Raptor distribution in relation to landscape composition ... - BES journal

Journal of Applied Ecology 0888\ 25\ 143Ð151

Raptor distribution in relation to landscape composition in semi!arid Mediterranean habitats JOSE ł NCHEZ!ZAPATA and JOSE ł A[ SA ł F[ CALVO Departamento de Ecolog(a e Hidrolog(a\ Universidad de Murcia\ Campus de Espinardo\ 29099 Murcia\ Spain

Summary 0[ Breeding sites of raptors were studied in relation to land!use and edge habitat using two di}erent scales in semi!arid Mediterranean landscapes in south!eastern Spain[ Habitat relationships were analysed using Generalized Linear Models[ 1[ The proportion of forest cover at a small scale was the best predictor for all species[ At a larger scale\ the proportion of forest cover was also a good predictor\ and the amount of edge habitat between forest and extensive agriculture was a very good predictor of booted and short!toed eagle densities[ 2[ Models for sedentary species of raptor were similar using both scales whereas trans! Saharan migrant raptors seemed to be more sensitive to larger landscape features that included longer edges between forest and extensive agriculture[ 3[ Habitat mosaics created by forestry and traditional farming were especially impor! tant for Mediterranean raptors[ Strengthening of the Agri!environmental Regulation "1967:81# will be necessary to compensate for agricultural intensi_cation proposals promoted under the Common Agricultural Policy "CAP#[ Key!words] birds of prey\ Common Agricultural Policy\ fragmentation\ modelling\ scale[ Journal of Applied Ecology "0888# 25\ 143Ð151

Introduction Most studies concerning habitat use by raptors focus on {microhabitat| variables such as tree charac! teristics\ ground cover or perches\ often measured at small detailed scales "Fuller 0868^ Andrew + Mosher 0871^ Cody 0874^ Verner\ Morrison + Ralph 0875#[ This has frequently been the pattern in analysing the distribution and densities of many species "Wiens 0878#\ although multiscale studies have shown that speciesÐhabitat associations may vary with scale "Wiens\ Rotenberry + Van Horne 0876^ Kotliar + Wiens 0889#[ Nevertheless\ the development of new techniques that can incorporate habitat data at large scales "Geographic Information Systems# and methods to model distribution and abundance of spec! ies "Generalized Linear Models# have resulted in new approaches to speciesÐhabitat relationships modelling "Vincent + Haworth 0872^ Verner\ Morrison + Ralph 0875^ Flather + Sauer 0885#[ Often\ these studies have been related to the conservation of rare and endang! ered species and the spatial projection of the results

Þ 0888 British Ecological Society

143

Correspondence] Jose A[ Sanchez!Zapata\ Departamento de Ecolog(a e Hidrolog(a\ Universidad de Murcia\ Campus de Espinardo\ 29099 Murcia\ Spain[ Fax] ¦ 23 57 25 28 52^ E!mail] tszapataÝfcu[um[es

on wide distribution ranges "Donazar\ Hiraldo + Bus! tamante 0882#[ Habitat fragmentation\ and especially forest frag! mentation\ is a topic related to habitat selection that has been central in conservation biology and land! scape ecology "Wilcove 0874^ McGarigal + McComb 0884#[ Forest fragmentation has been documented as a major cause of biodiversity loss and the decline of endangered forest and shrubsteppe birds "Saunders\ Hobbs + Margules 0880^ Knick + Rotenberry 0884#[ However\ in semi!arid Mediterranean regions in Spain\ forest fragmentation can be considered as a process associated both with ancestral human activi! ties "Le Honerou 0870# and major climate limitations resulting in patchy and changeable forest landscapes[ Today\ the policies of the Common Agricultural Pol! icy are changing many of these traditional practices[ Agricultural intensi_cation and land abandonment has detrimental e}ects for many bird species "Bignal + McCracken 0885^ Pain + Pienkowski 0886^ Blanco\ Tella + Torre 0887#[ These processes are especially intense in south!eastern Spain where interbasin water transfers promote the development of new intensive irrigation projects[ Unfortunately\ detailed studies on raptor habitat!relationships in semi!arid regions are few and many aspects of the ecology of some Med!

144 Landscape\ raptors and agricultural policies

iterranean raptors remain unknown "Sanchez!Zapata et al[ 0884# hence the impact of these processes cannot be judged[ The objetives of this paper are] "i# to analyse the distribution and density of forest raptors breeding in a semi!arid region of Spain^ "ii# to discuss the role of scale in the distribution and density patterns^ and "iii# to provide advice on landscape and agricultural management to bene_t raptors[

Materials and methods STUDY AREA

The study area covered the Murcia Region\ a 00 206 km1 area located in south!eastern Spain with an arid and semi!arid Mediterranean climate and mean annual rainfall ranging between 199 and 499 mm[ For! ests covered an area of 1070 km1 "08=17) of the region#[ The size of most forest patches was in the range 29Ð299 ha with only a few × 0999 ha[ The mini! mum patch size considered in this study was 04 ha and the largest was 12 731 ha "Fig[ 0#[ The main tree species is Pinus halepensis Mill[^ other species such as Quercus

Þ 0888 British Ecological Society\ Journal of Applied Ecology\ 25\ 143Ð151

Fig[ 0[ Distribution and shape of forest patches in the study area[

rotundifolia L[\ Pinus pinaster Aiton and Pinus nigra Arnold are much less common[

RAPTOR SPECIES AND CENSUS METHODS

Four tree!nesting raptor species were studied\ each with di}erent breeding ranges but similar body size[ The breeding distribution of booted eagle Hieraaetus pennatus Gmelin "maximum weight 864 g# and short! toed eagle Circaetus gallicus Gmelin "maximum weight 1299 g# is con_ned to the Mediterranean and Eurasia[ Both species are trans!Saharan migrants[ Common buzzard Buteo buteo L[ "maximum weight 0253 g# breeds across Europe and Asia[ European populations are mainly sedentary with local winter movements\ whereas many Asiatic populations migrate in winter[ The northern goshawk Accipiter gentilis L[ "maximum weight 0498 g# has a Holarctic breeding range[ Breeding populations are sedentary with some local winter movements "del Hoyo\ Elliot + Sargatal 0883#[ The breeding sites of tree!nesting raptors were cen! sused during 0880[ Complementary information for the period 0874Ð89 was also included[ The _eld work

145 J[A[ Sanchez! Zapata + J[F[ Calvo

was carried out at the start of the breeding season "from late February for common buzzards in coastal woodlands to early May for northern goshawks in higher mountain ranges# because breeding behaviour is conspicous at this time\ for instance nuptial displays and nest building\ making it easier to locate breeding sites[ Surveys carried out after these _rst stages of breeding may miss birds that have failed to breed "Newton 0868#[ Censusing methods included sightings with binocular and scopes from roads and tracks\ transects in areas of di.cult access\ nest searching\ enquires and play!back recording "Fuller + Mosher 0870^ Sanchez!Zapata et al[ 0884#[ Common buzzards\ booted and short!toed eagles were easily detected because they responded particularly well to tape rec! ordings of conspeci_cs and were very conspicous dur! ing the courtship "Sanchez!Zapata et al[ 0884#[ The northern goshawk is the hardest species to detect\ but on the other hand\ it is a scarce and valuable species "both for the naturalist and nest!robbers# and most breeding sites were well known[

function and a link function[ A linear predictor "LP# is de_ned as the sum of e}ects of the predictor vari! ables as follows] LP a ¦ bx0 ¦ cx1 ¦ ==[ where a\ b\ c [ [ [ are parameters to be estimated from the observed data and x0\ x1\ [ [ [ are the explanatory variables[ The error function will depend on the nature of the data[ For density response variables "number of breeding pairs# the Poisson distribution is an adequate error function "Vincent + Haworth 0872#[ One appro! piate link function for a Poisson distribution is the discrete Poisson function[ This means that the number of breeding sites in an area is a discrete\ s!shaped function when the linear predictor is a _rst!order poly! nomial or a bell!shaped function for second order polynomials[ We _tted each explanatory variable to the observed data and chose a 0) level of signi_cance "Nicholls 0878^ Donazar\ Hiraldo + Bustamante 0882^ Gibbons et al[ 0883#[ For regression analyses we used the program STATISTIX "Analytical Software 0881#[

HABITAT DATA

The location of breeding sites was incorporated into a Geographic Information System "GIS# using the Universal Transverse Mercator "UTM# grid of 0!km1 cells[ For the _rst small!scale landscape analysis the 0!km1 cells map was transformed into a 8!km1 "2 × 2 km# cells map\ so the regional map of 00 206! km1 cells was transformed into a map comprising 0270 cells each of 8 km1[ The large!scale landscape analysis focused on 099!km1 "09 × 09 km# cells[ The same GIS was used to characterize the breeding sites including the following groups of variables "Tables 0 and 1#] 0[ Land!use variables[ Di}erent types of land!use were obtained from maps produced by the Spanish Ministry of Agriculture "0 ] 199 999#[ 1[ Edge variables[ These were measured as the length "km# of edges between di}erent land uses using the digitized land!use map[ As a result of increasing the size of cells\ many of them included large areas of sea and adjacent regions that were not censused[ These cells were excluded from the data analysis\ resulting in the large!scale study having fewer breeding territories for all species "see Results#[

GENERALIZED LINEAR MODELS


We used Generalized Linear Models "GLMs# "Dob! son 0872^ McCullagh + Nelder 0878# to construct models of breeding densities[ GLMs allow for a wider range of relationships between the response and explanatory variables and the use of error for! mulations when the normal error for a traditional regression is not applicable[ Three components have to be de_ned for a GLM] a linear predictor\ an error

Results DISTRIBUTION AND ABUNDANCE OF FOREST RAPTORS

Common buzzard was the commonest species with 021 breeding sites in the 8!km1 grid and 090 territories in the 099!km1 grid[ Booted and short!toed eagles had a similar number of breeding sites both in the 8!km1 grid "66 and 57\ respectively# and the 099!km1 grid "57 and 42\ respectively#[ Northern goshawks were the scarcest species with 38 breeding sites in the 8!km1 grid and 26 in the 099!km1 grid[ Booted and short!toed eagles had the highest den! sities in the 099!km1 cells\ with up to 03 breeding sites per cell\ followed by common buzzard with a maximum of nine breeding sites per cell\ and northern goshawk with a maximum of three breeding sites per cell[

RELATIONSHIP BETWEEN RAPTORS AND LAND USE AT THE 8!KM 1 SCALE

The proportion of forest cover explained more of the variation in breeding densities than any other variable "Table 2#[ This relationship between forest cover\ com! mon buzzards and booted eagles was described by a bell!shaped function "Fig[ 1#[ The individual species models predicted maximum breeding densities of booted eagles in cells with around 74) of forest cover\ common buzzards in cells with around 79) of forest cover and northern goshawks and short!toed eagles in cells with 099) of forest cover[ Forest cover accounted for 29) and 12) of the variation in booted eagle and northern goshawk den!


Table 0[ Variables used to characterize the habitat of tree!nesting raptors Land!use variables] measured as the proportion ")# of land!use cover AINTA HINTA AEXTA HEXTA SHRUB FOREST SHF

Arboreous intensive agriculture[ Includes di}erent irrigated crops such as orange and lemon trees[ Herbaceous intensive agriculture[ Highly intensive crops such as tomato\ lettuce and melon[ Arboreous extensive agriculture[ Includes di}erent nonirrigated crops mostly almond\ olive and carob trees[ Herbaceous extensive agriculture[ Mostly barley and wheat _elds[ Includes a wide variety of native species[ Rosmarinus of_cinalis L[\ Quercus coccifera L[\ Rhamnus lycioides L[\ Stipa tenacissima L[ Mostly pine trees Pinus halepensis Mill[ Mixed shrub!forest[ Wide variety of shrubs with sparse "always under 19)# trees "mostly Pinus halepensis Mill[#[

Edge variables] measured as length "km# of edge between two di}erent land!uses EAIA FOIA IASH IASF FOEA EASH EASF FOSH FOSF SHSF

Edge between extensive and intensive agriculture[ Edge between forest and intensive agriculture[ Edge between intensive agriculture and shrub[ Edge between intensive agriculture and mixed shrub!forest[ Edge between forest and extensive agriculture[ Edge between extensive agriculture and shrub[ Edge between extensive agriculture and mixed shrub!forest[ Edge between forest and shrub[ Edge between forest and mixed shrub!forest[ Edge between shrub and mixed shrub!forest[

Table 1[ Mean and SD of the di}erent variables used to characterize the habitat[ See Table 0 for variable codes

Land use AINTA HINTA AEXTA HEXTA SHRUB FOREST SHF Edges IAEA FOIA IASH IASF FOEA EASH EASF FOSH FOSF SHSF


Scale 2 × 2 "N 0270#

09 × 09 "N 77#

Mean

SD

Mean

SD

3=32 5=35 09=58 23=77 08=99 03=63 4=87

9=22 9=32 9=34 9=73 9=48 9=53 9=23

4=52 6=80 01=82 16=33 08=78 05=63 5=11

9=75 0=24 0=26 1=07 0=43 0=70 9=63

0=18 9=07 9=69 9=96 0=73 3=16 0=99 9=73 9=68 9=16

9=96 9=91 9=94 9=90 9=98 0=24 9=95 9=94 9=95 9=92

05=89 1=47 09=94 9=73 10=50 40=07 00=11 09=20 09=27 2=07

0=48 9=40 0=39 9=10 1=41 2=65 0=49 0=94 0=57 9=40

sity models\ respectively\ with lower values for both common buzzard "04)# and short!toed eagle models "02)# "Table 2#[ The relationships with intensive agricultural cat! egories were null or negative for all species\ but the association with some categories of extensive agric! ulture was positive for booted and short!toed eagles[

RELATIONSHIP BETWEEN RAPTORS AND LAND USE AT THE 099!KM 1 SCALE

The proportion of forest cover also explained more of the variation in breeding densities of forest raptors at a large scale "Table 3#[ The relationship for all the species followed an s!shaped function "Fig[ 1#[ Maximum predicted densities were reached in cells with around 69) of forest cover\ which was the maximum forest cover found in 099!km1 cells in the study area[ Forest cover accounted for 33) of the variation in the booted eagle density model with lower values for short!toed eagle "13)#\ northern goshawk "19)# and common buzzard "01)# "Table 3#[ The relationships with intensive agriculture cat! egories were negative for all the species\ but the associ! ation with extensive agriculture categories was posi! tive for booted eagle and common buzzard "Table 3#[ RELATIONSHIP BETWEEN RAPTORS AND EDGE HABITATS AT THE 8!KM 1 SCALE

The length of edge habitat between forest and exten! sive agriculture "FOEA# was the best edge variable to describe breeding densities of forest raptors\ except for northern goshawks "Table 4#[ The relationship between FOEA and trans!Saharan migrants followed an s!shaped function\ whereas the relationship between FOEA and sedentary species followed a bell! shaped function "Fig[ 2#[ RELATIONSHIP BETWEEN RAPTORS AND EDGE HABITATS AT THE 099!KM 1 SCALE

The length of edge habitat between forest and exten! sive agriculture "FOEA# was an even better parameter


Table 2[ Response of forest raptors to land use at the 8 km1 scale[ See Table 0 for variable codes Hp) AINTA HINTA AEXTA HEXTA SHRUB FOREST SHF

NS 03\7 NS 7\5 NS 29\1 5\1

Ð ¦¦ ¦¦ Ð

Cg)

Bb)

NS NS NS 2\7 NS 02\1 NS

0\6 1\5 NS 4\9 NS 04\9 1\9

¦ ¦

Ag) Ð Ð Ð ¦¦ ¦

NS NS NS 4\8 NS 12\1 3\4

Ð ¦ ¦¦

Species^ Hp Hieraaetus pennatus\ Cg Circaetus gallicus\ Bb Buteo buteo\ Ag Accipiter gentilis[ ) Percentage of deviance explained[ NS\ not signi_cant\ P × 9=90^ \ P ³ 9=90^ \ P ³ 9=990[ ¦ positive s!shaped function[ ¦ ¦ positive bell!shaped function[ Ð negative s!shaped function[ Ð Ð negative bell!shaped function[

Fig[ 1[ Number of breeding pairs "per 099 km1# in relation to forest cover ")# at the 8!km1 "solid line# and 099!km1 "dashed line# scales[

Table 3[ Response of forest raptors to land use at the 099!km1 scale[ See Table 0 for variable codes Hp) AINTA HINTA AEXTA HEXTA SHRUB FOREST SHF Þ 0888 British Ecological Society\ Journal of Applied Ecology\ 25\ 143Ð151

3\7 19\8 5\1 5\4 10\5 33\0 3\7

Cg) Ð Ð ¦¦ ¦¦ ÐÐ ¦ ¦

NS 04\3 NS NS 00\0 13\9 01\3

Bb)

Ð

ÐÐ ¦ ¦¦

NS 5\4 7\5 NS 1\7 00\4 8\4

Ag)

Ð ¦¦ Ð ¦ ¦¦

NS 04\0 NS NS NS 19\2 7\0

Ð

¦ ¦



Table 4[ Response of forest raptors to di}erent categories of habitat edges at the 8!km1 scale[ See Table 0 for variable codes Hp) EAIA FOIA IASH IASF FOEA EASH EASF FOSH FOSF SHSF

4\3 NS NS NS 06\1 7\8 NS NS 7\8 NS

Ð

¦ Ð

¦¦

Cg)

Bb)

NS NS NS NS 6\7 NS NS NS 5\1 0\1

1\2 NS NS NS 6\8 0\2 0\0 NS 2\7 NS

¦

¦¦ ¦

Ag) ÐÐ

¦¦ Ð ¦ ¦¦

6\7 0\2 NS NS 6\1 3\9 NS NS 1\4 NS

Ð ¦

¦¦ Ð

¦


Fig[ 2[ Number of breeding pairs "per 099 km1# in relation to the length of edge between forest and extensive agriculture ")# at the smaller scale "8 km1#[

to describe breeding densities of tree!nesting raptors at a large scale "Table 5#[ This association for all species followed an s!shaped function "Fig[ 3#[ Maximum pre! dicted densities were reached in cells with the highest length of edges between forest and extensive agric! ulture[ FOEA accounted for a high proportion of the variation in booted and short!toed eagle density models "31) and 23)\ respectively#\ but this was far lower in the northern goshawk and common buzzard models "02) and 00)\ respectively# "Table 5#[

Discussion Þ 0888 British Ecological Society\ Journal of Applied Ecology\ 25\ 143Ð151

How the spatial con_guration of habitats a}ects the distribution and abundance of organisms is a question increasingly associated with landscape ecology "Fla! ther + Sauer 0885#[ Our ability to detect environ! mental heterogeneity depends on the scale of our

measurements\ whereas the ability of organisms to respond to such patchiness depends on how they per! ceive the environment "Wiens 0878#[ It seems that the response of raptors to habitat fragmentation took place at a larger scale than that of individual terri! tories\ as has been noted for other bird groups "Hinsley et al[ 0884^ Knick + Rotenberry 0884#[ Our initial analysis showed that breeding densities of four species were linked to forest cover but the response varied from an s!shaped function for the northern goshawk and the short!toed eagle to a bell! shaped function with di}erent optimal forest cover for the other two species[ When increasing the scale\ large agroforestry landscapes seemed to be selected by booted and short!toed eagles\ both trans!Saharan migrants[ Thus\ by increasing the scales the models _tted better for both these species[ On the other hand\ the model _tted better at the smaller scale for the northern goshawk\ but not so for the common buz! zard at either scale[ Di}erences between migrant and sedentary species may be related to territory size of the species but there is no information available on territory size of booted and short!toed eagles\ although preliminary results from radiotracking studies suggest that these are large "× 09 km1# "J[E[ Mart(nez\ unpublished data#[ The response of booted eagles to forest cover was a bell! shaped function at a small scale but an s!shaped func! tion at a larger scale[ This result seems contradictory but may re~ect the relationship with large agroforestry landscapes with high forest cover in small or frag! mented patches[ In our study area the maximum forest cover in a 099!km1 grid was 69)[ This could explain the relationship of booted and short!toed eagles with the forest edge!extensive agriculture variable and the high proportion of variation it explained[ The two trans!Saharan migrant raptors bred at high densities in more open woodlands than the two sed!


Table 5[ Response of forest raptors to di}erent categories of edge habitat at the 099!km1 scale[ See Table 0 for variable codes Hp) EAIA FOIA IASH IASF FOEA EASH EASF FOSH FOSF SHSF

03\4 NS 6\7 NS 30\6 02\4 5\5 4\3 02\2 3\20

Ð Ð ¦ ¦ ¦¦ ¦¦ ¦ Ð

Cg)

Bb)

06\2 NS NS NS 23\1 5\5 06\7 NS 05\1 NS

NS 1\8 NS NS 00\9 NS 09\8 NS NS NS

¦ Ð ¦¦ ¦¦

Ag)

¦

¦ ¦¦

NS NS NS NS 01\7 NS 03\9 NS 6\6 NS

¦¦ ¦ Ð ¦¦ ¦ ¦


Fig[ 3[ Number of breeding pairs "per 099 km1# in relation to the length of edge between forest and extensive agriculture ")# at the larger scale "099 km1#[


entary species[ This could be related to the important role of reptiles and edge!nesting birds in their diets and the incorporation of open and edge foraging habi! tats in the models "Sanchez!Zapata et al[ 0884#[ Range size of northern goshawks largely depends on habitat quality and food availability\ ranging from 19 to 46 km1 in boreal forests but being smaller "09Ð 19 km1# in fragmented woodlands in Central Europe "Kenward 0871^ Kenward + Widen 0878#[ The lower densities and sparse distribution of this species may be related to territorial behaviour and intraspeci_c competition "Solonen 0882# or with smaller!scale mic! rohabitat features "Beier + Drennan 0886#[ Range size\ as well as breeding performance\ of common buzzards are probably associated with rabbit Oryctolagus cuniculus L[ availability "Sanchez!Zapata et al[ 0884^ Austin + Houston 0886#[ Breeding sites of common buzzard were widespread across the study area\ typically occupying a variety of forest and agro! forestry landscapes "Gibbons et al[ 0883#[ Therefore\

it was not surprising that univariative models pro! vided such a poor _t[ Controversies about the role of competition in structuring animal communities "Schoener 0871^ Wiens 0878# may be associated with the use of a single scale for all species in the community "Wiens 0878#[ Habitat niche overlap occurs among many species\ and may be responsible for interspeci_c competition if food niche overlaps too[ Northern goshawk\ booted eagle and common buzzard have similar body sizes and prey on the same key species "rabbits\ pigeons Columba livia Gmelin\ and lizards Lacerta lepida Dau! din#\ although the proportion of these key species in the diet varies between raptor species and localities[ In south!eastern Spain\ trophic niche overlap was more than 69) "Sanchez!Zapata et al[ 0884#[ These raptor species also overlap in distribution[ Thus\ interspeci_c competition may be limiting their numbers[ Di}er! ences between habitat models may also re~ect partial e}ects of interspeci_c competition[ Interspeci_c relationships with Buteo and Bubo species may be one factor limiting breeding densities of northern gos! hawks in North America "Kenward 0885#\ whereas nest success of common buzzards may be in~uenced by interference from neighbouring buzzards in Cen! tral Europe "Kostrzewa 0885#[ Further research is needed to _nd out if there is any distributional e}ect of interspeci_c competion in this community of medium! sized forest raptors[ The migratory habits of the species may be deter! mined by an association with di}erent landscape structures[ Flather + Sauer "0885# found that Neo! tropical songbird migrants were more sensitive to landscape structure than were either temperate migrants or residents in North America[ They sug! gested that Neotropical migrants are dependent on landscape structure in a way that is unique when com! pared to other species with di}erent migratory habits[ Similar results have been obtained with Austral and trans!Saharan migrants "Hinsley et al[ 0884^ Chesser


+ Levey 0887# indicating that large!scale choices are more likely to be made by migrants[ They may not colonize {suitable| habitat patches if the surrounding landscape\ at the larger scale\ does not appear suitable "Hinsley et al[ 0884#[ Our results suggest a similar trend for trans!Saharan migrant raptors\ although the number of species and the study scale are much smaller[

MANAGEMENT IMPLICATIONS

In the Mediterranean Basin\ man!made _res and other practices such as clearcuttings or small!patch crops have in~uenced natural ecosystems since the middle Pleistocene "Le Honerou 0870#[ For several decades agricultural practices that created mosaic landscapes have increasingly been abandoned\ and mountain habitats are subject to less human impact "wood! cutting\ grazing and raising cattle\ crops# than pre! viously[ Agricultural policies in EEC countries tend either to favour the establishment of forest plantations instead of non!irrigated crops or to promote irrigated cultivation[ An increase in the number of tree plan! tations and irrigated cultivation may adversely impact trans!Saharan migrant raptors and other endangered Mediterranean raptor species\ such as the Bonelli|s and the Spanish imperial eagles "Ferrer + Harte 0886#[ Conservation of key species or habitats may target particular patches or landscape fragments for man! agement\ whereas programmes emphasizing species richness or complex communities may concentrate on preserving broader!scale landscape mosaics "Wiens 0878#[ Our results suggest that the conservation of Mediterranean raptors needs a regional approach that considers the habitat mosaics created by forestry and agriculture[ Similar relationships between widlife and extensive agriculture have been described for di}erent bird groups\ including raptors "Bignal + McCracken 0885^ Pain + Pienkowski 0886^ Blanco\ Tella + Torre 0887#[ The Zonal Programmes that enhance low! intensity farming systems supported under the CAP Agri!environmental Regulation "1967:81# should be encouraged[ This will mean more coordination and planning in the implementation measures but\ also\ the incorporation of conservation criteria in local decisions[

Acknowledgements


We thank Jose E[ Mart(nez\ Miguel A[ Sanchez\ Ser! gio Egu(a\ A[ Ortun½o\ R[ Mart(nez and D[ Carmona for their assistance with the _eld work[ Jose Antonio Donazar provided helpful suggestions on a previous draft[ Lee Alexander helped to improve the English[ We are indebted to M[ Ferrer and S[J[ Pettty for valuable comments on the manuscript[

References Analytical Software "0881# An Interactive Statistical Analysis Program for IBM PC[ Analytical Sotware\ St[ Paul[ MN[ Andrew\ J[M[ + Mosher\ J[A[ "0871# Bald Eagle nest site selection and nesting habitat in Maryland[ Journal of Wildlife Management\ 35\ 272Ð289[ Austin\ G[E[ + Houston\ D[C[ "0886# The breeding per! formance of the Buzzard Buteo buteo in Argyll\ Scotland and a comparison with other areas in Britain[ Bird Study\ 33\ 035Ð043[ Beier\ P[ + Drennan\ J[E[ "0886# Forest structure and prey abundance in foraging areas of northern goshawks[ Eco! logical Applications\ 6\ 453Ð460[ Bignal\ E[M[ + McCracken\ D[I[ "0885# Low!intensity farm! ing systems in the conservation of the countryside[ Journal of Applied Ecology\ 22\ 302Ð313[ Blanco\ G[\ Tella\ J[L[ + Torre\ I[ "0887# Traditional farming and key foraging habitats for chough Phyrrocorax phyr! rocorax conservation in a Spanish pseudosteppe land! scape[ Journal of Applied Ecology\ 24\ 121Ð128[ Chesser\ R[T[ + Levey\ D[J[ "0887# Austral migrants and the evolution of migration in New World birds] diet\ habitat and migration revisited[ American Naturalist\ 041\ 200Ð 208[ Cody\ M[L[\ ed[[ "0874# Habitat Selection in Birds[ Academic Press\ Inc[\ San Diego[ Dobson\ A[J[ "0872# Introduction to Statistical Modelling[ Chapman + Hall\ London[ Donazar\ J[A[\ Hiraldo\ F[ + Bustamante\ J[ "0882# Factors in~uencing nest site selection\ breeding density and breed! ing success in the bearded vulture Gypaetus barbatus[ Jour! nal of Applied Ecology\ 29\ 493Ð403[ Ferrer\ M[ + Harte\ M[ "0886# Habitat selection by immature Spanish imperial eagles during the dispersal period[ Jour! nal of Applied Ecology\ 23\ 0248Ð0253[ Flather\ C[H[ + Sauer\ J[R[ "0885# Using landscape ecology to test hypotheses about large!scale abundance patterns in migratory birds[ Ecology\ 66\ 17Ð24[ Fuller\ M[R[ "0868# Spatiotemporal ecology of four sympatric raptor species[ PhD Thesis\ University of Minnesota\ Min! nesota[ Fuller\ M[R[ + Mosher\ J[A[ "0870# Methods of detecting and counting raptors] a review[ Studies in Avian Biology\ 5\ 124Ð135[ Gibbons\ D[\ Gates\ S[\ Green\ R[E[\ Fuller\ R[J[ + Fuller\ R[M[ "0883# Buzzards Buteo buteo and ravens Corvus corax in the uplands of Britain] limits to distribution and abundance[ Ibis\ 026\ S64ÐS73[ Hinsley\ S[A[\ Bellamy\ P[E[\ Newton\ I[ + Sparks\ T[H[ "0884# Habitat and landscape factors in~uencing the pres! ence of individual breeding bird species in woodland frag! ments[ Journal of Avian Biology\ 15\ 83Ð093[ del Hoyo\ J[\ Elliot\ A[ + Sargatal\ J[\ eds[ "0883# Handbook of the Birds of the World[ Volume 1[ New World Vultures to Guineafowl[ Lynx Edicions\ Barcelona[ Kenward\ R[E[ "0871# Goshawk hunting behaviour and range size as a function of food and habitat availability[ Journal of Animal Ecology\ 40\ 58Ð79[ Kenward\ R[ "0885# Goshawk adaptation to deforestation] does Europe di}er from North America< Raptors in Human Landscapes "eds D[Bird\ D[Varland + J[Negro#\ pp[ 122Ð133[ Raptor Research Foundation[ Academic Press\ San Diego[ Kenward\ R[ + Widen\ P[ "0878# Do goshawks Accipiter gentilis need forests< Some conservation lessons from radio tracking[ Raptors in the Modern World "eds B!U[ Meyburg + R[D[ Chancellor# WWGBP\ Berlin[ Knick\ S[T[ + Rotenberry\ J[T[ "0884# Landscape charac! teristics of fragmented shrubsteppe habitats and breeding passerine birds[ Conservation Biology\ 8\ 0948Ð0960[



Kostrzewa\ A[ "0885# A comparative study of nest!site occu! pancy and breeding performance as indicators for nesting! habitat quality in three European raptor species[ Ethology Ecology and Evolution\ 7\ 0Ð07[ Kotliar\ N[B[ + Wiens\ J[A[ "0889# Multiple scales of patchi! ness and patch structure] a hierarchical framework for the study of heterogeneity[ Oikos\ 48\ 142Ð159[ Le Honerou\ H[N[ "0870# Impact of man and his animals on Mediterranean vegetation[ Ecosystems of the World\ II\ Mediterranean Shrublands "eds F[ Di Castri\ D[W[ Goodall + R[L[ Specht#\ pp[ 368Ð410[ Elsevier Scienti_c\ Amster! dam[ McCullagh\ P[ + Nelder\ J[A[ "0878# Generalised Linear Modelling[ Chapman + Hall\ London[ McGarigal\ K[ + McComb\ W[C[ "0884# Relationships between landscape structure and breeding birds in the Oregon Coast Range[ Ecological Monographs\ 54\ 124Ð 159[ Newton\ I[ "0868# Population Ecology of Raptors[ T[ + A[D[ Poyser[ Berkhamsted[ Nicholls\ A[O[ "0878# How to make biological surveys go further with generalised linear models[ Biological Con! servation\ 49\ 40Ð64[ Pain\ D[J[ + Pienkowski\ M[W[ "0886# Farming and Birds in Europe[ Academic Press\ London[ Sanchez!Zapata\ J[A[\ Sanchez\ M[A[\ Calvo\ J[F[ + Esteve\

M[A[ "0884# Ecolog(a de Las Aves de Presa de la Region de Murcia[ Servicio de Publicaciones de la Universidad de Murcia\ Murcia\ Spain[ Saunders\ D[A[\ Hobbs\ R[J[ + Margules\ C[R[ "0880# Bio! logical consequences of ecosystem fragmentation] a review[ Conservation Biology\ 4\ 07Ð21[ Schoener\ T[W[ "0871# The controversy over interspeci_c competition[ American Scientist\ 69\ 475Ð484[ Solonen\ T[ "0882# Spacing of birds of prey in southern Finland[ Ornis Fennica\ 69\ 018Ð032[ Verner\ J[\ Morrison\ M[L[ + Ralph\ C[J[ "0875# Wildlife 1999] Modelling Habitat Relationships of Terrestrial Ver! tebrates[ The University of Wisconsin press\ Wisconsin[ Vincent\ P[J[ + Haworth\ J[M[ "0872# Poisson regression models of species abundance[ Journal of Biogeography\ 09\ 042Ð059[ Wiens\ J[A[ "0878# Spatial scaling in ecology[ Functional Ecology\ 2\ 274Ð286[ Wiens\ J[A[\ Rotenberry\ J[T[ + Van Horne\ B[ "0876# Habi! tat occupancy patterns of North American shrubsteppe birds] the e}ects of spatial scale[ Oikos\ 37\ 021Ð036[ Wilcove\ D[S[ "0874# Nest predation in forest tracks and the decline of migratory songbirds[ Ecology\ 55\ 0100Ð0103[ Received 20 March 0887^ revision received 18 January 0888