Download - Tesis Carolina Soto Navarro - copia.pdf
Seediscussions,stats,andauthorprofilesforthispublicationat:http://www.researchgate.net/publication/272479844
Distribution,abundanceandinteractionpatternsbetweensympatriccarnivoresinamediterraneanprotectedarea
THESIS·FEBRUARY2013
READS
49
1AUTHOR:
CarolinaSotoNavarro
EstaciónBiológicadeDoñana
8PUBLICATIONS4CITATIONS
SEEPROFILE
Availablefrom:CarolinaSotoNavarro
Retrievedon:05October2015
Carolina Soto Navarro
Patrones de distribución, abundancia e interacciones entre
Carnívoros Simpátridosen un área mediterránea protegida
Carolina Soto Navarro
‘Patrones de distribución, abundancia e interacciones entre carnívoros simpátridos en un área mediterránea protegida’
© Carolina Soto NavarroIlustraciones: Carolina Soto NavarroDiseño y maquetación: Ramsés GarcíaFotografía: Carolina Soto Navarro, Ramsés García
Patrones de distribución, abundancia e interacciones entre carnívoros simpátridos en un
área mediterránea protegida
Distribution, abundance and interaction patterns between sympatric carnivores in a Mediterranean protected area
Tesis doctoralCarolina Soto Navarro
Sevilla, Diciembre 2012
Director de tesisDr. Francisco Palomares FernándezProfesor de InvestigaciónDepartamento de Biología de la ConservaciónEstación Biológica de Doñana-CSIC (EBD-CSIC)Sevilla-España
TutorDra. Mª José Leiva MoralesProfesora Titular Departamento de Biología Vegetal y EcologíaUniversidad de SevillaSevilla-España
A mis padres
A la tata nana
Para Ramsés
‘We cannot ease the burdens of the past, but we can atone by assuring the carnivores of the future’
Where the wild things were. W. Stolzenburg
Índice
ESTRUCTURA DE LA TESIS DOCTORAL............................................
INTRODUCCIÓN........................................................................................
Contexto de la tesis............................................................................. Marco teórico...................................................................................... Teoría de la selección de hábitat.............................................
El concepto de hábitat............................................................. El concepto de nicho ecológico..............................................
El estudio de la selección de hábitat....................................... Escalas espacio-temporales de investigación......................... De lo realmente importante y disponible para los individuos
Características ecológicas de las especies y mecanismos de coexistencia.............................................................................
De cómo simplemente encontrar el hábitat más idóneo suele nosersuficiente.......................................................................
Objetivos de la tesis doctoral..............................................................
Especies y área de estudio...................................................................
Las especies modelo................................................................
Área de estudio........................................................................
CHAPTER 1. Non-biological factors affecting track censuses: implications for sampling design and reliability.............................................
CHAPTER 2. Fine-scale habitat use and niche separation in a guild of sympatric carnivore species that differ in life-history traits ...........................
CHAPTER 3. Species abundances in a community of sympatric carnivores: a trade-off between habitat selection and interspecificinteractions?....................................................................................................
11
12
13
21
21
21
25
28
29
31
32
36
39
41
41
46
65
93
135
CHAPTER 4. Human-related factors regulate dog presence in protected areas: implications for conservation and management control...........................
CHAPTER 5. Surprising low abundance of European wildcats in a protected area of southwestern Spain.............................................................................
CONCLUSIONES........................................................................................
AGRADECIMIENTOS...............................................................................
157
185
209
213
11
ESTRUCTURA DE LA TESIS DOCTORAL
La presente tesis consta de los siguientes apartados: una Introducción al
estado de la cuestión y al marco teórico de la tesis; los Objetivos de la tesis; una
sección de Especies y área de estudio en la que se describen las especies modelo
y el área de estudio; redactados todos en castellano; cinco Capítulos Temáticos,
presentados en formato de manuscritos publicados y por publicar en revistas
internacionales, y por lo tanto escritos en inglés, aunque acompañados de sus
respectivos resúmenes encastellano;y,finalmente, lasConclusiones principales
obtenidas en la tesis redactadas en castellano.
Introducción
13
Introducción
Contexto de la tesis
Los seres humanos hemos cambiado la biosfera1 sustancialmente, hasta
el punto que algunos abogan por el reconocimiento de la época actual en la que
vivimos como una nueva era geológica, el ‘Antropoceno’ (Steffen et al. 2011,
Rockström et al. 2009, Zalasiewicz et al. 2011). Existen evidencias de que la Tierra
como ecosistema global puede estar acercándose a una transición crítica o punto
deinflexiónquepuedesuponeruncambiodeestado2 rápido a escala planetaria en
un período de décadas o siglos, si no se ha iniciado ya en la actualidad (Barnosky
et al. 2012). A pesar de que vivimos en una era geológica con la mayor riqueza
de especies y diversidad, nos encontramos en el comienzo de un fenómeno de
extinción en masas (Primack 2006). La principal causa de esta actual crisis a escala
global es el incremento de la alteración humana de la Tierra (Vitousek et al. 1997,
Primack 2006, Sinclair et al. 2006). De hecho, las tasas de crecimiento actuales de
la población mundial (77 millones de personas al año) así como el incremento de
las actividades humanas han supuesto cambios rápidos y drásticos en los paisajes
naturales (como la transformaciónde entreun tercioy lamitadde la superficie
terrestre (Vitousek et al. 1997, Hoekstra et al. 2005). El consumo excesivo de los
recursos y la expansión de las áreas humanizadas han llevado a la destrucción directa
1 El término ‘biosfera’ fue acuñado por el geólogo Eduard Suess en 1875, pero el concepto ecológico de biosfera procede de 1920 con Vladimir I. Vernadsky, precediendo a la introducción en 1935 del término ecosistemaporArthurTansley.Enestatesisempleoladefinicióndebiosferacomoecosistemaglobal.
2 En ecología, la teoría de estados estables alternativos predice que los ecosistemas pueden existir en múltiples ‘estados’ (conjunto de condiciones bióticas y abióticas únicas). Estos estados alternativos no son transitorios y por lo tanto se consideran como estables a escalas de tiempo ecológicamente relevantes. Los ecosistemas pueden sufrir una transición de un estado estable a otro, en lo que se conoce como un ‘cambio de estado’ (a veces llamado un cambio de fase o cambio de régimen), cuando son sometidos a perturbaciones. Uno de los cambios de estado más rápidos del planeta y el más reciente, ha sido la transición desde la última era glacial al presente período interglacial (Scheffer et al. 2009, Lenton 2012) que se produjo a lo largo de milenios (Hoek 2008).
14
Introducción
y fragmentación3 de hábitats naturales (por ejemplo, debido a la red de carreteras, la
expansión de las zonas agrícolas y ganaderas, y el desarrollo de las ciudades). De una
manera indirecta, procesos inducidos por el hombre también causan la degradación
de los hábitats de numerosas especies a través de la contaminación (pesticidas,
herbicidas, emisiones de industrias químicas y de automóviles que conducen a un
cambio climático anómalo (Houghton et al. 2001) y la introducción de especies
invasoras (Primack 2006). Debido a estos cambios drásticos en los ecosistemas,
muchas especies nativas se han extinguido o están en peligro de extinción debido
al impacto negativo de la reducción de hábitats y su degradación (Tilman et al.
1994, Brooks et al. 2002, Primack 2006). La transformación de la tierra es la fuerza
motriz de la amenaza para la biodiversidad a nivel mundial (Vitousek et al. 1997), y
este proceso está funcionando tan rápido que para la mayoría de las especies no hay
tiempoparaunaadaptaciónatalesmodificacionesatravésdeunacompensación
evolutiva (Teyssèdre 2004).
Almodificarelhábitatdenumerosasespecies, los sereshumanossehan
convertidoenunaparteintegraldesuentorno.Estainfluenciatieneefectoadiferentes
escalasespacialesynivelesbiológicos,talescomoladistribucióngeográficadelas
especies, la organización espacial de las poblaciones y el comportamiento individual
aescalafina.Porejemplo,lafragmentacióndehábitatspuedeimplicarladivisión
de las actuales poblaciones en subpoblaciones o metapoblaciones4 alterando así su
dinámica (por ejemplo, Banks et al. 2005). Este escenario puede restringir a muchas
3 Existeunaampliadiversidaddedefinicionesparaeltérminofragmentacióndehábitats.Enestatesislodefinirécomoelprocesoporelqueunhábitatgrandesetransformaenpequeñosparchesdelmismohábitat(véase la revisión de Fahrig 2003).
4UsaréladefinicióndeMoilanenetal.(1998):“una metapoblación es un conjunto de poblaciones locales que habitan distintos parches de hábitat distribuidos en un espacio definido”.
15
Introducción
especies a reservas naturales y áreas adyacentes en gran parte del mundo (Woodroffe
& Ginsberg 1998). No obstante, las áreas protegidas no están exentas en muchos
casos de las alteraciones relacionadas con las actividades humanas, bien porque
actividades humanas de alteración del medio han tenido lugar de forma común o
bienporquesufrenlasconsecuenciasdelainfluenciadelasáreasquelascircundan.
Además, para muchos de los ambientes y ecosistemas no existe una información
precisanifiablesobre lascomplejas interrelacionesqueregulanydeterminan la
distribución y abundancia de las especies, ni los efectos que diferentes factores
tienen sobre el individuo, las poblaciones o las comunidades.
Dentro de este contexto, la comprensión de la relación entre los organismos
ysuambientetieneimplicacionesfundamentalesenvariasdisciplinascientíficas
como la Ecología (por ejemplo, ¿cómo los cambios ambientales afectan a los
individuos o a la dinámica poblacional?), Evolución (por ejemplo, ¿cómo afectan
estoscambiosalfitnessdelosindividuos?,¿cómoseadaptanlasespeciesaestos
cambios, ¿cuáles son los fenotipos que mejor se adaptan a estos cambios?) y
Biología de la Conservación (por ejemplo, ¿cuáles son los “mejores” hábitats que
deben ser priorizados para la conservación de las especies?).
La importancia atribuida a los recursos naturales varía entre generaciones
junto con los cambios en la sociedad y las consecuencias de los niveles de explotación
anteriores (Conover 2002), pero el reconocimiento de que la biodiversidad desempeña
un papel esencial en el bienestar humano y en el equilibrio de los ecosistemas es
creciente, y ha llevado a la generación de medidas urgentes para incrementar la
conservación de especies y hábitats en todo el mundo, lo que también se traduce en
la cristalización de la era moderna de la ciencia y política conservacionistas. Desde
16
Introducción
que se acuñó el término en 1978 y se creó esta disciplina5, los estudios centrados
en la Biología de la Conservación han crecido exponencialmente hasta alcanzar la
relevancia actual que tiene esta área de investigación (Figura 1).
La distribución de las especies en el medio ambiente (también aplicable
para las poblaciones e individuos, pero me referiré a las especies en esta sección
por conveniencia de lectura) surge de la interacción entre eventos deterministas
y estocásticos (Corsi et al. 2000). De hecho, es el resultado de la interacción
entre eventos biológicos (por ejemplo, alimentación, reproducción o dispersión;
deterministas) y eventos impredecibles (por ejemplo, incendios, tormentas;
estocásticos). Durante estos eventos biológicos, los animales pueden elegir las
áreas que mejor satisfagan sus requerimientos ecológicos (por ejemplo, áreas con
abundantes recursos en épocas de escasez), es decir, pueden elegir hábitats a través
de sus movimientos. En un contexto ecológico, los requerimientos de las especies
seidentificanportantoenelmarcodelaselección de hábitat, un concepto que
desarrollaré más adelante.
5 El término Biología de la Conservación fue introducido como el título de una conferencia que tuvo lugar en la Universidad de California, San Diego en La Jolla, en 1978, organizado por los biólogos Bruce Wilcox y Michael E. Soulé. Conposterioridad,en1987,secreólaprimerasociedadcientíficaprofesional(The Society for Conservation Biology); “The Society is a response by professionals, mostly biological and social scientists, managers and administrators to the biological diversity crisis that will reach a crescendo in the first half of the twenty-first century. We assume that we are in time, and that by joining together with each other and with other well-intentioned persons and groups, the worst biological disaster in the last 65 million years can be averted. … Although we have varying philosophies, we share a faith in ourselves, as a species and as individuals, that we are equal to the challenge.… For these reasons we join together in professional alliance, in the service of each other, but also in the service of the less articulate members of our evolutionary tree” (Soulé 1987:4-5).
17
Introducción
La selección de hábitat es un concepto central en Ecología, debido a su
influenciaenladinámicadelaspoblacionesysupersistencia,lasinteraccionesentre
especies, y comunidades ecológicas (Morris 2003). En el contexto de la Biología de
la Conservación, el estudio de la selección de hábitat tiene por tanto implicaciones
importantes. Además, los modelos de hábitat son herramientas fiables para la
conservaciónymanejodeespeciesamenazadasquepermitenlaidentificaciónde
hábitats relativamente6 idóneos a proteger y/o gestionar adecuadamente para la
conservación de las especies.
6 He utilizado el término ‘relativamente’ para subrayar que hoy en día, debido a la alteración humana delpaisaje,laexpresiónde‘hábitatidóneo’debeserinterpretadaconcautela.Enmiopinión,serefierealo‘mejor dentro de lo malo’ más que simplemente a una calidad de hábitat buena para la especie según sus requerimientos. En esta tesis emplearé ‘la idoneidad de hábitat’ en este sentido.
Figura 1.Tendenciaenelusodeltérmino“BiologíadelaConservación”enlaspublicacionescientíficas.Los resultados proceden de una búsqueda en la ISI Web of Knowledge (http://isiknowledge.com/), con el término “conservation biology” como topic keyword.Lalíneaverticalsefijóenelaño1978.
18
Introducción
El objetivo de esta tesis es comprender los requerimientos ecológicos de un
gremio7 de mamíferos carnívoros silvestres y domésticos a nivel de población en un
área mediterránea protegida.
Los mamíferos constituyen un foco importante de conservación en una tasa
desproporcionadamente alta teniendo en cuenta que sólo representan un pequeño
porcentaje del número total de especies que existen en la tierra (alrededor de 4.500
especiesdemamíferosdemásde1.000.000deespeciestaxonómicasclasificadas
hasta 1970) (May 1992, Entwistle & Stephenson 2000). Los carnívoros, en
particular, constituyen un grupo de especies muy carismáticas que se han utilizado
como “especies bandera” (flagships)8 en muchos programas de conservación de
la biodiversidad y hábitats naturales. No obstante, su conservación se enfrenta a
diversos problemas. Generalmente presentan bajas densidades poblacionales, bajo
rendimiento reproductivo, períodos de gestación relativamente largos, necesitan
grandessuperficiesenunbuenestadodeconservación(notolerangrandeszonas
urbanasodealtaactividadhumana)ytiendenasermuyelusivos(loquedificulta
su estudio) (Cardillo et al. 2004, Karanth & Chellam 2009, Schipper et al. 2008).
Además, en algunos casos se consideran una amenaza para el ser humano o
sus actividades surgiendo así un conflicto de intereses (Inskip& Zimmerman
2009).Eseescenariodeconflictosrelegaaestasespeciesenmuchasocasiones
a reservas naturales y áreas adyacentes que deben permanecer ecológicamente
intactas y su gestión deber adoptar un enfoque metapoblacional que traspase las
fronteras de las reservas para evitar problemas como la endogamia y fenómenos
7 empleado en el sentido de un grupo de especies que utilizan recursos similares y por lo tanto, pueden competir.
8aunqueexistenmúltiplesacepcionesemplearéladefinicióndeHeywood1995;“especies carismáticas que sirven como símbolos o iconos para estimular la conciencia conservacionista”.
19
Introducción
estocásticos (como brotes epidémico) (Fernández et al. 2007) y asegurar así el
éxito conservacionista. Muchas especies de carnívoros han sufrido una disminución
dramática en los últimos pocos cientos de años y algunas se encuentran entre los
mamíferos terrestres más amenazados del planeta (ej. Ceballos & Ehrlich 2002,
Rodríguez & Delibes 2003, Naves et al. 2003). La pérdida de hábitat, el agotamiento
de sus presas, la explotación de pieles y la persecución directa han contribuido a la
disminución de muchas de estas especies en el pasado (Woodroffe 2001). Hoy en día,
la persecución directa y la disminución de sus potenciales presas son las amenazas
más inmediatas a corto plazo pero la pérdida de sus hábitats naturales y la mortalidad
adicionalrecientecausadaporeltráficoson,probablemente,lasmayoresamenazasa
largo plazo para su persistencia (Ginsberg 2001, Kerley et al. 2002, Burkey & Reed
2006).
Afortunadamente, el reciente reconocimiento de la importancia ecológica
de los depredadores y la falta de información para apoyar las estrategias de
recuperación ha llevado a una mayor preocupación por los efectos de los cambios
inducidos por el hombre en estas especies (Miller et al. 2001, Sergio et al. 2008). La
eliminación de un carnívoro superior en un ecosistema puede tener un impacto en la
abundancia relativa de las especies presa y en el gremio de carnívoros en general (por
ejemplo, Palomares et al. 1996), causando efectos en cascada a través de las cadenas
tróficashastalasplantas,afectandolasinteraccionesentrelasespeciesasícomola
estructurade lascomunidadesecológicasymodificando losprocesosbásicosde
funcionamiento de un ecosistema (por ejemplo, Estes & Palmisan 1974, Crooks
& Soulé 1999, Duffy et al. 2007, Delibes-Mateos et al. 2008). Un ejemplo clásico
de ello es la disminución de la densidad poblacional del puma (Puma concolor)
en el Parque Nacional Zion (Utah, EE.UU.), que condujo a un incremento en las
20
Introducción
densidades de venados mula o bura (su principal especie presa), una mayor presión
de herbivoría y en consecuencia un descenso en el reclutamiento de álamos de
rivera, un aumento en las tasas de erosión de los márgenes de rivera y una reducción
resultante de la abundancia de especies tanto acuáticas como terrestres (Ripple &
Beschta 2006). Otro ejemplo relevante de su importancia es el proceso conocido
como ‘liberación de mesodepredadores’ (mesopradator release)9 (por ejemplo,
Palomares et al. 1998, Gehrt & Prange 2007). Es decir, los carnívoros presentan un
papel claro en el mantenimiento directo o indirecto de la biodiversidad, mediante el
controldemesodepredadoresydiversificacióndelaspresas(Terborghetal.1999,
Miller et al. 2001). Estos estudios constituyen ejemplos de investigaciones holísticas
aniveldeecosistemayestimulanenfoquessimilaresendiferentesbiomas,afinde
comprender plenamente el papel de los mamíferos carnívoros en el funcionamiento
de los ecosistemas y su representatividad como especies clave (keystone species)10.
Esto sólo se puede lograr con investigaciones a largo plazo basadas en protocolos
estrictos de monitoreo de las especies y su medio ambiente (Yoccoz et al. 2001).
9 Las ideas relativas a la ‘liberación de mesodepredadores’ se remontan varias décadas, cuando los ecologistas comenzaron a observar que la eliminación de depredadores originaba explosiones poblacionales de otras especies inferiores (por ejemplo, Paine 1969, Pacala & Roughgarden 1984). El término fue acuñado por Soulé et al. (1988) para describir un proceso mediante el cual las poblaciones de mamíferos carnívoros de tamaño intermedio se hacían más prevalentes en ausencia de un carnívoro superior, y las poblaciones de diversas aves se veían deprimidas como consecuencia de ello. En esta tesis empleo el término de una manera másampliasegúnBrasharesetal.(2010)paradefinirlaexpansiónendensidadodistribución,oelcambioencomportamiento de un predador de rango medio como resultado de la disminución en densidad o distribución de un predador superior. Aunque la liberación de mesodepredadores se emplea normalmente en el contexto de lateoríatróficadecascadas(porejemplo,Bergeretal.2008,Brasharesetal.2010),setrataesencialmentedeuna interacción intragremial entre depredadores.
10 Paine 1969; “especies cuya presencia es crucial en el mantenimiento de la organización y diversidad de las comunidades ecológicas; especies excepcionales, en relación con el resto de la comunidad, en su importancia”.
21
Introducción
Marco teórico
Teoría de la selección de hábitat
Los patrones de distribución de las poblaciones en el medio natural son
el resultado de procesos que ocurren a diferentes escalas espaciales. La elección
individual de las características del medio ambiente es una de las fuerzas motrices
que opera a escalafina (Turchin 1998).Esta elección individual es inherente al
concepto de selección de hábitat,definidoporJohnson(1980)como“el proceso
por el cual un animal elige los componentes del hábitat a usar”. No obstante el
conceptodehábitatdebedefinirseyaclararseantesdedesarrollarelconceptode
selección de hábitat.
El concepto de hábitat
Aunque el concepto de hábitat es fundamental en Ecología, carece de una
clarayconsistentedefinición,apesarde losnumerososesfuerzosdeunificación
(Whittaker et al. 1973, Hall et al. 1997, Morris 2003, Kearney 2006), y su utilidad
es a veces incluso controvertida (Mitchell 2005). Básicamente, este término se usa
a menudo para describir el medio físico de las especies, poblaciones o individuos
a diferentes escalas espaciales. El hábitat se considera a veces sólo como una
descripción de la naturaleza física de un lugar (componentes abióticos y bióticos)
donde un organismo vive o puede potencialmente vivir (Kearney 2006, Morrison et
al.2006).Enocasionesladefinicióndehábitatincluyelosconceptosdepersistencia
de especies/poblaciones o la supervivencia y reproducción de los individuos
(Whittaker et al. 1973, Hall et al. 1997, Morris 2003). Sin embargo, todavía no
existeunconsensosobre laspropiedadesespecíficas individualesdelhábitat,ya
que esto relaciona la presencia de una especie, población o individuos con las
22
Introducción
características físicas y biológicas de un área (Hall et al. 1997). En mi opinión,
ladefinicióndeesteconceptodependeengranmedidadelcontextoenelquese
emplea. Cuando estamos hablando del hábitat de una especie (o de una población o
individuo) desde una perspectiva evolutiva, es evidente que el hábitat debe incluir el
fitness individual o la persistencia de las especies/poblaciones. En este contexto, la
respuesta a la pregunta: ‘¿Cuál es el hábitat de esta especie?’ incluye implícitamente
la noción de ‘buen’ hábitat en el que la especie puede persistir. Sin embargo, los
estudios de selección de hábitat rara vez consideran o miden ningún componente
relacionado con el fitness (pero ver McLoughlin et al. 2007), ya que es difícil
relacionar estas medidas con el hábitat de las especies. De hecho, los investigadores
tradicionalmente describen el hábitat de una especie, relacionando la presencia de
individuos, poblaciones o especies con las características de un área, asumiendo
que la presencia de la especie es un buen indicador de la calidad del hábitat. En este
contexto, el concepto de hábitat se emplea en un sentido más espacial. Por ejemplo,
consideremos una población en una dinámica de fuente-sumidero (para más detalles
véase Pulliam 1988), donde la población fuente (cuya reproducción local es mayor
que la mortalidad local) vive en una zona con características ambientales diferentes
que la población sumidero (cuya reproducción local no compensa el ritmo de la
mortalidad local). Sin la población fuente, la población sumidero no persistiría.
Por tanto, el área donde vive la población sumidero (a menudo denominado como
‘hábitatsumidero’)noseríaun‘hábitat’paralaespecieenelcontextodeladefinición
de rendimiento o persistencia.
Probablementeseaimposibleunificarestosconceptos,porloquemásbien
deberíadefinirseelconceptodehábitatespecíficamenteparacadaestudio,según
el contexto. Creo que la idea de persistencia o rendimiento de las especies debe
23
Introducción
referirse al concepto subyacente de calidad del hábitat, “la capacidad del medio
ambiente para proporcionar las condiciones apropiadas para la persistencia del
individuo y la población” (Hall et al. 1997). En este contexto, los hábitats sumidero
siguen siendo hábitats, pero de mala calidad, induciendo una alta mortalidad o una
bajareproducción.Porlotanto,preferiríaunadefinicióndehábitatquenoincluyera
la noción de persistencia, sino más bien las características físicas y biológicas de un
áreaenlaqueunaespecie(poblaciónoindividuo)puedevivir.Sinembargo,prefiero
pedir prestadoparte de la definiciónproporcionadaporWhittaker (1973), quien
destacó la propiedad multivariante de un hábitat: “Las m variables del ambiente
físico [biológico] y químico que forman gradientes espaciales en un paisaje o área
definida como ejes en un hábitat representado como un hiperespacio. La parte de
este hiperespacio que ocupa una determinada especie [en una escala particular
de espacio y tiempo] representa un hipervolumen que define su hábitat”. Esta
definiciónseacercabastantealadefinicióndenichoecológico(véaselasección
siguiente y Figura 2). Por lo tanto, en esta tesis trato de interpretar la selección del
hábitat de los animales haciendo con cautela inferencias acerca de la calidad del
mismo (aunque al tratarse de un área protegida, a pesar de que se llevan y se han
llevado a cabo de manera tradicional ciertos usos humanos del paisaje en la zona de
estudio, la alteración en cuanto a fragmentación y alteración del mismo se supone
menor que en otro tipo de áreas más humanizadas).
24
Introducción
Figura 2.Representaciónesquemáticadelnichoecológico.Lasflechasnegrasrepresentanvariablesambientales (e.g. cobertura de matorral, disponibilidad de presas), y por tanto el espacio ecológico. La elipse gris oscuro se corresponde con los valores de esas variables que están disponibles para la especie (población o individuos). La elipse gris claro representa el rango de valores usado por la especie, es decir, su nicho ecológico.
Siguiendoestadefinición,unparchedehábitatdescribiráunsubconjunto
del hábitat de la especie, es decir, una combinación particular de los componentes
(las variables del hábitat) que constituyen el hábitat de la especie. De acuerdo con
la cuestión de interés, las variables del hábitat pueden referirse como variables
ambientales (por ejemplo, elevación, tipo de vegetación, condiciones climáticas),
pero también puede integrar sus congéneres (por ejemplo, la densidad de población)
u otras especies (por ejemplo, la densidad de presas para un depredador). El término
‘parche’ se utiliza a menudo para describir áreas delimitadas que contienen una
cantidad limitada de recursos agregados en un ambiente pobre en recursos a
mayor escala (Cezilly & Benhamou 1996). Estos conceptos se pueden utilizar
para diferentes entidades, tales como especies, poblaciones, individuos o incluso
comunidades.
Niche(used)
E1
E3 E2
Environment(available)
25
Introducción
El concepto de nicho ecológico
Esteconceptotambiénsufredelafaltadeunadefiniciónunificadayseconfunde
a menudo con el concepto de hábitat. Fue desarrollado por primera vez por Grinnell
(1917) para referirse a todas las características del medio ambiente que le permiten a
unaespeciesobreviviryreproducirse(nóteselasimilitudconladefinicióndehábitat
mencionada, por ejemplo, Hall et al. 1997). Posteriormente, Elton (1927) introdujo
el papel funcional de la especie dentro de su comunidad, en su nueva definición
del concepto. Estos autores están detrás de las controversias pasadas y actuales.
¿Tenemos en cuenta el impacto de las especies en su medio ambiente y comunidad o
sólo el efecto del ambiente en la especie, es decir, el efecto de factores limitantes11en
la especie? Esto también depende del contexto. En 1957, Hutchinson formalizó el
conceptodenichoconunmodelogeométrico.Definióelnichocomoelhipervolumen
en el espacio multivariado de variables ambientales (el espacio ecológico, Figura 3)
dondeunaespeciepuedepersistir(Figura2).Estadefiniciónhacehincapiéenlagama
de condiciones ambientales necesarias para la persistencia de la especie, es decir, el
nicho Grineliano12 (que es similar al concepto de hábitat). En este contexto, el nicho
ecológico representa la posición de la especie en la gama de condiciones ambientales,
de manera que cada dimensión del nicho se corresponde por tanto con un subconjunto
de este rango potencial o realmente importante para la especie. Hutchinson no obstante
reconoció el papel potencial de la especie en su comunidad mediante la descripción de
dos tipos de nicho: el nicho fundamental y el nicho realizado.
11 los factores limitantes son “cualquier proceso [o factores] que afectan de una manera cuantificable el crecimiento de una población”(Messier1991),talycomorecursostróficos,refugiosocondicionesclimáticas.. 12definidoporGrinnell1917.
26
Introducción
Figura 3.(a)Espaciogeográficoy(b)ecológico.Lalocalización(normalmentedefinidapordoscoordenadas en el espacio; longitud y latitud) de una especie se emplea con frecuencia para analizar sus propiedades ecológicas en el espacio ecológico de variables ambientales (E1 a E3). (Adaptado de Calenge 2005).
El primer término corresponde al lugar ocupado por una especie en ausencia
de competencia. Sin embargo, el nicho fundamental rara vez se produce en la
naturaleza, ya que los ecosistemas están compuestos por conjuntos de especies
que coexisten e interactúan entre ellas. Por lo tanto, la presencia de una especie
no indica necesariamente que el hábitat sea el óptimo, pero es el resultado de un
trade-offentre lacalidaddelhábitaty lacompetencia intrae interespecíficaque
limitalosrecursoseinterfiereelaccesoaellos(VanHorne1983,Araujo&Guisan
2006, Soberón 2007). El segundo término tiene en cuenta dichas interacciones y
hace referencia a la distribución de la especie en su entorno, dada la presencia de
competidores, y por tanto se asume como más estrecho que el nicho fundamental.
Este concepto nos lleva a la idea de la partición de nicho, el mecanismo que permite
la coexistencia entre especies que habitan el mismo biotopo13 (Rosenzweig 1981).
13 Un biotopo es un área física con condiciones ambientales uniformes dónde viven un conjunto específicodeplantasyanimales.
(a) (b)E1
E3 E2
27
Introducción
Como resultado de la exclusión competitiva (Gause 1934), dos o más especies que
viven en la misma zona y tienen requisitos similares y utilizan los mismos factores
limitantes, puedenmodificar el uso de recursos (al menos una de las especies,
Gause 1934, Rosenzweig 1981), aunque esta idea está sujeta a controversia en la
literatura (Araujo & Guisan 2006). El uso de los conceptos nicho fundamental y
nicho realizado es frecuentemente confuso en la literatura (Soberón 2007) y su
utilidad ampliamente debatida (Araujo & Guisan 2006).
Enesta tesisheutilizadoel enfoquedeHutchinsonparadefinir elnicho
ecológico, ya que está íntimamente ligado al concepto de hábitat (cualquiera que
sealadefinicióndehábitat,ambassebasanenlarelaciónentreunaespecieylas
características del medio ambiente). Aunque los conceptos de nicho ecológico y
hábitat están relacionados con el espacio ecológico, a menudo están relacionados
conel espaciogeográfico (Figura3,Calenge2005,Araujo&Guisan2006).De
hecho,elestudiodelaubicacióndelasespeciesensuespaciogeográficopermite
laidentificacióndesuspropiedadesecológicas,ylaasociacióndelaspropiedades
ecológicas con factores espacialmente explícitos da lugar a la distribución
potencial de la especie (Araujo & Guisan 2006). En el marco de este concepto
se han desarrollado diversos métodos en relación con el objetivo de relacionar la
distribución de las especies con su medio ambiente (Hirzel et al. 2002, Calenge et
al. 2005, Basille et al. 2008, Calenge & Basille 2008, Calenge et al. 2008). Algunos
de estos métodos utilizan dos interesantes propiedades del nicho: la marginalidad
y especialización de las especies. La marginalidad es la posición de las especies en
losgradientesambientalesdisponibles.Porlotanto,serefierealaexcentricidaddel
nicho en comparación con el gradiente de componentes ambientales (Calenge et al.
2005). Por lo tanto, una especie marginal se encontrará en condiciones ambientales
28
Introducción
más atípicas (valores extremos del gradiente de una variable), mientras que las
especies no marginales usarán condiciones ambientales medias. La especialización
es la anchura del nicho, es decir, el grado de tolerancia de la especie al gradiente
ambiental. Cuanto más grande sea el nicho, mayor tolerancia presentará la especie,
mientras que cuánto más estrecho sea, más especializada estará la especie en el
uso de ciertos recursos. Estos conceptos son particularmente útiles para describir y
cuantificarlarelaciónentreunaespecieyelmedioambientedisponibleparaella.
En los últimos años, se han desarrollado numerosos análisis para la estimación del
nicho ecológico (Guisan & Zimmermann 2000, Calenge & Basille 2008). Aunque
este concepto ha sido desarrollado y utilizado a nivel de especie o población,
también puede ser generalizado a nivel individual.
El estudio de la selección de hábitat
Los estudios de selección de hábitat para identificar las características
ambientales que selecciona una especie, en el supuesto de que estas características
se han seleccionado debido a que proporcionan las mejores condiciones para la
supervivencia y la reproducción (Thomas & Taylor 2006), se abordan cada vez más
desde diferentes disciplinas (Evolución, Ecología y Conservación). La selección de
hábitat generalmente se investiga utilizando datos sobre el uso del espacio de una
especie dada. Las características del hábitat utilizado por la especie se comparan con
las de las zonas no utilizadas y más comúnmente con las de áreas que se consideran
como disponibles para la especie (Thomas & Taylor 1990, Manly et al. 2002). Se
dice que un hábitat es seleccionado cuando la proporción utilizada por los animales
es mayor que la proporción disponible. Por el contrario, se dice que un hábitat se
‘evita’ cuando la proporción de uso es menor que la proporción disponible. Sin
29
Introducción
embargo, como se indicó anteriormente las interpretaciones resultantes de estos
comparaciones requieren cierta precaución, porque no son una medida directa de
la calidad del hábitat (lo que requeriría algún tipo de medida relacionada con el
fitness14). Sin embargo, aunque la densidad de individuos podría ser un indicador
pobre de la calidad del hábitat en algunas condiciones (Van Horne 1983), la mayoría
de las veces es un buen indicador de la idoneidad (calidad) de un área en particular.
Escalas espacio-temporales de investigación
La naturaleza misma de la selección de hábitat es jerárquica. Diferentes
procesos actúan a diferentes escalas espacio-temporales, lo que resulta en una
selección diferencial del hábitat de acuerdo con la escala bajo consideración. Las
características de un área en la que se distribuye una población de una determinada
especie (selección de primer orden podrían no ser congruentes con las características
de los hábitats disponibles dentro del área de campeo de los individuos (selección
de tercer orden), porque los mecanismos implicados no son los mismos. La elección
de la escala de investigación (por ejemplo el rango geográfico de la especie, el
área ocupada por una población, la selección de hábitat individual dentro de las
áreas de campeo, etc) es crucial, ya que las inferencias a partir de los análisis a
una escala particular, están limitada a esa escala (Pendleton et al. 1998). Además,
la importancia de una escala particular puede ser diferente según la especie de
estudio. Para las especies generalistas, por ejemplo, las escalas grandes (escala de
paisaje,porejemplo)podríansermenosimportantesquelasescalasmásfinas(por
ejemplo, la selección de hábitat o recursos dentro del área de campeo) que para las
14aptitud,adecuaciónoeficaciabiológica
30
Introducción
especiesespecialistas,paralasqueelhábitatenelrangogeográficopuedeserde
crucial importancia. Sin embargo, como la mayoría de los procesos ecológicos, la
selección de hábitat a menudo se produce a más de una escala (Levin 1992).
Thomas & Taylor (1990) propusieron diferentes diseños de estudio para
las comparaciones entre hábitats usados y hábitats disponibles (o no usados)
teniendo en cuenta el organismo de estudio (población, individuos) y la escala
de investigación. El diseño de tipo I se utiliza para estudios a nivel de población
(selección de primer orden) cuando no se identifican los individuos. El uso del
hábitat y la disponibilidad de hábitat se miden a nivel de población. Los datos se
supone que son independientes (la presencia de la especie en un sitio particular
nodebeinfluirensupresenciaenotroslugares)yelaccesoalosrecursosesigual
para todos los individuos (deben por lo tanto, seguir una distribución ‘libre ideal’15
(Fretwell & Lucas 1969). Para este diseño se suelen emplear índices de presencia
de las especies (por ejemplo, heces, pelos, rastros).
Los diseños de tipo II, III y IV se utilizan para estudios a nivel individual
(selección de 2º, 3º y 4º orden). Los individuos se identifican (por ejemplo,
utilizando telemetría) y los datos de uso se miden por separado para cada uno de
los individuos. En el diseño de tipo II, la disponibilidad de hábitat es la misma
para todos los individuos (por ejemplo, la composición de áreas de campeo dentro
delrangogeográficodedistribucióndelasespecies),mientrasqueenlosdiseños
de tipo III y IV, la disponibilidad de hábitat se mide de forma independiente para
cada individuo. La disponibilidad de hábitat es constante durante el período de
estudioeneldiseño tipo III,por loque sedefineporunamedida (porejemplo,
15 La teoría de la distribución libre (ideal free distribution; IFD) formula que el número de individuos que se agregan en varios parches es proporcional a la cantidad de recursos disponibles en cada uno y predice que la distribución de los animales entre parches minimizará la competencia por los recursos y maximizará el fitness.
31
Introducción
el área de campeo de un individuo). En el diseño IV, hay un cambio temporal en
la disponibilidad de hábitat para un individuo dado, lo que requiere una serie de
medidas de disponibilidad (una medida por cada localización de un individuo). Este
último diseño fue creado con posterioridad por Erickson et al. (2001) para tener en
cuenta la creciente utilización de los nuevos tipos de datos proporcionados por la
telemetría, que permiten localizaciones frecuentes de los animales (por ejemplo,
una localización cada 30 minutos). En esta tesis empleo un diseño de estudio tipo
I para determinar los patrones de selección de hábitat de especies de carnívoros
silvestres y domésticos en un área protegida.
De lo realmente disponible e importante para los individuos
Medir la disponibilidad de hábitat para una especie requiere tener en cuenta
dosaspectosimportantes:laeleccióndelasvariablesbiológicassignificativaspara
la especie y los límites de la zona que consideraremos como disponible para la
especie en cuestión. La elección de las variables de hábitat a integrar en los análisis
es una tarea difícil (Lennon 1999, Guisan & Zimmermann 2000), y debe basarse
en un conocimiento previo profundo de la especie. Todas las variables de hábitat
consideradas limitantes para la especie deben incluirse. Sin embargo, la elección de
variables a menudo se basa en consideraciones logísticas, ya que algunas variables
son difíciles de medir (Mitchell 2005).
Como se mencionó anteriormente, la elección de la escala es muy importante
yconducealproblemainherentededefinicióndeladisponibilidaddehábitat.En
los estudios de selección de hábitat, la determinación de lo que es disponible para la
especie es un reto, ya que sólo los animales ‘saben’ lo que está realmente disponible.
En teoría, los investigadoresdebemosdefinirobjetivamente ladisponibilidadde
32
Introducción
hábitat, desde el punto de vista de la especie. Esto es crítico para la interpretación
de los análisis, porque al cambiar la disponibilidad cambiará la proporción de cada
uno de los componentes hábitat, y por tanto, la comparación entre la proporción
de uso y disponibilidad de este componente, especialmente si este componente
muestra agregación espacial (Porter & Church 1987). En la práctica, sin embargo, la
disponibilidadsedefineamenudosubjetivamente,debidoaladificultaddeevaluar
la percepción que tiene la especie del medio. Por ejemplo, una zona puede aparecer
disponibleparaunindividuodado,mientrasquelasinteraccionesinterespecíficas
(porejemplo, lapresenciadedepredadores)o interacciones intraespecíficas (por
ejemplo, la defensa del territorio) podría prevenir su uso o su acceso, respectivamente.
Enalgunoscasos,ladefinicióndedisponibilidaddebeserreducidasilacuestiónde
interésserefierealoscomponentesfísicosdelhábitat,osivariablesreferidasasus
congéneres, presas o depredadores deben ser incluidas.
LasescalasdeseleccióndefinidasporJohnson(1980)ayudanareduciresta
subjetividad, ya que tienen una base biológica, pero no la eliminan por completo
(Erickson et al. 2001). Por ejemplo, en la escala de establecimiento del área de
campeoenladistribucióngeográficadeunaespecie,loslímitesdeláreadeestudio
a menudo abarcan el área en la que se distribuye la población.
Características ecológicas de las especies y mecanismos de coexistencia
La selección natural ha inducido la aparición de estrategias más o
menos especializadas entre especies presentando un trade-off evolutivo entre la
especializaciónpararealizarunaspocasactividadeseficazmenteoelgeneralismo
para desarrollar muchas actividades de una manera menos efectiva (Levins 1968).
Algunas especies muestran amplias tolerancias ambientales y presentan una dieta
33
Introducción
muy variada (generalistas de hábitat y dieta), mientras que otras tienen tolerancias
ambientales muy específicas y estrechas y sólo consumen ciertos recursos
(especialistas de hábitat y dieta). Estas dos categorías de especies tienen diferentes
dinámicas poblacionales (Kolasa & Li 2003). Por ejemplo, la variación en la
densidad de población es mayor en los especialistas que en los generalistas (Kolasa
& Li 2003). Del mismo modo, los especialistas de hábitat utilizan unidades de
hábitat más pequeñas anidadas dentro de unidades de hábitat más grandes (Kolasa
& Pickett 1989). Esto tiene otra consecuencia ya que las especies que utilizan
pequeñas unidades de hábitat tienden a tener bajas densidades poblacionales como
consecuencia de la disminución de la eficiencia en la búsqueda de los parches
adecuados y de la mortalidad durante la dispersión (Kolasa & Romanuk 2005).
Así, la disponibilidad de hábitats adecuados parece afectar más a las especies
especialistas que a las más generalistas, las cuales utilizan una gama más amplia
de tipos de hábitats para satisfacer sus necesidades (Munday et al. 1997, Bean et
al. 2002). Obviamente existe un gradiente continuo en el nivel de especialización
de una especie entre el especialismo más extremo (como el caso del lince ibérico;
Lynx pardinus) y el completo generalismo (como el caso del zorro común; Vulpes
vulpes).
Ladefiniciónoperativadeespecializaciónqueusaréenestatesisseráeluso
de un subconjunto relativamente restringido de recursos o hábitats en comparación
con otras especies. Las especies especialistas suelen beneficiarse de ambientes
relativamente homogéneos o estables (en el espacio y/o tiempo) mientras que las
especiesgeneralistassuelenbeneficiarsedeambientesmásheterogéneos(Futuyma
& Moreno 1988, Kassen 2002, Marvier et al. 2004, Ostergard & Ehrlén 2005).
Las características ecológicas particulares de cada especie pueden afectar
34
Introducción
laeficienciadelaidoneidaddelhábitatolosmodelosdedistribucióndeespecies
(Stockwell & Peterson 2002). Por ejemplo, Hepinstall et al. (2002) sugieren que la
amplitud de nicho de las especies es importante porque para las especies generalistas
que utilizan diferentes hábitats se podría predecir su ocurrencia en todas partes
por los métodos de asociación de hábitat, mientras que las especies con nichos
estrechos son más propensas a ser predichas con mayor exactitud. Tsoar et al.
(2007), además, encontró que los rangos de distribución de las especies con nichos
ecológicos estrechos se puede modelar con mayor precisión que los de las especies
más generalistas. Otros autores (por ejemplo Cowley et al. 2000, Hepinstall et al.
2002, Brotons et al. 2004, Hernández et al. 2006, Brotons et al. 2007) también han
señalado que las especies con nichos ecológicos restringidos pueden ser modeladas
con mayor precisión que las especies más generalistas. Brotons et al. (2007) sugiere
que las especies que tienen distribuciones más amplias o utilizan una amplia variedad
de hábitats en un área no deben estar limitadas por los factores predictivos medidos
a la escala a la que se ajustan los modelos. Cowley et al. (2000) también encontró
que los modelos con mayor rendimiento son los de especies sedentarias que tienen
fuertes asociaciones de hábitat y que presentan una distribución generalizada en
esos hábitats.
El mecanismo por el que especies con diferentes historias de vida o
características ecológicas conviven en la misma área ha sido foco de estudio en
ecología. Diversos autores han discutido en detalle como la heterogeneidad espacial
promuevelacoexistenciaentreespeciespertenecientesalmismoniveltrófico(Levin
1974, Yodzis 1978, Hastings 1980). Los primeros modelos teóricos mostraron que
dos especies que comparten recursos no pueden coexistir en un solo parche, pero
pueden hacerlo cuando dos o más parches diferentes están presentes (Levin 1974).
35
Introducción
En la década de 1980 el concepto de que la heterogeneidad espacial promueve la
coexistencia fue estudiada con más detalle en una teoría desarrollada para explicar
la coexistencia entre especies basándose en las teorías de la selección óptima de
los recursos (optimal foraging theory) y la selección de hábitat, conocida como la
teoría Isoleg (Pimm & Rosenzweig 1981, Rosenzweig 1981). Esta teoría explica
cómo dos especies competidoras se distribuyen en hábitats de diferente calidad
segúnsusdensidadesintraeinterespecíficas.Unadelasprincipalesprediccionesde
la teoría isoleg es que la coexistencia se ve favorecida cuando una de las especies
competidoras es un especialista (comportándose por tanto de manera selectiva),
mientras que el otro es un generalista (actuando por tanto de forma oportunista)
(Rosenzweig 1987). Cuando el especialista es también la especie dominante, se
prevéquelacoexistenciaseveafavorecidaenáreasconunadiversidadsuficiente
de hábitats en las que la especie generalista subordinada pueda segregarse de la
especie especialista dominante. Un concepto similar se desarrolló con anterioridad
para los sistemas predador-presa – las presas pueden ‘buscar la seguridad’ frente a
los depredadores en zonas conocidas como refugios de depredación, lo que puede
ser crucial para la persistencia de ambas (Hassell & May 1973). Este principio
se puede aplicar igualmente a la competencia interespecífica y, en un entorno
heterogéneo, las especies con baja capacidad competitiva puede persistir mediante
el uso de refugios donde la competencia se reduce (Durant 1998).
No obstante, en áreas aparentemente homogéneas desde una perspectiva
o escala mayor, los mecanismos de coexistencia que deben desarrollar especies
similares(esdecir,pertenecientesalmismoniveltrófico)quelespermitansegregarse
yportantosubsistirnoresultantanevidentes.Estosmecanismopuedenreflejarse
en patrones diferenciados de uso del hábitat relativamente homogéneo a una escala
36
Introducción
másfina(aescaladeparche)ydebeversefavorecidoporlapresenciadeespecies
con diferentes grado de especialización en el gremio. Conocer el efecto del hábitat
en la coexistencia de las especies puede ser especialmente relevante cuando puede
afectar a las políticas de conservación y gestión, incluso en áreas protegidas.
De cómo simplemente encontrar el hábitat más idóneo puede no ser suficiente
Como ya referimos con anterioridad con las ideas Hutchinsonianas de nicho
fundamental y nicho realizado, la presencia de una especie no indica necesariamente
que el hábitat sea el óptimo debido a la existencia de un trade-off entre la calidad
del hábitat y las interacciones intra e interespecífica que limitan los recursos e
interfierenelaccesoaellos.Así,losmodelosdeseleccióndehábitat‘puros’16 no
siemprereflejanlaidoneidaddelhábitatparaunaespecie,oalmenosparaaquellas
especies para las cuales la presencia de otras especies pertenecientes al mismo nivel
tróficoylasinteraccionesconellaspuedanalterarsuusodelespacioylimitarsu
abundancia o distribución (e.g. Laurenson 1995, Lindström et al. 1995).
Así, bajo condiciones de competencia por el uso de los recursos17, la densidad
de individuos puede ser un indicador pobre de la calidad del hábitat para algunas
especies.
La competencia se suele clasificar tanto en competencia por explotación
como en competencia por interferencia (Park 1962). La competencia por explotación
se produce cuando una especie utiliza un recurso (por ejemplo consume una
16 con modelos de selección de hábitat ‘puros’ hago referencia a modelos en los que no se tienen en cuentalaspotencialesinteraccionescompetitivasexistentesentreespeciespertenecientesalmismoniveltróficoque la especie bajo estudio, sino solamente las características ‘anatómicas’ del paisaje que las rodea.
17entiéndasecomorecursocualquierfuentedelacuallaespecieobtengaunbeneficio,esdecirtantolas características del paisaje como la vegetación o disponibilidad de presas, como el propio espacio en sí mismo.
37
Introducción
presa específica) y con ello reduce la oportunidad de usar esemismo recurso a
otra especie. La competencia por interferencia implica sin embargo interacciones
comportamentales entre las especies como la predación intragremial (Polis et al.
1989) en el caso más extremo o bien mecanismos más sutiles como evitar los
hábitats más usados por el predador principal y mostrar preferencias por hábitats
menos productivos (Harrison et al. 1989, Thurber et al. 1992, Durant 1998, Fedriani
et al. 1999, Fuller & Keith 1981), ajuste de los patrones de actividad para reducir
los encuentros con el predador principal (Litvaitis 1992, Johnson et al. 1996) o
bien formar grupos para competir de una manera más exitosa por los recursos
y/o obtener ventajas antipredadoras (Kruuk 1975, Eaton 1979, Lamprecht 1981,
Gittleman 1989).
El tamaño corporal se considera normalmente el factor más influyente
que determina la dirección y fuerza de la dinámica intragremial (Polis et al. 1989,
Donadio & Buskirk 2006), siendo las especies de tamaño superior capaces de
excluiralasmáspequeñasdelosparchesdehábitatorecursostróficos.
Las especies de carnívoros de pequeño y mediano tamaño a menudo se
ven perjudicados por interacciones agresivas con otros miembros simpátridos de
mayor tamaño del gremio al que pertenecen (Palomares et al. 1996, Crooks & Soule
1999, Palomares & Caro 1999, Fedriani et al. 2000, Donadio & Buskirk 2006). Las
interacciones agresivas o predación intragremial entre los mamíferos carnívoros
es muy frecuente y en algunas especies, además de suponer un cambio en el uso
del hábitat de las especies menos aventajadas competitivamente tiene también un
impacto considerable en las tasas de mortalidad (Ralls & White 1995, Sovada et
al. 1995) y por tanto en las densidades relativas de dichas especies en presencia
de un competidor superior. En esta tesis analizaremos la existencia de potenciales
38
Introducción
interacciones agresivas o predación intragremial entre las especies que conforman
el gremio de mamíferos carnívoros de nuestra área de estudio para determinar cómo
predadoresinferiorespuedenmostrarunarespuestanuméricanegativareflejadaen
sus densidades relativas en presencia de un predador superior.
39
Introducción
OBJETIVOS DE LA TESIS DOCTORAL
El capítulo I está dedicado a consideraciones metodológicas, particularmente
en lo relativo a la necesidad de tener en cuenta variables metodológicas y/o
climáticas en los censos de rastros en sustratos arenosos y cómo su inclusión como
variablesadicionalesenmodelosdeseleccióndehábitatpuedemejorarlafiabilidad
de los resultados obtenidos.
Los objetivos biológicos de esta tesis cubren la selección de hábitat a nivel
de área de campeo de los individuos y la biología de la conservación y se presentan
en los capítulos del II al V.
En el capítulo II, el objetivo es determinar los patrones de selección de
hábitataescalafinadelasdistintasespeciesdecarnívorossilvestresconsiderándolas
en su conjunto. Particularmente el objetivo consiste en determinar cómo especies
con diferentes historias de vida (es decir, grado de especialismo de hábitat y
dieta) pueden convivir en un área estudiada relativamente homogénea a escala de
macrohábitat. Nos centramos en determinar el espacio ecológico ocupado por cada
una de las especies (nicho ecológico) dentro del hiperespacio representado por las
variables ambientales medidas en el área de estudio y en evaluar cómo factores
relacionadas con el tipo de vegetación, la disponibilidad de presas, la estructura del
paisaje o las perturbaciones humanas pueden afectar a los patrones diferenciales de
seleccióndehábitatentrelasdistintasespeciesaunaescalafina.
En el capítulo III pretendemos determinar si la presencia de un predador
superior puede afectar negativamente las densidades relativas de especies inferiores
unavezcontroladoporlaseleccióndehábitataescalafinadecadaespecie;efecto
potencialmente atribuible a la existencia de depredación intragremial.
40
Introducción
En el capítulo IV nos planteamos determinar si las rastros de perros
detectados durante los censos se correspondían con perros domésticos procedentes
de la matriz humanizada circundante al área protegida o se correspondían con perros
asilvestrados residentes en el interior del Parque, así como en evaluar las variables
ambientales y/o relacionadas con la presencia humana que determinan su uso del
espacio. Asimismo, discutimos las implicaciones de los resultados en términos de
gestión y manejo de especies domésticas en el área.
El capítulo V está dedicado a la presencia de gato montés en el Parque
Nacional. Esta especie resulta de especial interés debido a que no se han llevado a
cabo estudios relacionados con su presencia, abundancia y/o selección de hábitat
con anterioridad en el área protegida asumiéndose una especie muy escasa a pesar
de la potencial idoneidad del hábitat para la especie. En este capítulo realizamos
un estudio de fototrampeo para determinar su presencia y estatus de conservación.
Asimismo, discutiremos las posibles razones relacionadas con el actual estado de
la población en el área. Estos resultados sirven además como una herramienta para
ayudar a los gestores en la toma de decisiones en materia de conservación y gestión
de la población de dicha especie.
41
Introducción
ESPECIES & ÁREA DE ESTUDIO
Las especies modelo
Esta tesis se centra en el estudio de un gremio de carnívoros compuesto
por cuatro especies nativas (el lince ibérico Lynx Pardinus, el tejón europeo
Meles meles, el zorro común Vulpes vulpes y el gato montés Felis silvestris) y dos
introducidas en tiempos históricos (la gineta común Genetta genetta y el meloncillo
Herpestes ichneumon) (Figura 4) en un área protegida del suroeste de España; el
Parque Nacional de Doñana.
El zorro común (5-7 Kg.), debido a su marcado generalismo de hábitat y
dieta (Mitchell-Jones et al. 1999, Pita et al. 2009), es la especie más extendida en el
Mediterráneo. Presenta una alta resistencia ecológica gracias a su alta movilidad y
capacidadreproductiva(Blanco1998,Piñero2002)asícomounaaltaflexibilidad
en su ecología espacial; puede ocupar cualquier tipo de hábitat que le ofrezca un
mínimo de refugio y alimento adaptándose a cualquier tipo de cambio (Fedriani
1996, Piñero 2002). Como depredador oportunista, consume una amplia variedad de
recursostróficosysealimentadeacuerdoconladisponibilidaddepresas(Carvalho
& Gomes 2001, Cavallini & Lovari 1991, Delibes-Mateos et al. 2008).
El tejón europeo (7-8 Kg.) es un carnívoro social y territorial con una amplia
distribución en el Paleártico (Revilla & Palomares 2005). De actividad en su mayor
parte nocturna, es considerado comoun generalista trófico por la gran variedad
de recursos disponibles de los que hace uso (lombrices, insectos, frutos, gazapos;
Revilla & Palomares 2002). Pese a que en Europa central muestra preferencias por
los bosques templados y pastos asociados, se ha demostrado que también selecciona
las masas arbustivas como hábitat principal en climas más secos, como es el caso
del suroeste de la Península Ibérica (Revilla et al. 2000). Su presencia parece
42
Introducción
condicionada por la existencia de cobertura vegetal que oculte las madrigueras
(Revilla & Palomares 2005). El tamaño del área de campeo puede variar mucho
entre individuos y en función de la productividad de los hábitats ocupados, siendo
el tamaño medio de 525 hectáreas en el suroeste de la Península Ibérica (Rodríguez
et al. 1996).
El meloncillo (3 Kg.) es un carnívoro diurno procedente de África (Riquelme
Cantal-et al. 2008) cuya distribución en Europa se restringe al suroeste de la
Península Ibérica. Se alimenta en grupos presentando una dieta de amplio espectro
que incluye conejos, roedores, aves, reptiles y carroña (Palomares & Delibes 1991,
Palomares 1993, Zapata et al. 2007). Debido a su carácter diurno, la especie muestra
preferencias por zonas de matorral denso con una alta densidad de presas y refugio
evitando las zonas abiertas, por lo que se le ha considerado en cierto grado como
especialista de hábitat en zonas mediterráneas (Palomares & Delibes 1990, 1993).
El tamaño medio de su área de campeo es de 300 hectáreas (Palomares 1994).
Al igual que en el caso del meloncillo, la gineta (2 Kg.) también en una especie
introducida de África (Riquelme-Cantal et al. 2008). Es un carnívoro nocturno que
se alimenta de pequeños mamíferos, aves, reptiles y artrópodos, aunque la presa
principal puede variar en diferentes zonas de distribución (Virgós et al. 1999). En
nuestra zona de estudio, el Parque Nacional de Doñana, el análisis de la dieta ha
mostradounapreferenciapormicromamíferos,seguidodeaves,insectos,anfibios,
conejos y reptiles (Palomares & Delibes 1991). Aunque es muy adaptable en cuanto
al hábitat se ha descrito su uso de formaciones arboladas con cobertura arbustiva
(Palomares & Delibes 1994, Virgós & Casanovas 1997), así como que evita zonas
abiertas a no ser que exista vegetación de matorral cercana o parches aislados con
árboles y vegetación de sotobosque que puedan actuar como refugios para la especie
43
Introducción
(Rosalino & Santos-Reis 2002, Galantinho & Mira 2009). Debido a que presentan
ciertospatronesde seleccióndehábitatdefinidosen funciónde la zonaycierta
especialización local en la dieta, la gineta se considera entre el típico generalista y
el típico especialista de hábitat y dieta (Virgós, Llorente & Cortés 1999). El tamaño
medio del área de campeo en Doñana de la especie es de 541 hectáreas (Palomares
& Delibes 1994).
El gato montés (3-7 Kg.) es un carnívoro del cual, desde un punto de vista
científico, se ha sabido bastante poco hasta añosmuy recientes.Aunque puede
vivir en una gran variedad de hábitats, tradicionalmente se ha considerado como
una especie típicamente forestal (Guggisberg 1975, Ragni 1978, Blanco 1998). No
obstante, en climas más secos como la cuenca Mediterránea la especie se encuentra
en paisajes constituidos por mosaicos de matorral y pastizales con abundantes cursos
de agua y presas además de una alta cobertura de arbustos a escala de microhábitat
(Lozano et al. 2003). Se considera como un especialista facultativo de dieta; siendo
el conejo de monte (Oryctolagus cuniculus) la presa más abundante en su dieta
cuando está presente, pero consumiendo una alta proporción de roedores cuando los
conejos son escasos o ausentes (Moleon & Gil Sánchez 2003, Lozano et al. 2006,
Sarmento 1996).
Por último, el lince ibérico (9-15 Kg.) es la especie de felino más amenazada
del planeta (UICN 2008); es endémico de la Península Ibérica (Rodríguez & Delibes
1992, Ferreras et al. 2010) y en la actualidad existe únicamente en dos poblaciones
estables y reproductoras: Doñana y Sierra Morena oriental (Ferreras et al. 2010).
Ellinceibéricoesunespecialistatrófico,estrictamentedependientedelconejode
monte (Delibes et al. 2000). La densidad de conejos determina la densidad de linces
(Palomares 2001, Palomares et al. 2001). Además, asociado entre otras cosas a su
44
Introducción
sistema de caza y a la distribución natural del conejo, el lince ibérico también es un
especialista de hábitat, estando asociado al matorral mediterráneo (Palomares et al. 1991,
2000, Palomares 2001). La combinación de zonas de matorral denso y de extensiones
herbáceas o maquis abiertos que permitan la caza y el desplazamiento parece esencial
(Delibes et al. 2000). Esta especie constituye un claro ejemplo de contracción rango
extremo en las últimas décadas. Hay evidencias que indican un descenso continuo de la
especie debido al agotamiento severo de su principal presa por enfermedades (mixomatosis
y enfermedad hemorrágica vírica (Villafuerte et al. 1995, 1997) y el exceso de caza
(Rodríguez & Delibes 1992). La reducción y fragmentación del matorral mediterráneo
del que tanto el lince como su presa principal dependen, también ha causado un impacto
negativo sobre la especie (Rodriguez & Delibes 1992). Hoy en día, las poblaciones de
lince ibérico en declive persisten bajo condiciones de seria presión humana, incluyendo
reforestaciones con especies no nativas y rozas de matorral, persecución directa de la
especie, expansión de la red vial, construcción de presas y la urbanización del medio
natural (Rodriguez & Delibes 2004). El volumen creciente de tráfico es también el
responsable de las altas tasas de mortandad adicionales no naturales en el lince ibérico;
particularmente en los alrededores del Parque Nacional de Doñana (Ferreras et al. 1992).
De manera adicional, recientes estudios han demostrado que el deterioro concomitante
tantodelosatributosdemográficosdelaespeciecomodelascaracterísticasgenéticasde
la misma (es decir, tasas de mortalidad no traumática, tamaño de camada, supervivencia
de crías, edad de adquisición de territorio, proporción de sexos, tasas de endogamia y
diversidad genética) es consistente con un vórtice de extinción, y que esa coocurrencia,
conosininteracción,deldeteriorodemográficoygenéticopodríaexplicarlafaltadeéxito
de los esfuerzos de conservación de la población de lince ibérico o su lenta recuperación
en nuestra área de estudio (Palomares et al. 2012).
45
Introducción
Figura 4. Especies de estudio en el Parque Nacional de Doñana; (a) tejón europeo (Meles meles),
(b) zorro común (Vulpes vulpes), (c) gato montés (Felis silvestris), (d) lince ibérico (Lynx pardinus),
(e) gineta común (Genetta genetta) y (d) meloncillo (Herpestes ichneumon).
46
Introducción
Área de estudio
El nombre de Doñana surge para designar el conjunto de cotos de caza
mayor con paisaje de matorral y bosque situado en el entorno de las marismas del
río Guadalquivir en el suroeste de España, entre las provincias de Huelva y Sevilla
(Figura 5). El interés despertado por la fauna de este enclave conlleva la creación
de un área protegida en 1964, así como un centro de investigación, la Estación
Biológica de Doñana. En 1969 el área bajo protección aumenta hasta las 35.000
ha con la creación del Parque Nacional de Doñana. En 1982 el área protegida ve
aumentadadenuevosusuperficieconlacreacióndelParqueNaturaldeDoñanaen
el entorno del Parque Nacional (García-Novo & Martín-Cabrera 2005).
En la actualidad el Parque Nacional ocupa una extensión de 550 Km2
El área está designada como Reserva de la Biosfera (UNESCO), Humedal de
Importancia Internacional (Ramsar), Zona de Especial Protección para las Aves
Figura 5. Imagen general del área de estudio Parque Nacional de Doñana y su localización en el suroeste de España.
47
Introducción
(ZEPA - Natura 2000) y Patrimonio de la Humanidad (UNESCO). El clima en
Doñana es Mediterráneo sub-húmedo, con inviernos lluviosos y veranos secos. El
paisaje de Doñana ha sido transformado por la acción del hombre, con un mayor
alcance de estas transformaciones a lo largo del siglo XX. Como en muchos otros
humedales europeos, desde el siglo XIX se observan intentos de desecación y
puesta en cultivo de las marismas de Doñana. Sin embargo, la transformación en el
caso de Doñana fue particularmente lenta hasta mediados del siglo XX. Aunque las
primeras repoblaciones con pino piñonero en Doñana datan del siglo XIX, destacan
las llevadas a cabo con esta especie en la década de 1950, seguida de repoblaciones
de eucaliptos en la década de los 70 (García-Novo & Martín-Cabrera 2005). Así, el
alcornoque, el enebro y la sabina han sido sustituidos en muchas áreas de Doñana por el
pino y el eucalipto. El área ocupada por las marismas de agua dulce y salada tiene una
extensión actual de 30.000 ha (Enggass 1968, García-Novo & Martín-Cabrera 2005)
(Figura 6), suponiendo la quinta parte de las marismas naturales que aún subsisten en
España (García-Novo & Martín-Cabrera 2005). Además, el Parque Nacional presenta un
sistema de lagunas temporales y permanentes que consta de más de 3000 cuerpos de agua
en periodos de máxima inundación (Gómez-Rodríguez 2009). Estas zonas inundadas
constituyen uno de los hábitats de hibernación y cría más utilizados por miles de aves
europeas y africanas.
Figura 6. Vista aérea de la marisma de Doñana desde la denominada Torre del Palacio en época otoñal antes de las primeras lluvias.
Figura 5. Imagen general del área de estudio Parque Nacional de Doñana y su localización en el suroeste de España.
48
Introducción
49
Introducción
50
Introducción
Actualmente Doñana está formada por un mosaico de ecosistemas (playa, dunas,
cotos, marisma, matorral mediterráneo) (Figura 7). El monte o los cotos (Figura 8) es
el ecosistema más estable del Parque. Se localiza en su zona oeste y esta cubierto
deunespesomatorralmediterráneoqueocupamásdel50%de lasuperficiedel
área protegida y que según las especies dominantes, se denomina localmente como
‘monte blanco’ (jaras, jaguarzos, tomillos y romeros) o ‘monte negro’ (brecinas,
brezos, tojos y zarzas) (Figura 5). Este paisaje de monte mediterráneo constituye un
ambiente bastante homogéneo a escala de macrohábitat.
El tren de dunas móviles de Doñana, que avanzan hasta 6 metros por año,
está separado por depresiones intermedias con vegetación de matorral y pino
piñonero, que se denominan corrales, y se sitúa en la zona sur occidental del Parque
Nacional. También es de destacar el ecosistema conocido localmente como ‘La
Figura 7. Mosaico de ecosistemas del Parque Nacional de Doñana. (a) monte blanco, (b) monte
negro, (c) pastizal en La Vera de Doñana, (d) laguna temporal inundada en época invernal, (e)
pastizal en zona inundable desecada durante la época estival, (f) zonas de repoblación de pinos
piñoneros, (g) eucaliptar, (h) marisma inundada en época invernal, (i) inicio del tren de dunas
móviles, (j) pinares en corrales entre dunas, (k) dunas móviles (cerro de Los Ánsares de Doñana),
(l) vista de pinares de repoblación desde el inicio del tren de dunas móviles, (m) vista general de
zona de sabinares y matorral mediterráneo (n) vista general de La Vera de Doñana con pastizal y
alcornoques remanentes del bosque primigenio de Doñana.
Figura 8. Vista general de la zona de cotos o matorral mediterráneo en el Parque Nacional de Doñana. El matorral predominante en la imagen se corresponde con el denominado monte blanco (aulagas, jaguarzos y romeros como especies predominantes) aunque también se puede apreciar algunas manchas dispersas de monte negro (dominadas por brezales) (ver texto).
51
Introducción
Vera’, uno de los biotopos con más biodiversidad de Doñana (Figura 8).
Es un terreno de pastizales que constituye la zona de contacto entre la
marisma y el monte, y recorre el Parque Nacional de norte a sur. La capa freática
estámuypróximaalasuperficie,demodoqueformanumerosaslagunastemporales
en la estación húmeda. En la seca, sin embargo, solo aflora en los lugaresmás
deprimidos, generalmente ahondados ex profeso para proporcionar agua a la fauna
(zacallones).
Para dar una idea de la enorme riqueza de este espacio, baste mencionar que en
Doñana han sido avistadas 400 especies de aves y en lo que respecta a los demás
vertebrados,seencuentran29especiesdemamíferos,19dereptiles,12deanfibios
y 7 de peces, a las que añadir otras 60 del Estuario del Guadalquivir (García-
Novo & Martín-Cabrera 2005). Estas cifras, ciertamente altas para España, son
excepcionales para el subcontinente europeo.
REFERENCIASAraújo MB, Guisan A (2006) Five (or so) challenges for species distribution modelling. Journal of
Biogeography 33 (10):1677-1688.
Banks SC, Finlayson GR, Lawson SJ, Lindenmayer DB, Paetkau D, Ward SJ, Taylor AC (2005) The
effects of habitat fragmentation due to forestry plantation establishment on the demography
and genetic variation of a marsupial carnivore, Antechinus agilis. Biological Conservation
122 (4):581-597.
Barnosky AD, Hadly EA, Bascompte J et al (2012) Approaching a state shift in Earth’s
biosphere. Nature 486 (7401):52-58.
Basille M, Calenge C, Marboutin E, Andersen R, Gaillard JM (2008) Assessing habitat selection
usingmultivariatestatistics:Somerefinementsoftheecological-nichefactoranalysis.Ecol
Model 211 (1-2):233-240.
Bean K, Jones GP, Caley MJ (2002) Relationships among distribution, abundance and microhabitat
52
Introducción
specialisation inaguildofcoral reef triggerfish (familyBalistidae).MarEcol-ProgSer
233:263-272.
Berger KM, Gese EM, Berger J (2008) Indirect effects and traditional trophic cascades: A test
involving wolves, coyotes, and pronghorn. Ecology 89 (3):818-828.
Blanco JC (1998). Mamíferos de España. Geoplaneta, Barcelona.
Brashares JS, Prugh LR, Stoner CJ, Epps CW (2010) Ecological and conservation implications
of mesopredator release. In Terborgh J, Estes JA, eds. Trophic Cascades. Island Press.
Forthcoming.
Brooks TM, Mittermeier RA, Mittermeier CG, da Fonseca GAB, Rylands AB, Konstant WR, Flick
P,PilgrimJ,OldfieldS,MaginG,Hilton-TaylorC(2002)Habitatlossandextinctioninthe
hotspots of biodiversity. Conserv Biol 16 (4):909-923.
Brotons L, Herrando S, Pla M (2007) Updating bird species distribution at large spatial scales:
applications of habitat modelling to data from long-term monitoring programs. Diversity
and Distributions 13 (3):276-288.
Brotons L, Thuiller W, Araujo MB, Hirzel AH (2004) Presence-absence versus presence-only
modelling methods for predicting bird habitat suitability. Ecography 27 (4):437-448.
Burkey TV, Reed DH (2006) The effects of habitat fragmentation on extinction risk: mechanisms
and synthesis. Songklanakarin Journal of Science and Technology 28 (1):9-37.
Calenge C (2005) Des outils statistiques pour l’analyse des semis de points dans l’espace écologique.
Université Claude Bernard Lyon 1.
Calenge C, Basille M (2008) A general framework for the statistical exploration of the ecological
niche. Journal of Theoretical Biology, 252: 674–685.
Calenge C, Darmon G, Basille M, Loison A, Jullien JM (2008) The factorial decomposition of the
Mahalanobis distances in habitat selection studies. Ecology 89 (2):555-566.
Calenge C, Dufour AB, Maillard D (2005) K-select analysis: a new method to analyse habitat
selection in radio-tracking studies. Ecol Model 186 (2):143-153.
Cardillo M, Purvis A, Sechrest W, Gittleman JL, Bielby J, Mace GM (2004) Human population
density and extinction risk in the world’s carnivores. PLoS Biology 2 (7):909-914.
Carvalho JC, Gomes P (2001) Food habits and trophic niche overlap of the red fox, european wild
cat and common genet in the Peneda-Geres National Park. Galemys, 13, 39-48.
53
Introducción
CavalliniP,LovariS(1991)Environmental-FactorsInfluencingTheUseofHabitatInTheRedFox,
Vulpes-Vulpes. J Zool 223:323-339.
Ceballos G, Ehrlich PR (2002) Mammal population losses and the extinction crisis. Science 296
(5569):904-907.
Cezilly F, Benhamou S (1996) Optimal foraging strategies: A review. Rev Ecol-Terre Vie 51 (1):43-86.
ConoverM(2002)Resolvinghuman-wildlifeconflicts:thescienceofwildlifedamagemanagement.
Resolvinghuman-wildlifeconflicts: thescienceofwildlifedamagemanagement.Lewis
Publishers.
Corsi F, de Leeuw J, Skidmore A (2000) Modeling species distribution with GIS. Research techniques
in animal ecology: controversies and consequences. Columbia University Press.
Cowley MJR, Wilson RJ, Leon-Cortes JL, Gutierrez D, Bulman CR, Thomas CD (2000) Habitat-
basedstatisticalmodelsforpredictingthespatialdistributionofbutterfliesandday-flying
moths in a fragmented landscape. Journal of Applied Ecology 37:60-72.
Crooks KR, Soule ME (1999) Mesopredator release and avifaunal extinctions in a fragmented
system. Nature 400 (6744):563-566.
Delibes M, Rodríguez A, Ferreras P (2000) Action plan for the conservation of the Iberian lynx in
Europe (Lynx pardinus). Council of Europe Publishing, Strasbourg.
Delibes-Mateos M, de Simon JF, Villafuerte R, Ferreras P (2008) Feeding responses of the red
fox (Vulpes vulpes) to different wild rabbit (Oryctolagus cuniculus) densities: a regional
approach. Eur J Wildl Res 54 (1):71-78.
DonadioE,BuskirkSW(2006)Diet,morphology,andinterspecifickillingincarnivora.AmNat
167 (4):524-536.
Duffy JE, Cardinale BJ, France KE, McIntyre PB, Thebault E, Loreau M (2007) The functional role
of biodiversity in ecosystems: incorporating trophic complexity. Ecol Lett 10 (6):522-538.
Durant SM (1998) Competition refuges and coexistence: an example from Serengeti carnivores. J
Anim Ecol 67 (3):370-386.
Eaton RL (1979) Interference Competition among Carnivores - Model For The Evolution of Social-
Behavior. Carnivore 2 (MAR):9-16.
Elton G (1927) Animal Ecology. Animal Ecology. London (Sidgwick & Jackson).
Enggass PM (1968) Land reclamation and resettlement in the Guadalquivir Delta-Las Marismas.
54
Introducción
Economic Geography 44, 125-143.
Entwistle AC (2000) Flagships for future? Oryx 34: 239-240.
Entwistle AC, PJ Stephenson (2000) Small mammals & the conservation agenda. Pages 119-140
in A. Entwistle & N. Dunston, editors. Priorities for the Conservation of Mammalian
Diversity: Has the P&a Had its Day? Cambridge University Press, Cambridge, UK.
Erickson WP, McDonald TL, Gerow KG, Howlin S, Kern JW (2001) Statistical issues in resource selection
studies with radio-marked animals. Radio tracking and animal populations. Academic Press.
Estes JA, Palmisan JF (1974) Sea Otters - Their Role in Structuring Nearshore Communities.
Science 185 (4156):1058-1060.
Fahrig L (2003) Effects of habitat fragmentation on biodiversity. Annu Rev Ecol Evol Syst 34:487-515.
Fedriani JM (1996) Dieta anual del zorro, Vulpes vulpes, en dos hábitats del Parque Nacional de
Doñana. Doñana Acta Vertebrata, 23, 143-152.
Fedriani JM, Fuller TK, Sauvajot RM, York EC (2000) Competition and intraguild predation among
three sympatric carnivores. Oecologia 125 (2):258-270.
Fedriani JM, Palomares F, Delibes M (1999) Niche relations among three sympatric Mediterranean
carnivores. Oecologia 121 (1):138-148.
Fernandez N, Delibes M, Palomares F (2007) Habitat-related heterogeneity in breeding in a
metapopulation of the Iberian lynx. Ecography 30 (3):431-439.
Ferraras P, Rodriguez A, Palomares F, Delibes M (2010) Iberian lynx: the uncertain future of a
critically endangered cat. Biology and conservation of wild felids. Oxford University Press.
Ferreras P, Aldama JJ, Beltran JF, Delibes M (1992) Rates And Causes of Mortality in a Fragmented Population
of Iberian Lynx Felis-Pardina Temminck, 1824. Biological Conservation 61 (3):197-202.
Fretwell SD, Lucas HL (1969) On territorial behavior and other factors influencing habitat
distribution in birds. 1. Theoretical development. Acta Biotheoretica.
Fuller TK, Keith LB (1981) Woodland Caribou Population-Dynamics in Northeastern Alberta.
Journal of Wildlife Management 45 (1):197-213.
Futuyma DJ, Moreno G (1988) The Evolution of Ecological Specialization. Annu Rev Ecol Syst
19:207-233.
GalantinhoA,MiraA (2009)The influence of human, livestock, and ecological features on the
occurrence of genet (Genetta genetta): a case study on Mediterranean farmland. Ecol Res
55
Introducción
24 (3):671-685.
García-novo, F. y C. Martín-Cabrera -2005- Doñana: Agua y Biosfera. Doñana 2005, Confederación
hidrográficadelGuadalquivir,MinisteriodeMedioAmbiente.Madrid.
Gause GF (1934) The struggle for existence. The struggle for existence. (Williams & Wilkins).
Gehrt SD, Prange S (2007) Interference competition between coyotes and raccoons: a test of the
mesopredator release hypothesis. Behav Ecol 18 (1):204-214.
Ginsberg JR (2001) Setting Priorities for Carnivore Conservation: What Makes Carnivores Different? In:
Macdonald D, Gittleman J, Wayne R (Eds) Carnivore Conservation. Cambridge University Press.
Gittleman JL (1989) Carnivore group living: comparative trends. Carnivore behavior, ecology, and
evolution. Chapman & Hall, London & Cornell University Press, New York.
Gómez-RodríguezC (2009)Condicionantes ecológicos de la distribución de anfibios en el Parque
Nacional de Doñana. Tesis Doctoral, Departamento de Biología Animal, Ecología,
Parasitología, Edafología y Química Agrícola, Universidad de Salamanca, Salamanca, España.
Grinnell J (1917) The Niche-Relationships of the California Thrasher. The Auk 34 (4):427-433.
Guisan A, Zimmermann NE (2000) Predictive habitat distribution models in ecology. Ecol Model
135 (2-3):147-186.
Guggisberg CAW (1975). Wild cats of the world. David & Charles, Newton Abbot, London.
Hall LS, Krausman PR, Morrison ML (1997) The habitat concept and a plea for standard terminology.
Wildl Soc Bull 25 (1):173-182.
Harrison DJ, Bissonette JA, Sherburne JA (1989) Spatial Relationships Between Coyotes and Red
Foxes in Eastern Maine. Journal of Wildlife Management 53 (1):181-185.
Hassell MP, May RM (1973) Stability in Insect Host-Parasite Models. J Anim Ecol 42 (3):693-726.
Hastings A (1980) Disturbance, Coexistence, History, And Competition For Space. Theor Popul
Biol 18 (3):363-373.
Hepinstall JA, Krohn WB, Sader SA (2002) Effects of niche width on the performance and agreement
of avian habitat models. Predicting Species Occurrences: Issues of Accuracy and Scale.
Island Press, Washington.
Hernandez PA, Graham CH, Master LL, Albert DL (2006) The effect of sample size and species
characteristics on performance of different species distribution modeling methods.
Ecography 29 (5):773-785.
56
Introducción
Heywood, V.H. (1995) Global biodiversity assessment. Cambridge University Press, Cambridge.
Hirzel AH, Hausser J, Chessel D, Perrin N (2002) Ecological-niche factor analysis: How to compute
habitat-suitability maps without absence data? Ecology 83 (7):2027-2036.
Hoek WZ (2008) The Last Glacial-Interglacial Transition. Episodes 31 (2):226-229.
Hoekstra JM, Boucher TM, Ricketts TH, Roberts C (2005) Confronting a biome crisis: global
disparities of habitat loss and protection. Ecol Lett 8 (1):23-29.
Houghton, JT, Y Ding, DJ Griggs, M Noguer, PJ Van der Linden, D Xiaosu, Eds., 2001: Climate
Change 2001: The Scientific Basis: Contributions of Working group I to the Third
Assessment Report of the Intergovernmental Panel on Climate Change.Cambridge
University Press, 881 pp.
Hutchinson GE (1957). Concluding remarks. Cold Spring Harb. Symp. Quant. Biol. 22, 415–427.
InskipC,ZimmermannA(2009)ReviewHuman-felidconflict:areviewofpatternsandpriorities
worldwide. Oryx 43 (1):18-34.
Johnson DH (1980) The Comparison of Usage and Availability Measurements for Evaluating
Resource Preference. Ecology 61 (1):65-71.
Johnson WE, Fuller TK, Franklin WL (1996) Sympatry in canids: a review and assessment. Carnivore
behaviour, ecology and evolution, volume 2. Cornell University Press.
Karanth KU, Chellam R (2009) Carnivore conservation at the crossroads. Oryx 43 (1):1-2.
Kassen R (2002) The experimental evolution of specialists, generalists, and the maintenance of
diversity. J Evol Biol 15 (2):173-190.
Kearney M (2006) Habitat, environment and niche: what are we modelling? Oikos 115 (1):186-191.
Kerley LL, Goodrich JM, Miquelle DG, Smirnov EN, Quigley HB, Hornocker NG (2002) Effects of
roads and human disturbance on Amur tigers. Conserv Biol 16 (1):97-108.
Kolasa J, Li BL (2003) Removing the confounding effect of habitat specialization reveals
the stabilizing contribution of diversity to species variability. Proc R Soc B-Biol Sci
270:S198-S201.
Kolasa J, Pickett STA (1989) Ecological-Systems and the concept of Biological Organization. Proc
Natl Acad Sci USA 86 (22):8837-8841.
Kolasa J, Romanuk TN (2005) Assembly of unequals in the unequal world of a rock pool
metacommunity. Metacommunities: spatial dynamics and ecological communities.
57
Introducción
University of Chicago Press.
Kruuk H (1975) Functional aspects of social hunting by carnivores. Function and evolution in
behaviour. Essays in honour of Professor Niko Tinbergen, F.R.S. i-xxxi, l-393. Clarendon
Press, Oxford.
Lamprecht J (1981) The Function of Social Hunting in Larger Terrestrial Carnivores. Mammal
Review 11 (4):169-179.
Laurenson MK (1995) Implications of high offspring mortality for cheetah population dynamics.
Serengeti 2: dynamics, management and conservation of an ecosystem. University of
Chicago Press.
Lennon JJ (1999) Resource selection functions: taking space seriously? Trends Ecol Evol 14
(10):399-400.
Lenton TM, Livina VN, Dakos V, Scheffer M (2012) Climate bifurcation during the last deglaciation?
Clim Past 8 (4):1127-1139.
Levin SA (1974) Dispersion and Population Interactions. Am Nat 108 (960):207-228.
Levin SA (1992) The Problem of Pattern and Scale in Ecology. Ecology 73 (6):1943-1967.
Levins R (1968) Evolution in Changing Environments. Princeton University Press, Princeton, NJ.
Lindstrom ER, Brainerd SM, Helldin JO, Overskaug K (1995) Pine Marten Red Fox Interactions - A
Case of Intraguild Predation. Ann Zool Fenn 32 (1):123-130
Litvaitis JA (1992) Niche relations between coyotes and sympatric Carnivora. Pages 73–86 in
A. H. Boer, ed. Ecology and management of the eastern coyote.Wildlife Research Unit,
University of Brunswick, Canada.
Lozano J, Moleon M, Virgos E (2006) Biogeographical patterns in the diet of the wildcat,
Felis silvestris Schreber, in Eurasia: factors affecting the trophic diversity. Journal of
Biogeography 33 (6):1076-1085.
Lozano J, Virgós E, Malo AF, Huertas DL, Casanovas JG (2003) Importance of scrub–pastureland
mosaics for wild-living cats occurrence in a Mediterranean area: implications for the
conservation of the wildcat (Felis silvestris). Biodiversity & Conservation 12 (5):921-935.
Manly BFJ, McDonald LL, Thomas DL, McDonald TL, Erickson WP (2002) Resource selection by
animals:statisticaldesignandanalysisforfieldstudies.Secondedition.Resourceselection
by animals: statistical design and analysis for field studies. Second edition. Kluwer
58
Introducción
Academic Publishers.
Marvier M, Kareiva P, Neubert MG (2004) Habitat destruction, fragmentation, and disturbance promote
invasion by habitat generalists in a multispecies metapopulation. Risk Anal 24 (4):869-878.
May RM (1992) How many species inhabit the earth? Sci Amer 10: 18–24.
McLoughlin PD, Gaillard JM, Boyce MS, Bonenfant C, Messier F, Duncan P, Delorme D, Van
Moorter B, Said S, Klein F (2007) Lifetime reproductive success and composition of the
home range in a large herbivore. Ecology 88 (12):3192-3201.
Messier F (1991)The Significance of Limiting andRegulating Factors onTheDemography of
Moose and White-Tailed Deer. J Anim Ecol 60 (2):377-393.
Miller B, Dugelby B, Foreman D, Martinez del Rio C, Noss R, Phillips M, Reading R, Soule ME,
Terborgh J, Willcox L (2001) The importance of large carnivores to healthy ecosystems.
Endangered Species Update 18 (5):202-210.
Mitchell SC (2005) How useful is the concept of habitat? a critique. Oikos 110 (3):634-638.
Mitchell-Jones AJ, Amori G, Bogdanowicz W et al. (1999) The Atlas of European Mammals.
Academic Press, London, UK.
Moilanen A, Smith AT, Hanski I (1998) Long-term dynamics in a metapopulation of the American
pika. Am Nat 152 (4):530-542.
Moleon M, Gil-Sanchez JM (2003) Food habits of the wildcat (Felis silvestris) in a peculiar habitat:
the Mediterranean high mountain. J Zool 260:17-22.
Morris DW (2003) Toward an ecological synthesis: a case for habitat selection. Oecologia 136 (1):1-13.
Morrison ML, Marcot BG, Mannan RW (2006) Wildlife-habitat relationships: concepts and
applications. Third edition. Wildlife-habitat relationships: concepts and applications. Third
edition. Island Press.
Munday PL, Jones GP, Caley MJ (1997) Habitat specialisation and the distribution and abundance of
coral-dwelling gobies. Mar Ecol-Prog Ser 152 (1-3):227-239.
Naves J, Wiegand T, Revilla E, Delibes M (2003) Endangered species constrained by natural and
human factors: The case of brown bears in northern Spain. Conserv Biol 17 (5):1276-1289.
Ostergard H, Ehrlen J (2005) Among population variation in specialist and generalist seed predation
- the importance of host plant distribution, alternative hosts and environmental variation.
Oikos 111 (1):39-46.
59
Introducción
Pacala S, Roughgarden J (1984) Control Of Arthropod Abundance By Anolis Lizards On St-
Eustatius (Neth Antilles). Oecologia 64 (2):160-162.
Paine RT (1969) A Note on Trophic Complexity and Community Stability. Am Nat 103 (929):91-&.
Palomares F, Delibes M (1990) Habiatat preference of large grey mongoose, Herpestes ichneumon,
in Spain. Acta Theriologica, 35, 1-6.
Palomares F (1993) Opportunistic Feeding of The Egyptian Mongoose, Herpestes-Ichneumon, (L)
in Southwestern Spain. Rev Ecol-Terre Vie 48 (3):295-304.
Palomares F (1994) Site Fidelity and Effects of Body-Mass on Home-Range Size of Egyptian
Mongooses. Can J Zool-Rev Can Zool 72 (3):465-469.
Palomares F (2001) Vegetation structure and prey abundance requirements of the Iberian lynx:
implications for the design of reserves and corridors. Journal of Applied Ecology 38 (1):9-18.
Palomares F, CaroTM (1999) Interspecific killing amongmammalian carnivores.AmNat 153
(5):492-508.
Palomares F, Delibes M (1991) Dieta del meloncillo, Herpestes ichneumon, en Coto del Rey, Norte
delParqueNacionaldeDonana.Dofiana,ActaVertebrata,18,187-194.
Palomares F, Delibes M (1991) Alimentación del meloncillo Herpestes ichneumon y de la
gineta Genetta genetta en la Reserva Biológica de Doñana, SO de la península
Ibérica. Doñana Acta Vertebrata, 18 (1): 5-20.
Palomares F, Delibes M (1993) Resting Ecology and Behavior of Egyptian Mongooses (Herpestes-
Ichneumon) in Southwestern Spain. J Zool 230:557-566.
Palomares F, Delibes M (1994) Spatiotemporal Ecology and Behavior of European Genets in
Southwestern Spain. J Mammal 75 (3):714-724.
Palomares F, Delibes M, Revilla E, Calzada J, Fedriani JM (2001) Spatial ecology of Iberian lynx and
abundance of European rabbits in south-western Spain. Wildlife Monographs 148, 1–36.
Palomares F, Ferreras P, Fedriani JM, Delibes M (1996) Spatial relationships between Iberian lynx and
other carnivores in an area of south-western Spain. Journal of Applied Ecology 33 (1):5-13.
Palomares F, Ferreras P, Travaini A, Delibes M (1998) Co-existence between Iberian lynx and
Egyptian mongooses: estimating interaction strength by structural equation modelling and
testing by an observational study. J Anim Ecol 67 (6):967-978.
Palomares F, Godoy JA, Lopez-Bao JV, Rodriguez A, Roques S, Casas-Marce M, Revilla E, Delibes
60
Introducción
M (2012) Possible Extinction Vortex for a Population of Iberian Lynx on the Verge of
Extirpation. Conserv Biol 26 (4):689-697.
PalomaresF,RodriguezA,LaffitteR,DelibesM(1991)TheStatusandDistributionofTheIberian
Lynx Felis-Pardina (Temminck) in Coto Donana Area, Sw Spain. Biological Conservation
57 (2):159-169.
Park T (1962) Beetles, Competition, and Populations - Intricate Ecological Phenomenon is Brought
into Laboratory and Studied as an Experimental Model. Science 138 (3548):1369.
Pendleton GW, Titus K, DeGayner E, Flatten CJ, Lowell RE (1998) Compositional analysis and
GIS for study of habitat selection by goshawks in southeast Alaska. Journal of Agricultural
Biological and Environmental Statistics 3 (3):280-295.
Pimm SL, Rosenzweig ML (1981) Competitors and Habitat Use. Oikos 37 (1):1-6.
Pinero JR (2002) Mamíferos Carnívoros Ibéricos (Iberian Mammalian Carnivores). Lynx Edicions
PitaR,MiraA,MoreiraF,MorgadoR,BejaP(2009)Influenceoflandscapecharacteristicsoncarnivore
diversity and abundance in Mediterranean farmland. Agric Ecosyst Environ 132 (1-2):57-65.
Polis GA, Myers CA, Holt RD (1989) The Ecology and Evolution of Intraguild Predation - Potential
Competitors that Eat Each Other. Annu Rev Ecol Syst 20:297-330.
Porter WF, Church KE (1987) Effects of environnemental pattern on habitat preference analysis.
Journal of Wildlife Management 51, 681-685.
Primack RB (2006) Essentials of conservation biology. Fourth edition. Essentials of conservation
biology. Fourth edition. Sinauer Associates.
Pulliam HR (1988) Sources, Sinks, and Population Regulation. Am Nat 132 (5):652-661.
Ragni B (1978) Observations on the ecology and behaviour of the wild cat (Felis silvestris Schreber,
1777) in Italy. Carnivore Genetics Newsletters, 3: 270–274.
Ralls K, White PJ (1995) Predation on San-Joaquin Kit Foxes by Larger Canids. J Mammal 76
(3):723-729.
Revilla E, Palomares F (2002) Does local feeding specialization exist in Eurasian badgers? Can J
Zool-Rev Can Zool 80 (1):83-93.
Revilla E, Palomares F (2005) Patrones generales de organizacion social en el tejon eurasiático. In: Virgos E,
Revilla E, mangas JG, Domingo-Roura X (Eds.) Ecología y conservación del tejón en ecosistemas
mediterráneos. Sociedad Espanola para la Conservacion y Estudio de los Mamiferos SECEM.
61
Introducción
RevillaE, PalomaresF,DelibesM (2000)Defining key habitats for lowdensity populations of
Eurasian badgers in Mediterranean environments. Biological Conservation 95 (3):269-277.
Ripple WJ, Beschta RL (2006) Linking a cougar decline, trophic cascade, and catastrophic regime
shift in Zion National Park. Biological Conservation 133 (4):397-408.
Riquelme-Cantal JA, Simon-Vallejo MD, Palmqvist R, Cortes-Sanchez M (2008) The oldest
mongoose of Europe. J Archaeol Sci 35 (9):2471-2473.
Rockstrom J, Steffen W, Noone K, et al. (2009) A safe operating space for humanity. Nature 461
(7263):472-475.
Rodriguez A, Delibes M (1992) Current Range and Status of The Iberian Lynx Felis-Pardina
Temminck, 1824 in Spain. Biological Conservation 61 (3):189-196.
Rodriguez A, Delibes M (2003) Population fragmentation and extinction in the Iberian lynx.
Biological Conservation 109 (3):321-331.
Rodriguez A, Delibes M (2004) Patterns and causes of non-natural mortality in the Iberian lynx
during a 40-year period of range contraction. Biological Conservation 118 (2):151-161.
Rodriguez A, Martin R, Delibes M (1996) Space use and activity in a mediterranean population of
badgers Meles meles. Acta Theriol 41 (1):59-72.
Rosalino LM, Santos-Reis M (2002) Feeding habits of the common genet Genetta genetta (Carnivora
: Viverridae) in a semi-natural landscape of central Portugal. Mammalia 66 (2):195-205.
Rosenzweig ML (1981) A Theory of Habitat Selection. Ecology 62 (2):327-335.
Rosenzweig ML (1987) Habitat selection as a source of biological diversity. Evol Ecol 1 (4):315-330.
Sarmento P (1996) Feeding ecology of the European wildcat Felis silvestris in Portugal. Acta Theriol
41 (4):409-414.
Scheffer M, Bascompte J, Brock WA, Brovkin V, Carpenter SR, Dakos V, Held H, van Nes EH, Rietkerk
M, Sugihara G (2009) Early-warning signals for critical transitions. Nature 461 (7260):53-59.
Schipper J, Chanson JS, Chiozza F, et al. (2008) The status of the world’s land and marine mammals:
Diversity, threat, and knowledge. Science 322 (5899):225-230.
Sergio F, Caro T, Brown D, Clucas B, Hunter J, Ketchum J, McHugh K, Hiraldo F (2008) Top
Predators as Conservation Tools: Ecological Rationale,Assumptions, and Efficacy. In:
Annual Review of Ecology Evolution and Systematics, vol 39. Annual Review of Ecology
Evolution and Systematics. Annual Reviews, Palo Alto, pp 1-19.
62
Introducción
Sinclair ARE, Fryxell JM, Caughley G (2006) Wildlife ecology, conservation, and management. Second
edition. Wildlife ecology, conservation, and management. Second edition. Blackwell Publishing.
Soberon J (2007) Grinnellian and Eltonian niches and geographic distributions of species. Ecol Lett
10 (12):1115-1123.
Soulé ME (1987) History of the Society for Conservation Biology: how and why we got here.
Conserv Biol 1 (1):4-5
Soulé ME, Bolger DT, Alberts AC, Wright J, Sorice M, Hill S (1988). Reconstructed dynamics
of rapid extinctions of chaparral-requiring birds in urban habitat islands. Conservation
Biology 2: 75–91.
Sovada MA, Sargeant AB, Grier JW (1995) Differential Effects of Coyotes and Red Foxes on Duck
Nest Success. Journal Of Wildlife Management 59 (1):1-9.
Steffen W, Grinevald J, Crutzen P, McNeill J (2011) The Anthropocene: conceptual and historical
perspectives. Philos Trans R Soc A-Math Phys Eng Sci 369 (1938):842-867.
Stockwell DRB, Peterson AT (2002) Effects of sample size on accuracy of species distribution
models. Ecol Model 148 (1):1-13.
Terborgh J, Estes JA, Paquet P, Ralls K, Boyd-Heger D, Miller BJ, Noss RF (1999) The role of top
carnivoresinregulatingterrestrialecosystems.Continentalconservation:scientificfoundations
of regional reserve networks. Island Press.
Teyssèdre A (2004) Vers une sixième crise d’extinctions ?, in Barbault R. et Chevassus-au-Louis B.,
(eds.).Biodiversitéetchangementsglobaux.Enjeuxdesociétéetdéfispourlarecherche,
ADPF, Paris, France.
Thomas DL, Taylor EJ (1990) Study Designs and Tests for Comparing Resource Use and Availability.
Journal Of Wildlife Management 54 (2):322-330.
Thomas DL, Taylor EJ (2006) Study designs and tests for comparing resource use and availability
II. Journal of Wildlife Management 70 (2):324-336.
Thurber JM, Peterson RO, Woolington JD, Vucetich JA (1992) Coyote Coexistence With Wolves on
The Kenai Peninsula, Alaska. Can J Zool-Rev Can Zool 70 (12):2494-2498.
Tilman D, May RM, Lehman CL, Nowak MA (1994) Habitat Destruction and the Extinction Debt.
Nature 371 (6492):65-66.
Tsoar A, Allouche O, Steinitz O, Rotem D, Kadmon R (2007) A comparative evaluation of presence-
63
Introducción
only methods for modelling species distribution. Diversity and Distributions 13 (4):397-405.
Turchin P (1998) Quantitative analysis of movement: measuring and modeling population
redistribution in animals and plants. Quantitative analysis of movement: measuring and
modeling population redistribution in animals and plants. Sinauer Associates, Inc.
Vanhorne B (1983) Density as A Misleading Indicator of Habitat Quality. Journal of Wildlife
Management 47 (4):893-901.
Villafuerte R, Calvete C, Blanco JC, Lucientes J (1995) Incidence of viral hemorrhagic disease in
wild rabbit populations in Spain. Mammalia 59 (4):651-659.
Villafuerte R, Lazo A, Moreno S (1997) Influence of food abundance and quality on rabbit
fluctuations:Conservation andmanagement implications inDonanaNationalPark (SW
Spain). Rev Ecol-Terre Vie 52 (4):345-356.
Virgós E, Casanovas JG (1997) Habitat selection of genet Genetta genetta in the mountains of
central Spain. Acta Theriol 42 (2):169-177.
Virgós E, Llorente M, Cortes Y (1999) Geographical variation in genet (Genetta genetta L.) diet: a
literature review. Mammal Review 29 (2):117-126.
Vitousek PM, Mooney HA, Lubchenco J, Melillo JM (1997) Human domination of Earth’s
ecosystems. Science 277 (5325):494-499.
Whittake.Rh, Levin SA, Root RB (1973) Niche, Habitat, and Ecotope. Am Nat 107 (955):321-338.
Woodroffe R (2001) Strategies for carnivore conservation: lessons from contemporary extinctions.
In: Gittleman JL, Funk SM, Macdonald DW, Wayne RK (eds) Carnivore Conservation, vol
5. Conservation Biology Series. Cambridge Univ Press, Cambridge, pp 61-92.
Woodroffe R, Ginsberg JR (1998) Edge effects and the extinction of populations inside protected
areas. Science 280 (5372):2126-2128.
Yoccoz NG, Nichols JD, Boulinier T (2001) Monitoring of biological diversity in space and time.
Trends Ecol Evol 16 (8):446-453.
Yodzis P (1978) Competition for space and the structure of ecological communities. Springer-Verlag.
Zalasiewicz J, Williams M, Haywood A, Ellis M (2011) The Anthropocene: a new epoch of geological time?
Introduction. Philos Trans R Soc A-Math Phys Eng Sci 369 (1938):835-841.
Zapata SC, Travaini A, Ferreras P, Delibes M (2007) Analysis of trophic structure of two carnivore
assemblagesbymeansofguildidentification.EurJWildlRes53(4):276-286.
64
65
Soto C, Desniça S, Palomares F (2012) Non-biological factors affecting track
censuses: implications for sampling design and reliability. European Journal of
Wildlife Research 58:117–126
CHAPTER 1
Non-biological factors affecting track censuses: implications for sampling design and reliability
Factores no biológicos afectan los censos de rastros:
implicaciones para el diseño muestral y fiabilidad
66
Chapter 1
67
Chapter 1
ABSTRACT
Track census is a widely used method for rapid faunal assessments, which
assumes that differences in track count numbers mainly reflect differences in
species abundance due to some biological factors. However, some methodological
and climatic variables might affect results of track censuses. Here, we tested the
effect of climatic variables, such as maximum temperature, humidity, wind speed
or days since last rain, and methodological factors, such as censusing day period,
distance from transect to vegetation edge, substrate condition or observer on the
number of tracks of mammal carnivores and some of their potential prey detected in
sandy substrates. We sampled 2 x 2 km squares located within the scrubland area of
Doñana National Park (southwestern Spain) for carnivore and several potential prey
tracks. Our results showed differences in the number of tracks detected between
observersandasignificantinteractionbetweenobserversandthedayperiodwhen
censuses were carried out. Moreover, the variables increasing the quality of the
substrate (higher environmental humidity, lower wind speed and days since last
rain) led to a greater detection of carnivore tracks, but depending on the size of the
species sampled other variables, such as distance from transects to the vegetation
border,alsoaffectedresults.Werecommendrestrictingsamplingtocertainfixed
weather conditions when planning to monitor relative animal abundance from
track censuses. When not possible, climatic or methodological variables should be
included as covariates in analyses that try to test for the biological factors affecting
wildlife abundance, taking into account that these variables, which affect the number
of tracks detected could vary between years.
68
Chapter 1 Chapter 1
RESUMEN
Los censos de rastros constituyen un método ampliamente utilizado en
estudios de fauna y asumen que las diferencias en el número de rastros detectados
reflejandiferenciasenlaabundanciadelasespeciesdebidoafactoresbiológicos.
Sin embargo, algunas variables metodológicas y/o climáticas pueden afectar
los resultados de dichos censos de rastros. En este estudio analizamos el efecto
potencial de variables climáticas tales como la temperatura máxima, la humedad,
la velocidad del viento o los días transcurridos desde la última lluvia, así como de
variables metodológicas como el periodo del día en el que se realiza el censo, la
distancia desde el transecto al borde de la vegetación, las condiciones del sustrato
o el observador, en el número de rastros de diferentes especies de mamíferos
carnívoros así como de algunas de sus potenciales presas detectados en censos de
rastros en sustratos arenosos.
Censamos rastros de carnívoros y de diferentes presas potenciales en
cuadrículas de 2 x 2 km2 localizadas dentro del área de matorral del Parque Nacional
de Doñana (situado en el suroeste de España). Nuestros resultados mostraron
diferencias en el número de rastros detectados en los censos dependiendo del
observador,asícomounainteracciónsignificativaentreelobservadoryelperíodo
del día en el que se realizó el censo. Además, las variables que incrementaban la
calidad del sustrato (una mayor humedad ambiental y una menor velocidad del
viento, así como un número bajo de días transcurridos desde la última lluvia),
permitieron una mayor detección de rastros de carnívoros. No obstante, dependiendo
del tamaño de la especie, otras variables como la distancia del transecto al borde
de la vegetación también afectaron los resultados. Recomendamos restringir los
censos a ciertas condiciones climáticas determinadas cuando se planteen realizar
69
Chapter 1 Chapter 1
monitoreos de la abundancia relativa de especies basándose en datos obtenidos por
censos de rastros. En los casos en los que no sea posible, las variables climáticas
y/o metodológicas se deben incluir como covariables en los análisis que tengan
como objetivo analizar los factores biológicos que determinan la abundancia de las
especies, teniendo en cuenta que esas variables que pueden afectar los resultados de
censos, además pueden variar entre distintos años.
70
Chapter 1 Chapter 1
INTRODUCTION
Determining occurrence and estimating population abundance of species
is fundamental for their conservation, research and management (Caughley
and Sinclair 1994; Silveira et al. 2003; Sadlier et al. 2004). This is particularly
difficultandposesmanypracticalproblemsona largespatialscaleandin long-
term monitoring for cryptic species with nocturnal and solitary habits, large home
ranges, low-density populations and an elusive nature, such as most mammalian
carnivore species and their prey.
Few methods are suitable for monitoring elusive, low-density species (Mills
et al. 2000) in spite of the amount of available monitoring methods (Williams et
al. 2002; Liebenberg 2010), but indirect sampling methods such as track counts on
suitable natural substrates (e.g. snow, mud or sand) have been traditionally used to
overcome these problems (e.g. Stephenson 1986; Smallwood and Fitzhugh 1995;
Zaumyslova 2000; Gusset and Burgener 2005; Datta 2008; Funston et al. 2010).
Thebroadapplicationofnaturalsignsurveyssuchastrackcountshasfirmly
established their use as a tool for wildlife detection. Track surveys do not rely upon
special technology or equipment, can be relatively straightforward and quick to
conduct, and can easily incorporate multispecies and large geographic area objectives.
Moreover, track counts do not require a behavioural response to attractants or other
survey equipment, thus there are potentially fewer species-specific limitations
andbiasesinherenttotracksurveys(Longetal.2008).Nevertheless,field-based
species identificationmay be ambiguous or unfeasible so additional efforts and
highlyskilledandexperiencedtrackersareneededtovalidatetheidentificationof
species or individuals.Thisweakness related to species identification combined
with the limited availability of appropriate tracking mediums or conditions, the
71
Chapter 1 Chapter 1
ephemeral and weather-dependent character of tracks and the inconsistent survey
designs and quality control procedures, have resulted in a growing criticism of track
surveys and the need to improve survey efforts to meet more rigorous standards.
Although strong relationships between kilometric abundance indices (KAI),
obtained by spotlight counts, and population size have been previously reported
(e.g. Newsome et al. 1989; Short et al. 1997; Garel 2010), it is important to highlight
that track censuses can only be taken as indices of presence, relative abundance
or density estimators (Anderson 2001) and that such indices are rated closely to
true animal abundance across habitat types, observers, and other factors (see Gibbs
200). Herein, the number of tracks of certain species encountered on a transect will
depend on biological factors, such as their abundance, food density and distribution,
vegetationstructureandintra-orinterspecificinterference,includinghumans(e.g.
Odonoghue et al. 1997; Shapira et al. 2008; Bayne et al. 2008; Blaum et al. 2009)
but there are other classes of variables that affect the index (Buckland et al. 1993).
These variables are related to the observer, including the observer’s training and
experience, eyesight and fatigue level, the environment (i.e. climatic conditions and
local habitats) and aspects of the species itself, such as their body size (Anderson
2001; Mackenzie and Kendall 2002). Among the variables associated with the
environment, such as wind speed, temperature, humidity, cloud cover, time of
sunrise or days from the last rain or snow, many have been previously suggested as
potentialinfluencesontheresultsoftracksampling(Norton 1990; Hayward et al.
2002; Long et al. 2008). These non-biological factors constitute an important source
of error, as they affect the probability of detection and therefore the count. If they are
not considered when designing a monitoring program they can increase variance or
uncertainty for the estimates of relative abundance indices (Thompson et al. 1989).
72
Chapter 1 Chapter 1
Despitethepotentialinfluencethattheabove-mentionedfactorsmayhave
on track counts, specific studies on the subject are scarce (Jennelle et al. 2002;
Karanth et al. 2003, but see Stander 1998; Balme et al. 2009; Zielinski and Schlexer
2009). Nevertheless, there is a growing suggestion to include measures of precision
and estimates of the detection probability when using indices values (usually raw
counts) purporting to measure relative abundance (e.g. Anderson 2001; Rosenstock
et al. 2002; Engeman 2003).
Here, we studied how methodological and climatic factors affected the number
oftracksdetectedinsurveyscarriedoutonsand-basedsubstrates.Specifically,we
examined whether some climatic variables such as maximum temperature, average
relative humidity, maximum wind speed or days since last rain and methodological
factors such as censusing day period, distance from transect to vegetation edge,
substrate condition or observer may affect the number of tracks detected for a set of
seven mammalian carnivores and some of their potential prey on sandy substrates
in South-western Spain and how, depending on the size of the animals surveyed,
different factors could affect the detection probability.
73
Chapter 1 Chapter 1
METHODS
Study area
The study was carried out in Doñana National Park in southwestern Spain
(37º9´N, 6º26´W). This is a 550 km2flatsandyareaatsealevelborderedtothesouth
and west by the Atlantic Ocean and to the east by the Guadalquivir River mouth.
The climate is Mediterranean subhumid (i.e. characterised by mild wet winters
and hot dry summers), with an average annual rainfall of 500-600 mm. There are
three main environmental units in the park: marshland, dunes and Mediterranean
scrubland (Fig. 1). Track censuses were restricted to the scrubland area, which
is mainly characterised by heterogeneous patches of xerophytic species such as
Halimium sp. and Cistus sp., and hygrophytic ones such as Erica sp., with some
patches of Juniperus phoenicea and Pistacia lentiscus shrubs. Interspersed with the
scrubland there are scattered cork oak trees (Quercus suber) and wild olive trees
(Olea europea), and a few patches of pine Pinus pinea and eucalyptus Eucalyptus
sp. plantations. The Mediterranean scrubland represents approximately half of the
National Park surface area.
Carnivore species in our study area are the red fox (Vulpes vulpes), the
Eurasian badger (Meles meles), the Egyptian mongoose (Herpestes ichneumon), the
common genet (Genetta genetta), the polecat (Mustela putorius) the Iberian lynx
(Lynx pardinus), the European otter (Lutra lutra), wild and domestic cat (Felis sp.),
and domestic dog (Canis familiaris). Polecats and otters were excluded from our
study because of their low abundance.
Potential target prey species for hunting or consumption as carrion by the
carnivore community and sampled in our study were small mammals (i.e. Garden
dormouse (Eliomys quercinus), Southern Water Vole (Arvicola sapidus), bush rat
74
Chapter 1 Chapter 1
(Rattus rattus), Long-tailed Field Mouse (Apodemus sylvaticus) and other mice
(Mus spp.), but the most common are A. sylvaticus (Kufner and Moreno 1989)),
European rabbits (Oryctolagus cuniculus), red partridges (Alectoris rufa), domestic
cows (Bos Taurus) and horses (Equus caballus) and wild ungulates such as the
fallow deer (Dama dama), the red deer (Cervus elaphus) and the wild boar (Sus
scrofa).
Track sampling
Under a wider project that aims to study biological and anthropic
factors affecting wild and domestic carnivore abundance and distribution within
Doñana National Park, during the wet seasons of 2007-08 and 2008-09 (from
Figure 1. Map of the study area showing Doñana National Park in the southwestern Spain and the 2x2 km2, where carnivore and prey track censuses were carried out
75
Chapter 1 Chapter 1
November 2007 to May 2008 and from October 2008 to April 2009) three and two
observers sampled 59 and 57, respectively, 2 x 2 km squares all with at least 40% of
their surface located within the scrubland area of the park (Fig.1). Marshland area
was not sampled as its clay soils make it unsuitable for track censuses. The squares
with≥40%areaofopendunewereexcludedforthepresentstudysincevegetation
is clearly different from the rest of the scrubland area, which would add an extra
sourceofvariabilitytodata.Thethreeobserversofthefirstyearwereafieldworker
with 15 years of experience and two without previous experience, but that were
trainedfortwomonthsbytheexperiencedfieldworker.Duringthesecondyear,the
experienced observer and one of the others carried out the surveys.
We sampled for carnivore tracks in each square by slowly walking zigzag
(ca.1.5km/h)inatleast3km-longsandypaths(incartracksorfirebreaks).Once
a continuous track, that crossed side to side across the pathway, was detected, we
georeferenced it using a GPS. We noted location as a grid reference, date, and the
methodological variables, censusing day time (we established three block schedules;
early morning (from 8 a.m. to 12 a.m.), afternoon (from 12 a.m. to 3 p.m.) and
evening (from 5 p.m. to sunset), start time and end time for each census, and the
observer who carried out the census. In order to homogenize the number of tracks
detected per grid and maximize the probability of detection for each carnivore
species, we re-sampled the same path (leaving at least 7 days between samplings) a
secondtimeinafewsquaresuntilcompleting3kmifduringthefirstsamplingthere
was not enough available path within the square to achieve this distance. Thus, we
had more censuses than total number of 2x2 km squares. We always carried out
surveys at least 3 days after any rainfall.
76
Chapter 1 Chapter 1
We concentrated prey sampling within on month to avoid strong intermonthly
variations in abundance for some species (e.g. see Kufner 1986; Palomares et al.
2001 for small mammals and European rabbits, respectively). Thus, we carried out
the sampling of prey tracks in April 2009 along transects in every 2x2 km square
sampled for carnivore tracks. These transects for prey species were walked as they
were for carnivore tracks, were 25 m in length and approximately 1.7 m wide (i.e.
the area of a four-wheel-drive car) and were located in the middle of the census path
and separated by at least 300 m. We recorded the location as a grid reference, date,
observer, and the following methodological variables: distance from nearest border
of census transect to the closest vegetation border (not recorded for carnivore track
censuses as they were carried out by zigzag walking), pathway where transects
wereestablished(firebreaksorcartracks)andqualityofthesubstratefordetecting
tracks based on the presence of grass (we established two categories; good (when
grassy groundcover was less than 10% in any part of the 25 m transect) and fair
(when grassy groundcover in any part of the 25 m transect was between 11-30%).
We considered unsuitable for prey counts transects in which grassy groundcover
was more than 30% in some part of the transect.
The climatic variables result from an average of the maximum temperature
(calculated as the average of the maximum temperature measured on the census day
and the maximum temperature measured two consecutive days before the census
day), relative humidity, maximum wind speed, and the number of days since last
rain. The data was obtained from a station located inside Doñana National Park
(Control RM1 Meteorological Station; Latitude: 37º 1’18’’, Longitude: 6º 33’ 17’’
http://icts.ebd.csic.es).
77
Chapter 1 Chapter 1
Data analyses
In order to avoid linearity assumptions, we preliminary explored the shape of the
responseforeachlandscapevariablebeforefittingthemintothefinalequations(Austin,
2002).WiththisaimwefitGeneralisedAdditiveModels(GAMs)(HastieandTibshirani,
1990) using carnivore Kilometric Abundance Indexes and the number of prey tracks as
responsevariablesandfittingsmoothingsplineswith3degreesoffreedomtomodel
every climatic and methodological continuous effect. The smoothed variables were then
turned into suitable parametric terms guided by visual inspection of the partial residual
plots(Crawley,2005).Thepostulatedmodelswerethenfittothetrack-censusdataset
using general linear models (GLM) with the log link, negative binomial error structure
andlinearandnon-linearresponsestofixedeffectsinaccordancewiththeGAMresults.
We analyzed the effect of methodological variables observer (observer) and censusing
day time (day_time), and climatic variables maximum temperature (max_temp), average
relative humidity (humidity), maximum wind speed on census day (wind_speed), and
days since last rain (last_rain) on the carnivore Kilometric Abundance Index (KAI)).
We also included in the models the interactions between observer and day_time, as the
number of samplings carried out by each observer in each daily time period was different.
Correlations between predictors were always low (r<0.6)sowefittedfullmodels(i.e.
models including all the methodological and climatic variables). As for some 2x2 km
squares we carried out more than one census, our sampling unit was the census and not
the square. We examined the effect of the above variables on total carnivore abundance
index, small carnivore (from 1-5 kg of body mass) abundance index (including the
common genet, wild and domestic cats and the Egyptian mongoose) and medium-sized
carnivore (>7 kg of body mass) abundance index (including the red fox, the Eurasian
badger, the Iberian lynx and the domestic dog).
78
Chapter 1 Chapter 1
We followed a backward regression-model selection procedure excluding
variables contributing least to the model (i.e. variables with P > 0.3) before models
wererefitted.OnlyvariableswithP ≤0.05wereinterpretedasstatisticallysignificant.
Overdispersion was not a problem (ɸ was close to 1 (1.14 - 1.21)) for any of the models
(Zuur at al. 2009). We analyzed data separately for each study year as the number of
observers changed and to maximize the variability between conditions, which could
affect the number of tracks detected on sand substrates.
We also performed general linear models (GLM) with negative binomial errors
and log link function to analyse the effect of observer, distance from border of census
transects to the closest vegetation border (dist_veg.), type of path (place) where prey
censuseswerecarriedout(firebreaksorcartracks),qualityofthesubstrate(quality) and
climatic variables last_rain, max_temp, humidity and wind_speed on track counts data
of prey species. We also included in the models the interactions between observer and
quality and observer and place. Prey data were grouped as total prey, small prey (small
mammals), medium-sized prey (rabbits and partridges) and large prey (cows, horses and
ungulates). Correlation between predictors was low (r <0.5),sowefittedfullmodels.
Overdispersion was not a problem (ɸ = 1.04 - 1.24). The sample unit to adjust GLM was
the 25 m transects.
To simplify models and their understanding we also followed a backward
regression-model selection procedure excluding variables with P>0.3,andthenrefitted
the models.
All statistical analyses were performed using the SAS® 9.2 statistical software
(SAS Inst. Inc., Cary,NC),GAM andGLMwere fitted using the gam and genmod
procedures, respectively.
79
Chapter 1 Chapter 1
RESULTS
A total of 471 km were walked and 8,373 carnivore tracks were found during
surveys, with the red fox, the Eurasian badger and the Egyptian mongoose being
the most recorded species (Table 1). For prey, 5,000 tracks were detected on 11,575
m sampled (Table 2). Prey species more often detected were ungulates (i.e. fallow
deer, red deer, wild boar) and rabbits.
Forthefirstyear,thereweredifferencesinthenumberoftracksdetected
among observers for all small and medium carnivores, and the number of tracks
decreased when wind speed increased for total carnivores, increased when humidity
was higher for small carnivores, and was highest during the afternoon for medium-
sizedcarnivores(Table3,Fig.2).Asignificantinteractionwasalsodetectedbetween
thecensusingdayperiodandtheobserver,withthethreeobserversfindingmore
medium-sized carnivore tracks in the evening than in the morning or afternoon
(Table 3, Fig. 2d).
Number of censuses was 76 and 62 for each study year, respectively.
2007-2008 2008-2009
SpeciesPositive censuses
Total number of
tracks
KAI (mean±SD)
RangePositive censuses
Total number of tracks
KAI (mean±SD)
Range
Lynx pardinus 19 65 0.4 ± 0.9 0 - 5.4 14 75 0.4 ± 0.9 0 - 5.0
Meles meles 68 599 3.9 ± 5.1 0 - 20.6 53 666 3.7 ± 3.8 0 - 17.3
Herpestes ichneumon 60 631 4.0 ± 4.4 0 - 21.1 48 544 2.8 ± 3.3 0 - 12.5
Vulpes vulpes 76 3,138 17.7 ± 9.1 2.4 - 44.2 61 2,333 11.9 ± 6.5 1.2 - 32.6
Genetta genetta 21 160 0.8 ± 2.3 0 - 15.9 20 66 0.4 ± 0.7 0 - 2.8
Felis sp. 8 25 0.2 ± 0.8 0 - 5.7 17 27 0.1 ± 0.3 0 - 1.3
Canis familiaris 12 23 0.2 ± 0.7 0 - 5.9 11 21 0.1 ± 0.3 0 - 2.0
Table 1. Carnivore kilometric abundance index (tracks / km; KAI) in the scrubland area of Doñana National Park during the wet seasons of 2007-2008 and 2008-2009.
80
Chapter 1 Chapter 1
During the second year, more tracks were detected when humidity increased for total
and medium-sized carnivores, and fewer small carnivore tracks were recorded when
daily maximum temperature and days since last rain increased (Table 3, Fig. 3).
Most of the variables considered affected the total number of prey tracks
detected (Table 4). Thus, the number of tracks for total prey increased when the
daily maximum temperature, humidity and days since last rain increased (Fig.
4c), and decreased when wind speed and distance to vegetation edge was highest.
Furthermore, the number of total tracks was higher when samplings were carried
out in car tracks than infirebreaks (Fig. 4b), oneobserverdetectedmore tracks
thantheother(Fig.4a),andtherewasasignificantinteractionbetweenobserver
and place (Fig. 4d). Some interesting exceptions to this general pattern were found
when data were separated by type of prey (Table 4). For small prey, number of
tracks was not affected by wind speed, humidity, maximum daily temperature and
days since last rain, and the interaction between observer and place was not found.
For medium-sized prey the number of tracks found was not affected by wind speed,
daily maximum temperature and days from the last rain, but in this case quality of
the road did affect results, with a higher number of tracks being detected at transects
without grass. Finally, the number of large prey tracks was not affected by daily
maximum temperature, distance to vegetation, type of path, or observer, and there
was no interaction between observer and place.
DISCUSSION
Track censuses can provide a rapid and cost-effective assessment of relative
abundance of a species, but improving their performance and robustness by
understanding the factors that potentially affect them is a key step for the design of
81
Chapter 1 Chapter 1
sound monitoring programmes. Indexes derived from track surveys are partially a
function of animal abundance, but are also a function of a long list of methodological
and climatic variables and characteristics of the species being surveyed. Our results
support the hypothesis that non-biological factors can affect the number of animal
tracks detected in surveys, and must be taken into account when planning to study
animal abundance, distribution or the biological factors determining them.
The aim of this study is not to establish a standard protocol to carry out
track censuses in sandy substrates but to identify how different methodological
and climatic variables can affect the number of tracks detected in censuses.
Nevertheless, as a rule, our study shows how variables increasing the quality of the
substrate (e.g. higher environmental humidity, lower wind speed and few days since
last rain) allowed the detection of a greater number of tracks. Previous studies had
suggestedthatthesetypesofvariablescouldinfluenceresults,butnoquantitative
data had been provided (but see Hayward et al. 2002; Bayne et al. 2008). For
instance, to determine the presence of mountain lions Felis concolor californica,
through track surveys, Van Dyke et al. (1986) and Smallwood and Fitzhugh (1995)
recommended excluding desert areas or windy conditions. Other authors (Zielinski
and Kucera 1995; Bayne et al. 2005) have made recommendations for snow
tracking to detect the presence of a wide diversity of species such as American
Table 2. Total number of tracks detected and the mean index of abundance (tracks / 25 m) for each group of potential prey species along 463 25 m transects.
Species group No of tracks Mean ± SD RangeSmall mammals 233 0.5 ± 1.4 0 - 13Rabbits 2,132 4.6 ± 9.4 0 - 115Partridges 260 0.6 ± 1.3 0 - 11Cows/horses 210 0.4 ± 1.6 0 - 18Ungulates 2,165 4.7 ± 9.4 0 - 33
82
Chapter 1 Chapter 1
martens (Martes americana), Canada lynx (Lynx canadensis), wolverines (Gulo
gulo), coyotes (Canis lastrans), red foxes (Vulpes vulpes), bobcats (Lynx rufus),
grey wolves (Canis lupus) or mountain lions (Puma concolor) among others.
Recommendations include avoiding tracking during snowfall and waiting until the
second morning after a snowfall, carrying out tracking early in the morning during
periods of melting and freezing and not tracking on south-facing slopes.
Moreover, in Europe and other arid regions of Australia, India, South
America or South Africa where track surveys occurring in dust, mud or sand are more
broadly employed, many authors have also tried to standardize survey design and
Effects 2007-2008 2008-2009
Total carnivores
Small carnivores
Medium-sized carnivores
Total carnivores
Small carnivores
Medium-sized
carnivoresIntercept 3.7238 *** 0.5065 3.2957*** 2.4779** 2.0154 1.4906 wind_speed -0.1329* - -0.1316 0.1071 -0.1751 -0.0711 humidity - 0.0269** - 0.0194** - 0.0261** max_temp - - - -0.0376 -0.0937** -0.0095 last_rain - - - -0.0160 -0.048** -0.0023 day_time
early morning
afternoon
evening
24.5214
28.3601
35.5493
-
-
-
15.1443**
22.8087**
27.8652**
16.5645
20.3801
29.2495
-
-
-
15.1596
16.6357
23.1351 observer
1
2
3
20.7509***
27.8207***
38.8229***
3.4544***
2.4517***
11.5468***
16.1252**
25.1415**
28.3257**
-
-
-
3.5478
2.4450
20.0273
16.1825
observer*day_time ns - *** - - -
Table 3. Results of GLM analysis to test for the effect of several methodological and climatic variables on abundance indeces of total, small and medium-sized carnivores in Doñana National Park. Standard errors have been omitted to simplify the table. Variables with P > 0.3 excluded from the models are represented as (-). Least squares means (LS-Means) of the categorical fixedeffectsobserver and day time are shown. Non-est. means that the model could not calculate the parameter because of little data. (*P < 0.05; **P < 0.01; ***P<0.001;nsnotsignificant).
83
Chapter 1 Chapter 1
effortbysuspendingsearchesoftracksfor≥24hoursafterprecipitationorperiods
ofhighwinds(≥24km/h)andalsobyusingonly early morning observations for
analysis (e.g. Stander et al. 1998; Sargeant et al. 2005; Evans et al. 2009; Funston
et al. 2010).
There was some exception to the general rule stated. A positive correlation
was found between number of tracks recorded and days since last rain for large prey
species. Their large size rendered their tracks easily recognizable in poor substrate
conditions.
Table 4. Results of GLMM analysis to test for the effect of several methodological and climatic variables on abundance indeces of total, small and medium-sized and large prey in Doñana National Park. Standard errors have been omitted to simplify the table. Variables with P > 0.3 excluded from the models are represented as (-). Least squares means (LS-Means) of the categorical fixed effects place, quality and observer are shown. Non-est. means that the model could not calculate the parameter because of little data. (*P < 0.05; **P < 0.01; ***P < 0.001).
Effects Total prey Small prey Medium-sized prey Large prey
Intercept 2.4520*** -2.7311*** 0.7317* 2.4978***wind_speed -0.0005** - -0.0004 -0.0006**humidity 0.0001*** - 0.0001** 0.0001**max_temp 0.0002** - 0.0002 -last_rain 0.0439*** - - 0.0622***distance_veg -0.0666*** -0.1018* -0.1234*** -place firebreak car track
2.4818**2.8317**
-1.0304**-2.3403**
1.2910*1.7615*
--
quality without grass with grass
2.54092.7726
-1.3780-1.9927
1.8563**1.1962**
1.5986***2.3358***
observer 1 2
3.1145***2.1989***
-1.2665**-2.1042**
2.3213***0.7312***
--
observer*place *** - *** -
observer*quality Non-est. Non-est. Non-est. Non-est.
84
Chapter 1 Chapter 1
Other methodological variables also affected results depending on the size
of the species sampled. A higher number of prey species tracks were detected when
transects ran near a vegetation border or in car tracks. This result could be caused
by two different reasons. Vegetation must exert a protective effect against wind and
maintain a higher level of moisture on the sand, thereby increasing substrate quality
for detecting tracks. Also, medium and small prey may prefer to remain near a
vegetation edge to decrease predation risk (Hughes and Ward 1993, but see Moreno
et al. 1996).
Figure 2. Effects of wind speed on total carnivore tracks per km detected (a), of observer (least square means and their standard errors are represented) on total carnivore tracks per km (b), of relative humidity on small carnivore tracks per km (c), and differences in the number of medium-sized carnivore tracks per km detected by observers in each period of the day given as estimated leastsquaremeansandtheirstandarderrors(d)duringthefirststudyyear.FvaluescomputedasMSModel / MSError (i.e. Mean SquareModel/Mean SquareError), and their respective p-values are shown.
85
Chapter 1 Chapter 1
Nevertheless, it is necessary to note that our results were not consistent
among size groups or between years. This could be due to the fact that many of the
variables that affect detectability, and therefore the count, exhibit time trends (i.e.
vary between years) further confounding the value and interpretation of the index.
This is an important issue as it makes track censuses incomparable across different
climatic and methodological conditions.
Our results also showed differences in the number of tracks detected
amongst observers. Moreover, the differences were more apparent in poor
substrate conditions, when the tracks of nocturnal species would have suffered the
greatest deterioration due to time, sun or wind, aswassuggestedbythesignificant
interaction foundbetweenobserver and thecensusday time.This result reflects
the fact that some observers are able to detect more tracks at one period of the day
than others, probably depending on their performance. The effect of the observer in
the number of tracks detected could also be partially related to the census speed as
differences among observers in their average speed as a function of their experience
were detected (data not shown). Quality of data had been previously questioned
Figure 3. Effects of maximum temperature on small carnivore tracks per km (a), and of days since last rain on small carnivore tracks per km detected (b) for the second study year. F values computed as MSModel / MSError, and their respective p-values are shown.
86
Chapter 1 Chapter 1
depending on skill level of observers (Bider 1968; Smallwood and Fitzhugh 1995;
Anderson 2001; Wilson and Delahay 2001; Silveira et al. 2003). For this reason,
suggestions have been recently proposed to decrease differences among observers
or to evaluate observer skills (Sadlier et al. 2004; Evans 2006, 2009; Zielinski and
Schlexer 2009).
Different approaches might be used to reduce the possible influence
of methodological and climatic variables on track counts. One would be to limit
track censuses to some given weather conditions, which would ensure a constant
substrate quality for detecting tracks. Furthermore, to keep constant the time of the
Figure 4. Effects of observer (least square means and their standard errors are represented) on total prey tracks per 25 m transects (a), of censusing place on total prey tracks per km (least square means and their standard errors are represented) (b), of days since last rain on number of large prey tracks (c), and the interaction between observer and type of pathway sampled given as estimated least square means and their standard errors (d). F values computed as MSModel / MSError, and their respective p-values are shown.
87
Chapter 1 Chapter 1
day, days from the last rain or snow, observers involved in the sampling, or distance
to vegetation borders will also help to diminish variability in the number of tracks
recorded. Some of these suggestions have been approached by maintaining constant
substratequalitythroughtheuseofartificialsubstratesandbyerasingandresampling
newlylefttracksontransectsforagivenfixednumberofdays(Gruberatal.2008;
Watts et al. 2008; Russell et al. 2009; Zielinski and Schlexer 2009).
Avoiding such potentially confounding effects in data collection should be
a fundamental design concept (Engeman 2003). However, it is not always possible
to consider these meteorological or methodological issues when carrying out large-
scale samplings or when large volumes of data are needed for population estimation
procedures. Thus, we also propose that these possible meteorological or methodological
variables be recorded and to include them as covariates in any further statistical
analysis aiming to test for biological factors affecting relative animal abundance (also
see Smallwood and Fitzhugh 1995). Survey design and statistical rigor are important,
but we agree with previous authors (e.g. Thompson et al. 1998;Nichols et al. 2000;
Anderson 2001; Yoccuz et al. 2001; Mackenzie and Kendall 2002) that estimating
relative abundance based on indices alone (e.g., raw counts) is naive, and that the
focus of efforts ought to be on estimating detection probabilities as well.
Theseresultscouldbeappliedtoavarietyofresearchfields,bothfortesting
validity of pre-existing data and improving the suitability and performance of future
studies based on track surveys. Although the indices derived from track censuses
are only partially a function of animal abundance (Anderson 2001), if the variables
associated with the observer, the environment, and characteristics of the species
being surveyed are controlled, the reliability of the information extracted from these
methods may be improved.
88
Chapter 1 Chapter 1
ACKNOWLEDGMENTS
This research was funded by the projects CGL2004-00346/BOS (Spanish Ministry
of Education and Science) and 17/2005 (Spanish Ministry of the Environment; National Parks
Research Programme). Land-Rover España S.A. lent two vehicles for this work. We are very grateful
especiallytoJ.C.Rivillafortheirassistanceduringfieldwork.C.SotowasalsosupportedbyaJAE-
Predoc grant from the CSIC.
REFERENCES
Anderson DR (2001) The need to get the basics right in wildlife studies. Wildl Soc Bull 29:1294-1297.
Balme GA, Hunter LTB, Slotow R (2009) Evaluating Methods for Counting Cryptic Carnivores. J
Wildl Manag 73(3):433-441.
Austin MP (2002) Spatial prediction of species distribution: an interface between ecological theory
and statistical modelling. Ecol. Model. 157, 101–118.
Bayne E, Moses R, Boutin S (2005) Evaluation of winter tracking protocols as a method for
monitoring mammals in the Alberta Biodiversity Monitoring Program. Rep Alta
Biodivers Monit Progr, Vegreville, Alberta, Canada.
Bayne EM, Boutin S et al. (2008)EcologicalfactorsinfluencingthespatialpatternofCanadalynx
relative to its southern range edge in Alberta, Canada. Can J Zool-Rev Can Zool 86(10):
1189-1197.
Bider JR (1968) Animal activity in uncontrolled terrestrial communities as determined by a sand
transect technique. Ecol Monogr 38(3):269-308.
Blaum N, Tietjen B et al. (2009) Impact of Livestock Husbandry on Small- and Medium- Sized
Carnivores in Kalahari Savannah Rangelands. J Wildl Manag 73(1): 60-67.
Buckland ST, Anderson DR, Burnham KP and Laake JL (1993) Distance Sampling: Estimating
Abundance of Biological Populations. Chapman and Hall, London. 446pp
Caughley G, Sinclair ARE (1994) Wildlife ecology and Management. Blackwell Science
Publications, Oxford.
Crawley MJ (2005) Statistics: An Introduction using R. JohnWiley & Sons Ltd., Chichester.
Datta A, Anand MO, Naniwadekar R (2008) Empty forests: Large carnivore and prey
abundance in Namdapha National Park, north-east India. Biol Conserv 141(5):1429-1435.
89
Chapter 1 Chapter 1
Engeman RM (2003) More on the need to get the basics right: population indices. Wildl Soc Bull
31: 286-287.
Evans J (2006) Observer error in identifying species using indirect signs: analysis of a river otter
track survey technique. Thesis, Texas A&M University, College Station, USA.
Evans et al. (2009) Determining Observer Reliability in Counts of River Otter Tracks. J Wildl
Manag 73: 426-432.
Funston PJ, Frank L et al. (2010) Substrate and species constraints on the use of track incidences to
estimate African large carnivore abundance. J Zool 281(1):56-65.
Garel M, Bonenfant C, Hamann JL, Klein F, Gaillard JM (2010) Are abundance indices derived
from spotlight counts reliable to monitor red deer populations? Wildl Biol 16(1)
Gruber B, Reineking B et al. (2008) A new method for estimating visitation rates of cryptic animals
via repeated surveys of indirect signs. J Appl Ecol 45(2): 728-735.
Gibbs JP (2000) Monitoring populations. Pp. 213-247 in Research techniques in animal ecology:
controversies and consequences (L. Boitani & T. Fuller, eds.). Columbia University Press,
New York.
Gusset M, Burgener N (2005) Estimating larger carnivore numbers from track counts and
measurements. Afr J Ecol 43(4): 320-324.
Hastie, TJ, Tibshirani RJ (1990) Generalized Additive Models. Chapman & Hall, London.
Hayward GD, Miquelle DG et al. (2002) Monitoring Amur tiger populations: characteristics of track
surveys in snow. Wildl Soc Bull 30(4): 1150-1159.
Hughes JJ,Ward D (1993) Predation risk and distance to cover affect foraging behaviour in Namib
Desert gerbils. Anim Behav 46: 1243-1245.
Jennelle CS, Runge MC MacKenzie DI (2002) The use of photographic rates to estimate densities
of tigers and other cryptic mammals: a comment on misleading conclusions. Anim
Conserv 5:119-120.
KaranthKU,NicholsJDetal.(2003)Sciencedeficiencyinconservationpractice:themonitoringof
tiger populations in India. Anim Conserv 6:141-146.
Kufner MB (1986) Tamaño actividad, densidad relativa y preferencias de hábitat de los pequeños
y medianos mamíferos de Doñana, como factores condicionantes de su tasa de predación.
Thesis, Universidad Autónoma, Madrid, Spain.
90
Chapter 1 Chapter 1
Kufner MB, Moreno YS (1989) Abundancia y amplitud de los desplazamientos de Apodemus
sylvaticus encuatrobiotoposdeDoñanaquedifierenencoberturavegetal.Doñana,Acta
Vertebrata, 16: 179-181.
Liebenberg et al. (2010) Practical Tracking. Stackpole Books.
Long et al. 2008. Noninvasive Survey Methods for Carnivores. Island Press.
Mills LS, Citta J et al. (2000) Estimating animal abundance using non-invasive DNA sampling:
promise and pitfalls. Ecol Appl 10(1):283–294.
MacKenzie DI, and WL Kendall (2002) How should detection probability be incorporated into
estimates of relative abundance. Ecology 83: 2387-2393
Moreno S, Villafuerte R, Delibes M (1996) Cover is safe during the day but dangerous at night:
the use of vegetation by European wild rabbits. Can J Zool 74: 1656-1660.
Newsome AE, Parer I, Catling PC (1989) Prolonged prey suppression by carnivores-predator-
removal experiments. Oecologia 78:458–467.
Nichols JD, Hines JE, Sauer JR, Fallon FW, Fallon JE, and Heglund PJ (2000) A double-observer
approach for estimating detection probability and abundance from counts. Auk 117:393–408.
Norton PM (1990) How many leopards? A criticism of Martin and de Meulenaer’s population
estimates for Africa. S Afr J Sci 86:218–220.
Odonoghue M, Boutin S et al. (1997) Numerical responses of coyotes and lynx to the snowshoe hare
cycle. Oikos 80(1): 150-162.
Palomares F, Delibes M et al. (2001) Spatial ecology of Iberian lynx and abundance of European
rabbits in South western Spain. Wildl Monogr 148:1-36.
Rosenstock SS, Anderson DR, Giesen KM, Leukering T, and Carter MF (2002) Landbird counting
techniques: Current practices and an alternative. Auk 119:46–53.
Russell JC, Hasler N et al (2009) Automatic track recognition of footprints for identifying cryptic
species. Ecol 90, 2007–2013.
Sadlier LMJ, Webbon CC et al. (2004) Methods of monitoring red foxes Vulpes vulpes and badgers
Meles meles:arefieldsignstheanswer?MammRev34:75-98.
Sargeant GA, Sovada MA et al. (2005) Markov chain Monte Carlo estimation of species distributions:
a case study of the swift fox in western Kansas. J Wildl Manag 69:483– 497.
Shapira I, Sultan H et al. (2008) Agricultural farming alters predator-prey interactions in nearby
91
Chapter 1 Chapter 1
natural habitats. Anim Conserv 11(1): 1-8.
Short J, Turner B, Risbey DA, Carnamah R (1997) Control of feral cats for nature conservation. II.
Population reduction by poisoning. Wildl Res 24:703–714.
Silveira L, Jácomo ATA, Diniz-Filho JAF (2003) Camera trap, line transect census and track
surveys: a comparative evaluation. Biol Conserv 114: 351-355.
Smallwood KS, Fitzhugh EL (1995) A track count for estimating mountain lion Felis concolor
californica population trend. Biol Conserv 71:251–259.
Stander PE (1998) Spoor counts as indices of large carnivore populations: the relationship between
spoor frequency, sampling effort and true density. J Appl Ecol 35:378–385.
Stephenson RO (1986) Development of lynx population estimation techniqucs. Alaska Department
of Fish and Game, Project W-22-2, W-22-3, W-22-4, and W-22-5, Res. Final Report, 84 pp.
Thompson ID, Davidson IJ et al. (1989) Use of track transects to measure the relative occurrence of
some boreal mammals in uncut forest and regenerations stands. Can J Zool 67:1816-1823.
Thompson WL, White GC, and Gowan C (1998) Monitoring vertebrate populations. Academic
Press, San Diego, California, USA.
Van Dyke FG, Brocke RH, Shaw HG (1986) Use of road track counts as indexes of mountain lion
presence. J Wildl Manag 50 (1):102-109.
Watts CH, Thornburrow D et al. (2008) Tracking tunnels: a novel method for detecting a threatened
New Zealand giant weta (Orthoptera: Anostostomatidae). N Z J Ecol 31:92–97.
Williams BK, Nichols JD, Conroy MJ (2002) Analysis and management of animal populations:
modeling, estimation, and decision making. Academic Press, San Diego, USA: 1–1040.
Wilson GJ, Delahay RJ (2001) A review of methods to estimate the abundance of terrestrial
carnivoresusingfieldsignsandobservation.WildlRes28(2):151-164.
Yoccuz NG, Nichols JD, and Boulinier T (2001) Monitoring of biological diversity in space and
time. Trends in Ecol. and Evol. 16:446–453.
Zaumyslova O.Yu. (2000) Long term population-dynamics of ungulates in the Sikhote-Alin Reserve
from winter transect count data. In Analysis of long-term monitoring data of nature
componentsinNatureReservesintheRussianFarEast:70–79.Astafiev,A.A.(Ed.).
Vladivostok: Dalnauka, USAID.
ZielinskiWJ,KuceraTE,BarrettRH(1995)Americanmarten,fisher,lynx,andwolverine:survey
92
Chapter 1
methodsfortheirdetection.USDAforestservice,PacificSouthwestResearchStation
General Technical Report PSW-GTR-157, Albany, CA.
Zielinski WJ, Schlexer FV (2009) Inter-Observer Variation in Identifying Mammals from Their
Tracks at Enclosed Track Plate Stations. Northwest Science 83(4): 299-307.
93
CHAPTER 2
Fine-scale habitat use and niche separation in a guild of sympatric carnivore species that differ in life-history traits
Uso del hábitat a escala fina y segregación de nicho en un
gremio de carnívoros simpátridos que difieren en sus rasgos
de historia de vida
94
Chapter 2
95
Chapter 2
ABSTRACT
Heterogeneous landscapes offer many choices in habitat selection but in
landscapes where a general habitat type dominates, the resources distribution is
limited and opportunities for coexistence are likely to be the most restrictive. Subtle
patterns of habitat partitioning as well as coexistence between species with different
degree of specialisation may promote ecological separation and therefore sympatry
underthisscenario.Wetestedthehypothesisoffine-scalespatialpartitioningasa
functionofthespecies’degreeofspecialisationbycomparinghabitatuseoffive
sympatric carnivore species (Iberian lynx, Eurasian badger, Egyptian mongoose,
common genet and red fox) within a Mediterranean protected area dominated by
Mediterraneanscrublandinsouth-westernSpain.Wefirstdevelopedlogisticand
regression generalized linear mixed models to analyse habitat use and to assess the
environmental features that best describe the relationship between each species and
their environment and secondly we used OMI analysis to illustrate the ecological
spacefilledbyeachspeciesinamultidimensionalspace.Ourresultssuggestedthat
sympatryofcarnivorespeciesinDNPappearstobemediatedbyfine-scalehabitat
selection only for the most specialist species (lynxes and genets). Lynxes and
genets showed the narrowest and most marginal niches as well as niche segregation.
Lynxes were associated with zones of high density of bushes and ecotones between
bushes and pastureland whereas genets showed association to areas with high small
mammal abundance and pine forest with undergrowth bushes. Mongooses were
positively associated with dense bush cover whereas foxes and badgers showed a
non-specifichabitatusepatternandawidedistributionacross theenvironmental
conditions sampled in the study area. Foxes, badgers and mongooses overlapped in
niche position, but foxes and badgers had broader niches than mongooses suggesting
96
Chapter 2Chapter 2
that mongooses exhibit certain degree of habitat specialization.
Differences in fine-scale habitat use and ecological separation through
segregation along the ecological niche dimensions have been shown as key
mechanisms for the most specialist species (genets and lynxes) allowing coexistence
whereas for the most generalist species such as mongooses, foxes and badgers,
another mechanisms such as spatial-temporal segregation of activity patterns might
help to explain coexistence among them.
97
Chapter 2Chapter 2
RESUMEN
Los paisajes heterogéneos ofrecen diversas oportunidades a las especies
para la selección de hábitat, sin embargo, en paisajes dominados por un tipo de
hábitat principal, la distribución de los recursos es limitada y las oportunidades para
la coexistencia entre especies probablemente más restrictivas. Patrones sutiles de
particióndehábitatasícomolacoexistenciaentreespeciesquedifierenensugrado
deespecialismodehábitaty/otróficodebenfomentarlasimpatríaysegregaciónde
nicho bajo este escenario. Comprobamos la hipótesis de la partición espacial a escala
finaenfuncióndelgradodeespecialismode laespecieencuestióncomparando
los patrones de uso del hábitat de cinco especies de carnívoros simpátridos (lince
ibérico, tejón, zorro, gineta y meloncillo) en un área Mediterránea protegida en la
que la zona de estudio estuvo dominada por un tipo de hábitat principal; matorral
Mediterráneo, en el suroeste de España. En primer lugar ajustamos modelos lineares
generalizados mixtos logísticos y de regresión para analizar el patrón de selección
de hábitat y determinar las características ambientales que mejor describen las
relaciones de cada especie con su ambiente. Posteriormente realizamos un análisis
OMI para ilustrar el espacio ecológico ocupado por cada especie en un espacio
multimensional.
Nuestros resultados sugieren que la simpatría entre especies de carnívoros
en el Parque Nacional de Doñana parece estar mediada por selección de hábitat
aescalafinasolopara lasespeciesmásespecialistas(lincesyginetas).Lincesy
ginetas mostraron los nichos ecológicos más estrechos y marginales así como una
segregación de nicho entre ambas especies. Los linces mostraron asociación con
zonas con una densidad elevada de matorral alto y ecotonos entre matorral alto y
pastizales mientras que las ginetas aparecieron asociadas con áreas con una elevada
98
Chapter 2Chapter 2
disponibilidad de micromamíferos y pinares con sotobosque denso. Los meloncillos
estuvieron asociados a zonas con una cobertura densa de matorral alto mientras que
loszorrosy tejonesnomostraronunpatrónespecíficode seleccióndehábitata
escalafina.Zorros,tejonesymeloncillossolaparonenlaposiciónespacialdesus
nichos ecológicos, pero los zorros y tejones mostraron nichos más amplios que el
de los meloncillos, sugiriendo probablemente un mayor especialismo de hábitat de
estosúltimos.Lasdiferenciasenelusodelhábitataescalafinaylasegregación
de nicho resultaron ser mecanismos clave de coexistencia para las especies más
especialistas (linces y ginetas), mientras que para las especies más generalistas como
los meloncillo, zorros y tejones, otros mecanismos como la segregación espacio-
temporal en sus patrones de actividad deben ayudar a explicar los mecanismos de
coexistencia entre estas especies.
99
Chapter 2Chapter 2
INTRODUCTION
The competitive exclusion principle or Gause’s law (Gause 1934, Hardin
1960) states that two ecologically similar species cannot coexist unless they differ
sufficientlyinnicheseparation,thatis,inthewaytheyuseresources.Thus,some
degree of partitioning has to occur in the realized niche of coexisting species which
can occur at the temporal, trophic and/or habitat selection level. This is particularly
obvious in heterogeneous landscapes where spatial heterogeneity foments
coexistence between similar species (i.e., within the same trophic level), selection
of different habitats being one of the main processes that promotes sympatry (Levin
1974, Rosenzweig 1984). Nevertheless, mechanisms that facilitate coexistence
between similar species in apparently homogeneous landscapes where a general
habitat type dominates have been poorly explored.
Coexistence may be dependent on the availability of patches that support a set
ofresourcessufficienttofulfilspecies’basicneeds.Hence,accurate discrimination
and quality assessment of suitable habitats may require more detailed landscape
information to detect crucial habitat features not obvious at broader scales
(Fernández et al. 2003, Martin et al. 2010).
Additionally, life-history trait differences between coexisting species in
apparently homogeneous landscapes (i.e., where the spatial variation in the species’
biotic or abiotic environment is low) may help to understand the way species
exploitresources.Hence,besidesfine-scalehabitatusesegregation,coexistenceis
therefore possible if species exhibit differences in their life history traits that allow
niche differences in space (Brown and Wilson 1956, Hutchinson 1959, Chesson
2000). In other words, cohabitation may be privileged when species with different
degrees of habitat and/or trophic specialisation are present.
100
Chapter 2Chapter 2
To test the hypothesis of fine-scale spatial partitioning as a function of
the species’ degree of specialisationwe compared habitat use of five sympatric
carnivore species (Iberian lynx Lynx pardinus, Eurasian badger Meles meles,
Egyptian mongoose Herpestes ichneumon, common genet Genetta genetta and red
fox Vulpes vulpes) within a Mediterranean protected area with a main habitat type
in south-western Spain. We hypothesized that in an area with a main habitat type,
fine-scalehabitatusedifferencesmayallowsympatryofspecieswithinthesame
thropic level but that differ in the degree of specialisation. If we graph the ecological
space filled by each specieswithin an n-dimensional hypervolume representing
environmental variability, spatial segregation in their realized niches of the species
asafunctionoftheirdegreesofspecialisationmayberevealed.Wefirstdeveloped
habitatmodelsatfine-spatialscalestoassesstheenvironmentalfeaturesthatbest
describe the relationship between each species and its environment. Second, we
usedanordinationmethodtoillustratetheecologicalspacefilledbyeachspeciesin
amultidimensionalspace.Wefittedasetoffine-scalehabitatmodelsconsidering
prior knowledge of species ecology to test five hypotheses: (a) vegetation and
landscape structure are the best explanatory factors of species habitat use; (b) since
our study area is quite homogeneous in vegetation, we hypothesized that prey
availability is the single explanatory factor of habitat use; (c) human disturbance
is the most important variable describing habitat selection; (d) vegetation and
landscape structure and prey availability interact to explain habitat use; and (e) all
landscape descriptors, prey availability and human disturbance are relevant.
The study of the carnivore guild considered here is particularly interesting
because it will allow to compare different species within a continuum of
specialisation ranging from complete specialisation (i.e., the Iberian lynx) to full
101
Chapter 2Chapter 2
generalisation (i.e., the red fox). The Eurasian badger and the red fox are habitat and
thropic generalist species (Carvalho and Gomes 2001, Cavallini and Lovari 1991,
Kruuk and Parish 1985, Roper and Lups 1995, Balestrieri et al. 2004, Rosalino et
al. 2005, Amores 1975). The Egyptian mongoose is a habitat specialist but trophic
generalist (Palomares and Delibes 1991, Palomares and Delibes 1993, Zapata et al.
2007,) whereas the Iberian lynx is a highly habitat and trophic specialist species
(Palomares et al. 1991, Delibes et al. 2000, Palomares et al. 2001). The genet
meanwhile is considered between typical generalists and typical specialists (Virgós
et al. 1999).
We predicted that habitat use in lynxes and genets should be explained by
multiple variables due to their higher degree of habitat and trophic specialisation
whereas badgers and foxes should have fewer requirements and respond to single
explanatory factors such as prey availability or vegetation and landscape structure
because of their omnivorous diets and habitat generalism. For mongooses, as
habitat specialists but trophic generalists vegetation and landscape structure should
explain their pattern of habitat use. Therefore, genets and lynxes should show the
most marginal (i.e., located in extreme values of variable gradients) and narrowest
realized niches whereas foxes and badgers would have the most widespread and
widest niches among those possible. Mongooses meanwhile may show a non-
marginal niche, wider than lynxes and genets but narrower than foxes and badgers.
102
Chapter 2Chapter 2
Figure 1. The study area. Doñana National Park and its location in south-western Spain.
103
Chapter 2Chapter 2
METHODS
Study area
This study was located in Doñana National Pak (DNP), a fully protected
area in southwestern Spain (550 km237°9′N,6°26′W)(Fig.1).DNPisaflatsandy
area at sea level bordered to the south and west by the Atlantic Ocean and to the
east by the Guadalquivir River mouth. The climate is Mediterranean subhumid (i.e.,
characterized by mild wet winters and hot dry summers), with an average annual
rainfall of approximately 550 mm. Approximately half of the surface of DNP is
covered by Mediterranean scrubland, and the other half by marshland. There is
a dune system in the southern part of the scrubland area. Track censuses were
restricted to the scrubland biotope.
Mediterranean scrubland is dominated by hygrophytic species consisting
of very dense clumps of heathers Erica sp. up to 3 m high (Erico scoparidae –
Ullicetum australis and Erico ciliaris – Ullicetum (minaris) lusitanici associations)
and xerophytic species of up to 1.5 m high mainly consisting of Halimium sp, and
other such as Cistus sp., gorses Ulex sp. and rosemary Rosmarinus officinalis,
(Halimio halimifolii – Stauracanthetum genistoidis association). More mature
shrubland areas with Pistacia lentiscus and Myrtus communis can be found mainly
in the north and in the valleys of the dune system. Interpested between purely
scrubland areas there are a few and small patches of Eucalyptus camaldulensis and
pine, Pinis pinea plantations. Most of the dune valleys are also colonized by pines
Pinus pinea.
Human access into the park is regulated, but some low impact traditional
usesaremaintainedundercontrolincludingcattle-raising,apiculture,andfishing
with traditional methods. The northern and western edges of the protected area are
104
Chapter 2Chapter 2
inclosecontactwithhumansettlements,cropfieldsandahighlyintensiveusepaved
road. There are two villages close to the edge of the park limits. Populations in the
main suburban settlement (situated in the western vicinity) vary greatly between
winter and summer, as this area is mainly a summer resort occupied by about
250.000 people during the summer seasons. The other village situated in the north
is occupied by about 1,635 year-round residents, although on holydays the numbers
of people increase considerably and during a spring pilgrimage may even reach up
to one million people. There are also private large and medium-sized farms used for
agriculture as well as six visitor centers, hiking and cycling paths, recreation zones
and bird observatories in the nearby area.
Sampling protocol
The study area was divided into 69 2 x 2 km quadrants following UTM
coordinates. Quadrants were sampled throughout the wet seasons, from November
2007 to May 2008 and from October 2008 to April 2009. In each quadrant, a 3 km-
length survey route was slowly walked searching for carnivore and prey tracks. The
surveyrouteswerelocatedalongfirebreaksandcarroadsbetween2and12mwide.
Once a continuous track that crossed from one side to the other across the pathway
was detected, we georeferenced it using a global positioning system. We resampled
21 squares over the two sampling periods a second time (leaving at least 7 days
betweensamplings)untilcompleting3km,ifduringthefirstsamplingtherewere
insufficientavailablepathswithinthesquaretoachievethisdistance.Thus,aswe
had more censuses than quadrants we averaged track counts to get a single value per
quadrant. We always carried out surveys at least 3 days after any rainfall.
Potential target prey species of the carnivores studied were also sampled
105
Chapter 2Chapter 2
by trackcensuses.Preys sampledwere smallmammals (mostly long-tailedfield
mouse (Apodemus sylvaticus) according to Kufner and Moreno 1989), European
rabbits (Oryctolagus cuniculus), red partridges (Alectoris rufa), domestic cows
(Bos Taurus) and horses (Equus caballus) and wild ungulates such as the fallow
deer (Dama dama), the red deer (Cervus elaphus) and the wild boar (Sus scrofa).
Wild and domestic ungulates are rarely prey of any of the carnivore species studied
here, but we sampled them since they may provide on occasions an important food
sourceascarrionforsomeofthem.Forthefirststudyyear,thepreycensuseswere
carried out at each quadrant at the same time as the carnivore censuses. For the
second study year, we concentrated prey sampling in a one-month period to avoid
particularly apparent inter-monthly variations in abundance for some species (i.e.,
small mammals and European rabbits) (Kufner 1986, Palomares et al. 2001). Thus,
we carried out the sampling of prey tracks in April (corresponding to the intra-
annual abundance peak in both species) along transects in every quadrant sampled
for carnivore tracks. As Kilometric Abundance Index (KAI) of prey calculated
for each quadrant was obtained from censuses carried out in different months,
we applied a correction in order to homogenize prey indexes for the two study
years relativizing the first year’s KilometricAbundance Indexes toApril using
the abundance trend curve of the species that likely exhibit higher inter-monthly
variation in their relative densities (rabbits and small mammals) throughout the year
(Moreno et al. 2007, Moreno and Kufner 1988, Villafuerte et al. 1993, Villafuerte
and Moreno 1997). These transects for prey species were located in the middle of
the same routes walked for carnivore tracks, were 25 m in length and approximately
1.7 m wide (i.e., the area of a four-wheel-drive car) and separated by at least 300 m..
Thus, between 7 and 10 prey censuses were carried out per quadrant.
106
Chapter 2Chapter 2
Vegetation characteristics of quadrants were also sampled in the same places
where we censused prey species. There, we visually estimated the cover of short
shrubs (species such as Halimium sp. and Cistus sp.), tall shrubs (species such as
Erica sp., Juniperus phoenica and Pistacia lentiscus) and trees in a circle of 25 m
diameter around the sampling point. In addition we also measured other variables
related to habitat structure: average tree height and average tall- and short-shrub
height.
Predictive variables
Fourteen variables were selected to study habitat selection patterns for each
species (Table 1; see next section for arguments on variable selection).
For each quadrant, we averaged the value obtained at the sampling points both for
prey and vegetation indexes. We then used an index (Kilometre Abundance Index
of tracks (KAI) and percentage of coverage) to categorize different prey abundance
and vegetation type coverage in each quadrant.
The distance to water, distance to La Vera and distance to the anthropic edge
(see Table 1) were calculated from digitized roads, urban settlements, and water
source cover layers in DNP using a Euclidean distance-based approach (Perkin and
Conner2004,BensonandChamberlain2007).Roads includedfirewalls andcar
roads inside DNP. Urban settlements included towns and villages surrounding the
NationalPark.Watersourcesincludednaturalandartificialponds(i.e.,dugforthe
cattleinzoneswerethewatertableishigher)permanentlyflooded.Trafficindex
per quadrantwas derived fromdata on a previous study on the effect of traffic
on biodiversity in DNP (Román et al. 2010). Ecotones between pastureland and
scrublandweredefinedusinga1:10000fine-scalevegetationmapfortheyears1996-
107
Chapter 2Chapter 2
Table 1. Explanatory variables used to model relative abundance of Egyptian mongoose, Eurasian badger, and red fox and the probability of Iberian lynx and common genet presence.
Variable Code Definition UnitsVegetation tall shrub %B Mean cover of bushes per quadrant %short shrub + tall shrub %SB Mean cover of short shrub and bushes per quadrant %
trees %T Mean cover of trees per quadrant %Landscape
edges between tall shrub and pastureland eBP
Linear measure of the density of the ecotones between patches with bush cover >50% and patches with pasture cover >50%
m/ha
distance to water DW Measured in meters from the quadrant centre to the nearestpermanentfloodednaturalorartificialpond m
distance to La Vera DVMeasured in meters from the quadrant centre to the ecotone between the marshland and the Mediterranean scrubland
m
Prey availability
rabbits Ra Kilometric Abundance Index of rabbits per quadrant calculated as the mean of censuses per quadrant Tracks/km
small mammals SMKilometric Abundance Index of small mammals per quadrant calculated as the mean of censuses per quadrant
Tracks/km
Total prey TotKilometric Abundance Index of rabbits + partridges + small mammals + ungulates per quadrant calculated as the mean of censuses per quadrant
Tracks/km
distance to antropic edge DH
Measured in meters from the quadrant centre to the nearest protected areaborder influencedbyhumans(iexcluding the beach and marshland edges)
m
Human disturbance
Trafficindex TMean daily traffic (MDT), adjusted for road length(m)andtypeofroad(i) inthequadrant(=∑(MDTi * mi))
car/m
Humidity Hum Environmental humidity on census day %Observer Obs Observer who carried out censuses no units
108
Chapter 2Chapter 2
2006 obtained from the Sistema de Información Ambiental de Andalucía for the
Doñanaarea.Wereclassifiedvegetationunitsorpolygonsbasedonfourvegetation
attributes of physiognomy, species composition and density of each vegetation layer
within the polygon: (1) trees (P. pinea, Quercus suber and Eucaliptus spp.); (2)
tall shrubs (subsequently referred to as bushes) of mature Mediterranean shrubland
(e.g., P. lentiscus, M. communis) and also tall, thicket Erica spp.; (3) short shrubs
(H. halimifolium, Ulex spp., Stauracanthus genistoides) and (4) pastures. The result
wasa reclassifiedvegetationdigitalmapwith316polygonsand fourvegetation
attributes per polygon. The projection for all GIS layers and data was UTM 30S,
datum European 1950 (ED50). Hawth’s tools and Geopreocessing extension in
ArcInfo 9.3 (ESRI, Redlands, California, USA) was used to calculate distance-
based variables, to identify ecotones, and to calculate their density (Table 1).
Statistical analysis
We based our analyses on information-theoretic methods guided by the
view that ecological inference can best be approached by weighing evidence for
multiple working hypotheses simultaneously (Hilborn and Mangel 1997, Burnham
and Anderson 1998, Johnson and Omland 2004). In essence, these methods consist
of identifying a priori the alternative hypotheses for habitat selection and their
mathematical formulation, and then testing their support by fitting the relevant
equations to species distribution data and examining penalized maximum-likelihood
estimates (e.g., Fernández et al. 2003, Johnson et al. 2004).
We first specified a set of 16 candidate models that could potentially
predict species habitat use and distribution in DNP, therefore restricting the model
selection process to a few meaningful combinations of predictors of the species. For
109
Chapter 2Chapter 2
selecting predictor variables and formulating the candidate models, we considered
fiveworkinghypothesesaddressingthecriticalpointsofthedifferentspecieslife-
history traits and requirements: (1) vegetation and landscape structure, (2) prey
availability, (3) human disturbance, (4) vegetation and landscape structure + prey
availabilityand(5)global.Wedesignedfine-scalehabitatmodelsconsideringprior
knowledge of species ecology. The Eurasian badger and the red fox are generalist of
habitat use and diet (Carvalho and Gomes 2001, Cavallini and Lovari 1991, Delibes-
Mateos et al. 2008, Kruuk and Parish 1985, Roper and Lups 1995, Balestrieri et al.
2004, Rosalino et al. 2005, Amores 1975) although a certain degree of local feeding
specialisation has been reported for several badger populations and for different
resources (Kruuk and Parish 1981, Martín et al. 1995) as well as negative responses
to certain types of landscape fragmentation patterns (Virgós 2002). The Egyptian
mongoose is a trophic generalist (Zapata et al. 2007, Palomares and Delibes 1991),
but habitat specialists in Mediterranean areas (Palomares 1985, Palomares and
Delibes 1993). Mongooses actively avoid open areas and select those with dense
vegetation (bushes) for foraging and resting (Palomares and Delibes 1993). The
Iberian lynx is the most specialised in terms of habitat use and trophic niche of this
carnivore guild (Palomares et al. 1991, Ferreras et al. 1997, Delibes et al. 2000,
Palomares et al. 2001). Lynxes feed almost exclusively upon European wild rabbits
(Oryctolagus cuniculus) and need shrub vegetation patches to rest and breed.
Habitats sustaining stable lynx populations should ideally include 40% cover of
understorey vegetation half of which should be bushes (Palomares 2001, Delibes et
al. 2000). Moreover, the density of ecotones between shrubland and pastureland is
also a robust predictor of territory occurrence (Fernández et al. 2006, Fernández et al.
2007). Finally, the common genet is considered to be intermediate, falling between
110
Chapter 2Chapter 2
a typical generalist and a typical specialist species (Virgós et al. 1999). Genets need
bushes and hollow trees as sites for nocturnal and diurnal resting and feed mainly on
smallmammalssuchasthelong-tailedfieldmouse(Apodemus sylvaticus) (Delibes
1974, Palomares 1986, Palomares and Delibes 1988, Palomares and Delibes
1991). We also hypothesized that distance to permanent water resources may be an
important factor for all species particularly during the hottest months when many
surface water sources dry out. Additionally, proximity to humans and infrastructure
derivedfromtheiractivityaswellas traffic index inside theprotectedareamay
be detrimental to all the species because they produce higher mortality, degrade
the original Mediterranean ecosystems and involve a higher risk of predation or
competition with non-native carnivores (i.e., domestic dogs).
Highly correlated predictors (r > 0.6) were never included in the same
model.Inaddition,wefittedanintercept-onlyequationinordertotestimprovement
over the null model of no effect. Fitted models were compared using the Akaike
Information Criterion (AICc) and model weights (Burnham and Anderson 2002).
We ranked models by their AICc values and determined the model averaged
parameter estimates (Burham and Anderson 2002). The relative variable importance
of predictor variable j (wj) was determined as the sum of the wi across all models
where j occurs. Larger wj values indicate a higher relative importance of variable j
compared to other variables (Burnham and Anderson 2002).
Candidate model equations were fitted using Generalized Linear Mixed
Models (GLMM) in SAS 9.2 with logit-link and binomial (for lynxes and genets) or
negative binomial (for badgers, mongooses and foxes) error structure (McCullagh
and Nelder 1989). Observer and quadrant were modelled as random effects,
humidity as an additional explanatory variable (Soto et al. 2012), and the distance
111
Chapter 2Chapter 2
covered per quadrant during track censuses as an offset in all the models.
Additionally, although we selected 4 km2 grids for sampling in order to
diminish the possibility of the same individuals being sampled in neighbouring
grids (several of carnivore species sampled may have home ranges of this size;
Palomares 1994, Palomares and Delibes 1994), track counts at neighbouring
grids can be expected to show spatial autocorrelation. Therefore, we checked for
autocorrelation in our data through inspection of semi-variograms and Moran’s
I correlograms of non-spatial negative binomial and logistic generalised model
residuals. We performed analyses using an autocovariate (AC) method in the
GLIMMIX procedure (SAS 9.2, Littell et al. 1996) when spatial autocorrelation
in non-spatial generalised model residuals was detected (Hosmer and Lemeshow
2000).Theautocovariatemethodaccountsforfine-scalespatialvariationinthedata
byestimatinghowmuchtheresponseateachlocationreflectsresponsevaluesat
surrounding locations (Dormann 2007). This extra parameter is intended to capture
spatial autocorrelation originating from endogenous processes such as movement
of censused individuals between sampling sites (Smith 1994, Keitt et al. 2002,
Yamaguchi et al. 2003). For every quadrant in our study area, we calculated the
autocovariate in ArcGIS 9.3 (ESRI, Redlands, CA, USA) as:
where yj is the number of tracks at quadrant j and wj is the inverse Euclidean distance
between locations i and j. Hence, an autocovariate at location i is defined as a
weighted sum of observation records y at locations j in a neighbourhood determined
by Ni (Miller et al. 2007). The neighbourhood size may be informed by biological
112
Chapter 2Chapter 2
parameters, such as the species’ dispersal capacity (Knapp et al. 2003) if the cause
of spatial autocorrelation is known (or at least suspected). We hypothesised that
autocorrelation in our data may partially originated from movement of censused
individuals between sampling sites so we set the neighbourhood size to two quadrants
from each quadrant border to capture the average home range for all species. We
incorporated each autocovariate as an additional explanatory variable in the GLMM
models to account for the variation explained by space while maintaining the same
variable selection procedures as for spatially invariant models. Finally, we tested
autocovariate models for autocorrelation in the Pearson residuals, using Moran’s
I correlograms and semivariograms of the most parsimonious model. We used
variogram procedure in SAS 9.2 to conduct these tests.
To separate the ecological space filled by each species within an
n-dimensional hypervolume representing environmental variability at DNP, we
used an ordination method called the outlying mean index analysis (OMI) (Dolédec
et al. 2000). This method measures the marginality of a species’ habitat distribution
(niche position), i.e., the distance between the mean habitat conditions used by a
species (species centroid) and the mean habitat conditions across the study area
(origin of the niche hyperspace) as well as the species tolerance (niche breadth), i.e.,
the amplitude in the distribution of each species along the sampled environmental
gradients (Hurlbert 1978). The OMI index measures the niche position of each
species. Species with high values of OMI have marginal niches and are assumed
to be influenced by a subset of themeasured environmental variables (occur in
atypical habitats in a region), and those with low values have non-marginal niches
andindicatenospecificresponseofaspeciestotheenvironmentalvariables(occur
in typical habitats in a region); such species tend to be more common throughout
113
Chapter 2Chapter 2
the study area. The niche breadth or species tolerance is measured by an additional
variance term provided by this method. Low values of species tolerance mean that
a species is distributed across habitats with a limited range of conditions (specialist
species), while high values imply that a species is distributed across habitats with
widely varying environmental conditions (generalist species). Residual tolerance is
the variation in species occurrence not accounted for by the main gradient. Outlying
mean index is robust to unimodal, linear, or a mixture of species response curves and
is not biased against species-poor or low-abundance sites on the synthetic gradient.
Its interpretations are also robust to multicollinearity among the explanatory
variables (Dolédec et al. 2000). Wedeterminedsignificanceoftheoutlyingmean
index analysis at α=0.05basedupon aMonteCarlo simulation (Metropolis and
Ulam 1949), in which observed marginalities were statistically compared to 10,000
random permutation values of species marginalities or the null hypothesis that
species are distributed equivalently in relation to the environmental variables. The
OMI analysis was performed with the ADE4 library (Thioulouse et al. 1997) in the
‘R’ Software (R Development Core Team 2005).
RESULTS
We surveyed 471 km and 8,373 carnivore tracks were found, with foxes,
badgers and mongooses being the most frequent species (Fig. 2, Table 2). For prey,
5,000 tracks were detected on 11.6 km sampled. The most common prey species
were wild ungulates and rabbits (Table 2). The variables Ra and Tot were correlated
(r = 0.669, P < 0.001).
Based on the p-value of Moran’s I (P>0.05) and the semivariograms of
residuals, habitat use models for badgers, mongooses and foxes showed spatial
114
Chapter 2Chapter 2
Figure 2. Mapsofscaledabundances(0-1)(tracks/km)perquadrantofthefivecarnivorespecies for both study years in DNP. (a) Genetta genetta; (b) Lynx pardinus; (c) Herpestes icheumon; (d) Meles meles and (d) Vulpes vulpes.
115
Chapter 2Chapter 2
autocorrelation while it this was not detected for lynxes and genets. We therefore
fitted spatial negativebinomial generalisedmodels (i.e., using an autocovariate)
for badgers, mongooses and foxes and non-spatial logistic generalised models for
lynxes and genets.
Thebestapproximatingmodels(∆AICc<2)formongooses,badgersand
foxes belonged to the set of candidates designed with the hypothesis of vegetation
and landscape structure + prey availability as well as with the hypothesis of prey
availability (Table 3). For mongooses, top models included as predictors of their
relative abundance DV, DW, Tot, Ra, %SB, %B and eBP. Variables DW and DV had
the highest weights and were positively associated with the number of mongoose
tracks (Table 5). Variables Tot and %B were also positive and had relatively high
weights (wj > 0.4) (Table 5). For badgers, models included as predictors Ra, SM,
%SB, %B, %T, DW, V and eBP. All of these predictors were positive except for the
distance to La Vera. Ra was the variable with the highest weight (wj = 0.577) (Table
5) and was positively associated with the number of badger tracks. For the red fox
only one model was supported by data (Table 3). This model included as predictors
%B, %T, DW and SM and were all positively associated with the number of fox
tracks. Only two variables showed high weights; %B (wj = 0.809) and DW (wj =
0.871) (Table 5).
The best approximating models for lynxes and genets belonged to the set
of candidates designed with the hypothesis of vegetation and landscape structure,
vegetation and landscape structure + prey availability as well as with the global
hypothesis (Table 4). For genets, top models included as predictors %B, %T, DW,
SM, DH and T. Four variables included in top models (%B, %T, DW and SM) had
high weights (> 0.812; Table 5). %B, %T and SM were positively associated with the
116
Chapter 2Chapter 2
Posi
tive
quad
rant
sN
º tra
cks (
%)
KA
IAv
erag
eR
ange
Aver
age
Ran
geSp
ecie
#1st
2nd
1st
2nd
1st
1st
2nd
2nd
Lynx
par
dinu
s17
1774
(1.1
)73
(1.7
)0.
3±0.
90.
0-5.
40.
3 ±0
.90.
0-5.
0Vu
lpes
vul
pes
6961
4368
(65.
9)27
73 (6
4.3)
17.2
±9.1
2.4-
42.1
13.1
±10.
21.
6-77
.4H
erpe
stes
ichn
eum
on61
5099
6 (1
5.0)
617
(14.
3)3.
9±4.
60.
0-21
.23.
1±3.
70.
0-17
.2M
eles
mel
es55
5378
6 (1
1.8)
665
(15.
4)2.
6±3.
70.
0-19
.53.
3±3.
40.
0-17
.3G
enet
ta g
enet
ta29
2228
0 (4
.2)
83 (1
.9)
1.2±
2.5
0.0-
15.9
0.4±
0.8
0.0-
3.1
Felis
sp.ǂ
1221
48 (0
.7)
35 (0
.8)
0.1±
0.5
0.0-
3.9
0.2±
0.3
0.0-
1.4
Can
is fa
mili
aris
1512
38 (0
.6)
29 (0
.7)
0.2±
0.8
0.0-
5.9
0.2±
0.5
0.0-
3.4
Lutr
a lu
tra
1414
49 (0
.7)
37 (0
.9)
0.2±
0.7
0.0-
3.8
0.2±
0.4
0.0-
2.2
TOTA
L69
6566
3943
123.
2±5.
80.
0-42
.12.
6±4.
450.
0-77
.4O
ryct
olag
us c
unic
ulus
6965
3942
(43.
2)21
32 (4
2.6)
58.0
±59.
50-
1.1
32.8
±44.
60.
007-
1Sm
all m
amm
als*
4952
454
(5.0
)23
3 (4
.7)
6.7±
.40-
0.2
3.6±
.00-
0.1
Alec
tori
s ruf
a43
4484
8 (9
.3)
260
(5.2
)12
.5±2
1.6
0-0.
34.
0±.7
0-0.
07D
omes
tic u
ngul
ates
**21
2716
8 (1
.8)
210
(4.2
)2.
5±.2
0-0.
093.
2±.3
0-0.
1W
ild u
ngul
ates
***
6965
3720
(40.
7)21
65 (4
3.3)
54.7
±42.
60.
02-0
.633
.3±2
3.2
0.02
–0.4
TOTA
L69
6591
3250
0026
.9±2
7.9
0-1.
115
.4±1
7.0
0-1
*Elio
mys
que
rcin
us, A
rvic
ola
sapi
dus,
Rattu
s rat
tus,
Apod
emus
sylv
atic
us, M
us sp
p.
**Bo
s Tau
rus,
Equu
s cab
allu
s**
*Dam
a da
ma,
Cer
vus e
laph
usǂN
odistinctionbetweentrackso
fFel
is si
lves
tris
and
Fel
is c
atus
.#
Stud
y ye
ar (1
st (2
008-
2009
); 2n
d (2
009-
2010
))
Tabl
e 2.
Num
ber o
f pos
itive
2x2
km
qua
dran
ts (f
or 6
9 an
d 65
sam
pled
in e
ach
year
), to
tal t
rack
s fo
und
(Nº t
rack
s) a
nd p
erce
ntag
e re
gard
ing
tota
l tra
cks
foun
d (%
), an
d nu
mbe
r of t
rack
s pe
r km
(KAI
) for
eac
h ca
rniv
ore
and
thei
r pot
entia
l pre
y sp
ecie
s ce
nsus
ed in
Doñ
ana
Nat
iona
l Par
k du
ring
the
wet
seas
on fo
r eve
ry st
udy
year
.
117
Chapter 2Chapter 2
presence of genet tracks while DW was negatively associated. Finally, for lynxes
top models included as predictors %SB, %B, DW, eBP, V and Ra. Variables included
in the best models showed high weights (wj>0.8) (Table 5). %SB, %B, DW, eBP
and Ra were positively associated with the presence of lynx tracks while DV was
negatively associated.
Toillustratenicheseparationbetweenspecies,weusedthefirst twoaxes
of the outlying mean index analysis, which accounted for 99.95% of the total
explained environmental variability (Table 1, supplementary information). The
overall outlying mean index analysis (i.e. sensitivity of carnivores to environmental
variables)wassignificant(P < 0.0001).
ThepositionofthespeciesalongthefirsttwoOMIaxesispresentedinFig.
3. Genets and lynxes were clearly separated from mongooses, badgers and foxes.
Figure 3. OMI analysis.Ecological position of thefive carnivore species in the n-dimensionalhypervolumerepresentingfine-scaleenvironmentalvariabilityinDNP.(a)Lynx pardinus, (b) Genetta genetta, (c) Herpestes ichneumon, (d) Meles meles and (d) Vulpes vulpes.
118
Chapter 2Chapter 2
Tabl
e 3.
Sum
mar
y of
spa
tial n
egat
ive
bino
mia
l pre
dict
ive
mod
els
for
Egyp
tian
mon
goos
e, E
uras
ian
badg
er a
nd r
ed f
ox a
bund
ance
and
mod
el
sele
ctio
n es
timat
ors;
-2 lo
g(L)
=-2 log-likelihoodestim
ates;A
ICc=secondorderAkaike’sInformationCriterion;∆i=(A
ICc)i-(A
ICc)min;A
kaike
Wi=Akaikeweights.Topmodels(∆A
ICc≤2)aresh
owninbold.
Her
pest
es ic
hneu
mon
Mel
es m
eles
Vulp
es v
ulpe
sM
odel
ǂA
ICc
∆iW
iR
anki
ngA
ICc
∆iW
iR
anki
ngA
ICc
∆iW
iR
anki
ng
Nul
l mod
el
1
. Int
erce
pt o
nly
918.
5265
.55
0.00
1791
0.64
51.6
70.
0017
1277
.53
60.4
70.
0017
Vege
tatio
n an
d la
ndsc
ape
stru
ctur
e
2
. %SB
, %B,
DW
, eBP
, DV
880.
3527
.38
0.00
1488
5.97
27.0
00.
0011
1252
.19
35.1
30.
0013
3
. %B,
%T,
DW
884.
5431
.57
0.00
1588
6.03
27.0
60.
0012
1248
.20
31.1
40.
0011
4
. %SB
, %B,
DW
, DV
878.
3425
.37
0.00
1288
8.92
29.9
50.
0013
1250
.28
33.2
20.
0012
5
. DW
, DV
875.
0322
.06
0.00
1189
0.03
31.0
60.
0015
1252
.62
35.5
60.
0014
Prey
ava
ilabi
lity
6
. Ra
858.
835.
860.
035
858.
970.
000.
331
1221
.22
4.16
0.05
6
7
. SM
859.
796.
820.
028
861.
612.
640.
095
1221
.77
4.71
0.04
8
8
. Tot
862.
919.
940.
009
861.
132.
160.
114
1222
.05
4.99
0.03
9
Hum
an d
istu
rban
ce
9
. DH
, T88
5.97
33.0
00.
0016
889.
9430
.97
0.00
1412
53.6
936
.63
0.00
16
119
Chapter 2Chapter 2
Vege
tatio
n an
d la
ndsc
ape
stru
ctur
e +
prey
ava
ilabi
lity
1
0. %
SB, %
B, D
W, e
BP, D
V, R
a85
4.30
1.33
0.25
285
9.94
0.97
0.21
212
20.9
23.
860.
064
1
1. %
B, %
T, D
W, S
M85
8.92
5.95
0.02
686
0.64
1.67
0.14
312
17.0
60.
000.
421
1
2. %
SB, %
B, D
W, D
V, T
ot85
5.37
2.40
0.14
386
3.36
4.39
0.04
712
19.1
12.
050.
152
1
3. D
W, D
V, T
ot85
2.97
0.00
0.48
186
5.11
6.14
0.02
912
20.9
23.
860.
065
Glo
bal m
odel
s
1
4. %
SB, %
B, D
W, eB
P, DV
, Ra,
DH, T
857.
584.
610.
054
863.
314.
340.
046
1223
.07
6.01
0.02
10
1
5. %
B, %
T, D
W, S
M, D
H, T
863.
7810
.81
0.00
1086
4.74
5.77
0.02
812
21.3
94.
330.
057
1
6. %
SB, %
B, D
W, D
V, To
t, DH
, T85
9.56
6.59
0.02
786
7.54
8.57
0.00
1012
19.9
22.
860.
103
1
7. D
W, D
V, T
ot, D
H, T
879.
2826
.31
0.00
1389
4.15
35.1
80.
0016
1253
.63
36.5
70.
0015
ǂSeeTable1form
odeldefinitions
120
Chapter 2Chapter 2
Genets and lynxes had the highest marginality values and the lowest tolerance
to average habitat conditions of the synthetic gradients (Table 2, supplementary
information). Badgers and foxes had the lowest marginality and the highest
Table 4. Summary of non-spatial logistic predictive models for common genet and Iberian lynx occurrence, and model selection estimators; -2 log(L)=-2 log-likelihood estimates; AICc = second orderAkaike’sInformationCriterion;∆i = (AICc)i - (AICc)min; Akaike Wi = Akaike weights. Top models(∆AICc≤2)areshowninbold.
Genetta genetta Lynx pardinus
Modelǂ AICc ∆i Wi Rank AICc ∆i Wi Rank
Null model 1. Intercept only 175.23 18.59 0.00 17 151.89 44.15 0.00 15Vegetation and landscape structure 2. %SB, %B, DW, eBP, V 168.97 12.33 0.00 12 107.74 0.00 0.52 1 3. %B, %T, DW 163.47 6.83 0.02 8 141.70 33.96 0.00 10 4. %SB, %B, DW, V 170.63 13.99 0.00 13 121.26 13.52 0.00 5 5. DW, V 172.75 16.11 0.00 14 125.16 17.42 0.00 9Prey availability 6. Ra 173.89 17.25 0.00 16 151.71 43.97 0.00 14 7. SM 173.62 16.98 0.00 15 152.87 45.13 0.00 16 8. Tot 163.08 6.44 0.02 7 148.24 40.50 0.00 13Human pressure 9. DH, T 166.58 9.94 0.00 11 155.17 47.43 0.00 17
Vegetation and landscape structure + prey availability
10. %SB, %B, DW, eBP, V, Ra 164.69 8.05 0.01 9 109.53 1.79 0.21 2 11. %B, %T, DW, SM 157.25 0.61 0.34 2 142.21 34.47 0.00 11 12. %SB, %B, DW, V, Tot 162.11 5.47 0.03 4 122.23 14.49 0.00 6 13. DW, V, Tot 162.67 6.03 0.02 6 110.99 3.25 0.10 4
Global models 14. %SB, %B, DW, eBP, V, Ra, DH, T 162.32 5.68 0.03 5 110.02 2.28 0.17 3
15. %B, %T, DW, SM, DH, T 156.64 0.00 0.47 1 146.19 38.45 0.00 12
16. %SB, %B, DW, V, Tot, DH, T 161.11 4.47 0.05 3 122.93 15.19 0.00 8
17. DW, V, Tot, DH, T 164.91 8.27 0.01 10 122.72 14.98 0.00 7
ǂSeeTable1formodeldefinitions
121
Chapter 2Chapter 2
tolerance values whereas mongooses exhibited relatively low marginality and high
tolerance.Although theoverall responseofmongooseswasnonsignificant their
marginality and tolerance levels was between that of genets/lynxes and badgers/
foxes as predicted.
Ordination diagrams on the first two axes of the OMI describe the
environmentalgradientsthatbestdiscriminatedtheoccurrencesofthefivecarnivore
speciesinDNP(Fig.1,supplementaryinformation).ThefirstOMIaxiswasmost
influenced by prey availability and shrub cover far from human nuclei and the
proportion of tall shrubs and trees at the opposing end of the gradient. The second
OMIaxiswasmostinfluencedbydistancetothemainecotonebetweenmarshland
and scrubland (La Vera) and prey availability and percentage of ecotones between
pastureland and scrubland at the opposite end of the gradient (Table 2, supplementary
information). The presence of genet tracks was negatively associated with axis 1
(Fig. 3), and thus was positively related to patches distant from human nuclei and
with high tree and tall shrub coverage. Lynxes were positively associated with axis
1 and negatively with axis 2 and thus positively related to patches with high prey
availability (mainly rabbits) close to La Vera. Conversely, mongooses, badgers and
foxes were more likely to be found across a wide range of habitat types with varying
environmental conditions.
DISCUSSION
Heterogeneous landscapes offer many choices in habitat selection (i.e.
different axes over which species can differ), thus broadening opportunities for
coexistence. However, in quite homogeneous landscapes where a general habitat
type dominates, the distribution of the resources available for species is limited, and
122
Chapter 2Chapter 2
Tabl
e 5.
Var
iabl
e w
eigh
t (w
j),averagem
odelcoefficient(β)withaveragestandarderror(S
E(β))forvariablesin
cludedin
topmodels.
Varia
bles
are
des
crib
ed in
Tab
le 1
.
Her
pest
es
ichn
eum
onM
eles
mel
esVu
lpes
vul
pes
Gen
etta
gen
etta
Lynx
par
dinu
s
Vari
able
sβ
SE(β
)w
jβ
SE(β
)w
jβ
SE(β
)w
jβ
SE(β
)w
jβ
SE(β
)w
j
%B
0.17
0.12
0.48
0.33
0.26
0.45
0.35
0.25
0.81
0.10
0.06
0.95
0.32
0.12
0.90
%SB
-0.2
70.
240.
440.
140.
100.
291.
691.
20.
90
%T
0.05
0.01
0.83
eBP
-0.1
00.
080.
290.
420.
140.
241.
850.
690.
90
DW
0.08
0.04
0.96
0.25
0.19
0.46
0.29
0.22
0.87
-0.1
20.
070.
980.
480.
371.
00
V0.
610.
220.
93-0
.01
0.00
0.26
-2.5
60.
701.
00
Ra0.
160.
070.
320.
050.
030.
580.
020.
000.
38
SM0.
020.
000.
250.
000.
010.
510.
600.
180.
47
Tot
0.21
0.17
0.64
DH
0.01
0.04
0.56
T0.
010.
010.
210.
000.
000.
47-0
.16
0.14
0.56
HU
M0.
020.
011.
000.
020.
011.
000.
010.
021.
000.
020.
021.
000.
020.
021.
00
auto
cov
8.08
2.96
1.00
9.71
3.27
1.00
1.91
0.00
1.00
123
Chapter 2Chapter 2
opportunities for coexistence are likely to be the most restrictive. Hence, subtle
patterns of habitat partitioning as well as coexistence between species with different
degree of specialisation may promote ecological separation and therefore sympatry
under this scenario. A parsimony-based strategy for confronting different model
hypotheses allowed us to assess fine-scale landscape attributes linked to five
carnivore species in a protected area where species coexist in a largely similar
habitat type, the Mediterranean scrubland. These species exhibited spatial storage
or separate niches along a gradient of environmental variability according to their
degrees of specialization.
On the basis on OMI analysis, lynxes and genets are marked specialists and
showed the narrowest and most marginal niches throughout DNP’s environmental
conditions (Fig. 3). Additionally, lynxes and genets showed niche segregation
as a result of their differences in ecological preferences (Table 2 supplementary
information), which is supported by the fine-scale habitat use results. Results
of fine-scale habitat usemodels for lynxes and genetswere probably related to
the availability of prime resources for both species such as refuges and prey.
Lynxes and genets showed the most restricted habitat use pattern depending on
vegetation, landscape, prey and human-related variables. Lynxes were associated
with zones of high density of bushes likely representing areas of late-successional
Mediterranean communities, an uncommonm microhabitat in the area due to the
human transformation of the autochthonous vegetation (García-Novo and Martín-
Cabrera 2005). The density of ecotones between bushes and pastureland, which in
turn favour abundance of the main prey of lynx, the European rabbit (Fernández
et al. 2006, Fernández et al. 2007), was also a robust predictor of the presence of
lynx tracks, as well as the distance to the ecotone band between the Mediterranean
124
Chapter 2Chapter 2
scrubland La Vera, where refuge and grass availability favour a high rabbit density
in Doñana (Villafuerte and Moreno 1997, Fernández and Palomares 2000). As
expected, genets showed association to areas with high small mammal abundance
(acommunitymostlycomposedoflong-tailedfieldmouse),importantpreyitems
for the species in Mediterranean areas (Palomares and Delibes 1991), as well as
preference for pine forest areas with undergrowth bushes. These results agree with
previousfindingsfortheradio-trackedpopulationoftheIberianlynx(Palomareset
al. 2000, Palomares 2001, Fernández et al. 2003) and the common genet (Palomares
and Delibes 1988, 1991, 1994, Palomares et al. 1996) in the protected area. The low
abundance and the spatially structured distribution of both species in DNP (Fig.
2) could therefore be explained by the fact that genets and lynxes may be limited
in their home ranges to areas of simultaneous convergence of high small mammal
or rabbit availability as well as a high percentage of tall shrub cover and overall
understorey or a high percentage of tree cover, respectively.
Foxes andbadgers showedanon-specifichabitat usepattern and awide
distribution across the environmental conditions sampled in our study area as
shownbythelowweightsofpredictorinthefine-scalehabitatmodels(Table5)
and the OMI analysis results (Fig. 3) suggested. Rabbits, as well as bush density
and distance to water sources, were the most important predictors associated with
numbers of badger and red fox tracks, respectively. Badgers are trophic generalists
but in Mediterranean areas rabbits may constitute an important prey for the species
(Revilla and Palomares 2002). Foxes exhibited higher relative abundances in areas
with dense bush cover, a microhabitat that could potentially offer higher protection.
Mongooses were positively associated with dense bush cover and high
availability of different prey items as a whole (i.e., rabbits, small mammals and
125
Chapter 2Chapter 2
partridges) as well as with high availability of rabbits. It could be explained by the
fact that in spite of the trophic generalism of the species (Santos et al. 2007), wild
rabbits constitute an important prey item for the species in Mediterranean areas
(Palomares and Delibes 1991).
Foxes, badgers and mongooses overlapped in niche position, but foxes
and badgers had broader niches than mongooses (Fig. 3, Table 1 supplementary
information). In fact, mongooses exhibited and intermediate niche position and
niche breadth between that of genets/lynxes and that of badgers/foxes (Table 2,
supplementary information). These results suggest that mongooses exhibited certain
degree of habitat specialization.
The wider spatial niches exhibited by mongooses, foxes and badgers may
bearesponsetofluctuationsinresourceavailabilitycharacteristicofMediterranean
environments. Mediterraneity (Virgós and Casanovas 1999) may result in an
opportunistic behaviour in food gathering and habitat use broadening realized niches
of species and allowing coexistence between species with similar requirements by
switching to other temporarily available resources.
In summary, sympatry of carnivore species in DNP appears to be mediated by
fine-scalehabitatselectiononlyforthemostspecialistspecies(lynxesandgenets).
Badgers and foxes meanwhile did not show any clear pattern of specialization at
afinescaleasexpectedduetotheirhighlevelsofhabitatandtrophicgeneralism.
Although mongooses were more specialised in habitat selection than badgers and
foxes, for areas such as that studied here the three species can coexist with no
apparent differences in habitat use and/or realized niche segregation. Differences in
activity patterns of these three species might help to explain the coexistence among
them. Mongooses are almost exclusively diurnal in the study area (Palomares and
126
Chapter 2Chapter 2
Delibes 1993) while badgers are exclusively nocturnal (Revilla and Palomares
2002) and foxes nocturnal and crepuscular. Therefore, the carnivore community
structureinDNPseemsnottobedefinedonlybyhabitatusepartitioning.
Interspecific interactions among predators are another aspect that should
be explored and can also greatly shape the community structure. Mongooses, for
example, were strongly negatively associated with distance to La Vera, which might
be related with the high use that lynxes made of this area (Palomares et al. 1991,
Viota et al. 2012).
Underalandscapeanalysisschemethatwehavedefinedascoarse-scaled,
where Mediterranean shrubland can be considered as a homogeneous landscape,
we have been able to detect differences in suitability for carnivores in DNP after
examining fine scale habitat variables.Differences in fine-scale habitat use and
ecological separation through segregation along the ecological niche dimensions
have been shown as key mechanisms for some species allowing coexistence. Hence,
the study of spatial heterogeneity at small scales particularly in largely homogeneous
areas is essential to understanding the community structure of coexisting similar
species. To manage conservation areas that protect coexisting carnivore species,
wemustunderstandspecies-specifichabitatusepatternsandevaluatetherolethat
competitors play in determining these patterns.
ACKNOWLEDGEMENTS
This research was funded by the projects CGL2004-00346/BOS (Spanish Ministry of
Education and Science) and 17/2005 (Spanish Ministry of the Environment; National Parks Research
Program). Land-Rover España lent us two vehicles for this work. We are very grateful especially to J.C.
RivillaandS.DesniçaforassistanceduringfieldworkandtoM.Gonzálezforhervaluablesuggestions
on earlier versions of the manuscript. C. Soto was also supported by a JAE-Predoc grant from the CSIC.
127
Chapter 2Chapter 2
REFERENCES
Amores F (1975) Diet of the red fox (Vulpes vulpes) in the western Sierra Morena (south Spain).
Doñana Acta vert 2 (2):221-239.
Balestrieri A, Remonti L, Prigioni C (2004) Diet of the Eurasian badger (Meles meles) in an
agricultural riverine habitat (NW Italy). Hystrix 15 (2):3-12.
Benson JF, Chamberlain MJ (2007) Space use and habitat selection by female Louisiana black
bears in the Tensas River Basin of Louisiana. Journal of Wildlife Management 71
(1):117-126.
Brown WL, Wilson EO (1956) Character displacement. Systematic Zoology 5 (2):49-64.
Burnham KP, David R Anderson (1998) Model Selection and Inference: A Practical Information-
Theoretical Approach. New York: Springer-Verlag.
Burnham KP, David R Anderson (2002) Model Selection and Multimodel Inference: A Practical
Information-TheoreticalApproach. 2d ed. New York: Springer-Verlag.
CavalliniP,LovariS(1991)Environmental-FactorsInfluencingtheuseofHabitatintheredfox,
Vulpes vulpes. J Zool 223:323-339.
Carvalho JC, Gomes P (2001) Food habits and trophic niche overlap of the red fox, European wild
cat and common genet in the Peneda-Geres National Park. Galemys, 13 (2): 39-48.
Chesson P (2000) General theory of competitive coexistence in spatially-varying environments.
Theor Popul Biol 58 (3):211-237.
Delibes M (1974) Sobre la alimentación y biología de la gineta (Genetta genetta, L.) en España.
Doñana Acta Vertebrata, 1 (1): 143-199.
Delibes M, Rodríguez A, Ferreras P (2000) Action plan for the conservation of the Iberian lynx in
Europe (Lynx pardinus). Council of Europe Publishing, Strasbourg.
Delibes-Mateos M, de Simon JF, Villafuerte R, Ferreras P (2008) Feeding responses of the red fox
(Vulpes vulpes) to different wild rabbit (Oryctolagus cuniculus) densities: a regional
approach. Eur J Wildl Res 54 (1):71-78.
Doledec S, Chessel D, Gimaret-Carpentier C (2000) Niche separation in community analysis: A
new method. Ecology 81 (10):2914-2927.
Dormann CF, McPherson JM, Araujo MB, Bivand R, Bolliger J, Carl G, Davies RG, Hirzel A,
Jetz W, Kissling WD, Kuhn I, Ohlemuller R, Peres-Neto PR, Reineking B, Schroder B,
128
Chapter 2Chapter 2
Schurr FM, Wilson R (2007) Methods to account for spatial autocorrelation in the analysis
of species distributional data: a review. Ecography 30 (5):609-628.
Fernández N, Delibes M, Palomares F, Mladenoff DJ (2003) Identifying breeding habitat for the
Iberianlynx:Inferencesfromafine-scalespatialanalysis.EcolAppl13(5):1310-1324.
Fernández N, Delibes M, Palomares F (2007) Habitat-related heterogeneity in breeding in a
metapopulation of the Iberian lynx. Ecography 30: 431-439.
Fernández N, Delibes M, Palomares F (2006) Landscape evaluation in conservation: molecular
sampling and habitat modeling for the Iberian lynx. Ecological applications 16: 1037-1049.
Fernández N, Palomares F (2000) The selection of breeding dens by the endangered Iberian lynx
(Lynx pardinus): implications for its conservation. Biological Conservation 94 (1):51-61.
Ferreras P, Beltran JF, Aldama JJ, Delibes M (1997) Spatial organization and land tenure system of
the endangered Iberian lynx (Lynx pardinus). J Zool 243:163- 189.
Hardin G (1960) Competitive exclusion principle. Science 131 (3409):1292-1297.
García-novo, F. y C. Martín-Cabrera -2005- Doñana: Agua y Biosfera. Doñana 2005, Confederación
hidrográficadelGuadalquivir,MinisteriodeMedioAmbiente.Madrid.
Gause, G.F. (1934). The struggle for existence. Baltimore, MD: Williams & Wilkins.
Hilborn R, Mangel M (1997) The ecological detective. Confronting models with data. Monographs
in Population Biology 28: I-XVII, 1-315.
Hosmer, D.W., Lemeshow, S., 2000. Applied Logistic Regression.Wiley, NewYork.
Hurlbert SH (1978) Measurement of niche overlap and some relatives. Ecology 59 (1):67-77.
Hutchinson GE (1959) Homage to Santa-Rosalia or why are there so many kinds of animals. Am
Nat 93 (870):145-159.
Johnson CJ, Seip DR, Boyce MS (2004) A quantitative approach to conservation planning: using
resource selection functions to map the distribution of mountain caribou at multiple spatial
scales. Journal of Applied Ecology 41 (2):238-251.
Johnson JB, Omland KS (2004) Model selection in ecology and evolution. Trends Ecol Evol 19
(2):101-108.
Keitt TH, Bjornstad ON, Dixon PM, Citron-Pousty S (2002) Accounting for spatial pattern when
modeling organism-environment interactions. Ecography 25 (5):616-625.
Knapp RA, Matthews KR, Preisler HK, Jellison R (2003) Developing probabilistic models to
129
Chapter 2Chapter 2
predict amphibian site occupancy in a patchy landscape. Ecol Appl 13 (4):1069-1082.
Kruuk H, Parish T (1981) Feeding specialization of the european badger Meles meles in scotland.
J Anim Ecol 50 (3):773-788.
Kruuk H, Parish T (1985) Food, food availability and weight of badgers (Meles meles) in relation
to agricultural changes. Journal of Applied Ecology 22 (3):705-715.
Kufner MB (1986) Tamaño actividad, densidad relativa y preferencias de hábitat de los pequeños
y medianos mamíferos de Doñana, como factores condicionantes de su tasa de predación.
Tesis doctoral. Universidad Autónoma de Madrid.
Kufner MB, Moreno S (1989) Abundancia y amplitud de los desplazamientos de Apodemus
sylvaticusencuatrobitopesdeDoñanaquedifierenencoberturavegetal.Doñana
Acta Vertebrata 16 (1):179-181.
LittellRC,GAMilliken,WWStroup,RDWolfinger(1996)SASSystemforMixedModels,Cary,
NC: SAS Institute Inc.
Martin J, Basille M, Van Moorter B, Kindberg J, Allainé D, Swenson JE (2010) Coping with
human disturbance: spatial and temporal tactics of the brown bear (Ursus arctos).
Canadian Journal of Zoology 88 (9):875-883.
Martin R, Rodriguez A, Delibes M (1995) Local feeding specialization by badgers (Meles meles)
in a mediterranean environment. Oecologia 101 (1):45-50.
McCullagh, P. and Nelder, J. A. (1989). Generalized Linear Models, 2nd edition. London:
Chapman and Hall.
Metropolis N, Ulam S (1949) The monte carlo method. J Am Stat Assoc 44 (247):335-341.
Miller J, Franklin J, Aspinall R (2007) Incorporating spatial dependence in predictive vegetation
models. Ecol Model 202 (3-4):225-242.
MorenoS,BeltranJF,CotillaI,KuffnerB,LaffiteR,JordanG,AyalaJ,QuinteroC,JimenezA,
Castro F, Cabezas S, Villafuerte R (2007) Long-term decline of the European wild
rabbit (Oryctolagus cuniculus) in south-western Spain. Wildl Res 34 (8):652-658.
Moreno S, Kufner MB (1988) Seasonal patterns in the wood mouse-population in Mediterranean
scrubland. Acta Theriol 33 (1-11):79-85.
PalomaresF(1994)Sitefidelityandbodymassonhome-rangesizeofEgyptianmongooses.
Canad J Zool. 72, 465-469.
130
Chapter 2Chapter 2
Palomares F (1986) Ecología de la gineta y del meloncillo en el Parque Nacional de Doñana. Tesis
de licenciatura. Universidad de Granada. 186 pp.
Palomares F (2001) Vegetation structure and prey abundance requirements of the Iberian lynx:
implications for the design of reserves and corridors. Journal of Applied Ecology
38 (1):9-18.
Palomares F, Delibes M (1988) Time and Space Use by Two Common Genets (Genetta genetta) in
the Doñana National Park, Spain. J Mammal 69 (3):635-637.
Palomares F, Delibes M (1991) Dieta del meloncillo, Herpestes ichneumon, en Coto del Rey, norte
del Parque Nacional de Doñana. Doñana Acta Vertebrata, 18: 187-194.
Palomares F, Delibes M (1991). Ecología comparada de la gineta, Genetta genetta (L.) y el meloncillo,
Herpestes ichneumon, (L.) (Mammalia, Viverridae) en Doñana (SO de la Península
Ibérica). Bol. R. Soc. Esp. Hist. Nat. (Secc. Biol.), 87: 257-266.
Palomares F, Delibes M (1993) Key habitats for Egyptian mongooses in Doñana National-park,
south-western spain. Journal of Applied Ecology 30 (4):752-758.
Palomares F, Delibes M (1993) Resting ecology and behavior of Egyptian mongooses (Herpestes
ichneumon) in southwestern spain. J Zool 230:557-566
Palomares F, Delibes M (1994) Spatiotemporal ecology and behavior of european genets in
southwestern spain. J Mammal 75 (3):714-724.
Palomares F, Delibes M, Ferreras P, Fedriani JM, Calzada J, Revilla E (2000) Iberian lynx in a
fragmented landscape: Predispersal, dispersal, and postdispersal habitats. Conserv Biol 14
(3):809-818.
Palomares F, Delibes M, Revilla E, Calzada J, Fedriani JM (2001) Spatial ecology of Iberian lynx
and abundance of European rabbits in southwestern Spain. Wildl Monogr (148):1-36.
Palomares F, Ferreras P, Fedriani JM, Delibes M (1996) Spatial relationships between Iberian lynx
and other carnivores in an area of south-western Spain. Journal of Applied Ecology 33
(1):5-13.
PalomaresF,RodriguezA,LaffitteR,DelibesM(1991)ThestatusanddistributionoftheIberian
Lynx Felis-Pardina (Temminck) in Coto Doñana area, Sw Spain. Biological Conservation
57 (2):159-169.
Revilla E, Palomares F (2002) Does local feeding specialization exist in Eurasian badgers? Can J
131
Chapter 2Chapter 2
Zool-Rev Can Zool 80 (1):83-93.
Román J, A Barón, E Revilla (2010) Evaluación de los efectos del tránsito a motor sobre especies
y comunidades de interés en el Espacio Natural de Doñana. Consejería de Medio
Ambiente, Junta de Andalucía. Estación Biológica de Doñana, CSIC. 236 pp.
Roper TJ, Lups P (1995) Diet of badgers (Meles meles) in central switzerland - an analysis of
stomach contents. Z Saugetierkd-Int J Mamm Biol 60 (1):9-19.
Rosalino LM, Loureiro F, MacDonald DW, Santos-Reis M (2005) Dietary shifts of the badger
(Meles meles) in Mediterranean woodlands: an opportunistic forager with seasonal
specialisms. Mamm Biol 70 (1):12-23.
Rosenzweig ML, Abramsky Z, Brand S (1984) Estimating species interactions in heterogeneous
environments. Oikos 43 (3):329-340.
Santos MJ, Pinto BM, Santos-Reis M (2007) Trophic niche partitioning between two native and
two exotic carnivores in SW Portugal. Web Ecology 7:1-62.
Smith PA (1994) Autocorrelation in logistic-regression modeling of species distributions. Glob
Ecol Biogeogr Lett 4 (2):47-61.
Soto C, Desniça S, Palomares F (2012) Nonbiological factors affecting track censuses: implications
for sampling design and reliability.
Thioulouse J, Chessel D, Dolédec S, Olivier J (1997) Ade-4: a multivariate analysis and graphical
display software. Statistics and Computing, 7:75-83.
VillafuerteR,KufnerMB,DelibesM,MorenoS(1993)Environmental-factorsinfluencingthe
seasonal daily activity of the European rabbit (Oryctolagus cuniculus) in a Mediterranean
area. Mammalia 57 (3):341-347.
VillafuerteR,LazoA,MorenoS(1997)Influenceoffoodabundanceandqualityonrabbit
fluctuations:ConservationandmanagementimplicationsinDoñanaNationalPark(SW
Spain). Rev Ecol-Terre Vie 52 (4):345-356.
Villafuerte R, Moreno S (1997) Predation risk, cover type and group size in European rabbits in
Doñana (SW Spain).Acta Theriologica, 42 (2): 225-230.
Viota M, López-Bao JV, Palomares F, Rodríguez A (2012) Shift in microhabitat use as a mechanism
allowing the coexistence of victim and killer carnivore predators.
132
Chapter 2Chapter 2
Virgós E (2002) Are habitat generalists affected by forest fragmentation? A test with Eurasian
badgers (Meles meles) in coarse-grained fragmented landscapes of central Spain. J Zool
258 (3):313-318.
Virgós E, Casanovas JG (1999) Environmental constraints at the edge of a species distribution, the
Eurasian badger (Meles meles L.): a biogeographic approach. Journal of Biogeography 26
(3):559-564.
Virgós E, Llorente M, Cortes Y (1999) Geographical variation in genet (Genetta genetta L.) diet: a
literature review. Mammal Review 29 (2):117-126.
Yamaguchi N, Rushton S, Macdonald DW (2003) Habitat preferences of feral American mink in
the Upper Thames. J Mammal 84 (4):1356-1373.
Zapata SC, Travaini A, Ferreras P, Delibes M (2007) Analysis of trophic structure of two carnivore
assemblagesbymeansofguildidentification.EurJWildlRes53(4):276-286.
133
Chapter 2Chapter 2
Supplementary information
Table 1. Results of the outlying mean index analysis depicting relationships of occurrences of Carnivora species to the suite of environmental variables measured across sampling sites in DNP. Inertia = variance or weighted sum of squared distances to the origin of the environmental axes; OMI = outlying mean index (marginality) or the deviation of a particular species’ distribution from the overall mean habitat conditions (origin of outlying mean index axes), described by the environmental variables; Tol = tolerance index, which is analogous to ‘‘niche breadth’’ or spatial variance of an organism’s ‘‘niche’’ across the measured environmental variables—a function of all sampling sites with which the species is associated; RTol = residual tolerance. Italicized terms representthepercentagesofvariabilitycorrespondingtoaspecificstatistic.P = frequency based on number of random permutations (out of 10,000) that yielded a higher value than the observed outlying mean index (P ≤0.05indicatesasignificantinfluenceoftheenvironmentalvariablesforaspecies).
Species Inertia OMI Tol Rtol OMI Tol Rtol PLynx pardinus 13.22 5.19 1.35 5.73 39.30 17.40 43.30 0.00040Genetta genetta 15.66 4.13 1.61 6.54 26.40 31.90 41.70 0.00010Herpestes ichneumon 9.39 0.35 2.30 7.45 2.70 18.00 79.30 0.09329Meles meles 11.79 0.25 1.69 9.83 3.00 13.60 83.40 0.06429Vulpes vulpes 10.89 0.04 4.99 9.51 0.30 12.40 87.30 0.63804
Table 2. Loadingsof environmental variables for thefirst two axes of the outlyingmean indexanalysis. Values represent the best linear combination that explains occurrences of Carnivora at DNP. Variables are described in Table 1.
Variable Axis 1 Axis 2%SB 0.2014 0.0251%B -0.2361 -0.1876%T -0.2138 0.1877DH -0.5163 -0.1776Traf 0.1279 0.3091eBP 0.0237 -0.5020DW 0.2807 0.2649DV -0.0605 0.5255Ra 0.2431 -0.3693SM 0.0365 0.0165Tot 0.2635 -0.2575Eigenvalue 0.3766 0.0003Percentage variance explained 0.9986 0.0009
134
Chapter 2
SM SB
DW
DH
eBP
Ra
Tot
Traf
DV
T
B
Axis 1
Figure 1.Contributionofenvironmental,humanandpreyvariablestothefirstaxesproducedbythe OMI analysis. The length of the arrow describes the relative importance of each variable in the analysis, and the direction of the arrow indicates among-variable correlations.
135
CHAPTER 3
Species abundances in a community of sympatric carnivores: a trade-off between habitat selection and interspecific interactions?
Abundancia de especies en una comunidad de carnívoros
simpátridos: ¿existe un equilibrio entre la selección de
hábitat y las interacciones interespecíficas?
136
Chapter 3
137
Chapter 3
ABSTRACT
Inacarnivoreguild,theoccurrenceoflargerspeciescanexertsasignificant
effect on the relative abundances of smaller guild members, whereas similar-sized
species may potentially not exhibit any density-dependent relationships. When
modelling habitat use, the species relative abundance could be therefore a misleading
proxy of habitat quality if a larger predator may be limiting their abundance in the
potentially best optimal habitat.
We analyzed the potential effect of the co-occurrence of sympatric
carnivores in the habitat use pattern of five medium-sized carnivore species
(Iberian lynx, Eurasian badger, common genet, Egyptian mongoose and red fox) in
a Mediterranean protected area in southwest Spain (Doñana National Park; DNP),
by logistic and regression generalized linear mixed models. We carried out track
censuses during 2007-2008 and 2008-2009 within 2 x 2 km2 quadrants and we
found that the relative abundance of foxes, mongooses and genets were negatively
affected by the occurrence of the top predator of the carnivore guild, the Iberian
lynx. However, abundance of genets and mongooses were positively correlated
with the occurrence of mongooses and badgers, respectively, also improving the
explanatory power of the habitat use models of those species. Intraguild predation
as well as coincidence in the use of resources between species could explain the
patterns of co-occurrence found. We suggest that the particular identity of predator
species within a guild strongly affects the relative competitive advantage of each
guild member and we propose that relative abundances of predators at the same
trophic level should not be overlooked and included as covariates when studying
carnivore community structure and habitat use.
138
Chapter 3Chapter 3
RESUMEN
En un gremio de carnívoros, la ocurrencia de las especies más grandes puede
ejercer un efecto significativo sobre las abundancias relativas de los miembros más
pequeños de la comunidad, mientras que las especies de tamaño similar no deben presentar
aparentemente ninguna relación denso-dependiente entre ellas. En consecuencia, cuando
se modela el uso del hábitat por una especie, la abundancia relativa de la misma puede
ser un indicador erróneo de la calidad de hábitat si un depredador superior está limitando
la abundancia de la especie en su mejor hábitat potencial.
En este estudio analizamos el efecto potencial de la co-ocurrencia de carnívoros
simpátridos en el patrón de uso del hábitat de cinco especies de carnívoros de mediano
tamaño (lince ibérico, tejón, gineta, meloncillo y zorro) en un área protegida en el suroeste
de España (Parque Nacional de Doñana), a través de modelos lineares generalizados
mixtos logísticos y de regresión. Llevamos a cabo censos de rastros durante los años
2007-2008 y 2008-2009 en cuadrículas de 2 x 2 km2 y encontramos que la abundancia
relativa de zorros, meloncillos y ginetas se vio afectada negativamente por la ocurrencia
del carnívoro superior del gremio, el lince ibérico.
Sin embargo, la abundancia de ginetas y meloncillos se correlacionó positivamente
con la ocurrencia de meloncillos y tejones, respectivamente, mejorando también el poder
explicativo de los modelos de hábitat para dichas especies. La depredación intragremial
así como la coincidencia en el uso de los recursos entre especies pueden explicar dichos
patrones de co-ocurrencia detectados. Sugerimos que la identidad particular de una especie
de depredador dentro de un gremio afecta en gran medida su ventaja competitiva dentro
de la comunidad y proponemos que la abundancia relativa de predadores pertenecientes
almismoniveltróficodebeincluirsecomovariableexplicativaenestudiosdeusodel
hábitat y estructura de las comunidades de mamíferos carnívoros.
139
Chapter 3Chapter 3
INTRODUCTION
The statement that density of a species in a given habitat is a direct measure
of the quality of that habitat for the species is a widely accepted assumption in
habitat use studies (but see VanHorne 1983). Nevertheless, some species may not
exhibit the highest densities in their potentially best optimal habitats according to
the species life-history traits, as their relative abundances and pattern of habitat use
canbeaffectedbyinterspecificinteractionswithotherspeciesatthesametrophic
level (Case and Gilpin 1974, Estes and Palmisano 1974, Menge and Sutherland
1987, Durant 1998).
Among mammal carnivores, body size influences the outcomes of
interspecificinteractions(PalomaresandCaro1999).Withtheexceptionofsome
species that prey in packs (e.g., Rogers and Mech 1981, Paquet and Carbyn 1986),
interference competition is inflicted by a larger carnivore species on a smaller
one killing it (termed as intraguild predation (Polis et al. 1989)) or excluding it
from habitat patches or prey carcasses (Palomares and Caro 1999). These highly
asymmetrical interactions mediated by body size are common (e.g. Fedriani et al.
1999, Kamler et al. 2003, Durant et al. 2010) being the smaller species almost
invariably the loser due to their subordinate position (Palomares and Caro 1999,
Prugh et al. 2009, Roemer et al. 2009). Nevertheless, smaller members of a guild
may also kill cubs, young or subadult individuals of the larger species (Palomares
and Caro 1999). In the absence of predator species, small carnivores should be
ideally distributed based on habitat quality and preferred food availability (Van der
Meer and Ens 1997, Roemer et al. 2009), but in most natural communities smaller
species are potentially subject to top-down effects that mediate their ability to use
preferred habitat and limit their access to high-quality foraging areas (Palomares
140
Chapter 3Chapter 3
and Caro 1999, Ritchie and Johnson 2009). Moreover, in some carnivores,
intraguild predation has a considerable impact on small carnivore’ mortality rates
and consequently a negative effect on the relative abundance of the species in
the presence of the larger predator (Ralls and White 1995, Sovada et al. 1998).
In such cases, due to these potential predation-driven direct effects or fear-driven
indirect effects, the relative abundances of some predators may not be related to
prey availability or vegetation type. Densities of smaller members of a guild may
represent a trade-off between suitable areas that satisfy their basic requirements and
areas where probability of encounters with larger predators are null or lower (i.e., to
diminish the risk of being killed).
In this paper, we examine the potential effect of the co-occurrence of different
sympatric carnivores on the habitat use pattern of five small andmedium-sized
carnivore species in a Mediterranean protected area in southwest Spain, namely
DoñanaNationalPark(DNP).Onpreviouslyfittedfine-scalehabitatusemodelsfor
aguildoffivecarnivorespeciesweanalysedwhethertheinclusionoftherelative
abundances of any carnivore species improves the explanatory power of habitat
models (Johnson and Omland 2004, Burnham and Anderson 2002). The studied
carnivore guild is composed by the Iberian lynx (Lynx pardinus, 9-15 kg), Eurasian
badger (Meles meles, 7-8 kg), the red fox (Vulpes vulpes, 5-7 kg), the Egyptian
mongoose (Herpestes ichneumon, 3 kg) and the common genet (Genetta genetta,
2 kg). In DNP these species form a community of continuum degree of habitat
and trophic specialization predators with different degree of habitat use overlap.
It has been described aggressive interactions between some of these carnivores in
DNP, which always involved lynxes as the top predator. The Iberian lynx uses to
kill genets, mongooses and foxes (Palomares et al. 1996, Fedriani 1997), but no
141
Chapter 3Chapter 3
aggressive interactions between lynx and badger has been reported. On the other
hand, aggressive interactions between badgers, foxes, mongooses and genets have
been unreported, although according to the theory differences in body size among
some of these species could support such interactions (Palomares and Caro 1999,
Fedriani et al. 2000).
Specifically,wetestedthehypothesisthatinourcarnivoreguildthespatial
occurrenceoflargerspeciesexertsasignificanteffectontherelativeabundances
of other smaller guildmembers after controlling by species-specific habitat use
patterns. Based in our previous knowledge on the reported aggressive interaction
between species in the study area and in the general knowledge of the effect of
body size on carnivore interactions (reviewed in Palomares and Caro 1999), our
predictions were those outline in Figure 1: a) we expected a negative effect of the
lynx presence and/or abundance on the relative densities of foxes, mongooses and
genets; b) abundance of foxes and badgers should negatively affect abundance of
mongooses and genets and c) it should be no effect between badgers and foxes,
badgers and lynxes and mongooses and genets.
METHODS
Study area
DNP isa fullyprotectedandflat sandyareaat sea level in southwestern
Spain (550 km2 37°9′N,6°26′W), locatedon thewestbankof theGuadalquivir
River mouth. The climate is Mediterranean sub-humid, characterized by dry, hot
summers and mild, wet winters. Average annual rainfall is 500-600 mm. There are
three main biotopes in the park: scrubland, dunes, and marsh (Valverde 1958). The
dune area is situated at the western border of the protected area limited by the
142
Chapter 3Chapter 3
Atlantic Ocean and the marsh area at the northern and eastern borders limited by the
Guadalquivir River. The Mediterranean scrubland represents approximately half of
the National Park surface area and is mainly characterised by heterogeneous patches
of xerophytic species such as Halimium sp. and Cistus sp., and hydrophytic species
such as Erica sp., with some patches of Juniperus phoenica and Pistacia lentiscus
shrubs. Interspersed among the scrubland are scattered cork oak trees (Quercus
suber) and wild olive trees (Olea europea), and a few patches of pine Pinus pinea
Figure 1. Sketch of the possible expected interrelationships among the different carnivore species studied here. Single-headed arrow and straight lines represent aggressive interactions that have been previously reported in the literature for those species in the study area; single-headed arrow and dashed lines represent potential aggressive interactions based in the body size of the involved species but that have been previously unreported; and double-headed arrow and dashed lines represent potential neutral expected interactions between similar-sized species in the DNP carnivore guild.
143
Chapter 3Chapter 3
and eucalyptus Eucalyptus sp. plantations.
In addition to the studied species (i.e., red fox, Eurasian badger, Egyptian
mongoose, common genet and Iberian lynx), there are other carnivores such as
least weasel (Mustela nivalis), European polecat (Mustela putorius), Eurasian otter
(Lutra lutra) and wild cat (Felis silvestris), but they were excluded from the study
because of their scarcity in the area.
Sampling protocol
To assess the co-occurrence of carnivore species and to obtain relative
densities for foxes, badgers, lynxes, mongooses and genets we divided the study
area into 69 2 x 2 km quadrants following UTM coordinates (Figure 2). Quadrants
were sampled throughout the wet seasons of 2007-2008 and 2008-2009. In each
quadrant, a 3 km-length survey route was slowly walked searching for carnivore
tracks. Thesurveyrouteswerelocatedalongfirebreaksandcarroadsbetween2and
12 m wide. Once a continuous track that crossed from one side to the other across
the pathway was detected, we georeferenced it using a global positioning system.
For the two sampled seasons altogether we resampled 21 squares (leaving at least 7
days between samplings) a second time until completing 3 km, because there were
insufficientavailablepathswithinthesquaretoachievethisdistance.Thus,aswe
had more censuses than quadrants we averaged track counts to get a single value per
quadrant. We always carried out surveys at least 3 days after any rainfall.
Data about prey availability and vegetation variables included in habitat
models were obtained through track censuses in transects of 25 m in length and
approximately 1.7 m wide and from visually estimates in 25 m diameter circle
situated in the same places where transects for prey (see Soto et al. 2012 for more
details).
144
Chapter 3Chapter 3
Data analysis
To test the effect of the abundance of other species on the abundance of a
givenspeciesaccordingthisoutlinedinFigure1,wefirstfittedfine-scalehabitat
use models for each species (Chapter 2). After, we added the relative abundance
(Kilometric Abundance Index; KAI) or presence of the different carnivore species
to the best approximating habitat model describing abundance of each species
(hereafter null models) and observed whether models improved (i.e. a decrease of
the Akaike Information Criterion corrected for small sample size; AICc). According
totheinformation-theoreticapproach,modelswith∆AICc < 2 from the model with
the lowest AICc are considered to be virtually the same plausible models (Burnham
and Anderson 2002). Hence, in spite of the existence of certain model selection
uncertainty in multimodel inference, statistical models that decrease the AICc of
a reference model more than 2 units could be considered as more parsimonious
modelsandtobestfitthedata.Thus,forthespecificaimsofourstudy,when ∆AICc
< -2andmodelcoefficient(β) was negative we considered that there were support
to the predictions about the abundance or presence of a given species affected the
abundance or presence of the target species; when ∆AICc < -2andmodelcoefficient
(β) was positive we considered that both species could coincide in habitat resources
used,butnonegative interactionoccurredbetween them;andfinallywhen -2>
∆AICc < 2 we considered that abundance or presence of a given species did not
affect the abundance or presence of the target species. Note that if our hypothesis
is right, the inclusion of the relative abundance of presence of smaller species in
the model of habitat selection of larger species should have either no effect or if
any (i.e.,∆AICc > 2), β should be positive.Hence,we included as a control of
hypothesis all possible carnivore species in null models for each target species to
145
Chapter 3Chapter 3
test if this was the case.
Nullmodelsweredefinedbasedonprevioushabitatusestudiesforeachspeciesin
the DNP (Chapter 2). These models included different variables related to landscape
features, vegetation type, prey availability and human disturbance for each species
(see Table 1 for variable description). The general null model for the occurrence of
mongooses (HI), foxes (VV), badgers (MM) and for the presence of genets (GG)
and lynxes (LP) at site jwasspecifiedas:
logit(HI j)=φHI + α1DW j + α2DV j + α3Tot j + α4Hum j + ζHI autocovj
logit(VV j)=φVV + β1%B j + β2%T j + β3DW j + β4SMj + β5Humj + ζVV autocovj
logit(MM j)=φMM+γ1%B j+γ2%SB j+γ3eBP j+γ4DVj+γ5DW j+γ6Raj +γ7Hum j + ζMM autocovj
logit(GG j)=φGG+δ1%B j+δ 2%T j+δ3DW j+δ4SMj+δ5DH j+δ 6Traf j +δ7Hum j
logit(LP j)=φLP+ε1%B j+ε2%SB j+ε3eBP j+ε4DVj+ε5DW j+ε6Raj +ε7Hum j
whereφiisthespecies-specificintercept,thecoefficientsα1,..., α4, β1,..., β5,γ1,...,
γ7, δ1,..., δ7 and ε1,..., ε7 represent effects of the associated explicative variables
(Table 1) on species i, and ζi is the effect of the autocovariate on species i (see text
below).
Candidatemodel equationswere fitted usingGeneralizedLinearMixedModels
(GLMM) with binomial (for genets and lynxes) or negative binomial (for mongooses,
badgers and foxes) error structure using the glmmADMB and lmer library (Fournier
et al. 2012, Skaug et al. 2012) respectively in R 2.15.2 (R Development Core Team
2008). The dependent variable was the number of tracks/km. Observer and quadrant
were modelled as random effects, humidity as a covariate (Soto et al. 2012), and the
distance covered per quadrant during track censuses as an offsset. We incorporated
146
Chapter 3Chapter 3
an autocovariate as an additional explanatory variable in the GLMM models to
accountforfine-scalespatialvariationinthedatabyestimatingthedegreetowhich
the response at each location reflects response values at surrounding locations
(Dormann 2007). This extra parameter is intended to capture spatial autocorrelation
originating from endogenous processes such as movement of censused individuals
between sampling sites (Smith 1994, Keitt et al. 2002, Yamaguchi et al. 2003).
Table 1. Variablesusedtodefinefine-scalehabitatusenullmodelsforbadgers,mongooses,foxes,genets and lynxes.
Variable Code Definition Units
tall shrub %B Mean cover of bushes per quadrant calculated by averaging values obtained at the different sampling points of vegetation censuses %
trees %T Mean cover of trees per quadrant calculated by averaging values obtained at the different sampling points of vegetation censuses %
distance to water DW
Measured in meters using a Euclidean distance-based approach fromthequadrantcentretothenearestpermanentlyfloodednaturalorartificialpond(i.e.dugforthecattleatzoneswerethewatertableis higher) in a digitized water sources cover layer of DNP
m
distance to La Vera DV
Measured in meters from the quadrant centre to the ecotone between the marshland and the Mediterranean scrubland (locally called La Vera)
m
small mammals SM
Kilometric Abundance Index of small mammals per quadrant calculated as the mean between prey indexes at the different sampling points of prey censuses per quadrant
Tracks/km
rabbits RaKilometric Abundance Index of rabbits per quadrant calculated as the mean between prey indexes at the different sampling points of prey censuses per quadrant
Tracks/km
distance to antropic edge
DHMeasured in meters from the quadrant centre to the nearest protected area border influenced by humans (i.e., excluding the beach andmarshland edges)
m
Trafficindex T
Meandailytraffic(MDT),adjustedforroadlength(m)andtypeofroad(i)inthequadrant(=∑(MDTi*mi))derivedfromRománetal. 2010
car/m
Humidity HumEnvironmental humidity on census day obtained from a meteorological station located inside DNP. http://icts.ebd.csic.es %
Observer Obs Observer who carried out censuses no units
147
Chapter 3Chapter 3
RESULTS
Nullmodelsoffoxes,mongoosesandgenetssignificantlyimproved(∆AICc
< -2) after including the occurrence and/or presence of lynxes (Table 2). The
three species were less abundant in areas where lynxes exhibited a higher relative
abundance (Fig. 3a, c and e). Habitat models of foxes, genets (Fig. 3b) and badgers
also improved after including the relative density of mongooses (Table 2) but the
Figure 2. Map of the DNP showing its location in southwest of the Iberian Peninsula and the 2×2 km2 quadrants where carnivore track censuses were carried out during the wet seasons of 2007-2008 and 2008-2009. The dashed zone represents the marshland area of the DNP.
148
Chapter 3Chapter 3
three species exhibited higher relative densities in areas where mongooses were
also more abundant.
The inclusion of the presence of genets and the relative abundance of badgers
(Fig. 3d) also improved the habitat model of mongooses that were more abundant
in areas where genets were detected or where the occurrence of badgers was higher
(Table2).Theinclusionofanyofthesmallermembersoftheguildsignificantly
improved the habitat model of the Iberian lynx (Table 2).
DISCUSSION
Our results support the hypothesis that in a carnivore guild, the occurrence
ofsomespeciesexertsasignificanteffectontherelativeabundanceofotherguild
membersaftercontrollingbyspecies-specifichabitatusepatterns.Aswepreviously
hypothesized, the relative abundance of foxes, mongooses and genets in DNP was
negatively affected by the presence or occurrence of lynxes, the larger reported
intraguild predator for all of them (Palomares and Caro 1999). In fact, it had been
previously described for Doñana as foxes, mongooses and genets avoided temporal
or spatially highly used areas by lynxes (Palomares et al. 1996, 1998, Fedriani et
al. 1999, Viota et al. 2012). Similar results have been obtained in other carnivore
communities involving African wild dogs Lycaon pictus and cheetahs Acinonyx
jubatus with lions Panthera leo and spotted hyenas Crocuta crocuta (Laurenson
1995, Creel and Creel 1996, Durant 2000).
Also as predicted, the relative abundance of smaller species did not affect
the relative abundance of larger ones; in our case lynxes, badgers and foxes.
However, although theory predicts that larger species such as foxes and badgers
might negatively interact with genets and mongooses, results did not confirm
149
Chapter 3Chapter 3
Table 2. Results from negative binomial (foxes, badgers and mongooses) and logistic regression models (genets and lynxes) investigating the effects of the occurrence and/or presence of predator species on each carnivore species in the DNP. We report the small sample-size–adjusted Akaike’s information criteria (AICc),thedifferenceinAICcbetweeneachmodelandthenullmodel(∆AICc), the model coefficient (β) with standard error (SE) as well as its P-value of those models that decreasedthenullmodel≥2.VariablesaredescribedinTable1.
Species Models AICc ∆AICc β SE(β) P-value
Genetta genetta0. Intercept only 175.23 12.55 . . .1. Null model%B + %T + DW + SM + DH + Traf + Hum + autocov 162.68 0.00 . . .2. Null model + KAIl 159.75 -2.94 -1.14 0.72 0.053. Null model + KAIb 162.07 -0.61 . . .4. Null model + KAIf 161.07 -1.61 . . .5. Null model + KAIIm 153.01 -9.67 0.26 0.08 0.006. Null model + Lynx 160.62 -2.06 -1.13 0.59 0.08
Herpestes ichneumon
0. Intercept only 918.52 38.76 . . .1. Null modelDW + DV + Tot + Hum + autocov 879.76 0.00 . . .2. Null model + KAIl 877.23 -2.53 -0.31 0.14 0.033. Null model + KAIb 874.89 -4.87 0.06 0.02 0.014. Null model + KAIf 879.74 -0.02 . . .5. Null model + KAIg 879.63 -0.13 . . .6. Null model + Lynx 881.56 1.80 . . .7. Null model + Genet 863.83 -15.93 0.82 0.19 0.00
Vulpes vulpes0. Intercept only 1277.53 24.66 . . .1. Null model%B + %T + DW + SM + Hum + autocov 1252.87 0.00 . . .2. Null model + KAIl 1253.99 1.12 . . .3. Null model + KAIb 1254.09 1.22 . . .4. Null model + KAIm 1250.46 -2.41 0.03 0.01 0.035. Null model + KAIg 1254.84 1.97 . .6. Null model + Lynx 1249.69 -3.18 -0.27 0.12 0.027. Null model + Genet 1250.91 -1.96 . . .
150
Chapter 3Chapter 3
this prediction. Contrarily, positive correlations between some of these species
(mongooses with badgers and foxes) were found. The lacks of negative interactions
may be explained by the fact that the degree of intra-guild interference is thought
to depend on body-size differences: at small differences, attacks are less likely to
occur (Buskirk 1999, Donadio and Buskirk 2006), and the positive relationships
between species might simply be explained by the fact that they might exploit
the same resources or that the distribution of the resources used by each species,
although different, might overlap.
Meles meles0. Intercept only 910.64 9.29 . . .1. Null model%SB + %B + eBP + DW + DV + Ra + Hum + autocov 901.35 0.00 . . .2. Null model + KAIl 901.98 0.63 . . .3. Null model + KAIf 903.12 1.77 . . .4. Null model + KAIm 890.65 -10.70 0.11 0.03 0.005. Null model + KAIg 903.04 1.69 . . .6. Null model + Lynx 900.83 -0.52 . . .7. Null model + Genet 902.95 1.60 . . .
Lynx pardinus0. Intercept only 153.89 41.97 . . .1. Null model%B + %SB + DW + eBP + DV + Ra + Hum 111.92 0.00 . . .2. Null model + KAIb 112.27 0.35 . . .3. Null model + KAIf 111.00 -0.92 . . .4. Null model + KAIm 113.92 2.00 . . .5. Null model + KAIIg 110.43 -1.49 . . .6. Null model + Genet 112.21 0.29 . . .
KAIl; Kilometre Abundance Index of lynxKAIb; Kilometre Abundance Index of badgerKAIf; Kilometre Abundance Index of foxKAIm; Kilometre Abundance Index of mongooseKAIg; Kilometre Abundance Index of genetGenet: Presence/Absence of Genet (1/0)Lynx; Presence/Absence of Lynx (1/0)
151
Chapter 3Chapter 3
The same explanation might apply to the positive association found between
mongooses and genets; two similar-sized species. Genets and mongooses partially
overlap in their habitat use in DNP (Chapter 2). Furthermore, previous studies have
also shown that both mongooses and genets use similar microhabitats such as dense
bushes or thickets for activity and resting in Doñana (Palomares and Delibes 1993),
and that a large fraction of their diets may coincide (Palomares and Delibes 1991).
On the same way, although mongooses exhibit a certain higher degree of habitat
specialization than foxes and badgers, the niche overlap of all three predators is large
probably due to their trophic and/or habitat generalism (Chapter 2). Additionally,
mustelids are not species with a high aggressive nature (Palomares and Caro 1999),
so it is reasonable to think that they can coexist with smaller predators without any
interference.
Our results demonstrate that the inclusion of the relative abundance of other
carnivorespeciesinhabitatmodelsforagivenspeciesmaysignificantlyimprove
them. Models based on species relative densities that excluded other predators’
occurrence could be a misleading proxy of habitat quality when studying patterns of
habitat use by species, and this may be particularly important for smaller predators
that very often are intraguild prey in natural communities. Smaller species might be
more abundant at sub-optimal or marginal habitats due to predation-driven direct
or fear-driven indirect effects from larger predators. There are some examples of
other carnivore species that have been reported as exhibiting limited patterns of
distribution and abundance due to interference competition with larger predators
(Berger and Gese 2007, Peterson 1995, Thurber et al. 1992, White and Garrott
1997, Burrows 1995, Durant 2000). However, habitat use models for carnivore
species at the apex of their ecological communities that are based on vegetation
152
Chapter 3Chapter 3
and/or landscape features as well as on prey availability should be enough to design
adequate management plans that aim to restore habitat for the species.
Figure 3. Relationships between the probability of genet presence and the numbers of lynx (a) and mongoose tracks (b) detected; between the numbers of mongooses tracks and that of lynxes (c) and badgers (d); and between the number of fox tracks and the presence of lynx (least square means and their standard errors are represented) (e).
153
Chapter 3Chapter 3
ACKNOWLEDGMENTS
ThisstudywasfinancedbytheprojectsCGL2004-00346/BOS(SpanishMinistryofEducationand
Science) and 17/2005 (Spanish Ministry of the Environment; National Parks Research Programme),
and sponsored by Land-Rover España S.A. We are especially thankful to J.C. Rivilla and S. Desniça
fortheirassistanceduringfieldwork.C.SotoreceivedaJAEpredoctoralgrantfromCSIC(Spanish
National Research Council).
REFERENCES
Berger KM, Gese EM (2007) Does interference competition with wolves limit the distribution and
abundance of coyotes? Journal of Animal Ecology 76:1075–1085.
Burnham KP, Anderson DR (2002) Model selection and multimodel inference. A practical
information-theoretic approach. 2nd ed. Springer, New York.
Burrows R (1995) Demographic changes and social consequences in wild dogs, 1964–1992. In
Serengeti II. Dynamics, management and conservation of an ecosystem: 385–400.
Sinclair, A. R. E. & Arcese, P. (Eds). Chicago: University of Chicago Press.
Buskirk SW (1999) Mesocarnivores of Yellowstone. Pages 165–187 in T. W. Clark, A. Peyton
Curlee, S. C. Minta, and P. M. Kareiva, eds. Carnivores in ecosystems: the Yellowstone
experience. Yale University Press, London.
Case TJ, Gilpin ME (1974) Interference Competition and Niche Theory. Proc Natl Acad Sci U S A
71 (8):3073-3077.
Creel S, Creel NM (1996) Limitation of African wild dogs by competition with larger carnivores.
Conservation Biology, 10, 526–538.
DonadioE,BuskirkSW(2006)Diet,morphology,andinterspecifickillingincarnivora.AmNat
167:524–536.
Dormann CF, McPherson JM, Araujo MB, Bivand R, Bolliger J, Carl G, Davies RG, Hirzel A, Jetz
W, Kissling WD, Kuhn I, Ohlemuller R, Peres-Neto PR, Reineking B, Schroder B, Schurr
FM, Wilson R (2007) Methods to account for spatial autocorrelation in the analysis of
species distributional data: a review. Ecography 30 (5):609-628.
Durant SM (1998) Competition refuges and coexistence: an example from Serengeti carnivores. J
154
Chapter 3Chapter 3
Anim Ecol 67 (3):370-386.
Durant SM (2000) Living with the enemy: avoidance of hyenas and lions by cheetahs in the
Serengeti. Behavioral Ecology, 11, 624–632.
Durant SM, Craft ME, Foley C, Hampson K, Lobora AL, Msuha M, Eblate E, Bukombe J, McHetto
J, Pettorelli N (2010) Does size matter? An investigation of habitat use across a carnivore
assemblage in the Serengeti, Tanzania. J Anim Ecol 79 (5):1012-1022.
Estes JA, Palmisan JF (1974) Sea Otters - Their Role In Structuring Nearshore Communities.
Science 185 (4156):1058-1060.
FedrianiJM(1997)Relacionesinterespecíficasentreellinceibérico,Lynx pardina, el zorro,
Vulpes vulpes, y el tejón, Meles meles en el Parque Nacional de Doñana. Universidad de
Sevilla. Facultad de Ciencias Biológicas. 181 Pg.
Fedriani JM, Palomares F, Delibes M (1999) Niche relations among three sympatric Mediterranean
carnivores. Oecologia 121 (1):138-148.
Fedriani JM, Fuller TK, Sauvajot RM, York EC (2000) Competition and intraguild predation among
three sympatric carnivores. Oecologia 125:258–270.
Fournier DA, Skaug HJ, Ancheta J, Ianelli J, Magnusson A, Maunder M, Nielsen A and Sibert
J (2012). “AD Model Builder: using automatic differentiation for statistical inference of
highly parameterized complex nonlinear models.” Optim. Methods Softw., *27*, pp. 233-249.
Johnson JB, Omland KS (2004) Model selection in ecology and evolution. Trends Ecol Evol 19
(2):101-108.
Kamler JF, Ballard WB, Gilliland RL, Mote K (2003) Spatial relationships between swift foxes and
coyotes in northwestern Texas. Can J Zool-Rev Can Zool 81 (1):168-172.
Keitt TH, Bjornstad ON, Dixon PM, Citron-Pousty S (2002) Accounting for spatial pattern when
modeling organism-environment interactions. Ecography 25 (5):616-625.
Laurenson MK (1995). Implications of high offspring mortality for cheetah population dynamics.
Pages 385–399 in A. R. E. Sinclair and P. Arcese, eds. Serengeti II: dynamics, management
and conservation of an ecosystem. University of Chicago, Chicago.
Menge BA, Sutherland JP (1987) Comrnunity regulation: variation in disturbance, competition,
and predation in relation to environmental stress and recruitment. Am Nat 130: 730-757.
Palomares F, CaroTM (1999) Interspecific killing amongmammalian carnivores.AmNat 153
155
Chapter 3Chapter 3
(5):492-508.
Palomares F, Delibes M (1991) Ecología comparada de la gineta, Genetta genetta (L.) y el meloncillo,
Herpestes ichneumon, (L.) (Mammalia, Viverridae) en Doñana (SO de la Península Ibérica.
Bol. R. Soc. esp. Hist. Nat. (Secc. Biol.), 87: 257-266.
Palomares F, Delibes M (1993) Key Habitats for Egyptian Mongooses in Doñana-National-Park,
South-Western Spain. Journal of Applied Ecology 30 (4):752-758.
Palomares F, Ferreras P, Fedriani JM, Delibes M (1996) Spatial relationships between Iberian lynx and
other carnivores in an area of south-western Spain. Journal of Applied Ecology 33 (1):5-13.
Palomares F, Ferreras P, Travaini A, Delibes M (1998) Co-existence between Iberian lynx and
Egyptian mongooses: estimating interaction strength by structural equation modelling and
testing by an observational study. J Anim Ecol 67 (6):967-978.
Paquet PC, Carbyn LN (1986) Wolves, Canis-Lupus, Killing Denning Black Bears, Ursus-
Americanus, in The Riding-Mountain-National-Park Area. Can Field-Nat 100 (3):371-372.
PetersonRO(1995)Wolvesasinterspecificcompetitorsincanidecology.EcologyandConservation
of Wolves (eds L.N. Carbyn, S.H. Frittz & D.R. Seip), pp. 315–324. Circumpolar Press,
Edmonton, Alberta.
Polis GA, Myers CA, Holt RD (1989) The Ecology and Evolution of Intraguild Predation - Potential
Competitors that eat each other. Annu Rev Ecol Syst 20:297-330.
Prugh LR, Stoner CJ, Epps CW, Bean WT, Ripple WJ, Laliberte AS, Brashares JS (2009) The Rise
of the Mesopredator. Bioscience 59 (9):779-791.
R Development Core Team (2008). R: A language and environment for statistical computing. R
Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL
http://www.R-project.org.
Ralls K, White PJ (1995) Predation on San-Joaquin Kit Foxes by Larger Canids. J Mammal 76
(3):723-729.
Ritchie EG, Johnson CN (2009) Predator interactions, mesopredator release and biodiversity
conservation. Ecol Lett 12 (9):982-998.
Roemer GW, Gompper ME, Van Valkenburgh B (2009) The Ecological Role of the Mammalian
Mesocarnivore. Bioscience 59 (2):165-173.
Rogers LL, Mech LD (1981) Interactions of Wolves and Black Bears in Northeastern Minnesota. J
156
Chapter 3
Mammal 62 (2):434-436. doi:10.2307/1380735.
Soto C, Desniça S, Palomares F (2012) Nonbiological factors affecting track censuses: implications
for sampling design and reliability.
Skaug H, Fournier D, Nielsen A, Magnusson A and Bolker B (2012). Generalized Linear Mixed
Models using AD Model Builder.
Smith PA (1994) Autocorrelation in logistic regression modelling of species distributions. – Global
Ecol. Biodiv. Lett. 4: 47–61.
Sovada MA, Roy CG, Bright JB, Gillis JR (1998) Causes and rates of mortality of swift foxes in
western Kansas. Journal of Wildlife Management 62 (4):1300-1306.
Thurber JM, RO Peterson, JD Woolington and JA Vucetich (1992) Coyote coexistence with wolves
on the Kenai Peninsula, Alaska. Canadian Journal of Zoology 70:2494–2498.
Valverde JA (1958). An ecological sketch of the Coto Doñana. British Birds 51:1-23.
VanderMeer J, Ens BJ (1997) Models of interference and their consequences for the spatial
distribution of ideal and free predators. J Anim Ecol 66 (6):846-858.
Van Horne B (1983) Density as a Misleading Indicator of Habitat Quality. Journal of Wildlife
Management 47 (4):893-901.
Viota M, López-Bao JV, Palomares F, Rodríguez A (2012) Shift in microhabitat use as a mechanism
allowing the coexistence of victim and killer carnivore predators.
White PA, RA Garrott (1997) Factors regulating kit fox populations. Canadian Journal of Zoology
75: 1982–1988.
Yamaguchi N, Rushton S, Macdonald DW (2003) Habitat preferences of feral American mink in the
Upper Thames. J Mammal 84 (4):1356-1373.
157
Chapter 4
Soto C, Palomares F (2012) Human-related factors regulate dog presence in
protected areas: implications for conservation and management control. Submitted.
CHAPTER 4
Human-related factors regulate dog presence in protected areas: implications for conservation and management control
Factores humanos regulan la presencia de perros en las
áreas protegidas: implicaciones para la conservación y
gestión
158
Chapter 4
159
Chapter 4
ABSTRACT
The presence of domestic species such as dogs Canis familiaris in protected
areas represents a conservation problem due to competition, predation and/or
disease transmission to native species. This introduced species may increase their
ranging activity towards protected areas by the planning of new urban areas and
by the spread of houses and small urban settlements because they are abundant in
those environments. Dogs’ effects on wildlife in protected areas may depend on
their nature (domestic dogs vs. feral dogs), on where they are found and on the
factors controlling their space use. In order to improve our ability to design effective
control policies, we investigate the factors affecting detection of dog tracks in a
Mediterranean national park which protects the world’s most critically endangered
felid species, the Iberian lynx (Lynx pardinus).
We studied the presence or absence of dogs at 69 2x2 km grids and
analysed the associated environmental and/or human constraints by logistic
regression models. We failed to detect dogs in areas away from anthropogenic
edges (track census effort above 470 km) so according to our results, dogs at DNP
may be domestic dogs incurring occasionally into the protected area from the
surrounding matrix. The detection of dog tracks was dependent on the presence
of people and consequently on the resources they provide. The direct threats of
dogs to wildlife may therefore not spread throughout the entire reserve or along the
entire edge length. Any strategy aimed at reducing domestic dogs’ impact in areas
of conservation concern where feral dogs populations are not established should
focus on the presence of settlements and their spatial spread, local awareness and
regulation to encourage local people to restrain their dogs’ movements as well as
control measures on boundaries and areas close to human dwellings.
160
Chapter 4 Chapter 4
RESUMEN
La presencia de especies domésticas como los perros Canis familiaris
en las áreas protegidas representa un problema para la conservación debido a
competición, predación y/o transmisión de enfermedades a especies nativas. Estas
especies introducidas deben incrementar su actividad en el interior de las áreas
protegidascomoconsecuenciadelaplanificacióndenuevasáreasurbanasasícomo
de la expansión de zonas urbanizadas, debido a su abundancia en esos ambientes.
Los efectos de los perros en la fauna nativa en las áreas protegidas debe
depender de su naturaleza (si se trata de perros domésticos o perros asilvestrados),
de dónde se encuentran y de los factores que determinan su uso del espacio. Para
mejorar nuestra capacidad de diseño de políticas de control efectivas, analizamos
los factores que determinan la detección de rastros de perros en un Parque Nacional
que protege la especie de felino más amenazada a nivel mundial, el lince Ibérico
(Lynx pardinus).
Estudiamos la presencia o ausencia de perros en 69 cuadrículas de 2 x 2
km2 y analizamos los factores ambientales y/o humanos que pudieran determinar
su presencia a través de modelos de regresión logísticos. No detectamos perros en
zonas alejadas de los límites antrópicos del área protegida a pesar del esfuerzo de
muestreo (más de 470 Km. de caminos censados). Según nuestros resultados, los
perros presentes en el Parque Nacional de Doñana deben ser perros domésticos
que hacen incursiones de manera ocasional en el área protegida desde la matriz
antrópica circundante. La detección de rastros de perros dependió de la presencia
de humanos y consecuentemente de los recursos que proporcionan. Los efectos
directos de los perros en la fauna nativa no deben por tanto mostrar la misma
intensidad en toda la extensión del área protegida o a lo largo de toda la longitud
161
Chapter 4 Chapter 4
de sus límites. Cualquier plan de manejo que pretenda disminuir la incidencia de
perros domésticos en áreas protegidas en las cuales no existan poblaciones de
perros silvestres establecidos, debe focalizarse en la presencia de asentamientos
humanos y su localización espacial, en la concienciación y regulación local para
fomentar que los propietarios restrinjan los movimientos de sus mascotas, así como
en la localización de las medidas de control en los límites de las áreas protegidas
próximos a las viviendas o zonas habitadas.
162
Chapter 4 Chapter 4
INTRODUCTION
The introduction of non-native species into ecosystems represents one of
the causes of native species decline and endangerment (e.g., King 1985, Soulé
1990, Williamson 1999, IUCN 2012). Due to human population growth, urban
sprawl and the rapid urbanization of natural landscapes, humans and with them
their companion animals such as domestic dogs Canis familiaris may be closely in
contact with wildlife in many areas (Ordeñana et al. 2010) including places where
nature conservation is a priority, such as protected areas and national parks created
to protect populations of vulnerable or threatened species. These non-native species
introduced by human across the globe (Wandeler et al. 1993) may therefore increase
their ranging activity towards the remaining natural landscapes extending within
those areas the deleterious human-associated effects.
In areas of conservation concern presence of companion animals like domestic
dogs may pose distinct threats including competition, interbreeding, predation and
disease. Dogs harass and kill wildlife exhibiting in many cases a “surplus” killing
behaviour (e.g., Iverson 1978, Kruuk & Snell 1981, Manor & Saltz 2004, Banks
& Bryan 2007), compete with wildlife (e.g., Boitani 1983, Butler & du Toit 2002,
Butler et al. 2004, Vanak et al. 2009) and spread disease like rabies, parvovirus or
canine distemper (e.g., Sillero-Zubiri et al. 1996, Cleaveland et al. 2000, Fiorello
et al. 2004, Fiorello et al. 2006, Vanak & Gompper 2009). Moreover, dogs such
asmid-sized canids, can also exert a top-down influence on smaller carnivores
through interference competition or intraguild predation (e.g., Glen & Dickman
2005, Mitchell & Banks 2005, Vanak & Gompper 2009). Studying where domestic
dogs came from in protected areas is needed in order to manage them and prevent
their potential impacts on native fauna. Dogs at protected areas may exhibit varying
163
Chapter 4 Chapter 4
levelsofhuman reliance, fromdomesticdogswhoseneedsare satisfieddirectly
or indirectly by people (i.e., owned dogs, urban or rural free-roaming dogs and/
or village dogs (Vanak & Gompper 2009)) and spend time sporadically in natural
protected areas, to feral dogs living and reproducing freely in protected areas. The
effects of dogs on wildlife may therefore depend on their nature (i.e., domestic dogs
vs. feral dogs), on where they are found and on the factors controlling their numbers
and space use.
As a heavily human-subsidized species, domestic dogs exhibit higher
densities in areas close to human residences or places with a high density of houses
(e.g., Odell & Knight 2001, Ordeñana et al. 2010) as well as in areas near natural
reserve borders with agriculture, where rural human residences are nearby. In
contrast, they may exhibit lower densities in areas contiguous to large tracts of
native forest, which may be acting as a buffer to the entrance of dogs (Srbek-Araujo
& Chiarello 2008).
Domestic dogs’ records decrease from anthropogenic matrices to forest
patch edges (Torres & Prado 2010). Hence, if we consider protected areas as large
patches, presence and therefore direct negative effects of domestic dogs on native
fauna (i.e., predation and/or competition) may decrease from the anthropogenic
matrix to the protected area interior reaching their maximum at the borders. This
could strengthen the negative anthropogenic edge effect associated with these
artificialborderareas(Woodroffe&Ginsberg1998,Revillaetal.2001)diminishing
therefore the reserve’s effectiveness in conserving wildlife.
Feral dogs are meanwhile completely wild and independent of human-derived
materials as food sources (Nesbitt 1975, Green & Gipson 1994). They depend
almost exclusively on wild-caught food (e.g., Marsack & Greg 1990, Glen &
164
Chapter 4 Chapter 4
Dickman 2005, Mitchell & Banks 2005, Campos et al. 2007, Glen & Dickman
2008) and might not exhibit any human association. The direct threats of feral dogs
to wildlife may therefore spread throughout entire protected areas.
In this context we studied the patterns of detection of dog tracks and the
associated environmental and/or human constraints that could influence their
presence in a fully protected Mediterranean area, Doñana National Park (DNP),
with high potential for dog-arrival due to its proximity to urban and rural settlements
and with the potential of dog-settlement due to its size.
Our main research goals were to answer: (1) where are dogs present at
DNP? And (2) what factors predict dog presence? A priori, we hypothesised that
dogs using DNP might come either from a domestic dog population formed by
individuals that incur occasionally from the surrounding matrix, or from a feral
dogpopulation livingand reproducing freely. In thefirstcase,wewouldexpect
that dogs are heavily dependent on humans and are more abundant at the edges
of the protected area close to human settlements than far away from these edges.
In the second case, we would expect that dogs avoid human association, are more
evenly distributed throughout the protected area, and their presence or abundance
related to environmental features describing habitat suitability or potential wild
food availability. DNP is optimal for the design of a study of this type since 1) part
of the protected area is surrounded by human settlements and others are not, and 2)
it is large enough to potentially hold a feral dog population in its interior.
165
Chapter 4 Chapter 4
METHODS
Study area
ThestudywascarriedoutatDoñanaNationalPark(DNP),aflatsandyarea
located in south-western Spain (550 Km2,37º9´N,6º26´W)atsealevel.Wedefined
the anthropogenic DNP edge as the northern and western edges in close contact
withhumansettlements,cropfieldsandahigh-traffichighway,andthenaturalDNP
edge as the southern edge limiting with the Atlantic Ocean and the eastern edge
limiting with the Guadalquivir River through a 27,000 ha marshland area (Fig. 1).
Population in the main suburban settlement (situated in the western vicinity
and separated from the DNP by a paved road) undergoes great variation between
winter and summer, as it is mainly a summer resort occupied by about 2,710 people
in summer seasons. The village situated in the north, and separated by the DNP
from the marshland area, is occupied by about 1,635 year-round residents, although
during a spring pilgrimage, the number of visitors can reach up to one million
people. There are also private large and medium-sized farms used for agriculture
as well as six visitor centers, hiking and cycling paths, recreation zones and bird
observatories in the nearby area. The DNP is fenced but the fence is permeable to
small and medium-sized animals including dogs. Free access is forbidden to people
and access to the core area and the dirt-road network inside the DNP is restricted to
the park staff and researchers.
The climate is Mediterranean sub-humid, with mild, wet winters, and hot,
dry summers, and an average annual rainfall around 550 mm. There are three main
biotopes in the park: scrubland, dunes, and marsh (Valverde 1958). The dune area
is situated at the western border of the protected area limited by the Atlantic Ocean
and the marsh area at the northern and eastern borders limited by the Guadalquivir
166
Chapter 4 Chapter 4
Figure 1.StudyareadefinedbytheDoñanaNationalParksiteandthesurroundinganthropogenicarea. Dog tracks detected during track counts in 2×2 km2 cell grids between November 2007 and April 2009 are shown below in detail.
167
Chapter 4 Chapter 4
River. The Mediterranean scrubland represents approximately half of the National
Park surface area and is mainly characterised by heterogeneous patches of
xerophytic species such as Halimium sp. and Cistus sp., and hydrophytic ones such
as Erica sp., with some patches of Juniperus phoenica and Pistacia lentiscus shrubs.
Interspersed among the scrubland there are scattered cork oak trees (Quercus suber)
and wild olive trees (Olea europea), and a few patches of pine Pinus pinea and
eucalyptus Eucalyptus sp. plantations.
Among larger mammals wild boar (Sus scrofa), red deer (Cervus elaphus)
and fallow deer (Dama dama) are frequent. Wild carnivores include red fox (Vulpes
vulpes), Eurasian badger (Meles meles), Egyptian mongoose (Herpestes ichneumon),
common genet (Genetta genetta), least weasel (Mustela nivalis), European polecat
(Mustela putorius), Eurasian otter (Lutra lutra), wild cat (Felis silvestris) and
Iberian lynx (Lynx pardinus). 14 small-medium sized mammal species have been
recorded in DNP, as well as 397 bird species, approximately half of which are
breeding in the park.
Field methods
To evaluate detection of dog tracks we carried out systematic track surveys
on sandy paths at 69 2x2 km2 grid cells located within the entire scrubland and dune
areas of the protected area during the wet season of 2007-08 and 2008-09.
We sampled for dog tracks in each square by slowly walking (ca. 1.5 km/h)
atleast3kmalongavailablepathways(i.e.,sandyroadsandfirebreaks).Oncea
track was detected, we georeferenced it using a GPS. We re-sampled the same path
(leaving at least seven days between samplings) a second time in a few squares until
completing3kmifduringthefirstsamplingtherewereinsufficientavailablepaths
168
Chapter 4 Chapter 4
within the square to achieve this distance. We always carried out surveys at least
three days after any rainfall.
Environmental quality of the sampling grids to hold a feral dog population
was assessed by sampling potential prey availability and general habitat structure.
Feral dogs are habitat generalists and opportunistic foragers depending almost
exclusively on a wide variety of wild-caught food (e.g., Boitani et al. 1995, Marsack
& Greg 1990). Potential prey availability was estimated by counting tracks of small
mammals, European rabbits (Oryctolagus cuniculus), red partridges (Alectoris rufa),
domestic cows (Bos taurus) and horses (Equus caballus) and wild ungulates such
as the fallow deer (Dama dama), the red deer (Cervus elaphus) and the wild boar
(Sus scrofa). Prey such as small mammals, rabbits, partridges and young of wild
ungulates might be hunted by feral dogs, but adults of many species are consumed
as carrion (e.g., Sillero-Zubiri & Macdonald 1997, Butler et al. 2004, Aiyadurai &
Jhala 2006). Prey species were surveyed by walking between 7 and 10 25 m-long
transects of approximately 1.7 m wide (the width of a four-wheel-drive car) and
separatedbyat least300mwithineach2x2kmgrid.Duringthefirstyear,prey
transects were carried out throughout the wet season, when tracks from dogs were
surveyed, but during the second year, we concentrated samplings within the month
of April to avoid possible intermonthly variations in abundance for some species
(e.g., see Kufner 1986, Palomares et al. 2001 for small mammals and European
rabbits, respectively.
In order to identify main habitats at DNP (dunes, pine reforestation and
Mediterraneanscrubland),generalhabitatstructurewasrecordedforthefirstyear.
We estimated visually the percentage of open ground cover, and the percentage
and modal height of three vegetation categories: short shrub (xerophytic species
169
Chapter 4 Chapter 4
such as Halimium sp. and Cistus sp), tall shrub (Erica sp., Juniperus phoenica and
Pistacia lentiscus shrubs) and trees. The estimation of these variables related to
habitat structure were carried out in a circle of 15 m radius around the sampling
point every 300 m on transects walked for dog and prey tracks. For each square
sampled, we averaged the value obtained at the vegetation sampling points.
Data analyses
We examined different environmental and/or human-related factors
explaining the detection of dog tracks within each 2x2 km grid at DNP using
generalised linear models with a binomial error distribution and a logit link function
(logistic procedure in SAS® 9.2 (SAS Inst. Inc., Cary, NC)). Non-biological factors
(i.e. methodological and climatic variables) have been previously reported as
potentially affecting results of track censuses (Soto et al. 2012); therefore we also
incorporatedinthemodelfittingthesefactorstocontrolfortheirpotentialeffect.
Each grid cell was associated with a set of habitat variables as vegetation
type (dunes (> 60% of open groundcover on average inside the grid), pine forest (>
60% of pine vegetation on average) and Mediterranean shrub (> 60% of shrub (short
or tall) vegetation on average) and prey abundance (kilometric abundance index
of total prey) and with variables describing their location in DNP by calculating
Euclidean distance from the grid cell centre to every infrastructure.
We used a two-step approach to analyze data. First, we assessed which
methodological and/or climatic variables potentially affect dog detectability and
selectedthebest-fittingmodelusinganinformation-theoreticapproach(Burnharm
& Anderson 2002). Variables considered to be included in models were the
observer who carried out censuses (three and two observers for both study years
170
Chapter 4 Chapter 4
respectively) (Obs), relative humidity (%) on census day (Hum), days since last rain
(Rain), year (Year) and the maximum temperature (ºC) calculated as the average of
the maximum temperature on the census day and the maximum temperature two
consecutive days before the census day (Temp). Climatic data was obtained from a
meteorological station located inside DNP (Latitude: 37º 1’18’’, Longitude: 6º 33’
17’’) http://icts.ebd.csic.es.
Secondly,weusedthisbest-fittingmodelasanullmodeltodevelopaset
of a priori models of dog tracks’ detectability at DNP based on three groups of
hypotheses stemming from the different variables considered in relation to 1) the
possible human dependence of dog tracks’ detectability (i.e., dogs being domestic),
2) the possibility of dogs coming from a feral population (i.e., dog tracks’
detectability related to environmental and/or prey variables), or 3) dogs coming
from a combination of both domestic and feral populations. Variables included
in models were the minimum distance to nearest single house or visitors centre
(D_HOU), minimum distance to human settlements (D_VIL), distance to nearest
paved road (D_RD), distance to anthropogenic edge of DNP (D_ANT), distance to
natural edge of DNP (D_NAT), Kilometre Abundance Index of total prey (Pt) and
the vegetation category; dunes, pine forest, scrubland (Veg).
Initially, the correlation among variables was explored using Kendall’s tau
statistics, in order to eliminate highly correlated variables (tau > 0.4) and among
them; we retained the more ecologically meaningful ones.
We used the Akaike Information Criterion (AIC) corrected for a small sample
size (AICc) and the difference in AICc between each model and the model with the
lowest AICc (∆AICc) to rank the models according to their capacity to describe
the data parsimoniously (Burnharm & Anderson 2002). The model with the lowest
171
Chapter 4 Chapter 4
AICcandthosewith∆AICc≤2wereconsideredtobesupported.ΔAICcvalues
wereusedtocomputeAkaike’sweights(ωi), which is the weight of evidence that a
model is the best approximating model given the model set (Burnham & Anderson
2002)andisdefinedas
ωi =exp(-1/2∆i)/∑R r=1exp(-1/2∆r).
In addition, the relative variable importanceofpredictorvariablej(ωj) was
determinedasthesumoftheωiacrossallmodelswherejoccurs.Largerωj values
indicate a higher relative importance of variable j compared to other variables.
For each hypothesis we used data from both years andwe began by fitting all
variables included and then successively removing the terms that decreased the
AIC the most (Crawley 2002).
Finally,we explored the classification accuracy of the selectedmodel(s)
using the nonparametric estimate of the area under the curve (AUC) of receiver-
operating characteristic plots (Hosmer & Lemeshow 2000). AUC indices range
from 0.5 to 1, with ranges from 0.5 to 0.7 indicating poor discrimination, from 0.7
to 0.8 acceptable discrimination, from 0.8 to 0.9 good discrimination, and > 0.9
outstanding discrimination. The AUC measure from the ROC curve is considered
useful for comparing the performance of the detection of dog tracks-absence model
in a threshold-independent fashion (Fielding & Bell 1997).
172
Chapter 4 Chapter 4
RESULTS
A total of 471 km was walked and 72 dog tracks were found during surveys
(Fig. 1). We detected dog tracks at 16 and 12 grid cells surveyed in each study year,
respectively. Four anthropogenic variables were strongly correlated. In particular,
distance to the anthropogenic edge of the National Park (D_ANT), distance to
nearest village (D_VIL), distance to nearest paved road (D_ROAD) and distance
to the natural edge of DNP (D_NAT). Analyses were focused on distance to the
anthropogenic edge of DNP and distance to the natural edge of DNP.
The best-fittingmodel explaining detection of dog tracks based on non-
biological factors included humidity (Hum) and days since last rain (Rain). Humidity
waspositivebutnon-significant(oddsratio=1.035,χ2 = 2.159, P=0.142) whereas
dayssince last rainwasnegativelyandsignificantlycorrelatedwithdetectionof
dogtracks(oddsratio=0.931,χ2 = 4.286, P=0.035). Both predictors were therefore
included as covariables in further analyses.
The analysis of dog tracks’ detectability based on human-related, habitat
and prey variables showed that the a priori hypothesis best adjusted to data only
included human-related predictors.
The best model describing the detection of dog tracks at DNP after adjusting
for detection probability variables in the null model included the distance to the
anthropogenic edge of DNP (explaining 27.6 % of the deviance); the next model
included the distance to the anthropogenic edge of DNP and the distance to the
natural edge of DNP (models 1 and 3, Table 1).
Detectionofdogswassignificantlyandnegativelyassociatedtothedistance
totheanthropogenicedgeofDNP(oddsratio=0.737,χ2 = 8.020, P=0.005)
(Fig. 2). Equation for this model (model 1, Table1) is:
173
Chapter 4 Chapter 4
logit(P) = -3.32(±1.86) - 0.31(±0.11)D_ANT + 0.05(±0.02)Hum -
0.06(±0.04)Rain
where P is the probability of dog occurrence; values within parentheses are standard
errors.
The relative variable importance of this anthropogenic variable determined
asthesumoftheωiacrossallmodelswherethevariableoccurredwasωj=0.999.
The discriminating ability of the top model was AUC = 0.802 (P < 0.0001).
DISCUSSION
Results show that the detection of dog tracks at DNP was associated with
distance from the park boundary with human presence, a synthetic indicator of
Table 1. Selection results from logistic regression models investigating the effects of anthropogenic, habitat and a combination of all variables on detection of dog tracks at DNP. For the top models, we report the small sample-size-adjusted Akaike’s information criteria (AICc), the difference in AICc between each model with the lowest AICc(∆AICc) and the AICc weight (wi).
Model code Deviance AICc ∆AICc wi0. Null model 139.210 121.107 12.939 0.000Anthropogenic1. D_ANT, Hum, Rain 100.170 108.168 0.000 0.2762. D_NAT, D_ANT, D_HOU, Hum, Rain 97.470 109.469 1.301 0.1443. D_NAT, D_ANT, Hum, Rain 98.240 108.242 0.074 0.2664. D_ANT, D_HOU, Hum, Rain 99.750 109.747 1.579 0.125Habitat5. Pt, Hum, Rain 112.970 120.971 12.803 0.0006. Veg, Hum, Rain 106.720 122.265 14.097 0.0007. Pt, Veg, Hum, Rain 110.900 122.897 14.729 0.000Global8. D_NAT, D_ANT, D_HOU, Pt, Hum, Rain 97.950 111.953 3.785 0.0429. D_NAT, D_ANT, Pt, Hum, Rain 98.850 110.853 2.685 0.07210. D_ANT, Pt, Hum, Rain 100.810 110.811 2.643 0.074
174
Chapter 4 Chapter 4
humaninfluencethatcapturestheeffectofdistancetonearestvillageandnearest
paved road.
We found a high number of dog records near the reserve’s borders where
rural and suburban households were closer and we were unable to detect signs of
dogs far away from these anthropogenic DNP edges in spite of our large census
effort. Additionally, dog tracks’ detectability did not seem to be related to the
environmental variability of DNP such as vegetation type or prey availability.
This lack of association between dogs and variables describing habitat
suitability or potential wild food availability, their dependence on human-related
variables and their higher abundance at the edges of the protected area close to
human settlements compared to areas far away from these edges, support our
Figure 2. Probability of domestic dog tracks’ detectability as a function of distance to the anthropogenic edge of DNP during the wet seasons of 2007-08 and 2008-09.
175
Chapter 4 Chapter 4
hypothesis that dogs using DNP come from a domestic dog population formed by
individuals that arrive sporadically from the surrounding matrix (i.e., owned dogs,
urban/rural free-roaming dogs or village dogs (Vanak & Gompper 2009)) and not
from a feral dog population living and reproducing freely. The lack of association
between detection of dog tracks and wildlife food resources possibly also reveals
the dependence of dogs on human-derived materials, which is typical for the vast
majority of dog populations for which diet has been studied (Atickem 2003, Butler
et al. 2004, Vanak 2008, Vanak & Gompper 2009).
Feral dogs survive and reproduce independently of human assistance but
some feral dogs use human garbage for food (Green & Gipson 1994). A feral dog
population established inside DNP could therefore use the edge of the protected
area to access human subsidies. Nevertheless, the primary feature that distinguishes
feral from domestic dogs is the degree of reliance on humans, so if dogs using
DNP come from a feral dog population living and reproducing freely but accessing
human subsidies for food, we would expect dog detectability to be dependent
on habitat suitability and/or wild food availability and marginally dependent on
human-related variables.
In contrast with domestic dogs, feral dogs are highly social living in packs
or groups year round in most cases (Daniels & Bekoff 1989, Green & Gipson 1994).
In our study area, we only detected isolated dog’s tracks. Additionally, camera-
trapping studies conducted inside DNP during the same period detected only six
dogs,allofwhichwerefoundnearhumansettlementsandidentifiedasdomestic
animals based on their external physical appearance (personal observ.).
Additionally, the occurrence of dog tracks restricted to the DNP edge and
the low number of tracks detected far away from these reserve’s edges supports
176
Chapter 4 Chapter 4
the idea that the presence of domestic dogs at Doñana may be exacerbating the
anthropogenic edge effect associated with its border areas.
Some previous authors have also reported a higher occurrence of domestic dogs
near the edges of natural reserves compared to their interiors (Butler & du Toit
2004, Srbek-Araujo & Chiarello 2008, Lacerda et al. 2009, Marks & Duncan 2009).
In this sense, domestic dogs could be generally considered as human-derived
edge effect at protected areas. The direct threats of domestic dogs to wildlife may
therefore not spread throughout the entire protected area or along the entire edge
length reaching its maximum near the anthropogenic border areas.
Edge effects can be important in the dynamics of populations living in
fragmented landscapes because they may affect key population parameters, such
as survival and reproduction (Murcia 1995, Noss & Csuti 1997). The peripheries
of reserves thus function as population sinks (Revilla et al. 2001) and the resulting
edge effect can cause the decline or the extinction of protected carnivore populations
(Woodroffe & Ginsberg 1998). The higher occurrence of dogs in these border areas
due to human presence in the surrounding matrix might exacerbate this effect.
Nevertheless, although domestic dogs rarely leave the vicinity of human
dwellings (Vanak & Gompper 2009, Butler & du Toit 2002) and appear to exhibit
a range mainly limited to reserves’ borders, their daily activity pattern may involve
free-ranging that can bring them into contact with wildlife, especially when their
movementsarenotconfinedtoaproscribedoutdoorarea(Butleretal.2004,Vanak
2008).
Diseases from dogs that affect wild species (e.g., Woodroffe & Ginsberg
1999, Cleaveland et al. 2000) could be transmitted across the border of reserves
worsen therefore the direct edge effects represented by domestic dogs in protected
177
Chapter 4 Chapter 4
areas (i.e., the eventual predation and displacement of some species near the park
border). In consequence, domestic dogs maintaining highly virulent, multi-host
pathogenscan inducemortality towildcarnivoresowing to interspecificdisease
transmission between susceptible wild carnivores in the community even when the
wildlife population of interest may never have contact with them.
DNP houses one of the last metapopulation of the most globally endangered
felid species, the Iberian lynx (Lynx pardinus) (Palomares et al. 2011) so dogs
living or spending time inside the area come into contact with this wild endangered
carnivore posing a serious risk for its conservation (Ferreras et al. 1992, Meli et al.
2009 and 2010, Millán et al. 2009a and 2009b).
Domestic dogs are present in large numbers in urban, suburban and rural
areas, so, due to their high numbers, they can have a substantial impact on wildlife,
even when they do not need to hunt to survive.
This situation forms a complicated scenario for conservation biologists
especially in conservation areas with endangered endemic carnivore populations
and/or those where reserve size is too small in relation to the scale of the species’
movement. There is a huge demand for more knowledge about and experience with
this type of situation, in order to help prevent or diminish the impacts on native
fauna in natural reserves.
Conclusions and management implications
Domestic dogs in DNP and in the surrounding matrix can considerably
diminish the reserve’s effectiveness in conserving wildlife. The rapid urbanization
process close to conservation units and the growth of domestic dog populations is
an increasing worldwide conservation. Thus, transborder conservation measures
178
Chapter 4 Chapter 4
must be implemented in the protected areas.
InthespecificcaseofDNP,controllingthedogpopulationsisurgentandkey
for the wildlife protection in this National Park. The detection of dog tracks at DNP
was dependent on the presence of people and consequently on the resources they
provide such that the potential direct effects of dogs on wildlife may be stronger on
these anthropogenic boundaries. Management of domestic dogs in protected areas
where feral dogs’ populations are not established may therefore be focused at borders
and neighbourhoods close to human dwellings. We suggest that control measures
must include the restriction of domestic dog’s free-roaming activity through local
public awareness focusing on responsible ownership and biodiversity conservation,
the removal of un-owned dogs through systematic campaigns on reserve boundaries
near human settlements, as well as strengthening of pet policies.
ACKNOWLEDGMENTS
ThisstudywasfinancedbytheprojectsCGL2004-00346/BOS(SpanishMinistryofEducationand
Science) and 17/2005 (Spanish Ministry of the Environment; National Parks Research Programme),
and sponsored by Land-Rover España S.A. CS received a JAE predoctoral grant from CSIC
(Spanish National Research Council). We are especially thankful to J.C. Rivilla and S. Desniça for
theirassistanceduringfieldwork,toN.Fernándezforhisusefulinputsforanalysisofthedataand
valuable suggestions, and to C. Dickman and P. Ferreras for their comments on earlier versions of
the manuscript.
REFERENCES
Acevedo-Whitehouse, K. (2009). The importance of disease management programmes for wildlife
conservation. Anim. Conserv. 12, 185–186.
Aguirre, A.A., Starkey, E.E., Hansen, D.E. (1995). Wildlife diseases in national park ecosystems.
Wildlife Soc. B. 23, 415–419.
179
Chapter 4 Chapter 4
Aiyadurai, A., Jhala, Y.V. (2006). Foraging and habitat use by golden jackals (Canis aureus) in the
Bhal Region, Gujarat, India. J. Bombay Nat. Hist.Soc. 103, 5–12.
Atickem, A. (2003). The ecology of domestic dog (Canis familiaris) and its effect on the survival
of the Ethopian wolf (Canis simensis) in the Web Valley of the Bale Mountains National
Park, Ethopia. MS thesis, Addis Ababa University, Addis Ababa, Ethiopia.
Banks, P., Bryant, J. (2007). Four-Legged Friend or Foe? Dog Walking Displaces Native Birds. Biol.
Lett. 3, 611–613.
Boitani, L. (1983). Wolf and dog competition in Italy. Acta Zool. Fennica. 174, 259–264.
Boitani, L., Francisci, F., Ciucci, P., Andreoli, G. (1995). Population biology and ecology of feral
dogs in central Italy. In The Domestic Dog. Its Evolution, Behaviour, and Interactions
with People: 217–244. Serpell, J. (Ed.). Cambridge, UK: Cambridge University Press.
Burnharm, K.P., Anderson, D.R. (2002). Model selection and multimodel inference. A practical
information-theoretic approach. 2nd ed. New York: Springer.
Butler, J.R.A., du Toit, J.T. (2002). Diet of free-ranging Domestic Dogs (Canis familiaris) in rural
Zimbabwe: Implications for wild scavengers on the periphery of wildlife reserves.
Anim. Conserv. 5, 29–37.
Butler, J.R.A., du Toit, J.T., Bingham, J. (2004). Free-ranging domestic dogs (Canis familiaris) as
predators and prey in rural Zimbabwe: threats of competition and disease to large wild
carnivores. Biol. Conserv. 115, 369–378.
Campos, C.B., Esteves, C.F., Ferraz, K., Crawshaw, P.G., Verdade, L.M. (2007). Diet of free-ranging
cats and dogs in a suburban and rural environment, south-eastern Brazil. J. Zool. 273, 14–20.
Clark, T.W., Curlee, A.P., Reading, R.P. (1996). Crafting effective solutions to the large carnivore
conservation problem. Conserv. Biol. 10, 940–948.
Cleaveland, S., Appel, M.G.J., Chalmers, W.S.K., Chillingworth, C., Kaare, M., Dye, C. (2000).
Serological and demographic evidence for domestic dogs as a source of canine distemper
virus infection for Serengeti wildlife. Vet. Microbiol. 72, 217–227.
Crawley, M.J. (2002). Statistical computing: An introduction to data analysis using S-Plus.
Chichester, West Sussex, UK, Wiley.
Crooks, K.R. (2002). Relative sensitivities of mammalian carnivores to habitat fragmentation.
Conserv. Biol. 16, 488–502.
180
Chapter 4 Chapter 4
Ferreira, J.P., Leita, I., Santos-Reis, M., Revilla, E. (2011). Human-related factors regulate the
spatial ecology of domestic cats in sensitive areas for conservation. PlosOne 6, e25970.
Ferreras, P., Aldama, J.J, Beltrán, J.F., Delibes, M. (1992). Rates and causes of mortality in a
fragmented population of lberian lynx Felis pardina Temminck, 1824. Biol. Conserv. 61,197–202.
Fielding, A.H., Bell, J.F. (1997). A review of methods for the assessment of prediction errors in
conservation presence/absence models. Environ. Conserv. 24, 38–49.
Fiorello, C.V., Noss, A.J., Deem, S.L. (2006). Demography, hunting ecology, and pathogen exposure
of domestic dogs in the Isoso of Bolivia. Conserv. Biol. 20, 762–771.
Glen, A.S., Dickman, C.R. (2005). Complex interactions among mammalian carnivores in Australia,
and their implications for wildlife management. Biol. Rev. 80, 387–401.
Glen, A.S., Dickman, C.R. (2008). Niche overlap between marsupial and eutherian carnivores: does
competition threaten the endangered spotted-tailed quoll?. J.Appl. Ecol. 45, 700–707.
Green, J.S., Gipson, P.S. (1994). Feral dogs. In The Handbook: Prevention and Control of Wildlife
Damage :1–7. Hygnstrom, S. E., Timm R. M and Larson G. E. (Ed.). University of
Nebraska, Lincoln, NE, USA.
IUCN 2012. The IUCN Red List of Threatened Species. Version 2012.1.
<http://www.iucnredlist.org>. Downloaded on 25 June 2012.
Iverson, J.B. (1978). The impact of feral cats and dogs on populations of theWest Indian rock iguana,
Cyclura carinata. Biol. Conserv. 14, 63–73.
King, W.B. (1985). Island birds: will the future repeat the past?. In Conservation of island birds:
3–15. Moors, P. J. (Ed.). ICBP Technical Publication Number 33.
Kruuk, H., Snell, H. (1981). Prey selection by feral dogs from a population of marine iguanas
Amblyrhynchus cristatus. J. Appl. Ecol. 18, 197–204.
Kufner, M.B. (1986). Tamaño actividad, densidad relativa y preferencias de hábitat de los pequeños
y medianos mamíferos de Doñana, como factores condicionantes de su tasa de predación.
Thesis, Universidad Autónoma, Madrid, Spain.
Lacerda, A.C.R., Tomas, W.M., Marinho-Filho, J. (2009). Domestic dogs as an edge effect in the
Brasília National Park, Brazil: interactions with native mammals. Anim. Conserv. 12, 477–487.
Manor, R., Saltz, D. (2004). The impact of free-ranging dogs on gazelle kid/female ratio in a
fragmented area. Biol. Conserv. 119, 231–236.
181
Chapter 4 Chapter 4
Marks, B.K., Duncan, R.S. (2009). Use of Forest Edges by Free-ranging Cats and Dogs in an
Urban Forest Fragment. South. Nat. 8, 427–436.
Marsack, P., Greg, C. (1990). Feeding behavior and diet of dingoes in the Nullarbor region, Western
Australia. Aust. Wildlife Res. 17, 349–357.
McCallum, H. (1996). Immunocontraception for wildlife population control. Trends Ecol. Evol. 11,
491–493.
Meli, M.L., Cattoril, V., Martínez, F., López, G., Vargas, A., Simón, M.A., Zorrilla, I., Muñoz A.,
Palomares F., López-Bao, J.V., Pastor, J., Tandom, R., Willi, B., Hofmann-Lehmann R.,
Lutz, H. (2009). Feline leukaemia virus and other pathogens as important threats to the
survival of the critically endangered Iberian Lynx (Lynx pardinus). PLoS ONE. 4, e4744.
Meli,M.L.,Cattori,V.,Martınez,F.,López,G.,Vargas,A.,Palomares,F.,López-Bao,J.V., Hofmann-
Lehmann, R., Lutz, H. (2010). Feline leukemia virus infection: a threat for the survival of the
critically endangered Iberian lynx (Lynx pardinus). Vet. Immunol. Immunop. 134, 61–67.
Millán, J., Candela, M.G., Palomares, F., Cubero, M.J., Rodríguez, A., Barral, M., de la Fuente,
J., Almería, S., León-Vizcaíno, L (2009a). Disease threats to the endangered Iberian lynx
(Lynx pardinus). Vet. J. 182, 114–124.
Millán, J., Cabezón, O., Pabón, M., Dubey, J.P., Almería, S. (2009b). Seroprevalence of Toxoplasma
gondii and Neospora caninum in feral cats (Felis silvestris catus) in Majorca, Balearic
Islands, Spain. Vet. Parasitol. 165, 323–326.
Mitchell, B.D., Banks, P.B. (2005). Do wild dogs exclude foxes? Evidence for competition from
dietary and spatial overlaps. Austral Ecol. 30, 581–591.
Murcia, C. (1995). Edge effects in fragmented forests: implications for conservation. Trends Ecol.
Evol. 10, 58–62.
Nesbitt, W.H. (1975). Ecology of a feral dog pack on a wildlife refuge. In The Wild Canids : 391–
395. Fox, M. W. (Ed.). New York, USA: Van Nostrand Reinhold Company.
Neronov, V.M. Khlyap, L.A. Bobrov, V.V., Warshavsky, A.A. (2008). Alien species of mammals
and their impact on natural ecosystems in the biosphere reserves of Russia. Integrative
Zool. 3, 83–94.
Noss, R.F., Csuti B. (1997). Habitat fragmentation. In Principles of conservation biology: 269–
304. Meffe, G. K. and Carroll, C. R. (Eds.). Sunderland, Massachusetts: Sinauer
182
Chapter 4
Associates, Inc., Publishers.
Nowell, K., Jackson, P. (1996). Wild Cats: Status Survey and Conservation Action Plan. IUCN/
SSC, Gland, Switzerland.
Odell, E.A., Knight, R.L. (2001). Songbird and medium-sized mammal communities associated
with exurban development in Pitkin County, Colorado. Conserv. Biol. 15, 1143–1150.
Ordeñana, M.A., Crooks, K.R., Boydston, E.E., Fisher, R.N., Lyren, L.M., Siudyla, S., Haas,
C.D., Harris, S., Hathaway, S.A., Turschak, G.M., Miles, A.K., Van Vuren, D.H. (2010).
Effects of urbanization on carnivore species distribution and richness. J. Mammal. 91,
1322–1331.
Palomares, F., Delibes, M., Revilla E., Calzada, J., Fedriani, J.M. (2001). Spatial ecology of Iberian
lynx and abundance of European rabbits in Southwestern Spain. Wildlife Monogr.148,1–36.
Reed, S.E., Merenlender, A.M. (2011). Effects of Management of Domestic Dogs and Recreation on
Carnivores in Protected Areas in Northern California. Conserv. Biol. 25, 504–513.
Revilla, E., Palomares, F., Delibes, M. (2001). Edge-core effects and the effectiveness of traditional
reserves in conservation: Eurasian Badgers in Doñana National Park. Conserv. Biol. 15,
148–158.
Ryser-Degiorgis, M.P., Hofmann-Lehmann, R., Leutenegger, C.M., af Segerstad, C.H., Morner, T.,
Mattsson, R., Lutz, H. (2005). Epizootiologic investigations of selected infectious disease
agents in freeranging Eurasian lynx from Sweden. J. Wildlife Dis. 41, 58–66.
Schonewald-Cox, C., Azari, R., Blume, S. (1991). Scale, variable density and conservation planning
for mammalian carnivores. Conserv. Biol. 5, 491–495.
Sillero-Zubiri, C., King, A.A., MacDonald, D.W. (1996). Rabies and mortality in Ethiopian wolves
(Canis simensis). J. Wildlife Dis. 32, 80–86.
Sillero-Zubiri, C., Macdonald, D.W. (1997). The Ethiopian Wolf: Status Survey and Conservation Action
Plan. IUCN Canid Specialist Group, Gland, Switzerland, and Cambridge, UK.
Sleeman, J.M., Keane, J.M., Johnson, J.S., Brown, R.J., Woude, S.V. (2001). Feline leukemia virus
in a captive bobcat. J. Wildlife Dis. 37, 194–200.
Soto, C.A., Desnica, S., Palomares, F. (2010). Non-biological factors affecting track censuses;
implications for sampling design and reliability. Eur. J. Wildlife Res. 58, 117–126.
Soulé, M.E. (1990). The onslaught of alien species, and other challenges in the coming decades.
183
Chapter 4
Conserv. Biol. 4, 233–239.
Srbek-Araujo, A.C., Chiarello, A.G. (2008). Domestic dogs in Atlantic forest preserves of south-
eastern Brazil: a camera-trapping study on patterns of entrance and site occupancy rates.
Braz. J. Biol. 68, 771–779.
Torres, P.C., Prado, P.I. (2010). Domestic dogs in a fragmented landscape in the Brazilian Atlantic
Forest: abundance, habitat use and caring by owners. Braz. J. Biol. 70, 987–994.
Tyndale-Biscoe, C.H. (1994). Virus-vectored immunocontraception of feral mammals. Reprod.
Fert. Develop. 6, 281–287.
Valverde, J. A. (1958). An ecological sketch of the Coto Doñana. British Birds. 51, 1– 23.
Vanak, A.T. (2008). Intraguild interactions between native and domestic carnivores in central India.
PhD Dissertation, University of Missouri, Columbia, MO, USA.
Vanak, A.T., Gompper, M. E. (2009a). Dogs Canis familiaris as carnivores: their role and function
in intraguild competition. Mammal Rev. 39, 265–283.
Vanak, A.T., Gompper, M.E. (2009b). Dietary niche separation between sympatric free-ranging
domestic dogs and Indian foxes in central India. J. Mammal. 90, 1058–1065.
Vanak, A.T., Thaker, M., Gompper, M.E. (2009). Experimental examination of behavioural interactions
between free-ranging wild and domestic canids. Behav. Ecol. Sociobiol. 64, 279–287.
Wandeler, A.I., Matter, H.C., Kappeler, A., Budde, A. (1993). The Ecology of Dogs and Canine
Rabies: A Selective Review. Rev. Sci. Tech. O.I.E. 12, 51–71.
Whiteman, C.W., Matushima, E.R., Confalonieri, U.E.C., Palha, M.D.C., Silva, A.S.L., Monteiro,
V.C. (2007). Human and domestic animal populations as a potential threat to wild carnivore
conservation in a fragmented landscape from the Eastern Brazilian Amazon. Biol. Conserv.
138, 290–296.
Williamson, M. (1999). Invasions. Ecography. 22, 5–12.
Woodroffe, R., Ginsberg, J.R. (1998). Edge effects and the extinction of populations inside protected
areas. Science. 280, 2126–2128.
Woodroffe, R., Ginsberg, J.R. (1999). Conserving the African wild dog Lycaon pictus. I.Diagnosing
and treating causes of decline. Oryx, 33,132-142.
184
Chapter 4
ANNEX II
Domestic dogs sighted within Doñana National Park and photo-trapped during the study period.
185
CHAPTER 5
Surprising low abundance of European wildcats in a protected area of southwestern Spain
Abundancia sorprendentemente baja de gato montés en un
área protegida del suroeste de España
186
Chapter 5
187
Chapter 5
ABSTRACT
The wildcat Felis silvestris is a protected species in Europe but the lack
of information on its status in many areas of its distribution range is an obstacle
to conservation initiatives. To assess the status of the species over a 550 km2
Mediterranean protected area in south-western Spain; Doñana National Park, we
carried out track censuses during the wet season of 2007-08 and 2008-09 in 2 x 2
km2 quadrants and set camera traps from June 2008 to October 2010 in quadrants or
nearby quadrants where cat tracks were detected. We detected a total of 52 cat tracks
forbothstudyyearsand identifiedsixdifferent individualsusingmorphological
criteria from 28 photographs taken at 12 out of 166 trapping stations. We hypothesized
that the a priori surprising low abundance of the species in the protected area might
be likely multifold and could be explained by the decrease of rabbit population
in the DNP during the last decades, the isolation of DNP from the nearest natural
areas that could have slowed the recovery of wildcat populations after the species
declining at the beginning of the 20th century, an increased mortality rate over time
due to potential disease transmission from domestic cats and/or the competitive
exclusion of the species by the Iberian lynx (Lynx pardinus) in the Doñana area.
188
Chapter 5Chapter 5
RESUMEN
El gato montés Felis silvestris es una especie protegida en Europa pero la
escasa información sobre su estatus de conservación en muchas zonas de su área de
distribución, es un obstáculo para la conservación de la especie.
Para determinar el estado de conservación de la especie en un área
Mediterránea protegida situada en el suroeste de España; el Parque Nacional de
Doñana, realizamos censos de rastros durante la estación húmeda de los años 2007-
2008 y 2008-2009 en cuadrículas de 2 x 2 km2, y colocamos cámaras de foto-
trampeo entre junio del 2008 y octubre del 2010 en cuadrículas en las cuales se
detectaron rastros de gatos o en cuadrículas cercanas. Detectamos un total de 52
rastrosdegatoenambosañosdemuestreoeidentificamos6individuosatravés
de criterios morfológicos a partir de 28 fotografías obtenidos en 12 de las 166
estaciones de foto-trampeo colocadas. Sugerimos que la a priori sorprendentemente
baja abundancia de la especie en el área protegida debe estar originada por causas
múltiples como el descenso de las poblaciones de conejos en Doñana durante las
últimas décadas, el aislamiento de las zonas naturales más cercanas que podría
haber ralentizado la recuperación del gato montés después del declive de la especie
a inicios del siglo XX, una mayor tasa de mortalidad a lo largo del tiempo debido
a la potencial transmisión de enfermedades a través de gatos domésticos y/o una
exclusión competitiva por el lince ibérico (Lynx pardinus) en el área.
189
Chapter 5Chapter 5
INTRODUCTION
The European wildcat Felis silvestris is one of few wild felids in Europe,
but its conservation status is somewhat paradoxical. The wildcat is listed as of
Least Concern by the IUCN (2008) due to its wide distribution, ranging from the
Iberian Peninsula to Eastern Europe (Nowell and Jackson 1996, IUCN 2007).
Nevertheless, human-mediated habitat disturbance and large-scale hunting in the
early 20th century have led to severe local declines and extirpations in Europe
(Stahl and Léger 1992, Sunquist and Sunquist 2002), resulting in a fragmented
distribution (Stahl and Artois 1991, Nowell and Jackson 1996, Peichocki 2001).
Subsequent legal protection, under the Bern Convention (Appendix II 1979) and the
European Habitat Directive 92/43/EEC (EUROP 1992), has reduced the causes of
this decline and has led to a spontaneous recovery of European wildcat populations
in some parts of Europe (Stahl and Artois 1991). But despite this legal protection,
the wildcat continues to face number of threats throughout its range (Lozano 2009),
with human persecution (predator control) and habitat alteration (Lozano et al. 2007,
Virgós and Travaini 2005) likely the most important. Hence, the European wildcat
is considered to be “Near-Threatened” in the 25 member states of the European
Union (Temple and Terry 2007) including Spain, where wildcat subpopulations are
suspected to have decreased at a rate of >30% over three generations (Palomo and
Gisbert 2002).
In order to develop action plans for the conservation of the wildcat and
defineareaswhereconservationofwildcatsshouldbepriority, it isnecessary to
evaluate its distribution, abundance, ecological requirements and population status.
The aim of this study was to assess the presence of European wildcat in a protected
Mediterranean area, the Doñana National Park (DNP), where wildcats and other
190
Chapter 5Chapter 5
wildlifehavebeenprotectedformorethanfivedecades(thatis,apriorilowhuman
pressure and persecution), and where habitat and prey availability should be
potentially favourable to wildcat.
Although previous information is anecdotal and sporadic, it seems that the
abundance of the species has been very low long ago. Fifty years ago Valverde
(1967) considered the species as low abundant in the area. More recent available
information refers to a litter of three kittens found within the Doñana National
Park in 1997 when looking for Iberian lynx litters (N. Fernández, pers. comm.),
three individuals captured in 1999, 30 pictures of wildcats photo-trapped between
2000 and 2007 and a total of 52 direct sightings of the apparently wildcats between
1989 and 2000 collected by all personal working in the Doñana National Park
(Centro International de Estudios y Convenciones Ecológicas y Medioambientales
(CIECEM)) (see Annex III). Note that for the same period, other carnivores such as
foxes or lynxes were sighted in 1211 and 669 occasions, respectively.
The main objective of this study was to assess the wildcat’s current status
in the Doñana area and to discuss about the factors explaining its abundance.
Given how little is known about wildcat biology and the vulnerability of Iberian
populations, we aimed to provide baseline data to promote further research and
conservation of the wildcat in southern Spain.
METHODS
Study area
DNP is approximately a 550 km2 area in southwestern Spain bordered to
the south and west by the Atlantic Ocean and to the east by the Guadalquivir River
mouth.Theareaisflatandmostlynearsealevel;soilsarepredominantlysandyand
191
Chapter 5Chapter 5
of marine origin. The climate is Mediterranean subhumid and has marked seasons:
winters are mild and wet, and summers are hot and dry. Mean annual precipitation is
approximately 550 mm. Its situation between the Atlantic Ocean and Mediterranean
Sea and in southern Europe promotes some of the highest biological diversity on
the continent, particularly of vertebrate animals and vascular plants (Fernández-
Delgado 1997). There are three main biotopes in the park: scrubland, dunes, and
marsh (Valverde 1958). The dune area is situated at the western border of the
protected area where it is limited by the Atlantic Ocean and the marsh area lies at the
northern and eastern borders limited by the Guadalquivir River. The Mediterranean
scrubland represents approximately half of the National Park surface area and
is mainly characterised by heterogeneous patches of xerophytic species such as
Halimium sp. and Cistus sp., and hydrophytic ones such as Erica sp., with some
patches of Juniperus phoenica and Pistacia lentiscus shrubs. Interspersed among
the scrubland are scattered cork oak trees (Quercus suber) and wild olive trees (Olea
europea), and a few patches of pine Pinus pinea and eucalyptus Eucalyptus sp.
plantations. Vegetation in bare sand dunes is scarce and dune hollows are colonized
by pines Pinus pinea and varied scrubland species.
DNP is fully protected and access to the core area and the dirt-road network
inside DNP is restricted to the park staff and researchers. The northern and western
edgesofDNPareinclosecontactwithhumansettlements,cropfieldsandahigh
use paved road (Fig. 1). These surroundings support intense human activity, with
private large- and medium-sized farms used for agriculture as well as six visitor
centers, hiking and cycling paths, recreation zones and bird observatories in the
nearby area.
192
Chapter 5Chapter 5
Field methods
Weusedtracksurveysasafirstapproximationandphoto-trappingstudies
as a more directed methodology (since tracks of wildcats are not distinguishable
from that of domestic cats) to analyze the presence and/or occurrence of wildcats in
DNP. We carried out systematic track surveys at 69 and 67 2 x 2 km2 quadrant cells
located within the entire scrubland and dune areas of the DNP during the wet season
of 2007-08 and 2008-09, respectively. Marshland area was not sampled as its clay
soils make it unsuitable for track censuses. We sampled for cat tracks in each square
byslowlywalking(ca.1.5km/h)atleast3kmalongsandyroadsandfirebreaks.
Once a track was detected, we georeferenced it using a GPS. We re-sampled the
same path (leaving at least seven days between samplings) a second time in a few
squaresuntilcompleting3kmifduringthefirstsamplingtherewereinsufficient
available paths within the square to achieve this distance. We always carried out
surveys at least three days after any rainfall. Potential prey availability for wildcats
was also estimated by counting tracks of European rabbits (Oryctolagus cuniculus),
red partridges (Alectoris rufa) and small mammals (probably mostly long-tailed
fieldmouse(Apodemus sylvaticus) according to Kufner and Moreno 1989) in 25 m
long and approximately 1.7 m wide transects separated by at least 300 m (see Soto
et al. 2012). We also visually estimated variables related to vegetation type and
structure along transects in a circle of 15 m radius around the sampling point (see
Table 1 for variable description).
To asses the occurrence of wildcats and to determine if tracks detected
belonged to the species, we used camera-trapping techniques from June 2008 to
November 2010. Thirteen camera trap surveys with an average duration of 23 days
were conducted. To set the cameras, we selected quadrants and nearby quadrants
193
Chapter 5Chapter 5
Table 1. Variables used to compare means between quadrants with no evidence of cat presence, quadrants where wildcats were photographed and quadrants where cat tracks were detected during track censuses by non-parametric tests.
Variable Code Definition UnitsVegetationShort shrub %S Mean cover of short scrubland per quadrant* %Short shrub height S_h Average height of short scrubland per quadrant* m
Tall shrub %B Mean cover of bushes per quadrant* %Tall shrub height B_h Average height of bushes per quadrant* mTrees %T Mean cover of trees per quadrant* %Tree height T_h Average height of trees per quadrant* m
Landscape
Distance to water DW
Measured in meters using a Euclidean distance-based approach from the quadrant centre to the nearest permanentlyfloodednaturalorartificialpond(i.e.,dugfor the cattle at zones were the water table is higher) in a digitized water sources cover layer of DNP
m
Distance to La Vera DV
Measured in meters from the quadrant centre to the ecotone between the marshland and the Mediterranean scrubland (locally called La Vera)
m
Ecotones between pastureland and scrubland
eBP
Linear measure of the density of the ecotone between patches with bush cover >50% and patches with pasture cover>50%definedfroma reclassifiedfine-scale1:10000 vegetation map for the year 1996-2006 obtained from the Sistema de Información Ambiental de Andalucía
m/ha
Prey availabilityRabbits Ra Kilometric Abundance Index of rabbits per quadrant * Tracks/km
Small mammals SM Kilometric Abundance Index of small mammals per quadrant * Tracks/km
Total prey Tot Kilometric Abundance Index of rabbits + partridges + small mammals + ungulates per quadrant * Tracks/km
Human disturbance
Distance to antropic edge DH
Measured in meters from the quadrant centre to the nearestprotectedareaborderinfluencedbyhumans(i.e.excluding the beach and marshland edges)
m
Predators occurrenceKilometre abundance index of domestic dogs
KAId Number of dog tracks detected per km per quadrant m/ha
Kilometre abundance index of lynxes
KAIl Number of lynx tracks detected per km per quadrant tracks/km
* calculated by averaging values obtained at the different sampling points within quadrants
194
Chapter 5Chapter 5
where cat tracks were detected during track censuses. Camera traps were set
in the borders of car tracks and firebreaks, or in the edges of patches of dense
Mediterranean shrub with pasturelands (habitat potentially favourable for wildcats
(Lozano et al. 2003). In order to maximize the detection of wildcats we used scent
lures of valerian, catnip and canned sardines sprayed on a piece of cotton attached
toawoodenstakeata30–50cmheight,wetfish(sardines),liveprey(reportedas
themostefficientlureforsamplingsomefelidspecies(Guiltetal.2010,Garroteet
al. 2012)) such as rock pigeons (Columba livia) and rabbits (Oryctolagus cuniculus)
in wire cages inaccessible to wildcats, as well as no attractants. Cages of live prey
were approximately 100 x 50 x 50 cm and supplied with ample food and water
at least twice a week. On average, we set up 15 digital camera traps with passive
infraredmotionsensorsandautomaticflash(CuddebackDigitalScoutingModel
Expert®) per quadrant. Cameras were placed 20 cm above ground, at a distance
of 2-4 m from the lure with 300-400 m between them. We set camera traps with a
delay time of 1 min between successive photos, and checked at least twice per week
to replace attractant lures and twice a month for battery replacement.
Differentiation between wildcats and domestic cats was based on the general
physical appearance and on the pelage pattern of the individuals. Studies on the
European wildcat have indicated that camera trapping can be used to some extent
to determine the presence and abundance of this species and that individuals are
identifiablebasedontheirmorphology(RagniandPossenti1996,Monterrosoetal.
2005, Anile et al. 2007, Karanth et al. 2004). We considered photos as independent
events when taken more than 4 hours apart for the same individuals or if different
individualscouldbeidentified(O’Brienetal.2003).
195
Chapter 5Chapter 5
Figure 1. Locations of cat tracks detected during track censuses in 2 x 2 km2 quadrants during 2007-2009 (a); locations of the 166 trapping stations set during 2008-2010 (b), and the camera trap stations that provided pictures of wildcats and domestic cats, respectively, as well as the home range of the wildcat radio-tracked in DNP (c) are shown.
196
Chapter 5Chapter 5
We performed a Kruskal–Wallis test to determine whether different measures of
landscape structure, vegetation type, prey availability, human disturbance and the
occurrence of other carnivore species (i.e., domestic dogs (Canis familiaris) and
Iberian lynx (Lynx pardinus) as potentially negatively affecting wildcats (Palomares
and Caro 1999) (Table 1) differed across three categories of quadrants with different
wildcat presence: (0) quadrants with no evidence of wildcat presence (i.e., neither
photographs nor tracks), (1) quadrants where wildcats were photographed and
(2) quadrants where cat tracks were detected. Multiple post hoc comparisons of
meanrankswerealsoestimatedtodetectstatisticallysignificantdifferencesforall
pairs of groups (Siegel and Castellan 1988). Kruskal–Wallis analyses and post hoc
comparisons were conducted in SPSS software.
We also trapped a wildcat in a box-trap in December 2008 under a broader
project that aimed to study the effectiveness of red fox control actions within DNP.
The captured animal was chemically immobilized with a 0.75 ml dose (100mg*ml-1)
of tiletamine- zolazepam (Zoletil©, Virbac, Spain), measured, weighted, checked
for any sanitary disorders and sexed. Genetic analysis from blood samples collected
revealed that the individual was ‘pure’ or without any indication of parental
domestic heritage (Alves, P.C., pers. comm.). After handling, the individual was
maintained in the dark and returned to the capture location for release after complete
recoveryofreflexes(1–3h).Theindividualwasfittedwitharadio-collar(Wildlife
Materials, Inc., Carbondale, Illinois, USA), radio-tracked between December 2009
and March 2010 and located on average twice per week between 9:00 am and 2:00
pm. We used triangulation to determine the position of the individual (White and
Garrott 1990) and the minimum convex polygon method to estimate the home
range size based on all available locations (Mohr 1947, White and Garrott 1990)
197
Chapter 5Chapter 5
with Hawth’s Analysis Tools (Beyer 2004) in ArcGIS (ESRI, Redlands, CA). We
determine the main vegetation types included within the individual home range
froma1:10000fine-scalevegetationmapfortheyear1996-2006obtainedfromthe
Sistema de Información Ambiental de Andalucía (http://www.juntadeandalucia.es/
medioambiente/site/rediam/)
RESULTS
We detected a total of 25 and 27 cat tracks in 8 and 17 of 69 and 67 quadrants
censused in each year, respectively (Fig. 1). We set cameras at 166 different points
in 25 quadrants. Camera effort was 5761 trap-days with an average of 24.4 day/
cameras (SD= 8.02, range=1-33). We obtained a total of 2173 photographs in
whichweidentifiedmammals(n = 2050) or birds (n = 123). The red fox was the
most common species photographed followed by the Egyptian mongoose and the
common genet (Table 2). Thirty pictures of cats were taken at 12 of the 166 points
where we set camera trap stations. Twenty-eight of these pictures were of wildcats
and two pictures were of domestic cats. These camera traps were baited with
pigeons (n = 8) and wet sardines (n = 6) (Table 2). Wildcats were photographed
between20:00and07:00hours.Sixdifferentwildcatswereidentified(Fig.2)from
12 camera trap records (Table 3) and all were photographed only once. None of
the individuals could be sexed, nor could the presence of more than one individual
per trap-site be ascertained. Two-hundred six camera-trap days were required on
average to document the presence of an individual.
Radio-tracking effort produced 24 locations and a home range size of
approximately 24 km2 (Fig. 1). More than 60% of the individual home range
included areas of Mediterranean short scrubland (i.e., species such as Halimium sp.
198
Chapter 5Chapter 5
and Cistus sp.) and approximately 18% pine woodlands with understorey vegetation
(i.e., short shrubs and tall shrubs such as Erica sp., Juniperus phoenica and Pistacia
lentiscus).
We found differences between quadrants where wildcats were photographed,
quadrants with only cat tracks and quadrants with no evidence of cat presence for the
Table 2. Total number and percentage of pictures taken during camera-trapping surveys at DNP.
Specie N %
Felis silvestris 28 1.3Felis catus 2 0.1Lynx pardinus 3 0.1Genetta genetta 60 2.8Herpestes ichneumon 152 7.0Meles meles 26 1.2Vulpes vulpes 1284 59.1Canis familiaris 7 0.3Others* 611 28.1*other mammals and birds
Table 3. UTM coordinates (29S) of the camera-trap positions where wildcat were photographed. The period of time the camera was active and the number of wildcat pictures taken and baits used are also provided.
Trap station X Y Pictures Trap-days Bait
1 185585 4106480 7 22 Pigeon2 186451 4106914 1 22 Pigeon
3 185780 4100419 1 10 Pigeon
4 186990 4099547 1 29 Wet sardine5 184710 4101010 3 29 Wet sardine6 185476 4107561 1 29 Wet sardine7 185967 4108394 1 29 Pigeon8 185895 4108264 7 29 Wet sardine9 186503 4109315 1 29 Wet sardine10 183961 4117788 5 20 Pigeon11 190451 4105106 1 22 Pigeon12 189275 4105888 1 30 Pigeon
199
Chapter 5Chapter 5
variable distance to the anthropic edge of the protected area (Table 4). Mann-Whitney
U post-hoc tests revealed that quadrants where wildcats were photographed were
closer to the anthropic edge of the protected area than quadrants with no evidence of
cats and than quadrants where tracks were detected (Table 5). No differences were
found for any of the remaining variables analysed.
Figure 2. Individual wildcats distinguished on the basis of their external aspect in the Doñana National Park between 2008 and 2010.
200
Chapter 5Chapter 5
DISCUSSION
Thisstudyprovides thefirstsystematic informationon theoccurrenceof
wildcats throughout the DNP and reveals a priori surprising low abundance of the
species in the protected area. Wildcats like autochthonous Mediterranean scrubland
areas with scrub–pastureland mosaics (Lozano et al. 2003), and may rely for
feeding on rabbits (Gil-Sánchez et al. 1999, Lozano et al. 2006)) or small mammals
(Sarmento 1996, Moleón and Gil-Sánchez 2003, Carvalho and Gomes 2004). These
habitats and prey are common in many parts of the DNP, so one would expect that
wildcats were more abundant in the area than what we have found. Furthermore,
protection of the DNP for more than 5 decades has provided a safe place for wildcats.
Therefore, DNP should hold one of the largest wildcat populations in south western
Spain and be one of the most important areas for conservation of the species.
Wildcat scarcity in DNP might be due to several factors. First, during the
last decades wild rabbit population has decreased everywhere and in the DNP due to
diseases such as myxomatosis and Rabbit Hemorrhagic Disease (RHD) (Thompson
and King 1994, Villafuerte et al. 1995) and to changes in scrubland management
(Moreno and Villafuerte 1995). Abundance of the wildcat population might have
diminished for this reason. In fact, the large home range size of the radio-tracked
individual in our study area suggests low food abundance. However, there are some
areas where rabbits are abundant within the park, and wildcats can also consume
other alternative prey species such as small mammals (Lozano et al. 2003, Malo et
al. 2004), so this reason not fully explain the low abundance of wildcats in the area.
Secondly, the isolation of DNP from the nearest natural areas (Sierra Morena and
Cádiz)duetohumansettlements,widespreadfieldcropsortheGuadalquivirRiver
may also contribute to the low abundance of the species. Wildcats disappeared
201
Chapter 5Chapter 5
from many regions across its range and reached minimum levels at the beginning
of the 20th century (McOrist and Kitchener 1994). The recovery of the species
in several places was possible in the 1990s when anthropic pressure on wildcat
populations and their habitat was reduced (e.g., Parent 1975, Easterbee et al. 1991)
but the isolation and fragmented distribution of the species in Doñana may have
prevented the recovery in spite of the reduction in potential threats. Additionally,
Table 4. Results of the Kruskal Wallis Test to test for differences in the means of several variables between (0) quadrants with no evidence of cat presence, (1) quadrants where wildcats were photographedand(2)quadrantswherecattracksweredetected.Significantvariablesarerepresentedin bold.
Mean rank Kruskal Wallis Test
Variable 0 1 2 Chi-square P-value
Vegetation
%SB 37.03 39.58 29.54 2.574 0.276
%S 33.72 43.17 33.58 1.265 0.531
S_h 38.70 34.92 28.18 4.229 0.121
%B 37.77 35.00 29.54 2.590 0.274
B_h 35.36 27.67 34.86 0.796 0.672
%T 33.66 29.67 36.90 0.796 0.672
T_h 35.35 32.00 33.84 0.193 0.908
Human disturbance
DH 38.41 16.33 33.08 6.637 0.036
Landscape
eBP 38.24 32.83 29.36 3.065 0.216
DW 31.43 34.17 39.12 2.257 0.324
DV 30.81 50.00 36.24 5.168 0.075
Predators
KAId 32.43 40.67 36.08 2.437 0.296
KAIl 36.65 34.50 31.32 1.731 0.421
Prey
R 32.34 27.42 34.00 0.606 0.739
SM 29.25 25.00 38.98 4.963 0.084
Tot 32.40 28.83 33.56 0.312 0.856
202
Chapter 5Chapter 5
an increased mortality rate over time due to potential disease transmission from
domestic cats (e.g., McOrist et al. 1991) could be another causal factor explaining
wildcat scarcity in Doñana. The domestic cat population in the Doñana area might
have increased over the last decades due to the urbanization of the surrounding
neighbourhoods of the protected area. Hence, as in other natural areas (e.g., Ferreira
et al. 2011), feral and/or domestic may spend time sporadically within DNP during
their free-ranging activity potentially transmitting diseases to their wild relatives.
InfactourresultsconfirmthepresenceofdomesticcatswithintheDNPcloseto
humansettlements(Fig.1,Table4).Finally,althoughourresultsdonotconfirm
that the relative abundance of the Iberian lynx can negatively affect the detection
ofwildcats, interspecificinteractionsbetweenwildcatsandtheIberian lynxmay
have also led to the competitive exclusion of the species as carnivores persisting
at low population densities have been suggested to experience an increase in the
effect of intraguild predation (Creel and Creel 1998, Creel 2001, Creel et al. 2001,
Creel and Creel 2002). The European wildcat and the Iberian lynx may potentially
overlap in habitat use of Mediterranean scrubland but intraguild interaction with
the Iberian lynx may have resulted in habitat partitioning in that wildcats will avoid
habitat patches of high lynx densities to the detriment of its own success in prey
acquisition and access to the most suitable habitats. In turn, wildcats may have
been forced to enlarge their home ranges to continue hunting and may even roam
in human-occupied areas increasing their mortality risk. In fact our results revealed
that wildcats were detected more frequently near the anthropic edge of the area
suggesting that individuals could be ranging outside of DNP. Nevertheless, in spite
of the continued decrease of the Iberian lynx population in the Doñana area in
recent decades (Palomares et al. 2012) wildcats have not increased its abundance
203
Chapter 5Chapter 5
intheareaconfirmingthehypothesisthatthelowoccurrenceofthespeciesinDNP
can not be only attributed to Iberian lynx decrease.
In summary, although the DNP is optimal for a large wildcat population the
potential threats explained above may shed light on the low occurrence of the species
inthearea.Nevertheless,manythreatstothespeciesmayremainunidentifiedin
Doñana, so although its low population density makes field studies and direct
observationdifficult,moreresearchanddetailedinformationonoccurrenceaswell
as on wildcat habitat requirements based on radio-tracking efforts are necessary to
provide guidelines for management and conservation of the species in the area.
ACKNOWLEDGEMENTS
This research was funded by the projects CGL2004-00346/BOS (Spanish Ministry of
Education and Science) and 17/2005 (Spanish Ministry of the Environment; National Parks Research
Program). Land-Rover España lent us two vehicles for this work. We are very grateful especially to
J.C.RivillaandS.Desnicaforassistanceduringfieldwork.C.SotowasalsosupportedbyaJAE
Predoc grant from the CSIC.
REFERENCES
Anile S, Bizzarri L, Ragni B (2007) Experiences obtained from camera trapping the wildcat in Sicily
(Italy). Hystrix Italian Journal of Mammalogy Supplement V European Congress of
Mammalogy, 294.
Beyer HL (2004). Hawth’s Analysis Tools for ArcGIS. Available at www.spatialecology.com/htools.
Carvalho JC, Gomes P (2004) Feeding resource partitioning among four sympatric carnivores in
the Peneda-Gerês National Park (Portugal). Journal of Zoology 263: 275–283.
CreelSPG,CreelN(2001)Interspecificcompetitionandthepopulationbiologyofextinction
prone-carnivores. In: Carnivore Conservation. Eds. Gitleman, JL, Funk, SM
MacDonald, D Wayne RK. pp. 35-60. Cambridge University Press.
Creel S (2001) Four factors modifying the effect of competition on carnivore population dynamics
204
Chapter 5Chapter 5
as illustrated by African wild dogs. Conserv Biol 15 (1):271-274.
Creel S, Creel NM (1998) Six ecological factors that may limit African wild dogs, Lycaon pictus.
Animal Conservation 1 (1):1-9.
Creel S, Creel NM (2002) The African wild dog: behavior, ecology, and conservation. The African
wild dog: behavior, ecology, and conservation. Princeton University Press.
Daniels MJ, Balharry D, Hirst D, Kitchener AC, Aspinall RJ (1998) Morphological and pelage
characteristicsofwildlivingcatsinScotland:implicationsfordefiningthe‘wildcat’.J
Zool 244:231-247.
Duarte J, Vargas JM (2001) Son selectivos los controles de predadores en los cotos de caza?
Galemys 13 (special number): 1–9.
Easterbee N, Hepburn LV, Jefferies DJ (1991) Survey of the status and distribution of the wildcat
in Scotland, 1983-1987. Survey of the status and distribution of the wildcat in Scotland,
1983-1987. Nature Conservancy Council for Scotland, Edinburgh.
Garrote G, Gil-Sanchez JM, McCain EB, de Lillo S, Telleria JL, Simon MA (2012) The effect of
attractant lures in camera trapping: a case study of population estimates for the Iberian lynx
(Lynx pardinus). Eur J Wildl Res 58 (5):881-884.
Gil-Sanchez JM, Valenzuela G, Sanchez JF (1999) Iberian wild cat Felis silvestris tartessia predation on
rabbit Oryctolagus cuniculus: functional response and age selection. Acta Theriol 44 (4):421-428
Guilt F, Agudin S, El-Khadir N, Fernandez-Olalla M, Figueredo J, Dominguez FG, Garzon P,
Gonzalez G, Munoz-Igualada J, Oria J, Silvestre F (2010) Factors conditioning the camera-
trappingefficiencyfortheIberianlynx(Lynx pardinus). Eur J Wildl Res 56 (4):633-640.
Ferreira JP, Leitão I, Santos-Reis M, Revilla E (2011) Human-Related factors regulate the spatial ecology
of domestic cats in sensitive areas for conservation. PLoS ONE 6(10): e25970 (2011).
Ferrer M. 1993. El águila imperial. Quercus, Madrid, Spain.
Fernández-Delgado C (1997) Conservation management of an European Natural Area: Doñana
National Park, Spain. En: Groom, M.J., Meffe, G.K., Carroll, C.R. (Eds.), Principles of
Conservation Biology, pp. 536-543. Third Edition. Sinauer Associates, Inc. Publishers,
Sunderland, Massachusetts, USA.
IUCN (2007) European Mammal Assessment. IUCN, Gland, Switzerland and Cambridge, UK.
Jackson R.M., Roe J.D., Wangchuk R., Hunter D.O. 2005. Surveying Snow Leopard populations
205
Chapter 5Chapter 5
with emphasis on camera trapping. A handbook. Sonoma, California, The Snow Leopard
Conservancy: 73.
Karanth KU, Chundawat RS, Nichol JD, Kumar NS (2004) Estimation of tiger densities in the
tropical dry forests of Panna, Central India, using photographic capture-recapture sampling.
Animal Conservation 7:285-290.
Lozano J, Moleon M, Virgos E (2006) Biogeographical patterns in the diet of the wildcat, Felis
silvestris Schreber, in Eurasia: factors affecting the trophic diversity. Journal of
Biogeography 33 (6):1076-1085.
Lozano J, Virgos E, Cabezas-Diaz S, Mangas JG (2007) Increase of large game species in
Mediterranean areas: Is the European wildcat (Felis silvestris) facing a new threat?
Biological Conservation 138 (3-4):321-329.
Lozano J, Virgós E, Malo AF, Huertas DL, Casanovas JG (2003) Importance of scrub–pastureland
mosaics for wild-living cats occurrence in a Mediterranean area: implications for the
conservation of the wildcat (Felis silvestris). Biodivers Conserv 12 (5):921-935.
Lozano 2009. Gato montés–Felis silvestris. In: Enciclopedia Virtual de los Vertebrados Españoles.
(L. M. Carrascal and A. Salvador, Eds.). Museo Nacional de Ciencias Naturales, Madrid.
http://www.vertebradosibericos.org/
Malo AF, Lozano J, Huertas DL, Virgos E (2004) A change of diet from rodents to rabbits
(Oryctolagus cuniculus). Is the wildcat (Felis silvestris) a specialist predator? J Zool
263:401-407.
McOrist S, Kitchener AC (1994) Current Threats to the European Wildcat, Felis silvestris in
Scotland. Ambio 23 (4-5):243-245.
Mohr CO (1947) Table of equivalent populations of North American Small Mammals. Am Midl
Nat 37 (1):223-249.
Moleon M, Gil-Sanchez JM (2003) Food habits of the wildcat (Felis silvestris) in a peculiar habitat:
the Mediterranean high mountain. J Zool 260:17-22.
Monterroso, P., Sarmento, P., Ferreras, P. and Alves, P. C. 2005. Spatial distribution of the European
wildcat (Felis silvestris) in Vale do Guadiana Natural Park, South Portugal. Symposium:
Biology and Conservation of the European Wildcat (Felis silvestris) (ed M. Herrmann), pp.
17. NABU, Germany, January 21st –23rd 2005.
206
Chapter 5Chapter 5
McOrist, S., Boid, R., Jones, T.W., Hubbard, A.L., Easterbee, N., Jarret, O., 1991.Some viral and
protozoal diseases of the European wildcat Felis silvestris. Journal of Wildlife Diseases 27,
693–696.
Moreno S, Villafuerte R (1995) Traditional management of scrubland for the conservation of
rabbits Oryctolagus cuniculus and their predators in Doñana National Park, Spain.
Biological Conservation 73:81-85.
O’Brien TG, Kinnaird MF, Wibisono HT (2003) Crouching tigers, hidden prey: Sumatran tiger
and prey populations in a tropical forest landscape. Animal Conservation 6:131-139.
Palomo LJ and Gisbert J (2002) Atlas de los Mamíferos terrestres de España. Dirección General de
Conservación de la Naturaleza-SECEM-SECEMU, Madrid.
Parent GM (1975) La migration recente, a caractere invasionnel, du chat sauvage, Felis silvestris
Schreber, en Lorraine Belge. Mammalia 39 (2):251-288.
Peichocki, R. 2001. Lebensräume. Die Verbreitung der Wildkatze in Europa. In: H. Grabe and G.
Worel (eds), Die Wildkatze. Zurück auf leisen Pfoten, pp. 14-27. Buch and Kunstverlag
Oberpfalz, Amberg.
Ragni B. 2006. Il gatto selvatico. Salvati dall’Arca. A. Perdisa Ed., Bologna.
Rodriguez A, Delibes M (1992) Current range and status of the Iberian Lynx Felis-Pardina
Temminck, 1824 in Spain. Biological Conservation 61 (3):189-196.
Ruiz-Olmo J (1986) Dades sobre les causes de mortalitat dels carnívors (Mammalia) als massissos
del Montseny i del Montnegre i les seves rodalies. In: II trobada d’estudiosos del Montseny.
Diputación de Barcelona, Servei de Parcs Naturals, Barcelona, pp. 21–23.
Sarmento P (1996) Feeding ecology of the European wildcat Felis silvestris in Portugal. Acta
Theriologica 41: 409-414
Stahl P, Artois M (1991) Status and Conservation of the wild cat (Felis silvestris) in Europe and
around the Mediterranean rim. Council of Europe, Strasbourg.
Stahl P, Leger F (1992) Le chat sauvage (Felis silvestris, Schreber, 1777). En: Artois, M. y Maurin,
H. (Eds). Encyclopédie des Carnivores de France. Société Française pour l’Etude et la
Protection des Mammifères (S.F.E.P.M.), Bohallard, Puceul.
Stahl P, Artois M, Aubert MFA (1988) Organisation spatiale et déplacements des chats forestiers
adultes (Felis silvestris, Schreber, 1777) en Lorraine . Revue d’Écologie (Terre et
207
Chapter 5Chapter 5
Vie), 43: 113-131.
Sunquist M, Sunquist F (2002) Wild Cats of the World. The University of Chicago Press, Chicago.
TempleHJ,TerryA(2007)ThestatusanddistributionofEuropeanmammals.OfficeforOfficial
Publicationsof the European Communities, Luxembourg.
Villafuerte R, Calvete C, Blanco JC, Lucientes J (1995) Incidence of viral hemorrhagic disease in
wild rabbit populations in Spain. Mammalia 59 (4):651-659.
Valverde JA (1967) Estructura de una Comunidad de Vertebrados Terrestres. C.S.I.C., Madrid.
Virgos E, Travaini A (2005) Relationship between small-game hunting and carnivore diversity in
central Spain. Biodivers Conserv 14 (14):3475-3486.
White GC, Garrott RA (1990) Analysis of wildlife radio-tracking data. Analysis of wildlife radio-
tracking data. Academic Press Inc., San Diego, New York etc.
Wild cats: status survey and conservation action plan (1996). Wild cats: status survey and
conservation action plan. IUCN.
Wilson GJ, Delahay RJ (2001) A review of methods to estimate the abundance of terrestrial
carnivoresusingfieldsignsandobservation.WildlRes28(2):151-164.
208
Chapter 5
ANNEX III
Pictures of sporadic sightings of individual wildcats within the DNP (a and
b), wildcat radio-tracked in the study area between December 2009 and March
2010 (c) and wildcat photographed in March 2011 by the personal working in the
Doñana National Park (Centro International de Estudios y Convenciones Ecológicas
y Medioambientales (CIECEM)) (d). Credit pictures: CIECEM (a, b and d) and C.
Soto (c).
209
Conclusiones
210
Conclusiones Conclusiones
1. La abundancia relativa de especies obtenida a través de censos de rastros
depende de la abundancia en sí de las especies censadas pero también de
factores no biológicos condicionados por las condiciones climáticas los días
anteriores al censo y de factores relacionados con el método de censado.
2. Las variables que incrementan la calidad del sustrato sobre el que se
realizan los censos (elevada humedad ambiental, ausencia de viento fuerte,
o un número bajo de días transcurridos desde las últimas precipitaciones)
permiten una mayor detección de rastros en sustratos arenosos. Además,
en función del tamaño de las especies censadas, variables metodológicas
como la distancia del transecto sobre el que se realiza el censo al borde de
la vegetación más cercana o el observador, también afectan al número de
rastros encontrados.
3. Para disminuir la variabilidad en el número de rastros detectados y aumentar
lafiabilidadde losdatosobtenidos se recomienda restringir los censosa
ciertas condiciones climáticas y/o metodológicas, y si no es posible, incluir
en los modelos estadísticos las variables relacionadas con el clima y el
método que potencialmente pueden afectar el número de rastros encontrados
durante los censos.
4. De las cinco especies de carnívoros censadas en toda la zona de matorral del
Parque Nacional de Doñana, el zorro fue la especie más común y ubicua,
encontrándose en el 98.5% de las cuadrículas muestreadas. Le siguió el
tejón (97.1%), el meloncillo (94.2%), la gineta (59.4%) y el lince que ocupó
algo más de un tercio del área muestreada (37.6%).
211
Conclusiones Conclusiones
5. Modelosdeusodelhábitataescalafinaindicaronquelapresenciadelince
estuvo asociada a zonas con una alta cobertura de matorral alto y densidad
de ecotonos entre matorral alto y pastizal, mientras que la presencia de
ginetas se asoció a zonas de pinares con vegetación densa de sotobosque
así como con una alta abundancia de micromamíferos. Los zorros, tejones
y meloncillos mostraron un patrón de selección de hábitat menos marcado,
siendo la abundancia de presas totales y de conejo de monte, o la cobertura
de matorral alto algunos de los predictores de su abundancia y distribución.
6. Análisis de nicho indicaron que ginetas y linces son las especies más
especializadas con nichos ecológicos más estrechos, segregados y
marginales. Los zorros y tejones sin embargo, mostraron un generalismo
manifiesto,situándoseespacialmentesunichoecológicoentreeldezorros
y tejones por un lado y el de linces y ginetas por otro.
7. Cuando a los modelos generales de uso del hábitat se le añadieron las
variables de abundancia relativa o presencia de las otras especies de
carnívoros, se encontró una relación negativa entre la presencia del lince
y la abundancia relativa de zorros, meloncillos y ginetas. Sin embargo, la
abundancia de ginetas y meloncillos estuvo positivamente asociada con la
abundancia relativa de meloncillos y tejones, respectivamente.
8. Las asociaciones encontradas entre especies se enmarcan dentro de las
relacionesinterespecíficasconocidasdentrodelacomunidaddecarnívoros
del Parque Nacional, donde el lince actúa como depredador intragremial
de especies mas pequeñas como zorros, meloncillos y ginetas, y sugieren
212
Conclusiones
la importancia de incluir la presencia y/o abundancia de los grandes
depredadores en los estudios de uso y selección del hábitat de depredadores
pequeños y medianos.
9. Rastros de perros fueron detectados en el 17-23% de la superficie de
matorral del Parque Nacional de Doñana. Hubo mayor probabilidad de
encontrar perros en las áreas cercanas a los límites del Parque Nacional, y
no se encontraron rastros en zonas del interior del Parque, lo que sugiere
que se trata de perros procedentes de los núcleos urbanos y que no hay una
población silvestre establecida en el mismo.
10. Para un control efectivo de perros que usan el interior del Parque Nacional
deberíaactuarsesobrelosnúcleosurbanoscircundantes,asícomointensificar
las medidas de control en los límites del Parque Nacional.
11. La abundancia de gato montés en el Parque Nacional de Doñana fue
sorprendentemente baja a pesar del nivel de protección que ofrece el área de
estudio. Durante dos años de estudio se detectaron solo 52 rastros de gato en
algomásdeunterciodelascuadrículasmuestreadas,ysoloseidentificaron
6 individuos diferentes de gato montés para lo que fue necesario un esfuerzo
de muestreo de 24.4 cámaras/día para detectar cada uno de ellos.
12. La baja abundancia de gato montés en el Parque Nacional podría explicarse por uno
o más factores tales como el descenso de las poblaciones de conejo, el aislamiento
de la población, una posible alta incidencia de enfermedades debido al contacto
con gatos domésticos, y la competencia por interferencia con el lince ibérico.
213
Agradecimientos
En Biología de la Conservación, los estudios de selección de hábitat son cruciales para entender los mecanismos y/o proce-sos que determinan la distribución y abundancia de las espe-cies en el medio natural. Los mamíferos carnívoros constitu-yen un modelo de estudio ampliamente utilizado debido a su papel en los ecosistemas terrestres. Los objetivos generales de esta tesis son proporcionar información sobre los patrones de selección de hábitat a escala fina y las relaciones interespecífi-cas que puedan influenciarlos de distintas especies de carní-voros simpátridos que muestran diferentes rasgos de historias de vida en un área mediterránea protegida.