Goshawk Diet in a Mediterranean Area of Northeastern Spain

Home , List of birds of Spain, Pica (genus)

j. Raptor Res. 28(2):84-92 ¸ 1994 The Raptor ResearchFoundation, Inc.

GOSHAWK DIET IN A MEDITERRANEAN AREA OF NORTHEASTERN SPAIN

SANTI M^SOS^ Departamentde BiologiaAnimal, Facultatde Biologia,Universitat de Barcelona, AvingudaDiagonal, 645, 08028 Barcelona,Catalonia, Spain

ABSTR•CT.--The diet of the goshawk(Accipiter gentilis) is describedthroughout the year in La Segarra, a Mediterranean area of Catalonia (NE Spain) where one of the densestgoshawk populations recorded in Europewas found.Red-legged partridge (Alectoris rufa), Europeanrabbit (Oryctolaguscuniculus), wood pigeon( Columbapalumbus), jay ( Garrulusglandarius), magpie (Pica pica), thrushes( Turdusspp.) and red squirrel (Sciurusvulgaris) formed the bulk of the goshawkdiet. Nestling and fledglingbirds were very importantduring the breedingperiod, but the rabbit was the main sourceof biomassfor mostof the year, especiallyin winter. In the breedingseason, pairs in heavily forestedareas captured more squirrelsand less rabbits than those in lightly forestedareas. Changesin the diet involving a decreasein rabbit consumptionand an increasein the proportionof red-leggedpartridge were detectedfollowing a rabbit population crashcaused by the viral haemorrhagicdisease. KEY WORDS: Accipiter gentilis;goshawk; Spain; Mediterranean; food habits.

Dieta del azor en una zona mediterrfineadel norestede Espafia RESUMEN.--Sedescribe la dieta del azor (Accipitergentilis) a lo largo del afio en la Segarra,una zona mediterrfineade Catalufia (NE de Espafia)donde se encontr6 una de las poblacionesmils densasde azor hasta ahora registradasen Europa. La perdiz roja (Alectoristufa), el conejo(Oryctolagus cuniculus), la palomatorcaz (Columba palumbus), el arrendajo(Garrulus glandarius), la urraca (Picapica), los zorzales y mirlos (Turdusspp.) y la ardilla (Sciurusvulgaris) fueron las presasprincipales. Las avesj6venes constituyeronuna buena parte de las presasdel azor durante el perlodoreproductor, pero el conejofue la principal fuente de biomasadurante la mayor parte del afio, especialmenteen invierno. Las parejas de las zonasmils forestadas capturaron mils ardillas y menosconejos que las parejasde las zonasabiertas. La drfisticareducci6n de las poblacionesde conejocomo consecuencia de la pneumoniahemorrfigica vlrica, condujoa una disminuci6nde su consumoy a un aumentodel de perdiz.

The foodhabits of the goshawk(Accipiter gentilis) 31TCG61, and 31TCF59 in La Segarra County in the have been describedin northern and central Europe northeasternportion of the Iberian peninsula.The relief of the area is tabular, altitude lies between 500-800 m, (e.g., Sulkava 1964, Opdam et al. 1977, Wikman and the climate is a transition between continental Med- and Tarsa 1980, Marquis and Newton 1982, Gosz- iterranean and submediterranean. Natural vegetation czyfiskiand Pilatowski1986, Widgn 1987), but not communitiescover only 30% of the area, the remainder •n the Mediterranean region.Although there are some being occupiedmainly by cerealcrops. Depending on the exposure and soil characteristics,different types of sec- some general descriptionsof goshawk food habits ondary pine forest (Pinus nigra, P. sylvestris,and P. hale- from severalregions of Spain (Morillo and Lalanda pensis)or oak forest (Quercusfaginea and Q. ilex) cover 1972, Veiga 1982, Garrigues et al. 1990, Mafiosa the areas not suitable for agriculture. The southernand et al. 1990), theseare limited to the breedingseason easternparts of the study area are more forested,while and a detailed studyon this subjectin a Mediter- the northernand westernparts are mostlydevoted to crops Nest siteswere classifiedas heavily forestedif more than ranean area is still lacking. The objectivesof this 50% of the area within a 1-km radius of the nest was paper are (1) to describethe diet of a goshawkpop- coveredby wood or lightly forestedif that percentagewas ulation in a Mediterranean area of Catalonia, (2) less than 50%. to analyzediet changesthroughout the year, and (3) From 1987-89 the maximum number of goshawkpairs nesting simultaneouslyin a well-searched 176 km2 area to study diet variation in relation to changesin prey was 22, but the estimatedtotal population from the pat- availability betweenhabitat typesand years. terns of use and distribution of nest siteswas 26 pairs, giving a maximum densityof 1 pair/6.8 km2, one of the STUDY AREA AND GOSHAWK POPULATION highestin Europe (see Kalchreuter 1981, Thissen et al The study area was within the universal transverse 1981, Bijlsma 1991). The meannearest-neighbor distance mercator squares 31TCG50, 31TCG51, 31TCG60, betweenthe geometricmean locationsof the nestingsites

84 JUNE 1994 GOSHAWKDIET IN LA SEGARRA 85

Table 1. Percentageof prey obtained when analyzing was considered,therefore the biomassfigures in this paper goshawkdiet by differentmethods (see Methods section) refer to capturedbiomass. during the nestlingperiod at two nestsin 1989. (N = Quantificationof Diet During the BreedingSeason. number of prey individuals.) Between1985-89, I studiednestling diet (May-July) by repeated visits to nest sites to collect all prey remains (feathers, fur, and bones) and pellets at the nests and OBSER- known pluckingsites. Recently delivered or partially eaten VATIONS NEST prey were recordedas prey remains,but not collected.In FROM HIDE REMAINS PELLETS MIXED 1985 and 1986 nest visits were sporadic.From 1987-89, N = 74 N = 82 N = 29 N = 102 all nests containing chicks were visited every 4 d from hatching to a few days after fledging. To minimize dis- Reptilia 0.0 1.2 0.0 1.0 turbance, sampling was reduced during the laying and Phasianidae 21.6 25.6 6.9 21.6 incubationperiods (April). Columbidae 18.9 15.8 17.2 17.6 The identity of the prey remains and the minimum Estrigiformes 0.0 2.4 0.0 2.0 numberof prey individualsnecessary to explaintheir pres- Picidae 1.3 1.2 17.2 4.9 encewas established for eachvisit, accordingto the number of bonesor flight feathersencountered. All pelletsfrom a Turdidae 17.6 11.0 0.0 8.8 visit were pooled into a single sample and analyzed to- Corvidae 17.6 15.8 20.7 14.7 gether.The presenceof different prey typesin thesesam- Sturnidae 0.0 3.7 6.9 4.9 ples was recorded,but no attempt was made to quantify Other Passer. 9.5 3.7 6.9 4.9 the number of individualsrepresented. To avoid counting Other birds 4.0 2.4 3.4 2.9 the same prey individual twice in the same visit (i.e., in Leporidae 4.0 14.6 10.3 11.8 remains and in pellets), prey found in pellets were com- Sc•uridae 5.4 2.4 10.3 4.9 puted only if they had not been found as remains in the samevisit. I avoidedcounting the sameprey individual in successivevisits by comparingprey from successivecol- lections:prey found in pelletsor as an old remain were of everypair was 1535 m (SD = 455 m, range = 825- not consideredif they had been detectedin the previous 2800 m, N -- 26). Pair dispersion,measured by the G visit as a fresh or partially eaten prey. In both cases, statistic(Brown and Rothery 1978), showeda value of however, all methods of detection were recorded. 0 844, indicatinga regulardistribution of pairs(Tjernberg To assessthe reliability of the methodnoted above, 322.5 1985).The averagelaying date during the 1986-90 period, hours were spent observingin hides installed 15-20 m estimatedby inspectionof the nestsevery two daysfrom away from two nestsin 1989. During the nestlingperiod, mid-March until the first egg was laid, was 5 April + observation started at 1200 H, lasted until 1900 H and 7.96 d (N = 73, range = 21 March-29 April). was continuedthe followingday from 0500-1200 H. The processwas alternated between the two nests until the METHODS youngfledged. Only prey observedbeing deliveredto the Prey Identificationand Classification.Prey remains, nest were recorded and identified with the aid of a 20- bones,fur, nails and feathersfound on nests,plucking sites 60x telescope.Sixty-seven out of 74 (90.5%) prey were and pellets were identified by macroscopiccomparison identified to the species.The remaining sevenprey were with skeletonand skin referencecollections. Arthropods either unidentified small passerinesor nestling birds. The wereonly considered as possible goshawk prey when found results of these observations, which were assumed to be on the nests,but not in pellets, and were identified to an unbiasedsample of the nestling diet, were compared taxonomicorder. I tried to identify all vertebrateremains with the results obtained at the same nestsand year by to species.When possible,the sex and age of prey was pellet countsalone, prey remains alone, and the combi- recorded.For nidicolousbirds, I consideredthree age cat- nation of both as describedabove. The resultsgiven by egories:nestlings, fledglings, and adults. Young red-legged none of thesemethods were significantlydifferent from partridges(Alectoris tufa) wereconsidered nestlings if their those obtained by direct observation,but the combined size was lessthan three-fourths the adult size and fledg- method gave the nearestapproximation (X2 = 7.50, df = lingsif larger.Assigning avian prey to ageclass was based 6, P = 0.277; X2 = 11.47, df = 6, P = 0.075; X2 = 6.09, on size, plumage,feather characteristics, and degreeof df = 6, P = 0.412, respectively;Table 1). However, it ossification.The adult categorymight haveincluded some still overestimatedthe percentageof rabbitsin the diet and youngbirds no longerdistinguishable from adults.Eu- underestimatedthe proportion of thrushes (Turdus sp.) ropeanrabbits (Oryctolagus cuniculus) and red squirrels and other small birds (Table 1). (Sciurusvulgaris) were classifiedas young or adult ac- Quantification of Diet Outside the Breeding Season. cordingto sizeor degreeof ossification.When a preycould Diet outside the breeding season(August-March) was not be classifiedto age, it was not consideredin the age studied from 1986-88 by looking for prey remains at selectionanalysis. Prey found complete or nearlycomplete pluckingsites (Opdam et al. 1977, Ziesemer1983). I tried were weighedwith a springbalance. Otherwise, the live to standardizethe scanningpattern over different months biomasswas estimatedusing bibliographicinformation and to avoid finding prey of common buzzards (Buteo (Geroudet1946-57) or data from the studyarea according buteo)or sparrowhawks(Accipiter nisus) by scanningonly to the age and, if necessary,sex of the prey. No wastage goshawk nesting areas. Two monthly inspectionswere 86 SANTI MA•IOSA VOL. 28, NO. 2

Table 2. Preyitems of goshawkin La Segarraduring 1987-89. Weight in grams.Species with N < 10 are grouped and listed underneath.

N (%) TOTAL WEIGHT (%) Arthropodsa 8 (0.40) 17 (0.00) Reptiles 21 (1.05) 2906 (0.51) Lacertalepida 18 (0.90) 2728 (0.48) Otherreptiles b 3 (0.15) 178(0.03) Birds 1519 (75.85) 326 502 (56.90) Alectoristufa 362 (18.07) 140845 (24.54) Coturnix coturnix 21 (1.05) 2100 (0.37) Columbapalumbus 196 (9.79) 67 169 (11.70) Columbalivia 13 (0.65) 3950 (0.69) Unidentifiedpigeon 39 (1.95) 11470 (2.00) Streptopeliaturtur 28 (1.40) 3920 (0.68) Otusscops 27 (1.35) 2160 (0.38) Athene noctua 18 (0.90) 3060 (0.53) Picus viridis 31 (1.55) 6160 (1.07) Picoidesmajor 15 (0.75) 1200 (0.21) Turdus merula 134 (6.69) 10 989 (1.91) Turdus viscivorus 38 (1.90) 4232 (0.74) Unidentified thrush 25 (1.25) 1887 (0.33) Garrulusglandarius 184 (9.19) 28 197 (4.91) Picapica 54 (2.70) 9345 (1.63) Sturnusvulgaris 79 (3.94) 6516 (1.14) Fringillacoelebs 23 (1.15) 529 (0.09) Unidentifiedpassefine 87 (4.34) 3504 (0.61) Unidentified bird 36 (1.80) 3190 (0.56) Other birdsc 109 (5.44) 16 079 (2.80) Mammals 455 (22.72) 244 486 (42.60) Oryctolaguscuniculus 333 (16.63) 220 526 (38.43) Sciurusvulgaris 86 (4.29) 21 330 (3.72) Other mammalsd 36 (1.79) 2630 (0.46) Total 2003 573 858

Arthropods:Scolopendra sp., Orthopterans,Coleopterans. Other reptiles:Angms fragilis, Psammodromus algirus, unidentified reptiles. Otherbirds: Acczpiter gentills (nestlings from the samenest), Acczpzter nisus, Phasianus colchicus, Scolopax rusticola, Gallinula chloropus, Columbaoenas, Clamator glandarius, Cuculus canorus, Tyto alba, Strix aluco, unidentified owls, Caprimulgus europaeus, Caprimulgus sp, Meropsapiaster, Upupa epops, Galerida sp., Lullula arborea, unidentified lark, Luscinia megarhynchos, Turdus philomelos, Sylvza sp., Parus caeruleus,Parus major, Certhia brachydactila, Orzolus oriolus, Lanius excubitor, Corvus corone, unidentified crow, Passer domestzcus, Sermus serinus,Carduelis carduelis, unidentified Fringillidae, Miliarza calanalta. Other mammals:Crocidura russula, Eliomys quercinus, Microtus duodecimcostatus; Apodemus sylvaticus, Mus spretus,Rattus norvegzcus, Rattus rattus, unidentified mice, unidentified rodents.

made at 10 previouslyselected sites, but fresh remains rowhawk prey could have been confusedwith goshawk found in sporadicvisits to other nestingareas were also prey. They canbe distinguishedby the extentof the feather recorded. I recorded all fresh kills, bones, fur or feathers plucking (larger and usually scatteredin the goshawk) found, and establishedthe minimum number of prey nec- and the presenceof legsor bill remainsleft by the spar- essaryto explain their presenceaccording to the number rowhawk(Opdam 1975).When the prcdatoridentity could of bonesand flight feathers found. Becauseof the char- not be establishedwith confidence,the prey was not con- acteristics of the autumn and winter common buzzard diet sidered. •n Catalonia, consistingmainly of small mammals and Prey Availability Counts.European rabbit counts were Invertebrates(Mafiosa and Cordero 1992), little confusion carried out at dusk 1-5 times each month. A 19.7-km shouldhave arisen with that species.However, somespar- route (A) acrossthe whole studyarea was coveredwith a JUNE 1994 GOSHAWKDIET IN LA SEGARRA 87 vehicle,at a maximum speedof 40 km/h, from July 1987 + 235 g (N = 2003). Arthropodswere incidental to December 1989. All rabbits seen on the route were and in no casedid we have evidencethat they had recordedand abundancewas expressedas number of rabbits seenper kilometer. Another 25.4-km route (B), cov- been captured by the goshawk (i.e., they could be ering only the southof the studyarea, had beentraversed prey of goshawkprey). Reptileswere only consumed in the same way between October 1986-October 1988. during the nestlingperiod. The diet consistedalmost The results of the countsduring the period when both exclusivelyof endothermicvertebrate prey (98.9%). transectswere conductedsimultaneously (July 1987-Oc- Red-legged partridge, European rabbit, wood pi- tober 1988) were used to obtain a conversion index between them, which was used to obtain a estimate of rabbit geon(Columba palumbus), jay (Garrulusglandarius), abundancefor the whole area from October 1986-June magpie (Picapica), blackbird (Turdus merula), Eu- 1987 from countsconducted in route B. To obtain rough ropean starling (Sturnusvulgaris) and red squirrel estimatesof red-leggedpartridge abundance,I conducted formed the 71.3% of goshawks'captures. In terms car countsin April and May during the morningor before dusk, at a maximum speedof 20 km/h. A total of 57 km of biomass,the rabbit was the basicprey, followed in 10 countsof different length and location within the by the red-leggedpartridge and the wood pigeon, study area were done in 1987 and 102 km in 19 different whichaltogether accounted for 74.7% of the captured countsin 1989. Resultswere expressedas number of par- biomass. mdges seenper kilometer. SeasonalVariation. Only 1898 prey individuals Data Analysis and Statistics. Chi-square tests were usedto comparediet compositionby numbersof prey at couldbe assignedto a particular month of the year. different times of the year, and one-way analysisof vari- Frequenciesof capturevaried significantlyby season ance (ANOVA) combined with the Scheffe'stest (Zar (X2 = 144.34, df = 28, P < 0.001; Table 3). Rabbits 1984) were usedto compareaverage prey weights.Vari- and passetinesaccounted for more than 64% of the ations in the diet of the 1987-89 breedingseasons were analyzedby habitats (heavily forestedversus lightly for- prey in the January-April period.In May and June ested)and years. Prey were sortedaccording to the dif- passetinesand game birds were the main prey, but ferent nest sitesand years. Samplescontaining less than rabbits, pigeons,and corvids were also important. 20 prey were discarded(to reducebias causedby differ- In July, passefines lost their preponderantposition entialsampling), leaving 34 diet samplesfrom 18 different in the diet and gamebirds madeup the largestpro- nest sites, totaling 1590 prey items. The coefficientof variation betweensamples in the percentageof each prey portion of it, followed by pigeons,corvids and rab- type in eachsample was calculatedto determinethe degree bits. In the August-December period rabbits were of homogeneityin the consumptionof different groupsof again the main prey, followedby pigeonsand game prey. Chi-square testsfor mutual and partial indepen- birds. In terms of biomass, much less variation oc- dencein three-dimensionaltables were performedfollow- ing Zar (1984). When a two-dimensionalchi-square test curred,the rabbit beingthe dominantprey through- was globally significant,observed cell frequencieswere out the year, especiallyoutside the breedingseason. consideredto be significantlydifferent from the expected Gamebirds had a peakcontribution in May, pigeons frequencieswhen the absolutevalue of the standardized in the July-December period, and corvids in the residualwas >Z•/2. Statisticalsignificance level was set June-July period. at a = 0.05. Statisticalanalyses were performedwith SPSS (1990). The Shannon-Weaverindex (H', log base2) was Globally, diet was more diverse and contained usedto describedietary diversity(Margalef 1982). When smallerprey during the nestlingperiod (May-July, appropriate, mean + standard deviation are indicated. Table 3). ANOVA showedsignificant differences in the averageweight of prey betweenperiods (F4,1893 RESULTS = 16.63, P < 0.01), being lower in May, June and General Diet Description. Samplesfor the nest- July than in the January-April and August-De- ling period (May-July) included27 prey items in cemberperiods (Table 3). BetweenMay and July, 1985-86, 391 prey itemsfrom 13 nestsin 1987, 871 nestlingand fledgling birds accountedfor 37.5% of prey itemsfrom 26 nestsin 1988 and 452 prey items the 781 birds for which age could be determined,or from 12 nestsin 1989. Only 23 prey items were 18.1%of biomass(185 kg). Extrapolatingto all prey, obtainedduring the laying and incubation periods 28.8% of prey and 10.8% of the total biomasscap- (April), and 239 prey were determinedfor the Au- tured were youngbirds. The proportionof immature gust-March period. birds (both nidicolousand precocial)in relation to The diet of goshawksin La Segarraincluded 61 fully grown ones increasedfrom the beginning of differenttypes of prey (Table 2). Prey weightranged May to the end of July (x 2 = 32.38, df = 5, P < from only a few gramsto morethan 1000 g for some 0.001; Fig. 1) as the nestingseason progressed. For adult rabbits.The averageweight of prey was 286 nidicolousbirds alone, the proportion of nestlings 88 SANTI M^i•OS^ VOL. 28, NO. 2

Table 3. Percentagesby numbers(N) and weight (W) of different prey categoriesfound in the diet of the goshawk at different times of the year in La Segarra. (Only N% were testedfor significance.)

JAN-APRIL MAY JUNE JULY AuG- DECEMBER

N W N W N W N W N W

Phasianidae 13.4 16.3 22.4 35.3 17.0 24.6 21.5 17.4 20.3 19.6 Columbidae 14.9 14.4 10.5 12.9 12.8 13.6 16.5 18.8 21.9 a 21.1 Corvidae 0.7 a 0.4 9.4 5.3 14.6 a 8.4 16.2 a 10.5 8.6 3.5 Passeriformes 29.1 4.0 28.3 a 7.3 22.7 5.8 12.8 a 3.5 12.5 a 2.1 Other birds 2.2 a 1.1 4.7 a 1.6 9.9 6.5 12.1 a 5.8 7.8 3.2 Leporidae 35.1a 60.9 15.0 30.8 14.3 35.3 14.5 40.3 26.6a 49.9 Sciuridae 2.2 1.4 5.6 5.0 5.0 4.7 3.4 3.2 0.8 a 0.5 Other prey 2.2 1.3 4.0 1.7 3.7 1.0 3.0 0.2 1.6 0.1 Number of prey 134 446 893 297 128 Total weight (kg) 53.1 122.4 240.7 76.9 50.5 Averageweight (g) 396 + 276 274 + 207 270 + 235 259 + 241 395 + 256 H' 2.26 2.70 2.83 2.80 2.57 a Significantlydifferent from expectedfrequency. decreasedand that of fledglingsincreased throughout 124.80, df = 2, P = 0.01; Fig. 2). Similar but non- the breedingseason (May 32.8% and 14.3%, N = significanttrends were found for pigeonsand star- 119; June 21.3% and 27.1%, N = 314; July 13.6% lings, while thrushes showed a reverse trend (Fig. and 28.8%, N = 66, respectively;X 2 = 14.65, df = 2). The proportionof youngto adult rabbits in May 4, P < 0.005). The proportion of nestling corvids (100%, N = 38), June (69.6%, N = 79) and July decreasedfrom May to July as the proportionof (59.1%, N = 22) showeda significantdecrease (X 2 fledglingsincreased (X 2 = 23.66, df = 4, P < 0.01; = 17.33, df = 2, P < 0.01). The proportionof young Fig. 2). The proportionof youngpartridges captured to adult squirrelsin May (20%, N = 5), June (13%, increasedfrom May (0%) to July (74.6%, x • = N = 8) and July (0%, N = 5) did not show a significant trend (X2 = 1.04, df = 2, P = 0.594). 70 26 Year-to-year and Habitat Variation. Significant I NESTLINGS variation in diet compositionoccurred between the 60 J•] FLEDGLINGS 97 34 samplesanalyzed (X2 = 392.62, df = 231, P < 0.001). Accordingto the coefficientsof variation of 219 the different prey groups,game birds (C.V. = 31.2%), L 40 passerines(C.V. = 36.2%) and corvids (C.V. = • •o 45 159 43.3%) were the prey more homogeneouslyrepre- ,.,J sented in the samples, whereas squirrels (C.V. = z 20 89.2%) and other prey (C.V. = 105.8%) were the most unevenly consumedgroups. Pigeons (C.V. = 10 51.1%), rabbits(C.V. = 57.9%) and other birds (C.V. = 74.6%) showed intermediate levels of variation. i ii I II I II I II Year-to-year and habitat differencesin prey avail- APRIL MAY JUNE JULY ability might be partially responsiblefor this variation. A test for mutual independenceof prey com- INCUBATION[ YOUNGINNEST FLEDGE position, year, and habitat showeddependence of all three variables (X2 = 132.69, df = 37, P < 0.001). Figure 1. Proportion of young birds in relation to all birds (nidicolousand precocial)captured by goshawksin Test for partial independenceshowed habitat being successive2-wk periodsduring the breedingseason. Num- dependenton year and prey (X2 = 94.90, df = 23, bersabove the bars refer to the samplesize used to estimate P < 0.001), year beingdependent on prey and hab- age compositionin each case.The approximate timing of itat (X2 = 70.82, df = 30, P < 0.001) and prey being goshawkbreeding is shownunderneath. dependenton habitat and year (X2 = 121.93, df = JUNE 1994 GOSHAWKDIET IN LA SEGARRA 89

87 I ...... E• ...... E•] ...... MAY 1987 1988 1989 • 20

:i:

__ PARTRIDGE PIGEONS STARLING

JUN• '•2 4 8 ' 4 8 12 4

k40N*H Figure 3. Changes in the index of European rabbit

•z (Oryctolaguscuniculus) availability in La Segarra during the period 1986-89. ½z•ø L•'

L.• •-- 5 Table 4). When the three years were pooled,differencesbetween habitats remained significant(X 2 -- 66.24, df = 7, P < 0.001). Comparedwith lightly forestedareas, diet in heavily forestedareas included significantlyless rabbits and more squirrels(Table 4). Habitat differencesin the proportionof game birds, pigeons,corvids and passefineswere non-sig- ,JULY nificant,but consistentbetween years. Although the Q 20 trends were similar in both habitats, year-to-year variation was significantin the lightly forestedarea (X2 = 33.69, df = 14, P = 0.002) but not in the heavily forestedarea (X2 = 21.54, df = 14, P = 0.09). A decreasein dietary diversitywas noticedin both areasin 1989. After poolingthe two habitats, differencesbetween years remainedsignificant (X 2 = o 41.84, df = 14, P < 0.001). Diet in 1987 was char- PARTRIDGE PIGEONS STARLING CORVIDS TIIRUSHES acterized by a higher proportion of corvids,while F•gure 2. Percentagesof nestlings,fledglings, and adult diet in 1989 was characterizedby a decreasein the birdsof somerelevant groups in the diet of the goshawk •n May, June, and July. Numbers abovethe bars refer to proportionof rabbitsand an increasein that of game the samplesize usedto estimateage compositionin each birds (Table 4). The changesdetected in 1989 fol- case. lowed a declinein the availability of rabbits (Fig. 3), whereaspartridge availability had remainedcon- stant throughoutthe study period (1987:1.02 par- 35, P < 0.001). In consequence,three two-dimen- tridges/km;1989:0.94 partridges/km). sional tables comparing diet between habitats in- dependentlyfor eachyear, and two two-dimensional DISCUSSION tablescomparing diet betweenyears independently The diet of the goshawkin La Segarra showed for each habitat were tested.In all three years, diet three main peculiarities when compared to other differencesbetween heavily forestedand lightly for- European areas: (1) presenceof reptiles, (2) high estedareas were statisticallysignificant (1987, X2 = proportion of red-legged partridges, and (3) high 21.53, df = 7, P = 0.003; 1988, X2 = 39.39, df = 7, proportion of rabbits. The first characteristicwas P < 0.001; 1989, X2 = 18.83, df = 7, P = 0.009; also found in all Iberian localities studied (Morillo 90 SANTI MAI•OSA VOL. 28, NO. 2

Table 4. Goshawkdiet variation in La Segarraaccording to year and habitat. (L: lightly forestedarea; H: heavily forested area.)

1987 1988 1989

L H L H L H

Phasianidae 13.5 17.8 16.5 19.3 22.8 a 24.1 Columbidae 11.9 15.5 8.7 14.0 14.6 17.9 Corvidae 22.7 a 15.5 14.5 10.9 12.2 10.5 Passeriformes 16.7 20.1 20.2 24.6 22.8 30.9 Other birds 7.0 11.5 11.2 9.7 9.1 4.3 Leporidae 21.1b 9.8b 22.4b 10.1b 14.6a,b 4.9b Sciuridae 1.6 6.9 2.5 b 7.5 b 2.4 6.2 Other prey 5.4 2.9 4.0 3.9 1.6 1.2 N 185 174 401 414 254 162 H' 2.74 2.84 2.76 2.82 2.69 2.54

Significantlydifferent from expectedfrequency when comparedwith the samehabitat in other years. Significantlydifferent from expected frequency when compared with theother habitat in the sameyear. and Lalanda 1972, Veiga 1982, Garrigues et al. seasonaltrends detected in dietary diversityand av- 1990, Mafiosa et al. 1990), but only in some Eu- erageweight of prey are consistentwith thosefound ropeanlocalities (Sladek 1963, Goszczyfiskiand Pi- in other Europeanareas (Opdam et al. 1977, Widen latoski1986), andmight be correlatedwith the abun- 1987). This suggeststhat goshawkdiet composition dance of ocellatedlizards (Lacerta lepida) in the in La $egarra was largely determinedby the diver- Mediterranean regionsof the Iberian peninsula. sity and availability of vulnerableprey, which was Game birds,mostly red-legged partridges, was the higherin springand summer.The lowestproportion group most frequently capturedand the least vari- of resident and summer birds (Phasianidae, Corvi- able betweensamples and seasons,which might be dae and other birds) in the diet were reached in the causedby a certain degree of preferenceor local January-April period,and coincidedwith their low- abundanceof that prey. However, even taking into estpopulation levels. This was not foundfor pigeons considerationthat our methodologymay over esti- and passerines,in which the autumn and winter mate the frequencyof rabbits in the diet, in terms populationsmay be increasedby wintering or mi- of biomassthis was the more important prey species grant birds.The abundanceof youngbirds could be for goshawksin La Segarra especiallyoutside the as well a crucial factor determiningthe importance breeding season.Although rabbits have also been of different speciesin the spring and summerdiet, reportedas an importantprey for goshawksin other and the goshawkwould switch from one to another areasof Europe(Tinbergen 1936, Sladek1963, Br611 as they becomeavailable: from May-July, the total 1964, Marquis and Newton 1982), only the Iberian proportion of pigeonsand corvids in the diet in- localities studied so far shared this characteristic in creasedas the proportion of fledgling pigeonsand a consistentgeographic pattern. fledgling corvidsin the diet increased,whereas the The different methods used, as well as true sea- total proportion of passerinesdecreased as fledgling sonal trends, might be partially responsiblefor the thrushes(the main passerinegroup in the diet) de- differencesbetween breeding and non-breedingsea- creased.Also, the increasein the consumptionof son diet, becausesmaller or less conspicuousprey partridgesfrom June-July paralleled the increase might be hard to detectwhen searchingfor pluckings in the proportion of young partridges in the diet. outsidethe nestingseason (Opdam et al. 1977). Also, Similar importanceof youngbirds and mammalsin sevenof the 10 regularly surveyedpairs outsidethe the dietof goshawkshas been reported in otherregions breeding seasonwere in the lightly forestedarea, (Schnell 1958, Sulkava 1964, Opdam et al. 1977, which might have contributedto an overestimateof Wikman and Tarsa 1980, Reynolds and Meslow the proportion of rabbit in the diet of the whole 1984). population at this time of the year. However, the The versatility of feeding by the goshawkwas JUNE 1994 GOSHAWKDIET IN LA SEGARRA 91 furtheremphasized when comparing lightly forested the Departament d'Agricultura, Ramaderia i Pescaof the and heavily forestedareas. Rabbits and corvidson Generalitat de Catalunya. one hand, and squirrelsand pigeons(mainly wood pigeon)on the other, favor farmland and foresthab- LITERATURE CITED itats respectively,and goshawkstook advantageof AEBISCHER,N.J. AND P.A. ROBERTSON. 1993. Com- them differentially in each habitat type. However, positionalanalysis of habitat use from animal radio- the proportionin the diet of other prey types(game tracking data. Ecology74:1313-1325. birds) did not seem to reflect relative abundancein BIJLSMA,R.G. 1991. Trends in European goshawks eachenvironment, which might be a consequenceof (Accipitergentilis): an overview. Bird CensusNews 4. the unit-sum constraintof proportions(Aebischer 347. and Robertson 1993), or of differencesin the vul- BROWN, D. AND P. ROTHERY. 1978. Randomness and nerabilityof theseprey betweenhabitats, regardless local regularity of points in a plane. Biometrika 65. of their abundance. After the sudden decline on rab- 115-122. bit availability in La Segarra,probably as a con- BROLL,H. 1964. Das Leben Deutscher Greifvogel. 2 Aufi., Fischer. Stuttgart, Germany. sequenceof the outcomeof the viral haemorrhagic FERNANDEZ,C. 1993. Effect of the viral haemorrhagic disease(Mafiosa 1991), goshawksshowed a func- pneumoniaof the wild rabbit on the diet and breeding tional responseinvolving a reductionof rabbit con- successof the goldeneagle Aquila chrysaetos (L.). Rev. sumptionand an increasedpredation on red-legged Ecol. Terre Vie 48:323-329. partridge.This response,expressed as a proportional GARRIGUES,R., R. MARTINEZAND J.A. MORATA. 1990 changein rabbit consumption,was larger than that Introducci6nal estudiode la biologladel azor (Accipiter observedin golden eagles(Aquila chrysaetos;Fer- gentilis, L., 1758) en Albacete. Al-basit, Rev. Estud nfindez1993). In the goldeneagle (a rabbitspecialist Albacetenses 27:123-162. in Mediterraneanareas), diet diversityincreased fol- GEROUDET, P. 1946-57. La vie des oiseaux. Vol. 1-6 Collection de Poche. Les Beautes de la Nature. De- lowingrabbit populationcrash. This was lessin the lachaux and Niestle S.A., Paris, France. caseof the goshawk,an essentiallybird-eating raptor GOSZCZYiqSKI,J. AND T. PILATOWSKI. 1986. Diet of which seemsto prey opportunisticallyon rabbits. commonbuzzard (Buteobuteo L.) and goshawk(Ac- The effect on the partridge population of that in- cipitergentills) in the nestlingperiod. Ekol. Pol. 34. creasein goshawkpredation will dependon the nu- 655-667. merical responseof goshawkafter the rabbit pop- KALCHREUTER,H. 1981. The goshawk (Accipitergen- ulation crash. Further long-term monitoring of tilis) in westernEurope. Pages18-28 in R.E. Kenward breedingdensities, breeding success and diet of gos- and I.M. Lindsay lEDS.],Understanding the goshawk. hawks and their prey in La Segarrawould provide The International Associationfor Falconry and Con- a better understandingof the mechanismsunder- servationof Birds of Prey, Oxford, U.K. laying predator-preyinteractions in Mediterranean MAflOSA, S. 1991. Incid•ncia de la pneumonia vmrica del conill sobrela comunitatde rapinyaires segarrencs. agriculturallandscapes. El Medi Natural del Vallgs 3:141-149. -- AND P.J. CORDERO. 1992. Seasonaland sexual variation in the diet of the common buzzard in north- ACKNOWLEDGMENTS eastern Spain. J. Raptor Res. 26:235-238. I am grateful to Fernando Hiraldo and Xavier Ferrer, --, J. REALAND E. SANCHEZ.1990. Comparaci6 who providedguidance throughout this study,to Francesc Uribe, who grantedaccess to skin collectionsat the Museu de l'ecologiade duespoblacions d'astor Accipitergentihs de Zoologiade Barcelona,to JorgePuig for providingthe a Catalunya: el Vall•s Moian•s i la Segarra.El Medz observationtower, and to Joan Real for accessto his skin Natural del Vallgs 2:204-212. and skeletonreference collections. S. Sulkava,S.J. Petty MA•tGALEF,R. 1982. EcologœaEd. Omega,Barcelona, and an anonymousreferee made very usefulcomments on Spain. earlier versionsof the paper. Thanks are alsodue to Man- MARQUIS,M. AND I. NEWTON. 1982. The goshawk in dy Goddardand Marian Reed for improvingthe English Britain. Br. Birds 75:243-260. of the manuscript.The hospitalityand helpof peopleliving MORILLO, C. AND J. LALANDA. 1972. Primeros datos in La Segarrawere greatly appreciated.This studywas partially financedby a grant from the Comissi6Interde- sobrela ecologiade las Falconiformesen los Montes de Toledo. Bol. Estac. Cen. Ecol. 2:57-70. partamentalde Recercai Tecnologia(Generalitat de Ca- talunya), projectAR87-122, a FPI-grant from the Min- OPDAM,P. 1975. Inter and interaspecificdifferentiation isteriode Educaci6ny Ciencia.All nestvisits, hide building with respectto feedingecology in two sympatricspecies and nestobservations were conductedunder permission of of the genusAccipiter. Ardea 63:30-55. 92 SANTI M^•OS^ VOI,. 28, NO. 2

--, j. THISSEN,P. VE•tSCHURENaND G. MOSKENS. TJE•tNBE•G,M. 1985. Spacingof goldeneagle Aquila 1977. Feeding ecologyof a population of goshawk chrysaetosnests in relationto nestsite and food avail- (Accipitergentills). J. Ornithol.118:35-51. ability. Ibis 127:250-255. REYNOLDS,R.T. AND E.C. MESLOW. 1984. Partitioning V•.IG^, J.P. 1982. Ecologiade las rapacesde un ecosis- of foodand nichecharacteristics of coexistingAccipiter temade montafia.Aproximaci6n a su estructuracomu- during breeding.Auk 101:761-779. nitaria. TesisDoctoral. Univ. Complutensede Madrid, SCHNELL,J.H. 1958. Nesting behaviorand food habits Madrid, Spain. of goshawksin the Sierra Nevada of California. Condor WIDIgN, P. 1987. Goshawk predation during winter, 60:377-403. springand summerin a borealforest area of central SI•DEK, J. 1963. Beitragzur Nahrungs6kologiedes Hii- Sweden. Holarct. Ecol. 10:1-7. nerhabitchts.J. Ornithol. 81:44-94. WIKMAN,M. ANDV. TARSA. 1980. Food habitsof the SPSS-INC. 1990. SPSSreference guide. SPSS Inc., Chi- goshawkduring the breedingseason in southwestern cago, IL U.S.A. Finland 1969-77. Suorn. Riista 28:86-96. SULI•V^, S. 1964. Zur Nahrungsbiologiedes Habitchts, Z•t, J.H. 1984. Biostatisticalanalysis. Prentice Hall, Accipiterg. gentills(L.). AquiloSeT. Zool. 3:1-103. EnglewoodCliffs, NJ U.S.A. THISSEN,J., G. MOSKENSAND P. OPDAM. 1981. Trends ZIESEMER,F. 1983. Untersuchungenzum Einflussdes in the Dutch goshawkAccipiter gentills population and Habitchts(Accipiter gentilis) auf Populationenseinen their causes.Pages 28-43 in R.E. Kenward and I.M. Beutetiere.Verlag Gunter Hartmann, Kronshagen, Lindsay[EDS.], Understanding the goshawk.The In- Germany. ternationalAssociation for Falconryand Conservation of Birds of Prey, Oxford, U.K. TINBERGEN,L. 1936. Gegevensover het voedselvon nederlandsehaviken (Accipitergentills gallinarurn (Brehm)). Ardea 25:195-200. Received13 November1992; accepted24 February 1994