Habitat use for warren building by European rabbits (Oryctolagus cuniculus) in relation to landscape structure in a sand dune system

of 8

Please download to get full document.

View again

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
8 pages
0 downs
Habitat use for warren building by European rabbits (Oryctolagus cuniculus) in relation to landscape structure in a sand dune system
  Original article Habitat use for warren building by European rabbits(  Oryctolagus cuniculus  ) in relation to landscapestructure in a sand dune system Claudia M. Dellafiore*, Juan B. Gallego Ferna´ ndez, Sara Mun˜oz Valle´ s Departamento Biologı´a Vegetal y Ecologı´a, Universidad de Sevilla, Apdo. 1095, 41080 Sevilla, Spain a r t i c l e i n f o Article history: Received 12 January 2007Accepted 8 February 2008Published online 2 April 2008 Keywords: Habitat managementLandscapeGeographic information system(GIS)Warren building  a b s t r a c t Several conservation efforts are being made to recover European rabbit populations ( Oryc-tolagus cuniculus ) on the Iberian Peninsula. Some of them focus on burrow management;others involve building different types of warren. A few studies have examined siteselection for warren building, and these studies have considered only warren placementwithin sites and not the broader area surrounding these locations. The objective of thisstudy was to evaluate how landscape pattern determines habitat selection by rabbits forwarren building at different spatial scales. Landscape, home range scale, and microhabitatwere the spatial scales used in this study. Warrens were not uniformly distributed over thestudy area but, rather were concentrated in areas with a high abundance and cover of  Retama monosperma and high vegetation cover. Rabbits preferred digging warrens in areaswith low fragmentation and where patches are few, large, and contiguous. Based on ourresults, we suggest that a study of landscape structure should be carried out before designhabitat management, recovery or translocation programs. Such studies will need to takeinto account the physiognomy and size, shape, and continuity of patches in fragmentedlandscapes. Rabbit conservation programs must address areas that provide not only themaximum potential rate of intake, but also good soil and vegetation cover conditions forwarren building and suitable surrounding areas. ª 2008 Elsevier Masson SAS. All rights reserved. 1. Introduction The European rabbit ( Oryctolagus cuniculus ) is a native speciesof the Iberian Peninsula (Monnerot et al., 1994), and is a keyspecies in Spanish Mediterranean ecosystems. The body sizeof rabbits and their ease of capture make them an attractiveprey species. At least 39 predators consume rabbits on theIberian Peninsula (Moreno et al., 1996), including severalthreatened species such as the imperialeagle ( Aquila adalberti )and Iberian lynx ( Lynx pardinus )(Gonza´lez et al., 1990; Beltra´nand Delibes, 1991).Until the 1950s, rabbits were abundant on the IberianPeninsula; however, diseases such as myxomatosis and viralhemorrhagic disease have contributed to a progressive declineinrabbitpopulations(Mun ˜ oz,1960;Argu ¨ elloetal.,1988;Beltra´n,1991)tocurrentlevels,whicharesufficientlylowtothreatentheSpanishMediterraneanecosystem(Villafuerteetal.,1997).Sev-eral efforts are being made to recover rabbit populations, * Corresponding author. Tel.: þ 34 954557069.E-mail address:cdelaflor7@hotmail.com(C.M. Dellafiore). available at www.sciencedirect.comjournal homepage:www.elsevier.com/locate/actoec 1146-609X/$ – see front matter ª 2008 Elsevier Masson SAS. All rights reserved.doi:10.1016/j.actao.2008.02.002 acta oecologica 33 (2008) 372–379  including habitat management (Moreno and Villafuerte, 1995;Calvete and Estrada, 2004), translocation and recovery pro-grams (Villafuerte et al., 2001; Letty et al., 2003; Moreno et al.,2004),vaccinationcampaigns(Calveteetal.,2004a,b),andpred- ator control (Moreno and Villafuerte, 1995). Warren manage-ment has also received some attention as a potentialmanagement alternative (Calvete and Estrada, 2004).For prey animals, the characteristics of their habitat, suchas shelter possibilities and food availability, determine theirdistribution, abundance, and habitat use by individuals(Lagory, 1986; Lima, 1990; Watts, 1991). Habitat structure andthe distribution of suitable and unsuitable habitats are knownto affect the distribution of vertebrates (Rodrı´guez andAndre´n, 1999; Bowman et al., 2000; Reunanen et al., 2002). Inapplied ecology, an understanding of the relationshipsbetween animal distributions, habitat types, and landscapepatterns can help predict the evolution of animal populationswhen habitat changes occur.Foranimalprey,likerabbits,therefugeandfoodavailabilityare fundamental to understand the abundance, distribution,and habitat use of species (Lagory, 1986; Lima, 1990; Watts,1991). Numerous works have been carried out to explain howdifferent environmental variables affect the distribution and/or abundance and/or presence of rabbits in certain habitats(Lombardi et al., 2003; Monzo´n et al., 2004; Ferna´ndez, 2005)but few works have studied the habitat selection from the ref-uge point of view to different space scales.The objective of this study was to evaluate how landscapestructure determines habitat selection for warren building atdifferent spatial scales. Landscape, home range and micro-habitat were the scales used in this study. We address thehypothesis that rabbits are not uniformly distributed in ourstudy area and that rabbits select areas with a high cover of shrub to build warrens. 2. Materials and methods 2.1. Study area The study was conducted on the El Rompido spit, located atthe Piedras River estuary (province of Huelva, SW Spain, 37  12 0 N, 7  10 0 W). The spit stretches east for about 12 km run-ning parallel to the coast, with a width ranging between 300and700 m.Theareaconsistsof527 haofnaturalsandyterrainof which 293 ha is composed of inner stabilized dunes (Fig. 1).The climate has a Mediterranean pattern of winter rain andsummer drought, with a mean annual temperature of 18.2  C and a mean annual rainfall of 620 mm.The El Rompido spit has heterogeneous vegetation com-munities related to local geomorphology, including the beachand active dune system, salt marshes, inner stabilized dunes,slacks and tidal swales. These zones have been previouslyclassified into 29 environmental units according to functionaltrait and vegetation composition (Gallego Ferna´ndez et al.,2006); and include 6 units corresponding to the dune activezone, 10 units to the inner stabilized dune, and 13 units tothe active zone with tidal influence.The active dune zone is mainly covered by Ammophilaarenaria and Elymus farctus . Due to adverse conditions, vegeta-tion is sparse and grows low. A vegetation gradient inlandexists according to local environmental conditions. The innerstabilized dune zone constitutes the main area of the spitsurface (56%) and is largely covered by Retama monosperma (broom thicket), with plants growing to heights of 3.5 m andwith variable cover. The broom thicket occurs with chamae-phytic species such as Thymus carnosus , Artemisia crithmifolia ,and Helychrysum picardii , which show diverse distributionpatterns along the spit. The R. monosperma canopy contributesto the high number of herbaceous plants, largely representedbywinterannuals.Wetdepressionsarecoveredbyhygrophyticvegetation (usually not more that 1.5 m in height), dominatedby the perennial Scirpus holoschoenus and/or Juncus acutus and,occasionally, Scirpus maritimus (depending on water availabil-ity).Thesewetdepressionsarealsocolonizedbyseveralannualherbaceousspecies,especiallywinterannuals.Tidalswalesaredepressions between dune ridges connected with the riverchannelanddominatedbyhalophyticvegetationasinfluencedby water flooding regime and soil salinity gradients. 2.2. Description of spatial scale The spatial scales used to analyze the relationship betweenabundance and distribution of warrens and the landscape Fig. 1 – Geographical location of the study area and R. monosperma distribution (in black). acta oecologica 33 (2008) 372–379 373  structure are similar to those used byFerna´ndez (2005)whostudied the abundance of rabbits in relation to habitatvariables. These spatial scales were: (a) landscape contextmeasurement over environmental units, which representfunctionally homogeneous portions of landscape and areclassified according to geomorphology characteristics andvegetation composition. This scale would be associated withregulation of abundance of warrens at the population level.(b) The home range scale involves the area where rabbitstraverse and do their normal activities. The area used was1 ha because it is the maximum surface that rabbits usuallytravel during the breeding season in sand dunes and it isequivalent to the minimum home range size observed inrabbit populations in similar habitats (Kolb, 1991a). (c) Micro-habitat scale, the lower scale, is related to resource usagewithin the home range. 2.3. Data collection and statistical analysis A total of 110 parallel belt transects 10 m wide were estab-lished over the study area in 2003. The transects were sepa-rated from each other by 100 m, were perpendicular to thespit and crossed the whole width of the study area. Warrensobserved in each transect were georeferenced by a globalpositioning system (GPS, Garmin 12) and data were incorpo-rated into a geographic information system (GIS) program(ArcView 3.2). The transect line method was employed to esti-mate the density of warrens in the study area (Tellerı´a, 1986). A warren was defined as the structure where the rabbitproduced descendants (Southern, 1940; Chapman andSchneider, 1984); it serves as a social bond (Roberts, 1987) and give protection from adverse climatic conditions andattack by predators (Villafuerte et al., 1993). In this study, thewarrens could have one or more entrances and only activewarrens were considered. 2.3.1. Landscape context The number of warrens present was then obtained for eachenvironmental unit described byGallego Ferna´ndez et al. (2006). The chi-square test was used to determine if wildrabbitsuse the environment unit in proportion to its availabil-ity. TheNeu et al. (1974)method was employed to evaluatepreference or avoidance of a given habitat or environmentunit in relation to its availability. If the null hypothesis isrejected in the chi-square test, theNeu et al. (1974)methodcan discriminate between used or avoided habitats withmore frequency than expected, by using the statistical Z (Bonferroni normal statistics;Miller, 1966). 2.3.2. Home range The R. monosperma cover cartography was built from aerialphotographs (scale 1:5000) taken in 2001. This cartographywas created by digitalizing each broom with ArcView 3.2because it is the main shrub vegetation present in the studyarea. Landscape variables were measured within a circulararea with 1 ha of surface using ArcView 3.2. This sampling unit represented the rabbit home range.To predict the presence or absence of warrens from thelandscape variables described later on, 100 circles werecreated and randomly distributed along the study area.When two or more circles overlapped, only one was consid-ered to avoid pseudo replications.To establish the correlation between the abundance of warrens and the different landscape variables, one set of samples was obtained from circles created around each oneof the georeferenced warrens in the field.For each circle, the areas and perimeters of broom patcheswere calculated and, from these data, three landscape indiceswereproduced:perimeterarearatio,heterogeneity,anddiver-sityindex.Theseindiceswereselectedbecauseoftheirpoten-tial association with an essential resource for warren building such as protective vegetation against predators or exposure of warrens to the predator.A simple measure of shape complexity or landscapefragmentation was calculated by means of the perimeter–area ratio. Patches with elongated shapes or indented perim-eters had higher perimeter–area ratios than patches of thesame area with compact shape and unbroken perimeters. Inaddition, small patches had higher perimeter–area ratiosthan large patches of the same shape.Theheterogeneity(IC)anddiversity( D )indicesproposedbyBurelandBaudry(1999)werealsocalculated.TheICexpressestheleveloflandscapefragmentationintheformofcontinuity/discontinuity of vegetation cover at the spatial analysis level(home range), according to the existing number of vegetationpatches. IC is calculated as:IC ¼ log  10 À U 2 Á where U is the number of vegetation patches within eachcircle. IC increases with increasing  U and is independent of vegetation type. IC ¼ 0 when there is only one patch. D expresses the fragmentation of the vegetation andincreases when the number of patches is high and their sizeis small. D is calculated as: D ¼ X P i  N 2 i . S  X S i Âfð S  100 ÞÀ 50 S g where P i is the perimeterof each patch forvegetation type i , N i is the total number of patches for a given vegetation type, S isthe area of the circle (1 ha), and S i is the area of each patch. Inthis area, only one type of vegetation is present ( R. mono-sperma ), so diversity was calculated for that species.A logistic regressionmodel (Walker and Duncan, 1967) wasusedtopredictthepresenceorabsenceofwarrens(SPSS13.0).Thedatesusedforthisanalysiswereobtainedfromcirclesran-domly created overlapping with the georeferenced warrens.Thecircleswereclassifiedaccordingtothepresenceorabsenceof warrens. Logistic regression models are one of the mostcommonly used methods to model the relationship betweena dependent dichotomous variable and a set of independentvariables. The ability ofthe independent variables to discrimi-natebetweenthepresenceandabsenceofwarrenswasexam-ined by univariate logistic regression during the exploratorydata analysis phase. The Hosmer–Lemeshow test was used toevaluate the goodness-of-fit statistics to determine whetherthe model adequately described the data.To understand the relationships between each studyvariable and warren abundance, the Spearman correlation(SPSS 13.0) was calculated. For this analysis, data fromwarrens in the center of the circle were used. acta oecologica 33 (2008) 372–379 374  2.3.3. Microhabitat In the field, we observed if warrens had been dug under thebroom canopy. We calculated broom cover and total plantcover in 50  50 m plots, centered in each warren, using aerialphotographs (scale 1:5000) taken in 2001. Correlation analysiswas used to evaluate the relationship between the abundanceof warrens with R. monosperma and total plant cover. 3. Results 3.1. Warren density and landscape context In the study area, 113 warrens were recorded; the estimateddensity was 129 per km 2 , representing 444 Æ 161 warrens.Warrens were distributed in 8 of the 29 environmental units,and were not found in exact proportion to their occurrence( x 2 ¼ 32.9, p < 0.05). Rabbits built warrens in the units‘‘ R. monosperma with U. membranacea ’’ and ‘‘ R. monosperma with A. crithmifolia ,’’ which were used in a proportion higherthan expected ( Z ¼ 2.18, p < 0.01) (Table 1,Fig. 2). The dune scrub in the units, R. monosperma with T. carnosus , R. mono-sperma with H. picardi , and tidal swales were used in propor-tions lower than expected. Wet depressions and duneseepageareaswereusedinproportiontotheiravailability(Ta-ble 1,Fig. 2). No warrenswere present in the active dune zone. 3.2. Home range Significant differences were observed (Student’s t -test,  p < 0.05) in relation to the sites selected by wild rabbits forwarren building and the different landscape variables. Theplaces where warren building occurred had higher broomcover and lower fragmentation, that is, lower values of theperimeter–area ratio, patch number, IC, and diversity (Table2). The diversity index showed that rabbits would select areaswith a small number of large patches rather than a largenumber of smaller patches.Significantcorrelationswereobservedbetweenallindepen-dentvariables,exceptamongtheperimeter–arearationandIC(Table 3). The Hosmer–Lemeshow test showed that the modelfitthedataadequately(significantvalue0.195).Thelogisticre-gression models predicted that the probability of warren pres-ence is principally related to the perimeter arearatio (Table 4),indicatingthatwarrensaremorelikelytooccurwherethereishigh broom cover and low fragmentation (Fig. 3).The number of warrens was negatively correlated withperimeter area ratio ( r ¼À 0.90, p ¼ 0.037), number of patches( r ¼À 0.80, p ¼ 0.003), diversity index ( r ¼À 0.66, p ¼ 0.015)and IC index ( r ¼À 0.70, p ¼ 0.025). No correlation wasobserved with the vegetation cover area.Warrens were present in areas with lower perimeter arearatios, even though the study area showed higher fragmenta-tion.Whiletheperimeter arearatiosinthestudyareareachedvalues of 8, warrens were built only in locations wherefragmentation was lower than 1.05 (Fig. 4). 3.3. Microhabitat Ninety-eight percent of the warrens were observed to beunder a broom canopy and the remaining 2% were observedin tidal swales with 100% Juncus spp. cover.Broom cover ranged between 0% and 97% and total plantcover ranged between 10% and 98%. Significant correlationswere observed between abundance of warrens and total coverof vegetation ( r ¼ 0.59, p < 0.01) and between warrenabundance and broom cover ( r ¼ 0.58, p < 0.01). 4. Discussion Thepresence/absenceofpelletsisgenerallyusedtoassessthedistribution and/or abundance of rabbits (Monzo´n et al., 2004;Ferna´ndez, 2005). Since the objective of our work was to studyhabitat selection for building warrens, we observed the Table 1 – Warren occurrence in the environmental units of El Rompido spit (units used with proportion greater thanexpected are in boldface) Environment unit Totalsurfacearea(m 2 )Proportion of total m 2 Number of warrensobservedExpectednumberof warrensobservedProportionobservedin each areaConfidence intervalon proportion of occurrence(90% family confidencecoefficient) a Dune scrub 235689 0.07 1 8 0.01 À 0.01  p1  0.02 ( À ) R. monosperma with Thymus carnosus 511168 0.16 7 18 0.06 0.00  p2  0.08 ( À ) R. monosperma with  Artemisia critmifolia 1287682 0.41 74 46 0.65 0.53  p3  0.69 (  D  ) R. monosperma with Helichrysum picardii 468723 0.15 3 17 0.03 À 0.01  p4  0.04 ( À ) R. monosperma with Urticamembranacea 113780 0.04 18 4 0.16 0.07  p5  0.19 (  D  ) Slacks 327000 0.10 5 12 0.04 À 0.01  p6  0.06 ( À )Freshwater arising 20722 0.01 1 0.7 0.01 À 0.01  p7  0.02 nsTidal swales 214023 0.07 5 8 0.04 À 0.01  p8  0.06 ( À )a Z ¼ (1 À (0.1/2  8) ¼ 2.1894. acta oecologica 33 (2008) 372–379 375  distribution of warrens and their relationship to different en-vironmental and landscape variables.Our results showed that, at the different scales studied,warrens always appeared associated with high broom cover,but the reasons for these were different. In the landscapecontext, the warrens were not distributed either randomlyor uniformly. Warrens were concentrated in the centralzone, which corresponded to environmental units ‘‘ R. mono-sperma with A.critmifolia ’’and‘‘ R. monosperma with U.membra-nacea ’’ (Fig. 2). These two environmental units arecharacterized by high abundance and high cover of broomand other vegetation (Gallego Ferna´ndez et al., 2006).Monzo´n et al. (2004)observed that rabbits are more frequentlyobserved in shrubland than in forests and, according to ourresults, a shrubby understorey would be important as rabbitsselected this type of habitat to build warrens.At the home range scale, we observed that rabbitspreferred digging warrens in areas with high broom cover,low fragmentation (large size and low number of patches)and where there is continuity between patches (Table 2).Similar results were obtained byVirgos et al. (2003), whoobserved that the abundance of rabbits was associated withcontinuous areas and that, in fragmented areas, in the land-scape context, abundance was related to scrubland covering.We can surmise that rabbits selected these places becausethey are the only ones available in the study area, we can seethatthestudyareahad ahighfragmentation(especiallyinthebeginning and in the end) (Fig. 4) and that the rabbits activelychose sites with a high broom cover and low fragmentation. ** ns ***** -20.00-15.00-10.00-    R .  m  o  n  o  s  p  e  r  m  a  w   i   t   h   A .  c  r   i   t  m   i   f  o   l   i  a   R .  m  o  n  o  s  p  e  r  m  a  w   i   t   h   U .  m  e  m   b  r  a  n  a  c  e  a   F  r  e  s   h  w  a   t  e  r  a  r   i  s   i  n  g   S   l  a  c   k  s   F   i  n  g  e  r  m  a  r  s   h  e  s   D  u  n  e   S  c  r  u   b   R .  m  o  n  o  s  p  e  r  m  a  w   i   t   h   T .  c  a  r  n  o  s  u  s   R .  m  o  n  o  s  p  e  r  m  a  w   i   t   h   H .   P   i  c  a  r   d   i   i (EUU)    F  r  e  c  u  e  n  c  y   O   b  s  e  r  v  e   d  -   E  x  p  e  c   t  e   d Fig. 2 – Environmental unit utilization pattern obtained using Neu’s method. Environmental units used less than expectedare placed below the zero line; units used more than expected are placed above the zero line. Environmental units were first ranked and then plotted. EUU, environmental unit utilization.Table 2 – Averages and standard error of independent  variables associated with the presence or absence of warren in the study area Variables Without warren With warren Area (m 2 ) 2136.8 Æ 1883.2 4502.5 Æ 2358.8Perimeter/area 1.14 Æ 1.35 0.39 Æ 0.34Patches (no.) 33.04 Æ 20.1 23.03 Æ 21.4 D 27.87 Æ 28.69 11.54 Æ 21.60IC 1.39 Æ 0.39 1.09 Æ 0.60 D , diversity index; IC, heterogeneity index. Table 3 – Correlation matrix between the independent  variables Perimeter/areaPatchesnumber D IC Area Pearsoncorrelation À 0.484** À 0.246* À 0.322** À 0.391**Sig.(bilateral)0.000 0.017 0.02 0.000Perimeter/areaPearsoncorrelation À 0.30 0.225* À 0.022 Sig.(bilateral)0.777 0.30 0.835PatchesnumberPearsoncorrelation0.823** 0.877**Sig.(bilateral)0.000 0.000 D Pearsoncorrelation0.652**Sig.(bilateral)0.000Values in boldface indicate no significant correlation. Significantcorrelation levels: *5% and **1%. acta oecologica 33 (2008) 372–379 376
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks