sign in sign up
<<
Home , American crow, Eastern cottontail, Northern goshawk, Nuthatch, Red squirrel

j. RaptorRes. 39(3):264-273 ¸ 2005 The Raptor ResearchFoundation, Inc.

NORTHERN GOSHAWK DIET IN : AN ANALYSIS USING VIDEO RECORDING SYSTEMS

BRETT L. SMITHERS 1 Departmentof Range,Wildlife and FisheriesManagement, Tech University, Lubbock,TX 79409-2120 U.S.A.

CLINT W. BOAL U.S. GeologicalSurvey, Texas Cooperative and WildlifeResearch Unit, TexasTech University, Lubbock,TX 79409-2120 U.S.A.

DAVID E. ANDERSEN U.S. GeologicalSurvey, Minnesota Cooperative Fish and WildlifeResearch Unit, Universityof Minnesota, St. Paul, MN 55108-6124 U.S.A.

ABSTRACT.--Weused video-recording systemsto collect diet information at 13 (Ac- cipitergentilis) nests in Minnesota during the 2000, 2001, and 2002 breeding seasons.We collected4871 hr of video footage, from which 652 prey deliverieswere recorded. The majority of prey deliveries identifiedwere (62%), whereasbirds (38%) composeda smallerproportion of diet.Mammals accountedfor 61% of biomassdelivered, and avian prey items accountedfor 39% of prey biomass. Sciuridsand leporidsaccounted for 70% of the identifiedprey. Red (Tamiasciurus hudsonicus), easternchipmunk (Tamiasstriatus), and snowshoehare (Lepusamericanus) were the dominant mammals identified in the diet, while AmericanCrow (C0rvusbrachyrhynchos) and Ruffed (Bonasa umbellus) were the dominant avian prey delivered to nests.On average,breeding goshawksdelivered 2.12 prey items/d, and each deliveryaveraged 275 g for a total of 551 g delivered/d. However,daily (P < 0.001) and hourly (P = 0.01) deliveryrates varied among nests. Delivery rates (P = 0.01) and biomassdelivered (P = 0.038) increasedwith brood size.Diversity and equitabilityof prey usedwas similar among nests and was low throughout the studyarea, most likely due to the dominance of in the diet. KEYWORDS: NorthernGoshawk; gentilis; dieP,, Minnesota; prey diversity; red squirrel; Tamiasciurus hudsonicus.

DIETA DE ACCIPITER GENTILIS EN MINNESOTA: UN AN•LISIS BASADO EN SISTEMAS DE GRA- BACION EN VIDEO

RESUMEN.--Empleamossistemas de grabaci6n en video para recolectar informaci6n sobre la dieta de Acdpitergentilis en 13 nidos ubicadosen Minnesotadurante las temporadasreproductivas de 2000, 2001 y 2002. Obtuvimos4871 hr de grabaci0n,a partir de las cualesregistramos 652 entregasde presas.La mayoffa de las presasentregadas que identificamosfueron mamiferos (62%), mientrasque las aves (38%) representaronuna proporci6nmenor de la dieta.Los mamiferos y lasaves representaron el 61% y el 39% de la biomasaentregada, respectivamente. Los scifiridosy lep6ridosrepresentaron el 70% de las presasidentificadas. Los mamiferos predominantesidentificados en la dieta fueron Tamiasciurus hudsonicus,Tamias striatus y Lepusamericanus, mientras que las aves11evadas a los nidos predominante- mente fueron Corvusbrachyrhynchos y Bonasa umbellus. En promedio, los individuosnidificantes entre- garon 2.12 presas/d,y cada entrega tuvo un promedio de 275 g, para un total de 551 g entregados/d. Sin embargo,las tasasdiarias (P < 0.001) y horarias (P = 0.01) de entrega de presasvariaron entre nidos. Las tasasde entrega (P = 0.01) y la biomasaentregada (P = 0.038) incrementaroncon el tamafio de la nidada. La diversidady equitabilidadde las presasconsumidas fueron similaresentre nidosy bajas a travfs del firea de estudio,probablemente debido a la dominanciade T. hudsonicusen la diem. [Traducci6ndel equipo editorial]

Presentaddress and correspondingauthor: P.O. Box 1363,Meeker, CO 81641U.S.A.; email: [email protected]

264 SEPTEMBER 2005 BIOLOGY 265

The Northern Goshawk (Accipitergentilis) is a eastern Minnesota (46ø50'N, 92ø11'W) as describedby large, -dwelling raptor generally associated Boal et al. (2001) and Roberson(2001; Fig. 1). The study area elevation ranged from ca. 200-400 m. Mean summer with mature , coniferous, or mixed for- and winter temperatureswere 18øC and - 11øC,respec- ests (e.g., Bright-Smith and Mannan 1994, Siders tively, and maximum and minimum temperature records and Kennedy 1996, Beier and Drennan 1997, for the region were 40øCand -46øC, respectively(Daniel Squiresand Reynolds1997). Goshawkresearch in and Sullivan1981). Annual precipitationaveraged 60-70 cm. The studyarea was dominated by , mixed-hard- has been conducted primarily in wood, boreal, and second-growthforests with the western half of the continent (Boal et al. 2003). community typesinterspersed among forest stands(Tes- Consequently,there is little published literature ter 1995). describingecology of the speciesin the Western GoshawkNests. Nests included in this studywere con- Region (WGLR) of North America, sidered as sampling units and were selected from all known occupiednests in the studyarea (Boal et al. 2001). where it is currently listed as a Migratory Non- With the exception of one nest, where few data were col- game of Management Concern by the U.S. lected during 2000, diet information wasnot collectedat Fish and Wildlife Service (Region 3) and as a sen- any nest for more than one breeding season.Nests were sitivespecies by the U.S. ForestService (Region 9) selected randomly within the constraintsof accessibility due to lossof habitat (Reynoldset al. 1992). and to include different land ownerships.Thus, our sam- ple is not truly random and may not be representativeof Depending on region, season,and availability, the goshawkpopulation of our study area. However, to goshawkscapture a wide variety of prey and are examine the applicabilityof our diet data to the goshawk consideredprey generalists(Squires and Reynolds population as a whole, we examined prey diversity and 1997,Squires and Kennedy2005). Althoughbreed- overlap among nests. High overlap and low diversity would suggestprey use wassimilar among goshawkpairs ing-seasondiet composition has been studied for and that our data were representativeof the population many populations (e.g., Meng 1959, Grzybowski in general. and Eaton 1976, Boal and Mannan 1994, Younk Video Recording. We used VHS (Model SL 800, Se- and Bechard1994, Lewis2001), site-specificstudies curity Labs©, Noblesville,IN U.S.A.) and 8-mm video re- of diet are necessaryfor developingmanagement cording systems(Sony © Model M-350, Fuhrman Diversi- fied, Inc., Seabrook, TX U.S.A.) with color or strategiesfor goshawkpopulations at regional and black-and-white cameras (Model CCM-660W, Clover Elec- local levels (e.g., Reynoldset al. 1992). A number tronics©, Los Alamitos, CA U.S.A.). Cameras were m- of records exist of prey items collected opportu- stalled on nest within 0.6 m of the nest or, for cam- nisticallyat goshawknests in the WGLR (Eng and eras with zoom lenses, on an adjacent up to 9 m from the nest. Video recorders were placed in weather- Gullion 1962, Apfelbaum and Haney 1984, Martell proof casesca. 30 m from the base of each camera tree. and Dick 1996), but these reports are anecdotal Coaxial-video cables were used to convey power to and and provide a prey list rather than a quantitative transmit images from the cameras. Recorders were pro- assessmentof food habits (Roberson et al. 2003). grammed to record from 0530-2100 H (15.5 hr of foot- Methods used in goshawkfood habits research age) at the 48-hr (1.3 frames/sec) or the 72-hr (0 8 frames/sec) setting to optimize the amount of tape used have included indirect (i.e., identification of prey per samplingsession and battery life. We replacedtapes remains or contents of regurgitated pellets) and and batteries every 3-4 d. direct observationsof prey deliveries to nests Prey Identification. To identify prey delivered to nests, (Meng 1959, Grzybowskiand Eaton 1976, Bosa- we reviewedvideo footage until a prey deliveryoccurred, kowski and Smith 1992, Boal and Mannan 1994). then advanced frame by frame and freeze-framed to fa- cilitate prey identification.We identified avian and mam- Indirect methods of assessingraptor diet can lead malian prey by morphologicalfeatures and developed a to biased results (e.g., Bielefeldt et al. 1992), list of prey speciesdelivered by goshawksto all nests(Ta- whereasdirect methods should provide the least- ble 1). Goshawksmay cache prey and retrieve cached biasedresults (Collopy 1983, Marti 1987,Boal and prey items (Boal and Mannan 1994), which could bias estimatesof deliveryrates and proportional useof Mannan 1994). During the breeding seasonsof in the diets. We attempted to identify cached prey on 2000-02, we usedvideography as a modified meth- basisof a successive,iterative processthat included com- od of direct observationof prey deliveriesto ex- paring prey itemsusing flesh color,pelage or feather con- amine diet of Northern Goshawks in northern dition, and time of delivery from review of video footage, Minnesota. and then remove those items thought to be cached from analysis.

METHODS Age and Biomass Estimation. We assignedavian prey to age categories(e.g., adult, juvenile, or nestling) based StudyArea. The studyarea waslocated in the Lauren- on plumage (e.g., feathers and down) and amount of tian Mixed-Forest Province of north-central and north- sheathing on flight feathers (Reynolds and Meslow 266 SMITHERSET AL. VOL.39, No. 3

•V•AO ßLSP eDEE4 _ 'DIll2. MCD3 - 'PMT,, 3 2/ 1 . 1 WRI• STE eSH2

.

ß Ne• Loc•i on

• 25 50 i00 150 200

Figure 1. Study area and distribution of Northern Goshawknests in Minnesotawhere food habitsinformation was collectedduring the 2000-02breeding seasons. The three-letterdesignations indicate individual nests. Breeding seasondiet informationcollected at the DTR breedingarea was omitted from all analysesbecause of nestfailure. Breedingareas with similar prey composition are indicated with the samesuperscripts. Superscripts indicate cluster number (see Fig. 2).

1984). We categorizedmammalian prey as adultsor ju- mates were based on published information on mam- venilesbased on size (Bielefeldt et al. 1992). Becauseof malian and avian speciesoccurring in the studyarea difficultyin estimatingage of smallmammals, we consid- (Burt and Grossenheider1980, Jones and Birney 1988, ered all mammalssmaller than chipmunksto be adults. Dunning 1993, Dunn and Garrett 1997, Dunn 1999, Sib- Biomassfor partial prey itemswas calculated using the ley 2000). We calculatedmass for nestlingsfollowing proportion of prey deliveredto nests,and proportions Bielefeldtet al. (1992) using 100% of the adult massfor were estimatedqualitatively (e.g., 50% of adult size). warbler-sizedspecies, 65% of the adult massfor robin We estimatedbiomass for preyidentified to family,ge- andjay-sizedspecies, and 55% of the adultmass for large nus, or speciesand used the mean mass of both sexes such as grouse.We calculatedmass of juvenile (Reynolds and Meslow 1984, Lewis 2001). Biomassesti- red squirrel (Tamiasciurushudsonicus), eastern 2005 BIOLOGY 267

Table 1. Number, percent occurrence,and biomassof mammalianand avian prey deliveredto Northern Goshawk nests(N = 13) in Minnesota,2000-02. Values represent pooled number of prey identifiedat nestsduring the 2000, 2001, and 2002 breeding seasons.

BIOMASS PREYCATEGORY COMMONNAME N PERCENT (g) PERCENT Mammals Tamiasciurushudsonicus red squirrel 202 31.0 38046 23.6 Tamiasstriatus 95 14.6 8108 5.0 Lepusamericanus snowshoehare 31 4.8 41027 25.5 Sylvilagusfloridanus 7 1.1 7654 4.8 Sduruscarolinensis easterngray squirrel 3 0.5 1679 1.0 Peromyscusspp. 2 0.3 47 0.0 Family: i 0.2 18 0.0 Mustelafrenata long-tailed 1 0.2 210 0.1 Unknown (MSC1) a 8 1.2 186 0.1 Unknown mammal (MSC2) a 9 1.4 1720 1.1

Birds Corvusbrachyrhynchos American 37 5.7 14515 9.0 Bonasa umbellus 33 5.1 18448 11.5 Aythyaspp. diving 12 1.8 11360 7.1 Cyanodttacristata Blue 8 1.2 664 0.4 Fulica americana 6 0.9 3338 2.1 Turdusmigratorius 3 0.5 205 0.1 Quiscalusquiscula Common Grackle 3 0.5 341 0.2 Family:Icteridae blackbird 3 0.5 189 0.1 Picoidesspp. 3 0.5 199 0.1 Dryocopuspileatus 3 0.5 861 0.5 Unknown duckling 4 0.6 400 0.2 Butorides virescens Green 2 0.3 420 0.3 Perisoreuscanadensis GrayJay 2 0.3 142 0.1 Agelaiusphoeniceus Red-wingedBlackbird 2 0.3 105 0.1 varia Barred 1 0.2 394 0.2 Buteoplatypterus Broad-wingedHawk 1 0.2 455 0.3 : Calidris 1 0.2 73 0.0 Bucephalaclangula 1 0.2 900 0.6 Accipitercooperii Cooper'sHawk 1 0.2 439 0.3 Gallusspp. domesticchicken b I 0.2 Coccothraustesvespertinus EveningGrosbeak I 0.2 59 0.0 Pipilo erythrophthalmus I 0.2 41 0.0 Genus: Euphagus 1 0.2 63 0.0 Accipitergentilis Northern Goshawk 1 0.2 820 0.5 Picoidesvillosus Hairy Woodpecker 1 0.2 66 0.0 Charadriusvociferus Killdeer 1 0.2 97 0.1 Anasplatyrhynchos 1 0.2 1082 0.7 Sitta canadensis Red-breasted 1 0.2 10 0.0 Seiurusaurocapillus Ovenbird 1 0.2 19 0.0 Catharusfuscescens Veery 1 0.2 31 0.0 Unknown nestling 33 5.1 1190 0.7 Unknown bird (ASC1) a 18 2.8 173 0.1 Unknown bird (ASC2) a 23 3.5 1778 1.1 Unknown bird (ASC3) a 6 0.9 3459 2.1 Items not identified to class Mammalia or Aves 76 11.7

MSC1 -- mouse-sizedprey item; MSC2 -- red squirrel-sizedprey item; ASC1 = warbler-sized prey item: ASC2 = robin-sizedprey •tem; ASC3 = Ruffed Grouse-sizedprey item. Omitted from analysis. 268 SMITHERS ET AL. VOL. 39, NO. 3

( Tamiasstriatus), snowshoe (Lepusamericanus), and birds delivered among nests over 5-d intervals, because easterncottontail (Sylvilagusfloridanus) using 95% of the of missingdata among sampleddays, with a Kruskal-Wal- adult mass;if agescould not be determined reliably,we lis single-factorANOVA (Zar 1999). Becauseobservations assignedjuvenile massesto thesespecies. within breeding areas were not independent, we exam- To estimate biomass of unidentified prey, we pooled ined differencesin provisioningrates among breeding unidentified birds into three a prior/sizeclasses (SC) fol- areas with multivariate repeated measuresANOVA. We lowing Storer (1966) and Kennedyand Johnson (1986) used the General Linear Model (GLM) module of STA- that representedaverage mass of common speciesin our TISTICA (Version 6.0, StatSoft, Inc., Tulsa, OK U.S.A.) studyarea: SC1 = 10 g (e.g.,warbler-sized), SC2 = 77 g for all statisticalanalyses except calculationof diet over- (e.g.,robin-sized), and SC3 -- 576 g (e.g.,Ruffed Grouse lap and similarity,for which we used EcologicalMeth- [Bonasaumbellus]-sized). Similarly, we pooled unidenti- odology 6.1 (Exeter Software,Setauket, NY U.S.A.). An fied mammal prey into two a priorisize classes: SC1 = 23 alpha level of P = 0.05 was used for all statisticaltests, g (e.g., mouse-sized)and SC2 = 192 g (e.g., squirrel- and we present means and standarderrors. sized).

Prey and BiomassDelivery Rates. We calculateddeliv- RESULTS ery rates on the basisof number of prey deliveredper day,number of prey deliveredper nestlingper day,and Video Recording and Prey Identification. We in- number of prey deliveredper day at nestswith one, two, stalled video monitoring systemsat three, five, and and three nestlings.We calculatedbiomass estimates in the same manner. We calculatedmean delivery ratesover seven occupied goshawknests during the 2000, 5-d intervals from hatching to 5 d post-fledging (i.e., 2001, and 2002 field seasons,respectively. We from 0-45 d). placed camerasat nestswhen nestlingswere ca. 8 Prey Diversityand Overlap. We calculatedprey diver- d old (+1.18; range = 1-18 d). One of the 15 nests sxtyfor the study area using ungrouped prey categories (i..e,using each prey category identified to family,genus, failed within 3 d of camera placement and was re- or species separately). Because samples were smaller moved from analysis.Due to camera malfunctions, when examining individual nests, we generalized prey we were only able to collect 16 hr of footage at one into similar speciescategories (Lewis 2001) to calculate of the nests in 2000. We placed a camera at the prey diversityfor individual nests.The generalizedprey categoriesfor among-nestdiversity assessment were: (1) 2002 nest of the same pair, but pooled data from Scxurids,(2) blackbirds and Corvids, (3) Leporids, (4) both years as one nest area for analysis.Thus, our Ruffed Grouse, (5) diving (Aythyaspp.), (6) water sampleof 4801 hr (0•= 320 +- 42 hr/nest) of video and shore birds, (7) ,(8) Picidae, (9) - footage is derived from 13 nesting pairs of gos- iforms, (10) miscellaneousmammals (e.g., long-tailed . weasel [ Mustetafrenata] ) . We calculatedprey diversityusing Williams (1964) and We identified 59 (8.3%) of 711 prey deliveriesas MacArthur's (1972) modified form of the Simpson'sin- being retrievals of cached items. Of the 652 fresh dex (Simpson 1949) and diet equitability using Smith prey deliveries,we identified451 (69%) to the spe- and Wilson's index of evenness(Smith and Wilson 1996). cieslevel, 20 (3%) to genus,four to family (1%), We used prey identified to family, genus, or speciesto estimate diet overlap among nestswith the Simplified and four (1%) as unidentifiable ducklings(Table Morisita's Index of Overlap (Krebs 1999). Overlap mea- 1). Eighty(12%) birdsand 17 (3%) mammalswere suresare designedto measurethe degree that two spe- unidentifiablebeyond class,and we were unable to cies share a set of common resources or utilize the same identify 76 (12%) deliveries.The majorityof prey parts of the environment (Lawlor 1980). Overlap mea- deliveries identified to at least class(N = 576) were sures are scaled from zero to one, where zero overlap indicatesdissimilarity in resourceuse, and one indicates mammals (62%), whereasbirds (38%) comprised complete overlap (Krebs 1999). We also assessedsimilar- a smaller proportion of diet. lty in prey use among nestswith cluster analysisusing When considering only those deliveries identi- averagelinkage clustering(Romesburg 1984, Krebs 1999, McGarigal et al. 2000). As suggestedby Romesburg fied to family or finer resolution (i.e., to genus or (1984), we used the un-weightedpair-group method us- species;N = 476), the dominantprey specieswere ing arithmetic averages(UPGMA). red (41.2%), easternchipmunks (19.8%), StatisticalAnalysis. We used analysisof variance (AN- American (7.7%), Ruffed Grouse (6.9%), OVA) to examine relationshipsbetween delivery rate var- and snowshoe (6.5%). No other individual iables and brood size using log-transformeddata (Zar 1999). Biomassof prey delivered per day per nest was speciesaccounted for >5% of identifiedprey. As a transformedby taking the logarithmof biomassdelivered group, Sciuridsand Leporids(N = 338) accounted per day and adding 1.0 (Zar 1999). Normality of exper- for 70% of the identified prey.Among mammals, imental error wastested using the Shapiro-Wilktest pro- 51.8% were adults, 25.4% were juveniles,and we cedure, and assumptionsregarding homogenousvarianc- es were tested using Levene's test (Zar 1999). We were unable to estimate age for 22.8%. Of the examined differences in the number of mammals and birds, 36.7% were adults, 9.6% were juveniles, SEPTEMBER 2005 BIOLOGY 269

27.5% were nestlings,and we could not reliably brood size.On average,daily biomass delivered was estimateage for 26.2%. 509 g (+84 g) to nestswith one nestling, 555 g Biomass.In context of the prey speciesand bio- (-+42 g) to broodsof two, and 756 g (-+107 g) to massproportion usedby goshawksin our study,the broods of three. Despite greater amounts of bio- delivery of one domesticchicken (Gallusspp.) was massbeing provided to larger broods, this resulted unusualand the masswould dramaticallyinfluence in nestlingsin singlebroods receiving509 g (_+84 biomassestimates for avianprey. We thereforecon- g) of biomassper day,whereas nestlings in broods sidered it an outlier and deleted it from biomass of two each received278 g (+3 g) of biomassper estimates. day and nestlingsin broodsof three each receiving We estimatedthe total biomassof all prey deliv- 252 g (-+36 g) per day. eries at nestsas 161 kg. The mean massfor both Dietary Overlap. The diversityand equitabilityof avianand mammalianprey was 281 g (_+13.7,95% prey deliveredto nestswas low for the studyarea, confidenceinterval = 254-308 g). Although aver- as indicated by a reciprocal of the Simpson diver- age massof avian prey (i -- 292 g; range = 10- sity index (l/D) of 4.28 and a Smith and Wilson 1082 g) was similar to that for mammalian prey evennessindex (Evar)of 0.30. Similarly, diversity (275 g; range = 18-1361 g), avian prey accounted among nestswas low, with a mean value of 1/D = for only 39% of biomassdelivered whereasmam- 3.77 (+0.41, range = 2.09-7.35). The mean value mals accounted for 61% of biomass delivered. of Evar for all nestswas 0.56 (_+0.04,range = 0.36- Snowshoehare (25%), red squirrel (24%), Ruffed 0.80). Low prey diversityand evennessvalues may Grouse (11%), (9%), diving ducks be attributable to goshawkdiet being dominated (7%), chipmunk (5%), and eastern cottontail by red squirrelsand chipmunksin our study.Sim- (5%) accountedfor 86% of biomassused by gos- ilarly, there washigh dietary overlap (>0.8) among hawks.No other speciesaccounted for >5% of bio- breeding pairsof goshawksin our study(Table 2), mass. although one nesting area (LSP; Table 2) ap- Delivery Rates. Breeding goshawksdelivered peared to be measurablydifferent from the rest. 2.12 (_+0.14)prey per day (i.e., 0.14 deliveries/hr), Clusteranalysis indicated there were two groupsof each deliveryhad a mean massof 275 g (+20 g), breeding goshawkdiets that exhibited similar prey for a total of 551 g (_+50g) deliveredper day.How- composition and proportion of use (Fig. 2) al- ever, daily (F13,953= 3.44, P < 0.001) and hourly though, again, one nest (LSP; Fig. 2) appears to (Fl•,9•0 = 2.31, P = 0.01) delivery rates varied be an outlier. There wasno apparent relationship among nests. between overlap measuresand spatial distribution 1.3 ( -+0.1 ) prey itemswere deliveredper nestling of nestsacross the studyarea (Fig. 1, 2). per day, but delivery rates increasedwith brood DISCUSSION size (F2,271= 5.23, P = 0.01). Daily prey delivery rateswere 1.8 (+0.1) at nestswith one nestling, 2.3 Mammals were the dominant prey of breeding (+0.1) at nestswith two nestlings,and 2.5 (+0.2) goshawksin Minnesota,with red squirrelsand east- at nestswith three nestlings.Despite the increase ern appearing to be the most impor- in prey deliveriesamong nestswith larger broods, tant speciesin terms of both number delivered and there was an inverse relationshipbetween brood biomass. These two speciesalone accounted for sizeand the number of prey deliveredper nestling 62% of all prey identified to at least family and per day (r = -0.43, P < 0.05). Each nestling in 51% of prey identified to at leastclass. Several stud- singlebroods received a mean of 1.8 (+0.1) prey ies have documented red squirrels as important items per day,whereas each nestlingin broods of prey for goshawks(Squires and Kennedy 2005) two received only 1.2 (_+0.1) prey items per day, throughout their range. They may be especiallyim- and each nestlingin broodsof three receivedonly portant during the winter when other prey may be 0.9 (+0.1) prey items per day. lessavailable (Widfin et al. 1987). Squirrelsdomi- 322 g (_+32 g) of biomasswere delivered per nated goshawkdiets in Swedenin terms of number nestling. However, we observeda pattern of bio- (79%) and biomass(56%) during wintersof both mass delivered to broods of different sizes that was high and low squirrel abundance (Widfin et al. similar to that of number of prey delivered to 1987). Diet information for winter goshawksin the broods of different sizes;biomass delivered per WGLR is not available, but the extensive use of red nestlingper day (F9,6--- 5.96, P = 0.038) variedwith squirrels during the summer and the patterns of 270 SMITHERSET AL. VOL. 39, NO. 3

squirrel use during winter in other areas (Widen et al. 1987) suggestthis speciesmay be of year- round importance to goshawksin the region. In terms of biomass,snowshoe hares also appear to be important for goshawksin our study area, ac- countingfor 25% of the biomassdelivered to nests. Rabbitsand hares are alsoused extensivelyby gos- hawks throughout their range (Squires and Ken- nedy 2005). Ruffed Grousecomprised 5% of prey deliveries and 11% of biomass delivered to goshawknests during a 3-yr period of relatively low grouse abun- dance (Smithers 2003). There is anecdotal evi- dence that at least some goshawksin Minnesota may rely more heavilyon Ruffed Grousethan oth- er prey during some time periods (Eng and Gul- lion 1962, Apfelbaum and Haney 1984). Eng and Gullion (1962) focused on Ruffed Grouse mortal- ity and did not assessproportional use of grouse in the diet of goshawks,and Apfelbaum and Haney (1984) reported on prey remains collectedat a sin- gle nest in northern Minnesota.Because of the dif- ficulties in accurately quantifying the extent of grouse by goshawks(Eng and Gullion 1962) and the biasesassociated with determining raptor diets based on prey remains (Smithers 2003), the results of these studies need to be in- terpreted cautiously.We suspectthat the previous research on goshawkdiet for our study area, all collected by indirect methods (Eng and Gullion 1962, Apfelbaum and Haney 1984, Martell and Dick 1996), may overestimate the proportion of birds, especiallylarge birds suchas grouse, and un- derestimate the proportion of mammals in gos- diets. Qualitative review of the data suggeststhe mean delivery rate of 0.14 deliveries/hr to nestsin our studywas less than that observedin (0.25 deliveries/hr; Boal and Mannan 1994), (0.31 deliveries/hr; Younk and Bechard 1994) and two areas of southeast (0.30 and 0.23 deliv- eries/hr; Lewis 2001). However, although mean biomassper deliveryin our study(275 g) wasless than that in Arizona (307 g/delivery) where Le- porids and Sciuridswere the dominant prey (Boal and Mannan 1994), it was greater than the two ar- eas of Alaska (214 g and 173 g/delivery), where birds were the dominant prey (Lewis 2001). Our study indicates that goshawkswith larger broods provision with greater delivery rates and biomass.Biomass per nestlingwas similar between broodsof two and three (16.3-18.0 g/hr), but only SEPTEMBER2005 BIOLOGY 271

0.6

0.5

0:4

0.3

0,2

0.1 (2) (3) ,

0,0 L$? •N I-IAC SI•. • DIX •VAG $• •l•T MCD •A LSA' F•gure2. Clusteranalysis dendrogram for food habitsdata collected at Northern Goshawknests in Minnesotaduring the 2000, 2001, and 2002 breeding seasons.Parentheses indicate clusternumber (see Fig. 1). The LSP and WAG breedingareas exhibited the leastsimilarity of diet compositionamong breeding areas.

about half as much as that receivedby nestlingsin Based on observedfrequencies of prey use, this broods of one (33.0 g/hr). This posesan interest- would translateto the averagebreeding goshawk ing question regarding energetic aspectsof gos- pair capturing 29 red squirrels,14 eastern chip- hawk productivity;what is the minimum biomass/ munks, six American Crows, five snowshoe hares, hr necessaryto fledge young successfully?The five Ruffed Grouse, two diving ducks, one cotton- similarity between broods of two and broods of , one BlueJay, and 31 miscellaneoussmall birds three suggeststhat, at leastin our studyarea, and and mammals.To put this level of predation in at nestswith similar prey composition,a minimum context, all of these prey captureswould occur of 16-18 g of biomassper hr may be required for within a home range of 6376 ha for a goshawkpair successfulnesting. However, a finer assessmentof in the studyarea (Boal et al. 2003). nestling energeticswould likely require experi- Compositionand richnessof prey deliveredto mentation in a laboratory setting. nests was similar acrossthe study area, and esti- Given our prey use and delivery rate data, one mates of prey diversityand equitabilitywere gen- can make a generalizedprediction of the relative erallylow amongnests. We suspectthe high dietary impactof a breedingpair of goshawksin our study overlapand similarityof prey use amongbreeding area during the 45-d nestlingperiod. With an ex- areaswas most likely attributableto the dominance pecteddelivery rate of 2.1 prey/d overa 45-d nest- of red squirrelsand chipmunksin goshawkdiets. ling period, ca. 94 prey deliveriescan be expected. However, goshawkdiets were dominated by red 272 SM•WHERSEW AL. VOL. 39, NO. 3 squirrels and chipmunks, but , idency statusof Northern Goshawksbreeding in Min- Ruffed Grouse, and American Crow were also im- nesota. Condor 105:811-816. portant in terms of biomass. --AND R.W. MANNAN. 1994. Northern Goshawk diets As pointed out by Reynoldset al. (1992), raptor in ponderosapine forestson the KaibabPlateau. Stud. Avian Biol. 16:97-102. populations are often limited by prey availability BOSAKOWSKI,T. AND D.G. SMITH.1992. Comparativediets and their choice of foraging habitat is predicated of sympatricnesting raptors in the easterndeciduous on conditions in which prey are abundant and forestbiome. Can.J. Zool.70:984-992. available.Thus, an understandingof goshawkprey BRIGHT-SMITH,DJ. ANDR.W. MANNAN.1994. Habitat use speciesused and the relative importance of those by breeding male Northern Goshawksin Northern prey speciesis an important step toward develop- Arizona. Stud. Avian Biol. 16:58-65. ing managementplans for goshawks.By identifying BURT,W.H. ANDR.P. GROSSENHEIDER.1980. A field guide key prey species,as we have done here, forest man- to the mammals.3rd Ed. Houghton Mifflin Company, agerscan developa set of desirableconditions that Boston, MA U.S.A. fosterspresence of thosespecies while incorporat- COLLO?Y,M.W. 1983. A comparisonof direct observation and collectionsof prey remains in determining the ing structural aspectsof known goshawkforaging diet of Golden .J. Wildl.Manag. 47:360-368. habitat (e.g., Boal et al. 2001). Those desirablefor- DANIEL,G. ANDSULLIVAN, J. 1981. A Sierra Club natural- est conditions can be incorporated into goshawk ist'sguide to the north woodsof , , managementplans as one factor of foraging habi- Minnesota and southern . Sierra Club Books, tat (e.g., Reynoldset al. 1992) and facilitate con- San Franciso, CA U.S.A. servationof the species. DUNN,J.L. 1999. Field guide to the birds of North Amer- ica, 3rd Ed. National Geographic Society,Mary B. ACKNOWLEDGMENTS Dickinson,ED., ,DC U.S.A. A. Bellman, E Nicoletti,J. Ridelbauer,A. Roberson,M. • ANDK. G•mu•Tr. 1997. A field guide to warblers Solensky,L. Smithers,W. Steffans,C. Trembath, and A. of North America. Houghton Mifflin Company,Bos- Wester assistedwith various aspectsof this study.Person- ton, MA U.S.A. nel from the many cooperating agenciesand organiza- DUNNINg,J.B., J}t. 1993. CRC handbook of avian body tions provided assistanceand logisticalsupport during masses. CRC Press, Inc., Boca Raton, FL U.S.A. this project. They included R. Baker,J. Casson,J. Gal- ENG, R.L. AND G.W. GULLION. 1962. The predation of lagher, L. Grover, M. Hamady,J. Hines, M. Houser, E. goshawksupon Ruffed Grouse on the Cloquet Forest Lindquist, C. Mortensen, S. Mortensen, B. Ohlander, W. Research Center, Minnesota. WilsonBull. 74:227-241. Russ,R. Vora, and A. Williamson. Video equipment for GP,zx•owsm, J.A. ANDS.W. E•TON. 1976. Prey items of the 2000 and 2001 field seasonswas provided by Alaska Department of Fish and Game and the WisconsinDe- goshawksin southwesternNew York. WilsonBull. 88: partment of Natural Resources.Funding and logistical 669-670. support for this project was provided by the Minnesota JONES,J.K., J}t.ANt> E.C. BINNEY.1988. Handbook of mam- Department of Natural Resources,Minnesota Forest In- mals of the north-central states. Univ. Minnesota dustties, The National Council for Air and Steam Im- Press,Minneapolis, MN U.S.A. provement,Potlatch Corporation,the U.S. ForestService KENNEDY,P.L. ANDD.R. JOHNSON.1986. Prey size selec- Chippewa and Superior National ,and the U.S. tion in nestlingmale and female Cooper'sHawks. Wzl- Fish and Wildlife Service. son Bull. 98:110-115.

LITERATURE CITED KREBS,CJ. 1999. Ecological methodology, 2nd Ed. Ad- dison-WesleyEducational Publishers,Inc., Reading, APFELBAUM,S.I. AND A. HANEY.1984. Note on foraging MA U.S.A. and nesting habitatsof goshawks.Loon 56:132-133. L•w•o}t, L.R. 1980. Overlap, similarity and competition BEIE}t,P. ANDJ.E. DPd•NNAN.1997. Forest structure and coefficients.Ecolo• 61:245-251. prey abundance in foraging areas of Northern Gos- LEW•S,S.B. 2001. Breeding seasondiet of Northern Gos- hawks.Ecol. Appl. 7:564-571. hawksin southeastAlaska with a comparisonof tech- BIELEFELDT,J., R.N. ROSENFIELD,ANDJ.M. PAP?.1992. Un- niques used to examine raptor diet. M.S. thesis,Boise founded assumptionsabout diet of the Cooper's State Univ., Boise, ID U.S.A. Hawk. Condor 94:427-436. McG2•aUG•tL,K., S. CUSHMAN,AND S. STAFFORD.2000. Mul- BO•tL, C.W., D.E. ANDERSEN, AND P.L. KENNEDY. 2001. tivariate statisticsfor wildlife and ecology research. Home range and habitat use of Northern Goshawks Springer,, NY U.S.A. in Minnesota. Final Report. Minnesota Cooperative 1VIACARTHUR,R.H. 1972. Geographicalecology. Harper Fish and Wildlife Research Unit, University of Min- and Row, New York, NY U.S.A. nesota, St. Paul, MN U.S.A. M2mTI,C.D. 1987. Raptor food habitsstudies. Pages 67- --, --, AND--. 2003. Home range and res- 80 in B.A. Giron Pendleton, B.A. Millsap, K.W. , SE?WEMBER2005 BIOLOGY 273

and D.M. Bird [EDS.], Raptor management tech- SIMPSON,E.H. 1949. Measurement of diversity. Nature niques manual. National Wildlife Federation, Wash- 163:688. ington, DC U.S.A. SMITH,B. ANDJ.B. WILSON.1996. A consumer'sguide to MARTELL,M. ANDT. DICK. 1996. Nesting habitat charac- evenness indices. Oikos 76:70-82. teristicsof the Northern Goshawk(Accipitergentilis) in SMITHERS, B.L. 2003. Northern Goshawk food habits in Minnesota. Final Report to Minnesota Department of Minnesota:an analysisusing time-lapserecording sys- Natural Resources,Nongame Wildlife Program, Proj- tems.M.S. thesis,Texas Tech University,Lubbock, TX ect No. 9407382., Minnesota Department of Natural U.S.A. Resources, St. Paul, MN U.S.A. SQuIm•S,J.R. AND R.T. REYNOLDS.1997. Northern Gos- MENG,H. 1959. Food habits of nestingCooper's Hawks hawk (Accipitergentilis). In A. Poole and F. Gill lEDS.I, and goshawksin New York and .Wilson The birds of North America, No. 298. The Birds of Bull. 71:169-174. North America Inc., Philadelphia, PA U.S.A. REYNOLDS,R.T. AND E.C. MESLOW.1984. Partitioning of --AND P.L. KENNEDY. 2005. Northern Goshawk ecol- food and niche characteristicsof coexistingAccipiter ogy: an assessmentof current knowledge and infor- during breeding. 101:761-779. mation needs for conservation and management. --, R.T. Gg•a-I•M, M.H. REISER, R.L. B•tSSETr, P.L. Stud.Avian Biol. 31:in press. KENNEDY,D.A. BOVCE,JR., G. GOODWIN,R. SMITH,AND STATSOFT, INC. 2001. STATISTICA: the small book. E.L. .1992. Managementrecommendations for StatSoft, Inc., Tulsa, OK U.S.A. the Northern Goshawk in the southwestern United STORER,R.W. 1966. and food habits States.USDA ForestService, Gen. Tech. Rep. RMo217, in three North American .Auk 83:423-246. Ft. Collins, CO U.S.A. TESTER,J.R. 1995. Minnesota'snatural heritage: an eco- ROBERSON,A.M. 2001. Evaluating and developing survey logical perspective. Univ. Minnesota, Minneapolis, techniques using broadcast conspecific calls for MN U.S.A. Northern Goshawks in Minnesota. M.S. thesis, Univ. WIDEN, P., H. ANDREN, P. ANGELSTAM,AND E. LINDSTROM. Minnesota, St. Paul, MN U.S.A. 1987. The effect of prey vulnerability:goshawk pre- , D.E. ANDERSEN, AND P.L. KENNEDY. 2003. The dation and population fluctuationsof small game. O/- Northern Goshawk(Acdpiter gentills atricapillus) in the kos 49:233-235. western Great Lakes region: a technicalconservation WILLIAMS, C.B. 1964. Patterns in the balance of nature. assessment.Minnesota CooperativeFish and Wildlife Academic Press, London, U.K. Research Unit, Univ. Minnesota, St. Paul, MN U.S.A. YOUNK,J.V. ANDMJ. BECI-I•mD.1994. Breeding ecologyof ROMESBURG,H.C. 1984. Cluster analysisfor researchers. the Northern Goshawkin high-elevationaspen forest Lifetime Learning Publications,Belmont, CA U.S.A. of northern Nevada. Stud. Avian Biol. 16:119-121. SmLEY,D.A. 2000. The Sibley guide to birds. National Audubon Society,Alfred A. Knopf, Inc., New York, NY Z•m, J.H. 1999. Biostatisticalanalysis, 4th Ed. Prentice- U.S.A. Hall, Inc., EnglewoodCliffs, NJ U.S.A. SIDERS, M.S. AND P.L. KENNEDY. 1996. Forest structural characteristicsof accipiternesting habitat: is there an Received22 March 2004; accepted28 September 2004 allometric relationship?Condor 98:123-132. Guest Editor: Stephen DeStefano