Offshore wind energy and birds by Isadora Christel

PhD Thesis 2012

Offshore wind energy and birds: Integrating assessment tools in space and time Isadora Christel JimĂŠnez

Offshore wind energy and birds: integrating assessment tools in space and time Energia eòlica marina i aus: integració de les eines d’avaluació a l’espai i el temps

Energía eólica marina y aves: integración de herramientas de evaluación en el espacio y en el tiempo

Isadora Christel Jiménez García

Barcelona, Octubre 2012

ICH

cover and Book design: isadora christel

Facultat de Biologia - Departament de Biologia animal Programa de doctorat de Biodiversitat Bieni 2005-2007

Offshore wind energy and birds: integrating assessment tools in space and time Energia eòlica marina i aus: integració de les eines d’avaluació a l’espai i el temps

Energía eólica marina y aves: integración de herramientas de evaluación en el espacio y en el tiempo Memòria presentada per

Isadora Christel Jiménez García per a optar al títol de Doctora per la Universitat de Barcelona

Barcelona, Octubre 2012 Directors de Tesis

Dr. Xavier Ferrer Animal Biology Department Universitat de Barcelona (UB) Barcelona, Spain

Dr. David Vieites

Dr. Grégoire Certain

Museo Nacional de Ciencias Naturales Havforskningsinstituttet Consejo Superior de Investigaciones Institute of Marine Research (IMR) científicas (CSIC) Tromsø, Norway Madrid, Spain

The preparation of this thesis has been financially supported by a doctoral grant (APIF/2008) by the University of Barcelona

“Herzog dijo que hay más de un Anapurna en la vida de cada hombre y no siempre se alcanza la cumbre… pero si el esfuerzo nos permite vislumbrar algo de aquello que hay más allá del azul infinito, ya vale la pena. Al menos ayuda a vivir.” (1980, carta de Enric Benavente i Mata a mis abuelos)

TABLE OF CONTENTS ix xii

Acknowledgments How it all began

GENERAL INTRODUCTION 3 3 7 7 12 13 15 15 18 21

Introduction Offshore wind energy Environmental assessment Seabirds as indicators Objectives Supervisors’ report Methodological approach Seabirds surveys Study areas Modelling tools

RESEARCH PAPERS 27

Chapter 1: A refined methodology to estimate the vulnerability of seabird community to the establishment of offshore wind farms

Chapter 2: Wind farm Sensitivity Index for seabirds - A ssessing offshore wind energy development on the coasts of the Iberian Peninsula

Chapter 3: Seabird aggregative patterns: a new tool for offshore wind energy risk assessment

Chapter 4: Foraging movements of Audouin’s gull (Larus audouinii) in the Ebro Delta, NW Mediterranean: A preliminary satellite-tracking study

DISCUSSION & CONCLUSIONS 91 91 93 94 95 100 102

Discussion Vulnerability Index Aggregative Patterns Individual tracking Integrating tools in SEA and EIA Conclusions References

CATALAN SUMMARY APPENDIX

vii

ACKNOWLEDGMENTS

ACKNOWLEDGMENTS Agraïments, Agradecimientos, Remerciements

Estas líneas abren la tesis pero son las

però crec que podem sentir-nos orgullosos.

últimas que escribo. Es el final del camino.

Gràcies per la oportunitat d’arribar fins aquí,

Un camino largo, no siempre fácil, pero des-

per entendre i donar suport a la noia pràcti-

de luego lleno de gente que me ha ayudado,

ca que sóc i per confiar en mi tot i els dalta-

apoyado, animado y que de una forma u

baixos inevitables en tants anys.

otra ha estado a mi lado mientras este proyecto iba madurando.

Gracias a todos

Els principis van ser solitaris, lluny de la universitat i amb trasllats de despatx en despatx. Per sort els vaig poder compartir amb tu, Albert. També recordo el nostre primer té/cervesa a la UPC. No sempre ha estat fàcil

Mis primeros pasos por el departamento

treballar plegats. Tots dos som d’idees cla-

de Biología Animal empezaron mucho antes

res i tossuts per defensar-les, però crec que

de acabar la carrera, era 2001 y después de

hem superat prou bé les nostres diferències.

hablar con Domingo acabé haciendo cam-

Em quedo amb les reunions al bar que han

pañas de campo capturando codornices

solucionat més d’un problema metodològic,

y “descuartizando” tejones. Aún hoy soy

d’anàlisi o de com enfocar les nostres tesis.

hipersensible al canto de Coturnix cotur-

I bé, també amb unes quantes sessions de

nix. Domingo, eres un magnífico profesor

radar al Delta en companyia de Xesco.

de campo. El tiempo que colaboré contigo,

Xescuu! De ti me quedo con los intentos

Manel y con todos tus entonces doctorandos

por iniciarme en el mundo de la ornitología,

fue lo que me hizo decidirme a dar el paso

pero sobre todo, con tu buena música para

hacia la ciencia, así que para ser justos, ese

pasar las horas delante de un monitor de

es el principio del camino.

radar en medio de la nada.

Xavier, encara recordo entrar al teu des-

El traslado al departamento fue un cam-

patx per parlar del treball de Zoogeografia,

bio agradable. La tesis pasó de ser una lucha

l’últim treball de la carrera. Ha plogut molt

solitaria a una montaña rusa compartida

des de la idea per aquell treball fins arri-

con otros sufridos becarios. En compañía,

bar a aquesta tesi, el tema ha donat voltes

todo se hace más llevadero sobre todo si va

| servida por Jose Luis y Miguel, sois mis bio-

metodológica y si salió a flote, fue gracias a

barmans preferidos. En esta montaña rusa y

Marcos del Castillo y a Joan Navarro. Marcos,

desfile de compañeros de bar, pasillos y des-

gracias por todas esas horas de teléfono (y

pachos hay muchos nombres. Eloy, gracias

son muchísimas) hablando de como anali-

por entender a la perfección esos días en

zar una muestra que no daba para casi nada.

los que voy dando saltos por los pasillos

Joan, gracias por ayudar a desencallar el

y por darme ese abrazo siempre que lo he

tema y acompañarme en el proceso de pu

necesitado. Gemmmma, gràcies per donar

blicar mi primer artículo, se dice rápido pero

geni i figura als dies grisos de la facultat.

no fue un paso trivial en esta tesis.

Cotín, por descubrir mi wild side. Blanca,

Si hay algo a destacar, es el cambio de

por ser el eslabón que une a todos los de-

inflexión que supuso para mí ir a Noruega.

partamentos. Mari Carmen, por dar color

Norway... Norgue. In Trondheim I found my

a la facultad con tus viajes y experiencias.

third co-supervisor and the focus of the the-

Alberto, por ser el perfecto asesor estratega.

sis that I had been unconsciously looking

Olatz, por ese trabajo de campo en bikini en

for. Grégoire, merci de ton soutien. The first

el principio de los principios. Edu, por re-

day we met at the airport I already had the

cordarme que la ciencia al más puro estilo

feeling that we were going to get on well. I

“Konrad Lorenz” aún es posible, por hacerme

love your scientific stubbornness, your freak

sentir no tan R-freaky y por todas esas char-

side and your quality as supervisor, mentor,

las entre mis cafés y tus cigarros. Pero es in-

advisor… You helped me to jump without

justo que haya tan poco espacio, porque son

doubts to the world of R programming, and

muchos nombres y de todos podría decir

gave me such a quantity of useful tips and

algo: Manolo, Debs, Victor, Irene, Mario, los

comments for my future as a researcher that

(no tan)”nuevos”: Jose Manuel, Jaime, Fav, T,

I had professionally changed by the time I

Nicole, Urtzi… y los que ya dejasteis el nido:

went back to Barcelona. You have the mak-

Oriol, Laura “arpella”, Inés, Luigi, Rubén,

ings of a wonderful supervisor. Take more

Josep Lluis, Eva. Dejo muchos nombres por

students! Tusen takk also to Tycho, Inga,

poner… no me lo tengáis en cuenta porque

Duncan, Signe, Camilla and all and each of

hoy me acuerdo de todos vosotros.

the NINA colleagues that opened my mind

La tesis hizo un salto cualitativo cuando

and gave me so much to bring back home.

una llamada me informó de que, cuando ya

And well, the experience was wonderful,

había perdido toda esperanza de obtenerla,

but it was also possible thanks to you, Vidar.

me daban la ansiada beca predoctoral. Con

With you at home I didn’t feel the Norwe-

la beca también llegó mi primer co-director,

gian cold. You adopted the Spansk roommate

Vieites. Nos hemos visto muy poco (aún te

and immersed me completely into Norway

llamo por tu apellido!) pero ha sido tiempo

with our afternoons of talking, watching

altamente productivo. Ese retiro en la sierra

football, movies and partying. My second

madrileña fue una experiencia genial y sien-

stay in Norway was a really short one, but

to que el espacio-tiempo no se haya alinea-

those few weeks at the IMR in Tromsø (al-

do para poder interaccionar más. Aún con

most 70ºN) were also a very intense experi-

todo, me alegro de tenerte como “co-dire”.

ence full of new colleagues for science: Ben-

El primer artículo que publiqué (aunque

jamin, Alf Harbitz, Ulf and leisure: Maria,

último de la tesis) fue una pesadilla

Marina, Atal, Keka, Rune, Trond, Virginie,

ACKNOWLEDGMENTS Roland, Bas… thank you for everything and more.

Como ya he dicho, para mí una tesis es una montaña rusa. Tiene subidas y bajadas.

Una tesis avanza no sólo en el despacho.

Mis estancias en Noruega y el Máster en Co-

Mucha gente te ayuda incluso sin saberlo:

municación Científica están en las cimas,

mis compis bajo el agua (Olga, Laura, Núria,

pero también ha habido unas cuantas ba-

Patri, Silvia y Carlos); mis niñas, con nuestro

jadas vertiginosas y en esos momentos he

primer viernes de cada mes; y todos los ami-

tenido la fortuna de contar con tres refugios,

gos que pacientemente han visto pasar los

remansos de paz, fuentes de consejos y áni-

años sin preguntar cuando la acabaría y que

mos. Tres lugares en los que tal cual cruzar

tímidamente cada cierto tiempo se atrevían

la puerta se me ha permitido descargar de

a volver a preguntar de qué iba exactamente

mis hombros el peso del doctorado, los mie-

la tesis.

dos y todas las dudas. Uno está en Palamós-

Ya hacia el final descubrí toda una vo-

Gràcia-Centelles... allá dónde sea que Laura

cación, la comunicación científica. 2011 fue

prepare una infusión y Joan improvise una

un año intenso, pero junto a mis compis

cena. Vuestra casa siempre ha sido un refu-

de Máster pasó volando. MCC16 forever!

gio de calma y consejos en los momentos

No pongo vuestros nombres porque sois

más críticos. Otro está en Amsterdam. Alicia,

muchos, pero de verdad que esta tesis llega

gracias por todos estos años de amistad y

a buen puerto gracias a esa transfusión de

vuelos de ida y vuelta, que no es poco. Pero

energía, ganas y motivación que comparti-

independientemente del lugar, los brazos de

mos de 7 a 10 de la noche cada día.

Ivan han sido mi mejor refugio.

Y como no… padrins, tiets, primos, abue-

Ja són molts anys i tu has viscut al

los y padres, propios y políticos: Familia.

meu costat tots i cadascun dels moments

Gracias por acompañarme aún sin acabar de

d’aquest doctorat. La meva lògica de biòloga,

entender demasiado qué hacía o por cuánto

s’ha resistit a les teves pràctiques solucions

tiempo iba a “seguir estudiando”. Está cla-

d’enginyer, però anys i ioga ens han portat

ro que por más que lo he intentado no he

fins el dia que aquesta tesi s’acaba. Pensaves

conseguido convenceros de que esto ES UN

que la vida seria senzilla quan per fi acabés

TRABAJO así que: Abuelos, por fin la nieta

amb el doctorat? …doncs bé: Ara comença la

“ha dejado de estudiar”!

nostra aventura.

xii

HOW IT ALL BEGAN “What do you think? Will it have an impact?” “Oooh, so you moved to the dark side… you will say yes to the wind farm, won’t you?”

Those were the two most asked ques-

institutions and the offshore wind industry.

tions at the beginning of my research expe-

Some reports were helpful as they gave re-

rience. In August 2004 Dr. Xavier Ferrer was

commendations on survey methodologies

commissioned the Environmental Impact

but all their conclusions seemed a sequence

Assessment for an offshore wind farm. A

of “too many” descriptive distribution maps

project in front of the Ebro Delta area. The

summarized at the end with a few paragra-

developer company, Capital Energy Offsho-

phs according to the previous ornithological

re, paid for an exhaustive EIA and Albert

knowledge of the authors.

Cama and I suddenly found ourselves in the

As I see it, in this topic, our role as scien-

middle of a huge, potentially conflictive pro-

tists is to evaluate, in the most impartial and

ject. No complaints! In some countries, paid

objective way, the potential impacts of such

science is a luxury.

installations; and more importantly to do

People from Capital Energy, my family,

our best to quantify them. With this infor-

friends... all used to ask me the first ques-

mation, we have to inform decision-makers

tion. My biologist colleagues would look at

in the most clear synthetic way. A French

me with terrified faces and go for the second

supervisor of mine would argue that not

comment/question. My answers:

even scientists can reach real impartiality and objectivity, but I am satisfied if we make

“Yes, there will be an impact; we don’t need the study to say this”

a sincere effort to reach them. During these years, I have been somewhere in-between ornithologists, ecolo-

“No, it is not my job to decide that”

gists, conservationists, managers and business people. It is not easy when you do not

This situation was the seed of this the-

fit in a particular label, but at least it gives

sis because there was not much scientific

you a different point of view. After 8 years

literature on the topic. After some months

working at the University, I have learnt a few

of bibliographic search, all I could find was

things about myself: 1) I am definitively not

grey literature from governments, research

a passionate ornithologist (although now I

HOW IT ALL BEGAN can even follow their conversations); 2) I de-

or at least move them to the appendix of

finitively like methods and programming;

any Impact Assessment. There is a huge

and 3) I love visual communication of con-

part of fruitless work that has no space in

cepts because I have a taste for simplicity.

this dissertation and there is still a great

This thesis is the result of this.

part of work to be done. I know. But I got to

I wanted to bring some integrative tools

the end of the fourth year of my University

to summarize results in as fewer maps as

PhD grant and I took an unpaid extra half

possible. I wanted to transcend plain des-

year. Luckily for a scientist, what is still to

criptive distribution maps, â&#x20AC;&#x153;eradicateâ&#x20AC;? them

be done is just an opportunity for the future.

xiii

General introduction

INTRODUCTION

INTRODUCTION Offshore wind energy development and seabirds’ conservation: A management challenge

The world’s growing energy demand and

depending on the scenario (EC 2011b). One

climate change are two of the great challen-

of the policy measures to achieve this goal is

ges of this century. A trade-off between cli-

the Renewables Directive, which sets a tar-

mate change policies and competitiveness

get of 20% of energy consumption to come

is needed to find an economically viable low

from renewable sources in 2020.

carbon future. In this context, the European

In Europe, renewable energy sources re-

Commission defined the ‘Energy roadmap

present the 18% of all the energy production

2050’ that explores the possibilities to achie-

(Eurostat, 2009; Fig. 1a). Within Renewable

ve a low-carbon economy that at the same

electricity production, hydropower is the

time ensures a competitive, sustainable and

main source (54.5%) followed by wind power

secure energy supply (EC 2011a). The Euro-

(22.5%) (Observ’ER, 2011; Fig. 1b). By 2050,

pean Union is committed to reducing gre-

wind power is expected to provide more

enhouse gas emissions up to 80-95% below

electricity than any other technology (EC

1990 levels by 2050 (EC 2011b). It is impossible

2011b) and hence the potential contribu-

to forecast Europe’s long-term evolution but

tion of the marine environment for offshore

some of the possible low-carbon scenarios

wind energy development has received high

are (i) a highly energy efficient system, (ii) a

attention in the last decades.

system with a diversified supply of technologies including carbon capture and storage

Offshore wind Energy

facilities and/or nuclear power, and (iii) an scenario with a strong support to renewable

The first offshore wind farm was installed

energy sources. All the predictions for these

in Denmark in 1991. Since then, the sector

decarbonized Europe scenarios show that

had a rapid expansion (Fig. 2), particularly

electricity will have to play a greater role

in the north of Europe. So far, Europe has

than fossil fuels and the share of renewable

become the world leader in offshore wind

energy sources will rise substantially achie-

power with a total of 1371 offshore turbines

ving at least the 55% of the gross final ener-

spread across 53 wind farms in 10 countries

gy consumption in 2050 up to 64% or 97%

by the end of 2011 (EWEA 2012). The UK is

General introduction the country with the largest installed off-

Fig.1 a) EU energy

shore wind capacity, followed by Denmark,

production by source in 2009 (Eurostat 2009) RES=Renewable Energy Sources. b) Share of each resource in Renewable electricity generation in 2010 (Observâ&#x20AC;&#x2122;ER 2011).

Netherlands and Germany (Table 1). Interest in offshore wind energy is spreading beyond Europe. China, Japan, South Korea, USA and Israel have companies actively developing offshore wind turbines, although only China has three operational offshore wind farms. Most of the installed turbines have foundation structures. Floating models are being developed, and Norway and Portugal are the first countries that have a full-scale floating turbine installed. As the technology matures, offshore wind farms are expected to grow in size but also to be deployed further from the coast and in deeper waters, particularly if floating technology is further tested and its economic viability demonstrated. Current projects under construction have an average depth of 25 m and a distance to the shore of 33 km (EWEA 2012). This is possible because many of the actual OWF have been built in the North Sea that has a large part that lies on the European continental shelf (Fig. 3). This provides relatively large flat and shallow regions suitable for development (Henderson et al. 2003). In comparison with Northern Europe, the West coast of France,

Box 1 List of commonly used abbreviations.

abbreviations

the Iberian Peninsula and the Mediterranean Sea remain a challenge for OWF devel-

OWF: Offshore Wind Farm

opment. Although there are planned proj-

SEA: Strategic Environmental Assessment

ects for these areas, the available turbines

EIA: Environmental Impact Assessment

and foundation methods would require the construction of the wind farms much closer to the shore with a consequent increment of the conflicts to find optimal locations in terms of social acceptance, environmental impacts, conflicts of interest and national marine spatial planning. All these factors, together with a lack of funding, are slowing the offshore development of wind energy in West and South Europe. Indeed, the offshore wind energy is by no means free of conflicts. At a global scale, the

INTRODUCTION

4500

2500 2000 1500 1000

Cumulative capacity (MW)

3000

Fig.2

4000 3500

Cumulative offshore wind installations (MW) (EWEA, 2012).

500

Country

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

Total

No. of farms

No. of turbines

636

401

128

1371

Capacity installed (MW)

2094

857

247

200

195

164

3813

Table 1 European Operational Offshore wind farms by country (EWEA 2012). UK: United Kingdom; DK: Denmark; NL: Netherlands; DE: Germany; BE: Belgium; SE: Sweden; FI: Finland; IE: Ireland; NO: Norway; PT: Portugal.

Fig.3

Operational and planned offshore wind farms in Europe (EWEA 2011).

General introduction shift to renewable energies is widely accept-

bance of the seabed and fauna during OWF

ed as a necessary step to mitigate the ef-

construction and operation (Whitehouse et

fects of anthropogenically induced climate

al. 2010; Burkhard et al. 2011), the impacts on

change (King 2004; Rosenzweig et al. 2008).

fish larvae (Perrow et al. 2011), the unknown

At the local scale, however, the environmen-

effects of underwater noise on fish life and

tal impacts of wind energy development

sea mammals (Madsen et al. 2006; Bailey et

must be carefully considered (Gill 2005). In

al. 2010) and the effects at population-level

the field of marine management, there is

of collisions of birds with turbines (Fox et al.

a growing concern on the development of

2006; Desholm 2009) and disturbance (Dre-

offshore wind energy and its potential im-

witt and Langston 2006; Masden, Haydon, et

pacts on the marine ecosystem. Some of the

al. 2010).

aspects that are being studied are the distur-

Box 2 State of Offshore Wind Energy development in Spain

OFFSHORE WIND ENERGY IN SPAIN Spain has no operational offshore wind farms so far. Since the beginning of the offshore wind energy expansion in Northern Europe, different developer companies showed their interest on constructing offshore wind farms in the Spanish coasts. Despite the early private sector initiatives to promote its development, the Spanish government took the first legislative step forward in 2007. That year the Real Decreto 1028/2007 was published setting the compulsory administrative procedure that developers should follow to have the concession to construct an offshore wind farm in the Spanish coasts. As part of the necessary procedure a Strategic Environmental Assessment (SEA) for the Spanish coastline was commissioned. This study was published in 2009 (MARM and MITYC 2009) and included the definitive zonation map for offshore wind development areas. This map divided the Spanish coasts in 72 marine eolian areas (defined by one decimal degree squares). Within each area, the 24 first nautical miles were assessed according to multiple criteria and classified as suitable areas (in green), suitable areas with constraints (in yellow) and exclusion areas (in red).

The administrative concessions process is long and complex and has suffered several delays. To the date, the start of the application process is on hold, hence there is no official number of planned wind farms in Spain.

INTRODUCTION Environmental assessment

hore wind farms (Camphuysen et al. 2004). In the last years, as the sector has grown,

The European Union has a regulatory fra-

more reports have been published as well

mework (Directive 2001/42/EC) to standardi-

as research papers on the assessment of

ze the evaluation and monitoring of human

environment-OWF interaction of particular

activities in the ecosystem and to guarantee

wind farms (e.g. Desholm and Kahlert 2005;

a rational development of such activities

Perrow et al. 2011; Skeate et al. 2012) as well

including

considerations.

as reviews and general papers regarding

On a large scale, countries must develop a

SEAs (Elliott 2002; Fox et al. 2006; Punt et al.

Strategic Environmental Assessment (SEA)

2009; Masden, Fox, et al. 2010).

environmental

to plan their offshore wind farms network minimizing their ecological impact on the

Seabirds as indicators

coastal environment. At a local scale, each wind farm project requires an Environmen-

Ecological systems are highly diverse

tal Impact Assessment (EIA) of the possible

and complex. While ecological studies focus

negative impacts of the proposed project in

on this complexity, applied ecology requires

the marine environment.

methods that synthesize this complexity

The EIA concept was first introduced in a

in order to take actions that may have eco-

European Directive in 1985 (Directive 85/337/

nomic consequences (Piatt and Sydeman

EEC) but it was restricted to certain types of

2007). Such is the case of using indicator

projects. Years later, the need to deal with

species to simplify the monitoring and ma-

environmentally damaging decisions at na-

nagement processes for EIAs and SEAs. Ma-

tional levels developed into the Strategic

rine top predators are a key component of

Environmental Impact Assessment that was

marine ecosystem management (Boyd et

finally included in a European Directive in

al. 2006) and within top predators, seabirds

2001. Although all countries of the EU are

have become widespread indicators to eva-

implementing SEAs since 2004, EIAs have

luate potential effects of human activities

a longer tradition and clearer implementa-

at sea as well as ecosystem health (Cairns

tion procedures. This is also reflected in the

1987; Nettleship and Duffy 1993; Mallory et

environmental assessment of offshore wind

al. 2006).

energy.

Seabirds offer many advantages com-

For many years, the only available infor-

pared to other species. Considering an en-

mation on offshore wind farms assessments

vironment where most species are under

were reports focused on how to perform

water, seabirds are conspicuous animals,

EIAs of particular projects. The Danish expe-

they are easily surveyed during their move-

rience with the first wind farms was exten-

ments and in resting areas; and some spe-

sively reported by the National Environmen-

cies are easy to capture allowing individual

tal Research Institute (NERI) and their aerial

tracking and demographic studies (Piatt and

surveying methodology has become a stan-

Sydeman 2007). Moreover, most seabirds

dard for many EIAs (Noer et al. 2000). Later,

have specific legal protection frameworks

the COWRIE (Collaborative Offshore Wind

(e.g. Birds directive and Habitats directive

Research Into the Environment) from UK

in Europe) and are flagship species for the

commissioned a report to standardize the

public (Fox et al. 2006) which is reflected in

seabird surveys techniques for EIA of offs-

the abundance of comprehensive long-term

General introduction studies of their distribution at sea and popu-

des, the stationary structure or being caught

lation trends.

and injured in the pressure vortices created

Because of all this, the distribution and

by the rotor blades (Fox et al. 2006). Collision

abundance of seabirds are usually provided

risk depends on a range of factors related

as key information to support the establis-

to bird species (manoeuvrability, wing span,

hment of marine protected areas (Garthe

etc.), behaviour (e.g. nocturnal activity), pre-

et al. 2011; Arcos et al. 2012), to implement

sence in large numbers and weather condi-

fisheriesâ&#x20AC;&#x2122; management measures (Boyd et

tions reducing visibility. Collision mortality

al. 2006), to monitor the impact of oil and

is the most important hazard since direct

gas platforms at sea (Wiese et al. 2001), or to

mortality can potentially have rapid conse-

assess the impact of environmental disas-

quences at population levels. Nevertheless,

ters such as oil spills (Bretagnolle et al. 2004;

there is still limited information on the ac-

Moreno 2010). Thus, seabirds are suitable

tual numbers of bird collisions with offshore

indicators of the marine environment, and

wind farms, largely as a consequence of the

have become one of the keystones of the

technical difficulties to detect these colli-

decision-making process for the selection

sions at sea (Drewitt and Langston 2006).

of optimal areas for national offshore wind

Gradually, more remote technologies are

development and the impact assessment of

being included in the study of bird-turbine

particular OWF projects.

collisions at offshore wind farms. One of the most extended tools are S-band Radars,

Potential impacts on seabirds

although they cannot quantify collisions directly and depending on the study cannot

At the time of selecting development

provide species specific information (Chris-

areas, or when the location for a project is

tensen et al. 2004; Desholm et al. 2006). Yet,

settled, we can differentiate the effect of

radars are a useful tool to implement co-

OWF on two types of seabirds: migrant spe-

llision models (Desholm and Kahlert 2005;

cies that may encounter the wind farms on

Chamberlain et al. 2006). Thermal Animal

their migratory routes and breeding and

Detection Systems (TADS) are also an alter-

wintering species that have to interact with

native to gather information on actual co-

wind farms in their foraging grounds. Both

llision rates. This infrared based technology,

types of seabirds are susceptible to mul-

however, has been seldom applied and there

tiple anthropogenic impacts (e.g. Anderson

are few published studies on its performan-

et al. 2003; HĂźppop et al. 2006; Louzao et al.

ce (see Desholm et al. 2006 for a review).

2006), but the potential impacts of offshore wind farms on seabird communities can be

Habitat change

classified in three types; (i) direct mortality through collision, (ii) modification of their

This impact comprises the loss of habitat

physical habitat and (iii) avoidance due to

that would result from the presence of the

disturbance and barrier effects.

turbine bases, grid connection cabling and any other associated construction. The scale

Collision risk

of habitat loss is not generally perceived as a major concern whenever this is not produ-

Birds flying within a wind farm area are

ced in areas of high biodiversity or ecologi-

clearly at some risk of colliding with the bla-

cal importance (BirdLife International 2003).

INTRODUCTION

Resident species foraging flight paths

Flight path obstacles COLLISION RISK Injuries and casualties by collision with the turbine or by air turbulence

Reduced survival

Feeding habitat

Attraction

Migrant species migration flight paths

Habitat loss

HABITAT CHANGE

Barrier effect DISTURBANCE

‘Physical’ habitat gain with increased collision risk

‘Effective’ habitat loss

‘Physical’ habitat loss

Increased flight distance

Changes to annual breeding output and survival

New feeding habitat

Lost feeding habitat

Fig.4 The three major potential impacts of offshore wind farms on seabirds and their physical and ecological effects.

General introduction However, indirect habitat loss might also

factors depend, for instance, on the offshore

occur because of the turbine foundations

wind farm location with respect to impor-

on the seabed, or changes in habitat use by

tant habitats, design of the turbine array

humans. For instance, construction activi-

and distance between turbines. Moreover

ty and turbine’s distribution may affect the

seabirds may show different disturbance le-

site’s hydrology and have an impact over

vel depending on their diurnal and noctur-

greater areas (Percival 2003). There is uncer-

nal activity patterns (Desholm and Kahlert

tainty about the magnitude of such changes,

2005) or different weather conditions. Be-

but the damage may be significant especia-

havioural responses to the wind farms may

lly on feeding areas such as sandbanks in

vary between species but also between in-

shallow waters (Drewitt and Langston 2006).

dividuals of the same species according to

Turbine bases tend to have a ‘reef effect’

factors such as stage of life cycle (wintering,

that increases biodiversity through habitat

moulting and breeding), flock size or ten-

creation (Linley et al. 2007), but this may

dency to habituation.

influence floral and faunal communities

Even if disturbance and displacement

in complex ways generating both positive

occur, it may actually be inconsequential

and negative effects depending on the site

if there are abundant alternative habitats.

and the species (Perrow et al. 2011). Seabirds

However, offshore wind farms located in

might also be differently affected by these

migratory flyways or in local flight paths

changes in the habitat. While some specia-

might alter birds’ movements increasing

list species may lose important food sou-

their energy expenditure (Masden, Haydon,

rces, other opportunist species (e.g. gulls)

et al. 2010), this particular type of distur-

may increase their presence in the area to

bance is called the ‘barrier effect’. In fact,

exploit the new food source or, as it happens

observations in operational wind farms

with cormorants, seabirds may simply be at-

show that many birds chose to fly outside

tracted to turbine maintenance platforms to

the wind farm rather than fly between the

use them as perching structures (Kahlert et

turbines (Desholm and Kahlert 2005; Larsen

al. 2004). Nevertheless, this gain of habitat

and Guillemette 2007). Unfortunately, there

might be counterbalanced by higher colli-

is a lack of complete before-and-after con-

sion risk.

trol-impact studies (BACI) in many operational wind farms to properly quantify the

Disturbance

barrier and disturbance effects compared to baseline behaviour of seabirds (Drewitt and

The presence of turbines, as well as ves-

Langston 2006).

sels and people movements related to site construction and maintenance, can poten-

Gap of knowledge

tially deter some seabirds from using areas within and surrounding wind farms. These

The internationally agreed guidelines re-

displacements result in actual habitat loss

commend the assessment of collision risk

not because physical changes of the area but

with radar studies in strongly migratory

as a consequence of a behavioural respon-

areas (Desholm et al. 2006; Fox et al. 2006;

se. The scale of disturbance effects varies

Kunz et al. 2007) and density maps as a pro-

greatly depending on a wide range of factors

xy to assess the loss of foraging habitats by

(Drewitt and Langston 2006). Site-specific

avoidance or physical habitat modification

INTRODUCTION (Camphuysen et al. 2004; Fox and Petersen

garding seabird distribution and abundance,

2006).

data is usually reported as simple locations

Regarding collision risk assessment, ra-

or density grids. After a review of more than

dar technology is a powerful tool to improve

200 published studies, Tremblay et al. (2009)

our knowledge on spatio-temporal patterns

remarked that â&#x20AC;&#x153;the simple display of distri-

of some seabird groups. Data gathering from

bution data has been much more commonly

radars and the analysis of the outputs requi-

used than quantitative indicesâ&#x20AC;?. Indeed, few

re comprehensive studies that already have

studies have attempted to address analyti-

been addressed in several works (Desholm

cal and synthetic methods to extract ade-

2006; Brookes 2009; Mateos 2009). In con-

quate decisions at strategic (SEA) or local

trast, the use of density maps has fallen

(EIA) levels from seabird distribution data.

behind in the integration of the spatio-tem-

This thesis aims to contribute to fill in this

poral dimension of seabird patterns despite

gap in the methodological approach to the

seabird distribution maps play a prominent

use of seabird distribution data for Offshore

role in most EIA and SEAs assessments. Re-

Wind Energy Assessments.

General introduction

OBJECTIVES Only when you reach the end of the path, your footsteps become meaningful

Main objective The major objective of this thesis was to gain insight into analytical tools in space and time for offshore wind energy environmental assessment in order to provide practitioners with guidelines on how and when to apply them.

Specific objectives To achieve this objective, this thesis has been structured in four chapters and a global discussion that address the following specific objectives: 1.

Design and test a vulnerability index to assess the potential effects of offshore wind energy development on seabirds. (Chapter 1 and 2)

Develop a tool to integrate the spatial and temporal variability of seabirdsâ&#x20AC;&#x2122; abun-

Investigate the discrepancies between distribution and abundance maps and indivi-

dance at sea to quantify the potential impacts of offshore wind farms on seabirds. (Chapter 3) dual-based tracking of a flagship species. (Chapter 4) Provide practical guidelines on how to integrate the presented analytical tools in the design of SEAs and EIAs. (Discussion)

SUPERVISORS’ REPORT

Dr. Xavier Ferrer, Dr. David Vieites and Dr. Grégoire Certain co-supervisors of the PhD thesis entitled “Offshore wind energy and birds: integrating assessment tools in space and time” certify that the dissertation presented here has been carried out by Isadora Christel Jiménez García in its totality and grants her the right to defend her thesis in front of a scientific committee. As supervisors, we have participated in designing, guiding and correcting earlier drafts of the chapters and manuscripts written by the PhD candidate. The contribution of the PhD candidate to each manuscript is detailed below:

CHAPTER 1: A refined methodology to estimate the vulnerability of seabird community to the establishment of offshore wind farms G. Certain, I. Christel, B. Planque and V. Bretagnolle Journal of Applied Ecology. Impact Factor: 5 (In prep.) IC: Data analysis and writing.

CHAPTER 2: Wind farm Sensitivity Index for seabirds - Assessing offshore wind energy development on the coasts of the Iberian Peninsula I. Christel, A. Cama,G. Certain, J.M. Arcos, J. Bécares, B. Rodriguez, I. Ramirez, A. Meirinho, J. Andrade, D.R. Vieites and X. Ferrer Ecological applications. Impact Factor: 5.1 (In prep.) IC: Analytical study design, data analysis and writing

General introduction

CHAPTER 3: Seabird aggregative patterns: a new tool for offshore wind energy risk assessment I. Christel, G. Certain, A. Cama, D. R. Vieites and X. Ferrer Marine Pollution Bulletin. Impact Factor: 2.5 (Accepted) DOI: 10.1016/j.marpolbul.2012.11.005 IC: Study design, raw observational data processing, data analysis and writing

CHAPTER 4: Foraging movements of Audouin’s gull (Larus audouinii) in the Ebro Delta, NW Mediterranean: A preliminary satellite-tracking study I. Christel, J. Navarro, M. del Castillo, A. Cama and X. Ferrer Estuarine, Coastal and Shelf Science (2012) 96: 257-261. Impact Factor: 2.3 IC: Data collection and processing, analytical study design, data analysis and writing

We also certify that none of the manuscripts included in this PhD thesis has been used as a part of another PhD thesis.

Barcelona, 24th October 2012

Dr. Xavier Ferrer Animal Biology Department Universitat de Barcelona (UB) Barcelona, Spain

Dr. David Vieites

Dr. Grégoire Certain

Museo Nacional de Ciencias Naturales Havforskningsinstituttet Consejo Superior de Investigaciones Institute of Marine Research (IMR) científicas (CSIC) Tromsø, Norway Madrid, Spain

METHODOLOGICAL APPROACH

METHODOLOGICAL APPROACH Integrative tools: Simplifying ecological complexity

Seabird surveys

they do not provide food- have an attraction

From the existing census techniques, the

effect on birds which modifies at some de-

best available methods for obtaining bird

gree the original distribution of the seabirds

distribution and abundance at sea are air-

(Spear et al. 2004). Secondly, this method

craft and ship-based surveys. Boat surveys

requires longer times at sea to cover large

have been largely used following a standar-

areas. Aerial surveys, on the other hand, are

dised methodology (Tasker et al. 1984) with

particularly effective in a simultaneous co-

adaptations according to each particular

verage of large areas providing a snapshot

project. Aerial surveys of seabirds at sea has

of distribution and abundance (Camphuy-

had a rapid expansion in the last decade

sen et al. 2004) with a minimum attraction

and it has been highly influenced by the

or repulsion bias (Certain and Bretagnolle

Danish experience related to the EIA of offs-

2008). Furthermore, aerial surveys can reach

hore wind farms (Camphuysen et al. 2004).

distant inaccessible areas (e.g. shallow areas

So far, the methodology explained in their

or sandbanks) in short time spans with low

reports (e.g. Noer et al. 2000) has become an

per-kilometre costs (Camphuysen et al. 2004;

standard.

Garthe et al. 2011). This is possible due to

The choice between either surveying

the speed of aircrafts, but this speed is also

method depends on the specific research

the main disadvantage of the method. Ae-

objectives since each method has both ad-

rial surveys are performed at the minimum

vantages and disadvantages (see Camphuy-

flight speed that ensures flight safety and

sen et al. 2004 for a review). Boat surveys

provides enough observation time (usually

are especially adequate to make exhaustive

185 km/h). At this speed, there is a short ob-

counts, enabling better species identifica-

servation time that leads to some identifica-

tion with enough time to collect additio-

tion problems for a few species and reduced

nal information as age, behaviour or flight

count accuracy and miscounts of rare and

height. However, this method has two main

small species which are difficult to detect

disadvantages. Firstly, vessels at sea -even if

from the aircraft (Camphuysen et al. 2004;

General introduction Henkel et al. 2007). Moreover, additional in-

in the first and third chapter. Boat surveys

formation is not always easy to collect and

require more time but maximize the detec-

bird flight height cannot be calculated.

ted species richness (number of individual

In this thesis, both boat and aerial sur-

species or taxa identified on each survey)

veys have been used as source datasets of

(Henkel et al. 2007), a key feature to captu-

seabird distributions. Seabirds present dy-

re detailed biodiversity patterns. In the se-

namic and scale dependent distribution

cond chapter, which covers the coasts of the

patterns hence the surveys to tackle this

Iberian Peninsula, simultaneous and syste-

variability must be easily repeated in simi-

matically repeated surveys were not econo-

lar conditions. Aerial surveys outperform

mically viable. Therefore, a maximization of

for these spatio-temporal analyses as a par-

species detection through boat surveys was

ticular area can be surveyed several times

particularly important.

within a year, and therefore, have been used

Fig.5 The Cornide de Saavedra, one of the vessels used for boat surveys in Chapter 2. (Photo by J. Arcos)

Fig.6

Partenavia P68, airplane model used in the aerial surveys of Chapter 3. (Photo by A. Cama)

Both types of surveys aim to monitor a

METHODOLOGICAL APPROACH given area to see if seabirds use it, while it

and temporal dimension of the distribution

seems more intuitive to monitor seabirds

patterns of seabirds. Nevertheless, there are

to study how they are using an area (Perrow

some drawbacks for this methodology. Some

et al. 2006). This move from survey data to

of these devices have high costs, data usua-

tracking data requires a change from po-

lly depend on a small sample size and they

pulation-based studies to individual-based

require a large amount of analytical time. Fi-

studies and has become possible thanks to

nally only a limited number of seabird spe-

the use of electronic-based methodologies,

cies can be captured to attach the tagging

such as satellite tracking transmitters, GPS

methods and there are legal and conserva-

receivers or radio telemetry. Since the early

tion limits to capture protected bird species

1990s, telemetry utilization has constantly

(Perrow et al. 2006). This approach, however,

increased due to the advances in the minia-

provides fine-scale behavioural studies and

turization of the electronic devices (Trem-

could be especially useful if used together

blay et al. 2009).

with surveying methods such as boat or ae-

In the assessment of seabirds interaction

rial surveys (Tremblay et al. 2009) and there-

with offshore wind energy, telemetry is an

fore it was used in the fourth chapter of the

efficient approach to integrate the spatial

thesis.

Fig.7 Satellite tracking device attached to an Audouinâ&#x20AC;&#x2122;s Gull (Larus audouinii) (Photo by I. Christel)

General introduction and applicability for decision-making and

Study areas

management. The three study areas belong to French, Portuguese and Spanish waters

This Ph.D. tackles the issue of offshore

and have a potential for future offshore

wind energy development and birds inte-

wind energy development. Except for the

ractions from a methodological perspective

experimental floating turbine in Portugal,

with no focus on a particular area. However,

so far there is no constructed OWF in any

in order to present an analytical tool, real

of the study areas, which makes them rele-

data is far better than simulated datasets

vant examples on how to apply the analyti-

to understand the tool, its implementation

cal tools for future decision-making. A brief

Fig.8 Study areas: (a) the French continental shelf of the Bay of Biscay, (b) the coasts of the Iberian Peninsula and (c) the Ebro Delta continental shelf

Box 3 Glossary and diagram of the Continental margin and its domains. (Definitions by Maestro et al. 2012; Illustration by EncyclopĂŚdia Britannica, Inc.)

GLOSSARY The continental margin: Submerged prolongation of the continental crust up to the edge of the oceanic crust. The continental shelf: Flat surface with low depth gradient that extends up to the shelf break The continental slope / shelf break: Seaward zone where the seafloor depth gradient increases sharply.

METHODOLOGICAL APPROACH description of the three areas is given below.

Portuguese Society for the Study of Birds) and SEO/Birdlife (the Spanish Ornithologi-

Bay of Biscay

cal Society) in different stages between 1999 and 2011.

The Bay of Biscay is a gulf of the Atlan-

The location of the Iberian Peninsula,

tic Ocean that lies between Cape Ortegal in

surrounded by the Atlantic Ocean and the

Galicia, Spain (43.77ยบN, 7.89ยบW) and the is-

Mediterranean Sea, and the geomorpholo-

land of Ushant in Brittany, France (48.43ยบN,

gical and oceanographical diversity of its

5.18ยบW). Within this area, a region of 100000

continental margins, has significant im-

km2 of French territorial waters (Fig. 8a) was

plications regarding its climate and water

surveyed with 5000 lineal km of aerial tran-

mass circulation (Maestro et al. 2012 for a

sects on a monthly basis from October 2001

detailed revision). The continental margin of

to March 2002 and 4000 lineal km of boat

the Iberian Peninsula has well differentiated

transects in spring from 2003 to 2006.

regions conditioned by many oceanographi-

The study area covered the French con-

cal aspects like the Eastern North Atlantic

tinental shelf of the Bay of Biscay between

Upwelling Region and the Iberian Poleward

Penmarch in the north (47.75ยบN, 4.28ยบW) and

Current that have a strong influence in the

Bayonne in the south (43.497ยบN, 1.64ยบW).

Portuguese, Galician and the Bay of Biscay

Coastal and shelf break areas are the most

continental margins (Peliz et al. 2005; Llope

productive systems of the region (Lavin et al.

et al. 2006); the Mediterranean Outflow

2006; Certain et al. 2008). The Loire and Gi-

Water that flows from the Strait of Gibral-

ronde river run-offs are a source of nutrient-

tar along the continental slope of the Gulf of

rich fresh water (Planque et al. 2004) and the

Cรกdiz (Ribas-Ribas et al. 2011); the Modified

shelf break is an area of enhanced primary

Atlantic Water that affects the Alboran Sea;

production as the deep cooler waters reach

and other Mediterranean water masses that

the euphotic layer due to internal tides and

influence the Valencia, Catalan and Balearic

waves (Gerkema et al. 2004), particularly in

continental margins (Salat 1996). This ocea-

the southern area that is characterized by a

nographic settings affect the composition

deep canyon (Laborde et al. 1999).

and structure of plankton and all the com-

The community of seabirds recorded

ponents of the food web (Santos et al. 2007;

in this area during either aerial and boat

Cabal et al. 2008) up to the highest trophic

transects belong to eight families totaling

levels and therefore, seabirds. Indeed, the

30 species (see Table 2 for details) although

Iberian Peninsula hosts the highest diversi-

both surveys aimed the continental shelf

ty of seabirds in Europe. The community of

therefore, there might be unrecorded spe-

seabirds in this area has up to 39 usual spe-

cies near the coast, which wasnโ t surveyed

cies from ten different families (Table2) in

in detail.

addition to rare species that can eventually be found.

Iberian Peninsula coasts Ebro Delta This area of ca. 230000 km covers the 2

Spanish and Portuguese continental shelf

At a more local scale, the third area was

and spans over 7800 km of coastline (Fig. 8b).

located on the surroundings of the Ebro Del-

Boat surveys were carried out by SPEA (the

ta (40.7ยบ N, 0.75ยบ E; Fig. 8c). The study area co-

General introduction

Table 2 List of species detected in the three study areas. For each study area (BB= Bay of Biscay; IP= Iberian Peninsula; ED= Ebro Delta) a dot indicates the presence of the species in the boat surveys (B) or the aerial surveys (A). In the Ebro Delta aerial surveys the Razorbill and the Atlantic Puffin (Alcidae) could not be differentiated and therefore the species were recorded as a unique group.

METHODOLOGICAL APPROACH vered 1435 km2 of the continental shelf from

of its multiple hydrological and hydrographi-

l’Ametlla de Mar (24 km North; 40.86º N, 0.8º

cal characteristics and the patchy distribu-

E) to Peñíscola (51 km South; 40.35º N, 0.4º E).

tion of its biota (Steele 1989; González-Solís

It could be covered in a single day with an

and Shaffer 2009). The spatial and temporal

aircraft and the aerial surveys were carried

distribution of animals is the result from the

out monthly from April 2005 to March 2006.

combination of extrinsic processes, related

This area has a permanent upwelling due

to the influence of biotic and abiotic envi-

to the combination of the influence of the

ronmental factors, and intrinsic processes,

Liguro-Provençal-Catalan front, the sudden

related to population dynamics and intra-

broadening of the continental shelf and the

specific interactions (Bellier et al. 2010). Mo-

source of nutrients from the Ebro river ru-

reover, the spatial and temporal distribution

noff (Palomera 1992; Arcos 2001). The high

of seabirds is scale-dependent and patchy

productivity of the area supports an impor-

over a range of spatial and temporal sca-

tant fishing fleet which is a key feeding sou-

les (Hunt and Schneider 1987; Kotliar and

rce for breeding and wintering seabirds in

Wiens 1990; Fauchald et al. 2000), which is

the Ebro Delta (Arcos 2001; Arcos et al. 2008;

explained under the hierarchical patch dy-

Cama 2010). Moreover, the Ebro Delta is a

namic theory (Kotliar and Wiens 1990; Allen

wetland of international importance inclu-

and Hoekstra 1991; Wu and David 2002).

ded in the Ramsar Convention since 1993.

In a hierarchical patch dynamic system,

With 320 km2, it is the second most impor-

one would expect large-scale patterns to

tant wetland of the Western Mediterranean

be more stable and predictable because of

after the Camargue in France and the second

a high correlation with environmental va-

most important from the Iberian Penin-

riables that define a potential habitat (Hunt

sula after Doñana. The rice fields, lagoons,

and Schneider 1987; Bellier et al. 2010). At

saltpans and beaches of the Ebro Delta pro-

smaller spatial scales one might expect less

vide a variety of habitats for breeding and

predictable spatial patterns because smaller

wintering birds but also a stopover point for

patches with high densities of organisms

large numbers of migratory birds. In global,

are the result of a particular combination of

ca. 400 species of birds can be found in the

circumstantial variables that create a tem-

area (Bigas 2012); 18 of which could be de-

poral preferential habitat within the poten-

tected at sea from the aircraft (Table 2).

tial habitat (Bellier et al. 2010). Translating these theoretical concepts

Modelling tools

to applied ecology, the optimal assessment tools for seabirds-OWF interactions must

Data on the distribution of seabirds at

take into account this differential effect of

sea can be a useful tool for conservation and

spatial and temporal scales. At large-scale

environmental assessment whenever the

assessments, the observed distribution pat-

spatial data from seabird surveys represent

terns can be considered stable in time and a

a general pattern and not only a punctual

proxy to potential habitats and thus optimal

‘snapshot’ of a highly dynamic system (Fau-

for the demarcation of key areas of protec-

chald et al. 2002).

tion (e.g. Important Bird Areas, IBAs) and key

Despite its superficial homogeneity, the

areas for offshore wind energy development.

sea is a heterogeneous environment because

At regional or local-scale assessments, the

General introduction observed clustering of seabirds must be

dely applicable. Hence, instead of develo-

evaluated in its full temporal and spatial

ping a new index, this thesis examines the

variability as a preferential habitat and,

method in depth and suggests a refinement

consequently used to quantify -in terms of

on the mathematical formulation (Chap-

probability- the risk exposure to OWF.

ter 1) and makes recommendations on the optimal application of the index for its uti-

Large scale: Vulnerability Index

lization in any Strategic Environmental Assessment (Chapter 2).

The Strategic Environmental Assessment integrates data at broad scales; therefore,

Regional and local scale

we can assume that the temporal scale is not a priority whenever data from diffe-

At smaller scales, ornithological Environ-

rent years or periods can be pooled. Seabird

mental Impact Assessments usually focus

distributions might have different patterns

in migratory corridors and seabird habitat-

depending on the stage of life cycle (winte-

use strategies and the processes that are

ring, migrating and breeding) but in global

expected to influence seabird occurrence

their distribution is expected to be spatia-

or the availability of their prey. As in large

lly and temporally predictable (Fauchald et

scales, bird densities are used as a proxy of

al. 2002). In other words, at strategic levels

bird habitat to assess risk exposure to ha-

the main concern regarding seabirds-OWF

bitat loss or disturbance. Despite this is a

assessment is the spatial overlap of seabird

common practice, the effectiveness of this

distribution with key developing areas of

method is compromised by the assumption

OWF. This overlap is usually evaluated with

that at these scales the observed data fo-

the selection of presence/absence maps of

llows a normal distribution. In fact, animal

a few flagship species expected to be highly

count data are seldom normal. Seabird ae-

vulnerable to OWFs and general density

rial and boat surveys data are zero-inflated

maps with the global numbers of seabird

(Broek 1995; Pearce and Ferrier 2001; Barry

counts at sea. In this context, it becomes

and Welsh 2002) with a positive skew of

appropriate to apply an index to integrate

non-zero values, i.e. many counts of low to

all these information layers into a summari-

intermediate density and very few counts of

zing one.

high density (Fauchald et al. 2002; Mcsorley

Garthe and Hüppop (2004) proposed the

et al. 2005; Certain et al. 2007). For this rea-

Wind Farm Sensitivity Index (WSI) to map

son, the explicit consideration of temporal

the vulnerability of seabirds to offshore

and spatial variability of seabird occurrence

wind farms in a sea region. This index esti-

and density is necessary in any EIA to de-

mates the vulnerability of each species ac-

sign ecologically sound management strate-

cording to their sensitivity to collision risk,

gies at regional and local scales (Tobin 2004;

disturbance and their demographical and

Certain et al. 2007).

conservation status. This value is later com-

The third and fourth chapter of this

bined with the spatial abundance of each

thesis tackle this spatio-temporal varia-

species to obtain a vulnerability map.

bility through the application of Taylor’s

This method is general, simple and wi-

power law and the analysis of individual’s

METHODOLOGICAL APPROACH movements respectively.

Aggregative patterns

Individual tracking The second approach deals with spatiotemporal variability from an individual-ba-

Taylor’s power law is based on an em-

sed perspective. Seabird surveys are cons-

pirical relationship that expresses the de-

trained in space by the arrangement of the

pendency between the average measured

survey transects, and constrained in time by

in one point and the variance of the mea-

the moment of the day at which each transect

sures in this point (Taylor 1961). Although

is surveyed and the necessity of daylight for

its mathematical foundations have been

the counts. Individual tracking of seabirds,

widely discussed (Kendal 2004), this rela-

instead, is not restricted in either space or

tionship has been demonstrated for more

time. When applying multivariate mode-

than 400 species in taxa ranging from pro-

lling, temporal patterns in space use can be

tists to vertebrates (Kilpatrick and Ives 2003)

described not only monthly or seasonally

and it is true for both spatial data (repeated

but also within a circadian cycle. Moreover,

measures adjacent in space) and temporal

if a transmitter provides frequent locations,

data (repeated measures in a point over

it is possible to quantify the geometric pro-

time) (Taylor and Woiwod 1980, 1982; Taylor

perties of the path of a tagged animal (e.g.

et al. 1980). When calculated through spa-

speed, heading, turning angles)(Patterson et

ce, Taylor’s Power Law provides a measure

al. 2008). A particular combination of values

of the strength of the aggregative response

for these properties can be interpreted as a

of organisms (Jiménez et al. 2001; Östman,

behavioural mode (feeding, travelling bet-

2002). When calculated through time, it can

ween foraging patches, resting, etc.). Under

be used as an index of the temporal variabi-

this assumption, State-Space Models (SSM;

lity of the spatial distribution of organisms,

Jonsen et al. 2003) can be applied to calcu-

highlighting recurrent and occasional pre-

late the probability of an animal being in a

sence areas (Certain et al. 2007). Therefore

particular behavioural mode and later indi-

this method provides a useful framework

vidual decisions can be linked to population

to study the spatio-temporal variability in

distribution and applied to risk assessments

seabird surveys.

(Turchin 1998).

Research papers

Research papers resum L’energia eòlica marina és una de les fonts d’energia renovable més prometedores per al futur. No obstant això, l’establiment d’aquestes instal·lacions requereix una avaluació d’impacte detallada, en particular pel que fa a les poblacions d’aus marines. Fins on sabem, l’únic marc de treball disponible a gran escala que permet aquesta avaluació és el plantejament desenvolupat al 2004 per Garthe i Hüppop. Conceptualment es tracta d’un treball molt sòlid, però el tractament matemàtic dels conceptes no es correcte i cal que sigui actualitzat. L’estudi que es presenta en aquest capítol revisa el treball de Garthe i Hüppop destacant els supòsits en els que es fonamenta i els problemes d’interpretació associats als mateixos. Fet això, fem una reestructuració exhaustiva del marc matemàtic fent que sigui correcte tant en el seu aspecte formal (matemàtic) com en la seva interpretació ecològica. D’aquesta forma l’índex, ja de per sí molt últil es torna més adaptable i pràctic. La revisió que es presenta en aquest treball diferencia explícitament el risc de col·lisió i el risc de pertorbació; es basa en els desenvolupaments teòrics més recents d’ecologia de comunitats; i proposa una integració seqüencial dels efectes des d’un nivell d’espècie fins al de comunitat. Mitjançant el cas d’estudi de les aus marines del Golf de Biscaia (França) il·lustrem les limitacions del plantejament anterior i la utilitat de la nostra revisió de l’índex. En general, el marc refinat proporciona informació clara, complementària i sense ambigüitats que ha d’ajudar als gestors de l’àmbit marí en la pressa de decisió sobre les localitzacions òptimes per als parcs eòlics marins i l’avaluació dels possibles impactes que es pot esperar en determinades zones. A més, el mètode a través del qual integrem la vulnerabilitat de les espècies a nivell de la comunitat és de caire general, i podria ser fàcilment adaptat a qualsevol tipus d’impacte i comunitat animal més enllà del cas particular de les aus i l’energia eòlica marina

JOURNAL REFERENCE Paper to be submitted to the Journal of Applied Ecology

A refined methodology to estimate the vulnerability of seabird community to the establishment of offshore wind farms Grégoire Certain1, Isadora Christel2,3, Benjamin Planque 1 and Vincent Bretagnolle 4 1

Institute of Marine Research (IMR). PO box 6404, 9294 Tromsø,

Norway 2

Institute for Research on Biodiversity (IRBio) and Departament de Biologia Animal, Universitat de Barcelona (UB). Diagonal 645, E-08028 Barcelona, Spain.

Museo Nacional de Ciencias Naturales, Consejo Superior de Investigaciones Científicas. C/ José Gutiérrez Abascal 2, 28006, Madrid, Spain. 4 Centre national de la recherche scientifique (CNRS). FR–15 79360 Villiers en Bois, France. A bstract 1. Offshore wind energy is amongst the most promising renewable energy sources. However, the proper establishment of wind farms requires thorough impact assessment, in particular with regard to seabird populations. The only available framework for such assessment is the approach developed in 2004 by Garthe & Hüppop. Although conceptually sound, the approach is mathematically incorrect and needs to be updated. 2. This study Garthe & Hüppop’s approach, highlighting the hidden assumptions and interpretation problems associated to it. Our refined framework explicitly disentangles collision and disturbance risk, draws upon recent theoretical development in community ecology and proposes a sequential integration of the impact at the species and community level. To illustrate the pitfalls of the previous approach and the usefulness of our refined framework we use the study case of the seabird populations in the continental shelf of the Bay of Biscay, France. 3. The proposed diagnostic panel disentangles vulnerability in four components that provide managers with the pieces of information they would need to take informed decisions without implicitly masking some components of the impact. 4. Synthesis and applications: Overall, the refined framework provides clear, complementary and unambiguous information to managers about offshore wind farms location and the kind of impact to be expected. Furthermore, the method, which integrates vulnerability from species to community level, is general and has a potential application to any discipline where vulnerability assessments are needed.

Research papers INTRODUCTION

and the vulnerability index is weighted by population abundance.

The development of marine offsho-

In this study, we briefly review the origi-

re wind farms has increased significantly

nal approach (Garthe and Hüppop 2004) and

worldwide in the last decades following the

point towards its weaknesses. Then, we at-

need of decreasing carbon footprint through

tempt to solve the problems by proposing a

the exploitation of renewable energy sour-

new methodological approach that (1) expli-

ces (Punt et al. 2009; EWEA 2012). Ecological

citly distinguishes between risk factor and

effects of locating wind farms offshore can

aggravation factor, (2) allows the separation

be both detrimental and beneficial (Punt et

between collision risk and disturbance risk,

al. 2009; Snyder and Kaiser 2009). Among the

and (3) draws on recent development in

detrimental effects, wind farms are potential

functional diversity (Leinster and Cobbold

treats to marine seabirds in two different as-

2012) to produce and map a community vul-

pects: increased mortalities due to collision

nerability index based on the local relative

risk (Desholm and Kahlert 2005; Chamber-

frequencies of species within the seabird

lain et al. 2006; Fox et al. 2006), and increased

community. Finally, to compare the perfor-

energy expenditure and habitat loss through

mance of the refined approach with the ori-

disturbance (Exo et al. 2003; Garthe and Hüp-

ginal index, we apply the refined methodo-

pop 2004; Drewitt and Langston 2006; Mas-

logy to the Bay of Biscay by using data

den et al. 2010).

collected during an extensive seabird aerial

Spatial planning of wind farms requires

survey in the Bay of Biscay, France (Bretag-

quantitative assessment of these threats.

nolle et al. 2004; Certain et al. 2007; Certain

Garthe & Hüppop (2004) proposed the Wind

and Bretagnolle 2008).

Farm Sensitivity Index (WSI), the first index that mapped seabird community vulne-

METHODS

rability to win farms to inform on marine spatial planning of offshore wind energy.

Reviewing the Wind farm Sensitivity Index

The methodological framework combines two sources of information. The first sour-

The WSI proposed by Garthe & Hüppop

ce is an estimate of the vulnerability of each

(2004) has been successfully implemented

seabird population in a study area, based on

to detect areas where the seabird commu-

behavioural and demographical traits and

nity would be most vulnerable to the es-

conservation status. The second source is

tablishment of a wind farms (Garthe and

the spatial distribution of each population,

Hüppop 2004; Leopold and Dijkman 2010;

based on extensive at-sea surveys. Although

Christensen-dalsgaard et al. 2011). It is based

the method is general, simple, and widely

on a Seabird Sensitivity Index (SSI), designed

applicable, its mathematical formulation

to reflect the vulnerability of each seabird

contains hidden assumptions that might be

species to the establishment of offshore

problematic and lead to incorrect estimates

wind farms, and the at-sea abundances of

of vulnerability as well as biased identifica-

each seabird species (A). Let us consider an

tion of key concern areas. In particular, di-

area discretized in a succession of j = 1…L

fferent risks factors are given equal weight,

locations and populated by a set of i = 1…S

the index combines collision and disturban-

seabird populations. We can write:

ce risk with a multiplicative relationship

CHAPTER 1

(Eq.1)

These assumptions might be difficult to hold in a number of cases. First, for a given risk type, risk factors might not be independent, nor additive. We can distinguish two categories of risk factors: those that are di-

(Eq. 2)

rectly associated to the risk itself (i.e. time spent flying and flight altitude in the case of

Where the set of fri represent r = 1â&#x20AC;Ś9 risk (Eq. 3) factors for the ith seabird species. These risk

the collision risk) and those that are aggra-

factors can be grouped into three risk types:

vrability and nocturnal activity). This con-

risk related to collision (r = 1, 2, 3, 4), risk (Eq.re4)

ceptual difference is important: aggravation

lated to disturbance (r = 5, 6) and risk related

factors are not important in themselves, but

to the overall sensitivity of species (r = 7, 8,

they can increase a risk that already exists.

9). Table 1 synthesizes the different risk fac-

Following this reasoning, disturbance by

tors, each being measured on a relative scale (Eq. 5) ranging from 1 (minimum risk) to 5 (maxi-

ship and helicopter traffic (f5) can be viewed

mum risk). As it is formulated, the current

ty (f6) only matters if the species is disturbed

estimate of SSI and WSI makes the following

in the first instance. Finally, biogeographical

vation factors of the risk (i.e. flight manoeu-

as the real risk factor, while habitat flexibili-

assumptions:

population size (f7) and European conserva-

A1: All the risk factors associated to a gi(Eq.the 6) ven risk type are equally weighted, and

tion status (f9) both determine the overall

relationship between the risk factors of a gi-

while adult survival rate (f8) correlates to its

ven risk type is additive.

capacity to replenish the population if some

sensitivity of a species to any kind of impact,

A2: Each risk type is equally weighted,

increased mortality is experienced, hence f8

and the relationship between risk types is

can be defined as an aggravation factor. If we

multiplicative.

recognize that risk factors are not equal and

A3: The local importance of a given

they are subject to some hierarchy between

seabird species in the local measure of the

primary risk factors and aggravation factors,

vulnerability of the seabird community is

then the mathematical formulation of SSI

proportional to its local log abundance.

should be adapted to take into account the

Table 1 Risk factors according to which species vulnerability to wind farms is assessed. Detailed definitions for each risk factor can be found in Garthe & HĂźppop (2004).

Research papers potential caveats from assumption A1.

weight at the species level, and using the

Assumption A2 is conflictive as it sets

log-abundance of seabirds instead of their

that the three risk types (collision risk, dis-

abundance can be criticized. Applying the

turbance risk and overall species sensitivity)

weight at the species level introduces con-

are equally important and, furthermore, it

tradictions within the framework, because

assumes that they interact multiplicatively.

rare species are first up-weighted in the SSI

It is difficult to measure the relative impor-

through the factors f7 and f9, and then down-

tance of collision risk over disturbance risk,

weighted in the WSI because rare species

which justifies considering them as equal by

are usually the less abundant species. Con-

default despite collision mortality is often

versely, the importance of abundant species

considered to be the most important hazard

will be first down-weighted in the SSI, and

(Fox et al. 2006; Desholm 2009; Christel et al.

then up-weighted in the WSI. With the cu-

2012). Having a multiplicative relationship

rrent formulation, there is no control on the

between species overall vulnerability and

magnitude of up-weight / down-weight that

both collision and disturbance risk is also

each species will experience, which hinders

correct because the multiplication ensures

any interpretation of the spatio-temporal

that both risks are weighted by the overall

variations of the WSI. Finally, the use of log

sensitivity of each species. However, having

abundance assumes that the importance of

a multiplicative relationship between colli-

a single seabird in a location decreases ex-

sion and disturbance is arguable. Both risks

ponentially as the total number of seabird in

do not really depend on each other but are,

that location increases. A single individual

in fact, two independent aspects of the im-

in the middle of nowhere will have more

pact of offshore wind farms on seabirds.

weight, in proportion, than an individual

They have different ecological consequen-

located in a flock of one hundred seabirds.

ces (Drewitt and Langston 2006; Fox et al.

This assumption has neither ecological nor

2006) and might lead to very different ma-

management support.

nagement measures. By using a multiplica-

To take into account the potential ca-

tive relationship, if collision risk is high and

veats associated to assumptions A1, A2 and

disturbance risk is low, the resulting risk will

A3, we propose the refinement of the WSI

be much more lowered than it would with

framework in the following sections.

a simple additive relationship. Despite an additive relationship could be a solution,

Definitions

we question its usefulness. So far, there are no quantitative studies on the relative im-

Before presenting the mathematical re-

portance of the two risks, thus, we recom-

finement, it is useful to provide some clear

mend a sound option: the independent exa-

definitions and point toward the level of or-

mination of each collision and disturbance

ganisation at which they apply. Measures of

risk maps for decision-making until we can

risk and vulnerability will successively cross

measure their relative importance.

three levels of organisation: individuals, spe-

Assumption A3 gives more weight to

cies, and community. In this work, the word

species locally abundant. The intuitive no-

species is fairly equivalent to the word po-

tion behind this assumption is to prevent

pulation because most impact assessment

the installation of wind farms in areas whe-

are concerned with a delimited area and fo-

re seabirds aggregate. However, applying the

cus on the population of species within that

CHAPTER 1 area, not on the whole biogeographical distribution of the species.

Refining the Wind Farm Sensitivity Index

We will refer to a risk as a measure of the probability that an individual of a given

Combining factors with a power function

species suffers a given impact. For example, individual of a given species collides with a

Let us denote r the relative estimate of (Eq.1) a given risk, and let us assume that r is the

wind farm. We will use the term sensitivity to

combination of two factor: a primary risk

refer to the overall sensitivity of a given spe-

factor, a, and an aggravation factor, g. We

cies to any impact. We will use vulnerability

propose to link r to a and g through the fo(Eq. 2) llowing relationship:

collision risk refers to the probability that an

when the individual risk of suffering a given impact is integrated at a higher organisation level, distinguishing two levels: species vul-

(Eq. 3)

nerability and community vulnerability. Individual risks and population sensitivity will be estimated as a function of factors,

(Eq.g=0, 4) Under this formulation, r=a when

i.e. quantitative or semi-quantitative ele-

and then progressively increases as g in-

ments selected to measure a particular risk

creases. The parameter c is a measure of the

or the overall population sensitivity. We will

influence of g over r: the smaller the c, the

make the distinction between primary factors, i.e. factors directly controlling a risk or

more influence g will have on r (Fig.1). We (Eq. 5) suggest to use c=0.5 as a default and stron-

sensitivity, and aggravation factors, i.e. factors

gly recommend avoiding too small c values

that contribute to increase an already exis-

(i.e. <0.1, Fig.1). Under this formulation, r re-

ting risk or sensitivity.

mains bounded between 0 and 1.

All factors, risks, sensitivity and vulnerability measures will be expressed as relative probabilities, ranging between 0 and 1. A va-

6) Collision risk (ci ), disturbance risk (di ) and(Eq. species sensitivity (si )

lue of 0 indicates no risk, sensitivity or vulnerability, and a value of 1 is interpreted as

Let us denote the collision risk for the ith

maximum risk, sensitivity or vulnerability.

species ci , the disturbance risk di, and the

As these elements are all expressed on the

species sensitivity si. To obtain an estimate

same scale, they can be conveniently combi-

for each one, we rely on the set of estima-

ned through either averaging or multiplica-

ted risk factors for each species ff=(1...9)i, . The-

tion. We will use averaging when the values

se factors are produced following Garthe &

to be combined do not interact. We will use

H端ppop (2004) work and therefore their va-

multiplication when the values to be combi-

lues range from 1 to 5. As Eq.3 requires fac-

ned interact.

tors comprised between 0 and 1 they need to be rescaled (i.e. divided by 5) Collision risk results of the combination of the 4 first factors f1i to f4i (Table 1). Percentage of time spent flying (f1i) and percentage of time spent at the wind farm altitude when flying (f2i) can be seen as primary risk factors. Flight manoeuvrability (f3i) and noc-

Research papers

Fig.1 Theoretical effect of an aggravation factor (g) on a risk estimate (r) for different values of parameter c. Parameter c modulates the influence of the aggravation factor on the primary risk factor (a). In these figures, r = a when g = 0. (a,b) If c is low, then r is strongly dependent on the aggravation factor, whatever value takes a at the starting point of the curve. (c,d) If c is high, g has a lower effect on r.

turnal activity (f4i), on the other hand, can be seen as aggravation factor. ci is therefore obtained by applying Eq.3 with ai=f1i f2i and

and gi=f8i

Species vulnerability (vi )

gi=(f3i + f4i )/2. We use a multiplicative relationship between f1i and f2i because we as-

Once ci, di and si have been defined, they

sume they interact and we use an additive

can be combined to get an estimate of the

relationship between f3i and f4i because we

overall species vulnerability vi to wind farm.

assume they do not interact.

What we propose is to calculate the spe-

The disturbance risk di is the combination of a primary risk factor, the intensity of

cies vulnerability to a particular risk as the (Eq.1) product of the individual risk and species

the behavioural response to anthropic acti-

sensitivity. Then, calculate the overall spe-

vity (f5i), and an aggravation factor, the flexi-

cies vulnerability as a weighted mean of the

bility of habitat use (f6i). Therefore, di is ob-

risk-specific species vulnerabilities. In the (Eq. 2) context of seabird-wind farms interactions,

tained by applying Eq.3 with ai=f5i and gi=f6i. The species sensitivity si results from the combination of the 3 last factors f7i to f9i. The

this leads to the following expression: (Eq. 3)

biogeographical population size (f7i) and the species conservation status (f9i) can be seen as non-interacting primary risk factors. The

(Eq. 4)

natural survival rate of the species (f8i) can be seen as an aggravation factor. Therefore, si is obtained by applying Eq.3 with ai=f7i+f9i (Eq. 5)

CHAPTER 1 αc and αd are risk-specific weights contro-

weight to the most vulnerable species. Both

lling the influence of each risk. Setting αc = αd = 0.5 would mean that collision and dis-

the formulations of Hill (1973) and of Leins(Eq.1) ter & Cobbold (2012) contain a parameter, q,

turbance vulnerability are equally weighted.

which controls the sensitivity of the diver-

In our framework, vi is the direct equivalent

sity metric to the weighting parameter, i.e.

to the SSIi (Eq. 2).

in the case of Leinster & Cobbold (2012). The 2) greater is q, the higher is the weight of(Eq. simi-

Community vulnerability (Vj )

(Eq.1)

lar species over dissimilar ones. By adapting the diversity formulation to the vulnerability (Eq. 3)

The final step is to integrate the vulne-

context, the quantity (1-vi ) will be close to

rability of several species into a measure of

zero when the species are highly vulnera(Eq. 4) ble. As we precisely wish to give maximum

the vulnerability of a whole community, as (Eq. 2) it was originally attempted through the WSIj (Eq.1).

weight to the most vulnerable species, we set q=0. After replacing these parameters in

3) To do so, we build upon the recent (Eq. deve-

the original formulation and setting a spatial

lopment of Leinster & Cobbold (2012) that

context where community data are available (Eq. 5)

modifies the classical estimate of Hill’s di(Eq. 4) versity (Hill 1973) to take into account species similarity:

over j = 1…L locations, the overall community vulnerability to wind farms is written as follows:

(Eq. 6)

(Eq. 5)

Where pi is the relative frequency of the

Vj provides an estimate of the overall

ith species, and is a measure of the simila-

vulnerability of a community to offshore

rity between an individual of the ith species (Eq. 6) and an individual taken at random in the

wind farms in effective species number. It

community. is expressed between 0 (com-

lly abundant and fully vulnerable species

pletely dissimilar) and 1 (identical) and it is

that composes the community in a particu-

usually measured through a set of traits for

lar location. This formula can also be used

each species, as in classical functional di-

to measure the community vulnerability

can be interpreted as the number of equa-

versity studies (Leinster and Cobbold 2012).

to a given risk. Substituting vi for ci gives a

This index produces a diversity measure in

measure of the community vulnerability to

effective species number, i.e. the number of

collision, and substituting vi for di gives a

equally abundant species required to obtain

measure of the community vulnerability to

the same diversity measure. This is a re-

disturbance. This way, the overall communi-

commended practice as it greatly eases the

ty vulnerability map can be partitioned into

interpretation of the index (Tuomisto 2010;

each risk component.

Leinster and Cobbold 2012). The introduction of the term gives more weight to the

Seabird abundance at each location (Aj)

highly dissimilar species. In our framework, replacing by (1-vi ) will

In the original framework, the SSIi was

produce a diversity measure that gives more

multiplied by log(Aij) where Aij stands for

Research papers the abundance of each seabird species at

scheme was systematic with perpendicu-

each location, and summed over the spe-

lar transects separated by ca. 20km of each

cies (Eq.1). This unfortunately leads to mi-

other. Seabird observations were collected

sinterpretations the information concer-

continuously along the transect, including

ning abundance and species composition

species identification and number of indivi-

are mixed together. On the contrary, Vj fully

duals. The sampling design covered homo-

account for species composition and leaves

geneously the entire study area (ca. 100 000

abundance aside. The total seabird abun-

km2). For data processing, the transects were

dance at each location A.j is therefore a na-

sliced into segments of 20km, and the relati-

tural complement to Vj. It can be computed

ve abundance of each species, i.e. number of

from survey data and should be systemati-

counted individuals, was reported. To ease

cally presented together with Vj.

the comparison with previous studies, we use the whole ROMER dataset and the PEL-

Case study: Seabird populations in the Bay of Biscay.

GAS dataset from 2003 to 2008. Based on ROMER and PELGAS records, we first established the list of the 30 seabird

We applied both the original and refined

species identified in the surveys of the

framework to seabird populations on the

Bay of Biscay (Table S1). We also defined 7

French continental shelf of the Bay of Biscay,

groups for unidentified observations, assu-

which has been extensively sampled though

ming a species composition based on the

a series of aerial (‘ROMER’) and ship-based

species proportions of identified sightings

(‘PELGAS’) surveys (Bretagnolle et al. 2004;

(Table S1). For each species, the risk factors

Certain and Bretagnolle 2008; Certain et al.

ff=(1...9)i identified by Garthe & Hüppop (2004)

2011). These surveys resulted in a series of

were documented (Table S1). If possible, we

studies focusing on the spatial structure,

used the original values of Garthe & Hüppop

variability, and prey-predator relationships

(2004), otherwise we scored the species ac-

(Certain et al. 2007, 2011; Bellier et al. 2010,

cording to discussion carried out during

2012; Chadœuf et al. 2011). Details on both

expert meetings. For groups of unidentified

survey methodologies can be found in Cer-

seabirds, we used average values, weighted

tain et al. (2007), Certain and Bretagnolle

by species proportions in each group (Ta-

(2008) and Certain et al. (2011).

ble S1). Based on the risk factors, ci, di and

During ROMER surveys, strip-transect

si were computed for the 30 species and the

aerial surveys covered repeatedly the Bay of

7 groups. Then, overall species vulnerability

Biscay in winter, from October 2001-to March

vi was computed, as well as the original SSIi.

2002, offering a first exhaustive snapshot of

To reveal how vi differs from SSIi, we looked

the extent and abundance of the wintering

at the difference between the species rank

population of seabirds in the Bay of Biscay.

according to SSIi and the species rank ac-

From 2003 onward, observers recorded top

cording to vi. We computed correlations bet-

predator data on board of the RV-THALASSA

ween differences in rank and ci, di and si to

during the PELGAS cruises that are carried

search which risk was responsible for the

out each spring in the Bay of Biscay. In the

observed differences.

aerial and the boat surveys, the sampling

CHAPTER 1

the four maps, geostatistical interpolation

Diagnostic panels for the Bay of Biscay

(Cressie 1993; Pebesma and Wesseling 1998) and kriging were carried out to ease the re-

The refined framework we have develo-

presentation and interpretation of the spa-

ped is not supposed to produce one single

tial patterns. We interpret these maps in the

map, but rather to produce a few interpre-

context of wind farm impact assessment,

table maps, each capturing one key element

and point to the location on the continental

to be considered to assess the impact of

shelf where the impact on seabird popula-

offshore wind farms on seabird populations.

tions would be minimised.

We propose to use a diagnostic panel composed of 4 maps, each related to a specific component of the potential impact of offs-

Comparing the refined diagnostic maps with the previous WSI

hore wind farms. The two first maps present the two distinct and fundamental elements

To illustrate the differences between the

of the impact assessment: the overall vulne-

original and refined approach, we also com-

rability of seabird community (Vj ) and the

puted the original WSIj maps, together with

total seabird abundance map (A.j). Then, to

simple summed log abundance maps (i.e.

further inform management, we show the

removing SSIi in Eq.1). Showing both maps

two risk-specific components of Vj , namely,

illustrates how taking into account the com-

the vulnerability to collision and the vul-

munity vulnerability in effective species

nerability to disturbance. These two maps

number differs from the final vulnerability

highlight how both risks contribute to the

map in the original framework.

overall community vulnerability. To present

Fig.2 Comparison between the original Species Sensitivity Index (SSI) and the proposed overall Species Vulnerability Index (vi )

Research papers RESULTS Some species were more affected than

Comparison between SSIi and vi

others by the change of indices. For example, Skuas are considered more vulnerable

Fig.3 ROMER-based diagnostic panel. The grey rectangle points to the area that has lower impact in both ROMER and PELGAS diagnostic panels

For the sake of comparison, both SSIi and

with vi than with SSIi, with a difference in

vi were scaled between 0 and 1. When plot-

rank between -2 and -10, depending on the

ting these scaled values against each other,

species. On the contrary, auks are conside-

it is clear that the two indexes generally

red less vulnerable with vi, with a difference

agree on the classification of species (Fig.2a),

in rank between 2 and 9 (Table S1, last co-

even though nearly all species are located

lumn). The correlation between the diffe-

above the 1:1 line, suggesting that on avera-

rences in the ranking of species according to

ge, a seabird species is considered more vul-

both metrics and the components of vi was

nerable by the vi than by the SSIi. This is evi-

positive with collision risk ci (0.45, p=0.005,

dent in Fig.2b, where we see that the scaled

df=35), negative with disturbance risk di

distribution of the two indices differs, with

(-0.42, p=0.009), and not significant with spe-

the distribution of scaled vi being closer to 1.

cies sensitivity si (p=0.38).

CHAPTER 1

the coast than disturbance-vulnerable com-

Diagnostic Maps for Seabirds in the Bay of Biscay

munities. The inspection of PELGAS-based panel

Figure 3 and 4 present diagnostic pa-

offers a slightly different picture. The loca-

nels for seabirds in the Bay of Biscay based

tion of high and low vulnerability areas are

on aerial (ROMER) and boat surveys (PEL-

roughly the same (Fig.3a and Fig.4a), except

GAS), respectively. The ROMER-based panel

for a localised patch of high vulnerability in

highlights two main areas where the seabird

the North East, around Belle-Ile en Mer that

community is the most vulnerable, the nor-

was not visible from the ROMER-based pa-

thwest area and the south-eastern area

nel. The abundance map differs more clearly

(Fig.3a). In addition, the wintering popula-

(Fig.4b), with high abundances restricted to

tion of birds is widely spread in the Bay of

the Northernmost and coastal areas. Finally,

Biscay, leaving only few areas where seabird

vulnerability to collision (Fig.4c) and distur-

abundance is low (Fig.3b). Furthermore, the

bance (Fig.4d) present a rather similar pat-

ROMER panel clearly shows that vulnerabi-

tern, even though vulnerability to collision

lity to collision and disturbance differs in

is much more spread than vulnerability to

space, highlighting that collision-vulnerable

disturbance.

communities are distributed further from

Fig.4 PELGAS-based diagnostic panel. The grey rectangle points to the area that has lower impact in both ROMER and PELGAS diagnostic panels

Research papers

Fig.5 (a,b) Original Wind Farm Sensitivity Index maps calculated according to Garthe & H端ppop (2004) (c,d) total seabird log abundance maps.

The examination of ROMER and PELGAS

Comparing Diagnostic panels with WSIj maps

diagnostic maps reveal some differences between the wintering and spring situa-

Figure 5 (a, b) displays WSI maps as pro-

tions, however, in each case, the central part

posed by the original framework. They glo-

of the Bay of Biscay, identified in Figure 3

bally provide consistent information with

and 4 as a grey rectangle, is characterised by

the refined framework, but with more simi-

low abundances, and low to moderate vul-

larities to the disturbance risk maps than

nerability. As a synthetic result of this im-

the collision ones. Furthermore, the spatial

pact assessment, we suggest this area as an

patterns displayed by the WSIj (Fig.5a, b) are

informed choice for the location of offshore

extremely similar to the one displayed by

wind farms, as it seems to minimize the im-

the simple sum of log abundances (Fig.5c,

pact in both ROMER and PELGAS scenarios.

d), indicating that SSIi has in fact a negligible effect on the impact assessment according to the original framework.

CHAPTER 1 DISCUSSION

assessment method. Indeed, the method can be easily modified if, for example, addi-

Usefulness of the refined framework

tional studies appear concerning the way risk factors interact, or the relative impor-

The approach originally developed by

tance of collision over disturbance.

Garthe & H端ppop (2004) to assess the poten-

A relevant characteristic of our approach

tial impact of offshore wind farms on seabird

is that we do not attempt to synthesize the

populations has several interesting aspects.

information into one single map. Rather, we

The clear identification of species-specific

try to disentangle the different components

risk factors and the method for scaling them

of the information to present it in an inte-

is undoubtedly useful to synthesize quanti-

grated way to the managers. This is a key

tative and qualitative ecological information

aspect of communication between scien-

for impact assessment. It allows identifying

tists and managers. While scientists usually

which species are subject to which risks, it

try to identify all the aspects of a problem,

is a catalyst for expert meeting groups and

managers seek simple answers and synthe-

it is a major methodological tool to reach a

tic visualizations. This is one of the reasons

consensus between scientists and mana-

for the proliferation of indicator-based ap-

gers. However, the way this information is

proaches (Niemi and McDonald 2004; Piatt et

integrated later on and combined with sur-

al. 2007; Einoder 2009). Our case study illus-

vey data is not optimal. Relevant informa-

trates well that indeed, complex informa-

tion is lost in this integration, for example

tion related to the spatial distribution of 30

collision risk, and the original mathematical

seabird species can be synthesized in a few

formulation would, in fact, result in taking a

set of maps. However, information reduc-

decision based only on summed log-abun-

tion has to be carefully designed and firmly

dance patterns instead of accounting for the

theoretically grounded. Reducing complex

additional information provided by the tho-

problems to a single scale or a big formula

rough documentation of all the risk factors

may result in an intractable mixing of in-

and the computation of the SSIi. The refined

formation that either becomes difficult to

framework solves these issues by explicitly

interpret or strongly underestimates crucial

defining primary risk factor and aggravation

aspects of a problem.

factors, by treating each risk separately, by integrating them sequentially first at spe-

Wind farm impact assessment in the Bay of Biscay

cies level and then at population level, and finally, by explicitly differentiating community composition from abundance.

Our study clearly identifies areas of high and low expected impact on seabirds for the

With the proposed diagnostic panel, we

establishment of offshore wind farms, and it

provide managers with all the pieces of in-

provides as well a qualitative assessment of

formation they would need to take informed

the kind of impact to be expected. However,

decision without implicitly masking some

the reader should be aware that the quality

components of the impact. Because the fra-

of such an evaluation depends on the quali-

mework is clearly mathematically defined

ty of the data. We have no doubt that ROMER

and all the assumptions are stated and writ-

and PELGAS surveys provided state-of-the-

ten, we provide scientist with a transparent

art data on seabird populations. However,

and tractable method to improve the impact

these surveys have spatio-temporal limi-

Research papers tations that need to be clearly stated. First,

plicit link between an identified impact and

both surveys focused on the continental

a biodiversity-related metric at community

shelf and therefore, they do not document

level we aim to contribute not only to the

with detail the coastal community, which is

field of offshore wind energy assessment

the reason why we do not map abundance

but also to numerous disciplines where vul-

or vulnerabilities near the coast. Second, the

nerability assessments are needed.

timing of the surveys also limits the interpretation of our results. The ROMER surveys

REFERENCES

focused on the wintering period, which is the period during which the seabird com-

Bellier E, Certain G, Planque B, Monestiez B, Bre-

munity reaches the largest numbers in the

tagnolle V (2010) Modelling habitat selection

Bay of Biscay (Certain et al. 2007). PELGAS

at multiple scales with multivariate geosta-

surveys offer the spring perspective, when

tistics: an application to seabirds in open sea.

some of the main seabird taxa present in

Oikos 119:988–999

winter have left the area for breeding (e.g.

Bellier E, Monestiez P, Certain G, Chadoeuf J, Bre-

auks). Therefore, we stress that the maps

tagnolle V (2012) Decomposing the heteroge-

presented in this study can only serve for

neity of species distributions into multiple

risk assessment during these time periods,

scales: a hierarchical framework for large-sca-

but further surveys should be carried out and analysed to provide a better picture of the impact for the whole year.

le count surveys. Ecography 35:839–854 Bretagnolle V, Certain G, Houte S, Métais M (2004) Distribution maps and minimum abundance estimates for wintering auks in the Bay of Bis-

Potential for generality

cay , based on aerial surveys. Aquatic Living Resources 17:353–360

The refined approach proposed in this

Certain G, Bellier E, Planque B, Bretagnolle V (2007)

paper has potential applications in a wide

Characterising the temporal variability of the

range of disciplines beyond offshore wind

spatial distribution of animals: an application

energy impact assessment. The adaptation

to seabirds at sea. Ecography 30:695–708

of the work of Leinster & Cobbold (2012) that

Certain G, Bretagnolle V (2008) Monitoring seabirds

we introduce (Eq.6) can in fact be applied

population in marine ecosystem: The use of

to estimate the vulnerability of any kind of

strip-transect aerial surveys. Remote Sensing

community to any kind of impact. If a spe-

of Environment 112:3314–3322

cies-specific vulnerability parameter (such

Certain G, Skarpaas O, Bjerke J, Framstad E, Lind-

as vi) is available and community distribu-

holm M, Nilsen J-E, Norderhaug A, Oug E, Pe-

tion and abundance data has been collec-

dersen H-C, Schartau A-K, van der Meeren G.,

ted, the application of Eq.6 is then straight-

Aslaksen I, Engen S, Garnasjordet P-A, Kvaloy

forward.

P, Lillegard M, Yoccoz N, Nybo S (2011) The Na-

Garthe & Hüppop’s method has already

ture Index: A general framework for synthe-

been adapted to evaluate other types of ha-

sizing knowledge on the state of biodiversity.

zards in different taxa (e.g. Noguera et al.,

PloS one 6:e18930

2010; Stelzenmüller et al., 2010; Sonntag et

Chadœuf J, Certain G, Bellier E, Bar-Hen a., Cou-

al., 2012), particularly the species-specific

teron P, Monestiez P, Bretagnolle V (2011) Es-

vulnerability calculation. By improving the

timating inter-group interaction radius for

method formulation and including the ex-

point processes with nested spatial structu-

CHAPTER 1 res. Computational Statistics & Data Analysis

the effects of European marine offshore wind

55:627–640

farms on birds. Ibis 148:129–144

Chamberlain D, Rehfisch M, Fox AD, Desholm M,

Garthe S, Hüppop O (2004) Scaling possible adver-

Anthony S (2006) The effect of avoidance rates

se effects of marine wind farms on seabirds:

on bird mortality predictions made by wind

developing and applying a vulnerabiblity in-

turbine collision risk models. Ibis 148:198–202

dex. Journal of Applied Ecology 41:724–734

Christel I, Certain G, Cama A, Vieites DR, Ferrer

Hill MO (1973) Diversity and Evenness : A Unifying

X (2012) Seabird aggregative response: a new

Notation and Its Consequences. 54:427–432

tool for offshore wind energy risk assessment.

Leinster T, Cobbold C (2012) Measuring diversity:

Marine Pollution Bulletin. DOI: 10.1016/j.mar-

the importance of species similarity. Ecology

polbul.2012.11.005 (In press)

93:477–489

Christensen-dalsgaard S, Lorentsen S, Hanssen

Leopold MFL, Dijkman EM (2010) Offshore wind

F, Systad GH (2011) Marine wind farms and

farms and seabirds in the Dutch Sector of

seabirds. Updated screening of potential con-

the North Sea: What are the best and the

flict areas on a national scale. NINA Report

worst locations for future develpment? Ins-

616. <http://mhk.pnnl.gov/wiki/images/8/83/

titute for Marine Resources & Ecosystem

Visual_and_Radar_Observations_of_Birds.

Studies.<http://www.we-at-sea.org/leden/

pdf> (accessed 24.10.12)

docs/reports/RL2-2%202007-003%20Effect%20

Cressie N (1993) Statistics for spatial data. J. Wiley, USA Desholm M (2009) Avian sensitivity to mortali-

of%20offshore%20wind%20farms%20on%20 seabirds%20in%20the%20Dutch%20Nord%20 Sea.pdf> (accessed 24.10.12)

ty: prioritising migratory bird species for as-

Masden E a, Haydon DT, Fox AD, Furness RW

sessment at proposed wind farms. Journal of

(2010) Barriers to movement: Modelling ener-

Environmental Management 90:2672–9

getic costs of avoiding marine wind farms

Desholm M, Kahlert J (2005) Avian collision risk at an offshore wind farm. Biology Letters 1:296– 298 Drewitt A, Langston RW (2006) Assessing the impacts of wind farms on birds. Ibis 148:29–42

amongst breeding seabirds. Marine Pollution Bulletin 60:1085–91 Niemi G, McDonald ME (2004) Application of Ecological Indicators. Annual Review of Ecology, Evolution, and Systematics 35:89–111

Einoder L (2009) A review of the use of seabirds as

Noguera JC, Pérez I, Mínguez E (2010) Impact of te-

indicators in fisheries and ecosystem mana-

rrestrial wind farms on diurnal raptors: deve-

gement. Fisheries Research 95:6–13

loping a spatial vulnerability index and poten-

EWEA (2012) The European offshore wind indus-

tial vulnerability maps. Ardeola 57:41–53

try key 2011 trends and statistics. European

Pebesma E, Wesseling CG (1998) Gstat: A program

Wind Energy Association. <http://www.ewea.

for geostatistical modelling, prediction and si-

org/fileadmin/ewea_documents/images/gra-

mulation. Computers & Geosciences 24:17–31

phs_maps_tables/EuOffwind2011.pdf> (accessed 24.10.12) Exo KM, Hüppop O, Garthe S (2003) Birds and offshore wind farms: a hot topic in marine ecology. Wader Study Group Bulletin 100:50–53

Piatt JF, Harding A, Shultz M, Speckman S, van Pelt T, Drew GS, Kettle A (2007) Seabirds as indicators of marine food supplies: Cairns revisited. Marine Ecology Progress Series 352:221–234 Punt MJ, Groeneveld R a., van Ierland EC, Stel

Fox AD, Desholm M, Kahlert J, Christensen TK,

JH (2009) Spatial planning of offshore wind

Petersen IK (2006) Information needs to sup-

farms: A windfall to marine environmental

port environmental impact assessment of

protection? Ecological Economics 69:93–103

Research papers Snyder B, Kaiser MJ (2009) Ecological and econo-

Stelzenmüller V, Ellis JR, Rogers SI (2010) Towards

mic cost-benefit analysis of offshore wind

a spatially explicit risk assessment for marine

energy. Renewable Energy 34:1567–1578

management: Assessing the vulnerability of

Sonntag N, Schwemmer H, Fock HO, Bellebaum J, Garthe S (2012) Seabirds, set-nets, and conser-

fish to aggregate extraction. Biological Conservation 143:230–238

vation management: assessment of conflict

Tuomisto H (2010) A diversity of beta diversities:

potential and vulnerability of birds to bycatch

straightening up a concept gone awry. Part 1.

in gillnets. ICES Journal of Marine Science

Defining beta diversity as a function of alpha

69:578–589

and gamma diversity. Ecography 33:2–22

CHAPTER 1

Table S1 Supplementary data with the list of the 30 detected seabird species and their proportion (Prop), the score of the nine vulnerability factors (f1...9), the Species Sensitivity Index (SSI) Collision risk (ci), disturbance risk (di), species sensitivity (si), Species vulnerability (vi) and species rank differences between SSI and vi (â&#x2C6;&#x2020;). See the text for details.

Research papers

resum L’Índex de Sensibilitat als parcs eòlics (WSI) de les aus marines és una eina feta en el context dels sectors alemanys del mar Bàltic i del Nord. Va ser creat amb la finalitat de proporcionar una eina de decisió a escales grans per a l’avaluació ambiental estratègica de l’energia eòlica marina. A continuació, es mostra com es pot millorar l’índex d’una Avaluació Ambiental Estratègica (AAE) per al desenvolupament d’energia eòlica marina en un context de gran escala. El WSI integra la informació basada en les densitats d’aus marines a la zona d’estudi amb un Índex de Sensibilitat Espècies (SSI) als parcs eòlics. Aquest índex es calcula tenint en compte nou factors, que es deriven dels atributs de les espècies que semblen definir la sensibilitat de l’ocell amb els parcs eòlics. Es van dur a terme censos des de barca. Després de calcular el SSI per a cada espècie que es troba en l’àrea d’estudi, s’aplica l’índex als mapes de densitats locals obtinguts mitjançant els censos des de vaixell. Per prendre una decisió interessa treballar amb el mínim nombre de mapes que sintetitzin completament la realitat ecològica d’un àrea. Aquest index disposa d’aquesta característica integradora i el fa especialment interessant en l’avaluació ambiental estratègica d’un àrea determinada. Atès que hi ha una manca d’informació per conèixer l’abast de l’impacte real dels parcs eòlics en alta mar, es recomana aquest índex com un mètode molt útil en la majoria d’estudis d’avaluació ambiental estratègica mentre no es desenvolupi una eina millor per aquest tipus d’avaluació.

JOURNAL REFERENCE Paper to be submitted to Ecological Applications

Wind farm Sensitivity Index for seabirds Assessing offshore wind energy development on the coasts of the Iberian Peninsula Isadora Christel1,2, Albert Cama3, Grégoire Certain4, J. M. Arcos3, J. Bécares3, B. Rodriguez3, I. Ramirez5, A. Meirinho5, J. Andrade5, David R. Vieites2,6 and Xavier Ferrer1 Institute for Research on Biodiversity (IRBio) and Departament de Biologia Animal, Universitat de Barcelona (UB). Diagonal 645, E-08028 Barcelona, Spain. 2 Museo Nacional de Ciencias Naturales, Consejo Superior de Investigaciones Científicas. C/ José Gutiérrez Abascal 2, 28006, Madrid, Spain. 3 SEO/BirdLife, Delegació de Catalunya, C/Murcia 2-8, Local 13, 08026 Barcelona, Spain. 4 Institute of Marine Research (IMR). PO box 6404, 9294 Tromsø, Norway 5 Sociedade Portuguesa para o Estudo das Aves. Avenida João Crisóstomo N18 4D | 1000-179 Lisboa – Portugal. 6 REFER Biodiversity Chair, University of Porto, CIBIO, Campus Agrário de Vairão, R. Padre Armando Quintas, 4485-661 Vairão, Portugal. 1

A bstract The Windfarm Sensitivity Index (WSI) for seabirds is a tool made in the context of the German sectors at the Baltic and North Seas. It was created in order to provide a tool of decision for a broad-scale Strategic Environmental Assessment for offshore wind energy. Here, it is showed how this Index can improve an Strategic Environmental Assessment (SEA) for offshore wind energy development in a large scale context. The WSI integrates the information based on the seabirds densities in the study area with a Species Sensitivity Index (SSI) to windfarms. Such Index was calculated taking into account nine factors, which derive from the species attributes that seem to define the bird sensitivity to windfarms. Boat surveys were carried out. After calculating the SSI for each species found in our surveys, we applied it to the maps of local densities obtained by means of boat surveys. When making a decision, the fewer number of fully explicative maps are always desirable. This integrative characteristic of the index makes it especially interesting in the environmental assessment of a proposed offshore windfarm. Since there is a lack of information to know the extent of the real impact of offshore windfarms, we recommend this index as a very useful method in most SEA until the moment we will have a better tool for assessment.

Research papers INTRODUCTION

Thus, seabirds seem a suitable indicator of the marine environment, and have become

Renewable energies are viewed as an en-

one of the keystones of the decision-making

vironmental benign alternative to the energy

process for the selection of optimal areas for

production based on fossil fuels (Inger et al.

national offshore wind development.

2009). The potential of the marine environ-

The Wind Farm Sensitivity Index (WSI)

ment, and particularly the offshore wind en-

was the first index that used seabirds to as-

ergy development, has received high atten-

sess at large scale the suitability or unsuit-

tion in the last years. Europe has become the

ability of a sea region for the construction of

world leader in offshore wind power with

offshore wind farms. The index, developed

a total of 1,371 offshore turbines totalling

by Garthe and H端ppop (2004), takes into

3,812.6 MW spread across 53 wind farms in

account two crucial points in the evalua-

10 countries by the end of 2011 (EWEA 2012).

tion of future impacts. First, it takes into

At a global scale, the shift to renewable ener-

account seabirds abundances and areas of

gies is widely accepted as a step to mitigate

high density which is the information com-

the effects of anthropogenically induced cli-

monly used to inform SEAs and EIAs on off-

mate change (King 2004; Rosenzweig et al.

shore wind energy. Second, the abundance

2008). At the local scale, however, the envi-

of seabirds is corrected by a specific value

ronmental impacts of wind energy develop-

(SSI, Species Sensitivity Index) that quanti-

ment must be carefully considered. Indeed,

fies the sensitivity of each seabird species to

the European legislation requires Strategic

the presence of an offshore wind farm or its

Environmental Assessments (SEAs) of na-

construction. This way the presence of few

tional wind energy plans impacts on wildlife

individuals of flagship species can be ac-

(Directive 2001/42/EC).

counted as well as the massive presence of

Among the different topics that SEAs must address, wind farms and birds interac-

common species, with no conservation concern but relevant because their numbers.

tions are an issue of great concern (Garthe

Despite the WSI relevance as a practical

and H端ppop 2004; Fox et al. 2006). Seabirds

assessment tool, the peer-reviewed papers

are susceptible to multiple anthropogenic

that have actually used it are adaptations

impacts in their migratory routes and forag-

of the index to evaluate other types of haz-

ing grounds (Anderson et al. 2003; H端ppop

ards (e.g. Noguera et al., 2010; Stelzenm端ller

et al. 2006; Louzao et al. 2006). In the case of

et al., 2010; Sonntag et al., 2012) and so far

offshore wind farms these potential impacts

there is no paper showing its application in

are direct mortality through collision, bar-

a different geographic area and only some

rier effects and foraging habitat loss (Fox et

reports (e.g. Leopold and Dijkman, 2010;

al. 2006).

Christensen-dalsgaard et al., 2011). A recent

Beyond conservation concerns, seabirds

revision on the mathematical formulation

have also become useful indicators to evalu-

of the WSI has presented a new calculation

ate the potential effects of human activi-

of the Index (Certain et al. en prep.) In this

ties in marine ecosystems (Piatt et al. 2007).

paper, we present the refined Wind farm

Compared to other marine species, seabirds

Vulnerability Index applied to the coasts of

are highly visible species with specific legal

the Iberian Peninsula, a study area that su-

protection frameworks and comprehensive

rrounds more than 7000 km and covers the

long-term studies of their distribution at sea.

continental coasts of Portugal and Spain.

CHAPTER 2 The Iberian Peninsula hosts the high-

order to make the Index more general and

est diversity of seabirds in Europe, mainly

applicable to other biogegraphic areas in Eu-

because its waters cover different biogeo-

rope and other continents; (ii) provide new

graphical regions. Among these species

Species vulnerability values to expand and

some have their breeding stronghold in the

improve the original SSI table by including

Iberian Peninsula (e.g. the Balearic Shear-

the diversity of Atlantic and Mediterranean

water Puffinus mauretanicus and Audouinâ&#x20AC;&#x2122;s

species detected in the area and (iii) make

Gull Larus audouinii) and are flag species be-

recommendations for the future develop-

cause their conservations status. Many of

ment of offshore wind energy in the Iberian

the species are listed in the Annex I of the

coasts with full awareness of ecological im-

European Community Birds Directive and

pacts.

are main targets of conservation projects (Ramirez et al. 2008; Arcos et al. 2009). So far,

METHOD

in the whole Iberian Peninsula there is only one experimental floating turbine installed

Study area & vessel-based data

in Portugal in June 2012. Therefore this is an opportunity to inform policy-makers and

At-sea seabird surveys were conducted

practitioners on how to design the optimal

in different vessel expeditions coordinated

zonation to allow a rational offshore wind

by the Spanish Ornithological Society (SEO/

energy development respectful with the

Birdlife) and the Portuguese Society for the

marine ecosystem.

Study of Birds (SPEA). The surveys were ca-

The main aims of this paper are to: (i) ap-

rried out from 1999 to 2011 covering the Spa-

ply the index refinement developed by Cer-

nish and Portuguese continental coasts (up

tain et al. (in prep.) and suggest some changes

to 100 nautical miles offshore) (Fig 1a). The

in the factors used to calculate the SSI in

total surveyed area covered more than 25000

Fig. 1 a) Situation Map. b) Study regions marked by solid lines; I: Spanish North Atlantic Ocean, II: Portuguese North Atlantic Ocean, III: Gulf of Cadiz and Alboran Sea and IV: Mediterranean Sea. b) Grid location

Research papers

Table 1 Total surveyed area (Km2) by year and month

Km2 (Table 1). Seabird counts followed stan-

farms using nine factors: flight manoeu-

dardized strip-transect techniques (Tasker

vrability, flight altitude, percentage of time

et al. 1984) adapted to the study area charac-

flying, nocturnal flight activity, sensitivity

teristics (Louzao et al. 2006). The observers

towards disturbance by ship and helicop-

covered a 300 m strip transect band at each

ter traffic, flexibility in habitat use, biogeo-

side of the boat when visibility and wind

graphical population size, adult survival

conditions were adequate. All seabirds ob-

rate, and conservation status (Garthe and

served within the survey transect were re-

Hüppop 2004). Each factor was scored on a

corded and summed into 10 minutes survey

5-point scale where 1 indicated low vulne-

bins.

rability and 5 indicated high vulnerability.

The study area had a latitudinal span of

Following Garthe and Hüppop methodology,

10 decimal degrees (from 35º N to 45º N) and

when no empirical data was available, the

a longitudinal span of 15 decimal degrees

factors (5 out of 9) were given a subjective

(from 11º W to 4º E). It was divided in a re-

value partially based on bibliography -when

gular grid at four different scales (2º, 1º, 0.5º

available- and the authors experience on

and 0.25º) (Fig.1c). Four main regions have

the species. These scores where then sub-

been considered for local discussion accor-

mitted for assessment to 10 independent

ding to their oceanographic features and po-

experts with at-sea experience. After the

litical boundaries: the Bay of Biscay and the

independent

Galician Atlantic coast (Region I), the Portu-

where revised. When close species with si-

guese Atlantic coast (Region II), the Gulf of

milar characteristics had different values

Cádiz and Alboran Sea (Region III) and the

the experts where consulted again, and the

Mediterranean Sea (Region IV) (Fig.1b).

values were corrected if a consensus was

evaluation,

species

scores

reached. The nine factors are shortly descri-

Species vulnerability calculation

bed below, with emphasis on those factors with changes from its original definition. A

The original WSI is derived from distri-

more detailed description of the factors can

butional data of seabirds’ counts at sea and

be found in the original paper (Garthe and

a Species-specific Sensitivity index (SSI). SSI

Hüppop 2004).

evaluates the species’ vulnerability to wind

CHAPTER 2 (f1) Percentage of time flying This factor aims to assess how much

evaluated by the experts.

(f5) Disturbance by ship and helicopter traffic

time a species present in a wind farm area is susceptible to collision. This factor was

The ship and helicopter traffic during

obtained from the behavioural data collec-

construction and maintenance of wind

ted during the surveys. Species were scored:

farms is expected to have some disturbance

1, if 0-20% of the individuals were flying; 2,

effect on species provoking escape, avoidan-

21-40%; 3, 41-60%; 4, 61-80% and 5, 81-100%.

ce or fleeing behaviours. There is almost no

After collecting and ranking behavioural

information on the issue; hence the factor

data for each species, some discrepancies

was evaluated subjectively from hardly any

where found between similar species. The-

behavioural response (score 1) to strong be-

se differences were artefacts of data related

havioural reactions (score 5).

to differential detection. In these cases the scores where equalized for the group using

(f6) Flexibility in habitat use

the better sampled species. This factor takes into account the ha-

(f2) Flight altitude

bitat preferences of species. Those species occupying large sea areas and no specific

This factor estimates how often a spe-

habitat preferences (e.g. gulls) are expected

cies flies within the range of the blades of

to be less sensitive to offshore wind farms

the turbines. The altitudes were classified

than those species relying on specific habi-

as follows: 1, 0-5m; 2, 5-10m; 3, 10-20m; 4,

tat features (e.g. sea ducks feeding on banks

20-50m and 5, 50-100m. The original factor

on shallow grounds). Therefore species were

was based on real data from flight altitude

classified from very flexible in habitat use

assessments. Since this information was

(score 1) to reliant on specific habitat cha-

not available in our surveys the experts were

racteristics (score 5) and again evaluated by

asked for the most frequent altitude class.

the experts.

(f3) Flight manoeuvrability

(f7) Biogeographical population size

This factor takes into account the flight

Population sizes were obtained for each

ability of a species to avoid collision with

species from Birdlife publications (Bird-

wind farms at sea. Species were classified

Life International 2004, 2012). Species were

from very high flight manoeuvrability (sco-

scored: 1 for populations exceeding 3 mi-

re 1) to low flight manoeuvrability (score 5),

llion individuals; 2 for 1-3 million indivi-

and this classification was sent for evalua-

duals; 3 for 500000-1 million individuals; 4

tion to the experts.

for 100000-500000 individuals and 5 for less than 100000 individuals.

(f4) Nocturnal flight activity

(f8) Adult survival rate

Nocturnal flight activity was classified from hardly any flight activity at night (score

Additional mortality due to collisions is

1) to high flight activity at night (score 5) and

likely to affect species with high annual sur-

Research papers vival rates rather than species with low sur-

cell and plotted.

vival rates. The factor was classified as fo-

The count data of all the years was poo-

llows: 1, ≤ 0.75; 2, >0.75-0.80; 3, >0.80-0.85; 4,

led and summarized in four temporal scena-

> 0.85-0.90; 5, > 0.90. The survival rates were

rios: the whole year, breeding season (March

obtained from Garthe and Hüppop (2004),

to June), post-breeding season (July to Octo-

Schreiber and Burger (2002) and Álvarez and

ber) and wintering season (November to Fe-

Velando (2007). When the rate was not avai-

bruary). Following the recommendations of

lable the values from closely related species

the WSI revision (Certain et al. in prep.), for

were taken.

each temporal scenario we used a diagnostic panel showing the overall vulnerability,

(f9) Conservation status

abundance and collision and disturbance vulnerability maps to inform on the strate-

The original factor reflected the European threat and conservation status using

gic assessment of the offshore wind energy development in the Iberian Peninsula.

part of the SPEC (Species of European Con-

Instead of using different legends for

cern) categories. For a more general index

each map, the vulnerability values and

with applicability to any part of the world

seabird density were plotted in a colour gra-

we used the IUCN conservation criteria. Ac-

dient where each colour indicates a particu-

cording to their conservation status species

lar percentile. For the local discussion, the

where scored: 1 for Least Concern; 2, Near

collision and disturbance vulnerability at

threatened; 3, Vulnerable; 4, Endangered

the 0.25º scale were split in the four defined

and 5, Critically Endangered.

regions and plotted with independent co-

With all the species scores, instead of

lour gradient scales.

calculating the SSI value, we calculated the overall species vulnerability (vi), the species

RESULTS

vulnerability to collision (vci) and the species vulnerability to disturbance (vdi) for each

A total of 41 different species were coun-

species according to the formulas described

ted in the surveys. The most abundant spe-

in Certain et al. (in prep.) that differentiated

cies were the Northern gannet Morus bas-

risk factors in primary risk factors (f1, f2, f5, f7,

sanus (32807 individuals), the Yellow-legged

f9) and aggravation factors (f3, f4, f6, f8).

gull Larus michahellis (20449 individuals) and the Balearic shearwater Puffinus mauretani-

Distributional data and community vulnerability maps

cus (12621 individuals). The species showed

Once the species vulnerability score

the Balearic shearwater were the species

is calculated for each species, the index is

with the highest sensitivity while the Black-

applied to the distributional data to calcula-

headed gull Larus ridibundus and the Atlantic

te the overall community vulnerability (Vj),

puffin Fratercula arctica ranked the lowest.

a wide range of vulnerability values (Table 2). The Audouin’s gull Larus audouinii and

the community vulnerability to collision (Vcj)

The overall community vulnerability

and the community vulnerability to distur-

(Vj) had a similar pattern across grid scales

bance (Vdj). The total seabird abundance per

(Fig.2). Nevertheless, at the broader scale

grid cell was obtained by dividing the sum of

(Fig.2a) coastal cells could not be differentia-

individuals by the total surveyed area in the

ted from offshore cells. This leads to a biased

CHAPTER 2

Table 2 Species scores for each risk factor (f1-9). Overall species vulnerability scores (vi), species vulnerability to collision (vci) and vulnerability to disturbance (vdi) for Atlantic and Mediterranean species detected.

Research papers

OVERALL VULNERABILITY ( Vj )

Fig. 2 Overall community vulnerability for all the year at different scales

1º

2º

interpretation of the vulnerability patterns.

Percentile

Low 25

0.5 º

High 100

0.25 º

In the wintering months (November to Fe-

For the whole year, the areas of highest

bruary) the southern coasts of the Iberian

vulnerability, i.e. with Vj values over the

Peninsula showed the higher vulnerability

50th percentile, were in the Portuguese At-

levels (East and West, between 36º-40º). The

lantic coast and the Mediterranean coasts

eastern part of region I, corresponding to the

(Fig.3ia,c,d). The vulnerability values near

Bay of Biscay, was only surveyed during the

the coast were generally higher than tho-

post-breeding months hence no informa-

se further offshore in the Atlantic coasts,

tion is available for the breeding and winte-

while the Mediterranean coasts had high

ring season.

vulnerability in almost all grid cells. Seabird

The regional vulnerability maps at 0.25º

abundance was higher in the coastal cells of

scale (Fig.4 and 5) highlight that regions as-

Portugal, the Gulf of Cadiz and the Northern

sessed in detail have areas with constant

part of the Mediterranean region (Fig.3ib).

high vulnerability and other areas that tend

The abundance pattern was not similar to

to be either vulnerable to collision or dis-

any of the vulnerability maps of the diag-

turbance. Within Region I and II, areas with

nostic panel.

high collision vulnerability tend to be more

Regarding the temporal evolution of ove-

offshore than areas with high disturbance

rall vulnerable areas, the coastal southern

vulnerability. In general vulnerable areas fit

part of the Portuguese region (between 37-

well with the already defined Marine Impor-

39º N) and the area between the Ebro Delta

tant Bird Areas (IBAs)

and Cape Nao (between 39-41º N) remained vulnerable through the three different pe-

DISCUSSION

riods, whereas other areas increased their vulnerability only in particular periods of

SSI and WSI modifications

the year (Fig.3a). In the breeding months (March to June), the northern half of the

The technological advances in remote

Mediterranean region and the Alboran Sea

sensing has fostered the study of seabird

showed the higher levels of vulnerability to

movements at sea (Ropert-Coudert and

collision and disturbance (Fig.3iic,d). In the

Wilson 2005; Louzao et al. 2009; Christel et

post-breeding months (July to October), the

al. 2012). Nevertheless there is still a lack of

Mediterranean region showed moderate to

information about seabirds’ behaviour in

low levels of vulnerability to disturbance

offshore areas and how this behaviour can

while it increased in all the western coast

be affected by the presence of offshore wind

of the Iberian Peninsula (between 37-44º N).

farms (Desholm and Kahlert 2005; Perrow

a) Overall Vulnerability (Vj)

b) Total seabird abundance (Aj) c) Vulnerability to Collision (Vc j)

d) Vulnerability to Disturbance (Vd j)

CHAPTER 2 | 53

fig. 3 Diagnostic panels at large scale (1ยบ grid scale) in four temporal scenarios.

All year

(i)

Breeding

(Mar-Jun)

(ii)

Post-breeding

(Jul-Oct)

(iii)

Wintering

(Nov-Feb)

(iv)

Research papers et al. 2011). Thus, the SSI methodology that

values over the 60th percentile were defined

merges real data with expert-based scores is

as ‘concern’ areas and those over the 80th

a useful tool to evaluate species sensitivity

percentile were defined as ‘major concern’

to wind farms when no quantitative data is

areas. In the reports that have applied this

available.

methodology these thresholds have been

The scoring method of the SSI is flexible

retained. In the papers that have adapted

and can easily be adapted to data availabili-

the methodology a similar system has been

ty and circumstances of a particular study,

used except in one case (Stelzenmüller et al.

however, we found necessary to make an

2010) where the results were rescaled to a 1

important change in the conservation sta-

to 10 qualitative scale. Although three levels

tus factor. We suggest using the IUCN Red

seems an intuitive output for decision-ma-

list categories by default instead of the SPEC

king, the selection of the threshold percen-

(Species of European Concern) categories

tiles that divides concern areas from low/no

and only choose other conservation classi-

concern areas is subjective. Besides, the 60th

fications if all the study species fall in the

and 80th percentiles might not be the ade-

same category which was the case of Garthe

quate cut-off values for all biogeographic re-

and Hüppop’s study. The SPEC categories are

gions. Instead of generalizing these values,

not as widely recognized as the IUCN cate-

we propose to plot the community vulnera-

gories. Moreover, this classification criterion

bility maps ranking the cells from the lowest

is only useful in a European framework and

value to the highest. Plotting a ranking

offshore wind energy development and any

(which is equivalent to plot the percentiles)

recommended tool should aim to use inter-

allows a comparative analysis between high

national standards.

and low concern areas, retains the percenti-

Concerning the WSI graphical output,

le information and, at the same time, avoids

the original methodology suggested three

the subjective definition of a threshold.

levels of categorization where cells with WSI

Community Vulnerability maps and Offshore

Vulnerability to Collision (Vc j)

Fig. 4 Community vulnerability to collision by regions (0.25º grid scale; All year). Each region is plotted with an independent colour scale. Spanish and Portuguese Marine IBAs overlaid in black.

Percentile

Low 0

High 100

Region I

Region II

Region IV

Region III

CHAPTER 2

Despite the data set limitations the gen-

Wind Energy planning

eral pattern indicates that offshore areas in the Atlantic regions have lower vulnerabil-

The 2ยบ grid is too coarse to capture the

ity, while the coastal Atlantic areas and the

vulnerability patterns. With this scale, off-

Mediterranean regions have the highest vul-

shore areas with very low vulnerability

nerability. This pattern can be explained by

would be assessed as high concern areas.

the characteristics of the continental mar-

Therefore, we suggest using the 1ยบ or 0.50ยบ

gin. Surveys in the Atlantic coast cover the

scales instead. Since the Spanish SEA for

continental shelf and the shelf break. In the

offshore wind energy already defined ma-

Mediterranean coast the continental shelf is

rine zones of 1ยบ (MARM and MITYC 2009) we

wider, particularly in the area between the

chose to use this grid for large scale assess-

Ebro delta and Cape Nao, thus most of the

ment of the vulnerability.

surveys only cover the continental shelf. If

The discrepancies between abundance

offshore wind farms were to be planned in

maps and the vulnerability maps (Fig.3b)

the Mediterranean region the vulnerability

highlights the limitations of using only

index can be applied at a lower scale (e.g.

abundance maps for decision-making and

Fig.4,5). At this scale, the area North to the

the advantage of using also vulnerability in-

Ebro Delta and the area between Cape of

dexes. With the diagnostic panels at large

Gata and Cape Palos are the areas with lower

scale, different areas can be compared in

vulnerability which is consistent with areas

order to demarcate areas with low expected

with a narrow continental shelf. According

impact on seabirds as optimal development

to this, the best advice for the future devel-

areas. In the study area, region I is the area

opment of the offshore energy in the coasts

with the lower expected impact. However,

of the Iberian Peninsula is to avoid the con-

this is probably biased by the lack of surveys

tinental shelf and promote the development

in the breeding and wintering seasons.

of floating offshore wind farms which would

Vulnerability to Disturbance (Vd j)

Percentile

Low 0

High 100

Region I

Region II

Region IV

Region III

Fig. 5 Community vulnerability to disturbance by regions (0.25ยบ grid scale; All year). Each region is plotted with an independent colour scale. Spanish and Portuguese Marine IBAs overlaid in black.

Research papers be the optimal option to minimize the im-

spatial patterns than density maps. This is

pact on the seabird community.

because vulnerability indexes are able to

By assessing independently collision

integrate different layers of information.

and disturbance vulnerability further stud-

This is particularly useful for policy makers

ies may be suggested accordingly for the

and conservation practitioners that need

selected areas for offshore wind energy

analytical tools to take informed decision

development. In areas with high collision

on offshore wind energy spatial planning at

vulnerability during migration season radar

large and regional scales.

studies should be performed, while in areas with high species abundance and high

REFERENCES

disturbance vulnerability habitat modelling and satellite tracking methods could be per-

Anderson DJ, Huyvaert KP, Wood DR (2003) At-sea

formed to improve the knowledge of the off-

distribution of waved albatrosses and the Ga-

shore distribution of seabirds.

lapagos Marine Reserve. Biological Conservation 110:367–373

CONCLUSION

Arcos J, Bécares J, Rodríguez B, Ruiz A (2009) Áreas Importantes para la Conservación de las Aves

The species vulnerability table presented

marinas en España. LIFE04NAT/ES/000049-So-

in this paper could be of major interest in

ciedad Española de Ornitología (SEO/Birdlife),

the future if offshore wind energy is to be

Madrid

developed in Spain, Portugal or any other

BirdLife International (2004) Birds in Europe: po-

country with similar diversity of seabirds.

pulation estimates, trends and conservation

Besides the application of the revised for-

status (I Burfield and F Van Bommel, Eds.). Bir-

mulation of the Index (Certain et al. in prep.)

dLife International, Cambridge <http://www.

we suggest avoiding the percentile threshold

birdlife.org/eu/pdfs/BirdLife_Bern_windfarms.

criterion used by Garthe and Hüppop (2004)

pdf> (accessed 24.10.12)

and using a ranking instead that would keep

BirdLife International (2012) IUCN Red List for

the percentile information of the original

birds. BirdLife International. Available: http://

work without setting a subjective threshold

www.birdlife.org (accessed 17 July 2012), Cam-

that may lead to biased interpretations of

bridge

the outputs.

Certain G, Christel I, Planque B, Bretagnolle V. A

With this study we show the utility of the

refined methodology to estimate the vulnera-

vulnerability indexes as comparative tools

bility of seabird community to the establish-

at large scales but also applicable at smaller

ment of offshore wind farms. (in prep.)

scales when detailed assessment is requi-

Christel I, Navarro J, del Castillo M, Cama A, Ferrer

red. Applying the refined WSI to the Iberian

X (2012) Foraging movements of Audouin’s

coasts we show how the index is useful to

gull (Larus audouinii) in the Ebro Delta, NW

highlight key development regions for offs-

Mediterranean: A preliminary satellite-trac-

hore wind energy as well as exclusion areas

king study. Estuarine, Coastal and Shelf Scien-

due to high vulnerability for seabirds.

ce 96:257–261

Vulnerability maps, which are partly

Christensen-dalsgaard S, Lorentsen S, Hanssen

based on abundance maps, show different

F, Systad GH (2011) Marine wind farms and

CHAPTER 2 seabirds. Updated screening of potential con-

of%20offshore%20wind%20farms%20on%20

flict areas on a national scale. NINA Report

seabirds%20in%20the%20Dutch%20Nord%20

616. <http://mhk.pnnl.gov/wiki/images/8/83/ Visual_and_Radar_Observations_of_Birds. pdf>(accessed 24.10.12)

Sea.pdf> (accessed 24.10.12) Louzao M, Bécares J, Rodríguez B, Hyrenbach K, Ruiz a, Arcos J (2009) Combining vessel-based

Desholm M, Kahlert J (2005) Avian collision risk at

surveys and tracking data to identify key ma-

an offshore wind farm. Biology Letters 1:296–

rine areas for seabirds. Marine Ecology Pro-

298

gress Series 391:183–197

EWEA (2012) The European offshore wind indus-

Louzao M, Hyrenbach KD, Arcos J, Abelló P, de Sola

try key 2011 trends and statistics. European

LG, Oro D (2006) Oceanographic habitat of an

Wind Energy Association. <http://www.ewea.

endangered Mediterranean procellariiform:

org/fileadmin/ewea_documents/images/gra-

Implications for marine protected areas. Eco-

phs_maps_tables/EuOffwind2011.pdf> (accessed 24.10.12)

logical Applications 16:1683–1695 MARM, MITYC (2009) Estudio estratégico am-

Fox AD, Desholm M, Kahlert J, Christensen TK,

biental del litoral español para la instalación

Petersen IK (2006) Information needs to sup-

de parques eólicos marinos. <http://www.

port environmental impact assessment of

aeeolica.org/uploads/documents/562-estudio-

the effects of European marine offshore wind

estrategico-ambiental-del-litoral-espanol-

farms on birds. Ibis 148:129–144

para-la-instalacion-de-parques-eolicos-mari-

Garthe S, Hüppop O (2004) Scaling possible adver-

nos_mityc.pdf> (accessed 24.10.12)

se effects of marine wind farms on seabirds:

Noguera JC, Pérez I, Mínguez E (2010) Impact of te-

developing and applying a vulnerabiblity in-

rrestrial wind farms on diurnal raptors: deve-

dex. Journal of Applied Ecology 41:724–734

loping a spatial vulnerability index and poten-

Hüppop O, Dierschke J, Exo KM, Fredrich E, Hill

tial vulnerability maps. Ardeola 57:41–53

R (2006) Bird migration studies and potential

Perrow MR, Gilroy JJ, Skeate ER, Tomlinson ML

collision risk with offshore wind turbines. Ibis

(2011) Effects of the construction of Scroby

148:90–109

Sands offshore wind farm on the prey base of

Inger R, Attrill MJ, Bearhop S, Broderick AC, Ja-

Little tern Sternula albifrons at its most im-

mes Grecian W, Hodgson DJ, Mills C, Sheehan

portant UK colony. Marine Pollution Bulletin

E, Votier SC, Witt MJ, Godley BJ (2009) Marine

62:1661–70

renewable energy: potential benefits to biodi-

Piatt JF, Harding A, Shultz M, Speckman S, van Pelt

versity? An urgent call for research. Journal of

T, Drew GS, Kettle A (2007) Seabirds as indica-

Applied Ecology 46:1145–1153

tors of marine food supplies: Cairns revisited.

King DA (2004) Climate Change Science : Adapt, Mitigate, or Ignore? Science 303:176–177

Marine Ecology Progress Series 352:221–234 Ramirez I, Geraldes P, Meirinho A, Amorim P, Paiva

Leopold MFL, Dijkman EM (2010) Offshore wind

VH (2008) Important Areas for Seabirds in Por-

farms and seabirds in the Dutch Sector of

tugal. Project LIFE04NAT/PT/000213. Socieda-

the North Sea: What are the best and the

de Portuguesa Para o Estudo das Aves, Lisboa

worst locations for future develpment? Ins-

Ropert-Coudert Y, Wilson RP (2005) Trends and

titute for Marine Resources & Ecosystem

perspectives in animal-attached remote sen-

Studies.

sing. Frontiers in Ecology and the Environ-

<http://www.we-at-sea.org/leden/

docs/reports/RL2-2%202007-003%20Effect%20

ment 3:437–444

Research papers Rosenzweig C, Karoly D, Vicarelli M, Neofotis P, Wu

a spatially explicit risk assessment for marine

Q, Casassa G, Menzel A, Root TL, Estrella N, Se-

management: Assessing the vulnerability of

guin B, Tryjanowski P, Liu C, Rawlins S, Imeson

fish to aggregate extraction. Biological Con-

A (2008) Attributing physical and biological

servation 143:230–238

impacts to anthropogenic climate change. Nature 453:353–357 Schreiber EA, Burger J (Eds) (2002) Biology of Marine Birds. CRC Press, Boca Raton

Tasker ML, Jones PH, Dixon T m, Blake BF (1984) Counting seabirds at sea from ships: a review of methods employed and a suggestion for a standardized approach. The Auk 101:567–577

Sonntag N, Schwemmer H, Fock HO, Bellebaum J,

Álvarez D, Velando A (2007) El cormorán moñudo

Garthe S (2012) Seabirds, set-nets, and conser-

en España. Población en 2006-2007 y método

vation management: assessment of conflict

de censo. Madrid

potential and vulnerability of birds to bycatch in gillnets. ICES Journal of Marine Science 69:578–589 Stelzenmüller V, Ellis JR, Rogers SI (2010) Towards

CHAPTER 2

Research papers

resu M El desenvolupament de l’energia eòlica marina ha fomentat el debat sobre l’impacte potencial d’aquestes infraestructures sobre les aus marines. En aquest context apareix la necessitat de trobar indicadors que determinin l’efecte i extensió d’aquests impactes. La majoria d’Estudis d’Impacte Ambiental (EIA) presenten mapes de distribució i densitat d’aus, però molt pocs intenten representar de manera explícita els impactes potencials en l’espai i el temps. Mitjançant la relació entre mitjana i variància descrita per Taylor (Taylor ‘s power law) i models lineals mixtos es pot modelar la variabilitat espai-temporal dels patrons de distribució de les aus marines. Els models resultants descriuen el grau d’agregació de les aus al mar el que permet diferenciar zones de transició d’àrees d’alimentació. Aquesta distinció, al seu torn, es pot utilitzar per definir zones amb un alt risc de col · lisió i zones de potencial pèrdua d’hàbitat en el cas de construir un parc eòlic marí. Amb el Delta de l’Ebre com a cas d’estudi il · lustrem la utilitat d’aquest mètode i comentem els avantatges dels mapes d’impacte potencial respecte als mapes d’abundància.

JOURNAL REFERENCE Christel I, Certain G, Cama A, Vieites DR, Ferrer X (2012) Seabird aggregative response: a new tool for offshore wind energy risk assessment. Marine Pollution Bulletin. DOI: 10.1016/j. marpolbul.2012.11.005 (In press)

Seabird aggregative patterns: a new tool for offshore wind energy risk assessment Isadora Christel1,2, Grégoire Certain3,4, Albert Cama1,2, David R. Vieites2,5 and Xavier Ferrer1 1

Institute for Research on Biodiversity (IRBio) and Departament de Biologia Animal, Universitat de Barcelona (UB). Diagonal 645, E-08028 Barcelona, Spain.

Museo Nacional de Ciencias Naturales, Consejo Superior de Investigaciones Científicas. C/ José Gutiérrez Abascal 2, 28006, Madrid, Spain. 3 Norwegian Institute for Nature Research (NINA). Tungasletta 2, NO-7047 Trondheim, Norway. 4 Institute of Marine Research (IMR). PO box 6404, 9294 Tromsø, Norway 5 REFER Biodiversity Chair, University of Porto, CIBIO, Campus Agrário de Vairão, R. Padre Armando Quintas, 4485-661 Vairão, Portugal. A bstract The emerging development of offshore wind energy has raised public concern over its impact on seabird communities. There is a need for an adequate methodology to determine its potential impacts on seabirds. Environmental Impact Assessments (EIA) are mostly relying on a succession of plain density maps without integrated interpretation of seabird spatio-temporal variability. Using Taylor’s power law coupled with mixed effect models, the spatio-temporal variability of species’ distributions can be synthesized in a measure of the aggregation levels of individuals over time and space. Applying the method to a seabird aerial survey in the Ebro Delta, NW Mediterranean Sea, we were able to make an explicit distinction between transitional and feeding areas to define and map the potential impacts of an offshore wind farm project. We use the Ebro Delta study case to discuss the advantages of potential impacts maps over density maps, as well as to illustrate how these potential impact maps can be applied to inform on concern levels, optimal EIA design and monitoring in the assessment of local offshore wind energy projects.

Research papers INTRODUCTION

Assessment (EIA) of its potential impact in the marine environment, including the risk

Studies on marine top predators are today considered as a key component of ma-

imposed on avian populations (Bright et al., 2008; Masden et al., 2010).

rine ecosystem management (Boyd et al.,

The potential impacts of offshore wind

2006). Within top predators, seabirds are

farms on seabird communities are complex.

good indicators of ecosystem health (Cairns,

Fox et al. (2006) provided a conceptual clas-

1987; Mallory et al., 2006; Nettleship and

sification of these impacts, distinguishing

Duffy, 1993) and are useful indicators to

between (1) avoidance, (2) modification of

evaluate potential effects of human activi-

the physical habitats, and (3) direct mor-

ties at sea. Most seabirds are flagship species

tality trough collision. Most EIA guidelines

for the public (Fox et al., 2006) and have clear

suggest radar studies to assess collision risk

protection criteria collected in protection di-

in strongly migratory areas (Desholm et al.,

rectives like the Birds directive (79/409/EEC)

2006; Fox et al., 2006; Kunz et al., 2007) and

and Habitats directive (92/43/EEC) in Europe.

density maps as a proxy to loss of foraging

Their distribution and abundance are usu-

habitats by avoidance and physical habitat

ally provided as key information to support

modification (Camphuysen et al., 2004; Fox

the establishment of marine protected ar-

and Petersen, 2006). However, density maps

eas, to implement fisheries’ management

do not provide a full understanding of the

measures (Boyd et al., 2006), to assess the

underlying behavioral patterns related to

impact of environmental disasters such as

their movements. Seabirds often present dy-

oil spills (Bretagnolle et al., 2004; Moreno,

namic and complex spatial patterns at sea

2011) or to monitor the impact of oil and gas

which are far from being understood. When

platforms at sea (Wiese et al., 2001).

foraging, many species of seabirds are usu-

In the last years, offshore wind energy

ally characterized by an important aggre-

has emerged as a priority field in many

gative behavior (Buckley, 1997; Grünbaum

European countries to meet Europe’s 2020

and Veit, 2003), with birds forming flocks of

agenda that promotes renewable energies

hundreds of individuals. On the contrary, a

to mitigate the effects of climate change;

lower aggregative behavior is expected in

hence offshore wind farms will likely ex-

transitional areas solely used as flight paths

perience an important increase in the near

between feeding areas and their resting or

future. However, in the field of marine man-

breeding areas. While density maps focus

agement there is a growing concern on the

on high concentrations of seabirds as po-

development of offshore wind energy and

tential risk areas, we propose the explicit

its potential impacts on coastal seabird

distinction between transitional and forag-

populations, mainly because of possible col-

ing areas as a key step to better predict and

lisions with windmills (Fox et al., 2006). On

classify the risk of wind farm establishment

a large scale, countries might develop “Stra-

on seabird populations. In transitional ar-

tegic Environmental Assessments” (SEA) to

eas, the main risk will be direct collision and

plan their offshore wind farms network in

mortality (Desholm and Kahlert, 2005; Hüp-

a way that minimizes their ecological im-

pop et al., 2006). In foraging areas, the risk of

pact on the coastal environment (Directive

direct collisions is increased and potentially

2001/42/EC). At a local scale, each wind farm

associated with a displacement from their

project requires an Environmental Impact

preferred feeding areas, resulting in habitat

CHAPTER 3 loss (Masden et al., 2010; Perrow et al., 2011).

farms. Finally, we discuss how the resulting

In 2004, the proposal of an offshore

potential impacts maps provide a frame to

wind farm project in front of the Ebro Delta

inform on EIA design and monitoring in the

(North-Western Mediterranean, Fig. 1a) em-

context of an offshore wind farm proposal.

phasized the necessity for adequate indicators to determine the extent and effect of

METHOD

potential impacts on its seabird community. Here, we use the slope of the Taylor’s power

Study area & survey method

law as a measure of the aggregative patterns of seabirds to identify transitional and feed-

The Ebro Delta (NW Mediterranean, Fig.

ing areas, and map the risk accordingly. The

1) is a very productive area because of a

slope of the Taylor’s power law (Taylor, 1961;

permanent upwelling, result of the sudden

Taylor and Woiwod, 1982) provides a conve-

broadening of the shelf (up to 70km) in com-

nient measure of the aggregation levels of

bination with the influence of the Liguro-

animals (see Kendal, 2004 for a review). It

Provençal-Catalán front and nutrients car-

has already been used in a spatio-temporal

ried by the Ebro river runoff (Arcos et al.,

context with seabirds (Certain et al., 2007)

2001; Palomera, 1992). This high productiv-

and has proved to be useful to describe the

ity supports an important fishing fleet with

temporal variability associated to the spatial

a high trawling activity (Arcos et al., 2001;

distribution of seabirds at multiple scales.

Louzao et al., 2006; Palomera, 1992) which

Here, using the Ebro Delta as a case study,

in turn has been pointed as a key resource

we first show how to take into account the

for seabirds (Arcos, 2001; Arcos et al., 2008).

aggregative properties of seabird distribu-

However, the trawling activity is regulated

tions together with abundance maps. Sec-

with temporal moratoria in the area. Fish-

ond, we point the advantages of this method

ing moratoria affects the northern area (B1-

as an integrative tool to summarize in few

2 and B14-16, Fig. 1) in May and June and

maps the spatial and temporal variability

the southern area (B3-B13) during July and

of the potential impacts of offshore wind

August, and influences the distribution of

Fig.1 (a) Situation map. (b) Survey design of aerial transects and projected offshore wind farm location. (c) Block design of the study area showing inner and outer classification and block Id. The main breeding colonies location, harbours and the Ebro River are shown.

Research papers some species.

Table 1. Descriptive data

aircraft, Partenavia P68, and the aircraft GPS

Seven monthly aerial surveys were car-

was used for navigation along the transect

ried out from March 2005 to September 2005

tracks. The cruising speed was set at c. 100

on the continental shelf around the Ebro

knots (185 km/h) with respect to the air

Delta (40.7º N, 0.75º E). The surveys covered a

speed and average flying height was 300 feet

total area of 1435 km2 from L’Ametlla de Mar

(100 m). Along the transects, all observed

harbour (24 km North; 40.86º N, 0.8º E) to Pe-

bird flocks were recorded with a voice re-

ñíscola (51 km South; 40.35º N, 0.4º E) (Fig. 1).

corder, stating information on species (or

The entire shelf area can be covered in a sin-

the lowest taxonomical level determinable),

gle day using this approach, and availabil-

number of individuals, behaviour (e.g. flying,

ity biases due to attraction and avoidance

flushing, sitting on water, feeding on trawler

movements of seabirds were minimized. In

discards), age whenever possible, transect

this study, we used the standard seabird aer-

strip, date and time. The presence of trawl-

ial survey methodology described by Noer et

ers was also recorded. These recordings

al. (2000).

were geo-referenced later with the transect

The survey area was covered by 45 tran-

track information provided by a GPS and a

sects systematically arranged in parallel

Turbo Pascal application (Ib Krag Petersen

lines running perpendicular to the coast,

pers. com.). In the moments of maximum

to follow the dominant sea depth gradient,

glare or any other adverse light situation,

and flown at 2 km intervals. During the sur-

the counting was interrupted. Since counts

veys, two observers, one at each side of the

results are highly sensitive to meteorology,

aircraft, covered 1 km strip at each side. The

no surveys were conducted when Beaufort

surveys were conducted from a twin-engine

Sea state was greater than one.

CHAPTER 3

Fig.2 Data preparation and selected scales.

as well as the Common Tern population.

Study model

Both species are present in the area between March and September. The species arrive to

We focused the study on the three most

the breeding grounds in March and April,

abundant seabird species in the area, the

being the peak of the breeding season be-

Yellow-legged Gull (Larus michahellis), the

tween May and June, after which there is a

Audouin’s Gull (Larus auouinii) and the Com-

variable post-fledging period with dispersive

mon Tern (Sterna hirundo). These three spe-

behaviour until they start their migration

cies represented the 93% of all detected in-

from late August to October (Cama, 2010). In

dividuals at sea. Moreover, they can be used

2005, the main colony for the Common Tern

as model species for their different foraging

was in the north of the Ebro Delta with 3361

and feeding strategies. The Yellow-legged

breeding pairs, and the main colony for the

Gull is a scavenger that makes extensive

Yellow-legged and the Audouin’s gulls was

use of trawler’s discards (Oro et al., 1995),

in the South of the Ebro Delta having 9850

and their foraging habits are strongly influ-

and 13850 breeding pairs respectively (Fig. 1)

enced by trawlers predictability (Cama et al.,

(Cama, 2010).

2012). The Audouin’s Gull is an opportunist species that exploits small pelagic fish (Oro,

Data preparation

1998 and references therein), but also makes use of trawler’s discards and terrestrial

First, transects were sliced into segments

food sources (Christel et al., 2012; Navarro

of 0.5 km length, each segment containing

et al., 2010; Oro and Ruiz, 1997). The Com-

the number of birds counted for each spe-

mon Tern, conversely, only preys actively on

cies (Fig. 2). This length corresponds to the

shoals of small pelagic fish (Cramp and Sim-

minimum scale at which the information

mons, 2004).

could be located, according to the survey

The Yellow-legged Gull population in the

protocol (Noer et al., 2000). Second, mean

Ebro Delta is sedentary. The species breeds

and variance of bird abundance of the seg-

from mid-March to April at the Punta de la

ments were computed within grid cells of 3.5

Banya peninsula (Fig.1). Some individuals of

km wide. Only grid cells containing a mini-

Audouin’s Gull are in the area all over the

mum of 10 segments and at least two non-

year, but the main population is migratory

zero abundance values were included in the

Research papers analyses. Finally, the study area was further

of organisms (Jiménez et al., 2001; Östman,

divided in blocks of 11km. These 11km cor-

2002). Usually, slope values range between

respond to the scale at which management

1 and 2 when estimated in space (Engen

information is extracted, and was set as a

et al., 2008; Kendal, 2004). When calculated

trade-off for sample size. We searched the

through time, it can be used as an index of

finer scale that had at least N=8 grid cells

the temporal variability of the spatial distri-

into each block so that to fit a power law and

bution of organisms, highlighting recurrent

provide detailed information for manage-

and occasional presence areas (Certain et

ment. At this selected block size, the pure

al., 2007).

coastal areas could be distinguished from

Coupling Taylor’s power law with linear

areas located more offshore, where bird ac-

mixed effect models (LME) allows the inves-

tivity might differ in abundance and behav-

tigation of the spatio-temporal variability of

ior.

TPL slope and consequently the variability of the aggregative patterns of organisms,

Modeling of seabird aggregative pattern

avoiding confusing effects of changes in animal abundances (Certain et al., 2007).

Pioneered work of Taylor (1961) on the spatial and temporal variability of animal

The simplest model within this framework is:

abundance provides an useful framework to study the spatio-temporal heterogeneity

log (Vbmsj) = A × log (Mbmsj) + B + εbmsj

of a population within its habitat (Kendal,

j= 1,…,nbms

(Eq. 3)

2004; Kilpatrick and Ives, 2003; Taylor and Woiwod, 1980, 1982). Taylor’s power law (re-

Where the slope and the intercept are

ferred as TPL from here onwards) states that

supposed constant through space (b), time

the variance in abundance (V) is proportion-

(m) and the three species (s). However, the

al to a power of the mean abundance (M):

slope might vary according to one or several of these factors. The effect of these factors

V=b×Ma

(Eq. 1)

and all the possible combinations can be introduced in the model as a grouping factor

Which in the logarithmic scale becomes

with a random effect on the slope (Pinheiro

a linear regression, where a is the slope and

and Bates, 2000). Starting from the simplest

log b is the intercept:

possible model (Eq. 3), different models were developed by the sequential addition of ran-

log (V) = a × log (M) + log (b)

(Eq. 2)

dom effects on the slope. The most complete model could be written as the following:

In this context, the slope a is considered an aggregation index. If individuals are ran-

log (Vbmsj) = (A + abms) × log (Mbmsj) + B + εbmsj

domly distributed the slope equals 1, if in-

j= 1,…,nbms

(Eq. 4)

dividuals show some degree of aggregation the slope increases accordingly (Engen et

Where A is the fixed slope, abms is the

al., 2008; Kendal, 2004; Kilpatrick and Ives,

random effect on the slope of block, month

2003). When calculated through space, the

and species together, B is the fixed intercept,

TPL slope can be interpreted as a measure

nbms is the number of observations on a bms

of the strength of the aggregative response

combination, and the εbmsj are independent

CHAPTER 3 N(0,σ2) error terms. Forward stepwise mod-

sels) or sitting on water; and used to calcu-

el selection was applied. Each model was

late the proportion of birds flying. The GAM

compared with the null model with a likeli-

analysis was carried out with the mgcv

hood ratio test to check whether or not the

package (Wood, 2006) following a forward

inclusion of a new grouping factor was out-

stepwise model selection based on the min-

performing the previous one (Pinheiro and

imization of AIC and the analysis of devi-

Bates, 2000). We retained the simplest model

ance between models. The number of knots

for which the inclusion of any new group-

in the smooth functions was minimized to

ing factor did not result in a significant im-

five to avoid overfitting.

provement of the model. All data processing and model developments were performed in R (R Development Core Team, 2008) with package lme4 (Bates et al., 2011).

Fig.3 Monthly ave-

To visualize the slope variation across

rage and standard error of TPL slope values for all blocks. The general mean slope for the full area and months is showed in grey dotted line.

months, the predicted slope values of the optimal mixed effect model were averaged by month (Fig. 3). Spatial variation of the slope was summarized for the breeding (March to June) and post-breeding (July to September) seasons, and was calculated with the average of the slope values for the corresponding months and plotted together with the density map of the number of individuals per month and block area (Fig. 4).

Behavioral interpretation of the aggregative pattern

RESULTS

We employed a generalized additive model (GAM) to test if the resulting slope

The result of the model-selection pro-

values of the mixed effect model had any

cess for the mixed-effect model is presen-

significant linear or non-linear correlation

ted in table 1. The optimal model retained

with the density of seabirds, the recorded

block and month as the grouping factor

behavior, the flock size or the presence of

with a random effect on the slope. Species

trawlers. For each block in a given month,

had no significant effect on the slope. The

we extracted the density of seabirds (total

fixed estimated slope for all the area was

observed individuals divided by block area),

1.923 and the predicted slope values ranged

mean and variance of the size of the ob-

between 1.712 and 2.049.

served flocks and the number of trawlers.

The monthly evolution of Taylor’s pow-

The behavioral information recorded with

er law slope (Fig. 3) can be summarized in

the observations was classified in two cat-

two sequences. First, from March to June

egories: i) Flying, ii) fishing (on shoals or ves-

(Mean±SE = 1.89±0.02), and especially in

Research papers

AIC is the Akaike information criterion, and Dev indicates the percentage of deviance explained by each model. Abbreviations for the formula terms are: Var, Logarithm of the variance (Dependent variable); Mean, logarithm of the mean; B, Block; M, month; S, Species. The model emboldened was selected as the optimal one.

Table 2. Forward-selection procedure used to find the best set of predictors of Taylor’s power law slope in the mixed effects model.

May and June, slope values are low, suggest-

According to the optimal selected GAM -135, deviance explained=

ing low aggregation levels of the seabirds

model (AIC=

populations at sea. Then, TPL slope increases

29.3%), the predicted slope values show a

markedly from July to September (Mean±SE

linear negative correlation with the pro-

= 1.97±0.01), suggesting strong aggregations

portion of flying birds (Estimate=-0.12826,

of seabirds at sea.

p=0.006) and a non-linear correlation with

As two time periods (March-June and

the interaction of mean flock size and den-

July-September) were clearly distinguished

sity (p=0.005, 3.734 estimated degrees of

in the temporal evolution of the TPL slope

freedom). When factors were examined one

in the area, spatial maps of TPL slope were

by one, mean flock size was the main driver

drawn for these two periods (Fig. 4). From

of slope changes (p<0.0001, 22.8% deviance

March to June, blocks near the colonies (B4-

explained) followed by proportion of flying

5, B10 and B1-2) had slope values under the

birds (p<0.0001, 14.8%). Density had no sig-

average, suggesting a lower aggregative be-

nificant effect (p=0.216, 1.73%).

havior, while areas in the outer blocks near

Thus, an increase in the mean flock size

the river mouth (B13-12) had higher slopes.

in a block increases TPL slope, but blocks

From July to September, the slope values

with high numbers of flying birds are more

around the colonies increased switching to

likely to have lower slopes. The main effect

a more aggregated pattern (except B4). The

of density is on its interaction with flock size.

outer blocks north and south to the river

For low density values, an increase on the

mouth (B11-16) retained and intensified

mean flock size has a logarithmic increase

their aggregated pattern. Southern blocks

in TPL slope. For densities greater than five

(B6-9) did not show any constant pattern in

individuals per km2 the increase becomes

slope values between seasons.

linear (Fig. 5).

CHAPTER 3

Fig.4 Spatial structure of TPL slope values for the two seasons. Only values with intervals (MeanÂąSE) significantly different to the general average (1.923) are plotted. In the background density values (average ind/km2) are shown in grey scale.

suggesting strong aggregations of birds at

DISCUSSION

sea. The seasonal pattern of TPL slope suggests that these two periods can be used to

Aggregative patterns in time and space

summarize and highlight the main behavioural and aggregative patterns of the popu-

The species of seabirds observed in the

lation. Indeed, seasonal scenarios are easier

continental shelf were mainly gulls and

to communicate than a detailed sequence

terns in high numbers. Despite the select-

of monthly representations and are useful,

ed species have different feeding sources

for instance, to recommend mitigation mea-

and foraging behaviors, the species of both

sures. Regarding the spatial structure of the

groups are central place foragers (Orians

aggregative patterns at both seasons (Fig. 4),

and Pearson, 1979) and use the presence of

the differences between the blocks near the

conspecifics as a cue to find food patches

breeding colonies, the blocks in the outer

at sea (Paiva et al., 2007; Ward and Zahavi,

side of the study area and the southern-

1973). This common behavior probably ex-

most blocks can be explained respectively

plains the similarity in the aggregative pat-

by the vital cycle requirements, the feeding

terns between species.

sources distribution, and trawling moratoria

The temporal evolution of the aggrega-

influence.

tive patterns (Fig. 3) is strongly correlated

The area near the colonies has low aggre-

to the life cycle of the species. The marked

gative levels during the breeding season. At

decrease in TPL slope in May and June co-

this moment of the year the species perform

incides with the chick-rearing period, when

frequent and shorter foraging trips than the

most of the pairs of the three species already

rest of the year (Paiva et al., 2007). This re-

have chicks and perform short and frequent

sults in a constant transit of individuals fly-

foraging trips resulting in low aggregation

ing from and to the colony minimizing the

levels at sea. After the breeding season in

time spent foraging. Once the chick-rear-

June, TPL slope increases markedly as birds

ing period ends, the aggregative patterns

are freed from chick-rearing constraints,

around colonies increase; although the area

Research papers might still be a highly transited area, it is

1992; Sierra et al., 2002) characterized by the

probably combined with some groups of in-

abundance of clupeoids (mainly sardine Sar-

dividuals feeding or spending time resting

dine pilchardus and snchovy Engraulis encrasi-

in these areas.

cholus). The presence of fish shoals and the

The aggregated pattern of the outer

proximity of the common tern colony make

blocks is likely to be driven by the presence of

this area an optimal feeding ground for the

feeding sources. Previous studies in the area

Common tern (Hernández-Matías, 2003). Be-

have reported the extensive use of trawlers

sides, the presence of trawlers makes the

discards by gulls (Arcos et al., 2001; Cama et

northern area attractive for Audouin’s gull

al., 2012; Navarro et al., 2010; Pedrocchi et al.,

and Yellow-legged gull during the trawling

2002) and the diet preferences of common

moratoria in the southern blocks (Cama,

terns (Arcos et al., 2002; Hernández-Matías,

2010).

2003), but no work so far had characterized

The southern area has two main har-

the spatio-temporal distribution of their ag-

bours (Vinarós and Benicarló) and an im-

gregation at sea. The aggregation in the out-

portant daily discarding activity from 15 to

er blocks south of the river mouth is due to

16 h which attracts large flocks of seabirds

a higher presence of trawlers and big flocks

(Cama, 2010). However, southernmost blocks

of birds associated (see trawlers’ distribu-

show a less consistent pattern between sea-

tion in Cama et al., 2012). The area north

sons and none differentiates significantly

of the river mouth is a highly productive

from the average aggregation level. This is

area due to the Ebro river runoff (Palomera,

most likely an effect of trawling moratoria on seabirds’ presence in the area.

Fig.5 Contour plots of the predicted TPL slope according to density of birds and mean flock size, (a) for a fixed 20% of birds flying and (b) for an 80%. The average TPL slope for the whole study area indicated with dashed line. Results based on the semiparametric GAM model of TPL slope with a linear estimator for proportion of birds flying and a smooth term for the interaction of mean flock size and density.

Behavioral interpretation of the aggregative pattern Taylor’s power law slope is widely accepted as an aggregation index, which is corroborated by its correlation with mean flock size. The GAM analysis allows us to be more precise in the interpretation of TPL slope in the case of seabird populations in the Ebro Delta. Low TPL slopes are related to areas with high percentages of birds flying, usually individually or forming small flocks (Fig. 5b). Hence, areas with weak aggregative patterns can be considered transitional or flight path areas. High TPL slopes are found in areas with high percentages of birds feeding and –to a lesser extent– sitting on water, mainly forming big flocks (Fig. 5a). This indicates that areas with strong aggregative patterns are mostly feeding areas

CHAPTER 3

Fig.6 Potential risk map (according to transitional and feeding areas) and concern levels for offshore wind farm placement.

where the presence of birdsâ&#x20AC;&#x2122; aggregations is

areas will eventually have big groups of sea-

driven by the availability of punctual feeding

birds feeding. The presence of offshore tur-

opportunities (fish shoals near the surface

bines in these areas would result in habitat

or discarding trawlers in this study case).

loss threat for species with a strong avoidance response or an increased collision risk

Density maps and Potential impact maps

for the species that venture between the wind turbines (Fig. 6).

Density maps define areas with high

This is particularly true for the assess-

numbers of birds but they do not provide in-

ment of areas of known importance for

formation on the dynamic and complex spa-

breeding populations. However, we suggest

tial patterns of seabirds. At the Ebro Delta, if

the necessity of applying this methodol-

a key protection area had to be selected ac-

ogy to flyway corridors or areas with a dif-

cording to density maps (Fig. 4 background)

ferent composition of species (e.g. plunge-

the northern area would be the one of high-

divers like gannets or surface-divers like sea

est concern, mainly based on high abun-

ducks) to investigate any possible difference

dances during the post-breeding season.

in the interpretation of the potential risks

Despite the intense pattern of movements

associated to the observed aggregative pat-

near the colonies, during the breeding peri-

terns. Nevertheless, to consider the infor-

od the pattern would be masked by low den-

mation on the second order properties of

sity values. Since the aggregative pattern is

speciesâ&#x20AC;&#x2122; distributions (i.e. social aggregation)

a reflection of behavior, it has a direct ap-

provides further information to managers in

plication in the demarcation of areas of high

terms of potential impacts of offshore wind

potential risk for seabirds. Areas revealed as

farms than solely focusing on the first order

main travelling areas are highly susceptible

properties (i.e. density).

to collision threat. Areas pointed as foraging

Research papers ence and abundance is crucial to describe

Concern levels and monitoring protocol

the factors driving the aggregative patterns which in turn can be used to evalu-

The placement of an offshore wind farm

ate barrier effects and the energetic costs of

is a management decision that takes into

avoidance (Masden et al., 2010). In the Ebro

account many socio-economic and environ-

Delta study case, the movement of trawlers

mental factors. To facilitate the inclusion

through time likely changes the distribu-

of seabirds in the decision-making process

tion patterns of gulls at sea as the birds fol-

we propose a ranking of the areas according

low them (e.g. Cama et al. 2012), hence an

to their potential impacts on the seabirdsâ&#x20AC;&#x2122;

offshore wind farm would only change the

populations. Based on recommendations in

distribution of this feeding source, having

the available literature, each level of concern

little impact on gulls. By contrast, the distri-

should be associated to a set of required

bution of terns is driven by productivity and

types of surveys for the EIA and the compul-

fish shoals availability (Paiva et al., 2007). If

sory monitoring for pre- and post-construc-

high numbers of the species were observed

tion of accepted wind farms.

in the area selected for the wind farm, the

Collision mortality is often considered

expected habitat loss could have harmful ef-

to be the most important hazard (Fox et al.,

fects on the population and should be taken

2006). Accordingly, those areas with both

into account in the EIA.

Collision and Habitat loss risk would have

For any project placed in L2 areas, be-

the higher concern level (L3); areas with col-

sides L1 and L0 monitoring, collision risk

lision risk would have the next concern level

models should be calculated. This requires

(L2) followed by areas with habitat loss risk

-depending on the particular case- point

(L1).

transect surveys of flight height and direc-

Any project placed in L0 areas would re-

tion (Camphuysen et al., 2004), surveillance

quire the monitoring of the nearest colonies

radars (vertical and/or horizontal) or infra-

to obtain estimates of demographic and

red camera systems (Desholm et al., 2006).

population sizes to assess how the popu-

In these cases it is especially important an

lations respond to the offshore wind farm

adequate monitoring in pre- and post-cons-

(Kunz et al., 2007).

truction to evaluate predictions made in EIA,

Any project placed in L1 areas, besides

to allow adaptive management of the wind

the previous recommended monitoring,

farm but also to quantify the cumulative im-

should in addition include distribution and

pacts on migratory species (Fox et al., 2006).

habitat modeling (Drewitt and Langston,

Finally, for any project placed too near or

2006; Fox and Petersen, 2006). Part of the

inside Level 3 areas, satellite-tracking stud-

modeling can be based on the already avail-

ies should be also carried out in order to as-

able aerial surveys. However, species with

sess quantitatively the intensity of tracks in

limited numbers or the very small species

the transitional areas and the recurrence of

(e.g. storm petrels, shearwaters, alcids) are

feeding areas. The selected species should

likely missed by aerial surveys and only

be preferentially a flagship or keystone spe-

found in boat surveys (Camphuysen et al.,

cies, because of their conservation status or

2004). Hence, L1 areas with planned wind

relevance in the community. In the Ebro Del-

turbines should be assessed with detailed

ta, for instance, the near threatened Audou-

boat-based surveys. Modeling species pres-

inâ&#x20AC;&#x2122;s gull (Larus audouinii) would be the target

CHAPTER 3 of the study since the colony holds 12000-

ve the decision-making process. To know

13000 breeding pairs, ca. 65% of the world’s

the compulsory monitoring in a selected

total population of this species (Christel et

site might help to decide the optimal loca-

al., 2012; Oro et al., 2009).

tion of offshore wind farms minimizing not only the impact on the seabird community

CONCLUSION

but also the future monitoring costs. Once the placement is decided, the same results

The design of EIA and monitoring sur-

could be used to inform the EIA.

veys is not always an easy, straightforward decision. This is particularly true in areas

ACKNOWLEDGEMENTS

where there is scarce knowledge on the distribution and abundance of the seabirds’

The Offshore wind farm project in front

community at sea. Unlike boat surveys,

of the Ebro Delta was rejected in 2009 be-

aerial surveys provide an extensive covera-

cause it was within an exclusion area in the

ge in a short period of time that offers an

final zonation for the offshore wind energy

image of the presence of seabirds at sea in

in Spain published that year (SEA according

a particular moment (Certain and Bretagno-

to the Real Decreto 1028/2007).

lle, 2008; Drewitt and Langston, 2006). This

We are grateful to T. Anker-Nilssen and

characteristic feature of aerial surveys can

E. Bellier for their useful comments in the

unveil distribution patterns that differ with

first steps of this paper. We appreciate the

the previous knowledge, which is often ba-

work of X. Macià as dedicated observer in

sed in observations from land or ship-based

all the aerial surveys. Special thanks to I.

surveys with more limited area coverage

Bruteig (NINA, Trondheim), O.T. Albert and

per survey than flights. Taylor’s power law

B. Planque (IMR, Tromsø) for welcoming IC

applied on aerial surveys provides a con-

into their groups during the development of

venient analysis tool to ensure the optimal

this paper. Capital Energy Offshore funded

allocation in time and space of resources in

the aerial surveys through agreement with

order to obtain the most detailed knowledge

Fundació Bosch i Gimpera (Contract 304683).

for the EIA of future offshore wind farms on

IC was funded by a PhWD fellowship of the

seabirds.

University of Barcelona. AC was funded by a

Although presented for a local scale, we think that this methodology would be very

PhD fellowship of the Government of Catalonia (2009FIC75).

useful in the four steps of offshore wind energy development: the SEA of offshore

REFERENCES

wind energy development, the decision-making on wind farm projects placement, the

Arcos, J., 2001. Foraging ecology of seabirds at sea:

EIA design and the monitoring planning of

Significance of commercial fisheries in the NW

accepted projects. At a broader scale than

Mediterranean. Thesis, University of Barcelo-

the one presented here, this tool could be

na.

used in the marine spatial planning to select

(accessed 28.02.12)

<http://www.tdx.cat/handle/10803/775>

development regions that avoid the areas

Arcos, J., Louzao, M., Oro, D., 2008. Fisheries ecosys-

identified with potential impacts. Within a

tem impacts and management in the Medite-

selected region, stating a clear monitoring

rranean: seabirds point of view. American Fis-

protocol prior to placement would impro-

heries Society Symposium 49, 1471-1479.

Research papers Arcos, J., Oro, D., Sol, D., 2001. Competition bet-

10.1111/j.1600-0587.2011.06756.x.

ween the yellow-legged gull Larus cachinnans

Camphuysen, C.J., Fox, A.D., Leopold, M.F., Peter-

and Audouin’s gull Larus audouinii associated

sen, I.K., 2004. Towards standardised seabirds

with commercial fishing vessels: the influence

at sea census techniques in connection with

of season and fishing fleet. Marine Biology 139,

environmental impact assessments for offs-

807-816.

hore wind farms in the U.K., Cowrie Report.

Arcos, J., Ruiz, X., Bearhop, S., Furness, R.W., 2002.

London.

Mercury levels in seabirds and their fish prey

Certain, G., Bellier, E., Planque, B., Bretagnolle, V.,

at the Ebro delta (NW Mediterranean): the role

2007. Characterising the temporal variabili-

of trawler discards as source of contamina-

ty of the spatial distribution of animals: an

tion. Marine Ecology Progress Series 232, 281-

application to seabirds at sea. Ecography 30,

290.

695-708.

Bates, D.M., Maechler, M., Bolker, B., 2011. lme4:

Certain, G., Bretagnolle, V., 2008. Monitoring

Linear mixed-effect models using S4 classes.

seabirds population in marine ecosystem: The

R package version 0.999375-42.

use of strip-transect aerial surveys. Remote

Boyd, I.L., Wanless, S., Camphuysen, C.J., 2006. Top

Sensing of Environment 112, 3314-3322.

predators in marine ecosystems: their role

Christel, I., Navarro, J., del Castillo, M., Cama,

in monitoring and management. Cambridge

A., Ferrer, X., 2012. Foraging movements of

University Press.

Audouin’s gull (Larus audouinii) in the Ebro

Bretagnolle, V., Certain, G., Houte, S., Métais, M.,

Delta, NW Mediterranean: A preliminary sa-

2004. Distribution maps and minimum abun-

tellite-tracking study. Estuarine, Coastal and

dance estimates for wintering auks in the Bay

Shelf Science 96, 257-261.

of Biscay, based on aerial surveys. Aquatic Living Resources 353-360.

Cramp, S., Simmons, K.E.L., 2004. BWPi. Birds of the Western Paleartic on Interactive DVD-

Bright, J., Langston, R.W., Bullman, R., Evans, R., Gardner, S., Pearcehiggins, J., 2008. Map of bird

ROM. Birdguides, Oxford University Press, Oxford, U.K.

sensitivities to wind farms in Scotland: A tool

Desholm, M., Fox, A.D., Beasley, P.D.L., Kahlert, J.,

to aid planning and conservation. Biological

2006. Remote techniques for counting and es-

Conservation 141, 2342-2356.

timating the number of bird-wind turbine co-

Buckley, N.J., 1997. Spatial-concentration effects

llisions at sea: a review. Ibis 148, 76-89.

and the importance of local enhancement in

Desholm, M., Kahlert, J., 2005. Avian collision risk

the evolution of colonial breeding in seabirds.

at an offshore wind farm. Biology Letters 1,

American Naturalist 149, 1091–1112.

296-298.

Cairns, D., 1987. Seabirds as indicators of marine food supplies. Biological Oceanography 5, 261271.

Drewitt, A., Langston, R.W., 2006. Assessing the impacts of wind farms on birds. Ibis 148, 29-42. Engen, S., Lande, R., Sæther, B.-E., 2008. A general

Cama, A., 2010. Seabird distribution patterns and ecological factors driving larid presence at the Ebro delta shelf (NW Mediterranean). Thesis, University of Barcelona.

model for analyzing Taylor’s apatial acaling laws. Ecology 89, 2612-2622. Fox, A.D., Desholm, M., Kahlert, J., Christensen, T.K., Petersen, I.K., 2006. Information needs to

Cama, A., Abellana, R., Christel, I., Ferrer, X.,

support environmental impact assessment of

Vieites, D.R., 2012. Living on predictability:

the effects of European marine offshore wind

modeling the density distribution of efficient

foraging

seabirds.

Ecography.

DOI:

farms on birds. Ibis 148, 129-144. Fox, A.D., Petersen, I.K., 2006. Assessing the de-

CHAPTER 3 gree of habitat loss to marine birds from

studies. Environmental Monitoring and As-

the development of offshore wind farms, in:

sessment 113, 31-48.

Boere, G.C., Galbaraith, C.A., Stroud, D.A. (Eds.),

Masden, E. a, Haydon, D.T., Fox, A.D., Furness, R.W.,

Waterbirds Around the World. The Stationery

2010. Barriers to movement: Modelling ener-

Office, Edimburg, UK, pp. 801-804.

getic costs of avoiding marine wind farms

Grünbaum, D., Veit, R.R., 2003. Black-browed albatrosses foraging on antarctic krill: Density-dependence through local enhancement? Ecology 84, 3265–3275.

amongst breeding seabirds. Marine Pollution Bulletin 60, 1085-91. Moreno, R., Jover, L., Diez, C., Sanpera, C., 2011. Seabird feathers as monitors of the levels and

Hernández-Matías, A., 2003. An approach to colo-

persistence of heavy metal pollution after the

niality in seabirds: The common tern Sterna

Prestige oil spill. Environmental Pollution 159,

hirundo as a case study. Thesis, University of

2454-60.

Barcelona.

Navarro, J., Oro, D., Bertolero, A., Genovart, M.,

Hüppop, O., Dierschke, J., Exo, K.-michael, Fre-

Delgado, A., Forero, M.G., 2010. Age and sexual

drich, E., Hill, R., 2006. Bird migration studies

differences in the exploitation of two anthro-

and potential collision risk with offshore wind

pogenic food resources for an opportunistic

turbines. Ibis 148, 90-109. Jiménez, J., Rossi, J.P., Lavelle, P., 2001. Spatial distribution of earthworms in acid-soil savannas of the eastern plains of Colombia. Applied Soil Ecology 17, 267-278.

seabird. Marine Biology 157, 2453–2459. Nettleship, D.N., Duffy, D.C., 1993. Seabird populations. Elsevier Applied Science, London. Noer, H., Christensen, T.K., Clausager, I., Petersen, I.K., 2000. Effects on birds of an offshore wind

Kendal, W.S., 2004. Taylor’s ecological power law

park at Horns Rev: Environmental impact

as a consequence of scale invariant exponen-

assessment. NERI report, MEE-NERI. <http://

tial dispersion models. Ecological Complexity

www.hornsrev.dk/miljoeforhold/miljoerap-

1, 193–209.

porter/Baggrundsrapport_18.pdf>

Kilpatrick, A., Ives, A., 2003. Species interactions can explain Taylor’s power law for ecological time series. Nature 422, 65–68.

(accessed

28.02. 12) Orians, G.H., Pearson, N.E., 1979. On the theory of central place foraging, in: Horn, D.J., Mitchell,

Kunz, T.H., Arnett, E.B., Cooper, B.M., Erickson,

R.D., Stairs, G.R. (eds) (Eds.), Analysis of Ecolo-

W.P., Larkin, R.P., Mabee, T., Morrison, M.L.,

gical Ecosystems. Colombus: Ohio State Uni-

Strickland, M.D., Szewczak, J.M., 2007. Assessing Impacts of Wind-Energy Development on

versity Press, pp. 155-177. Oro, D., 1998. Larus audouinii Audouin’s Gull, BWP

Nocturnally Active Birds and Bats: A Guidance

Update (pp. 47-61). Oxford University Press.

Document. Journal of Wildlife Management

Oro, D., Bosch, M., Ruiz, X., 1995. Effects of trawl

71, 2449-2486. Louzao, M., Hyrenbach, K.D., Arcos, J., Abelló, P., de Sola, L.G., Oro, D., 2006. Oceanographic habitat

moratorium on the breeding paramenters of the Yellow-legged Gull Larus cachinnans. Ibis 137, 547-549

of an endangered Mediterranean procellarii-

Oro, D., Pérez-Rodríguez, A., Martínez-Vilalta, A.,

form: Implications for marine protected areas.

Bertolero, A., Vidal, F., 2009. Interference com-

Ecological Applications 16, 1683-1695.

petition in a threatened seabird community: A

Mallory, M.L., Gilchrist, H.G., Braune, B.M., Gaston, A.J., 2006. Marine birds as indicators of

paradox for a successful conservation. Biological Conservation 142, 1830-1835.

Arctic marine ecosystem health: linking the

Oro, D., Ruiz, X., 1997. Exploitation of trawler dis-

Northern Ecosystem Initiative to long-term

cards by breeding seabirds in the north-wes-

Research papers tern Mediterranean: differences between the

Sierra, J., Sánchez-Arcilla, A., GonzálezDelRio, J.,

Ebro Delta and the Balearic Islands areas. ICES

Flos, J., Movellán, E., Mösso, C., Martínez, R.,

Journal of Marine Science: Journal du Conseil

Rodilla, M., Falco, S., Romero, I., 2002. Spatial

54, 695-707.

distribution of nutrients in the Ebro estuary

Paiva, V.H., Ramos, J. a., Martins, J., Almeida, A., Carvalho, A., 2007. Foraging habitat selection by Little Terns Sternula albifrons in an estuarine lagoon system of southern Portugal. Ibis 150, 18-31. Palomera, I., 1992. Spawning of anchovy Engraulis

and plume. Continental Shelf Research 22, 361-378. Taylor, L.R., 1961. Aggregation, Variance and the Mean. Nature 189, 732-735. Taylor, L.R., Woiwod, I.P., 1980. Temporal Stability as a Density-Dependent Species Characteris-

encrasicolus in the Northwestern Mediterra-

tic. The Journal of Animal Ecology 49, 209-224.

nean relative to hydrographic features in the

Taylor, L.R., Woiwod, I.P., 1982. Comparative Sy-

region. Marine Ecology Progress Series 79, 215-

noptic Dynamics. I. Relationships between

223.

Inter- and Intra-specific spatial and temporal

Pedrocchi, V., Oro, D., González-solís, J., Ruiz, X., Jover, L., 2002. Differences in diet between the

variance/mean population parameters. The Journal of Animal Ecology 51, 879-906.

two largest breeding colonies of Audouin’s

Ward, P., Zahavi, A., 1973. The importance of cer-

gulls: the effects of fishery activities. Scientia

tain assemblages of birds as “information-

Marina 66, 313-320.

centres” for food-finding. Ibis 115, 517-534.

Perrow, M.R., Gilroy, J.J., Skeate, E.R., Tomlinson,

Wiese, F.K., Montevecchi, W.., Davoren, G.., Huett-

M.L., 2011. Effects of the construction of Scroby

mann, F., Diamond, A.., Linke, J., 2001. Seabirds

Sands offshore wind farm on the prey base of

at risk around offshore oil platforms in the

Little tern Sternula albifrons at its most im-

North-west Atlantic. Marine Pollution Bulletin

portant UK colony. Marine Pollution Bulletin

42, 1285-1290.

62, 1661-70. Pinheiro, J.., Bates, D.M., 2000. Mixed-Effects Models in S and S-PLUS, Front matter. SpringerVerlag, New York.

Wood, S., 2006. Generalized additive models: an introduction with R. Chapman and Hall/CRC, Boca Raton. Östman, O., 2002. Distribution of bird cherry-oat

R Development Core Team, 2008. R: A Language

aphids (Rhopalosiphum padi (L.)) in relation to

and Environment for Statistical Computing.

landscape and farming practices. Agriculture,

Vienna Austria R Foundation for Statistical

Ecosystems & Environment 93, 67-71.

Computing.

CHAPTER 3

Research papers resum Conèixer les estratègies d’alimentació dels depredadors marins és essencial per comprendre els factors intrínsecs que controlen la seva distribució, abundància i la seva funció ecològica en l’ecosistema marí. En el següent capítol, es va investigar per primera vegada els moviments de cerca d’aliment i els patrons d’activitat de la gavina corsa Larus audouinii mitjançant l’ús de dades de localització per satèl·l it a partir de vuit adults reproductors a la colònia principal de l’espècie a tot el món (el Delta de l’Ebre, Mediterrània NO). Les gavines marcades s’alimentaven a la zona marina propera a la colònia de cria (62% dels llocs d’alimentació) i a l’àrea terrestre del Delta de l’Ebre (principalment els camps d’arròs, el 38% dels llocs d’alimentació). Els patrons d’activitat de cerca d’aliment va canviant significativament al llarg del dia; El seu mínim va del capvespre fins a la primera meitat de la nit (19-1 h, el 32% dels llocs actius) i és més alt durant la resta del dia (1-19 h; 75,5 ± 4,3% ubicacions d’actives). Aquests resultats confirmen la plasticitat alimentària d’aquesta au marina i, en base a la informació anterior sobre els hàbits alimentaris d’aquesta espècie, hipotetitzem sobre com els seus patrons d’activitat temporal i l’ús que fa de l’hàbitat podrien estar associats amb variacions en la disponibilitat de recursos alimentaris marins (per exemple, les migracions verticals diàries dels peixos pelàgics) i de l’explotació dels recursos terrestres (per exemple, crancs de riu americà Procambarus clarkii).

JOURNAL REFERENCE Christel I, Navarro J, del Castillo M, Cama A, Ferrer X (2012) Foraging movements of Audouin’s gull (Larus audouinii) in the Ebro Delta, NW Mediterranean: A preliminary satellite-tracking study. Estuarine, Coastal and Shelf Science 96:257–261 PDF available in the Appendix, page 134

Foraging movements of Audouin’s gull (Larus audouinii) in the Ebro Delta, NW Mediterranean: A preliminary satellite-tracking study Isadora Christel1,2, Joan Navarro3, Marcos del Castillo4, Albert Cama1,2 and Xavier Ferrer1 1

Institute for Research on Biodiversity (IRBio) and Departament de Biologia Animal, Universitat de Barcelona (UB). Diagonal 645, E-08028 Barcelona, Spain.

Museo Nacional de Ciencias Naturales, Consejo Superior de Investigaciones Científicas. C/ José Gutiérrez Abascal 2, 28006, Madrid, Spain. 3 Institut de Ciències del Mar (ICM-CSIC), P. Marítim de la Barceloneta 37-49, 08002 Barcelona, Spain. 4 C/ Vilanova 8A, 07002 Palma de Mallorca, Mallorca, Illes Balears, Spain.

A bstract Knowing the foraging strategies of marine predators is essential to understand the intrinsic factors controlling their distribution, abundance and their ecological function within the marine ecosystem. Here, we investigated for the first time the foraging movements and activity patterns of Audouin’s gull Larus audouinii by using satellite-tracking data from eight breeding adults in the main colony of the species worldwide (Ebro Delta, NW Mediterranean). Tagged gulls foraged in the marine area close to the breeding colony (62% of foraging locations) and in the terrestrial area of the Ebro Delta (mainly rice fields; 38% of foraging locations). The foraging activity patterns changed significantly throughout the day; lower from dusk through the first half of the night (19-1 h; 32% of active locations) and higher during the rest of the day (1-19 h; 75.5±4.3% of active locations). These results confirm the foraging plasticity of this seabird and, based on previous information about the dietary habits of this species, we hypothesize how its time-dependent activity patterns and habitat use could be associated with variations in the availability of marine food resources (e.g. diel vertical migrations of pelagic fish) and the exploitation of terrestrial resources (e.g. American crayfish Procambarus clarkii).

Research papers Introduction

cially relevant in breeding populations located in areas where diverse trophic resources

An important issue in the feeding ecolo-

are highly available (e.g. Oro and Ruiz, 1997;

gy of marine predators is the degree of plas-

Oro et al., 1999; Navarro et al., 2010), which is

ticity of their foraging behaviour. In general,

the case of the breeding population located

specialist predators are constrained to fora-

in the Ebro Delta (Fig.1. NW Mediterranean).

ge on a specific habitat and time of day de-

This colony holds around 12000-13000 bree-

termined by a specific prey availability (Fu-

ding pairs of Audouin’s gull, ca. 65% of the

tuyma and Moreno, 1988; Krebs and Davies,

total world population (Oro et al., 2009). The

1993; Julliard et al., 2006). Under changing

marine ecosystem of the Ebro Delta is one of

conditions of prey availability, specialists

the most important fishing grounds in the

are able to dapt their foraging strategy by

Mediterranean Sea, resulting in one of the

extending foraging range or time spent fo-

largest fishing fleets in this region, which

raging (e.g. Oro et al., 1997; Lewis et al., 2001;

generates a high quantity of fisheries dis-

Schwemmer and Garthe, 2008). By contrast,

cards (Coll et al., 2008). Moreover, freshwater

generalist predators have the ability to ex-

resources such as the invasive American

ploit different trophic resources and, con-

crayfish Procambarus clarkii in the rice fields

sequently, they present higher plasticity in

of the Ebro Delta are abundant and easily

their foraging strategies (Krebs and Davies,

available (Gutierrez-Yurrita et al., 1999), pro-

1993; Boyd et al., 2006; Julliard et al., 2006).

viding an alternative and proficient trophic

This opportunistic behaviour allows genera-

resource for the species (Oro et al., 1996b;

lists to modify their foraging strategies (i.e.

Longoni, 2010; Navarro et al., 2010).

exploited habitat, range or temporal pat-

Although the diet habits of the Audouin’s

terns) according, for instance, to the varying

gull are well known (e.g. Oro et al., 1997; Pe-

degree of competition for food. Indeed, the

drocchi et al., 2002; Sanpera et al., 2007; Na-

foraging plasticity of marine predators has

varro et al. 2010), detailed information on

allowed these organisms to benefit from

the foraging movements is biased toward

anthropogenic food resources (e.g. fisheries

studies based on ship surveys (e.g. Abelló

discards, refuse dumps or introduced prey

and Oro,1998; Arcos et al., 2001; Abelló et al.,

species; Tablado et al., 2010; Ramos et al.,

2003), which are strongly biased by the in-

2011; Wagner and Boersma, 2011).

fluenceof fishery discards and underestima-

Amongst

marine

predators,

the

te the importance of land habitat utilization.

Audouin’s gull Larus audouinii is a good

The only previous telemetric study (radio-

example of an opportunist species that ex-

tracking) already pointed to the apparent

hibits clear plasticity in its diet habits. This

importance of the terrestrial habitat for the

Mediterranean endemic species exploits

breeding population of the Ebro Delta colony

small pelagic fish (their main prey, see Oro

(Mañosa et al., 2004).

1998 and references therein), but also al-

Here, we present preliminary results of

ternative anthropogenic resources such as

the first satellite-tracking study of Audouin’s

demersal or benthonic fish from fisheries

gull during the breeding season in its lar-

discards or invasive freshwater crabs from

gest breeding colony (Ebro Delta). This pa-

terrestrial habitat (Oro et al., 1996a; Oro and

per aims to quantify the foraging range of

Ruiz, 1997; Oro et al., 1999; Navarro et al.,

Audouin’s gull, evaluate the habitat utili-

2010). This opportunistic behaviour is espe-

zation of marine and terrestrial areas and

CHAPTER 4

Fig.1 (a) Breeding

identify the temporal patterns of the fora-

battery powered “Platform Transmitter Ter-

ging activity of the species. Based on pre-

minals” (PTTs; North Star Science and Tech-

vious information about the dietary habits

nology, LLC) during the chick-rearing period

of this species, we also hypothesize how the

(May) of 2006 (Table 1). We captured all birds

observed foraging movements could be at-

on the nest by using a drop trap (Mills and

tributed to the exploitation of different tro-

Ryder, 1979) during late incubation to redu-

phic resources in the Ebro Delta marine and

ce the risk of desertion. Once trapped, each

terrestrial ecosystems.

individual was sexed, weighed, ringed and tagged with a PTT. The attached PTTs weig-

Material and methods

hed 20 g and were programmed to be active in a 6 h on/5 h off duty cycle to get infor-

Fieldwork procedures

mation on the foraging locations during one month. The PTT was fixed to the mid-dorsal

The study was carried out at the natural

feathers of the mantle using Tesa tape (Wil-

reserve of Punta de la Banya in the Ebro Del-

son et al., 1997). With this method the PTT

ta Natural Park, North Western Mediterra-

falls off after one month without the neces-

nean Sea (Fig. 1, 40º33’N, 0º39’E). Punta de la

sity to recapture the instrumented bird. The

Banya is a flat sandy peninsula of 2,514 ha,

entire transmitter equipment represented

partially occupied by saltworks and connec-

between 3 and 4% of the Audouin’s gull’s

ted to extensive rice field areas (20,000 ha)

body mass, so the potential effects of an

by a 5 km-long narrow sand bar. To examine

additional weight on the gull’s movement

the foraging activity, we satellite-tracked 8

were minimized (e.g. Phillips et al., 2003;

breeding birds (4 males and 4 females) using

Passos et al., 2010).

areas of the Mediterranean endemic Audouin’s gull Larus audouinii and study area: Ebro Delta, NW Mediterranean. (BirdLife International, 2011) (b) Map of the Ebro Delta area indicating the Audouin’s gull colony position with an asterisk and 1 km buffer area around la “Punta de la Banya” peninsula, the rice fields and wetlands shaded in dark gray and the location of the main harbours. (c) Foraging locations of 7 satellite-tracked Audouin’s gulls during the breeding period of 2006. To better visualize the foraging locations’ range the Minimum Convex polygon (short dashed line) is shown beside the 95% (solid line) and 50% (long dashed line) kernel polygons.

Research papers

Table.1 Summary information of PTTs performance

and the first kilometer around it were “fora-

Satellite-tracking data and statistical analyses

ging locations” (we assumed that the birds were feeding to recover the body condition

Data on the position of each PTT were

lost during the incubation bout). Finally, we

obtained from ARGOS system (CLS, Tou-

calculated the 95% fixed-kernel estimates of

louse, France) and imported to ArcView 3.2

the foraging area and the maximum fora-

(ESRI) using the Argos Tool extension (Pota-

ging distance from the colony.

pov and Dubinin, 2005). Each position was

We employed logistic regression – a ge-

classified according to its estimated error:

neralized linear model (GLM) – to test the fo-

Type 0 (>1000 m), Type 1 (350-1000 m), Type

raging activity and habitat use. First, we tes-

2 (150-350 m), Type 3 (0-150 m), and Types A

ted a model with the proportion of foraging

and B (without an estimated error) (ARGOS,

locations as the dependent variable, and

2006). Initial data filtering involved calcula-

we selected as the explanatory variable the

ting velocities between successive satellite

“time of day” -categorized in 6-hour inter-

locations, and rejecting those for which the

vals (1-7 h; 7-13 h; 13-19 h; 19-1 h)- with the

velocity exceeded a threshold of 50 m·s-1,

7-13 h interval as the reference level. Then,

the maximum velocity described for this

we analyzed habitat use by testing the effect

species (Rosén and Hedenström, 2001). By

of the explanatory variable “time of day” on

this procedure, up to 8 % of the locations

the dependent variable “terrestrial vs. ma-

were filtered; all of them from the low-qua-

rine proportion of foraging locations”. The

lity accuracy class “B”.

analyses were carried out using R software

To gain insight into the foraging activi-

(R Development Core Team, 2008), calling

ty of the tagged Audouin’s gulls we sorted

the “glm” function with binomial error dis-

the locations into three classes, according

tribution and its default logit link function.

to their spatial position. PTT locations in-

A likelihood ratio test was used to compa-

side the “Punta de la Banya” peninsula or

re the resulting model with the null model

within the first kilometer around it were

(without any variable) and to assess the sig-

classified into the “colony locations” group.

nificance of the explanatory variable “time

In contrast, the locations outside the colony

of day”.

CHAPTER 4

Results

Fig.2 Example of foraging trajectories for the individual “58980” (see Table 1 for more information)

We obtained a total of 89 filtered PTT locations spanning a period of 13 consecutive days. One of the eight PTTs failed to give any location probably due to a battery failure, and the performance of the remaining PTTs was heterogeneous (see Table 1). Due to sample size limitations individual variability was not included in the analysis, but the movements of one of the tracked individuals is shown in Figure 2 to illustrate the

day, they foraged mainly in marine rather

general pattern of the foraging movements.

than terrestrial habitat (1-7h: 50% marine,

The

the

28% terrestrial habitat; 7-13h: 59% marine,

Audouin’s gulls was 5400 km2 (95% fixed-

foraging

area

covered

19% terrestrial habitat; 19-1h: 23% marine,

kernel density estimate), covering both the

9% terrestrial habitat)(Fig. 3a).

marine area of the Ebro Delta (ca. 3300 km2) and the terrestrial area (ca. 2100 km2) (Fig.

Discussion

1c). The maximum foraging distance covered ranged from 20.5 to 81.7 km (mean ± sd

Satellite-tracked Audouin’s gulls covered

= 51.5 ± 24.3 km) and was similar for both

a foraging area that ranges 80 km, span-

marine and terrestrial locations (T-Student

ning both marine and terrestrial habitats. It

test, T = 1.44, df = 56, p = 0.15).

has been widely described previously that

The foraging activity changed signifi-

breeding Audouin’s gulls cover large ran-

cantly over the course of the day (Likelihood

ges when foraging. There are records of in-

Ratio Test, χ =13.79, df =3, p=0.003). Tagged

dividuals foraging at 70 to 150 km from the

gulls were more active at 7-13 h (78.1%), at

breeding colony during the breeding season

1-7 h (77.8% of the total locations in this pe-

(Baccetti et al., 2000; Mañosa et al., 2004),

riod, p = 0.65), and 13-19h (70.6%, p = 0.56),

and data from vessel counts suggest that

all of them significantly different from the

individuals forage during the day and night

19-1 h interval (31.8%, p= 0.001), i.e., the fora-

even further offshore (Abelló and Oro, 1998;

ging activity diminished during the first half

Arcos and Oro, 1996). However, the species’

of the night (Fig. 3b). Moreover, we found

terrestrial foraging movements had been

that the proportion of foraging locations in

scarcely described (Ruiz et al., 1996; Mañosa

marine vs. terrestrial habitats changed du-

et al., 2004).

ring the day. Although the time of day was

It is well documented that Audouin’s

not significant as a global explanatory va-

gulls forage during the night in marine

riable, the model indicated a significant di-

habitats preying on small pelagic fish and

fference between the 13-19 h interval and

exploiting discards provided by nocturnal

the reference level 7-13 h (p= 0.04) (Fig. 3c).

fisheries (e.g. Witt et al. 1981; Mañosa et al.

Between 13h and 19 h, Audouin’s gulls fora-

2004; Arcos et al., 2008). However, our re-

ged mainly in terrestrial (41%) rather than in

sults highlight that the species’ nocturnal

marine habitat (29%); during the rest of the

activity is not homogeneous throughout

Research papers the night (see Fig.3). Satellite-tracked gulls were mainly located in the breeding colony during the hours before and after dusk (19-

Fig.3 (a) Activity (foraging in marine or terrestrial habitat; or located in the colony) during a 24h cycle of 7 satellite-tracked Audouin’s gulls during the breeding period in Ebro Delta colony. (b,c) Mean and 95% confidence interval, of the foraging probability and foraging in marine habitat probability respectively, according to the GLM models. * indicates a significant difference of the time block probability compared to the reference level 7-13 h.

1 h). In the period after midnight to dawn (1-7 h) they increased their foraging activity, which then remained constant and high during the day. These results, coupled with the nocturnal arrival and departure times from the breeding colony described in Mañosa et al. (2004), confirm a peak of activity between midnight and dawn. Attendance to purse seiners during the night is considered a strategy that is only significant during trawling moratorium and winter periods (Arcos and Oro, 2002), neither of which were covered during our study; therefore, the individuals located at sea during the night were probably feeding on small pelagic fish. Accordingly, the nocturnal foraging habits of the Audouin’s gull would still rely on the capture of small pelagic fish (Witt et al., 1981; Oro, 1998), a resource that might not be available throughout the night, but only in the hours before dawn due to the diel vertical migration of the shoals (Blaxter and Hunter, 1982; Oro, 1998). With regard to diurnal activity, tagged birds showed a high foraging activity with an unexpected constant presence in terrestrial habitats (generally rice fields or wetlands) in addition to the expected presence in marine habitat (Oro, 1998). The fact that all tagged individuals could be found in both habitats suggests that the use of terrestrial habitat was not due to the casual behaviour of a single individual. This result supports previous studies that describe the use of the rice fields of the Ebro Delta by the Audouin’s gull (Ruiz et al., 1996; Mañosa et al., 2004; Longoni, 2010), probably related to the exploitation of the exotic American crayfish (Navarro et al., 2010), which is very abundant in the rice fields of the Ebro Delta (Gutierrez-Yurrita et al., 1999). Although

CHAPTER 4 many studies have demonstrated that the

Park team (T.Curcó, C. Vidal and F. Blanch).

Audouin’s gull exploits trawler discards (Oro

S. Young revised the English. Research funds

et al., 1997; Arcos, 2001; Cama, 2010), the fo-

were provided by a project funded by Capi-

raging activity of our satellite-tracked indi-

tal Energy through agreement with Funda-

viduals was higher inland than at sea in a

ció Bosch i Gimpera (Contract 304683). I. C.

period of time that includes the discarding

was funded by a PhD fellowship of the Uni-

peak of the trawling fleet (from 15 to 16 h;

versity of Barcelona. J. N. was supported by

Cama, 2010). This result suggests that te-

a postdoctoral contract of Juan de la Cierva

rrestrial foraging has become an alternative

program (MICINN-JDC, Spanish Ministry of

food source to trawling discards (Navarro et

Science and Innovation). A. C. was funded

al., 2010), probably prompted by the inter-

by a PhD fellowship of the Government of

ference competition for fisheries discards:

Catalonia (2009FIC75).

namely, intraspecific competition (due to an increasing population density), and inters-

References

pecific competition with the sympatric and dominant Yellow legged gull Larus michahellis (e.g. Arcos et al., 2001). In conclusion, the present study

Abelló, P., Oro D., 1998. Offshore distribution of seabirds in the northwestern Mediterranean in June 1995. Colonial Waterbirds 21, 422-426.

shows that Audouin’s gull foraged in both

Abelló, P., Arcos, J., Gil Sola, L., 2003. Geographical

marine and terrestrial habitats and showed

patterns of seabird attendance to a research

activity during both night and day. These re-

trawler along the Iberian Mediterranean coast.

sults confirm the high foraging plasticity of

Scientia Marina 67, 69–75.

Audouin’s gull, a species once defined as a

Arcos, J., 2001. Foraging ecology of seabirds at sea:

specialist nocturnal forager that has beco-

significance of commercial fisheries in the

me an opportunist on fisheries discards and

NW Mediterranean. PhD Thesis, University of

terrestrial resources. However, due the limi-

Barcelona.

ted sample size we suggest the necessity of

Arcos, J., Oro, D., 1996. Changes in foraging range

conducting more studies using biologging

of Audouin’s gulls Larus audouinii in relation to

methodologies (such as PTTs or GPS) to con-

a trawler moratorium in the western Medite-

firm the observed patterns and to gain new

rranean. Colonial Waterbirds 1, 128-131.

insight into the foraging ecology of this endangered seabird.

Arcos, J., Oro, D., Sol, D., 2001. Competition between the yellow-legged gull Larus cachinnans and Audouin’s gull Larus audouinii associated

Acknowledgments

with commercial fishing vessels: the influence of season and fishing fleet. Marine Biology 139,

The birds were tagged with a permit

807-816.

from the Environmental Department of the

Arcos, J., Oro, D., 2002. Significance of nocturnal

Catalonian Government. We are grateful to

purse seine fisheries for seabirds: a case study

D. Oro, L. Cardador and J. M. Arcos for their

off the Ebro Delta (NW Mediterranean). Marine

comments to improve this manuscript. F.

Biology 141, 277–286.

Zino, C. Carboneras and J. González-Solís for

Arcos, J., Louzao, M., Oro, D., 2008. Fishery ecosys-

their comments about attachment methods.

tem impacts and management in the medite-

We also appreciate the help of X. Macià, R.

rranean: seabirds point of view. American Fis-

Loras, S. Mañosa and the Ebro Delta Natural

heries Society Symposium 49, 1471-1479.

Research papers ARGOS, 2006. Argos User’s Manual. ARGOS CLS, Toulouse, France. http://www.argossystem. org/manual/link. Last accessed 24 November 2011. Baccetti, N., Dall’Antonia, L., Magnani, A., Serra, L., 2000. Foraging routes of Audouin’s Gulls

the best of a bad situation?. Crustacean Issues, Balkema, Rotterdam, pp. 161–192. Julliard, R., Clavel, J., Devictor, V., Jiguet, F., Couvet, D., 2006. Spatial segregation of specialists and generalists in bird communities. Ecology letters 9, 1237-44.

Larus audouinii from two Sardinian colonies.

Krebs, J.R., Davies, N.B., 1993. An introduction to

In: Sultana, J., Yésou, P. (Eds.), Monitoring and

behavioural ecology. Wiley-Blackwell publis-

conservation of birds, mammals and sea turt-

hing, Oxford, 420 pp.

les of the Mediterranean and Black Seas. En-

Lewis, S., Sherratt, T.N., Hamer, K.C., Wanless, S.,

vironment Protection Department, Malta, pp.

2001. Evidence of intra-specific competition

150-158.

for food in a pelagic seabird. Nature 412, 816-

BirdLife Iternational, 2011. Distribution maps of Birds of the World. BridLife International, Cambridge, UK. Blaxter, J.H.S., Hunter, J.R., 1982. The biology of clupeoid fishes. Advances in Marine Biology 20, 1-223. Boyd, I.L., Wanless, S., Camphuysen, C.J. (Eds.), 2006. Top predators in marine ecosystems:

819. Longoni, V., 2010. Rice Fields and Waterbirds in the Mediterranean Region and the Middle East. Waterbirds 33, 83-96. Mañosa, S., Oro, D., Ruiz, X., 2004. Activity patterns and foraging behaviour of Audouin’s gulls in the Ebro Delta, NW Mediterranean. Scientia Marina 68, 605-614.

their role in monitoring and management.

Mills, J.A., Ryder, J.P., 1979. Trap for Capturing Sho-

Cambridge University Press, Cambridge, 378

re and Seabirds. Bird-Banding, 50, 121-123.

pp.

Navarro, J., Oro, D., Bertolero, A., Genovart, M., Del-

Cama, A., 2010. Seabird distribution patterns and

gado, A., Forero, M.G., 2010. Age and sexual

ecological factors driving larid presence at

differences in the exploitation of two anthro-

the Ebro delta shelf (NW Mediterranean). PhD

pogenic food resources for an opportunistic

Thesis, University of Barcelona.

seabird. Marine Biology 157, 2453-2459.

Coll, M., Palomera, I., Tudela, S., Dowd, M., 2008.

Orians, G.H., Pearson, N.E., 1979. On the theory

Food-web dynamics in the South Catalan Sea

of central place foraging. In: Horn, D.J., Stairs,

ecosystem (NW Mediterranean) for 1978–2003.

G.R., Mitchell, R.D. (Eds.), Analysis of Ecological

Ecological Modelling 217, 95-116.

Ecosystems. Ohio State University Press, Co-

Futuyma, D.J., Moreno, G., 1988. The evolution of ecological specialization. Annual Review of Ecology and Systematics 19, 207–233. Garthe, S., Camphuysen, C.J., Furness, R.W., 1996.

lombus, pp. 155-177. Oro, D., 1998. Larus audouinii Audouin’s Gull. BWP Update. Oxford University Press, Oxford, pp. 47-61

Amounts discarded by commercial fisheries

Oro, D., Jover, L., Ruiz, X., 1996a. Influence of

and their significance as food for seabirds in

trawling activity on the breeding ecology of

the North Sea. Marine Ecology Progress Series

a threatened seabird, Audouin’s gull Larus

136, 1-11.

audouinii. Marine Ecology Progress Series 139,

Gutierrez-Yurrita, P., Martínez, J., Ilhéu, M., Bravo-

19-29.

Utrera, M., Bernardo, J., Montes, C., 1999. The

Oro, D., Genovart, M., Ruiz, X., Jiménez, J., García-

status of crayfish populations in Spain and

Gans, J., 1996b. Differences in diet, population

Portugal. In: Gerardi, F., Holdich, D. (Eds.), Cra-

increase and breeding performance between

yfish in Europe as alien species: How to make

two colonies of Audouin’s gulls Larus audouinii

CHAPTER 4 during breeding seasons affected by a trawling

conservation management sciences. Diversity

moratorium. Journal of Avian Biology 27, 245-

and Distributions 17, 1-12.

251.

R Development Core Team, 2008. R: A Language

Oro, D., Ruiz, X., 1997. Exploitation of trawler dis-

and Environment for Statistical Computing.

cards by breeding seabirds in the north-wes-

R Foundation for Statistical Computing, Vien-

tern Mediterranean: differences between the

na, Austria. ISBN 3-900051-07-0, URL http://

Ebro Delta and the Balearic Islands areas. ICES Journal of Marine Science 54, 695-707.

www.R-project.org/. Rosén, M., Hedenström, A., 2001. Testing predic-

Oro, D., Ruiz, X., Jover, L., Pedrocchi, V., González-

tions from flight mechanical theory : a case

Solís, J., 1997. Diet and adult time budgets of

study of Cory’s shearwater and Audouin’s gull.

Audouin’s gull Larus audouinii in response to changes in commercial fisheries. Ibis 139, 631637. Oro, D., Pradel, R., Lebreton, J.D., 1999. Food availability and nest predation influence life history traits in Audouin’s gull, Larus audouinii. Oecologia 118, 438-445.

Acta ethologica 3, 135-140. Ruiz, X., Oro, D., Martínez-Vilalta, A., Jover, L., 1996. Feeding ecology of Audouin’s gull Larus audouinii in the Ebro Delta. Colonial Waterbirds 19, 68-74. Sanpera, C., Ruiz, X., Moreno, R., Jover, L., Waldron, S., 2007. Mercury and stable isotopes

Oro, D., Cam, E., Pradel, R., Martínez-Abraín, A.,

in feathers of Audouin’s gulls as indicators

2004. Influence of food availability on demo-

of feeding habits and migratory connectivity.

graphy and local population dynamics in a

The Condor 109, 268-275.

long-lived seabird. Proceedings of the Royal Society of London 271, 387-396.

Schwemmer, P., Garthe, S., 2008. Regular habitat switch as an important feeding strategy of an

Oro, D., Pérez-Rodríguez, A., Martínez-Vilalta, A.,

opportunistic seabird species at the interface

Bertolero, A., Vidal, F., 2009. Interference com-

between land and sea. Estuarine, Coastal and

petition in a threatened seabird community: A paradox for a successful conservation. Biological Conservation 142, 1830-1835.

Shelf Science. 77, 12-22. Tablado, Z., Tella, J.L., Sánchez-Zapata, J.A., Hiraldo, F., 2010. The paradox of the long-term posi-

Passos, C., Navarro, J., Giudici, A., González-Solís,

tive effects of a North American Crayfish on a

J., 2010. Effects of Extra Mass on the Pelagic

European community of predators. Conserva-

Behaviour of a Seabird. The Auk 127, 100-107.

tion biology 24, 1230-1238.

Pedrocchi, V., Oro, D., González-Solís, J., Ruiz, X.,

Wagner, E.L., Boersma, PD., 2011. Effects of fishe-

Jover, L., 2002. Differences in diet between the

ries on seabird community ecology. Reviews in

two largest breeding colonies of Audouin’s gulls: the effects of fishery activities. Scientia Marina 66, 313-320.

Fisheries Science 19, 157-163. Wilson, R.P., Pütz, K., Peters, G., Culik, B., Scolaro, J.A., Charrassin, J.B., Ropert-Coudert, Y.,

Phillips, R., Xavier, J.C., Croxall, J.P., Burger, A.E.,

1997. Long-term attachment of transmitting

2003. Effects of satellite transmitters on alba-

and recording devices to penguins and other

trosses and petrels. The Auk 120, 1082-1090.

seabirds. Wildlife Society Bulletin 25, 101-106.

Potapov, E., Dubinin, M., 2005. Argos-tools manual.

Witt, H.H., Crespo, J., de Juana, E., Varela, J.M., 1981.

The Falcon Research Institute, National Avian

Comparative feeding ecology of Audouin’s Gull

Research Center, Carmarthen, UK.

Larus audouinii and the Herring Gull L. argenta-

Ramos, R., Ramírez, F., Carrasco, J.P., Jover, L., 2011. Insights into the spatiotemporal component of feeding ecology: an isotopic approach for

tus in the Mediterranean. Ibis 123, 519-526.

Discussion & Conclusions

DISCUSSION

DISCUSSION Gaining insight: What I learned from the experience

Vulnerability Index

not important in themselves, but as an aggravation factor that can increase a risk if it

The work of Garthe and HĂźppop (2004)

already exists (e.g. flight manoeuvrability).

proposed the quantitative assessment of

In other words, if a seabird species never

the vulnerability of a seabird community

flies within a height with risk of collision, it

to wind farms. The index framework calcu-

is irrelevant the manoeuvrability of the spe-

lated this vulnerability through the Species

cies since it will never have to avoid a turbi-

Sensitivity Index (SSI) that focuses on the

ne. This means that factors have a hierarchi-

species vulnerability (at individual and po-

cal structure between primary risk factors

pulation levels) and the Wind Farm Sensiti-

and aggravation factors that cannot be dealt

vity Index (WSI) that applies the SSI to esti-

with an additive formulation. Therefore, an

mate the community vulnerability. However,

alternative power function is suggested to

as I show in Chapter 1, the mathematical

improve the estimate of collision and dis-

formulation of the original index contains

turbance risk.

hidden assumptions at both speciesâ&#x20AC;&#x2122; and

The second assumption was that each

community levels that might lead to inco-

type of risk (collision, disturbance and po-

rrect estimates of vulnerability and a biased

pulation sensitivity) was equally important

identification of key areas.

and had a multiplicative relationship. It is

The first assumption was that all risk

difficult to measure the relative importan-

factors associated to a given type of risk had

ce of collision risk over disturbance risk,

equal importance and had an additive rela-

which justifies considering them as equal

tionship. Nevertheless, there is a conceptual

by default despite collision mortality is of-

difference between the factors included in a

ten considered to be the most important

particular type of risk. Taking collision risk

hazard (Fox et al. 2006; Christel, Certain, et

as an example, we find two types of risk

al. 2012). Having a multiplicative relations-

factors: those directly associated to the risk

hip between collision and disturbance risk

itself (e.g. percentage of time spent at high

is conflictive because these two types of risk

altitude when flying) and factors that are

do not depend on each other. They are two

Discussion & Conclusions independent aspects of the potential impact

larity of these species.

of an OWF on seabirds and thus, collision

After incorporating the changes in the

and disturbance impact have different eco-

index formulation and applying the index to

logical consequences. By linking them with

a large study area (the coasts of the Iberian

a multiplicative relationship might result in

Peninsula), there are some conclusions and

the underestimation of a risk effect, only be-

recommendations that can be drawn from

cause the other risk is very low. In fact, this is

the experience.

the case in Garthe and Hüppop’s calculation

This type of index is eminently compa-

of the SSI which is more correlated to the

rative therefore, it is better to apply it in the

disturbance risk than to the collision risk. If

broader scale possible. Iberian Peninsula is a

the two types of risks have to be combined

geographical unit and regardless the borders

in a single map I would rather use an additi-

it is interesting to see how it is in the con-

ve relationship. However, our results suggest

junct. Based on boat surveys means maxi-

against such combination but rather the in-

mum probabilities of detecting all species,

dividual examination of both collision risk

which ensures the reliability of the index.

maps and disturbance risk maps to reach an informed decision for management.

Placement decision is a trade-off between impact and benefits and it is not

The final step of the index is to integrate

scientists but managers decision. Although

the species vulnerability into a measure of

Garthe and Hüppop (2004) set a percenti-

vulnerability of a whole community. To do

le criterion for defining concern and major

so, the third assumption was that the con-

concern areas, I think that it is better to rank

tribution to the community vulnerability of

the output accordingly to their WSI avoiding

a given seabird species is proportional to the

the delimitation of a subjective value.

log abundance of that species in a particular

By applying the index at different sca-

location. The rationale for this approach was

les we find consistent patterns that remain

to prevent the installation of wind farms in

constant through the scales although loo-

areas with high aggregations of seabirds.

sing detail with larger scales. However, it

However, abundant species are usually tho-

shows that different scales can be used to

se with lower SSI values while rare species

different management purposes. The uti-

are those with higher SSI values. For a given

lization of a large scale WSI grid, with grid

location, SSI and abundance affect the final

cells between 1º and 0.5º, seems more appro-

WSI value in opposite directions, which hin-

priate for the definition of optimal develop-

ders the interpretation of the variations of

ment areas, while small scale WSI grids, e.g.

the index. Moreover, the use of log abundan-

0.25º, are better for the demarcation of areas

ce instead of plain abundance assumes that

of high vulnerability or areas of high con-

a single individual has more weight, in pro-

cern if there was a project to install an OWF

portion, than an individual located in a flock

within them. Small-scale application of the

of one hundred seabirds, which has neither

WSI seems optimal for hazard location and

ecological nor environmental support. To

the definition of high risk areas that must

solve these inconsistencies, a major chan-

be specifically excluded for the construction

ge in the formulation is suggested based on

of any OWF and the indirect influence of the

the work of Leinster and Cobbold (2012) that

cumulative impact of OWF.

presents a diversity measure based not only

Although this method is an integrati-

on species abundance but also on the simi-

ve tool and the example of the study area

DISCUSSION shows its utility, any technique based on

‘barrier effect’ (Masden, Haydon, et al. 2010;

boat or aerial surveys has some methodo-

Perrow et al. 2011). Therefore, after quanti-

logical limitations that have been described

fying the aggregative pattern in a given area,

in the methodological approach section.

the potential risk can be evaluated and used

Because of this, it is advisable to comple-

to rank regions within this area according to

ment these indexes with new methods like

different levels of concern. Areas with both

satellite tracking to complement the infor-

collision and habitat loss risk would have

mation with the offshore distribution of

the higher concern level (L3); areas with co-

flagship species, or add additional decision

llision risk would have the next concern le-

layers, e.g. Important Bird Areas.

vel (L2) followed by areas with habitat loss risk (L1). This classification can be used later

Aggregative patterns

to choose the optimal location of an OWF (by choosing areas with minimum concern)

Abundance maps define areas with high

or to define a required monitoring protocol

numbers of birds, which is relevant infor-

for an accepted OWF location according to

mation in the assessment of offshore wind

the concern level of this location. Moreover,

farms locations, but they do not provide

there is a temporal evolution of the aggrega-

information on the dynamic and complex

tive patterns and it is correlated to the life

spatial patterns of seabirds at sea. Whi-

cycle of the species. Whether we use seaso-

le density maps focus on the detection of

nal scenarios (e.g. breeding, post-breeding,

high concentrations of seabirds as potential

migration, wintering) or a set of critical

risk areas, the application of Taylor’s power

months, temporal scenarios are easier to

law allows the explicit distinction between

communicate.

transitional and foraging areas over time.

marize key information that can highlight

Taylor’s power law is widely accepted as an

the potential impact of an OWF in a sensi-

aggregation index in time and space, which

tive moment in a seabird life cycle or can be

is corroborated in Chapter 3 by its correla-

used to recommend mitigation measures

tion with mean flock size. Areas with weak

during critical months.

Temporal scenarios sum-

aggregative patterns can be considered tran-

The application of this tool and the in-

sitional or flight path areas while high ag-

terpretation of its results are particularly

gregative patterns are mostly feeding areas

useful for the assessment of areas with lar-

determined by the punctual availability of a

ge breeding populations. However, it would

feeding opportunity.

be advisable to apply this method in other

By linking the aggregative patterns with

scenarios to investigate any possible diffe-

a particular behaviour, we can better predict

rences in the interpretation of the potential

and classify the risk of wind farm establish-

risks associated to the observed aggregati-

ment on a seabird population or community.

ve patterns. For instance, areas with a di-

In transitional areas, the main risk will be

fferent composition of species like plunge-

direct collision and mortality (Desholm and

divers (e.g. gannets) or surface-divers (e.g.

Kahlert 2005; Hüppop et al. 2006). In foraging

sea ducks) might show different spatial and

areas, the presence of turbines would result

temporal patterns. It would also be desirable

in habitat loss for species with a strong avoi-

to study the outputs of the tool in a migra-

dance response or an increased collision

tory corridor. Finally, although the method

risk for the species that experience a low

results are consistent with the ornithologi-

Discussion & Conclusions cal observations and behavioural data in the

distances, diving patterns) Chapter 4 draws

Ebro Delta, I would encourage the applica-

attention to the plasticity on the temporal

tion of telemetry data to test with an inde-

activity patterns and the habitat use and

pendent data set the behavioural interpreta-

this plasticity has to be addressed in the as-

tion of the Taylor’s power law.

sessment of any offshore wind farm.

Nevertheless, the study presented in

Many studies have shown that the activi-

Chapter 3 demonstrates that considering

ty pattern of seabirds is not constant throug-

the information on the second order proper-

hout the day (e.g. Garthe et al. 2003; Cama,

ties of species’ distributions (i.e. the social

2010; Cama et al. 2012; Christel, Navarro, et

aggregation) provides further information

al. 2012). Some seabirds, for instance, re-

for the assessment of potential impacts of

lay on small pelagic fish, but this resource

offshore wind farms than solely focusing on

might only be available in the hours before

the first order properties (i.e. density).

dawn when the shoals perform their vertical migration (Blaxter and Hunter 1982). In this

Individual tracking

case, an area with recurrent aggregations of seabirds foraging on these shoals would

Seabird locations and seabird behaviour

not be detected by surveys which usually

are distinct, and the latter is an important

start after dawn. Therefore, seabird surveys,

component that can be extracted from in-

which must be performed with daylight and

dividual tracking data types (Tremblay et al.

usually following a constant schedule, are

2009). One of the aims of Chapter 4 was to

not always sufficient to capture the variabi-

analyse the behavioural component with

lity of seabird circadian cycles.

State-Space Models (SSM; Jonsen et al. 2003)

The foraging plasticity of seabirds, regar-

on satellite tracking data as a final alterna-

ding their habitat use and prey selection, is

tive on spatio-temporal assessment tools.

sometimes underestimated. In conditions of

However, the performance of the Platform

reduced prey availability, specialised seabird

Transmitter Terminals (PTTs) was poor and

species usually modify their feeding strate-

heterogeneous among devices, probably due

gy by extending their foraging area, the time

to battery problems. The final sample size

spent at sea or reducing the time between

was very low and the time span between lo-

trips (e.g. Lewis et al. 2001; Schwemmer and

cations too long to perform State-Space Mo-

Garthe, 2008). Generalist species may chan-

dels or to test the conclusions of Chapter 3.

ge their foraging habitats or shift their diet

Despite the technical problems and limi-

(e.g. González-Solís et al. 1997; Schwemmer

ted sample size, some general conclusions

and Garthe, 2008; Navarro et al. 2010). In

can be drawn from the results of the spatial

Chapter 4, satellite tracking data shows that

and temporal analysis of the movements of

Audouin’s Gull has large foraging areas and

the Audouin’s Gull (Larus audouinii). When

makes an extensive use of both terrestrial

foraging, seabirds have to overcome the va-

and marine habitats, despite the terrestrial

riability on the distribution, abundance, mo-

movements had been scarcely described

bility and predictability of their food sources

before. Large numbers of Audouin’s Gulls

(Bell 1991). To do so, seabird species show a

were detected in the continental platform

certain degree of plasticity on their at-sea

of the Ebro Delta (Chapter 3, Table 1), but

behaviour. Among many possible behaviou-

aerial and boat surveys are limited in time

ral responses (e.g. trip duration, travelling

and extension and hence unable to capture

DISCUSSION the individual’s variations in habitat use or the temporal activity patterns during dusk,

(1) Key development regions

night and dawn. To produce a proper estimation of the

To design a rational development of offs-

potential effects of an OWF on seabirds it is

hore wind energy regarding seabirds, there

necessary to complement distribution and

is a need for a large-scale strategy defined

abundance maps based on surveys with

at least at national level. At this scale it is

individual tracking. Since it is impractical

optimal to apply the refined Wind Farm Sen-

to apply this technique to all species in a

sitivity Index described in Chapters 1 and 2.

seabird community I suggest to carry out

The main limitation is the availability of ex-

telemetry studies to high concern or highly

tensive surveys to cover the whole strategic

vulnerable species present in areas with a

area. This is usually possible through boat

potential interaction with offshore wind

surveys that in many cases are performed

farms (see also Perrow et al. 2006; Wilson

within the frame of multidisciplinary ves-

et al. 2009; Burger et al. 2011). In the study

sels that record oceanographic, large marine

case of the Ebro Delta, Audouin’s Gull was

mammals’ biogeography and other ecolo-

a good candidate because ca. 65% of the

gical information in the same surveys. This

total world population breeds in the area

constrains the availability of replications of

(Oro et al. 2009). Nevertheless, the Balea-

the surveys in different temporal scenarios,

ric Shearwater Puffinus mauretanicus would

but it is countered by the high seabird bio-

have been also highly relevant since it is

diversity detection in boat surveys. In some

Critically Endangered according to the IUCN

cases, there are aerial surveys available for

red list and the species makes an extensive

all the strategic area. When this is possible,

use of the Ebro Delta continental platform

temporal scenarios can also be included in

as foraging grounds despite not breeding in

the Strategic Environmental Assessments.

the area.

With the surveys’ results, the Vulnerability Index produces four diagnostic panels

Integrating tools in SEA and EIA

that capture the key elements to be considered in the strategic assessment: colli-

Based on the tools presented in this the-

sion and disturbance vulnerability, the ove-

sis, the reviewed literature and the acquired

rall community vulnerability and the total

experience, I present a suggested integration

abundance map. With this information it is

of the methods in the sequential decision-

possible to detect the areas with the lowest

making process for offshore wind energy

vulnerability and hence the environmenta-

development regarding seabirds. This inte-

lly optimal regions for offshore wind energy

gration, summarized in Figure 9 and explai-

development.

ned below, starts at large scale in Strategic

The focus of this thesis is on seabirds

Environmental Assessments until the final

as marine ecosystem indicators; however,

Environmental Impact Assessments at local

many other layers of information can be

scale.

overlapped to select these key development regions. Concerning seabirds, cumulative

DISCUSSION & CONCLUSIONS

fig.9 Suggested integration of the analytical tools presented in this thesis in the decision-making process for Offshore Wind Energy development.

Boat Surveys N E

W S

Aerial Surveys Individual tracking Radar studies N

W S

N E

Flight height N

W S

Flight direction

DISCUSSION impact assessments or main migratory rou-

conservation concern or highly vulnerable

tes are relevant information, but also infor-

species it might be advisable to strengthen

mation concerning large marine animals

the assessment decision with individual

(turtles, cetaceans, etc.), fish distribution,

tracking information of flagship species.

benthos, etc. Moreover, there are many socio-economical parameters to be taken into

(3) Optimal OWF location

account. All these parameters fall outside the scope of this thesis and hence they are

Within the selected region, aerial sur-

not commented in the presented decision-

veys allow temporal repetitions that provi-

making process although the framework

de an idea of the seasonal patterns, which

could be expanded for a more general ma-

we usually lack at large-scale surveys (e.g.

rine spatial planning that would include all

surveys with one coverage per year). The

of them.

aggregative patterns analysis presented in Chapter 3 takes into account this seasona-

(2) Key exclusion areas

lity and allows the mapping of the potential impacts of offshore wind farms by defining

Once a development region is selec-

transitional areas (with high collision risk)

ted, the Vulnerability Index can be applied

and feeding areas (with high disturbance

within the region. Since it is a comparative

and habitat loss risk). Associated to this po-

index, even in low vulnerability areas it will

tential risk map I suggest a classification of

be able to detect areas of higher vulnerabi-

sectors of the region according to three con-

lity. Applied at regional scale, I recommend

cern levels (L0, L1, L2 and L3).

the evaluation of the areas with higher vul-

These concern levels are directly related

nerability in order to decide if those areas

to the ecological impact of the presence of

should be excluded for OWF locations.

wind turbines in a sector, but instead of only

At this scale, the decision may be based

pointing to the potential impact, our su-

on the same surveys used at large scale. Ne-

ggestion is to link this concern levels with a

vertheless, if aerial surveys can be carried

predefined compulsory monitoring protocol,

out for the particular region it would be op-

should the OWF project be accepted (Table

timal to include the temporal scale in the

3). Each concern level should require its own

definition of key exclusion areas. Although

monitoring methods besides the recom-

not strictly necessary, if the assessed region

mended monitoring of the previous levels.

is also renowned by the presence of a high

By doing this, the selection of the optimal

Table 3 Recommended monitoring according to the concern level defined with the aggregative pattern analysis

Discussion & Conclusions location of a wind farm, which is usually

If the assessed area has both high collision

carried out by a developer company, would

and habitat loss risk and there are flagship

be based not only on a technical cost/benefit

species near, it might be advisable to carry

analysis, but also on the evaluation of the

out individual tracking information to better

increasing costs associated to the selection

define the temporal and spatial patterns of

of a less ecological favourable location. This

this species near the wind farm location.

decision system could promote the selection of low concern sectors, and would, in

(5) Monitoring

any case, provide the exploiting company with the required monitoring tools for pro-

Regardless of the concern level of the

per management. Nevertheless, even if a

selected location, when a wind farm is ope-

wind farm was planned in a high concern

rational there is a need for studies that mo-

area (L3) and accepted, this area would have

nitor the unpredictable effects of the insta-

been selected among the less vulnerable

llation in the environment. The availability

areas defined in the previous steps at large

of this monitoring is crucial to carry out any

scale.

adaptive management action once the wind

This analysis is recommended for aerial

farm is functional. The developer company

survey data sets. Boat surveys are unable to

is given, in advance, a clear picture of the

provide a â&#x20AC;&#x2DC;snapshotâ&#x20AC;&#x2122; of seabird distribution

minimum monitoring costs that are going

patterns since they require a long period

to be associated to the location of the cons-

of time to cover a large region which also

tructed OWF, being able to include them in

makes unpractical the replication of surveys

the annual budget for the wind farm main-

through the months to capture the temporal

tenance. However, the monitoring plan gi-

dimension of the aggregative patterns.

ven in Table 3 can always be adjusted or expanded according to the results of each EIA.

(4) Offshore Wind Farm EIA Once a site for an offshore wind farm is selected, it is compulsory to carry out an EIA. In this case, the potential risk map generated in Step 3 helps to identify the necessary studies to assess collision and disturbance risk, and the optimal location of sampling sites. If a project falls within an area with habitat loss risk, the project should be assessed by modelling species presence and abundance with, for instance, detailed boatbased surveys in small areas that cover the wind farm location and its surroundings. Areas with collision risk should be assessed with collision risk models that could require point transect surveys of flight height and direction, surveillance radars (vertical and/ or horizontal) or infrared camera systems.

DISCUSSION

FURTHER WORK Taylorâ&#x20AC;&#x2122;s power law applied to the Vulnerability Index: If the vulnerability index was applied to an aerial data set with enough temporal repetitions, Taylorâ&#x20AC;&#x2122;s power law methodology could be applied to a better assessment of the temporal variability of the vulnerability patterns. For instance, it could be used to define recurrent high vulnerable areas from occasional vulnerable areas. Geostatistics: The objective of this thesis was to provide new analytical tools that improved the assessment of Offshore Wind Energy development based on distribution and abundance maps. Although this thesis emphasizes the necessity to supersede plain density maps with more integrative methodologies, distribution maps are key layers of information and sometimes essential in any SEA or EIA. The most common visualization of these maps is on a grid. However, this visualization is often noisy which may obscure the underlying spatial patterns and their interpretation. To complement the tools described in this thesis the use of geostatistical interpolation and kriging should be promoted to ease the visualization of abundance maps and its ecological interpretation. Besides the simple representation of raw abundance data, distribution models are also becoming a common tool in many studies to predict density at a given point in time and space. To do so, quite often models are fit assuming independence of the data, but these models should take into account the spatial autocorrelation (SAC). When the values of a variable (e.g. seabird counts) are sampled at nearby locations, the values at each location are not completely independent from each other. In other words, if a grid cell has a thousand gulls because there is a feeding source available, neighbouring cells will also have high numbers of gulls because of the influence of that grid cell. Geostatistical interpolation methods (e.g.Variograms and Kriging) account for this spatial pattern and counteract this bias (Cressie 1993; Pebesma and Wesseling 1998). I am already working in the application of spatio-temporal variogram models (FernĂĄndez-Casal et al. 2003) to quantify not only the spatial but also the the spatio-temporal autocorrelation of survey data. The application of this technique would improve our knowledge of seabird surveys limitations and optimal survey design to reduce this bias. Other taxa and environmental assessment contexts: The Vulnerability index and the aggregative pattern analysis have been developed and assessed for seabirds. Nevertheless, these analytical tools can be adapted to any organism whenever distribution or abundance maps are available. Widening the scope of the tools would lead to a better integration of different taxa in the Environmental Assessments of offshore wind farms instead of presenting them as independent sections of SEAs or EIAs. Furthermore, these tools are not necessarily restricted to the assessment of Offshore Wind Energy and could be fitted for other environmental assessment purposes.

Box 4 Further work

100

Discussion & Conclusions

CONCLUSIONS â&#x20AC;&#x153;It always takes awfully long time to understand unbelievably simple thingsâ&#x20AC;? Joe Chung

1 2

The mathematical formulation of Garthe and HĂźppop (2004) WSI contains hidden assumptions at both species and community levels that might lead to incorrect estimates of vulnerability and a biased identification of key areas. The original WSI can be improved by (i) the explicit distinction between direct and aggravation factors (ii) the independent identification of collision and disturbance risk areas, and (iii) the incorporation of recent developments in functional diversity to produce a vulnerability map based on local relative frequencies of species. The use of our refined methodology allows a much more balanced assessment of the vulnerability.

The refined approach can be applied to estimate the vulnerability of any kind of community to any kind of impact, provided that a measure of the speciesspecific vulnerability to that impact is defined and community distribution data has been collected.

The application of the refined Vulnerability Index to the coasts of the Iberian Peninsula shows that the index calculated at different grid scales can be used to different management purposes: the definition of optimal development regions at large scale and the demarcation of areas of high vulnerability at regional scales.

CONCLUSIONS

5 6 7 8

Taylor’s power law slope can measure seabirds’ aggregative pattern in time and space that can be used to define recurrent transitional and feeding areas in areas for the potential location of an OWF. By demarcating transitional areas, potential collision risk may be assessed, while demarcating foraging areas defines areas with potential habitat loss. The resulting potential impacts map can be employed to inform on concern levels, optimal EIA design and defining the required monitoring of accepted projects.

The telemetric study of the Audouin’s Gull (Larus audouinii) highlights that individual-based studies may show individual differences in habitat use, the exploitation of alternative food sources and the potential capability of species to switch their foraging grounds, should the individuals find a barrier in their preferential habitats.

9 10 11

Seabird surveys are constrained in space and time. Any assessment study based on aerial or boat surveys should be complemented with telemetry data to account for seabirds’ plasticity and the limitations of any systematic survey method. It is advisable to identify flagship or high concern species as target species to perform telemetric studies.

The decision making process for offshore wind energy development can be improved with the implementation of a sequential action protocol that includes the presented tools of this thesis: 1st) Application of the vulnerability index at large scales to define key development regions; 2nd) Application of the vulnerability index at regional scales to define key exclusion areas; 3rd) Application of the aggregative patterns analysis (Taylor’s power Law) at regional scales to define the potential risk map and inform on optimal OWF locations; 4th) Application of the potential risk map to design and implement the EIA at local scale; 5th) Application of the potential risk map to define the compulsory monitoring program if a project is accepted.

101

102

Discussion & Conclusions

REFERENCES

Allen T, Hoekstra T (1991) Role of heterogeneity

Thompson PM (2010) Assessing underwater

in scaling of ecological systems under analy-

noise levels during pile-driving at an offshore

sis. In: Koalsa J, Pichett STA (eds) Ecological

windfarm and its potential effects on marine

heterogeneity. Springer-Verlag, New York, pp

mammals. Marine Pollution Bulletin 60:888–

47–67

897

Anderson DJ, Huyvaert KP, Wood DR (2003) At-

Barry S, Welsh AH (2002) Generalized additive

sea distribution of waved albatrosses and the

modelling and zero inflated count data. Eco-

Galapagos Marine Reserve. Biological Conser-

logical Modelling 157:179–188

vation 110:367–373 Arcos J (2001) Foraging ecology of seabirds at sea:

Bell WJ (1991) Searching behaviour. Chapman & Hall, London

Significance of commercial fisheries in the

Bellier E, Certain G, Planque B, Monestiez B, Bret-

NW Mediterranean. Thesis. Universitat de

agnolle V (2010) Modelling habitat selection

Barcelona.

<http://www.tdx.cat/bitstream/

at multiple scales with multivariate geosta-

handle/10803/775/TOL38.pdf?sequence=1>

tistics: an application to seabirds in open sea.

(accessed 24.10.12)

Oikos 119:988–999

Arcos J, Bécares J, Villero D, Brotons L, Rodríguez B,

Bigas D (2012) Llista Patró dels ocells del delta

Ruiz A (2012) Assessing the location and sta-

de l’Ebre. Categories A, B i C. Categories D i E.

bility of foraging hotspots for pelagic seabirds:

Parc Natural del Delta de l’Ebre, Departament

An approach to identify marine Important Bird

d’Agricultura, Ramaderia, Pesca, Alimentación

Areas (IBAs) in Spain. Biological Conservation

i Medi Natural, Generalitat de Catalunya

(In Press) DOI: 10.1016/j.biocon.2011.12.011

BirdLife International (2003) Windfarms and

Arcos J, Louzao M, Oro D (2008) Fisheries ecosys-

Birds: An analysis of the effects of windfarms

tem impacts and management in the Medi-

on birds, and guidance environmental assess-

terranean: seabirds point of view (J Nielsen, J

ment criteria and site selection issues. Report

Dodson, K Friedland, T Harmon, N Hughes, J

written by BirdLife International on behalf

Musick, and E Verspoor, Eds.). American Fish-

of the Bern Covention. <http://www.birdlife.

eries Society Symposium 49:1471–1479

org/eu/pdfs/BirdLife_Bern_windfarms.pdf>

Bailey H, Senior B, Simmons D, Rusin J, Picken G,

(accessed 24.10.12)

REFERENCES Blaxter JHS, Hunter JR (1982) The biology of clupeoid fishes. Advances in Marine Biology 20:1– 223

birds. Ecography 35:912–921 Camphuysen CJ, Fox AD, Leopold MF, Petersen IK (2004) Towards standardised seabirds at

Boyd IL, Wanless S, Camphuysen CJ (2006) Top

sea census techniques in connection with

predators in marine ecosystems: their role

environmental impact assessments for off-

in monitoring and management. Cambridge

shore wind farms in the U.K. COWRIE Report

University Press

No. BAM02-2002. London. <http://jncc.defra.

Bretagnolle V, Certain G, Houte S, Métais M (2004) Distribution maps and minimum abundance

gov.uk/pdf/Camphuysenetal2004_COWRIEmethods.PDF> (accessed 24.10.12)

estimates for wintering auks in the Bay of Bis-

Certain G, Bellier E, Planque B, Bretagnolle V (2007)

cay, based on aerial surveys. Aquatic Living

Characterising the temporal variability of the

Resources 17:353–360

spatial distribution of animals: an application

Broek JV den (1995) A score test for zero inflation in a Poisson distribution. Biometrics 51:738– 743 Brookes K (2009) Assessment of methods used to investigate the impact of offshore wind farms on seabirds. Thesis. University of Aberdeen

to seabirds at sea. Ecography 30:695–708 Certain G, Bretagnolle V (2008) Monitoring seabirds population in marine ecosystem: The use of strip-transect aerial surveys. Remote Sensing of Environment 112:3314–3322 Certain G, Ridoux V, Canneyt O van, Bretagnolle

Burger J, Niles LJ, Porter RR, Dey AD, Koch S, Gor-

V (2008) Delphinid spatial distribution and

don C (2012) Using a shore bird (red knot) fit-

abundance estimates over the shelf of the

ted with geolocators to evaluate a conceptual

Bay of Biscay. ICES Journal of Marine Science

risk model focusing on offshore wind. Renew-

65:656–666

able Energy 43:370–377

Chamberlain D, Rehfisch M, Fox AD, Desholm M,

Burkhard B, Opitz S, Lenhart H, Ahrendt K, Garthe

Anthony S (2006) The effect of avoidance rates

S, Mendel B, Windhorst W (2011) Ecosystem

on bird mortality predictions made by wind

based modeling and indication of ecologi-

turbine collision risk models. Ibis 148:198–202

cal integrity in the German North Sea—Case

Christel I, Certain G, Cama A, Vieites DR, Ferrer

study offshore wind parks. Ecological Indica-

X (2012) Seabird aggregative response: a new

tors 11:168–174

tool for offshore wind energy risk assessment.

Cabal J, González-Nuevo G, Nogueira E (2008) Mesozooplankton species distribution in the NW

Marine Pollution Bulletin (In press). DOI: 10.1016/j.marpolbul.2012.11.005

and N Iberian shelf during spring 2004: Rela-

Christel I, Navarro J, del Castillo M, Cama A, Fer-

tionship with frontal structures. Journal of

rer X (2012) Foraging movements of Audouin’s

Marine Systems 72:282–297

gull (Larus audouinii) in the Ebro Delta, NW

Cairns DK (1987) Seabirds as indicators of ma-

Mediterranean: A preliminary satellite-track-

rine food supplies. Biological Oceanography

ing study. Estuarine, Coastal and Shelf Science

5:261–271

96:257–261

Cama A (2010) Seabird distribution patterns and

Christensen TK, Hounisen JP, Clausager I, Peters-

ecological factors driving larid presence at the

en IK (2004) Visual and radar observations of

Ebro delta shelf (NW Mediterranean). Thesis.

birds in relation to collision risk at the Horns

Universitat de Barcelona.

Rev Offshore Wind Farm. NERI Report. <http://

Cama A, Abellana R, Christel I, Ferrer X, Vieites DR

mhk.pnnl.gov/wiki/images/8/83/Visual_and_

(2012) Living on predictability: modeling the

Radar_Observations_of_Birds.pdf>

density distribution of efficient foraging sea-

24.10.12)

(accessed

103

104

Discussion & Conclusions Cressie N (1993) Statistics for spatial data. J. Wiley, USA

Eurostat (2009) Panorama of energy. Energy statistics to support EU policies and solutions (2009

Desholm M (2006) Wind farm related mortality among avian migrants. Thesis. University of Copenhagen

Edition).

<http://epp.eurostat.ec.europa.eu/

cache/ITY_OFFPUB/KS-GH-09-001/EN/KS-GH09-001-EN.PDF> (accessed 24.10.12)

Desholm M (2009) Avian sensitivity to mortali-

Fauchald P, Erikstad KE, Skarsfjord H (2000) Scale-

ty: prioritising migratory bird species for as-

Dependent Predator-Prey Interactions: The

sessment at proposed wind farms. Journal of

Hierarchical Spatial Distribution of Seabirds

Environmental Management 90:2672–9

and Prey. Ecology 81:773

Desholm M, Fox AD, Beasley PDL, Kahlert J (2006)

Fauchald P, Erikstad KE, Systad GH (2002) Seabirds

Remote techniques for counting and estima-

and marine oil incidents: is it possible to

ting the number of bird-wind turbine colli-

predict the spatial distribution of pelagic

sions at sea: a review. Ibis 148:76–89

seabirds? Journal of Applied Ecology 39:349–

Desholm M, Kahlert J (2005) Avian collision risk at an offshore wind farm. Biology Letters 1:296– 298

360 Fernández-Casal R, Gonzalez-Manteiga W, Febrero-Bande M (2003) Flexible spatio-temporal

Drewitt A, Langston RW (2006) Assessing the impacts of wind farms on birds. Ibis 148:29–42

stationary variogram models. Statistics and Computing 13:127–136

EC (2011a) A Roadmap for moving to a competi-

Fox AD, Desholm M, Kahlert J, Christensen TK,

tive low carbon economy in 2050. COM (2011)

Petersen IK (2006) Information needs to sup-

112 of 8 March 2011. European Commission.

port environmental impact assessment of

<http://eur-lex.europa.eu/LexUriServ/Lex-

the effects of European marine offshore wind

UriServ.do?uri=CELEX:52011DC0112:EN:NOT> (accessed 24.10.12)

farms on birds. Ibis 148:129–144 Fox AD, Petersen IK (2006) Assessing the degree

EC (2011b) Energy roadmap 2050. COM (2011) 885

of habitat loss to marine birds from the de-

of 15 December 2011. European Commission.

velopment of offshore wind farms. In: Boere

DOI:

<http://ec.europa.eu/

GC, Galbaraith CA, Stroud DA (eds) Waterbirds

energy/publications/doc/2012_energy_road-

around the world. The Stationery Office, Edim-

10.2833/10759

map_2050_en.pdf> (accessed 24.10.12)

burg, UK, pp 801–804

EWEA (2011) European offshore wind projects

Garthe S, Benvenuti S, Montevecchi WA (2003)

map. 3rd Edition. November 2011. EWEA.

Temporal patterns of foraging activities of

<http://www.ewea.org/fileadmin/ewea_docu-

northern gannets, Morus bassanus, in the

ments/images/graphs_maps_tables/EuOff-

northwest Atlantic Ocean. Canadian journal

wind2011.pdf> (accessed 24.10.12)

of Zoology 81:453–461

EWEA (2012) The European offshore wind indus-

Garthe S, Hüppop O (2004) Scaling possible adverse

try key 2011 trends and statistics. European

effects of marine wind farms on seabirds: de-

Wind Energy Association. <http://www.ewea.

veloping and applying a vulnerabiblity index.

org/fileadmin/ewea_documents/documents/

Journal of Applied Ecology 41:724–734

publications/statistics/EWEA_stats_offshore_2011_01.pdf> (accessed 24.10.12)

Garthe S, Markones N, Mendel B, Sonntag N, Krause JC (2011) Protected areas for seabirds

Elliott M (2002) The role of the DPSIR approach and

in German offshore waters: Designation, ret-

conceptual models in marine environmental

rospective consideration and current perspec-

management: an example for offshore wind

tives. Biological Conservation (In Press). DOI:

power. Marine Pollution Bulletin 44:iii–vii

10.1016/j.biocon.2011.12.002

REFERENCES Gerkema T, Lam F a., Maas LRM (2004) Internal tides in the Bay of Biscay: conversion rates and seasonal effects. Deep-Sea Research II 51:2995–3008 Gill AB (2005) Offshore renewable energy: ecological implications of generating electricity in the coastal zone. Journal of Applied Ecology 42:605–615 González-Solís J, Oro D, Jover L, Ruiz X, Pedrocchi V (1997) Trophic niche width and overlap of two sympatric gulls in the southwestern mediterranean. Oecologia 112:75–80

shore_Wind_Farm_at_Rodsand.pdf> (accessed 24.10.12) Kendal WS (2004) Taylor’s ecological power law as a consequence of scale invariant exponential dispersion models. Ecological Complexity 1:193–209 Kilpatrick A, Ives A (2003) Species interactions can explain Taylor’s power law for ecological time series. Nature 422:65–68 King DA (2004) Climate Change Science : Adapt, Mitigate, or Ignore? Science 303:176–177 Kotliar N, Wiens J (1990) Multiples scales of

González-Solís J, Shaffer S (2009) Spatiotemporal

patchiness and patch structure: a hierarchi-

dynamics of seabirds in the marine environ-

cal framework for the study of heterogeneity.

ment. Marine Ecology Progress Series 391:117– 120

Oikos 59:253–260 Kunz TH, Arnett EB, Cooper BM, Erickson WP,

Henderson AR, Morgan C, Smith B, Sorensen HC,

Larkin RP, Mabee T, Morrison ML, Strickland

Barthelmie RJ, Boesmans B (2003) Offshore

MD, Szewczak JM (2007) Assessing Impacts

Wind Energy in Europe? A Review of the State-

of Wind-Energy Development on Nocturnally

of-the-Art. Wind Energy 6:35–52

Active Birds and Bats: A Guidance Document.

Henkel LA, Ford RG, Tyler WB, Davis JN (2007)

Journal of Wildlife Management 71:2449–2486

Comparison of aerial and boat-based survey

Laborde P, Urrutia J, Valencia V (1999) Seasonal

methods for Marbled murrelets Brachyramphus

variability of primary production in the Cap-

marmoratus and other marine birds. Marine

Ferret Canyon area (Bay of Biscay) during the

Ornithology 35:145–151

ECOFER cruises. Deep-Sea Research II 46:2057–

Hunt GL, Schneider DC (1987) Scale-dependent

2079

processes in the physical and biological en-

Larsen JK, Guillemette M (2007) Effects of wind

vironment of marine birds. In: Croxall JP (ed)

turbines on flight behaviour of wintering

Seabirds: Feeding Ecology and Role in Marine

common eiders: implications for habitat use

Ecosystems. Cambridge Univ Press

and collision risk. Journal of Applied Ecology

Hüppop O, Dierschke J, Exo KM, Fredrich E, Hill

44:516–522

R (2006) Bird migration studies and potential

Lavin A, Valdes L, Sanchez F, Abaunza P, Forest A,

collision risk with offshore wind turbines. Ibis

Boucher J, Lazure P, Jegou AM (2006) The Bay of

148:90–109

Biscay: the encountering of the ocean and the

Jonsen I, Myers R, Flemming J (2003) Meta-analysis

shelf. In: Robinson AP, Brink K (eds) The Sea

of animal movement using state-space mod-

Volume 14B: The Global Coastal Ocean. Inter-

els. Ecology 84:3055–3063

disciplinary Regional Studies and Syntheses.

Kahlert J, Petersen IK, Fox AD, Desholm M, Clau-

Harvard University Press, pp 935–1001

sager I (2004) Investigations of Birds During

Leinster T, Cobbold C (2012) Measuring diversity:

Construction and Operation of Nysted Off-

the importance of species similarity. Ecology

shore Wind Farm at Rødsand. Annual Status

93:477–489

Report 2003. NERI. <http://mhk.pnnl.gov/wiki/

Lewis S, Sherratt TN, Hamer KC, Wanless S (2001)

images/5/58/Investigations_of_Birds_during_

Evidence of intra-specific competition for food

Construction_and_Operation_of_Nysted_Off-

in a pelagic seabird. Nature 412:816–819

105

106

Discussion & Conclusions Linley EAS, Wilding TA, Black K, Hawkins AJS,

113:31–48

Mangi S (2007) Review of reef effects of off-

Masden E a, Fox AD, Furness RW, Bullman R, Hay-

shore wind farm structures and potential for

don DT (2010) Cumulative impact assessments

enhancement and mitigation. Report from

and bird/wind farm interactions: Developing a

PML Applications Ltd. and the Scottish Asso-

conceptual framework. Environmental Impact

ciation for Marine Science to the Department

Assessment Review 30:1–7

for Business, Enterprise and Regulatory Re-

Masden E a, Haydon DT, Fox AD, Furness RW

form (BERR). <http://webarchive.nationalar-

(2010) Barriers to movement: Modelling ener-

chives.gov.uk/+/http://www.berr.gov.uk/files/

getic costs of avoiding marine wind farms

file43528.pdf> (accessed 24.10.12)

amongst breeding seabirds. Marine Pollution

Llope M, Anadón R, Viesca L, Quevedo M, González-

Bulletin 60:1085–91

Quirós R, Stenseth NC (2006) Hydrography

Mateos M (2009) Radar technology applied to the

of the southern Bay of Biscay shelf-break

study of seabird migration across the Strait of

region: Integrating the multiscale physical

Gibraltar. Thesis. Universidad de Cadiz

variability over the period 1993–2003. Jour-

Mcsorley CA, Webb A, Dean BJ, Reid JB (2005) UK

nal of Geophysical Research 111:C09021. DOI:

inshore marine Special Protection Areas: a

10.1029/2005JC002963

methodological evaluation of site selection

Louzao M, Hyrenbach KD, Arcos J, Abelló P, de Sola

and definition of an interest feature using line

LG, Oro D (2006) Oceanographic habitat of an

transect data. JNCC Report No 344. <http://jncc.

endangered Mediterranean procellariiform:

defra.gov.uk/page-3512> (accessed 24.10.12)

Implications for marine protected areas. Eco-

Moreno R (2010) Seabirds as bioindicators of coas-

logical Applications 16:1683–1695 MARM, MITYC (2009) Estudio estratégico am-

tal ecosystem recovery following the Prestige oil spill. Thesis. Universitat de Barcelona

biental del litoral español para la instalación

Navarro J, Oro D, Bertolero A, Genovart M, Delgado

de parques eólicos marinos. <http://www.

A, Forero MG (2010) Age and sexual differences

aeeolica.org/uploads/documents/562-estudio-

in the exploitation of two anthropogenic food

estrategico-ambiental-del-litoral-espanol-

resources for an opportunistic seabird. Marine

para-la-instalacion-de-parques-eolicos-mari-

Biology 157:2453–2459

nos_mityc.pdf> (accessed 24.10.12) Madsen P, Wahlberg M, Tougaard J (2006) Wind

Nettleship DN, Duffy DC (1993) Seabird populations. Elsevier Applied Science, London

turbine underwater noise and marine mam-

Noer H, Christensen TK, Clausager I, Petersen IK

mals: implications of current knowledge and

(2000) Effects on birds of an offshore wind

data needs. Marine Ecology Progress Series

park at Horns Rev: Environmental impact

309:279–295

assessment. NERI Report. NERI report, MEE-

Maestro A, López-Martínez J, Llave E, Bohoyo F,

NERI. <http://www.hornsrev.dk/Miljoeforhold/

Acosta J, Hernández-Molina FJ, Muñoz A, Jané

miljoerapporter/Baggrundsrapport_18.pdf>

G (2012) Geomorphology of the Iberian Continental Margin. Geomorphology (In Press). DOI: 10.1016/j.geomorph.2012.08.022 Mallory ML, Gilchrist HG, Braune BM, Gaston AJ (2006) Marine birds as indicators of Arctic

(accessed 24.10.12) Observ’ER (2011) The state of renewable energies in Europe (11th EurObserv’ER Report, 2011). <http://www.eurobserv-er.org/pdf/barobilan11.pdf> (accessed 24.10.12)

marine ecosystem health: linking the North-

Oro D, Pérez-Rodríguez A, Martínez-Vilalta A, Ber-

ern Ecosystem Initiative to long-term studies.

tolero A, Vidal F (2009) Interference competi-

Environmental monitoring and assessment

tion in a threatened seabird community: A pa-

REFERENCES radox for a successful conservation. Biological Conservation 142:1830–1835

ress Series 352:199–204 Planque B, Lazure P, Jégou A (2004) Detecting hy-

Palomera I (1992) Spawning of anchovy Engraulis

drological landscapes over the Bay of Biscay

encrasicolus in the Northwestern Mediterra-

continental shelf in spring. Climate research

nean relative to hydrographic features in the region. Marine Ecology Progress Series 79:215– 223 Patterson T a, Thomas L, Wilcox C, Ovaskainen O,

28:41–52 Punt MJ, Groeneveld R a., van Ierland EC, Stel JH (2009) Spatial planning of offshore wind farms: A windfall to marine environmental

Matthiopoulos J (2008) State-space models of

protection? Ecological Economics 69:93–103

individual animal movement. Trends in Ecol-

Ribas-Ribas M, Gómez-Parra A, Forja JM (2011)

ogy & Evolution 23:87–94

Spatio-temporal variability of the dissolved

Pearce J, Ferrier S (2001) The practical value of

organic carbon and nitrogen in a coastal area

modelling relative abundance of species for

affected by river input: The north eastern

regional conservation planning: a case study.

shelf of the Gulf of Cádiz (SW Iberian Penin-

Biological Conservation 98:33–43

sula). Marine Chemistry 126:295–308

Pebesma E, Wesseling CG (1998) Gstat: A program

Rosenzweig C, Karoly D, Vicarelli M, Neofotis P, Wu

for geostatistical modelling, prediction and

Q, Casassa G, Menzel A, Root TL, Estrella N, Se-

simulation. Computers & Geosciences 24:17–

guin B, Tryjanowski P, Liu C, Rawlins S, Imeson

A (2008) Attributing physical and biological

Peliz A, Dubert J, Santos a. MP, Oliveira PB, Le Cann B (2005) Winter upper ocean circulation in the

impacts to anthropogenic climate change. Nature 453:353–357

Western Iberian Basin—Fronts, Eddies and

Salat J (1996) Review of hydrographic environmen-

Poleward Flows: an overview. Deep Sea Re-

tal factors that may influence anchovy habi-

search Part I 52:621–646

tats in northwestern Mediterranean. Scientia

Percival SM (2003) Birds and wind farms in Ire-

Marina 60:21–32

land: A review of potential issues and impact

Santos AMP, Chícharo A, Dos Santos A, Moita T,

assessment. Report comissioned by the Sus-

Oliveira PB, Peliz Á, Ré P (2007) Physical–bio-

tainable Energy Authority of Ireland. <http://

logical interactions in the life history of small

www.seai.ie/Archive1/Files_Misc/Assess-

pelagic fish in the Western Iberia Upwelling

mentMethodologyBirdsIreland.pdf> (accessed

Ecosystem. Progress In Oceanography 74:192–

24.10.12)

209

Perrow MR, Gilroy JJ, Skeate ER, Tomlinson ML

Schwemmer P, Garthe S (2008) Regular habitat

(2011) Effects of the construction of Scroby

switch as an important feeding strategy of an

Sands offshore wind farm on the prey base

opportunistic seabird species at the interface

of Little tern Sternula albifrons at its most im-

between land and sea. Estuarine, Coastal and

portant UK colony. Marine Pollution Bulletin

Shelf Science 77:12–22

62:1661–70

Skeate ER, Perrow MR, Gilroy JJ (2012) Likely ef-

Perrow MR, Skeate ER, Lines P, Brown D, Tomlinson

fects of construction of Scroby Sands offshore

ML (2006) Radio telemetry as a tool for impact

wind farm on a mixed population of harbour

assessment of wind farms: the case of Little

Phoca vitulina and grey Halichoerus grypus

Terns Sterna albifrons at Scroby Sands, Norfolk,

seals. Marine pollution bulletin (In Press) DOI:

UK. Ibis 148:57–75

10.1016/j.marpolbul.2012.01.029

Piatt JF, Sydeman WJ (2007) Seabirds as indicators

Spear L, Ainley D, Hardesty B (2004) Reducing bi-

of marine ecosystems. Marine Ecology Prog-

ases affecting at-sea surveys of seabirds: use

107

108

Discussion & Conclusions of multiple observer teams. Marine Ornithology 157:147–157 Steele JH (1989) The ocean “landscape.” Landscape Ecology 3:185–192

Turchin P (1998) Quantitative Analysis of Movement: Measuring and Modeling Population Redistribution in Animals and Plants. Sinauer Whitehouse RJS, Harris JM, Sutherland J, Rees J

Tasker ML, Jones PH, Dixon T m, Blake BF (1984)

(2010) The nature of scour development and

Counting seabirds at sea from ships: a review

scour protection at offshore windfarm foun-

of methods employed and a suggestion for a standardized approach. The Auk 101:567–577 Taylor LR (1961) Aggregation, Variance and the Mean. Nature 189:732–735 Taylor LR, Woiwod IP (1980) Temporal Stability as a Density-Dependent Species Characteristic. The Journal of Animal Ecology 49:209–224

dations. Marine Pollution Bulletin 62:73–88 Wiese FK, Montevecchi W., Davoren G., Huettmann F, Diamond A., Linke J (2001) Seabirds at risk around offshore oil platforms in the North-west Atlantic. Marine Pollution Bulletin 42:1285–1290 Wilson LJ, Mcsorley CA, Gray CM, Dean BJ, Dunn

Taylor LR, Woiwod IP (1982) Comparative Synoptic

TE, Webb A, Reid JB (2009) Radio-telemetry as

Dynamics. I. Relationships between Inter- and

a tool to define protected areas for seabirds in

Intra-specific spatial and temporal variance/

the marine environment. Biological Conserva-

mean population parameters. The Journal of Animal Ecology 51:879–906

tion 142:1808–1817 Wu J, David JL (2002) A spatially explicit hierarchi-

Taylor LR, Woiwod IP, Perry JN (1980) Variance and

cal approach to modeling complex ecological

the Large Scale Spatial Stability of Aphids,

systems: theory and applications. Ecological

Moths and Birds. The Journal of Animal Ecol-

Modelling 153:7–26

ogy 49:831–854 Tobin PC (2004) Estimation of the spatial autocorrelation function: consequences of sampling dynamic populations in space and time. Ecography 27:767–775 Tremblay Y, Bertrand S, Henry R, Kappes M, Costa D, Shaffer S (2009) Analytical approaches to investigating seabird–environment interactions: a review. Marine Ecology Progress Series 391:153–163

REFERENCES

109

Catalan summary

INTRODUCCIÓ

La creixent demanda mundial d’energia

la participació de les fonts d’energia renova-

i el canvi climàtic són dos dels grans reptes

bles s’incrementarà substancialment, fins a

d’aquest segle. En aquest escenari és neces-

assolir un 55%, 64 o 97% del consum d’ener-

sari trobar un equilibri entre les polítiques

gia final bruta el 2050, segons l’escenari (CE

de canvi climàtic i la competitivitat, per tal

2011b). Una de les mesures polítiques per

que la reducció de les emissions de carboni

aconseguir aquest objectiu és la Directiva

sigui econòmicament viable. En aquest con-

sobre energies renovables, que fixa com a

text, la Comissió Europea va definir el «full

objectiu que el 20% del consum energètic

de ruta Energia 2050», que explora les possi-

provingui de fonts renovables el 2020.

bilitats d’aconseguir una economia baixa en

A Europa, les energies renovables repre-

carboni alhora que s’assegura un subminis-

senten el 18% de tota la producció energè-

trament d’energia competitiva, sostenible i

tica (Eurostat 2009; Fig 1a, pàg.4). L’energia

segura (CE 2011). La Unió Europea s’ha com-

hidroelèctrica és la font principal de produc-

promès per al 2050 a reduir les emissions de

ció d’energies renovables (54,5%), seguida de

gasos d’efecte hivernacle fins a un 80-95%

l’energia eòlica (22,5%) (Observ’ER 2011, Fig

per sota dels nivells de 1990 (CE 2011b). És

1b, pàg.4). L’any 2050 s’espera que l’energia

impossible predir els canvis que es produi-

eòlica proporcioni més electricitat que qual-

ran a Europa a llarg termini, però alguns dels

sevol altra tecnologia (CE 2011b) i per tant la

possibles escenaris són: (i) un sistema d’alta

contribució potencial del medi marí per al

eficiència energètica, (ii) un sistema amb

desenvolupament de l’energia eòlica ha re-

una oferta diversificada de tecnologia, com

but una gran atenció a les últimes dècades.

ara la captura de carboni i les instal·lacions d’emmagatzematge o l’energia nuclear, i

Energia eòlica marina

(iii) un fort suport a les fonts d’energia renovables. Tots els pronòstics per reduir les

El primer parc eòlic marí es va instal·lar

emissions de carboni a Europa mostren que

a Dinamarca el 1991. Des de llavors, el sec-

l’electricitat haurà de tenir un paper més

tor s’ha expandit ràpidament i en particular

important que els combustibles fòssils i que

al nord d’Europa (Fig 2, pàg.5). Avui en dia,

113

114

Quadre

Catalan Summary

Estat de desenvolupament de l’energia eòlica marina a Espanya

Europa és el líder mundial en energia eòlica

turbines eòliques al mar, encara que només

marina, amb un total de 1.371 turbines a alta

la Xina té tres parcs eòlics marins operatius.

mar distribuïdes en 53 parcs eòlics en 10 pa-

La majoria de les turbines instal·lades

ïsos a finals del 2011 (EWEA 2012). El Regne

tenen fonaments al fons marí. S’estan de-

Unit és el país amb la major potència instal·

senvolupant models flotants, i Noruega i

lada d’energia eòlica marina, seguit per Di-

Portugal són els primers països que tenen

namarca, Països Baixos i Alemanya (Taula 1,

instal·lada una turbina flotant a gran esca-

pàg.5). L’interès per l’energia eòlica marina

la. A mesura que la tecnologia es desenvo-

s’està estenent més enllà d’Europa. A Xina,

lupi, s’espera que els parcs eòlics marins

Japó, Corea del Sud, EUA i Israel hi ha empre-

creixin en grandària i que es despleguin més

ses que treballen en el desenvolupament de

lluny de la costa i en aigües més profundes,

ENERGIA EÒLICA MARINA A ESPANYA Actualment Espanya no disposa de parcs eòlics marins operatius. Des del començament de l’expansió de l’energia eòlica en alta mar a l’Europa Septentrional, diferents empreses han mostrat el seu interès en la construcció de parcs eòlics marins a les costes espanyoles. Malgrat les iniciatives del sector privat per promoure el seu desenvolupament, el govern espanyol va establir l’any 2007 el procediment administratiu obligatori per aconseguir la concessió per construir un parc eòlic marí a les costes espanyoles (Reial Decret 1028/2007). El procediment establia com a necessari una Avaluació Ambiental Estratègica (AAE) de la costa espanyola. Aquest estudi va ser publicat el 2009 (MARM i el MITYC, 2009) i va incloure el mapa de zonificació definitiva per a les àrees de desenvolupament de l’energia eòlica marina. Aquest mapa divideix les costes espanyoles en 72 àrees (definits per quadrats d’un grau decimal). Les primeres 24 milles nàutiques de cada àrea es van avaluar d’acord a múltiples criteris per a classificar les zones en tres categories: aptes (en verd), àrees adients amb restriccions (en groc) i les zones d’exclusió (en vermell).

El procés de concessió administrativa llarg i complex ha patit diversos retards. Avui en dia a Espanya no hi ha un nombre oficial de parcs eòlics previstos.

INTRODUCCIÓ sobretot si se’n demostra la viabilitat econò-

durant la construcció i operació OWF (Whi-

mica dels models flotants. Els projectes ac-

tehouse et al. 2010; Burkhard et al.,2011) i els

tuals en construcció es troben a una fondà-

efectes sobre les larves de peixos (Perrow et

ria mitjana de 25 metres i a una distància de

al. 2011). A més, es desconeixen els efectes

la costa de 33 km (EWEA 2012), ja que molts

del soroll submarí sobre la vida dels peixos,

dels OWF (Parcs Eòlics Marins; sigles de

tortugues i mamífers marins (Madsen et al.

Offshore Wind Farms) s’han construït al Mar

2006; Bailey et al. 2010), els efectes a nivell

del Nord, que té una gran part a la platafor-

de la població de les col·lisions d’aus amb

ma continental europea (Fig 3, pàg.5). Això

les turbines (Fox et al. 2006; Desholm 2009)

proporciona regions planes i superfícies re-

i els efectes de la pertorbació sobre diversos

lativament grans adequades per al desenvo-

grups animals (Drewitt & Langston 2006;

lupament d’OWF (Henderson et al. 2003). A

Masden, Haydon, et al. 2010).

diferència del nord d’Europa, la costa occidental de França, de la Península Ibèrica i la

L’avaluació ambiental

Mediterrània segueixen sent un desafiament per al desenvolupament de l’energia eòlica

La Unió Europea compta amb un marc

marina. Encara que hi ha projectes previstos

normatiu (Directiva 2001/42/CE) per es-

per a aquestes zones, les turbines disponi-

tandarditzar l’avaluació i el seguiment de

bles i els mètodes de fonamentació requeri-

les activitats humanes en els ecosistemes

rien la construcció de parcs eòlics molt més

i garantir un desenvolupament racional

a prop de la línia de costa, amb un conse-

d’aquestes activitats, incloent considera-

güent increment dels conflictes per trobar

cions ambientals. A gran escala, els països

llocs òptims, ja sigui per l’acceptació social,

han de desenvolupar una Avaluació Ambi-

els impactes ambientals, els conflictes de in-

ental Estratègica (AAE) per planificar la seva

terès nacional o la planificació espacial ma-

xarxa de parcs eòlics marins i minimitzar

rina. Tots aquests factors, juntament amb

el seu impacte ecològic sobre el medi am-

la manca de finançament, estan frenant el

bient costaner. A nivell local, cada projecte

desenvolupament de l’energia eòlica en alta

de parc eòlic requereix una Estudi d’Impacte

mar a l’Europa occidental i meridional.

Ambiental (EIA) dels possibles impactes ne-

L’energia eòlica marina, de fet, no està

gatius en el medi marí del projecte proposat.

exempta de conflictes. A escala global, el

Durant molts anys, l’única informació

canvi cap a les energies renovables és ac-

disponible sobre els parcs eòlics en alta mar

ceptat àmpliament com un pas necessari

eren informes centrats en la forma de rea-

per mitigar els efectes del canvi climàtic

litzar els EIA de projectes particulars. L’ex-

antropogènic (King 2004; Rosenzweig et al.

periència danesa amb els primers parcs eò-

2008). A escala local, però, cal considerar

lics i la seva metodologia de mostreig aeri

acuradament els impactes ambientals del

s’ha convertit en un referent per a molts

desenvolupament de l’energia eòlica (Gill

EIA (Noer et al. 2000). Més tard, el COWRIE

2005). En el camp de la gestió del medi marí,

(Collaborative Offshore Wind Research Into

hi ha una creixent preocupació sobre el des-

the Environment) del Regne Unit va encar-

envolupament de l’energia eòlica al mar i

regar un informe per estandarditzar les tèc-

els seus possibles impactes en l’ecosistema

niques de censos d’aus marines per l’EIA de

marí. Alguns dels aspectes que s’estan estu-

parcs eòlics a alta mar (Camphuysen et al.

diant són l’alteració del fons i fauna marines

2004). En els últims anys, i com que el sec-

115

116

Catalan Summary tor ha crescut, s’han publicat reportatges i

ara la Directiva Aus i la Directiva Hàbitats a

treballs d’investigació sobre l’avaluació de

Europa) i són espècies emblemàtiques per al

la interacció amb el medi ambient de parcs

públic (Fox et al. 2006). Per aquests motius

eòlics particulars (p. ex. Desholm & Kahlert

abunden els estudis a llarg termini de la dis-

2005; Perrow et al. 2011; Skeate et al. 2012),

tribució en el mar i les tendències poblacio-

i també revisions i treballs generals relacio-

nals dels ocells marins.

nades amb l’AAE (Elliott 2002; Fox et al. 2006; Punt et al. 2009; Masden, Fox, et al. 2010).

Per tot això, la distribució i abundància de les aus marines esdevé informació clau per donar suport a les àrees marines protegides

Les aus marines com indicadors

(Garthe et al. 2011; Arcos et al. 2012), per aplicar les mesures de gestió pesquera (Boyd et

Els ecosistemes marins tenen nivells de

al. 2006), per monitoritzar l’impacte de les

biodiversitat elevats i a la vegada són molt

plataformes de petroli i gas al mar (Wiese

complexos ecològicament. Mentre que els

et al. 2001), o per avaluar l’impacte dels de-

estudis ecològics es centren en aquesta

sastres ambientals com ara vessaments de

complexitat, l’ecologia aplicada requereix

petroli (Bretagnolle et al. 2004; Moreno 2010)

de mètodes que sintetitzen aquesta comple-

. Per tant, les aus marines són indicadors

xitat per tal de prendre mesures que poden

adequats del medi marí, i s’han convertit

tenir conseqüències econòmiques (Piatt &

en una de les pedres angulars del procés de

Sydeman 2007). Aquest és el cas de la uti-

presa de decisions per a la selecció d’àrees

lització d’espècies indicadores per simpli-

òptimes per al desenvolupament nacional

ficar els processos de supervisió i de gestió

d’energia eòlica marina i l’avaluació d’im-

per als EIA i AAE. Els principals depredadors

pacte dels projectes concrets de parcs eòlics

marins són un component clau de la gestió

marins.

dels ecosistemes marins (Boyd et al. 2006.), i dins dels principals depredadors, les aus

Els impactes potencials sobre les aus marines

marines han esdevingut els indicadors generalitzats per avaluar els efectes potencials

A l’hora de seleccionar les àrees de des-

de les activitats humanes al mar, així com la

envolupament, o quan la ubicació d’un pro-

salut de l’ecosistema (Cairns 1987; Nettles-

jecte es resol, podem diferenciar l’efecte de

hip & Duffy 1993; Mallory et al. 2006).

OWF en dos tipus d’aus: i) espècies migratò-

Les aus marines ofereixen molts avan-

ries que poden ensopegar amb els parcs eò-

tatges en comparació amb altres espèci-

lics en les seves rutes migratòries i ii) espèci-

es. Considerant un entorn on la majoria de

es que tenen els quarters de cria i hivernada

les espècies estan sota l’aigua, les aus ma-

prop d’un parc eòlic marí. Tots dos tipus

rines són animals visibles que es poden es-

d’aus són susceptibles a múltiples impactes

tudiar fàcilment. A més, com que algunes

antropogènics (Anderson et al. 2003; Hüppop

espècies tenen facilitat de captura, es poden

et al. 2006; Louzao et al. 2006). Els impactes

realitzar seguiments individuals i estudis

potencials dels parcs eòlics marins sobre les

demogràfics (Piatt & Sydeman 2007). D’altra

comunitats d’aus marines es poden classifi-

banda, la majoria de les aus marines tenen

car en tres tipus (Fig 4, pàg.9), (i) mortalitat

determinats marcs legals de protecció (com

directa per col·lisió, (ii) modificació del seu

INTRODUCCIÓ hàbitat físic i (iii) efectes de pertorbació i de

turbines pot afectar la hidrologia del lloc i

barrera.

impactar grans àrees (Percival 2003). Hi ha incertesa sobre la magnitud d’aquests can-

Risc de col·lisió

vis, però el dany pot ser significatiu, especialment a àrees d’alimentació, com ara bancs

Els ocells que volen per la zona del parc eòlic tenen clarament un cert risc de col·

de sorra en aigües poc profundes (Drewitt & Langston 2006).

lisió amb les aspes i l’estructura estacionà-

Les bases de les turbines tendeixen a te-

ria, o de ser atrapats i ferits en els vòrtex de

nir un «efecte escull» que augmenta la bi-

pressió creats per les pales del rotor (Fox et

odiversitat mitjançant la creació d’hàbitat

al. 2006). El risc de col·lisió depèn d’una sè-

(Linley et al. 2007), però això pot influir en

rie de factors relacionats amb les espècies

les comunitats de flora i fauna de manera

d’aus (maniobrabilitat, envergadura, etc.),

complexa generant efectes tant positius

comportament (per exemple, activitat noc-

com negatius, depenent del lloc i de l’es-

turna), la presència en grans estols i les

pècie (Perrow et al. 2011). Les aus marines

condicions meteorològiques que redueixen

també poden veure’s afectades de manera

la visibilitat. La mortalitat per col·lisió és el

diferent pels canvis en l’hàbitat. Mentre que

perill més important, ja que la mortalitat

algunes espècies especialistes poden perdre

directa pot tenir potencialment conseqüèn-

importants fonts d’aliment, altres espècies

cies ràpides en els nivells de població. No

oportunistes (com ara les gavines) poden

obstant això, encara hi ha poca informació

augmentar la seva presència a la zona per

sobre el nombre real de col·lisions d’aus

explotar la nova font d’aliment. D’altres aus

amb parcs eòlics en alta mar, en gran part

marines (com succeeix amb els corbmarins)

com a conseqüència de les dificultats tèc-

poden veure’s atretes per les plataformes

niques per a detectar les col·lisions al mar

de manteniment de turbines que utilitzen

(Drewitt & Langston 2006).

com estructures de descans (Kahlert et al. 2004). No obstant això, aquest guany d’hà-

Modificació de l’hàbitat

bitat podria ser contrarestat per un risc de col·lisió superior.

Aquest impacte comprèn la pèrdua d’hàbitat resultat de la presència de les ba-

Pertorbació

ses de turbina, els cables de connexió a la xarxa i qualsevol altra construcció associa-

La presència de les turbines, així com

da. La magnitud de la pèrdua d’hàbitat no

els moviments dels vaixells i de les perso-

es considera generalment com una de les

nes relacionades amb la construcció i man-

principals preocupacions sempre que no es

teniment del lloc, pot dissuadir algunes

produeixi en zones d’alta biodiversitat o im-

aus marines de l’ús de zones del parc eòlic

portància ecològica (BirdLife International

i els seus voltants. L’escala dels efectes de

2003). No obstant això, també pot haver-hi

pertorbació varia molt en funció d’una àm-

una pèrdua d’hàbitat indirecta a causa dels

plia gamma de factors (Drewitt & Langston

fonaments de la turbina sobre el fons del

2006), com ara el disseny de la matriu de la

mar, o pel canvi en l’ús de l’hàbitat que en

turbina, la distància entre les turbines, els

fan els humans. Per exemple, l’activitat de

patrons d’activitat (nocturna o diürna) de

la construcció i la distribució espacial de les

les aus marines (Desholm & Kahlert 2005)

117

118

Catalan Summary o les condicions climàtiques. Les respostes

densitat per avaluar la pèrdua de hàbitats

comportamentals als parcs eòlics no només

d’alimentació i la modificació de l’hàbitat fí-

poden variar entre les espècies, sinó tam-

sic (Camphuysen et al. 2004; Fox & Petersen

bé entre individus de la mateixa espècie en

2006).

funció de factors com ara l’etapa del cicle de

Per a avaluar el risc de col·lisió la tecno-

vida (hivernada, muda i de cria), la mida o la

logia de radar és una eina poderosa, ja que

tendència a l’habituació.

permet millorar el nostre coneixement sobre

Encara que es produeixi pertorbació i

patrons en l’espai i el temps d’alguns grups

desplaçament, el seu efecte pot ser intrans-

d’aus marines. La recopilació de dades de

cendent si abunden els hàbitats alterna-

radars i l’anàlisi dels resultats requereixen

tius. No obstant això, els parcs eòlics marins

estudis integrals que ja s’han abordat (Des-

situats a les rutes migratòries o en trajectò-

holm 2006; Brookes 2009; Mateos 2009). Per

ries de vol locals podrien alterar els movi-

contra, l’ús de mapes de densitat s’ha que-

ments de les aus i augmentar la seva des-

dat enrere en la integració de la dimensió

pesa d’energia (Masden, Haydon, et al. 2010).

espai-temporal dels patrons de les aus ma-

Aquest tipus de trastorn s’anomena “efecte

rines, tot i que els mapes de distribució de

barrera”. De fet, les observacions en parcs

les aus marines tenen un paper prominent

eòlics operatius mostren que molts ocells

en la majoria d’EIA i les AAE. Respecte a la

decideixen volar fora del parc eòlic en lloc

distribució d’aus marines i la seva abun-

de volar entre les turbines (Desholm & Kah-

dància, generalment es representa com a

lert 2005; Larsen & Guillemette 2007). Desa-

simples localitzacions o quadrícules de den-

fortunadament, hi ha una manca d’informa-

sitat. Després d’una revisió de més de 200

ció completa abans i després de l’impacte en

estudis publicats, Tremblay et al. (2009) van

molts parcs eòlics operatius per quantificar

assenyalar que «la visualització senzilla de

adequadament l’efecte barrera i els efectes

les dades de distribució ha estat molt més

de les pertorbacions en comparació amb

freqüent que els índexs quantitatius». De

el comportament bàsic de les aus marines

fet, pocs estudis han tractat d’abordar els

(Drewitt & Langston 2006).

mètodes analítics i sintètics per extreure les decisions adequades estratègiques (AAE) o

Noves línies de recerca

locals (EIA) dels nivells de les dades de distribució d’aus marines. Aquesta tesi pretén

Les directrius acordades internacional-

contribuir a omplir aquest buit en l’enfoca-

ment recomanen l’avaluació del risc de

ment metodològic per a l’ús de les dades de

col·lisió amb estudis de radar en àrees for-

distribució d’aus marines en alta mar per les

tament migratòries (Desholm et al. 2006;

Avaluacions d’Energia Eòlica.

Fox et al. 2006; Kunz et al. 2007) i mapes de

OBJECTIUS

Objectiu principal L’objectiu principal d’aquesta tesi és aprofundir en les eines analítiques en l’espai i el temps per a l’avaluació ambiental de l’energia eòlica marina a fi de proporcionar als professionals les directrius sobre com i quan aplicar-les.

Objectius específics Per aconseguir aquest objectiu, la present tesi s’ha estructurat en quatre capítols i una anàlisi global que aborden els següents objectius específics:

Dissenyar i posar a prova un índex de vulnerabilitat per avaluar els efectes potencials de desenvolupament d’energia eòlica marina a les aus marines. (Capítols 1 i 2)

Desenvolupar una eina per integrar la variabilitat espaial i temporal de l’abundància

Investigar les discrepàncies entre els mapes de distribució i abundància i el segui-

d’aus marines al mar per quantificar els impactes potencials dels parcs eòlics marins sobre les aus marines. (Capítol 3) ment basat en individus d’una espècie emblemàtica. (Capítol 4) Proporcionar directrius pràctiques sobre la manera d’integrar les eines analítiques presentades en el disseny d’AAE i EIA. (Discussió)

119

120

Catalan Summary

PLANTEJAMENT METODOLÒGIC

Censos d’aus marines

vol. No obstant això, aquest mètode té dos

De les tècniques de cens existents, els

inconvenients principals. En primer lloc, els

millors mètodes disponibles per a la ob-

vaixells al mar, encara que no proporcionin

tenció de la distribució i abundància d’aus

aliments, tenen un efecte d’atracció sobre

al mar són els censos des d’avioneta i des

les aus que modifica en algun grau la dis-

d’embarcació. Els censos en vaixell han

tribució original de les aus marines (Spear

estat àmpliament utilitzats seguint una

et al. 2004). En segon lloc, aquest mètode

metodologia estandarditzada (Tasker et al.

requereix un temps més llarg al mar per

1984), amb adaptacions d’acord a cada pro-

cobrir grans àrees. Els estudis aeris, d’altra

jecte en particular. Els censos aeris d’aus

banda, són particularment eficaços en una

marines al mar han tingut una ràpida ex-

cobertura simultània de grans àrees que

pansió en l’última dècada. Han estat forta-

proporcionen una instantània de distribu-

ment influenciats per l’experiència danesa

ció i abundància (Camphuysen et al. 2004)

en relació amb l’avaluació de l’impacte am-

amb un mínim d’atracció o repulsió (Certain

biental dels parcs eòlics a alta mar (Cam-

& Bretagnolle 2008). D’altra banda, els reco-

phuysen et al. 2004). Fins ara, el mètode ex-

neixements aeris permeten estudiar zones

posat en els seus informes (p. ex. Noer et al.

llunyanes de difícil accés (com ara, zones

2000) s’ha convertit en un referent.

poc profundes o bancs de sorra) en intervals

L’elecció entre un o altre mètode depèn

de temps curts i de manera poc costosa

dels objectius específics de la investigació,

(Camphuysen et al. 2004; Garthe et al.

ja que cada mètode té els seus avantatges i

2011). Això és possible gràcies a la velocitat

desavantatges (Camphuysen et al. 2004 per a

dels avions, però aquesta velocitat és també

una revisió completa). Els censos en vaixell

el principal inconvenient del mètode. Els es-

són especialment adequats per fer recompt-

tudis aeris es duen a terme a una velocitat

es exhaustius, el que permet una millor

mínima que garanteix la seguretat de vol i

identificació de les espècies amb el temps

proporciona suficient temps d’observació

suficient per recollir informació addicional

(en general 185 km/h). A aquesta velocitat,

com l’edat, el comportament o l’alçada de

el temps d’observació és curt i genera prob-

PLANTEJAMENT METODOLÒGIC lemes d’identificació d’algunes espècies i

l’energia eòlica en alta mar amb les aus ma-

redueix la capacitat de detecció d’espècies

rines, la telemetria esdevé un mètode eficaç

rares i petites que són difícils de detectar

per integrar la dimensió espacial i temporal

des d’avioneta (Camphuysen et al. 2004;

dels patrons de distribució de les aus ma-

Henkel et al. 2007). A més, la informació ad-

rines. No obstant això, hi ha alguns incon-

dicional no sempre és fàcil de recollir i no es

venients per a aquesta metodologia. Alguns

pot calcular l’alçada de vol.

d’aquests dispositius tenen alts costos; la

En aquesta tesi s’han utilitzat censos des

mida mostral és petita, i per tant els anàlisis

de vaixell i reconeixements aeris com a font

són molt costosos en temps. A més només

de dades de distribució d’aus marines. Les

un nombre limitat d’espècies d’aus marines

aus marines presenten patrons dinàmics de-

poden ser capturades per fixar els mètodes

pendents de l’escala de distribució, per això

de marcatge (Perrow et al. 2006.). Finalment

calen conjunts de dades que permetin fer

també hi ha problemes legals i de conser-

front a aquesta variabilitat i que es puguin

vació a l’hora de capturar espècies prote-

repetir fàcilment en condicions similars. Els

gides. Aquest enfocament, que s’ha utilitzat

estudis aeris permeten obtenir dades d’una

en el quart capítol de la tesi, proporciona

àrea en particular diverses vegades dins

estudis de comportament a escala fina i

d’un any (capítol primer i tercer). Els censos

resulta especialment útil si s’utilitza junta-

des de vaixells requereixen més temps però

ment amb mètodes com censos aeris i des

maximitzen la riquesa d’espècies detecta-

de vaixell (Tremblay et al. 2009).

des (nombre d’espècies o tàxons identificats en cada enquesta) (Henkel et al. 2007), una

Les àrees d’estudi

característica clau per capturar els patrons detallats de biodiversitat. En el segon capí-

Aquesta tesi doctoral aborda la qüestió

tol, l’àrea d’estudi abasta les costes de la

del desenvolupament d’energia eòlica ma-

Península Ibèrica. Censos repetits simultà-

rina i les interaccions amb les aus, des d’una

niament i sistemàticament eren econòmica-

perspectiva metodològica, sense centrar-se

ment inviables. Per tant, la maximització de

en una àrea particular. No obstant això, per

la detecció de les espècies a través de censos

tal de presentar un instrument d’anàlisi,

amb vaixell era particularment important.

entendre l’eina, la seva aplicació i aplica-

Tots dos tipus de censos en una àrea

bilitat per a la presa de decisions i la gestió,

determinada permeten veure si les aus

les dades reals són preferibles als conjunts

l’utilitzen. El que sembla més intuïtiu però,

de dades simulades. Les tres àrees d’estudi

és controlar les aus marines per estudiar

pertanyen a les aigües franceses, portugue-

com estan explotant una àrea (Perrow et al.

ses i espanyoles i tenen un gran potencial

2006). Per això s’han utilitzat mitjans elec-

per al futur desenvolupament d’energia

trònics de seguiment, com ara transmissors

eòlica marina. A excepció de la turbina ex-

de localització per satèl·lit, receptors GPS o

perimental flotant a Portugal, fins ara no hi

ràdio telemetria. Des de principis de 1990,

ha cap OWF construït a les àrees d’estudi, el

ha augmentat molt l’ús de la telemetria a

que les converteix en exemples rellevants de

causa dels avenços en la miniaturització

com aplicar les eines d’anàlisi per a la futura

dels dispositius electrònics (Tremblay et al.

presa de decisions. A continuació s’exposa

2009).

una breu descripció de les tres àrees.

l’avaluació

interacció

121

122

Catalan Summary Golf de Biscaia

voltada per l’Oceà Atlàntic i el Mar Mediterrani, i la diversitat geomorfològica i ocean-

El Golf de Biscaia és un golf de l’Oceà

ogràfica dels seus marges continentals, té

Atlàntic que s’estén entre el cap Ortegal a

implicacions significatives sobre el seu clima

Galícia, Espanya (43.77 º N, 7.89 º W) i l’illa

i la circulació de masses d’aigua (Maestro et

d’Ouessant, a Bretanya, França (48.43 º N,

al. 2012 per a una revisió detallada). El marge

18/05 º W). Dins d’aquesta zona, una regió de

continental de la Península Ibèrica té divers-

100.000 km 2 (Fig.8a, pàg 18) es va cobrir amb

es regions ben diferenciades: i) la regió est

5000 km lineals de transsectes aeris mensu-

de l’Atlàntic Nord i l’aflorament ibèric cap al

als des d’octubre de 2001 a març de 2002 i

pol actual, que tenen una forta influència en

amb 4000 km lineals de transsectes des de

els marges continentals portuguès, gallec i

vaixell a la primavera del 2003 al 2006.

del Golf de Biscaia (Peliz et al. 2005; Llope et

L’àrea d’estudi cobreix la plataforma

al. 2006); ii) la sortida de l’aigua del Mediter-

continental francesa del Golf de Biscaia en-

rani que flueix des de l’estret de Gibraltar al

tre Penmarch al nord (47.75 º N, 28/04 º W)

llarg del talús continental del Golf de Cadis

i Baiona al sud (43.497 º N, 1.64 º W). Les

(Ribas-Ribas et al. 2011); iii) els corrents de

àrees de descans costaneres i de la plata-

l’Atlàntic que afecten el Mar d’Alborán , i iv)

forma són els sistemes més productius de

altres masses d’aigua mediterrànies que in-

la regió (Certain et al. 2008). Les desembo-

flueixen en els marges continentals valèn-

cadures dels rius Loira i Gironde són una

cia, català i balear (Salat 1996). Aquesta con-

font d’aigua rica en nutrients provinents de

figuració oceanogràfica afecta la composició

l’aigua dolça (Planque et al. 2004) i la vora de

i l’estructura del plàncton i de tots els com-

la plataforma és una àrea d’elevada produc-

ponents de la cadena alimentària (Santos et

ció primària, on les aigües profundes més

al. 2007; Cabal et al. 2008) fins als nivells tròf-

fredes arriben a la capa eufòtica, a causa de

ics més alts i, per tant, les aus marines. De

les marees internes i les onades (Gerkema et

fet, la Península Ibèrica acull la major diver-

al. 2004), especialment a la zona sud, carac-

sitat d’aus marines d’Europa. La comunitat

teritzada per un profund canó, el Cap Ferret

d’aus marines en aquesta zona manté fins a

(Laborde et al. 1999).

39 espècies habituals (Taula 2, pàg 20), a més

La comunitat d’aus marines d’aquesta

de diverses espècies rares.

àrea es pot agrupar en vuit famílies i un total de 30 espècies (Taula 2, pàg 20).

Costes de la Península Ibèrica

Delta de l’Ebre La tercera àrea, d’escala més local, es troba al voltant del delta de l’Ebre (40,7 º N, 0.75

Aquesta àrea, d’aproximadament 230.000

ºE) (Fig.8a, pàg 18). L’àrea d’estudi cobreix

km , cobreix la plataforma continental espa-

1.435 km 2 de la plataforma continental des

nyola i portuguesa i s’estén sobre 7.800 km

del port de l’Ametlla de Mar (24 km al nord;

de costa (Fig.8b, pàg 18). Els censos es van

40,86 º N, 0,8 º E) fins a Peníscola (51 km al

dur a terme en vaixell per l’SPEA (Societat

sud; 40,35 º N, 0,4 º E). D’aquest àrea, que pot

Portuguesa per a l’Estudi de les Aus) i SEO

ser coberta en un sol dia amb avioneta, se’n

/ Birdlife (Societat Espanyola d’Ornitologia)

van fer reconeixements aeris un cop al mes,

en diferents etapes entre 1999 i 2011.

des de l’abril de 2005 a març del 2006.

La ubicació de la Península Ibèrica, en-

Aquesta zona compta amb un aflora-

PLANTEJAMENT METODOLÒGIC ment permanent gràcies a la combinació

bució espacial i temporal dels animals és

de la influència del front liguro-provençal-

el resultat de la combinació de processos

català, la sobtada ampliació de la plata-

extrínsecs, relacionats amb la influència

forma continental i la font de nutrients del

dels factors ambientals biòtics i abiòtics,

riu Ebre (Palomera 1992; Arcos 2001). L’alta

i processos intrínsecs, relacionats amb la

productivitat de la zona acull una important

dinàmica de la població i de les interaccions

flota pesquera, font d’alimentació clau per a

intraespecífiques (Bellier et al. 2010). A mes,

la cria i hivernada d’aus marines al Delta de

les distribucions també depenen de l’escala

l’Ebre (Arcos 2001; Arcos et al. 2008). D’altra

d’estudi.

banda, el delta de l’Ebre és una zona humi-

Així,

sistema

d’agrupacions

da d’importància internacional inclosa en

jeràrquiques dinàmiques (Kotliar & Wiens

el Conveni de Ramsar des de 1993. Amb 320

1990; Allen i Hoekstra 1991; Wu & David

km , el Delta és la segona zona humida més

2002), els patrons a gran escala seran més

important de la Mediterrània occidental

estables i predictibles a causa d’una alta

després de la Camarga a França i la segona

correlació amb les variables ambientals que

més important de la Península Ibèrica

defineixen els hàbitats potencials (Hunt &

després de Doñana. Els arrossars, llacunes,

Schneider 1987; Bellier et al. 2010). En canvi,

salines i platges del delta de l’Ebre ofereixen

a escales espacials més petites, els patrons

una varietat d’hàbitats de cria i hivernada

són menys predictibles ja que depenen de

de les aus, però també parada una escala

combinacions particulars de variables cir-

d’aturada per a un gran nombre d’aus mi-

cumstancials que creen un hàbitat temporal

gratòries. En global, s’hi poden trobar al

preferencial dins l’hàbitat potencial (Bellier

voltant de 400 espècies d’aus (Bigas 2012),

et al. 2010).

18 de les quals es van poder detectar al mar des d’avioneta (Taula 2, pàg 20).

Per traduir aquests conceptes teòrics de l’ecologia aplicada calen els instruments que permetin avaluar de forma òptima les

Eines de modelització

interaccions entre aus marines i els parcs eòlics marins. Aquests hauran de tenir en

L’eficàcia de la utilització de dades so-

compte l’efecte diferencial de les escales

bre la distribució de les aus marines al mar

espacials i temporals. En avaluacions a gran

com a eina per a la conservació i valoració

escala, es pot considerar que els patrons

del medi ambient depèn de si les dades es-

de distribució observats són estables en el

pacials a partir dels censos d’aus marines

temps i representen els hàbitats potencials.

representen un patró general o només una

Per tant, permeten delimitar de forma òpti-

’instantània’ puntual d’un sistema altament

ma les àrees clau de protecció (p. ex. Impor-

dinàmic (Fauchald et al. 2002).

tant Bird Areas; IBAs) i les àrees clau per al

Malgrat la seva homogeneïtat superfi-

desenvolupament d’energia eòlica marina.

cial, el mar és un entorn heterogeni a causa

En les avaluacions a escala regionals o lo-

de les seves múltiples característiques hi-

cal, cal avaluar l’agrupació observada d’aus

drològiques, hidrogràfiques i la distribu-

marines en la seva variabilitat temporal i es-

ció desigual de la seva biota (Steele 1989;

pacial per quantificar (amb probabilitats) el

González-Solís & Shaffer 2009). La distri-

risc d’exposició a a l’OWF.

123

124

Catalan Summary desenvolupar un nou índex, en aquesta

Gran escala: Índex de Vulnerabilitat

tesi s’analitza el mètode en profunditat i es suggereix un refinament de la seva

L’Avaluació Ambiental Estratègica integ-

formulació matemàtica (capítol 1). A més, es

ra dades a escales molt grans, de forma que

formulen recomanacions sobre l’aplicació

podem assumir que l’escala temporal no és

òptima de l’índex per a la seva utilització en

una prioritat i podem combinar les dades de

qualsevol avaluació ambiental estratègica

diferents anys o períodes. Les distribucions

(capítol 2).

d’aus marines poden tenir patrons diferents depenent de l’etapa del cicle de vida (hiver-

Escala regional i local

nada, migració i reproducció), però s’espera que en conjunt, la seva distribució sigui es-

escales

més

petites,

els

estudis

pacial i temporalment predictible (Fauchald

d’impacte ambiental es centren general-

et al. 2002). En altres paraules, a nivell es-

ment en l’ús de l’hàbitat per part de les aus

tratègic la principal preocupació referent

marines, així com en les estratègies i els

a l’avaluació de la interacció entre les aus

processos que poden influir en la presència

marines i els OWF és la superposició espa-

d’aus marines o la disponibilitat de les seves

cial de la distribució de les aus marines amb

preses. A mesura que s’augmenta l’escala, la

el desenvolupament d’àrees clau d’OWF.

densitat d’aus s’utilitza com a estimador de

Aquest tema generalment es tracta amb la

l’hàbitat potencial ornitològic per avaluar

selecció de mapes de presència/absència

l’exposició al risc de pèrdua d’hàbitat o per-

d’una espècie emblemàtica o altament vul-

torbació. Malgrat ser una pràctica comuna,

nerable als OWFs i mapes generals de den-

l’eficàcia d’aquest mètode es veu compro-

sitat amb les xifres globals dels recomptes

mesa si les dades observades no segueixen

d’aus marines al mar. En aquest context, és

una distribució normal. De fet, les dades de

convenient aplicar un índex per integrar i

comptatge d’animals rarament són normals

resumir totes aquestes capes d’informació.

(hi ha molts sectors sense individus i la dis-

Garthe i Hüppop (2004) van proposar

tribució és contagiosa). Per tant, per al dis-

l’Índex de Sensibilitat de parcs eòlics (WSI)

seny d’estratègies de gestió ecològicament

per cartografiar la vulnerabilitat de les aus

racionals a escala regional i local de qual-

marines als parcs eòlics marins. Aquest ín-

sevol EIA cal considerar explícitament la

dex estima primer la vulnerabilitat de cada

densitat i la variabilitat temporal i espacial

espècie en funció de la seva sensibilitat als

dels ocells marins (Tobin 2004; Certain et al.

riscos de col·lisió i pertorbacions, i després

2007.).

en funció de la seva demografia i el seu es-

Els capítols tercer i quart d’aquesta tesi

tat de conservació. Aquest valor es combina

se centren en aquesta variabilitat espacial

amb l’abundància espacial de cada espècie

i temporal mitjançant l’aplicació de la Llei

per obtenir un mapa de vulnerabilitat.

Exponencial de Taylor i l’anàlisi dels movi-

Aquest mètode és general, simple i d’aplicació àmplia. Per tant, en lloc de

ments a nivell individual, respectivament.

DISCUSSIÓ

Índex de vulnerabilitat

de vol dedicat a gran altitud) i els factors que

El treball de Garthe i Hüppop (2004) va

no són importants en si mateixos, sinó com

proposar l’avaluació quantitativa de la vul-

a agreujants que poden augmentar el risc

nerabilitat de les comunitats d’aus marines

preexistent (per exemple la maniobra de

als parcs eòlics. Aquesta vulnerabilitat es

vol). En altres paraules, si una espècie marina

calcula mitjançant de l’índex de Sensibilitat

mai vola en una altura amb risc de col·lisió,

de les Espècies (SSI, Species Sensitivity In-

és irrellevant la seva maniobrabilitat, ja que

dex) que centrat en la vulnerabilitat de les

no li caldrà evitar una turbina. Això significa

espècies (a nivell individual i poblacional) i

que els factors jeràrquica estan estructurats

l’Índex de Sensibilitat als Parcs Eòlics (WSI,

entre els factors de risc primaris i els factors

Windfarm Sensitivity Index). No obstant

de agreujament. Per tant, els factors de risc

això, com s’ha demostrat en el capítol 1, la

no poden ser tractats amb una formulació

formulació matemàtica de l’índex principal

additiva. Per això es suggereix una funció al-

té supòsits implícits, tant a nivell d’espècies

ternativa que permeti millorar l’estima del

com de la comunitat, que podrien conduir a

risc de col·lisió i pertorbació.

estimacions incorrectes de la vulnerabilitat i

La segona premissa diu que tots els ti-

a una identificació parcial de les àrees clau.

pus de risc (de col·lisió, de sensibilitat a les

La primera suposició és que tots els fac-

pertorbació i sensibilitat poblacional) tenen

tors de risc associats a un determinat ti-

la mateixa importància, i una relació multi-

pus de risc tenen la mateixa importància i

plicativa. La dificultat de mesurar la impor-

mantenen una relació additiva. No obstant

tància relativa del risc de col·lisió sobre el

això, hi ha una diferència conceptual entre

risc de pertorbació justifica considerar-los

els factors inclosos en un tipus particular de

similars, tot i que sovint es considera que el

risc. Prenent el risc de col·lisió com exemple,

perill més important és la mortalitat per xoc

ens trobem amb dos tipus de factors de risc:

(Fox et al. 2006; Christel, Certain et al. 2012).

els que estan directament associats al propi

Relacionar multiplicativament la col·lisió i

risc (per exemple, el percentatge de temps

el risc de pertorbació és conflictiu, perquè

125

126

Catalan Summary aquests dos tipus de riscos no depenen l’un

també en la similitud d’aquestes espècies.

de l’altre. Són dos aspectes independents relatius al impacte potencial d’un OWF sobre

Patrons d’agregació

les aus marines i, per tant, presenten conseqüències ecològiques diferents. Al vincular-

Els mapes d’abundància permeten defi-

los de manera multiplicativa es pot subesti-

nir àrees riques en ocells (informació relle-

mar l’efecte d’un dels riscos, només perquè

vant en l’avaluació de les ubicacions dels

l’altre risc és molt baix. Per això, si els dos

parcs eòlics en alta mar), però no proporcio-

tipus de riscos han de ser combinats en un

nen informació sobre els patrons espacials

únic mapa és recomanable utilitzar una re-

dinàmics i complexos de les aus marines al

lació additiva. Malgrat això, considerem que

mar. Si bé els mapes de densitat permeten

per arribar a una decisió informada per a la

centrar-se en la detecció d’altes concentra-

gestió és preferible examinar ambdues cate-

cions d’aus marines com àrees de risc po-

gories, la del risc de col·lisió més la del risc

tencial, l’aplicació de la llei exponencial de

de pertorbació.

Taylor permet distingir explícitament en

El pas final de l’índex és la integració de la

el temps entre les zones de transició i les

vulnerabilitat de les espècies en una mesura

d’alimentació. En el capítol 3 hem corro-

de la vulnerabilitat de tota la comunitat. Per

borat la Llei exponencial de Taylor com a

això, el tercer supòsit és que la contribució

índex d’agregació en el temps i l’espai. Les

d’una determina espècie a la vulnerabilitat

àrees amb patrons d’agregació febles po-

de la comunitat és proporcional al logaritme

den ser considerades com àrees de transició

de l’abundància d’exemplars d’aquesta es-

o corredors de vol mentre que els patrons

pècie en un lloc determinat. Aquest enfoca-

d’agregació elevats són majoritàriament zo-

ment volia impedir la instal·lació de parcs

nes d’alimentació determinades per la dis-

eòlics en zones de grans concentracions

ponibilitat puntual de recursos alimentaris.

d’ocells marins. No obstant això, les espè-

La vinculació dels patrons d’agregació

cies abundants generalment tenen valors

amb un comportament particular pot pre-

més baixos de SSI mentre, al contrari, les

dir i classificar millor el risc que representa

espècies rares presenten valors més alts. Per

per a les aus marines (a nivell poblacional o

a un indret donat, SSI i abundància poden

de comunitat) l’establiment de parcs eòlics

afectar el valor final de WSI en direccions

marins. A les zones de transició, el princi-

oposades, fet que dificulta la interpretació

pal risc serà la col·lisió directa i la mortali-

de les variacions en l’índex. A més, l’ús del

tat (Desholm & Kahlert 2005; Hüppop et al.

logaritme de l’abundància en comptes de

2006.). En canvi, a les zones d’alimentació,

l’abundància simple assumeix que un indi-

la presència de les turbines es traduiria en

vidu solitari té més pes, en proporció, que

la pèrdua d’hàbitat per a les espècies amb

un individu situat en un estol d’un centenar

una forta resposta d’evitació o d’un aug-

d’aus, i aquest supòsit no té cap suport ni

ment de risc de col·lisió per a les espècies

ecològic ni ambiental. Per resoldre aques-

que experimenten un suau “efecte barrera”

tes inconsistències, es suggereix un canvi

(Masden, Haydon, et al. 2010; Perrow et al.

important en la formulació basant-se en

2011). Per tant, després de la quantificació

el treball de Leinster i Cobbold (2012) que

del patró d’agregació donada en un deter-

presenta una mesura de diversitat basada

minat sector, es pot avaluar el risc potencial

no només en l’abundància d’espècies, sinó

per les regions dins d’aquesta àrea d’acord

DISCUSSIÓ amb diferents nivells de gravetat o perill. Les

llei exponencial de Taylor.

àrees amb risc de col·lisió i pèrdua d’hàbitat

No obstant això, l’estudi presentat en el

obtindrien el nivell major de gravetat (L3),

capítol 3 demostra que la informació sobre

les àrees amb risc de col·lisió tindrien el se-

les propietats de segon ordre de distribució

güent nivell de preocupació (L2), seguida de

de les espècies (és a dir, l’agregació social)

les àrees amb risc de pèrdua d’hàbitat (L1).

proporciona informació addicional a les

Aquesta classificació es pot utilitzar després

propietats de primer ordre (densitat) per a

per triar la ubicació òptima d’un OWF (se-

l’avaluació dels impactes potencials dels

leccionant zones amb mínim perill) o per

parcs eòlics marins.

definir un protocol de monitoratge requerit per a una ubicació OWF. A més, hi ha una

Seguiment individual

evolució temporal dels patrons d’agregació que es correlaciona amb el cicle de vida de

El comportament dels ocells marins és un

l’espècie. Si s’utilitza escenaris estacionals

component important que es pot extreure

(p. ex. cria, dispersió post-nupcial, migració

del monitoratge d’ocells marcats individual-

o hivernada) o un conjunt de mesos crítics,

ment (Tremblay et al. 2009). Un dels objec-

els escenaris temporals són més fàcils de

tius del capítol 4 era estudiar el comporta-

comunicar perquè resumeixen la informa-

ment mitjançant models SSM (State-Space

ció clau que pot posar de relleu l’impacte

Models) (Jonsen et al. 2003) sobre dades de

potencial d’un OWF en un moment delicat

rastreig per satèl·lit com una alternativa fi-

de la vida dels ocells marins o es pot utilit-

nal a les eines d’avaluació espai-temporals.

zar per recomanar mesures de mitigació du-

No obstant això, el rendiment dels transmis-

rant els mesos crítics.

sors (PTT) va ser molt limitat i heterogeni

L’aplicació d’aquesta eina i la interpreta-

entre dispositius, probablement a causa de

ció dels seus resultats són particularment

problemes de bateria. La mida mostral molt

útils per a l’avaluació d’àrees amb grans

baixa i l’interval de temps entre geolocalit-

poblacions reproductores. No obstant això,

zacions successives va ser massa llarg per

seria aconsellable aplicar aquest mèto-

aplicar els State-Space Models o per posar a

de en altres escenaris per investigar pos-

prova les conclusions del capítol 3.

sibles diferències en la interpretació dels

Malgrat els problemes tècnics i la mida

riscos potencials associats amb els patrons

limitada de la mostra, podem extreure al-

d’agregació observats. També seria inte-

gunes conclusions generals dels resultats

ressant aplicar aquesta eina a sectors amb

de l’anàlisi espacial i temporal dels movi-

composició d’espècies diverses tal com ca-

ments de la gavina corsa (Larus audouinii).

bussadors (p.ex. mascarells) o cabussadors

Per alimentar-se, les aus marines han de

superficials (p. ex. ànecs marins) o en zones

compensar la variabilitat en la distribució,

amb un important corredor migratori. Final-

l’abundància, la mobilitat i la predictibili-

ment, tot i que els resultats del mètode són

tat de les seves fonts d’aliment (Bell, 1991).

consistents amb les observacions ornitolò-

marines Aquesta variabilitat implica que

giques i dades de comportament al Delta de

els ocells marins mostrin un cert grau de

l’Ebre, seria molt recomanable l’aplicació de

plasticitat en el seu comportament al mar.

les dades de telemetria per testar amb un

Entre les moltes possibles respostes com-

conjunt de dades independents la interpre-

portamentals (p. ex. temps de vol, distàncies

tació del comportament animal en base a la

sobrevolades, patrons de capbussament), el

127

128

Catalan Summary capítol 4 destaca la plasticitat en els patrons

es detecta un gran nombre de gavines cor-

d’activitat temporal i l’ús de l’hàbitat i en

ses (capítol 3, Taula 1), però els censos aeris

conseqüència cal abordar la plasticitat en

i de barca tenen una limitació en el temps

l’avaluació de qualsevol parc eòlic marí.

i en l’extensió que poden cobrir, i per tant

Molts estudis han demostrat que el patró

no permeten captar les variacions a nivell

d’activitat de les aus marines no és constant

individual en l’ús de l’hàbitat o els patrons

al llarg del dia (p. ex. Garthe et al. 2003; Cama

d’activitat temporal durant el capvespre, la

2010; Cama et al. 2012; Christel, Navarro, et

nit i l’albada.

al. 2012). Per tant, els censos d’aus marines,

Per estimar de forma adient els efectes

que han de ser efectuats amb la llum del

potencials d’un parc eòlic sobre les aus ma-

dia i generalment seguint un horari cons-

rines, cal complementar els mapes de distri-

tant, no sempre són suficients per capturar

bució i abundància amb dades de seguiment

la variabilitat de les aus marines en funció

individual. Com que és impossible aplicar

dels cicles circadians. Alguns ocells marins,

aquesta tècnica a totes les espècies d’una

per exemple, s’alimenten de petits peixos

comunitat d’aus marines aconsello que els

pelàgics, però aquest recurs només és dis-

estudis telemètrics es duguin a terme esco-

ponible a les hores prèvies a l’alba, quan

llint espècies altament vulnerables o ame-

els bancs realitzen la seva migració vertical

naçades presents a zones una amb poten-

(Blaxter & Hunter 1982). En aquest cas, els

cial interacció amb els parcs eòlics marins

censos, que generalment comencen després

(vegeu també Perrow et al. 2006; Wilson et al.

de l’alba, no permetrien detectar una zona

2009; Burger et al. 2011). En el cas d’estudi del

amb agregacions recurrents d’aus marines

Delta de l’Ebre, la gavina corsa era una bona

en aquests bancs.

candidata perquè gairebé el 65% de la po-

De vegades se subestima la plasticitat de

blació mundial cria a l’àrea (Oró et al. 2009).

l’alimentació de les aus marines, tant en re-

No obstant això, la baldriga balear (Puffinus

ferència al seu ús de l’hàbitat com a la selec-

mauretanicus) hauria estat també molt relle-

ció de preses. Quan la disponibilitat de les

vant, ja que està en perill greu (CR) segons la

preses és reduïda, les espècies especialistes

llista vermella de la UICN i , encara que no

solen modificar la seva estratègia alimen-

cria a la zona, l’espècie explota extensiva-

tària mitjançant l’extensió de la seva zona

ment la plataforma continental del Delta de

d’alimentació, el temps que passen al mar

l’Ebre com a zona d’alimentació.

o la reducció del temps entre viatges (Lewis et al. 2001; Schwemmer & Garthe 2008). Per contra, les espècies generalistes poden can-

Integració de les eines d’avaluació ambiental

viar els seus hàbitats d’alimentació o variar la seva dieta (p. ex. González-Solís et al. 1997;

En base a les eines presentades en

Schwemmer & Garthe 2008; Navarro et al.

aquesta tesi, la bibliografia consultada i

2010). En el capítol 4, les dades de seguiment

l’experiència adquirida, presento una possi-

per satèl·lit mostren que la gavina corsa té

ble integració dels mètodes descrits al llarg

àmplies àrees d’alimentació i explota in-

de la seqüència de presa de decisions per al

tensivament ambdós hàbitats (el terrestre i

desenvolupament d’energia eòlica marina

el marí), tot i que els moviments terrestres

en referència a la problemàtica ornitològi-

data gairebé no havien estat descrits. A la

ca. Aquesta integració s’inicia a gran esca-

plataforma continental del Delta de l’Ebre

la a les Avaluacions Ambientals Estratègi-

DISCUSSIÓ ques, finalitzant a escala local en els Estudis d’Impacte Ambiental (Fig. 9, pàg. 92).

Aquesta tesi es centra en els ocells marins com indicadors dels ecosistemes marins, però a aquesta informació es poden su-

(1) Regions de desenvolupament òptimes

perposar moltes altres capes d’informació a l’hora de seleccionar les regions òptimes de

Per dissenyar un desenvolupament ra-

desenvolupament. En el tema ornitològic,

cional de l’energia eòlica marina compati-

les avaluacions d’impactes acumulats o les

ble amb la conservació ornitològica cal una

dades sobre les principals rutes migratòries

estratègia prèvia a gran escala definida al-

són rellevants, però també la informació re-

menys, a nivell nacional. A aquesta escala és

lativa als grans animals marins (tortugues,

òptim aplicar l’Índex de vulnerabilitat des-

cetacis, etc.), la distribució dels peixos, el

crit als capítols 1 i 2. El factor limitant és la

bentos, etc. D’altra banda, hi ha molts parà-

disponibilitat de censos que hagin mostrejat

metres socioeconòmics a tenir en compte,

el conjunt de l’àrea estratègica. Això sol ser

els quals queden fora de l’àmbit d’aquesta

possible mitjançant censos des de vaixell

tesi i per tant no s’inclouen en la seqüència

que en molts casos es duen a terme dins el

de presa de decisions.Les eines analítiques

marc de projectes multidisciplinaris que en

presentades podrien però aplicar-se també

una mateixa campanya registren, per exem-

a una avaluació més global de l’espai marí

ple, dades oceanogràfiques, biogeogràfiques

que inclogui també tots els paràmetres bio-

de mamífers marins, i altres paràmetres

geogràfics i socioeconòmics.

ecològics. Això limita la disponibilitat de rèpliques d’aquests censos per a diferents

(2) Àrees d’exclusió

escenaris temporals, però és contrarestat per la acurada aproximació a la biodiversi-

Un cop seleccionada una regió potencial,

tat dels ocells marins que dóna els mostre-

l’Índex de vulnerabilitat també es pot aplicar

jos de vaixell. En alguns casos, però, hi ha

a dita regió. Com que és un índex compara-

disponibles censos aeris per al conjunt de

tiu, fins i tot en zones de baixa vulnerabilitat

l’àrea estratègica. Quan això és possible, els

pot detectar zones més sensibles. Aplicat a

escenaris temporals també poden ser inclo-

escala regional, recomano l’avaluació de les

sos en les Avaluacions Ambientals Estratè-

àrees de major vulnerabilitat per tal de deci-

giques.

dir si aquestes àrees han de ser excloses per

Amb els resultats dels censos, l’índex

a les futures opción d’ ubicació de OWF.

de vulnerabilitat produeix quatre panells

A aquesta escala, la decisió pot basar-

diagnòstics que capturen els elements clau

se en els mateixos censos utilitzats a gran

a tenir en compte per a l’avaluació estratè-

escala. No obstant, si es poden fer censos

gica: la vulnerabilitat a les col·lisions i les

aeris a la regió, seria òptim incloure l’escala

pertorbacions, la vulnerabilitat general de

temporal en la definició d’aquestes àrees

la comunitat i el mapa d’abundància total.

d’exclusió. Encara que no és estrictament

Amb aquesta informació, és possible detec-

necessari, si se sap que la regió avaluada té

tar les àrees amb la menor vulnerabilitat i

una forta presència d’espècies amenaçades

per tant les regions ambientalment òptimes

podria ser aconsellable reforçar l’avaluació

per al desenvolupament de l’energia eòlica

amb informació de seguiment individual de

marina.

les espècies bandera.

129

130

Catalan Summary una regió gran el que també impossibilita la

(3) Selecció d’una ubicació òptima

rèplica dels censos al llarg dels mesos per capturar la dimensió temporal dels patrons

Dins de la regió seleccionada, l’anàlisi

d’agregació.

dels patrons d’agregació presentat al capítol 3, permet cartografiar els impactes poten-

(4) Estudi d’Impacte Ambiental

cials dels parcs eòlics marins mitjançant la

definició de zones de transició (amb alt risc

Quan la localització d’un parc eòlic marí

de col·lisió) i d’alimentació (amb alt risc de

ja s’ha seleccionat, és obligatori dur a terme

pertorbació i pèrdua d’hàbitat). Associat a

un EIA. En aquest cas, el mapa de risc po-

aquest mapa de risc potencial es suggereix

tencial generat a l’etapa 3 ajuda a identifi-

una classificació dels diferents sectors de la

car els estudis necessaris per avaluar el risc

regió en funció de tres nivells de preocupa-

de col·lisió i pertorbació, i la ubicació òpti-

ció (L0, L1, L2 i L3).

ma dels punts de mostreig. Si un projecte

Aquests nivells estan directament rela-

es troba dins d’una àrea amb risc de pèrdua

cionats amb l’impacte ecològic de la presèn-

d’hàbitat, s’hauria d’estudiar la presència i

cia de les turbines en un sector, però en lloc

abundància d’espècies mitjançant una mo-

d’apuntar simplement a l’impacte potencial,

delització de l’hàbitat, per exemple, realit-

la proposta és vincular aquests nivells amb

zant censos de barca en àrees petites que

un protocol obligatori de monitoratge si un

cobreixen la ubicació del parc eòlic i els

projecte de parc eòlic és acceptat (Taula 3,

seus voltants. Les zones amb risc de col·lisió

pàg. 93). Cada nivell de perill exigeix els seus

s’haurien d’avaluar amb models de risc de

propis mètodes de seguiment, a més dels

col·lisió que podrien requerir mostrejos

punts recomanats als nivells inferiors. En fer

puntuals o transsectes per determinar alça-

això, la selecció de la ubicació òptima d’un

da i direccions de vol, radars de vigilància

parc eòlic (que usualment porta a terme una

(vertical i/o horitzontal) o sistemes amb cà-

empresa promotora) es basaria no només en

meres d’infrarojos. Si l’àrea avaluada té una

l’anàlisi tècnic del cost/benefici sino també

alt risc de col·lisió i de pèrdua d’hàbitats i

en l’increment dels costos de monitoratge

a més hi ha espècies emblemàtiques al vol-

associats a la selecció d’una ubicació menys

tant, seria convenient dur a terme estudis

favorable ecològicament. Aquest sistema de

de seguiment individual per definir millor

decisió podria promoure la selecció dels sec-

els patrons temporals i espacials d’aquesta

tors de baix perill. No obstant, fins i tot si

espècie en els entorns del futur parc eòlic.

es decideix ubicar un parc eòlic en una àrea de gran perill (L3), cal recordar que aquesta

(5) Estudis de seguiment

localització haurà estat seleccionada entre les àrees menys vulnerables definides a les etapes anteriors.

Independentment del nivell de perill de la ubicació seleccionada, quan un parc eòlic

Es recomanable fer aquesta anàlisi so-

ja està operatiu calen estudis de seguiment

bre dades de censos aeris, donat que els

per detectar qualsevol efecte imprevisible

censos de barca no permeten captar una

de la instal·lació en el medi ambient. La dis-

“instantània” dels patrons de distribució de

ponibilitat d’aquest monitoratge és crucial

les aus marines. Els censos de barca reque-

per dur a terme qualsevol acció de gestió

reixen un llarg període de temps per cobrir

adaptativa una vegada el parc eòlic funcio-

DISCUSSIÓ ni. En definir previament el monitoratge que

seu pressupost anual pel manteniment del

caldrà fer, l’empresa promotora pot tenir

parc eòlic. Cal remarcar que el pla de vigi-

una idea clara dels costos mínims de control

lància proposat a la Taula 3 sempre es pot

que estaran associats a la ubicació del parc

ajustar o expandir d’acord amb els resultats

eòlic construït, i per tant incloure’ls en el

de cada EIA.

131

Appendix

134

Appendix

Estuarine, Coastal and Shelf Science 96 (2012) 257e261

Contents lists available at SciVerse ScienceDirect

Estuarine, Coastal and Shelf Science journal homepage: www.elsevier.com/locate/ecss

Foraging movements of Audouin’s gull (Larus audouinii) in the Ebro Delta, NW Mediterranean: A preliminary satellite-tracking study Isadora Christel a, b, *, Joan Navarro c, Marcos del Castillo d, Albert Cama a, b, Xavier Ferrer a a

Departament de Biologia Animal, Facultat de Biologia, Universitat de Barcelona, Diagonal 645, 08028 Barcelona, Spain Museo Nacional de Ciencias Naturales, Consejo Superior de Investigaciones Cientíﬁcas (CSIC), C/José Gutiérrez Abascal 2, 28006 Madrid, Spain Institut de Ciències del Mar (ICM-CSIC), P. Marítim de la Barceloneta 37-49, 08002 Barcelona, Spain d C/Vilanova 8A, 07002 Palma de Mallorca, Mallorca, Illes Balears, Spain b c

a r t i c l e i n f o

a b s t r a c t

Article history: Received 27 April 2011 Accepted 8 November 2011 Available online 18 November 2011

A knowledge of the foraging strategies of marine predators is essential to understand the intrinsic factors controlling their distribution, abundance and their ecological function within the marine ecosystem. Here, we investigated for the first time the foraging movements and activity patterns of Audouin’s gull Larus audouinii by using satellite-tracking data from eight breeding adults in the main colony of the species worldwide (Ebro Delta, NW Mediterranean). Tagged gulls foraged in the marine area close to the breeding colony (62% of foraging locations) and in the terrestrial area of the Ebro Delta (mainly rice fields; 38% of foraging locations). The foraging activity patterns changed significantly throughout the day; lower from dusk through the first half of the night (19-1 h; 32% of active locations) and higher during the rest of the day (1e19 h; 75.5 4.3% of active locations). These results confirm the foraging plasticity of this seabird and, based on previous information about the dietary habits of this species, we hypothesize how its time-dependent activity patterns and habitat use could be associated with variations in the availability of marine food resources (e.g. diel vertical migrations of pelagic fish) and the exploitation of terrestrial resources (e.g. American crayfish Procambarus clarkii). 2011 Elsevier Ltd. All rights reserved.

Keywords: Ebro Delta foraging activity foraging distribution habitat use marine birds marine habitat Mediterranean Sea rice ﬁelds

1. Introduction An important issue in the feeding ecology of marine predators is the degree of plasticity of their foraging behavior. In general, specialist predators are constrained to forage on a speciﬁc habitat and time of day determined by a speciﬁc prey availability (Futuyma and Moreno, 1988; Krebs and Davies, 1993; Julliard et al., 2006). Under changing conditions of prey availability, specialists are able to adapt their foraging strategy by extending foraging range or time spent foraging (e.g. Oro et al., 1997; Lewis et al., 2001; Schwemmer and Garthe, 2008). By contrast, generalist predators have the ability to exploit different trophic resources and, consequently, they present higher plasticity in their foraging strategies (Krebs and Davies, 1993; Boyd et al., 2006; Julliard et al., 2006). This opportunistic behavior allows generalists to modify their foraging strategies (i.e. exploited habitat, range or temporal patterns) according, for instance, to the varying degree of competition for food. Indeed, * Corresponding author. Departament de Biologia Animal, Facultat de Biologia, Universitat de Barcelona, Diagonal 643, 08028 Barcelona, Spain. E-mail addresses: isadora.jimenez@gmail.com, isadora.jimenez@ub.edu (I. Christel). 0272-7714/$ e see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.ecss.2011.11.019

the foraging plasticity of marine predators has allowed these organisms to benefit from anthropogenic food resources (e.g. fisheries discards, refuse dumps or introduced prey species; Tablado et al., 2010; Ramos et al., 2011; Wagner and Boersma, 2011). Amongst marine predators, the Audouin’s gull Larus audouinii is a good example of an opportunist species that exhibits clear plasticity in its diet habits. This Mediterranean endemic species exploits small pelagic fish (their main prey, see Oro, 1998 and references therein), but also alternative anthropogenic resources such as demersal or benthonic fish from fisheries discards or invasive freshwater crabs from terrestrial habitat (Oro et al., 1996a, 1999; Oro and Ruiz, 1997; Navarro et al., 2010). This opportunistic behavior is especially relevant in breeding populations located in areas where diverse trophic resources are highly available (e.g. Oro and Ruiz, 1997; Oro et al., 1999; Navarro et al., 2010), which is the case of the breeding population located in the Ebro Delta (Fig. 1. NW Mediterranean). This colony supports ca. 12000e13000 breeding pairs of Audouin’s gull, ca. 65% of the total world population (Oro et al., 2009). The marine ecosystem of the Ebro Delta is one of the most important fishing grounds in the Mediterranean Sea, resulting in one of the largest fishing fleets in this region, which generates a high quantity of fisheries discards (Coll et al., 2008).

258

I. Christel et al. / Estuarine, Coastal and Shelf Science 96 (2012) 257e261

Moreover, freshwater resources such as the invasive American crayfish Procambarus clarkii in the rice fields of the Ebro Delta are abundant and easily available (Gutierrez-Yurrita et al., 1999), providing an alternative and proficient trophic resource for the species (Oro et al., 1996b; Longoni, 2010; Navarro et al., 2010). Although the diet habits of the Audouin’s gull are well known (e.g. Oro et al., 1997; Pedrocchi et al., 2002; Sanpera et al., 2007; Navarro et al., 2010), detailed information on the foraging movements is biased toward studies based on ship surveys (e.g. Abelló and Oro, 1998; Arcos et al., 2001; Abelló et al., 2003), which are strongly biased by the influence of fishery discards and underestimate the importance of land habitat utilization. The only previous telemetric study (radio-tracking) already pointed to the apparent importance of the terrestrial habitat for the breeding population of the Ebro Delta colony (Mañosa et al., 2004). Here, we present preliminary results of the first satellitetracking study of Audouin’s gull during the breeding season in its largest breeding colony (Ebro Delta). This paper aims to quantify the foraging range of Audouin’s gull, evaluate the habitat utilization of marine and terrestrial areas and identify the temporal patterns of the foraging activity of the species. Based on previous information about the dietary habits of this species, we also hypothesize how the observed foraging movements could be attributed to the exploitation of different trophic resources in the Ebro Delta marine and terrestrial ecosystems. 2. Material and methods 2.1. Fieldwork procedures The study was carried out at the natural reserve of Punta de la Banya in the Ebro Delta Natural Park, North Western Mediterranean Sea (Fig. 1, 40� 330 N, 0� 39’E). Punta de la Banya is a flat sandy peninsula of 2514 ha, partially occupied by saltworks and

connected to extensive rice ﬁeld areas (20,000 ha) by a 5 km-long narrow sand bar. To examine the foraging activity, we satellitetracked 8 breeding birds (4 males and 4 females) using battery powered “Platform Transmitter Terminals” (PTTs; North Star Science and Technology, LLC) during the chick-rearing period (May) of 2006 (Table 1). We captured all birds on the nest by using a drop trap (Mills and Ryder, 1979) during late incubation to reduce the risk of desertion. Once trapped, each individual was sexed, weighed, ringed and tagged with a PTT. The attached PTTs weighed 20 g and were programmed to be active in a 6 h on/5 h off duty cycle to get information on the foraging locations during one month. The PTT was ﬁxed to the mid-dorsal feathers of the mantle using Tesa tape (Wilson et al., 1997). With this method the PTT falls off after one month without the necessity to recapture the instrumented bird. The entire transmitter equipment represented between 3 and 4% of the Audouin’s gull’s body mass, so the potential effects of an additional weight on the gull’s movement were minimized (e.g. Phillips et al., 2003; Passos et al., 2010).

2.2. Satellite-tracking data and statistical analyses Data on the position of each PTT were obtained from ARGOS system (CLS, Toulouse, France) and imported to ArcView 3.2 (ESRI) using the Argos Tool extension (Potapov and Dubinin, 2005). Each position was classified according to its estimated error: Type 0 (>1000 m), Type 1 (350e1000 m), Type 2 (150e350 m), Type 3 (0e150 m), and Types A and B (without an estimated error) (ARGOS, 2006). Initial data filtering involved calculating velocities between successive satellite locations, and rejecting those for which the velocity exceeded a threshold of 50 m s�1, the maximum velocity described for this species (Rosén and Hedenström, 2001). By this procedure, up to 8% of the locations were filtered; all of them from the low-quality accuracy class “B”.

Fig. 1. (a) Breeding areas of the Mediterranean endemic Audouin’s gull Larus audouinii and study area: Ebro Delta, NW Mediterranean. (BirdLife International, 2011) (b) Map of the Ebro Delta area indicating the Audouin’s gull colony position with an asterisk and 1 km buffer area around la "Punta de la Banya" peninsula, the rice ﬁelds and wetlands shaded in dark gray and the location of the main harbors. (c) Foraging locations of 7 satellite-tracked Audouin’s gulls during the breeding period of 2006. To better visualize the foraging locations’ range the Minimum Convex polygon (short dashed line) is shown beside the 95% (solid line) and 50% (long dashed line) kernel polygons.

135

136

Appendix

I. Christel et al. / Estuarine, Coastal and Shelf Science 96 (2012) 257e261 Table 1 Summary information of PTTs performance. PTT Id

Sex

Tracking days

First location

Last location

Total locations

58978 58979 58980 58981 58982 58983 58984 58985 Total

_ _ _ _ \ \ \ \

2 1 10 2 7 0 10 3 13

15/05/2006 19/05/2006 18/05/2006 15/05/2006 15/05/2006 e 15/05/2006 18/05/2006 15/05/2006

16/05/2006 19/05/2006 27/05/2006 16/05/2006 21/05/2006 e 24/05/2006 20/05/2006 27/05/2006

6 2 31 4 6 e 32 8 89

To gain an insight into the foraging activity of the tagged Audouin’s gulls we sorted the locations into three classes, according to their spatial position. PTT locations inside the “Punta de la Banya” peninsula or within the first kilometer around it were classified into the “colony locations” group. In contrast, the locations outside the colony and the first kilometer around it were “foraging locations” (we assumed that the birds were feeding to recover the body condition lost during the incubation bout). Finally, we calculated the 95% fixed-kernel estimates of the foraging area and the maximum foraging distance from the colony. We employed logistic regression e a generalized linear model (GLM) e to test the foraging activity and habitat use. First, we tested a model with the proportion of foraging locations as the dependent variable, and we selected as the explanatory variable the “time of day” -categorized in 6-h intervals (1e7 h; 7e13 h; 13e19 h; 19-1 h)with the 7e13 h interval as the reference level. Then we analyzed habitat use by testing the effect of the explanatory variable “time of day” on the dependent variable “terrestrial vs. marine proportion of foraging locations”. The analyses were carried out using R software (R Development Core Team, 2008), calling the “glm” function with binomial error distribution and its default logit link function. A likelihood ratio test was used to compare the resulting model with the null model (without any variable) and to assess the significance of the explanatory variable “time of day”. 3. Results We obtained a total of 89 filtered PTT locations spanning a period of 13 consecutive days. One of the eight PTTs failed to give any location probably due to a battery failure, and the performance of the remaining PTTs was heterogeneous (see Table 1). Due to sample size limitations individual variability was not included in the

Fig. 2. Example of foraging trajectories for the individual “58980” (see Table 1 for more information).

259

analysis, but the movements of one of the tracked individuals is shown in Fig. 2 to illustrate the general pattern of the foraging movements. The foraging area covered by the Audouin’s gulls was 5400 km2 (95% fixed-kernel density estimate), covering both the marine area of the Ebro Delta (ca. 3300 km2) and the terrestrial area (ca. 2100 km2) (Fig. 1c). The maximum foraging distance covered ranged from 20.5 to 81.7 km (mean � sd ¼ 51.5 � 24.3 km) and was similar for both marine and terrestrial locations (T-Student test, T ¼ 1.44, df ¼ 56, p ¼ 0.15). The foraging activity changed significantly over the course of the day (Likelihood Ratio Test, c2 ¼ 13.79, df ¼ 3, p ¼ 0.003). Tagged gulls were more active at 7e13 h (78.1%), at 1e7 h (77.8% of the total locations in this period, p ¼ 0.65), and 13e19 h (70.6%, p ¼ 0.56), all of them significantly different from the 19-1 h interval (31.8%, p ¼ 0.001), i.e., the foraging activity diminished during the first half of the night (Fig. 3b). Moreover, we found that the proportion of foraging locations in marine vs. terrestrial habitats changed during the day. Although the time of day was not significant as a global explanatory variable, the model indicated a significant difference between the 13e19 h interval and the reference level 7e13 h (p ¼ 0.04) (Fig. 3c). Between 13 h and 19 h, Audouin’s gulls foraged mainly in terrestrial (41%) rather than in marine habitat (29%); during the rest of the day, they foraged mainly in marine rather than terrestrial habitat (1e7 h: 50% marine, 28% terrestrial habitat; 7e13 h: 59% marine, 19% terrestrial habitat; 19-1 h: 23% marine, 9% terrestrial habitat) (Fig. 3a). 4. Discussion Satellite-tracked Audouin’s gulls covered a foraging area that ranges 80 km, spanning both marine and terrestrial habitats. It has been widely described previously that breeding Audouin’s gulls cover large ranges when foraging. There are records of individuals foraging at 70e150 km from the breeding colony during the breeding season (Baccetti et al., 2000; Mañosa et al., 2004), and data from vessel counts suggest that individuals forage during the day and night even further offshore (Arcos and Oro, 1996; Abelló and Oro, 1998). However, the species’ terrestrial foraging movements had been scarcely described (Ruiz et al., 1996; Mañosa et al., 2004). It is well documented that Audouin’s gulls forage during the night in marine habitats preying on small pelagic fish and exploiting discards provided by nocturnal fisheries (e.g. Witt et al., 1981; Mañosa et al., 2004; Arcos et al., 2008). However, our results highlight that the species’ nocturnal activity is not homogeneous throughout the night (see Fig. 3). Satellite-tracked gulls were mainly located in the breeding colony during the hours before and after dusk (19-1 h). In the period after midnight to dawn (1e7 h) they increased their foraging activity, which then remained constant and high during the day. These results, coupled with the nocturnal arrival and departure times from the breeding colony described in Mañosa et al. (2004), confirm a peak of activity between midnight and dawn. Attendance to purse seiners during the night is considered a strategy that is only significant during trawling moratorium and winter periods (Arcos and Oro, 2002), neither of which were covered during our study; therefore, the individuals located at sea during the night were probably feeding on small pelagic fish. Accordingly, the nocturnal foraging habits of the Audouin’s gull would still rely on the capture of small pelagic fish (Witt et al., 1981; Oro, 1998), a resource that might not be available throughout the night, but only in the hours before dawn due to the diel vertical migration of the shoals (Blaxter and Hunter, 1982; Oro, 1998). With regard to diurnal activity, tagged birds showed a high foraging activity with an unexpected constant presence in

260

I. Christel et al. / Estuarine, Coastal and Shelf Science 96 (2012) 257e261

terrestrial habitats (generally rice fields or wetlands) in addition to the expected presence in marine habitat (Oro, 1998). The fact that all tagged individuals could be found in both habitats suggests that the use of terrestrial habitat was not due to the casual behavior of a single individual. This result supports previous studies that describe the use of the rice fields of the Ebro Delta by the Audouin’s gull (Ruiz et al., 1996; Mañosa et al., 2004; Longoni, 2010), probably related to the exploitation of the exotic American crayfish (Navarro et al., 2010), which is very abundant in the rice fields of the Ebro Delta (Gutierrez-Yurrita et al., 1999). Although many studies have demonstrated that the Audouin’s gull exploits trawler discards (Oro et al., 1997; Arcos, 2001; Cama, 2010), the foraging activity of our satellite-tracked individuals was higher inland than at sea in a period of time that includes the discarding peak of the trawling fleet (from 15 to 16 h; Cama, 2010). This result suggests that terrestrial foraging has become an alternative food source to trawling discards (Navarro et al., 2010), probably prompted by the interference competition for fisheries discards: namely, intraspecific competition (due to an increasing population density), and interspecific competition with the sympatric and dominant Yellow legged gull Larus michahellis (e.g. Arcos et al., 2001). In conclusion, the present study shows that Audouin’s gull foraged in both marine and terrestrial habitats and showed activity during both night and day. These results confirm the high foraging plasticity of Audouin’s gull, a species once defined as a specialist nocturnal forager that has become an opportunist on fisheries discards and terrestrial resources. However, due the limited sample size we suggest the necessity of conducting more studies using biologging methodologies (such as PTTs or GPS) to confirm the observed patterns and to gain new insight into the foraging ecology of this endangered seabird. Acknowledgments The birds were tagged with a permit from the Environmental Department of the Catalonian Government. We are grateful to D. Oro, L. Cardador and J. M. Arcos for their comments to improve this manuscript. F. Zino, C. Carboneras and J. González-Solís for their comments about attachment methods. We also appreciate the help of X. Macià, R. Loras, S. Mañosa and the Ebro Delta Natural Park team (T.Curcó, C. Vidal and F. Blanch). S. Young revised the English. Research funds were provided by a project funded by Capital Energy through agreement with Fundació Bosch i Gimpera (Contract 304683). I. C. was funded by a PhD fellowship of the University of Barcelona. J. N. was supported by a postdoctoral contract of Juan de la Cierva program (MICINN-JDC, Spanish Ministry of Science and Innovation). A. C. was funded by a PhD fellowship of the Government of Catalonia (2009FIC75). References

Fig. 3. (a) Activity (foraging in marine or terrestrial habitat; or located in the colony) during a 24 h cycle of 7 satellite-tracked Audouin’s gulls during the breeding period in Ebro Delta colony. (b), (c) Mean and 95% conﬁdence interval, of the foraging probability and foraging in marine habitat probability respectively, according to the GLM models. * indicates a signiﬁcant difference of the time block probability compared to the reference level 7e13 h.

Abelló, P., Oro, D., 1998. Offshore distribution of seabirds in the northwestern Mediterranean in June 1995. Colonial Waterbirds 21, 422e426. Abelló, P., Arcos, J., Gil de Sola, L., 2003. Geographical patterns of seabird attendance to a research trawler along the Iberian Mediterranean coast. Scientia Marina 67, 69e75. Arcos, J., 2001. Foraging ecology of seabirds at sea: significance of commercial fisheries in the NW Mediterranean. PhD thesis, University of Barcelona. Arcos, J., Oro, D., 1996. Changes in foraging range of Audouin’s gulls Larus audouinii in relation to a trawler moratorium in the western Mediterranean. Colonial Waterbirds 1, 128e131. Arcos, J., Oro, D., Sol, D., 2001. Competition between the yellow-legged gull Larus cachinnans and Audouin’s gull Larus audouinii associated with commercial fishing vessels: the influence of season and fishing fleet. Marine Biology 139, 807e816. Arcos, J., Oro, D., 2002. Significance of nocturnal purse seine fisheries for seabirds: a case study off the Ebro Delta (NW Mediterranean). Marine Biology 141, 277e286.

137

138

Appendix

I. Christel et al. / Estuarine, Coastal and Shelf Science 96 (2012) 257e261 ARGOS, 2006. Argos User’s Manual. ARGOS CLS, Toulouse, France. http://www.argossystem.org/manual/link. Last accessed 24 November 2011. Arcos, J., Louzao, M., Oro, D., 2008. Fishery ecosystem impacts and management in the mediterranean: seabirds point of view. American Fisheries Society Symposium 49, 1471e1479. Baccetti, N., Dall’Antonia, L., Magnani, A., Serra, L., 2000. Foraging routes of Audouin’s Gulls Larus audouinii from two Sardinian colonies. In: Sultana, J., Yésou, P. (Eds.), Monitoring and Conservation of Birds, Mammals and Sea Turtles of the Mediterranean and Black Seas. Environment Protection Department, Malta, pp. 150e158. BirdLife International, 2011. Distribution Maps of Birds of the World. BridLife International, Cambridge, UK. Blaxter, J.H.S., Hunter, J.R., 1982. The biology of clupeoid fishes. Advances in Marine Biology 20, 1e223. Boyd, I.L., Wanless, S., Camphuysen, C.J. (Eds.), 2006. Top Predators in Marine Ecosystems: Their Role in Monitoring and Management. Cambridge University Press, Cambridge, p. 378. Cama, A., 2010. Seabird distribution patterns and ecological factors driving larid presence at the Ebro delta shelf (NW Mediterranean). PhD thesis, University of Barcelona. Coll, M., Palomera, I., Tudela, S., Dowd, M., 2008. Food-web dynamics in the South Catalan Sea ecosystem (NW Mediterranean) for 1978e2003. Ecological Modelling 217, 95e116. Futuyma, D.J., Moreno, G., 1988. The evolution of ecological specialization. Annual Review of Ecology and Systematics 19, 207e233. Gutierrez-Yurrita, P., Martínez, J., Ilhéu, M., Bravo-Utrera, M., Bernardo, J., Montes, C., 1999. The status of crayfish populations in Spain and Portugal. In: Gerardi, F., Holdich, D. (Eds.), Crayfish in Europe as Alien Species: How to Make the Best of a Bad Situation? Crustacean Issues, Balkema, Rotterdam, pp. 161e192. Julliard, R., Clavel, J., Devictor, V., Jiguet, F., Couvet, D., 2006. Spatial segregation of specialists and generalists in bird communities. Ecology Letters 9, 1237e1244. Krebs, J.R., Davies, N.B., 1993. An Introduction to Behavioural Ecology. WileyBlackwell publishing, Oxford, p. 420. Lewis, S., Sherratt, T.N., Hamer, K.C., Wanless, S., 2001. Evidence of intra-specific competition for food in a pelagic seabird. Nature 412, 816e819. Longoni, V., 2010. Rice fields and waterbirds in the Mediterranean region and the Middle East. Waterbirds 33, 83e96. Mañosa, S., Oro, D., Ruiz, X., 2004. Activity patterns and foraging behaviour of Audouin’s gulls in the Ebro Delta, NW Mediterranean. Scientia Marina 68, 605e614. Mills, J.A., Ryder, J.P., 1979. Trap for capturing shore and seabirds. Bird-Banding 50, 121e123. Navarro, J., Oro, D., Bertolero, A., Genovart, M., Delgado, A., Forero, M.G., 2010. Age and sexual differences in the exploitation of two anthropogenic food resources for an opportunistic seabird. Marine Biology 157, 2453e2459. Oro, D., 1998. Larus audouinii Audouin’s Gull. BWP Update. Oxford University Press, Oxford, pp. 47e61. Oro, D., Jover, L., Ruiz, X., 1996a. Influence of trawling activity on the breeding ecology of a threatened seabird, Audouin’s gull Larus audouinii. Marine Ecology Progress Series 139, 19e29. Oro, D., Genovart, M., Ruiz, X., Jiménez, J., García-Gans, J., 1996b. Differences in diet, population increase and breeding performance between two colonies of

261

Audouin’s gulls Larus audouinii during breeding seasons affected by a trawling moratorium. Journal of Avian Biology 27, 245e251. Oro, D., Ruiz, X., 1997. Exploitation of trawler discards by breeding seabirds in the north-western Mediterranean: differences between the Ebro Delta and the Balearic Islands areas. ICES Journal of Marine Science 54, 695e707. Oro, D., Ruiz, X., Jover, L., Pedrocchi, V., González-Solís, J., 1997. Diet and adult time budgets of Audouin’s gull Larus audouinii in response to changes in commercial fisheries. Ibis 139, 631e637. Oro, D., Pradel, R., Lebreton, J.D., 1999. Food availability and nest predation influence life history traits in Audouin’s gull, Larus audouinii. Oecologia 118, 438e445. Oro, D., Pérez-Rodríguez, A., Martínez-Vilalta, A., Bertolero, A., Vidal, F., 2009. Interference competition in a threatened seabird community: a paradox for a successful conservation. Biological Conservation 142, 1830e1835. Passos, C., Navarro, J., Giudici, A., González-Solís, J., 2010. Effects of extra mass on the pelagic behavior of a seabird. The Auk 127, 100e107. Pedrocchi, V., Oro, D., González-Solís, J., Ruiz, X., Jover, L., 2002. Differences in diet between the two largest breeding colonies of Audouin’s gulls: the effects of fishery activities. Scientia Marina 66, 313e320. Phillips, R., Xavier, J.C., Croxall, J.P., Burger, A.E., 2003. Effects of satellite transmitters on albatrosses and petrels. The Auk 120, 1082e1090. Potapov, E., Dubinin, M., 2005. Argos-tools Manual. The Falcon Research Institute. National Avian Research Center, Carmarthen, UK. Ramos, R., Ramírez, F., Carrasco, J.P., Jover, L., 2011. Insights into the spatiotemporal component of feeding ecology: an isotopic approach for conservation management sciences. Diversity and Distributions 17, 1e12. R Development Core Team, 2008. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria, ISBN 3900051-07-0. URL. http://www.R-project.org/. Rosén, M., Hedenström, A., 2001. Testing predictions from flight mechanical theory: a case study of Cory’s shearwater and Audouin’s gull. Acta ethologica 3, 135e140. Ruiz, X., Oro, D., Martínez-Vilalta, A., Jover, L., 1996. Feeding ecology of Audouin’s gull Larus audouinii in the Ebro Delta. Colonial Waterbirds 19, 68e74. Sanpera, C., Ruiz, X., Moreno, R., Jover, L., Waldron, S., 2007. Mercury and stable isotopes in feathers of Audouin’s gulls as indicators of feeding habits and migratory connectivity. The Condor 109, 268e275. Schwemmer, P., Garthe, S., 2008. Regular habitat switch as an important feeding strategy of an opportunistic seabird species at the interface between land and sea. Estuarine, Coastal and Shelf Science 77, 12e22. Tablado, Z., Tella, J.L., Sánchez-Zapata, J.A., Hiraldo, F., 2010. The paradox of the longterm positive effects of a North American Crayfish on a European community of predators. Conservation Biology 24, 1230e1238. Wagner, E.L., Boersma, P.D., 2011. Effects of fisheries on seabird community ecology. Reviews in Fisheries Science 19, 157e163. Wilson, R.P., Pütz, K., Peters, G., Culik, B., Scolaro, J.A., Charrassin, J.B., RopertCoudert, Y., 1997. Long-term attachment of transmitting and recording devices to penguins and other seabirds. Wildlife Society Bulletin 25, 101e106. Witt, H.H., Crespo, J., de Juana, E., Varela, J.M., 1981. Comparative feeding ecology of Audouin’s gull Larus audouinii and the Herring gull L. argentatus in the Mediterranean. Ibis 123, 519e526.

Offshore wind energy and birds: Integrating assessment tools in space and time:

Chapter 1: A refined methodology to estimate the vulnerability of seabird community to the establishment of offshore wind farms

· Chapter 2: Wind farm Sensitivity Index for seabirds - Assessing offshore wind energy development on the coasts of the Iberian Peninsula

· Chapter 3: Seabird aggregative patterns: a new tool for offshore wind energy risk assessment

· Chapter 4: Foraging movements of Audouin’s gull (Larus audouinii) in the Ebro Delta, NW Mediterranean: A preliminary satellitetracking study