSAC SHOAC SHRAC TB

Bare soil percentagl! Dead bunch grass cover Green bunch grass cover Total bunch grass cover Dead biomass percl!ntage bunch grasses Stem rosette cover Fire age Forb cover Grazing intensity Cover of ground rosette layer Short grass cover Shrub cover Tussock base cover

Submodel BUNCH, development bunch grass cover Variables BUOCON BUOLIT BUYCON BUYMC BUYPROD BUYSEN GRTB GSBUY Parameters BUOGF BUYGF ·. BUYRGR FSB GSB LSBUO LSBUY RESD TBRGR TITB

Grazing effect on dead bunch grass cover Effect of litter fall on dead bunch grass cover Grazing effect on green bunch grass cover Maximum green bunch grass cover Increase of green bunch grass cover b.y production Effect of senescence on green bunch grass cover Growth rate tussock base Grazing stimulus of bunch grass production Grazing factor of dead bunch grasses Grazing factor of green bunch grasses Relative growth rate green bunch grass cover Fire stimulus factor of bunch grass production Grazing stimulu~ factor of bunch grass production Life span dead bunch leaves Life span green bunch grass leaves Reserve depletion rate of tussock bases Relative growth rate of tussock bases Factor of trampling impact on tussock bases

% % %

month %

(cowdr./50 m~ % % % %

Unit /parameter value %/month %/month %/month %

%/month %/month %/month

1.94 • 10·3 1.11 • w-3 0.053 /month 1.12 0.22 40 months 20 months 0.15 %/month 0.025/month 0.032

Submodel FSP, development stem rosette cover Variables ESPCON ESPMC

Effect of grazing on stem rosette cover Actual maximum stem rosette cover 215

%/mqnth %

Docwnenration simulatioll model Incrc:ase of stem rosette cov er by production Grazmg stimulus of t s em rosetre production

ESPPROD GSESp

Documenration simulation model

farametm ~L

ESPGF ESPMAX ESPRGR FSE GSE PF

Submodel SHR

36.1% 4• 3.0 0.22 1.54

Grazing stimulus of h s rub production Effect of graz.i Actual m · ng on shrub cover axJmum shrub cover Increase of shrub cover by production

. Fire stimulus factor of s hrub production r: . Grazing stimulus tactor . of shrub . Grazmg productiOn factor of sh b s ru Potential maxim Relative growth ~~ sh~b cover Limitation of sh be o shrub cover ru growth by bunch grasses . · on of shrub LlmJtati growth by stem rosettes

Submodel SHO

%/month. % %/month 1.55 0.50 1.5 • JO·l 17.8 % 0.012/month 0.08 0.46

' development short grass cover

GSSHO SHOCON SHOMC SHOPROD

10·~::

GSSHR SHRCON SHRMc SHRPROD

0.35 0.15 •

' development shrub cover

FSSR GSSR SHRGF SHRMAX SHRRGR SRBL SREL

Subinodel ROS, development ground rosette layer

Limitation of stem rosette Grazing factor of stem growth by bunch grasses Potential m . m ste rosettes ax1mu Relative growth rate ofmsterosetle cover . m rosette cover FIre stimulus tiacto f r o stem rose1te production . G · razmg stimulus factor of stem rosette production Power factor

Grazing stimulus of h

Effect of grazing on short grass production ActUal maximu h s ort grass cover

Increase of sho~ s ort grass cover grass cover by production

Parameters GSSO Graz!ng stimulus factor of SHOGF ~hort grass production Graz.mg factor of short grasses SHOMAX Potential m . axJmum short grass cover SHORGR Relative growth . SOBL Limitation of sh rate short grass cover SOEL L!m!tation of sh~~ :::: g;owth by bunch grass cover SOTL LJmJtation of short grass g owth by stem rosettes . growth by tussock bases

%/month % %/monih 0.58 8.52. JO·l

37.0% 0.16/month 0.51

Variables GSROS ROSCON ROSMC ROSPROD

Grazing stimulus of production ground rosette layer Effect of grazing on ground rosette layer Actual maximum cover of ground rosette layer Increase of ground rosette layer by production

Parameters FSR GSR ROBL ROEL ROSGF ROSL ROSRGR ROTL

1.17 Fire stimulus factor of production ground rosette layer 0.20 layer rosette ground tion produc of factor us stimul g Grazin . ·o.99 cover -grass bunch by Limitation of ground rosette growth 0.10 Limitation of ground rosette growth by stem rosettes 9.0 • 10"1 Grazing factor of ground rosette layer 0.99 Limitation of ground rosette growth by short grasses 0.08/month Relative growth rate of cover of the ground rosette layer 0.65 bases k tussoc by growth Limitation of ground roselle

%/month %

%/month

Submodel FOR, development (medium tall} forb cover Variables FORCON FORMC FORPROD

Grazing effect on forb cover Actual maximum forb cover Increase of forb cover by production .

% %/month

Parameters FAL FBL . FEL FORGF . FORRGR FSF

Limitation of forb growth by shrubs Limitation of forb growth by bunch grass cover Limitation of forb growth by stem rosette cover Grazing factor of forbs Relative growth rate of forb cover Fire stimulus factor of forb production

0.3 0.7 0.2 0.016 0.095/month 1.15

e Submodel BAR, development percentage bare soil surfac Parameters

BAL BBL BEL BFL

Limitation of bare soil percentage by shrubs Limitation of bare soil percentage by bunch grasses Limitation of bare soil percentage stem rosettes . Limitation of bare soil percentage by forbs

0.20

0.15

216 217

0.103 0.70 ' o.o52

0.368

•

Documentation simulation model

Submodel GRI, calculation of grazing intensity Variables VA GRIAV us~

BUTAC{l) . Vegetation attractiv~ness, or forage availability Average grazing intensity · Use factor, determining the degree of utilization

: '

Submodel FIREAGE, simulation of fire occurrence

Forcing function FI(TIME) Burning function that toggles the possibility of fire Variables BURN FUEL

Burning potential Fuel load

+ BUOAC(l)

IF (FIREAGE(I) .LE. 48) THEN GRTB = -RESD ELSE GRTB TBRGR- TITB*LOGlO(GRI(I)+l) ENDIF

(%) (cow dr./50m 2)

= BUY AC(l)

DTB{I) = DTB{l) + GRTB TB(l) = AMAXl (O.l , TB(I)) TB{l) = AMIN1(30.0, TB(I)) DEPAC(l)

% %

Model equations in SENECA 10

= lOO*BUOAC(l)/BUTAC(I)

IF (FIREAGE(l) .EQ. 0) THEN BUYAC(I) = 0.1 * TB(I) BUOAC(I) = 0 BUTAC(l) = BUYAC(I) + BUOAC(I)_ DEPAC(I) = 0 ENDIF CONTINUE END

C Submodel BUNCH C Development bunch-grass cover DO 101=1 ,4 BUYSEN = (1/LSBUY) * BUYAC(I) BUOLIT = (1/LSBUO) * BUOAC(I) BUOCON = BUOGF * GRI(I) • BUOAC(I) DBUOAC(I) = DBUOAC(I) + BUYSEN - BUOLIT - BUOC ON BUOAC(I) = AMIN 1(80.0, BUOAC(I)) BUOAC(I) = AMAXl(O.O, BUOAC(I)) GSBUY = GSB*LOG(GRI(I)+ 1) + 1 IF (FIREAGE(I) .LE. 48) THEN BUYPROD = FSBUY * LOG(49/(FIREAGE(I)+ 1)+ 1.7183 )* & GSBUY * BUYRGR * TB(I) ELSE BUYPROD = GSBUY * BUYRGR * TB(I) ENDIF BUY CON = BUYGF * GRI(I) • BUYAC(I) . DBUYAC(I) = DBUYAC{I) + BUYPROD - BUYSEN BUYCON BUYMC = 80.0 - BUOAC{I) BUYAC(I) = AMINl(BUYMC, BUYAC(l)) BUYAC(I) = AMAXl(O.l, BUYAC(I))

C Submodel ESP C Development Espeletia or stem rosette cover DOlO 1=1,4 ESPMC = ESPMAX- EBL*BUTAC(l) ESPMC = AMAXl(O.l,ESPMC) GSESP = GSE*LOG(GRI(I)+ 1) + 1 IF (FIREAGE(l).LE.48) THEN ESPPROD = FSE*LOG(49/(FIREAGE(I)+ 1) + 1.7183)*GSE SP* ESPRGR*ESPAC(I)*(l -ESPAC(I)/ESPMC) & ELSE ESPPROD = GSESP*ESPRGR*ESPAC(I)*(l-ESPAC(I)/ESPM C) ENDIF ESPCON = ESPGF*ESPAC(l)*GRI(I)**PF DESPAC(I)

= DESPAC(I) + ESPPROD- ESPCON

ESPAC{l) = AMINl(ESPMC, ESPAC(I)) ESPAC{l) = AMAXl(O.l, ESPAC(l))

218 219

·.. · j

Documentation simulation model

IF (FIREAGE(I) .EQ. 0) THEN ESPAC(I) = 0.5 * ESPAC(I) ENDIF CONITNUE

SHOAC(I) = AMIN1(SHOMC, SHOAC(I)} SHOA.C(l) = AMAX1(0.1, SHOAC(I))

END

lF (FIREAGE(I) .EQ. 0) THEN SHOAC(I) = 0.1 * SHOAC(I) ENDIF

C Submodel SHR C Development shrub cover

DO 10 1=1,4 SHRMC = SRRM AX- SRBL*BUTAC(I)- SREL*ESPAC(I) SRRMC = AMAXI(O.l, SHRMC) GSSHR = GSSR * LOG(GRI(I)+ 1) + 1 IF (FIREAGE(I).LE.48) THEN *FSSR*LOG(49/(FIREAGE(I)+ 1)+ 1. 7183) SHRPROD = GSSHR *SHRAC(I)*(l-SHRAC(I)/SHRMC)

CONTINUE END

C Submodel ROS C Development of cover of ground rosette layer ROEL*ESPAC(I) - ROSL*SHOAC(I) 10 I= 1,4 DOROSMC = (100 - ROBL*BUTAC(I) - ROTL*TB(I)) * 1.1 & ROSMC = AMAX1(0.1, ROSMC) GSROS = GSR*LOG(GRI(I)+ I)+ 1

-----Â·-,;.:~ ,,,

ELSE

= GSSHR*SHRRGR*SHRAC(I)*(l-SHRAC(I)/SHRMC) SHRPR IF OD END

GE(I) .LE. 48) THEONS*LOG(49/(FIREAGE(I)+ l))*ROSRGR* IF (FIREA ROSPROD = GSROS*FSR ROSAC(I)*(l-ROSAC(I)/ROSMC) &

SRRCON = SHRGF*SHRAC(I)*GRI(I)

ELSE ROSPROD

DSHRAC(I) = DSHRAC(I) + SHRPROD - SHRCON

i~~~N = ROSGF * ROSAC(I) * LOG(GRI(I)+ 1)

SHRAC(I) = AMIN1(SHRMC, SHRAC(I)) SHRAC(I) = AMAXI(O.l, SHRAC(I))

DROSAC(I)

IF (FIREAGE(I) .EQ. 0) THEN SHRAC(I) = 0. I * SHRAC(I) END IF CONTINUE

= DROSAC(I) + ROSPROD.: ROSCON

ROSAC(I) = AMIN1(ROSMC, ROSAC(I)) ROSAC(I) = AMAXI(O.l, ROSAC(I))

END

C Submodel SHO C Development of short grass cover

=GSROS*ROSRGR *ROSAC(I)*(l-~OSAC(I)/ROSMC)

IF (FIREAGE(I) .EQ. 0) THEN ROSAC(I) = 0.1 * ROSAC(I) ENDIF .CONTINUE

END

DO 10 I= 1,4 ) SHOMC = SHOMAX- SOTL*TB(I) - SOBL*BVTAC(I)- SOEL*ESPAC(I SHOMC = AMAXI(O.l, SHOMC) GSSHO = GSSO*LOG(GRI(I)+ 1) + 1 C) SHOPROD = GSSHO*SHORGR * SHOAC(I)*(l-SHOAC(I)ISHOM SHOCON = SHOGF * SHOAC(I) * GRI(I)

C Submode1FOR C Development of (tall) forb cover -FEL"'ESPAC(I) -FAL*SHRAC(I) 10 I= 1,4 DOFORMC = 100 -FBL*BUTAC(I) -TB(I) -SHOAC(I) -ROSAC(I) & FORMC = AMAX1(0.1, FORMC) ORAC(I)*

48)*TLHOENG(49/(FIREAGE(I)+ l))*FORRGR*F GE(I)= .LE. lF (FIREA FSFOR FORPROD

DSHOAC(I) = DSHOAC(I) + SHOPROD - SHOCON &

(1-FORAC(I)/FORMC)

220 221

Documentation simulation model ELSE FORPROD = FORRGR • FORAC(l)-(1-EORAqi)/F ORMC) ENDI F _

FORCON

= FORGF * FORAC(I) * LOGIO(GRI(I)+l)

Documeruarion simulaJion model C Submodel GRAZ C Calculation of grazing intensity DO 10 1=1,4 VA= 2*BUYAC(I) +BUOAC(I) + 4*SHO. AC(I) + 2*ROSAC(I) +FORAC(I) GRIAV = EXP(0.005*VA +0.98) GRI(I) = USEF * GRIA V

DFORAC(I) = DFORAC(I) + FORPROD - FORCON FORAC(i) = AMIN1(FORMC, FORAC(I)) FORAC(I) = AMAX1(0.1, FORAC(I)) lF (FIREAGE(I) .EQ. 0) THEN

FORAC(I) = 0.05 • FORAC(I) ENDIF 10 CONTINUE END

IF (FIREAGE(I) .EQ. 0) THEN GRI(I) = 0 END IF CONTINUE

END

C Submodel BAR C Development!of bare soil percentage DO 10 I= 1,4 . BARAC(I). = 100 -TB(I) -BBL*BUTAC(I) -BEL*ESPA C(I) & -BAL*SHRAC(l) -SHOAC(I) -ROSAC(I) -BFL*FO~C( I) BARAC(I)

= AMAXl(O, BARAC(I))

lF (FIREAGE(I) .EQ. 0) THEN

BARAC(I) :.: 100

ENDIF

CONTINUE

END C Submodel FIRE C Simulation of fire occurence DO 10 I= 1,4 FUEL = BUTAC(I)*DEPAC(I)*O.Ol + 0.7*ESPAC(I) BURN = FI(fiME) * FUEL IF (BURN .GE. 32) THEN FIREAGE(I) = 0

ELSE

DFIREAGE(I) = DFIREAGE(I) + 1

ENDIF

CONTINUE END

222

223

SUMMARY The high mountain ecosystem of the northern Andes, the ' paramo', i-s under increasing human influepce. Gmdual changes in the original bunch-gmss vegetatio n take place in relation to burning and gmzing, but mpid tmnsfonnation thrQugh the cultivation of native tubers also occurs. The paramo zone is of vital importance to the human population of the inter-Andean valleys as it regulates the water flow to the lower (densely populated) areas. Within the boundaries of Los Nevados National Pa'rk in the Cordillera Ccntml of Colombia , a small population of farmers take care of a number of isolated farms. Livestock are raised for beef production, and a related land use is the burning of vegetation to stimulate the growth of fresh gra.ss shoots. Although it was generally assumed that these practices cause degradati on of the ecosystem, hardly any knowledge concerning the rate and spatial extension of these processes was available to those responsible for the management of the paramo. In this dissertation, vegetation patterns in relation to burning and gmzing are modelled . Both spatial and temporal aspects of human-influenced vegetation dynamics are described. The objectives included the development of spatial models of the management regime, of both cattle distribution and frre history; the explanation of variation in actual vegetatio n patterns; and the prediction of the response of vegetation to changes in management variables. Besides these scientific objectives, an important aim was to translate the results into implications for management. In the region of Los Nevados, the conservation of biological diversity and the maintenance of a stable Qydrological system have been considered top priorities by the management organizationsÂˇ concerned. Clustering and ordination techniques were applied to analyze the variation in vegetation sOllcture and floristic composition in relation to management variables. The zonal vegetation of the investigated pararno proper (3900 to 4200 m) is dominated by species of the bunch or tussock grass Calamagrosris and the stem rosette Espeleria harrwegiana ssp. cemroandina. On moderate slopes, the vegetation opens up under an intermediate gmzing intensity. In these situations, the cover, height, and diameter of the bunch grasses decrease, whereas the percentage of bare soil and tmmpling impact in the form of terracettes increase. On more gentle slopes and flat terrain, short matted grasslands and herblands develop. The most importan\ short-growing species are Lachemilla orbicu/ata, Ca/amagrostis coarcrara, and Aciachne Âˇacicularis. A list of plant species showing either a positive or negative correlation with burning and grazing variables is presented. A vegetation map was prepared on the basis of aerial photographs at scales I: 25,000 to I: 33,000. This map shows that Âą 50% of the original dense bunch grasslands have been changed into more open vegeiatio n structures due to burning, grazing, or a combination of both. Tlie floristic composition of the vegetation changes more slowly, over time spans of several decades or more. The grazing behaviour of cattle was studied in field observations. In this way, tbe botanical composition of cattle diet was determined Forage preference tends to be related to the quality of the consumed plant groups in terms of average digestibility and crude protein content. Short grasses, forbs, and species of the ground rosette layer are preferred to bunch grasses. A low forage quality (7.5% crude protein), a low cow productivity index (35 kg per 100 kg of body weight per year), the long distances caule walk to meet their feeding requirements agree with the characteristics of an extensive low-production grazing system.

225

Swnmary

Summary

Although the Espeleria stem rosettes appa rently possess certain unique fire adaptive adverse impacts of management on aspec ts of population dynamics were aerno~ISIT:or 'i.-l.~·~·:.w·ii!l especially in the case of combined burni ng and high intensity grazing. An adult grow Jh of 8.8· em per year was meas11red in undis turbed situations. An increased growth rate, included a fire-induced stimulation facto r was observed at burned sites. After mortality increases, especially among the tallest individuals: the mortality rate can than twice the natural rate. After initial high juvenile mortality dl!e to fire, increased survival and higher" juvenile growth quick ly compensate for this loss. The clustering . individuals was observed in intensively graze d situations; this may provide protection ·• · trampling and scraping impact by cattle. Increased mortality rates were likewise obser ved. · The bunch grasses of Ca/amagrostis are characterized by a low productivity. Ym:rr,,.,; ,<: recovery takes about 10 years on average. Changes in diameter distribution and the per. cental!e:'.,{ of fragmented bunches (or 'tussocks') best reflect management impact. The fra;gffilent< uiotn tussocks can be related to burning, grazi ng, or a combination of both. The temp or~ induced fragmentation of tussocks into separate regenerating tillers is often main tained grazing. When grazing pressure continues to be high, the tussock bases gradually disin tegrate: · This fragmentation is due to trampling impa ct and the depletion of reserves by consumptio n.' The above findings were used to develop a simulation model of vegetation developm ent under · different management strategies. Functiona l plant groups wereTecognized, representin g groups of species with a similar response ' to ecologi®l and management variables. Iterative calibration on the basis of time series of field data served to improve the structure of the model and estimate parameter values. The model can be used to predict the course of vegetation development under different burning and grazing strategies. The outco mes are consistent with the results of the spatial models described below.

The critical level below which the bunch grasses a.re sus~a i~c~ ppcar under intensive grazmg (wnhm JOis at a~out 0.16 A.U.0a. ~'h.e tUSSOCkScan fd P'tdl J..-.trt-. u••~ years), whereas any tncrease . in tussock base cover is slow· Landscape heterogeneity in the form of patch iness is maintained by low fire frequenc~e s, l~w intensit razing or a combination of both. The farmers also tend to bum t~e t~rrat n u.mts referr:d cattie, such as the lateral mora ines and glaciated lava fields. Bummg IS pracused ~~ interme~ate distances from the fincas, as the low fuel load nearby does not all?w the spread of fire. Under the influence of highe r grazing intensities, a short matted vegetauon has frequently developed close to the fincas.

. . · f ariation in biological diversity is Besides landscape heterogenetty, 1uauo n o bv f · I £ onservation purposestheTheeva for releves ge num er o vascu ar P!ant taxa . . ::;~~~~~~i~t~~bed bunch grasslands. is 38.5avera (± 3.1). There is a weak tendency of increasi.ng diversity under light grazing intensities . Unde r moderate to intense grazlflg, ~~we~er, sp~c; ~~ richness is lower - as is also the case when frequent fires occur. Potato cu ~vanon an highest grazing intensities are related to high proportions of introduced spectes (up to 30%) and a decrease in the total number of taxa. · A scheme with a summary of the long-te~ ~ffec ts of burning and grazing IY)anagement on various aspects of the paramo ecosystem ts gtven on page 204.

A spatial model of canle distribution was deve loped on the basis of multiple regression on point observations. A measure of forage avail ability (vegetation attractiveness for graiing) was derived from the weighted sum of functiona l plant groups, taking into account forage quality and preference. This variable best expla ined the variation in grazing intensity. The other significant predictors of grazing intensity were (in order of importance): slope, stock ing density, terrain preference of cattle, and tl_le distance to the finca. Attribute maps of these ·variables were prepared and combined in the form of a map of cattle distribution to predict grazing intensity. The fire history of a pilot area where burning activity is important was reconstruc ted by analyzing a time series of aerial photograph s of 10 different years between 1959 and 19R9. An accurate GIS-based procedure was deve loped for the purpose of retrospective moni toring. The average fire frequency for the pilot area was estimated at 1.0 to 1.6/100 yr. Fire frequency has not diminished since the creat ion of the national park. Forage availabilit y tend~ to decrease at fire frequencies of more than once a decade. The farmers' justificatio n of burning practices, i.e. to increase forage availability, is only valid at lower frequencie s. To conserve stable regional and local hydro logical systems, the protective cover of the soil surface must be maintained. Assuming that the bunch-grass layer provides the best prote ction , it is important to note that the bunch grass es degrade under the present grazing inten sity (0.37 A.U./ha for utilized vegetation patches).

226

227

Resumen

RESUMEN El ecosistema de alta montana al norte de los Andes, el paramo, esta sometido cada-vez m~s a Ia ·influencia humana. En relaci6n a pr.icticas de quema y pas10reo, cambios graduales han tenido Iugar en Ia vegetaci6n original del pajonal, pero transformaciones rapidas debido.al culuvo de tuberculos nativos tambien ocurren en Ia region. La zona de paramo es de vital .:. · importancia ~ara Ia poblaci6n de los valles intramontanQs andinos ya que estos regulari' el . flujo de agua hacia las zonas mas bajas (densamente pobladas). En Ia Cordillera Central de ·. ·. Colombia, dentro de los Iimites del Parque Nacional Natural Los Nevados, una pequeiia. ' poblaci6n de campesinos esta encargada del cuidado de un numero de fincas aisladas. El · ganado es criado para Ia producci6n de came, y Ia quema de Ia vegetaci6n es un uso de Ia .. tierra asociado con esta actividad para estimular el crecimiento de brotes frescos de pasta. ;..· pesar de que generalmente se asume que estas practicas son causantes de Ia degradaci6n del ecosistema, diffcilmente los responsables del manejo del paramo tenlan conocimiento exacto de Ia tasa y Ia extension espacial del proceso. En esta disertaci6n, son modelados los patrones de vegetaci6n en relaci6n con Ia quema y el ."' pastoreo. Asf mismo son descritos tanto los aspectos espaciales como temporales relacionados con Ia influencia antr6pica en Ia didmica de Ia vegetaci6n natural. Los objetivm; de est<i " investigaci6n incluyen el desarrollo de modelos espaciales del regimen de manejo, tanto para Ia distribuci6n del ganado como para Ia historia de las quemas; Ia explicaci6n de Ia variaci6n en patrones de vegetaci6n actual; y Ia predicci6n de Ia respuesta de Ia vegetaci6n a los cambios en las variables de manejo. Ademas de los objetivos cientificos, un fin imponante fue traducir los resultados dentro del contexto de manejo. En Ia regi6n de Los Nevados, Ia conservaci6n de Ia diversidad biol6gica y el mantenimiento de un sistema hidrol6gico estable han sido considerados aspectos prioritarios por pane de las organizaciones de manejo involucradas. Tecnicas de agrupaci6n y ordinales fueron aplicadas con el fin de analizar Ia variad6n en Ia estructura de Ia vegetaci6n y Ia composici6n floristica en relaci6n con las variables de manejo. La zona de vegetaci6n del paramo propiamente dicho (3900 a 4200 m) estci dominada por especies gramineas de Calamagrostis que crecen en forma de macolla y el frailej6n Espelecia harrwegiana ssp. centroandina. En. Iaderas de inclinaci6n moderada, Ia vegetaci6n se abre bajo una intensidad de pastoreo intermedia. En estas situaciones disminuyen Ia cobertura, Ia altura y el diametro de Ia macolla, mientras aumentan el porcentaje de suelo desnudo y el impacto-de pisoteo manifestado en las formas de terracetas. En laderas menos inclinadas y terrenos pianos se desarrollan esteras d~ gramlneas y herbaceas conas. Las especies de tallo cono mas imponantes son Lachemilla (lrbiculata, Calamagrosris coarctata y Aciaclme acicularis. Se presenta una lista de especies de plantas que muestran ya sea una correlaci6n positiva o negativa con las variables de quema y pastoreo. El mapa de vegetaci6n fue preparado basado en fotografias aereas con escalas que van del 1: 25,000 al 1: 33,000. Este mapa muestra que aproximadamente e1 50% del pajonal denso ha sido tra~sforrnado a estructuras de vegetaci6n mas ralas debido a las pr.icticas de quema y pastoreo 6 Ia combinaci6n de ambas. Los cambios en !:1 composici6n florfstica de Ia vegetaci6n son mas lentos, estos ocurren sobre un perfodo de tiempo de varias decadas o mas. Los hcibitos de pastoreo del ganado fueron estudiados a traves de observaciones de campo. De esta forma fue determinada Ia composici6n botanica de Ia dicta del ganado. 228

La prcfcrcncia de forrajc ticnde a estar relacionada con Ia calidad de los grupos de plantas consumidas en tcrminos del promedio de digcstibilic.lad y el promedio de contenido brutO de protefna. Los pastos conos, las hierbas y las cspecies cubridoras del suelo son preferidos en comparaci6n con las macollas. Un forraje de baja calidad (7.5% protelna cruda), un bajo fndice de productividad vacuna (35 kg por cada 100 kg de peso corporal por aiio), y las largas distancias que el ganado camina para satisfacer los requerimientos de nutrici6n concuerdan con las caracteristi cas de un sistema de pastoreo extensive de baja producci6n. A pesar de que aparentemente los frailejones de Espeleria posecn ciertos rasgos unicos de adaptaci6n al fuego, se demosrr6 que el manejo produce impactos adversos en aspectos de Ia dimimica de Ia poblacion, especialmente en el caso donde se combina Ia quema con el pastoreo de alta intensidad: La tasa de crecimiento adulto de 8.8 em por ai'io fue medida en situaciones sin penurbar. Se observ6 un incremento en Ia tasa de crecimiento en Iocalidades · afectadas por el fuego Io cual incluye un fac tor de estimulaci6n inducido por el fuego. Despues de la quema Ia mortalidad aumenta especialmente entre los individuos mas altos: Ia tasa de monalidad puede ser mas de dos veces mayor que Ia tasa natural. Entre los juveniles, Ia alta mortalidad inicial en sitios de quema se ve r.ipidamente compensada por el mayor numero de plantufas que sobreviven y por el mayor ritmo de crecimiento de los juveniles en estos sitios. La ocurrencia de individuos en grupos se observ6 en situaciones de pastoreo intensivo, posiblemente como forma de proteccion contra el impacto del ganado. a traves del pisoteo y raspaduras. Los indices de mortalidad se ven incrementados en esta situaci6n a! igual que lo observado despues de una quema. · Las macollas de Calamagrosris se caracterizan por una productividad baja. Despues del fuego Ia recuperaci6n tarda un promedio de 10 aiios. Los cambios en Ia distribuci6n del diametro y el porcentaje de fragmentaci6n de las macollas reflejan de Ia mejor manera el impacto del manejo. La fragmeiilaci6n de las macollas de pas to puede ser relacionada con la quema, el pastoreo o Ia combinaci6n de ambos. La fragmen taci6n temporal causada por .el fuego en macollas con grupos de retoiios separados es frecuentemente mantenida por el pastoreo. Cuando Ia intensidad de pastoreo continua siendo alta, las bases de Ia macolla se desintegran • gradualmente. Esta fragm~ntaci6n es debida al impacto de pisoteo y a! agotamiento de las reservas por el consumo. · Los resultados arriba mencionados fueron usados para crear un modelo de simulaci6n del desarrollo de Ia vegetaci6n bajo diferentes estrategias de manejo. Grupos funcionales de plantas fueron reconocidos, representando grupos de especies con una respuesta similar a las variables ecol6gicas y del manejo. La calibraci6n iterativa con base en secuencias temporales de datos colectados en el campo sirvi6 para mejorar Ia estructura del modelo y esrimar los valores de los parametros. El modelo puede ser aplicado para predecir el curso del desarrollo de Ia vegetaci6n bajo diferentes estrategias de quema y pastoreo. ·Las respuestas son consistentes con los resultados de los modelos espaciales descritos a continuaci6n. Un modelo espacial de Ia distribuci6n del ganado fue desarrollado con base en analisis regresi6n multiple sobre los puntas de observaci6n. Una medida d.e Ia disponibilidad de forraje (valor arractivo de la vegetaci6n para pastoreo) fue derivada de Ia suma segtln el peso de los grupos funcionales de plantas, tomando en cuenta Ia calidad y Ia preferencia del forraje. Esta variable explica de la mejor manera Ia variaci6n en Ia intensidad del pastoreo.

229

--·- --- ---·- ··-- . ·- -------.._;, :, 0

Resumen Los pronosticadores mas significante s de Ia intensidad de paswreo son (en orden de .' imponancia): Ia pendiente, Ia densidad de Ia manada, Ia preferencia de re.m:no por el ganado, . y Ia distancia a Ia finca. Mapas de arriburos de estas variables fueron preparad os y' · combinados en forma de un mapa de Ia distribuci6n del ganado para predecir Ia intensidad ·.':: de pastoreo. La historia de Ia quema de un area pHoto donde Ia actividad de incendio es importan te fue •: reconstruida a traves del ancilisis de una serie de fotografias aereas de 10 aiios diferentes entre 1959 y 1989. Un procedimiento preciso basado en SIG fue desarrollado con el prop6sito de ' monitoreo retrospec tive. La frecuencia de quema promedio para el area piloto fue estimada en 1.0 a 1.6/100 anos. La frecuencia de incendios no ha disminuido desde Ia creaci6n del _· parque nacional. La disponibilidad del forraje tiende a disminuir. cuando Ia frecuencia de · incendios es mayor de una vez cada diez ailos. La justificaci6n de los campesin os para Ia pnictica de incendios, con el objetivo de incrementar Ia disponibilidad del forraje, es solamente vc11ida a frecuencias mas bajas. Para conservar sistemas hidrol6gicos estables regionales y locales, Ia cobettura protectora de Ia superficie del suelo debe de ser mantenida. Asurniendo que Ia capa de macollas provee Ia mejor protecci6n, es imponante mencionar que el pajonal se degrada bajo Ia presente intensidad de pastoreo (0.37 A.U/ba para parches de vegeta-ci6n utilizados). El nivel critico bajo el cual las macollas se sostienen es aproximadamente de 0.16 A.U/ba. Las macollas putden desaparecer rapidame nte bajo pastoreo intensive (dentro de 50 anos), mientras cualquier incremento en Ia cobenura de Ia base de Ia macolla de pasto es Iento. La heterogeneidad del paisaje en Ia forma de un mosaico de vegetaci6n es mantenida por bajas frecuencias de quema, una baja intensidad de pastoreo o por Ia combinac i6n de ambos. Tambien los campesinos tienden a quemar las unidades de terrene preferidas por el ganado, tales como morrenas laterales y campos de lava con rasgos glaciales. Los incendios son practicados a distancias intermedias de las fincas, ya que cerca Ia baja carga de combustible no perrnita Ia dispersi6n del fuego. Bajo Ia influencia de intensidades altas de pastoreo, esteras _de vegetaci6n corta se han desarroll ado frecuentemente cerca de las fincas. Ademas de Ia heterogeneidad del paisaje, Ia evaluaci6n de Ia variaci6n en Ia diversidad biol6gica es imponante para Ia conservaci6n. El numero promedio de taxa de plantas vasculares para los levantam ientos correspondientes a pajonales que no han sido perturbados es de 38.5 (± 3.1 ). Hay una debil tendencia de incremento en Ia diversidad bajo intensidades de pastoreo livianas. Sin embargo bajo pastoreo moderado a intense Ia riqueza de especies es mas baja como es tambien el caso cuando incendios ocurren frecuentemente. El cultivo de papas y el pastoreo a las intensidades mas altas esllin relacionados a altas proporciones de especies introducidas (hasta un 30%) y una disrninuci6n en el numero _total de taxa. Un esquema con un resumen de los efectos a largo plazo del manejo de quema y pastoreo en varies aspectos del ecosistema de paramo se presenta en Ia pagina 204.

230

SAMENVATIING Het ecosysteem van het hooggebergte in de noordelijke Andes, de paramo: wor~t in toenemende mate door de mens be'invloed. Ten gevolge van brand en begrazmg vmden geleidelijke ~eranderingen plaats in de oorspronkel!jke horstgrasvegetatie, maar een .snelle transformatie door de teelt van lokale knolgewassen komt ook voor. De paramozo ne IS van vitaal belang voor de menselijke bevolking in de inte:-Andiene valleie~, aangezien deze de waterstroom naar de lager gelegen (dicht bevolkte) ge~teden reguleert. ~mnen de grenzen ~an het Nationale Park Los Nevados in de Centrale Cordillera van Colombia, beheert een kleme bevolkingsgroep van boeren een aan1al ge'isokerd gelegen boerderijen (finc~'s) . .De ~oeren bedrijven veeteelt voor de produktie van vlee.s en een vorm .van landgebrw k dte htemnee samenhangt is het branden van de vegetaue om de groe1 van verse grasscheuten te bevorderen.' Hoewel in het algemeen werd aangenomen dat deze praktijken leiden tot degradatie van het ecosysteem, beschikten dicgenen, ~ie verantwoordelij~ zijn voor h:t be~~er van de pa_ramo, nauwelijks over kennis ten aanzten van de snelhetd en de rutmteltjke uitbreiding van deze processen. In dit proefschrift worden vegetatiepatronen in relatie tot brand..en begrazing ge~odellee.rd. Zowel ruimtelijke, als tijdsaspekten van de door de mens bemvloe.de v~getaned ynarruek worden beschreve n. De doelstellingen omvatten de ontwikkeling van rutmtehjk e_modellen van het behcersregime; van zowel de verspreiding van het vee als de br~dge~~ -hiedenis; de verklaring van de variatie" in de ·huidige vegetaticpatronen; en de voorspelhng van de rcsp~ns van de vegetatie op veranderingen in beheersvaria~elen: ~aast deze :-vetensch appehjke~ . doelstellingen, was de vertaling van de resultaten naar tmphcaues ten aanZien van ?et ~heer een belangrijk doe!. Voor het gebied van Los Nevados is aan het behoud van b101o.gt~ch~ diversiteit en de handhaving van een stabie1 hydrologisch systeem de hoogste pnontett toegekend door de betreffende beheersorganisaties. · Cluster- en ordin~tietechnieken werden toegepast om de variatie in vegetaties truktuur ~n floristische samenste lling in relatie tot de beheersvariabelen te analyseren. De zo~ale vegetatte van de onderzochte paramo proper (van 3900 to 4200 m).wordt gedo~neerd d<>?r soorten van het horstgras Calamagrosris en de stamcompostet . Espelerra harrwegw na_. ssp. centroandina. Op gematigde hellingen krijgt de vegetatie een .open karakter ~IJ een gemiddelde begrazing sintensite it. In deze situatie nemen de bedekkmg, hoogte en .at~eter ~an de horstgrassen af, terwijl hc;t percentage kale ~ond en_ het effek~ van betredmg m ~e vorm van terrein-.treden (terracettes) toeneemt. Op ltcht glo01ende hellmgen en vlak terrem, ontwikkelen zich kone gras- en kruidrnanen. De bel.c~!lgrijkste s~rten me_t een. kone groeivorm zijn Lachemilla orbiculara, Calamagrostis coarctat? en Acrachn: acrcularrs. Een lijst met plantensoorten die een positieve dan wei een negaueve correlaue met b~d- en begrazingsvariabelen vertonen, is opgenomen. Een vegetatiekaart we«) gema~t op basts van luchtfoto's op schalen varierend van 1: 25,000 .tot 1: ~3,000: Deze kaart laat ~e~ dat ±50% van de oorspronkelijke dichte horstgraslanden ts veranderd m meer open vegetaues trukt~n als gevolg van branden, begrazing, of beide..De floristische s~enstelling,yan, .~e vegetatte veranden langzamer, over een tijdsbestek van enkele tientallen Jaren of meer.

231

.., Samenvauing Het graasgedrag van hct vee werd bcstu deerd tijdens veldwaamemingen. Op deze werd de botanische samerrstclling van het dieet van het vc:e bepaald. Voedselvoorkeu r het algemeen gerclateerd aan de kwaliteit van de geconsumeerde plantengroepen in van gemiddelde verteerbaarheid en ruw eiwi tgehalte. Een !age voedselkwaliteit (7 eiwit), een !age produktiviteitsindex (35 kg per jaar per 100 kg lichaamsgewicht voor volwassen koe) en de lange afstanden die het vee loopt om te voorzien ·· ~ voedselbehoeften, komen overeen met de kenmerken van een extensief ' graassysteem. Ondanks het feit dat de Espeletia starncomp osieten bepaalde unieke eigenschappen blijk ~n bezitten die te interpreteren zijn als aanpassin gen aan brand, werden de schadelijke · van het beheer voor aspekten van de populatiedynamiek aangetoond, met name . bij combinatie van branden en intensieve begrazing. Bij volwassen exemplaren werd lengtegroei van 8.8 em per jaar geme ten in ongestoorde situaties. Op afgebrande plekk en een toegenomen groeisnelheid, inclusief een brand-stimuleringsfaktor, waargenomen. Na brand is de mortaliteit hoger, vooral onde r de grotere exemplaren: de mortaliteit ·is sommige gevallen meer dan verdubbeld . Na een aanvankelijk hoge mortalitei t onder juveniele exemplaren als gevolg van brand , wordt dit verlies snel gecompensee~d door een_· verhoogd overlevingspercentage van zaail ingen en een toegenome n juveniele groei . Een' geclusterde verspreiding van individuen werd waargenomen in intensief begraasde situaties, hetgeen bcscherming kan bieden tegen de gevolgen van vertrapping en het schuren van vee · tegen de stammen. Een verhoogde mortalitei t werd eveneens waargenomen. De horstgrassen van het genus Cala magrostis worden gekenmerkt door een !age produktiviteit. Herstel na brand duurt gemiddeld 10 jaar. De veranderingen diameterverdeling en in het percentage in gefragmenteerde horstgrassen, wcerspieg elen de gevolgen van het beheer het beste_De fragm entatie van horstgrassen kan gerelateerd zijn aan brand, begrazing, of een combinatie van beide. De tijdelijke door brand veroorzaak te fragmentatie van horstgrassen in gescheiden , herstellende uitlopers wordt dikwijls in stand gehouden door begrazing. Wanneer de begra zingsintensiteit aanhoudend hoog is, valle n de horstgrasbases geleidelijk uiteen. Deze fragm entatie is het gevolg van betreding en uitpu tting van de reserves door consumptie. De bovenstaande bevindingen werden gebruikt om een simulatiemodel van de vege tatieomwikkeling onder verschillen de beheersstrategieen te ontwik.kelen. Funk tionele plantengroepen werden onderscheiden, welk e overeenkomen met groepen soorten die een vergelijkbare res pons met be trekking tot ecolo gische, en beheersvariabelen vertonen. lterat ieve calibratie op basis van tijdseries van veldg egcvens diende om de modelstru ktuur te verbe teren en parameter-waarden ..te schatten. Het mode l kan gebruikt worden om het verloop van de vegetatieontwikkeling onder verschillen de brand - en graasstrategieen te voorspell en. De uitkomsten zijn consistent met de resul taten van de onderstaand beschreven ruim telijke modellen. Een ruimtelijk model van de verspreiding van het vee ..verd ontwikkeld op basis van 'multiple regression' op de puntwaarnemingen. Een maat voor de beschikbaarheid van voer (de aantrekkelijkh eid van de vegetatie voor begrazing) werd afgeleid van de gewogen som van de funktionele plantengroepen, rekening houdend met voedselkwaliteit en -voorkeur . Deze variabele verklaarde de variatie in begra zingsintensiteit het beste . '

Samenvacting Andere significunt~ voorspellers van . d be razingsintensiteit waren (in volgorde van e g.· ~k r van hct vee en de afstand tot de belangrijkheid): hellingshock, veebezettmrkg, ~erreltn_vb~t ~~artcn en geco'mbineerd in de vorm ·· D . 'abel cn wcrd cn boerdenj. eze van n u · . . verwhe t matom de begrazingsintensiteit te voorspe11en. van een kaart van de verspreldm g van et vee, . d waar branden een belangrijke aktiviteit De brandgeschiedenis van een prodgebl~' is, werd d . met luchtfoto's van 10 verschillende jaren gereconstrueerd via de analyse van ~en IIJ seneGIS gebaseerde procedure werd ontw ikkeld tusse n 1959 en 198?. Een na~w.keunge: o~ een D emiddelde brandfrekwentie voor het ten behoeve van retrospekueve momton~~ .' ~ ;randfrekwemie is niet verminderd sin~~ proefgebied werd gcsch~t op 1.0 tot 1;/l~;:i be:chikbaarheid neigt naar vermindering blJ de stichting van het nauonale park. e 10 . D rechtvaardiging volgens de boeren van brandfrekwenties van meer dan eens I dJaar. de lb schikbaarheid te verhogen, de ·praktijk van het brand:n, met het per is al!een doe e voe se e geldig bij lagere frekwenues. . . le en locale hydrologische Ten behoeve van het be~oud van stable mcn, moet le reglo~lak in stand gehouden wordsyste en. Aannemend . de beschermende bedekkmg van het bode ~op~ ie.d t dient men in acht te nemen dat de dat de horstgraslaag de beste beschermm g . • . fha voor stukken begrazmgsmtensl'tel't (0 ·37 AU horstgrassen degrade~n ~nder de hUl'dige · · . vegetatie die in gebrUik ZIJn voor begrazlhng). De kritische drempel waaronder de horstgrassen k nnen snel verowijnen onder intensieve [ha overleven, 1.s ongevee rO· 16 AU · .: De orstoe rassen. ud bcdekking van begrazing (binnen 50 jaar), ·teTWI de horstgras bases JI een toenarne m e langzaam verloopt. ' van een · d moza'iek-patroo (' pate h'10ess') wordt De heterogeniteit van het landschap m e v~rm I" ht begrazing of een ncomb inatie van beide in stand gehouden door !age brandfrek~en u~s, diC : e de voorkeur hebben van het vee, zoals. De boeren branden doorgaans de terrem-ee~ 1\e~ ~te lava's Branden wordt toegepast op de laterale morenen en de vroeger, met IJ,S . e e korte ~fstand de geringe hoev eelheid gemiddelde afstanden van de ~n~a s, aange zt~n bof 11 Dichtbij de finca 's heeft zich brandbaar materiaal de verspreJdm~ van bran ·~~:e ~nder invloed van hogere . veelvuldig een korte vegetaue-mat ontwl e begrazingsintensiteiten. · · · · biologische diversiteit belangrijk Naast landschapsheterogeniteit, is de eval~ . ~~el~ana:~~~~~: onder de vaatplante n is 38.5 (± ten behoeve van natuurbehoud. Het ge~l .e e akke tendens van toenemende diversiteit 3. 1) voor ongest~rde ~orsq~ra~landen... r IS ~e~ z~t intensieve begrazing echte r, is de bij Iichte begrazmgsmtensltelten. BIJ mat gal . eer branden frekwent voorkomen. .. d I at eveneens het gev IS wann s~ennJ~ o~ e~g:; ~o:gste begrazingsintensiteit en zijn gerelateerd aan een groot aan deeI ~an· :~:tr:uceerde soorten (tot 30%) en een afnam e van het totale aantal taxa. . . enva Een schemausch o~erzlcht met ~-~1 sa: aspettin van de lange-termijn effekten van brand en kt;n van het paramo ecosysteem is gegev begrazing ten aanz1en van verse 1 en en op . (W' u:::O ~l pagma 204. t OL0~-1 .~· . ···•· ,..1•1'\'... \II:. c . .. .

o·

~1~

...

'"~u\ tJI'U!3tl ..

.. •.

.· l

232

233 '

~·~~~.\

~ 1\ .,;§ "

_,,1. l'\f\..Cl

:.··'\li-> _;. '

· ••v'IH 60

.· ... ·

\_·:~ .... ~\·: :-'

.:·'."". ···'··

~ r ·1

tDv"" . 1C>W.I~\Q' i'~-""""'"

Appendix C Table 3.2. TWINSPAN FLORISTIC VEGETATION TABLE of the upper watershed of RĂo OtĂşn, Los Nevados National Park, Colombia. A B C D E F G H I s s 1 s s s s s s s s s s b s b s s s s s s s s s s s b b s b b s s s b b b b b b b b b b b s s s s b b b b b b b b b b b b b b b b b b b b b b b b b b b s s s s s s s s s s s 1 b b b b b b b b b s s s s b s s s s s s s s s s s s s b s s s s s s s s s s s s s s b b b b b s s s s s b b b b b s s b s s s s s s s s s s s b s s s s s b s 0 9 9 9 1 1 2 3 5 8 9 2 8 2 8 8 9 3 5 2 0 7 7 7 7 4 4 1 4 4 0 7 8 1 2 2 4 4 4 4 3 5 5 6 0 1 1 4 0 0 0 0 0 0 0 1 1 1 2 3 3 3 3 4 5 5 5 5 5 5 6 6 6 6 6 2 2 2 2 3 4 4 8 8 8 8 0 0 1 1 1 1 1 2 4 5 0 0 3 5 3 3 3 7 7 7 2 3 3 3 3 4 4 5 1 0 0 1 1 2 2 1 1 1 0 2 4 5 9 2 3 3 5 6 9 9 4 4 6 3 2 2 3 6 4 5 0 4 5 5 9 6 6 6 6 6 6 7 2 5 5 6 6 9 6 7 1 6 7 8 7 8 6 9 6 9 0 0 4 1 5 8 9 5 1 0 7 5 6 7 8 0 1 9 8 9 6 4 0 7 2 3 2 3 5 7 1 0 9 5 2 5 6 3 1 2 3 4 5 6 7 6 8 9 5 2 3 6 9 6 2 3 4 5 7 8 1 2 3 4 7 1 2 3 5 4 0 1 1 2 6 7 0 8 0 1 2 3 4 4 4 1 3 4 3 4 4 2 6 2 3 9 9 0 1 7 8 4 5 7 5 8 9 0 1 4 7 2 3 4 5 8 2 5 1 7 0 7 6 8 2 3 7 8 5 8 6 8 5 0 6 0 9 9 2 3 4 1 2 3 4 6 7 1 9 8 9 0 8 5 6 0 Elevation Terrain Slope Mottling Grazing intensity Fire age Sructure

8 C 4 0 t

9 C 2 0 e

9 C 1 0 e

9 V 0 0 c

5 D 0 0 c

5 D 0 4 0 c

6 V 0 0 c

6 V 0 0 9 c

4 D 0 0 c

5 L 1 2 c

6 F 0 0 c

5 D 0 1 c

5 G 1 5 0 c

5 D 1 6 0 a

5 G 1 6 1 c

5 L 1 5 0 f

5 G 1 5 2 f

4 V 0 2 3 f

5 L 1 4 2 f

5 F 1 5 3 1 o

6 L 2 4 3 d

6 G 3 3 0 3 d

6 G 2 2 1 1 o

6 G 3 4 1 6 d

5 G 2 4 1 1 o

6 L 2 2 0 1 o

6 G 2 2 1 3 h

6 L 0 6 4 2 m

5 L 1 2 2 3 d

5 L 1 2 1 4 m

5 G 4 4 0 5 d

6 G 3 4 0 9 d

7 G 3 2 0 d

5 B 3 2 0 3 d

5 B 2 0 1 o

6 B 2 4 0 6 h

6 G 2 4 0 4 h

5 B 3 2 1 1 o

6 C 2 2 2 3 o

4 L 3 0 1 o

4 L 2 1 1 o

5 L 2 1 7 o

3 B 3 0 3 e

5 L 0 3 m

5 G 4 2 4 o

5 L 3 2 1 5 d

5 L 3 3 1 o

5 L 1 4 3 4 o

4 L 2 4 3 3 o

4 L 2 0 4 d

5 L 2 2 5 h

6 L 1 2 2 3 m

5 L 2 2 m

4 B 3 0 1 o

4 B 3 0 5 h

4 B 1 1 3 d

5 B 3 2 1 d

5 B 3 2 1 o

5 B 3 2 1 2 o

4 L 2 1 h

3 B 2 2 m

4 L 2 1 4 h

4 L 2 0 h

5 L 2 1 1 m

5 L 2 2 3 o

6 L 2 2 2 o

5 L 2 2 1 o

5 L 1 1 5 h

4 L 0 3 3 o

4 L 3 2 5 o

4 L 2 1 1 o

5 L 3 0 2 h

4 B 1 1 3 m

4 B 2 0 2 o

4 L 3 0 3 d

6 B 0 0 5 d

4 B 0 0 1 o

4 B 0 0 8 d

5 B 4 0 3 d

4 L 3 4 1 d

4 G 2 4 2 3 m

4 G 2 2 2 4 m

6 L 2 2 3 6 o

6 L 2 2 2 9 h

6 L 2 2 1 9 m

5 L 4 2 1 m

8 G 4 0 m

4 B 0 2 8 m

4 B 4 4 o

4 B 3 2 1 9 h

4 B 0 3 5 h

4 B 0 2 2 3 m

4 B 1 3 5 m

4 B 1 2 5 h

5 B 3 0 4 d

4 B 3 3 8 o

4 L 3 0 d

4 L 2 2 o

4 L 3 4 2 1 o

4 L 2 0 o

3 L 2 2 m

6 G 0 4 g

5 L 0 4 g

6 G 0 3 2 m

6 G 0 3 2 g

6 G 0 5 2 g

6 L 0 4 m

6 G 0 4 g

6 L 0 4 9 m

5 G 1 2 2 s

5 G 1 3 2 9 s

5 L 2 4 2 m

5 L 1 2 2 m

6 L 2 1 9 m

4 B 0 3 3 6 m

4 L 4 4 1 a

5 L 2 5 0 m

4 L 2 4 3 a

5 L 0 4 2 a

5 L 3 3 9 a

6 L 2 4 2 a

5 L 0 5 0 g

5 L 0 5 0 s

5 L 0 5 2 g

4 L 2 2 3 s

5 G 1 2 4 m

4 G 0 4 4 m

3 G 2 4 1 s

4 C 1 4 3 s

3 L 1 2 s

3 L 2 2 0 m

3 B 0 1 s

4 B 0 4 9 s

4 L 2 2 s

4 L 2 2 2 f

4 L 2 2 3 s

3 D 2 4 0 a

3 G 2 2 2 s

1 E 2 2 2 t

4 L 3 0 s

3 L 3 4 t

3 L 1 2 0 s

4 B 3 1 g

3 L 3 0 s

4 G 2 0 s

3 L 0 2 4 s

4 V 0 4 s

4 C 3 2 0 f

5 G 0 2 0 t

5 G 0 2 3 t

4 C 2 2 f

2 E 0 4 3 t

2 C 3 2 2 t

2 C 3 3 3 f

2 C 0 3 3 t

1 E 1 3 3 t

1 E 0 0 f

2 G 1 3 3 t

3 L 1 0 e

4 F 0 4 4 s

4 F 0 2 2 s

2 G 2 4 4 t

1 B 3 1 1 t

4 F 0 3 3 s

5 L 0 3 3 e

1 F 0 0 f

56 71 72 122 17 132 59 188 190 9 40 86 216 131 78

Cerastium floccosum Draba hallii Draba pachythyrsa Lachemilla holosericea Arenaria spec. 2 Lupinus alopecuroides Colobanthus quitensis var.2 Senecio canescens Senecio latiflorus Agrostis foliata Calamagrostis ligulata Festuca dolichophylla Werneria humilis Lucilia kunthiana Erigeron chionophilus

3 2 3 2 -

2 2 1 1 1 1 3 2 2 2 2 1 1 2

2 2 3 1 1 2 2 1 1 2 1 2

3 1 1 2 1 1

1 -

4 -

1 -

2 1 1 2

1 -

2 -

167 70 155 8 151 43

Plantago rigida Distichia muscoides Oritrophium limnophilum Agrostis breviculmis Niphogeton lingula Calamagrostis spec.

5 2 2 1

2 5 1 2 -

5 1 1

7 3 2 1

6 1 1

4 2

7 1 -

6 1 1

7 1 -

6 -

4 1

1 -

3 1

1 -

145 210 160 38

Muhlenbergia cleefii Valeriana plantaginea Pedicularis incurva Calamagrostis intermedia

2 -

1 -

2 -

2 1 1

2 1 3

1 -

2 2 -

2 -

1 -

3 -

1 -

69 149 215 175 161 66 41 63 120 173 14

Disterigma empetrifolium Nertera granadensis Werneria crassa Ranunculus peruvianus Pentacalia vaccinioides Diplostephium schultzii Calamagrostis macrophylla Cotula mexicana Lachemilla galioides Potentilla heterosepala Aphanactis piloselloides

2 -

1 -

3 -

1 2 2 1 1 -

4 3 2 -

1 3 3 1 1 -

2 2 3 1 2 -

4 1 1 2 3 -

4 1 4 1 -

3 1 1 2 2 2 -

3 1 2 1 3 1 1 -

2 1 1 3 1 -

2 2 1 1

1 2 3 1 -

1 1 2 2 2 -

1 2 3 2 1

2 1 1 1 1

1 -

1 1 -

1 1

1 -

2 -

1 -

2 -

1 -

2 -

1 -

2 -

1 -

1 1

1 1 1 -

1 1 1

1 -

1 1 -

1 -

1 1 -

2 1 -

2 1

1 -

1 1 -

1 -

1 1 -

2 -

1 1 1 -

1 -

1 1 -

1 -

1 1 1 1 -

1 1 2 -

1 1

1 -

1 2 -

1 1 -

1 -

1 2 -

61 62 144

Cortaderia bifida Cortaderia nitida Muehlenbeckia vulcanica

- - - - - - - - - - 1 - - - - - - - - - - 3 2 2 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 2 - - - - - - - - - 2 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 1 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 2 - - - - - - - - - - - - - - - - - - 7 1 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 2 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 3 3 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 2 - 2 - - - - - - - - - - - - - - 1 - 1 - - - - - - - - - - - - - - - - - - - - - - - - - - - -

18 135 111 104 49 52 121

Azorella aretioides Lupinus tolimensis Huperzia cruenta Halenia spec. Carex bonplandii Carex tristicha Lachemilla hispidula

3 -

2 3 -

1 2 1 4 -

1 1 1 2 4 -

1 2 1

2 1 2 1 5 2

4 1 1

1 1 1 1 3 2

1 2 1 3 1

4 1 1

2 2 2

1 2 1

1 1 2 1

8 -

1 1 1 1 6 2

2 1 7 1

6 2 1

1 5 -

2 2 1 3 4 2

2 2 1

2 2 2 3 -

2 3 1 1 2 3 2

2 1 1 1 2 3 1

1 4 1

2 1 2 5 1

2 1 1 3 1

1 1 2 3 1

1 2 2 1

1 1 1

1 1 1 2

2 1

2 1 -

1 -

2 2

2 1

2 -

1 -

1 3 -

2 1 3 -

1 1

1 -

1 1

1 1 1

1 -

2 1 1 1

1 -

1 1

1 -

1 1 -

1 -

1 2 -

1 -

1 1 -

2 -

1 -

2 1 -

2 2

1 -

2 -

1 -

1 1 -

1 -

1 1 1

2 1 -

1 1 -

2 1 2 -

1 2 -

1 -

1 1 1

1 -

2 2 -

1 -

1 2 -

2 -

1 -

1 1 1

2 1

2 1 -

2 -

1 -

2 -

1 -

34 114 28 92 21

Calamagrostis recta Hypericum lancioides Bartsia pedicularoides Gentiana sedifolia Baccharis caespitosa

3 -

2 1 -

2 2 -

2 3 1 -

4 1 1 1

2 3 1 -

3 3 1 1 1

3 2 2 1 -

3 1 1 2

2 1 1 2

2 1 1 -

2 3 2

1 4 1 1 -

3 3 1 1

3 1 1 -

2 2 2 -

2 -

2 2 1

1 2

1 2 2 1 1

1 2 -

1 -

2 1 1 1 1

2 1 -

1 -

1 1 2 1 -

2 1 2 1

1 1 -

1 2 1 1 1

1 2 1

2 1 2 2

3 3 2 1

3 2 1 1

2 2 1 1

2 2 1 -

1 1 -

1 -

1 2 1

4 1 -

2 -

1 -

2 1 -

1 -

3 -

3 1 -

2 1 1 -

3 1 1 1 -

2 1 2

4 1 2

3 1 1

1 1 1

2 -

3 1

2 -

2 2 1 -

2 1 1 -

2 1 -

1 -

3 1 1

2 1 1 2

1 1 -

2 1 1 -

1 -

2 1 -

1 1 -

2 -

1 2 -

1 -

3 1 -

3 1 1

2 2 2 1 -

1 2 1 -

3 1 1

1 1

3 1 1 1

3 1 2 1

2 1 1 -

1 -

2 1 1 -

1 2 2 -

4 -

1 1 1 -

2 1 2 1 1

3 2 -

2 -

3 1

2 1 1 3

3 -

1 -

1 1 -

1 1 1

1 1 -

3 1 1

2 1 -

1 -

2 1 1

1 1 1

1 1

1 2 1 1 1

2 1 1 1

4 2 1

1 1 1

2 2 1

2 1 -

3 1 -

2 2 1 -

3 2 -

1 1

1 -

2 1

1 1

1 -

1 1 -

1 -

2 1 -

1 -

2 1

2 -

192 79 189 117 152 95

Senecio repens Erigeron ecuadoriensis Senecio formosus Jamesonia goudotii Ophioglossum crotalophoroides Geranium multipartitum

2 -

1 -

2 1 -

1 1 -

1 -

2 -

2 1 1 -

1 2 2 -

2 1 1

1 1 -

1 1 1 1 -

2 1

1 1 1 -

1 1 -

1 -

1 1

1 1 1 -

2 2 2 1 -

1 1 -

1 1 1 2 1 -

2 2 2 1 -

1 1 1 2

1 1 1 -

1 1 2

2 1 1 -

1 -

1 1 -

2 -

1 1 -

2 -

1 1 -

1 1 1 -

1 1 -

1 1 1 -

1 1 1 1 1 -

1 -

1 2 1 -

1 1 -

1 1 1 -

1 1 -

1 -

1 1 -

2 1 -

2 1 1 -

1 1 1 -

1 1 -

1 -

1 1 1 -

1 2 -

1 1 -

1 -

2 2 -

1 -

1 1 1 -

1 1 -

1 1 1 1

1 1 1 1 1

1 1 1 1

1 1 1 1 1

1 1 1 1 -

1 -

1 1 -

1 -

1 1 1 -

1 1 2 2 -

1 -

1 2 -

1 1 1 -

1 -

1 1 -

1 1 2

1 1

2 1 -

2 2

2 1

1 2

1 1 1

2 1 -

1 1 1 -

1 -

1 1 1 1

2 2 2 1 1

1 1 -

2 -

1 -

2 -

1 -

1 2 -

1 1

1 1 -

1 1

1 -

1 1 -

1 -

2 1 2 1 -

1 -

208 93 32 124

Uncinia meridensis Gentianella dasyantha Bromus lanatus Lachemilla mandoniana

5 -

1 3 -

2 2

2 -

2 1

2 2 1 2

1 1 -

2 -

2 1 2

4 1 1 2

6 2 1

3 1 1 2

2 1 1 1

4 2 2 2

2 1 2

1 2

1 1 -

2 -

2 1 -

2 -

1 -

1 1

1 -

1 2

1 1

2 -

1 2 -

1 -

2 -

1 -

2 1 -

2 -

1 1 -

1 -

1 1 1

1 2 2 -

2 1

2 1 1

1 1 -

1 -

2 2 -

1 -

2 -

1 -

1 2 -

2 -

1 1 1

1 -

2 -

1 1 -

2 1 -

2 -

1 1

2 1 -

1 -

1 2 1 -

1 -

2 -

1 1 -

1 -

2 -

1 -

1 3

187 57 154 176

Satureja nubigena Cerastium subspicatum Oreomyrrhis andicola Ranunculus praemorsus

1 -

1 2 2

1 1

1 -

1 1

1 1 1 1

2 1

1 1 1 -

1 1 1

2 -

1 1 -

1 -

1 1

1 -

1 2 -

1 1 -

1 -

1 2

1 1 -

1 -

1 1

1 -

2 1 2

2 1

1 2 1

1 -

1 1

1 -

2 -

1 -

2 -

1 -

3 1 1 -

2 1 1

2 1 1 -

2 1 1

2 1 -

2 1 1

2 1 -

1 1 -

2 1

2 -

1 1 -

2 1 1 1

1 1 -

2 1 1

1 -

2 -

2 2 1

1 1 1 1

3 2 1

2 1 1

2 1 1 1

1 1 1 -

2 1 1

1 1 1

1 1

2 1 1

1 1 1

1 1 1 1

2 1 1 1

1 1 1 1

3 1 1

1 1 1

1 1 1 -

1 -

2 1 1 -

1 1 -

1 -

2 1 -

1 1 -

1 -

2 1 2 1

1 1 1

1 1 1 1

1 1

1 2 1 2

1 1 -

2 2 -

1 2 -

1 2 1

3 2 -

5 1 2

1 1 1 1

1 1 1

1 2

1 1 1 -

1 2

2 1 1 1

1 1 2

1 2 1

2 1 -

1 -

10 84 20

Agrostis haenkeana Festuca andicola Azorella multifida

- - - - - - - - 1 - - 2 3 - 2 1 3 - 1 - - - - - - - - - 1 - - - - - - - - - - - 2 2 - 1 - - - - - - - - - - - - - - - - 1 - - - - - - - - - - - - - - - - - - - - - 1 - 3 - - 2 - - - - - - - - - 3 - - 2 - - - - 2 - - 1 - - - - - - - - - - - - - - - 3 - 2 3 1 3 3 - - - 3 2 2 1 - - 2 3 - 2 2 2 - 2 2 2 - - - - - - - - 2 - - - - 2 - 3 - - - - - - - - - 3 - - - - 2 1 - - - - - - - - - - - - 1 - - - - - - - - - - 1 - - - - - - - - - - - - - - - - - - 1 2 - - - - - - - - 1 - 2 - - - - - - - - - - - - - - 3 - 1 - 2 - 2 - - 1 - - - - 2 - - - - 2 - - - - - - - - - 1 - - - - - - - - - - - 4 2 4 2 2 4 3 - - - - - 2 - 2 1 1 - - 2 - - - - - - - - - - - 1 - - - - - - - - - - - - - - - 1 - 1 - 1 - - - - 1 - - - - - - - - - 1 - - - - - - - - - - - 1 - 2 - - - - - 1 - - 1 - - - - 1 1 - - - - - - - - - 1 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 1 - - 1 1 - - - - 3 - 1 - 1 - - - - - - - - - - - - - - - - - - - - - - - - -

Festuca sublimis

- - - - 5 4 4 4 4 4 - 2 2 3 2 2 2 4 - - 1 - - 1 - - 1 - 3 3 1 3 - 5 - - 1 - - - 4 2 1 2 - 5 - 3 3 3 2 2 3 2 3 3 3 4 - 3 3 3 1 - 1 1 - 3 2 3 1 3 2 3 3 - - 2 4 - 1 2 - - 3 3 - - 3 - 3 - - 4 - 1 7 4 - 1 4 - - 3 - - - - - - - - - - - - - - - - - - - - - 4 5 1 - 1 2 3 - - - - 2 1 - - - - - - - - 1 - - - - - - - - - - - - - - - - - - -

35 80 116 27 197 164 207 51

Calamagrostis coarctata Eryngium humile Hypochoeris sessiliflora Baccharis tricuneata Sisyrinchium trinerve Pernettya prostrata Trisetum spicatum Carex pichinchensis

2 3 -

156 26

Oritrophium peruvianum Baccharis rupicola

- - - - - - - - - - - - - - - - 1 - - 2 2 2 2 2 1 2 2 1 1 1 - - 1 - 1 - 1 - 1 - - - - 1 - - 1 1 1 2 1 1 1 1 1 - - 1 1 - - 1 - - 1 2 2 - - 1 1 1 - - - - - - - - - - 1 - - - 1 - - - - - - - - - - - - - - - - - - - 1 1 2 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 1 1 1 1 - 1 - 1 1 - - 1 1 - 1 - 1 - - 2 - 2 1 1 1 1 1 - 2 1 1 2 2 2 2 2 2 2 2 - 1 1 2 1 1 1 - 1 2 1 - 1 2 2 1 - - - - - - - 1 - - - - - - - 1 2 - - 1 - - - - - - 2 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 2 - 1 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

36 82 83 140 53 115 12 96

Calamagrostis effusa Espeletia hartwegiana Espeletia hartwegiana juv. Lycopodium clavatum Castilleja pumila Hypericum laricifolium Aa colombiana Geranium sibbaldioides

2 -

1 -

1 1 1

1 -

1 1 2 1 -

4 -

1 1

1 2

4 3 1 1 -

8 3 2 1 1 1

8 3 1 2 1 -

5 2 2 -

8 2 2 2 1 -

5 3 1 1 -

4 2 1 -

6 1 1 1 -

5 1 1 1 2

7 1 1 1 1 -

5 1 1 2 1 -

9 3 1 3

8 3 1 1 1 1 1 -

7 3 2 1 2

7 1 1 1 2 -

3 2 1 1 -

7 1 2 1 -

6 2 1

5 2 1 1 -

4 3 1 1 1

2 1 1 1 -

3 3 1 1 1 -

2 1 2 -

1 2 1 -

3 1 1 1 2

4 3 2 1 -

7 3 1 1 1 1

7 2 2 1 -

3 2 2 2 -

5 2 2 2 -

8 3 2 2 1 1 -

5 2 1 1 -

3 1 1 1

5 2 1 1 1 -

5 2 1 2 1 2

5 2 1 2 2 -

7 2 2 2 1 1 1 -

5 1 1 1 1 1 1 -

5 2 1 2 1 1 1 -

4 2 1 1 1 2

5 1 2 1 1 -

3 3 2 2 1 1 -

5 2 2 1 -

6 1 1 -

4 2 1 1 1 -

5 3 1 1 1

3 2 1 1 -

4 2 1 -

5 3 1 1 1

4 3 1 2 1 1 -

4 2 1 -

4 1 1 1 1 1 1 -

6 1 1 2 2 1 -

5 2 1 1 2 -

4 2 2 2 1 1 -

9 1 1 1 2 -

8 3 1 -

3 2 -

8 1 1 1

6 1 2 2 1 -

8 -

5 1 2 1 1 1

6 1 1 1 1

3 2 2 2

6 4 1 1 1 1 -

4 2 2 1 1 1

5 2 1 -

3 2 2 1 1

3 2 2 -

2 3 1 1 1

7 1 1 2 1 2 -

6 3 2 1 1 -

4 2 -

5 4 2 2 -

5 2 2 2 1 -

7 2 1 1 1 -

2 1 1 -

3 -

2 1 -

4 1 1 1 -

5 1 -

2 2 -

4 3 1

2 2 1

2 2 -

5 1 1 1

1 4 2 1 -

7 1

1 2 2 1 1 -

3 1 1 1 1

3 1 -

6 2 2 1

1 4 1 3

4 2 1 -

4 1 1 1 -

5 -

3 1 1 -

3 1

1 -

1 1 1 -

3 1 -

2 1 2

1 1 -

1 1

4 2 1 1 -

1 4 2 2 1 1 -

2 2 1 -

1 1 1 -

1 1 1

1 -

1 3 -

2 1

2 -

1 -

1 1 1 -

1 -

133 108 97 29 42 30 100 222

Lupinus microphyllus Hieracium avilae Gnaphalium antennarioides Belloa longifolia Calamagrostis planifolia Bidens triplinervia Achyrocline alata Lachemilla tanacetifolia

1 -

1 1

1 -

1 2 -

1 1

1 2 1

1 1 1 2 -

1 2 1

1 2 2 1 1 -

1 2 1 -

1 1 1 -

2 1 2 -

1 1 1 1

1 2 1

3 1 2 1 2 1 2

1 1 1 2 1 1

1 1 2 2 1 1

1 1

2 2 2 2 1 1 1 1

1 1 1

1 1 1 1 1 1

2 1 1 -

1 2 1 2 1

1 1 1 1

2 2 1 -

1 1 1 2 2 2

1 1 1 1

1 2 1 2 1 1 2

1 1 1 2 1 1 3

1 1 1 1 1 1

1 1 1 1 1 1 1

1 2 1 1 -

2 2 2 2 1 2 1

1 2 2 1 1 -

1 1 2 1 1 2 2

2 2 2 2 2 2 1 2

2 2 2 1 1 2 1 1

2 2 2 1 2 2

3 2 1 1 2 1 2

2 2 1 2 2 1 2 1

1 1 1 1 2 1 1 2

2 2 2 1 1 1 1

2 1 1 1 1 1 1

2 2 1 2 1 1

2 2 1 2 1 2 2

2 2 2 2 1 1 1 2

2 2 1 1 1

2 2 1 1 2 2 1

2 2 2 2 1 1 1 1

2 1 2 1 1 2 1 1

1 2 1 1 1 2 1

1 1 1 2 2 2 2

1 1 2 2 1 2

1 1 2 1 1 2

1 1 1 1 1 1

3 1 1 2 1 1 1

4 1 1 1 1 2 1 2

1 1 2 1 2 2 1 1

1 2 1 1 2 2 1 1

3 2 1 1 1 1 1

2 2 2 1 2 1 1 1

2 2 2 1 1 1

1 1 1 -

1 2 1 -

2 2 1 -

1 2 2 -

3 2 1 1 2 -

3 2 1 2 1 1 1

3 2 2 1 1 1 2

3 1 1 2 1 2 1

2 2 1 1 1

1 1 2 1 1

1 1 1 1 2 1

2 2 1

3 1 2 1 2 2 1

2 1 1 1 3 1 1

2 1 1 2 1 1

3 1 1 2 1 1 1

4 1 2 2 1 1

2 2 2 1 1 2 1

4 2 1 1 1 2 1

2 1 1 2 1 1 -

2 2 2 2 -

1 2 1 2

3 1 2 2 1 2

1 2 1 1

2 2 2 1 2 1 -

2 1 1

1 1 -

1 -

2 1

2 3 1

2 1 1 1

3 1 1 -

1 2 1

2 1 1 1 2 1

3 1 1 2 2 1

3 1 1 2 1 1

3 1 1 1 1 1

1 1 1 1 1 2

2 1 1 1 2 1 1

2 2 2 1 -

2 2 2 -

2 1 1 -

1 -

2 1 1 -

1 -

2 1 2 2 1

4 1 2 2 2

1 2 1 -

2 1 1 2 1 1

1 1 -

1 1 1 2 1

2 2 1 1 1 1 1

5 1 1 1 -

1 1 1 -

6 2 1 1 2 1 1

2 1 2 3

1 1 1 1 1

1 -

2 1 2 2 2 -

3 1 1 1 1 -

1 1 1 2 1 1

3 1 1 1 -

1 1 1 2 1 -

1 2 1

1 1 1

2 -

1 1 1

2 -

1 -

1 1 -

1 1

1 -

1 2 -

137 99 15 147 134 16 74

Luzula racemosa Gnaphalium graveolens cf. Arenaria serpens Myrrhidendron glaucescens Lupinus sp. Arenaria spec.1 Elaphoglossum mathewsii

2 -

2 1

1 -

2 -

1 -

1 1 -

1 -

1 1 -

2 1 1

1 1 1 1

1 -

1 1 -

1 -

2 1

1 1 -

1 -

1 1 -

2 2 -

2 1 -

1 2 -

1 -

1 1 1 -

1 1 -

1 -

1 1 -

1 2 -

1 2 1 -

1 -

1 1 -

1 -

2 -

Escallonia myrtilloides Baccharis genistelloides Baccharis caesp./genist.

- - - - - - - 1 - 1 - 1 - 2 - 1 1 - - - - - - - - - - - - - - 1 - - - - - - - - 1 - - - - - - - - - - - - - - 1 - - - - - - - - - - - - 1 - - - 2 - - 3 2 1 1 - - - - - 1 1 - 2 1 2 2 - 1 1 2 2 - - - - 1 - - - - - - - - 2 1 3 3 3 3 3 2 4 5 6 4 1 - - 1 - - - 3 - - 1 - - - - 2 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 1 - - - - - 1 - - - - - - - - - - 1 - - - - - - - - - - - 1 1 - - - - 2 - - 1 - - - 1 1 - - 1 1 - 2 1 - 1 1 - 1 2 1 1 1 1 - 2 2 - 2 2 2 2 2 2 2 2 - 2 1 2 2 3 1 - 1 3 2 3 2 - 2 - - - - - 1 - 1 1 1 2 1 2 1 1 1 2 - 2 1 1 - - 3 2 1 1 2 1 2 1 - 1 1 1 2 1 - - - 1 1 - - - - - - - - - - - - - - 1 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 1 - - - - - - - - - 1 - - - - - 1 - - - - - - - 1 1 - - 1 1 - 1 1 - - 1 - - - - - - 1 - - - 2 - - - - 2 - 1 1 - 1 - - - - - - - - - - - - - - - 1 - - - - - - - - - - - 1 - - - - - - - - 3 - - - 1 1 1 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

4 170 102 85

Aciachne acicularis Poa subspicata Gnaphalium sp. Festuca breviaristata

2 -

1 1

2 -

1 -

4 -

1 1 -

1 -

3 -

3 1 -

1 -

1 1 -

4 1 -

1 -

4 1 -

6 1 -

4 -

1 -

3 -

1 -

5 1 -

4 -

4 1 1 -

4 -

1 -

8 1 -

4 6

4 3

4 6

6 -

1 -

3 -

2 -

1 -

2 -

1 -

166 180 148 107 221

Plantago linearis Galium corymbosum Nasella pubiflora Hesperomeles heterophylla Poa spec.

1 -

2 1 -

1 -

2 -

1 -

2 1 -

2 2 2 -

1 1

2 -

1 -

2 -

1 -

1 1 -

1 -

4 -

1 -

3 -

2 -

1 -

2 1 -

1 -

1 1 -

1 -

1 1 -

2 -

1 -

2 2 1 -

2 1 1 -

1 1 1 -

1 -

1 2 2

1 -

2 -

2 1 2 1 2

3 1 1 1 1

2 1 1 -

2 1 3 -

2 2 -

1 -

1 1 -

1 -

2 -

11 98 119 60 150 179

Agrostis tolucensis Gamochaeta purpurea Lachemilla andina Conyza uliginosa Niphogeton dissecta Galium hypocarpium

1 -

2 -

1 -

1 1 1 -

1 -

2 1 1 -

1 2 1 1

2 -

1 1 -

1 -

2 1 -

1 1 1 1

2 1 1 1 -

2 1 2 1 1 -

2 2 1 1 1 -

1 1 1 1 -

1 1 1 1 1

2 1 -

2 1 1 -

1 1 1 -

2 1 1 1 -

2 2 1 1 -

1 1 -

2 2 1 1 1 -

2 2 1 1 -

2 2 3 1 2 1

2 1 1 1

2 1 1 -

1 1 1 1 1 -

3 1 3 1

2 1 1 -

3 2 2 1 -

2 1 1 1 -

2 2 2 1 -

2 2 2 1 1 -

2 2 1 1 -

2 1 2 1 -

2 1 2 2 1 -

1 1 2 -

2 2 1 2 -

2 3 2 1 1 -

3 2 1 1 -

2 1 1 2 1 1

2 2 1 1 -

2 1 2 1 -

2 2 1 -

3 1 2 1 1 1

2 2 2 1 -

3 1 2 1 -

3 2 2 -

2 2 1 1 -

1 1 -

5 2 2 1 -

3 1 2 1 -

3 1 1 1 1 -

2 1 2 1 1 -

3 1 2 2 -

1 1 1

1 1 1 1

1 1 -

1 1 2 1 1 -

1 1 1 1 -

2 1 2 1 1

1 2 2 1 -

1 2 1 1

1 2 1 1 -

1 1 2 1 -

1 1 1 1 1

3 2 2 2 2 1

4 -

2 2 1 -

2 2 1 1 1

1 2 1 1 -

1 1 1 -

2 1 1 1 1

2 2 1 1 1 -

1 2 2 1 1 -

2 1 2 1 1

2 2 3 1 1

2 2 2 1 1

2 2 1 1

1 1 2 -

3 1 1 1 1 1

1 1

2 -

1 1 -

2 1 1 -

1 2 1 -

1 1 -

2 1 1 -

1 1 1 -

2 1 1

1 2 2 -

1 1 1 1 1 -

1 -

2 1 1 1

2 1 -

1 1

2 1

2 1 1 -

1 1

1 -

2 1 1 -

2 2 1 1 -

1 -

1 1 1 1 1 -

1 1 1 1

2 2 1 2

2 1 1 1 -

1 1 -

2 1 1 1 1 -

3 2 2 1 1

3 2 1 1 -

1 -

1 1 -

1 1

2 2 1 1 -

2 1 1 1

2 1 1 2 -

1 1 2 1

1 1 1 1

1 1 1 1 -

3 -

1 1 2 -

1 1 1 -

3 2 1 2 1 -

2 1 1

2 1 1 1 -

1 1 1 1 -

1 1 1 -

2 1 -

1 1 1 1

2 2 1 1

1 2 2 1 2

2 2 -

1 1

1 2 1 1

2 -

1 -

125 185 94

Lachemilla orbiculata Rumex acetosella Geranium colombianum

- - - - - - - - - - - - 1 - 1 5 - 2 - - - - - - - - - 3 1 1 - - - - - - - - - - 2 - 4 4 - - 1 - - - - 2 - - - 1 - - - 2 1 1 1 - - - - - 1 3 - - - - - - 1 1 - - 3 1 1 - 1 1 - 3 1 - - 3 - 1 - - 5 5 6 3 6 2 5 - 4 4 4 3 3 3 8 5 3 4 7 5 4 3 8 9 3 6 8 5 5 6 8 8 9 8 6 8 9 8 3 6 7 8 5 6 7 8 7 9 5 6 9 4 2 3 1 3 - 3 4 4 2 3 4 9 9 6 5 5 5 3 2 - - - - - - - - - - - - - - - - 1 1 1 - - - - - - - 2 1 1 - - - - - - - - 2 - 2 2 2 2 2 1 3 2 2 1 2 2 2 2 1 1 - 1 - 1 1 1 1 2 2 1 2 1 1 2 2 2 1 1 1 - 2 - - 2 2 2 2 1 1 1 - 2 2 - 2 2 1 2 - 1 - 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 1 2 2 2 2 1 1 2 1 2 1 2 2 2 2 1 2 2 2 1 1 5 2 1 2 1 2 2 1 2 2 3 2 1 6 4 4 8 4 6 5 3 4 5 2 3 3 - 2 1 - 1 - - - - - - - - - 1 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 1 - - - 2 - 1 1 - - - 1 - - - - - - - 2 - 1 1 - - - - - - - - - - 1 - 1 - - 2 - 2 - - - - - 2 2 - 1 2 1 1 - 3 - - 3 - 2 - - - - - - - - - 1 1 - - 1 1 - 1 - 2 - - - - - 1 - 1 1 1 2 2 1 2 - 1 2 2 1 2 2 1 1 2 - 1 - - - - 1 2 2 1 2 2 2 3 1 - - 2 2 - - -

202 212 55 58

Taraxacum cf. officinale Veronica serpyllifolia Cerastium arvense Colobanthus quitensis var.1

2 -

1 -

2 -

1 -

1 1 1

1 -

2 -

1 -

2 -

1 -

1 1 -

1 -

1 1 -

1 -

1 1 1

1 1 -

1 -

2 -

1 1 1

1 2 1

1 -

1 1 -

1 -

2 1 -

1 2 1 2

1 1 -

2 1 -

1 1 2 2

1 1 -

1 2 -

1 1 1 -

1 2 -

1 -

1 1 2 -

2 1 -

1 1 -

2 1 -

2 1 1 1

2 1 1 -

1 2 2 -

1 1 2 -

1 2 -

1 1 2 -

1 2 2 -

2 1 -

1 1 1 1

2 1 -

2 2 2 1

1 2 1 -

1 2 1

1 -

2 2 1 -

2 2 2 -

2 2 1 -

2 2 -

2 1 3 -

1 1 2 -

1 1 -

2 -

1 1 -

2 1 1 -

2 1 2 -

1 -

2 -

1 2 -

206 109 218 19 200 112

Trisetum irazuense Hieracium tolimense Grammitis moniliformis Azorella crenata Stachys elliptica Hydrocotyle bonplandii

1 -

1 1

1 -

1 1 -

2 1 -

1 1 -

1 -

1 1 1 -

2 -

1 -

2 1 -

2 1 1

2 -

1 -

1 2 -

1 -

1 1

1 1 -

1 -

2 -

1 -

3 -

2 1 1

2 -

1 -

2 -

2 13 169 204

Acaena ovalifolia Anthoxanthum odoratum Poa pratensis Trifolium cf. amabile

2 1 -

1 2 1

1 -

2 -

1 -

2 -

1 -

1 1

2 1 3

1 1 -

1 -

2 1

1 4 1

2 -

3 -

4 1 -

7 1

3 2 4

3 7 3

2 1 2 2

2 2

3 2 1

2 2 5

3 5

2 2 -

4 3

5 3

2 1

3 5 3

1 2 -

2 2 2

8 2 2

1 8 1 1

4 1 3

8 3

1 1 -

1 6 2 4

1 5 2 -

4 -

1 -

168 118 64 113 129 165 110

Poa annua Juncus bufonius Dactylis glomerata Hydrocotyle sp. Lolium perenne Plantago australis Holcus lanatus

1 -

2 -

1 -

2 -

1 -

2 -

1 1 -

1 -

2 -

1 1 -

2 2 -

1 1 -

1 -

2 2 -

2 2 1

1 -

1 1

3 2 4 -

1 3 6 -

3 2 2 4 -

1 2 3 1 -

2 -

1 2 2 -

2 2 4 2 2 -

2 2 -

3 2 -

2 2 -

3 2 1 1 -

2 3 -

4 3 -

1 2 2 -

220 7 91 128 199 31 194 213 195 105

Lachemilla spec. Agrostis araucana Gaultheria spec. Lolium multiflorum Solanum tuberosum Bomarea spec. Sherardia arvensis Viola spec. Tragopogon spec. Halimolobos hispidula

2 -

1 -

1 6 -

2 1 -

1 2 1 -

1 3 4 -

2 2 2 -

2 1 -

1 1 2 1 -

1 1 -

2 1 1 -

3 2 1 1 -

2 -

1 -

1 1 1 1

2 1 -

196 123 223 174 157 127 25 77

Sisyrinchium convolutum Lachemilla killipii Sibthorpia repens Ranunculus flagelliformis Oxalis spec.1 Lachemilla pectinata Baccharis revoluta Epilobium denticulatum

3 -

2 -

1 -

2 -

1 -

2 2 1

1 -

2 -

1 2

39 75 139 54 143 191 50 171 153 193 209 224 68 182 47 158 201 130 159 88 106 3 183 87 186 203 214 33 23 141 142 162 184 198 5 48 76 172 6

Calamagrostis jamesonii Elatine ecuadoriensis Luzula vulcanica cf. Castilleja fiss. var. pygmea Montia meridensis Pentacalia reissiana Carex microglochin Poa trivialis Oreobolus cleefii cf. Senecio sp. Uncinia spec. Agropyron attenuatum Diplostephium sp. Ribes leptostachum Cardamine spec.1 Oxalis spec.2 Stellaria cuspidata Lolium temulentum cf. Paspalum bonplandianum Festuca rubra cf. Hesperomeles lanuginosa Saracha quitensis Ribes spec. Festuca procera Rumex crispus Tibouchina spec. Weinmannia spec. Buddleja sp. Baccharis caldasiana Miconia salicifolia Monnina spec. Peperomia hartwegiana Rubus sp. Solanum colombianum cf. Carex spec. Cardamine spec. 2 Eleocharis stenocarpa Polypogon elongatus Agrostis alba cf.

2 2 2 1 1 -

1 -

1 1 -

1 1 1 -

2 1 -

4 -

1 1 1 -

4 -

1 -

1 3 3 1 1 1 1 1 1 -

3 -

3 3 2 1

0 0 . . . . .

0 1 0 0 . . .

0 1 0 1 0 0 .

0 1 0 1 0 1 0

0 1 0 1 0 1 1

0 1 0 1 1 0 .

0 1 0 1 1 1 .

0 1 1 0 0 0 0

0 1 1 0 0 0 1

0 1 1 0 0 1 0

0 1 1 0 0 1 1

0 1 1 0 1 0 0

0 1 1 0 1 0 1

0 1 1 0 1 1 0

0 1 1 0 1 1 1

0 1 1 1 0 0 0

0 1 1 1 0 0 1

0 1 1 1 0 1 0

0 1 1 1 0 1 1

0 1 1 1 1 0 0

0 1 1 1 1 0 1

0 1 1 1 1 1 .

1 0 0 0 . . .

1 0 0 1 0 0 .

1 0 0 1 0 1 .

1 0 0 1 1 . .

1 0 1 0 0 0 .

1 0 1 0 0 1 0

1 0 1 0 0 1 1

1 0 1 0 1 . .

1 0 1 1 . . .

1 1 . . . . .

81 24 22

1 2 -

4 1 -

1 1 1 2

2 1 1 1 -

3 1 -

2 1 1 1 1 -

1 1 2 2 1 1

1 2 1 1 1 1

1 2 2 1 2 -

2 1 1 -

2 1 1 1 1 1 2

2 2 1 2 1 1

3 1 1 2 1 -

1 1 -

2 2 1 1 1 1 -

4 2 2 1 1 1 1 -

4 2 1 2 2 -

1 2 1 1 1 2 -

1 2 2 1 1 2 1 2

2 2 1 1 2 2

2 2 1 2 2

1 2 1 2 1 2 2

1 2 1 2 2 2 -

2 1 1 1 1 2

1 1 1 1 1 1

5 2 2 1 2 2 -

3 2 2 1 1 1 1

4 2 2 1 1 3 1 1

2 2 1 1 3 1 -

1 1 2 2 1 2 1 1

2 1 1 3 1 -

1 1 1 1 1 2 2

2 2 2 1 1 -

3 2 2 1 3 1 -

1 1 1 1 2 1 2

1 2 2 1 2 1 2

1 1 1 1 1 1 2

1 1 2 1

2 2 1 1 1 2 2 2

5 2 1 1 2 2 2 -

1 1 1 2 2

4 1 1 1 1 3 -

3 2 1 2 2 1

1 2 2 2 1 2 2

2 2 1 1 1 2

4 2 2 1 1 1 -

2 2 2 1 2 2

1 2 2 2 1 2 1 2

1 2 2 1 2 2

2 2 2 1 1 8 2

1 1 2 1 1 2 1

3 2 2 2 2 2 2 1

1 2 1 2 1 3 2

2 2 2 2 1 2 1 1

1 1 1 1 2 2

1 2 1 1 2 1 2

2 2 2 2 2 2 2

2 1 1 2 1 3 2

2 1 1 2 1 4 1 2

1 1 1 1 2 1 2

1 2 1 2 2 1 2

1 2 2 1 1 2 1 2

2 2 1 1 1

1 2 2 1 2 1 2

1 2 1 1 2 1 3

1 1 1 1 2

4 2 2 2 2 3 2 -

3 3 2 1 -

1 2 1 1 2 2 1 1

1 1 1 1 3 1 1

1 2 1 1 4 2 1

1 2 1 2 4 2 1

1 2 1 1 2 2

2 2 1 1 3 2 -

2 3 -

1 2 3 -

2 2 1 1 3 1 -

2 2 2 3 2 -

2 2 2 1 2 1 -

1 2 2 2 2 2

2 2 1 1 2 2 -

1 1 1 1 2 1 -

1 2 2 3 2 -

3 1 2 3 1 4 2 1

1 2 1 2 -

4 2 2 2 1 2 -

1 2 1 2 2 -

1 1 2 2 2 -

1 2 1 1 2 2 -

3 2 2 2 1 2 2 -

3 2 2 2 2 2 2 -

2 2 1 3 1 -

2 1 2 1 -

2 1 1 1 1

1 1 1 3 1 1

1 2 2 1 1 3 -

1 1 1 2 1 1 2

3 2 1 1 1 3 2 1

8 1 2 1 1 2 -

8 2 2 -

8 1 2 2 1 2 -

8 1 2 1 1 2 -

7 1 2 2 -

5 1 2 1 1 1 2 -

6 1 2 1 1 2 -

5 1 2 1 2

5 1 2 1 1 1 1 -

5 2 2 2 1 2 -

4 1 2 3 2 1 2 -

5 1 2 2 1 1 1 1

3 1 2 2 1 2 -

6 2 2 2 1 1 2 -

3 2 2 2 1 2 -

3 2 2 1 1

7 1 1 2 1 1 -

4 1 1 3 -

3 1 1 3 1 3 -

3 1 2 3 1 2 -

5 1 1 -

3 1 1 -

5 1 1 -

1 2 2 2 1 1 2 -

1 1 2 1 1 1 1 1

3 1 1 2 1 1 -

1 1 1 1 2 1 2 2

2 1 2 1 2 1 2 -

2 1 1 1 1 3 1

2 1 1 1 2 1 -

1 1 1 2 5 2 -

4 1 1 1 1 2 -

1 2 1 1 1 2 1 -

1 1 1 -

1 1 1 1 1 -

3 1 1 1 1 2

1 1 1 1 1 2 1

3 1 2 4

2 2 2 1 2

2 1 1 1 1 -

2 1 1 1 2 1

1 1 1 1 1 3 -

1 1 1 1

3 1 1 1 1

1 -

1 1 -

2 -

1 -

1 1 1

1 -

2 -

3 1 -

Total number of vascular taxa

1 1 2 2 2 2 1 3 3 3 3 3 3 4 3 2 4 3 4 2 3 3 2 3 2 1 2 3 4 4 2 5 3 3 2 3 3 2 3 1 4 3 4 3 2 3 2 3 4 3 3 3 3 4 4 4 4 4 4 4 5 3 3 4 3 3 3 3 3 3 4 3 4 3 3 2 2 2 3 3 3 4 4 3 4 4 4 4 3 4 4 3 4 5 4 3 1 3 3 4 4 3 2 2 2 2 2 3 3 3 4 3 4 4 4 3 2 3 2 2 2 1 1 1 4 3 3 5 3 4 4 3 3 4 2 3 4 4 2 3 3 3 3 2 2 2 1 3 1 2 2 2 2 2 2 1 2 2 2 2 3 1 1 2 6 9 6 6 4 7 9 3 7 4 0 0 6 4 5 6 5 4 4 6 6 0 0 3 5 8 8 8 1 5 9 1 4 8 2 1 0 6 3 7 3 2 1 9 4 4 4 5 2 9 2 8 6 2 2 5 0 1 2 4 0 5 9 5 5 1 4 4 7 6 3 8 0 6 0 8 1 8 3 8 8 0 4 9 7 0 2 2 8 2 2 2 3 1 0 2 6 9 2 1 5 3 2 7 6 7 8 0 5 8 2 4 3 3 3 2 5 1 6 4 8 8 6 5 1 5 0 3 8 4 8 7 0 2 8 5 6 0 8 6 7 8 0 7 6 8 3 1 9 4 1 3 0 0 1 6 2 2 0 8 8 1 6 0 6 1

Number of introduced weeds

1 2 . . . . . . . . . 1 . . . 1 . . 2 4 1 . . . . . . . 1 1 1 . . . . . . . . 1 . 2 1 3 1 2 1 1 1 1 1 1 2 1 2 1 1 . 1 . 2 4 2 3 2 1 1 2 1 1 2 1 1 1 1 1 . 2 . . 1 1 2 2 1 1 1 . 3 2 . 1 1 1 2 . 1 . 3 1 3 5 1 2 1 3 2 2 1 2 1 1 2 1 2 1 1 2 1 2 2 1 1 1 1 5 2 3 7 2 5 5 3 2 4 3 4 2 7 4 5 6 5 3 4 7 8 4 5 6 6 2 7 8 9 8 5 0 8 5 3 3 7 8 2 4 6

Number of native weeds/crops

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 3 3 2 3 1 5 1 4 6 4 3 4 8 7 0 7 9 8 9 5 4 7 3 4 4 3 3 7 0 2 7 4 6 0 5 7 6 9 8 4 5 9 1 1 6 0 5 9 0 1 9 2 2 3 2 2 0 0 0 5 7 1 3 4 9 9 0 8 8 1 2 3 1 2 9 7 6 9 0 1 2 1 5 4 6 7 0 4 2 0 3 2 5 1 0 9 6 2 1 3 7 3 8 8 6 1 1 1 2 2 4 3 3 4 5 4 9 1 7 9 2 6 6 6 6 4 4 8 6 4 6 2 1 5 3 6 2 0 9 4 3 4 3 9 9 9 5 8 7 7 0 7 7 5 5 7 6 7 7 4 3 6 9 4 7 4