Unfounded Assumptions About Diet of the Cooper's Hawk

The Condor94~427-436 0 The Cooper Ornithological Society 1992

UNFOUNDED ASSUMPTIONS ABOUT DIET OF THE COOPER’S HAWK’

JOHN BIELEFELDT Park Planning, Racine County Public Works Division, Sturtevant, WI 53177

ROBERT N. ROSENFIELD College of Natural Resources,University of Wisconsin-StevensPoint, StevensPoint, WI 54481

JOSEPH M. PAPP 17960 W. BeresRoad, New Berlin, WI 53146

Abstract. Prior analysesand reviews of diet in the Cooper’s Hawk (Accipitercooperii) have suggestedthat avian prey usually provides a majority of items and biomass, but this interpretation may be compromised by methodological problems. In Wisconsin, we found that indirect collection of prey remains near nests(92% birds) overestimated the proportion of avian items in comparison to direct observation of prey deliveries to nestlings(5 l-68% birds). Previous studiesusing indirect methods may have overstated the frequency of birds in the diet. Also, most avian items brought to nestlingsin Wisconsin, as elsewhere, were young birds. Some prior studies relying on indirect methods and using prey species’ adult mass to calculateavian biomass have probably accentuatedthe methodologicalbias toward birds among prey remains. In Wisconsin, birds provided a minority of prey biomass (40- 46%) in two of our three sampling pools. Proportions of avian items and biomassare highly variable in the small set of other direct dietary studiesof the Cooper’s Hawk, all from the breeding season. Becauseof methodological, seasonal,geographic, and other limitations, existingdata do not warrant an assumptionthat birds prevail in the diet. Becausevulnerable ground foraging animals (as well as youngbirds) appear to constitutemost of the breeding seasondiet, we alsoquestion the assumption-basedmainly on breedingseason studies- that Cooper’sHawks feed principally on preyof high“ agility,” asproposed in sometheories of reversedsexual size dimorphism. Key words: Coopers’ Hawk; Accipitercooperii; nestling diet; dietary methods.

INTRODUCTION Methodological differences, however, may Many investigators have enumerated prey spe- cloud the comparability or reliability of previous cies or prey items taken by Cooper’s Hawks (Ac- dietary studies. Stomach analyses (Fisher 1893 cipiter cooperii). Although reptiles or mammals [see also Sherrod 19781, Duncan 1966, Storer have been numerically preponderant among prey 1966) have employed data from hawks of all ages in some individual hawks and some areas (e.g., collectedover wide geographicareas at every time Fitch et al. 1946, Reynolds and Meslow 1984), of year. Seasonal, regional, or age-related biases reviews have typically reported (Sherrod 1978, might exist in such data. Specimensfrom poultry Jones 1979, Rosenfield 1988) that birds are the or gamebird farms may also give an unrepresen- most frequent prey in proportions ranging from tative sample of hawks’ ages and stomach con- 55% (Duncan 1966, as recalculated by Jones tents (Meng 1959). Other dietary work on Coo- 1979) to 9 1% (Craighead and Craighead 1956). per’s Hawks has been done almost exclusively Only four studies,all in the breeding season,have in the breeding season.Methods include analyses estimated prey biomass. Avian prey reportedly of the crop contents of nestlings (e.g., Errington accounted for the majority of biomass in three 1933), direct observations of food deliveries to instances (Kennedy 1980, Millsap 198 1, Toland nestlings(e.g., Snyder and Wiley 1976), collation 1985) but a minority in the remaining case of prey remains in pellets, nests,or plucking areas (Reynolds and Meslow 1984). near nests (e.g., Meng 1959), or combinations of these techniques (e.g., Craighead and Craighead 1956). The potential biases of these methods in ’ ’ Received6 September199 1. Accepted15 January analyzing falconiform diets are well recognized 1992. (Marti 1987) but remain unevaluated in Coo-

Wl’ 428 J. BIELEFELDT, R. N. ROSENFIELD AND J. M. PAPP per’s Hawks. Here we contrast direct observa- Johnson (1986) assumed that prey items deliv- tions and collection of prey remains as methods ered to nests by female Cooper’s Hawks were of assessingthe prey of breeding Cooper’s Hawks also captured by females if the vocalizations in Wisconsin. We use the former method to com- characteristically accompanying male-female pare biomass of avian and non-avian prey, and prey transfers did not precede the delivery. We we also estimate biomass for prey items tabu- do not separate deliveries into captures by sex lated by Hamerstrom and Hamerstrom (195 1) under this criterion becauseboth sexescache and in analyses of the crop contents of young Coo- retrieve uneaten or untransferred prey during the per’s Hawks in Michigan. breeding season(pers. observ.). Although we have We suggestthat dietary studies based on prey not seen one sex retrieve items cached by the remains may be biased toward avian items. We other, we assume that females retrieve prey also show that most prey items delivered to nest- cached by males at regularly used transfer sites. ling Cooper’s Hawks are sub-adult individuals It nevertheless seems likely that most delivered and ground-foraging species.Review of previous items were captured by males, as indicated by dietary studies indicates that generalizations other nesting seasonstudies (e.g., Reynolds and about a predominant classofprey or the “agility” Meslow 1984), becausefemales rarely leave the of prey are not warranted by available data. nest site until nestlings reach 2 14 days of age (pers. observ.). After excluding five items (1.5%) METHODS that could not be identified as either avian or All original data were collected in Wisconsin, non-avian prey, we estimated biomass for nest almost entirely in central or southeasterncoun- deliveries as follows. Biomass here refers to the ties within 85 km of the sites where nest deliv- live mass of an item, not the mass delivered to eries were observed. We did not attempt to eval- the nest or consumed by nestlings. uate prey availability. Nearly all mammals (95%) were identifiable to genusor species;most were easternchipmunks NEST DELIVERIES (Tamias striatus). We calculated mass of “adult” We recorded prey deliveries to nestlings aged 5- (SC3) chipmunks from an independent sample 28 days from blinds erected within 10 m of four (n = 26, 88-126 g, K = 107 g) collected in late nests in two study areas: two nests in the semi- May to mid-July in both study areas; we also urban Amherst area in Portage County in June- weighed sub-sets of headless and eviscerated July 1986 and 1987, and two nests about 185 specimens.Juvenal chipmunks may be two-thirds km farther south in the Kettle Moraine State grown in Wisconsin in the latter half of June Forest in Waukesha County in June 1986. The (Jackson 196 1); the median date of our blind same marked male but different females bred at observations was 27 June, and we often had the Amherst area in both years. problems in gaugingthe age and size of delivered During 44 all-day and eight part-day watches chipmunks at the margin between SC2 and 3. (734 hours), we noted species,age, and condition Among 25 reliably aged deliveries, seven (28%) (intact, headless,eviscerated) of prey when pos- were juveniles in SC2. We extrapolated this pro- sible. We initially categorized items into three a portion to all deliveries, and calculated a mean priori size classes(SC) following Kennedy and mass of 74 g for SC2 chipmunks from condition Johnson (1986) and Storer (1966) using known and mass of five deliveries, assuming that head mass and bulk of familiar species as reference and visceral masseswere proportionally similar points: SC1 I 27 g, SC2 = 28-91 g, and SC3 = to those of adults. A mean mass of 98 g for all 92-2 16 g. Becausewe encountered only one prey chipmunks was derived from these estimates. item > 132 g among nest deliveries, we here trun- Local specimensof similar age were also used to cate SC3 at 92-l 32 g. We also recorded 40 prey estimate mass of thirteen-lined ground squir- deliveries and five items retrieved from adult or rels (Spermophilus tridecemlineatu.s)-all juve- nestling crops during late incubation (10 items) niles-at 55 g. We assumed mice (Peromyscus and nestling (35) stages,respectively, at 35 ad- spp.) and voles (Micro&s spp.) to be adults, and ditional Wisconsin nests, 198 l-l 989. We com- a single gray squirrel (Sciurus carolinensis)to be bine these observations as a fifth sample of de- a three-fourthsgrown juvenile; in thesecases mass liveries to “other” nests. was estimated from Craighead and Craighead Snyder and Wiley (1976) and Kennedy and (1956:429) and Jackson(196 1). Six mammals of DIET ASSUMPTIONS ABOUT COOPERS’ HAWKS 429 unknown identity were assigneda mean mass of items include remains from the same reoccupied 60 g on the basis of relative size and preceding nesting areasover as many as four years (in some estimates. casesthe same marked adult[s] over 2-3 years), All of 202 avian items delivered to nests were we treat the data as independent becauseno one assignable to size class but only one-third were nesting area provided >9 items across 10 years identifiable to genus or species.Adult mass of of collecting remains, and becausenests yielding identifiable bird specieswas taken from region- remains involved 2 96 different, individually ally and seasonallyappropriate samplesin Clench marked adult hawks. Because of small annual and Lieberman (1978) and Dunning (1984). sample sizes at each stageof breeding, we com- When possible,age of avian prey was determined bined years. Becauseproportions of avian prey by feather sheathingin nestlingsand recent fledg- were very similar (91-93%) we also combined lings, or distinctive juvenal plumages in inde- incubation (35 items), nestling (59) and fledgling pendent young. From the literature on growth (38) stages. rates, we estimated mass of identifiable birds by Although we often searchedfor regularly-used size and age classesas follows. Most speciesin plucking sites near the nest at the incubation SC 1 fledgeat 190% of adult mass, so we ignored stage,we did not revisit these sites on a system- age in these cases.For ground foraging birds, the atic basis to collect prey remains. Most remains usual category of avian prey of Cooper’s Hawks were thus found incidentally during other work. in our samples (see Results), we calculated mass of altricial, open-nesting speciesat nestling/fledg- PREY FORAGING HEIGHTS ling, independent juvenal, and unknown agesas In all data sets, we used personal observations 70, 90, and 95% of adult mass, respectively, for to partition identifiable prey items (excluding smaller species (28-59 g) in SC2; 65, 85, and nestlings)into two categories:taxa that primarily 90% for larger species(60-9 1 g) in SC2; and 55, (e.g., American Robin, Turdus migratorius) or 80, and 85% for species in SC3 as truncated. frequently (e.g., Blue Jay, Cyanocitta cristata) Woodpeckers and most hole-nesting passerines forage on the ground, and those that do not (e.g., also fledge at near-adult size, so we estimated Northern Oriole, Zcterusgalbula). both juveniles and birds of unknown age at 95% of adult mass in such species.Mean mass of 32 RESULTS avian prey deliveries in SC2, weighed or iden- We did not detect prey items other than birds tified and aged in hand, was 65 g. We use this and mammals. figure in estimating biomass of unidentified birds in SC2. Unidentified birds in SC 1 and SC3 were NEST DELIVERIES assumed to average 25 and 115 g, respectively. Avian prey formed half or more of delivered We applied the same proceduresin estimating items at all four Wisconsin nestsbut proportions biomass for prey speciesand ages listed by Ha- of avian items were greater at the two Amherst merstrom and Hamerstrom (195 l), except that nests than the two Kettle Moraine nests (Table unidentified birds (10.3% of total items) were 1). Differences among nests are significant (x2 = excluded, tentatively identified birds (6.5%) were 9.37, df = 3, P < 0.05) and remain significant if assigned a mean mass (as above) by size class, the Amherst nests, where the same male bred in unidentified Sciuruswere assumedto be juvenal both years, are considered non-independent (x2 gray squirrels, and “immature” Ring-necked = 7.08, df = 2, P < 0.05). Inter-year differences Pheasants (Phasianus colchicus)were assumed at Amherst were not significant (P > 0.10). We to be 4-5 week old poults with a mass of 150 g. thus pool Amherst nests (68% avian items, n = 194) and Kettle Moraine nests(5 1% avian items, PREY REMAINS n = 90), respectively, in subsequent analyses. We tabulated remains of 132 prey items found The composite sample from 35 “other” Wiscon- in or < 100 m from 69 active nests in 39 geo- sin nests (56% avian items, n = 45) is treated as graphically separate nesting areas, 1980-l 989, a separatepool. during incubation through fledgling stages of Avian biomass accountedfor 58% of total prey breeding. These items were mainly pluckings biomass among nest deliveries at Amherst, 40% (> 80%), occasionally skeletal remains or uneat- in the Kettle Moraine, and 46% at “other” Wis- en prey; we did not collect pellets. Although these consin nests (Table 2). Eastern chipmunks were 430 J. BIELEFELDT, R. N. ROSENFIELD AND J. M. PAPP

TABLE 1. Numbersand frequency(%) of mammalianand avian preyitems among Wisconsin nest deliveries.

Kettle Moraine Amherst I 2 I 2 Other nests Mammalian E. chipmunk 19 24 23 28 18 Other 0 1 7 5 2 Total 19 (50) 25 (48) 30 (28) 33 (38) 20 (44) Avian SC1 1 6 17 11 SC2 18 19 56 37 2: SC3 0 2 5 5 1 Total 19 (50) ?i (52) 78 (72) 53 (62) 25 (56) Total 38 52 108 86 45 strongly predominant among mammalian items fledglings at four nests in Michigan, 194 l-l 946. in all Wisconsin nest delivery samples (Table l), Proportions of avian items were not significantly accounting for 36-59% of total prey biomass (Ta- different among nests (P > 0.25). Our further ble 2). analyses of their data exclude 27 unidentified birds of unknown size. PREY REMAINS We used their lists of prey speciesand prey Avian items provided 92% of total items among ages to calculate that avian prey accounted for Wisconsin prey remains at incubation through 7 1% of total prey biomass (Table 2). Chipmunks fledgling stagesof breeding (Table 3). This pro- constituted 59% of mammalian items (n = 41) portion-virtually identical (93%, n = 59) if the but only 11% of total prey biomass in Michigan nestling stageis consideredalone- is much larger (Table 2). These proportions, while relatively than any obtained in our direct observations of large, are all substantially less than comparable prey deliveries to nestlings. As in nest deliveries, figures for chipmunks in Wisconsin nest deliv- chipmunks were the leading item among mam- eries. malian prey remains. PREY AGES CROP CONTENT ANALYSES Data on the ages of mammalian prey are scant Hamerstrom and Hamerstrom (195 1) reported becauseof problems in ageingchipmunks in both that birds provided 221 (84%) of 262 prey items Wisconsin (seeMethods) and Michigan (Hamer- retrieved from crops of nestlings and tethered Strom and Hamerstrom 195 1) but it appearsthat

TABLE 2. Biomass(g) and relativebiomass (%) of mammalianand avian preyamong Wisconsin nest deliveries and Michigancrop content analyses.

Wisconsin Kettle Moraine Amherst Other nests Michigan Mammalian E. chipmunk 4,214 4,998 1,764 2,352 Other 60 846 64 3,624 Total 4,274 (60) 5,844 (42) 1,828(54) 5,976 (29) Avian SC1 175 739 0 1,050 SC2 2,413 6,031 1,419 6,407 SC3 230 1,157 125 3,055 >sc3 0 0 0 4,235 Total 2,8 18 (40) 7,927 (58) 1,544(46) 14,747(71) Total 7,092 13,771 3,372 20,723 DIET ASSUMPTIONS ABOUT COOPERS HAWKS 431

TABLE 3. Numbersand frequency(O/o) of mamma- prey remains, and crop retrievals. The exception lian and avian items amongWisconsin prey remains was a flying squirrel (Glaucomysvolans) in Mich- (n = 132). igan. The proportion of identifiable avian prey items Mammalian Avian E. chip- that primarily or frequently forage on the ground munk Other Total SCI SC2 SC3 >sc3 Total was uniformly high among nest deliveries (94%, 9 1 lO(8) 1 82 31 8 122 (92) n = 53) and prey remains (93%, n = 107) in Wisconsin as well as crop retrievals (82%, n = 169) in Michigan. up to 43% of mid-sized mammals (chipmunks DISCUSSION and ground squirrels, IZ = 37) and 100% of larger PREY FREQUENCY AND BIOMASS mammals (sciurids and lepids, IZ = 7) were sub- adults among ageablemammals in the combined Errington (1932) proposedthat indirect methods nest delivery and crop content data of these stud- are less accurate than direct techniques in eval- ies. uating the diets of nestling falconiforms, and oth- Among avian prey in Michigan, 108 (74%) of er researchershave concurred (Snyder and Wiley 145 ageable items were young of the year, and 1976, Marti 1987, Rosenbergand Cooper 1990). 11 (10%) of these 108 young birds were nestlings. Our results for Cooper’s Hawks also agree: an Among avian items both ageableand identifiable indirect sample of prey remains (92% avian items) (n = 136), young birds (73% of items) accounted from Wisconsin nest sites apparently overesti- for 62% of avian biomass. mated the frequency of birds in nestling diets in A bias against adults might exist among uge- comparison to direct observations of nest deliv- able birds, however, for speciesin which feather eries (51-68% avian items). This result is un- sheathing (rather than distinctive juvenal plum- surprising. It presumably occurs because avian age) is the only available criterion of age; many pluckings are larger, often more colorful, and adults might be consigned to an “age unknown” hence more conspicuous than mammalian or category in such cases. More conservatively, other non-avian remains. The colorful Blue Jay, young of the year accounted for ~56% of total for example, accountedfor 39 items (36%) among avian items (n = 194) and 253% of avian bio- identifiable avian remains (n = 108) but only mass among identifiable items (n = 177) when nine items (15%) among identifiable birds (n = birds of unknown age are included in the Mich- 60) delivered to nests. igan data. A similarly conservative estimate for Certain biases may exist in our own samples, Wisconsin comes from 42 reasonably intact avi- but we contend that none of them is likely to an items examined in hand as nest deliveries or obviate our methodological results. Most prey found in nestsas prey remains; at least 24 (I 57%) remains were collected incidentally during other of these were young of the year and approxi- work. More intensive searcheswould perhaps mately half of the young birds were nestlings or have discovered larger numbers of inconspicu- recent fledglings. Identifiable sub-adult birds av- ous mammalian items, yet it is also possible that eraged 86% of adult mass in Michigan (n = 74) persistentsearching would have found more small and 82% in Wisconsin (n = 25) for species in or drably-colored birds. Also, our nest delivery which adult mass falls within SC2 and 3 (28-132 data might be skewed toward certain types of g). All speciesin this calculation were altricial; prey if some items (e.g., small birds) were selec- median adult mass of these items was 79 g in tively consumed by adult hawks rather than de- both states. livered to nestlings. A similar bias would arise among prey remains to the extent that such items PREY FORAGING HEIGHTS are consumed where captured, which is nearly Most small to mid-sized mammals in Wisconsin always beyond the immediate vicinity of the nest and Michigan are at least partly terrestrial in (Rosenfield et al. 199 1, pers. observ.). Moreover, foraging habits. It is not surprising that mammals a very largenumber of selectivelyconsumed birds foraging primarily or frequently on the ground (12 per day per adult at the nests watched from accounted for all but one of the 17 1 identifiable blinds) would be required to eliminate the pro- mammalian prey items (chiefly chipmunks) in portional disparity of avian items in nest deliv- the combined two-state tally for nest deliveries, eries vs. prey remains. 432 J. BIELEFELDT, R. N. ROSENFIELD AND J. M. PAPP

Finally, our samplesof nest deliveries and prey mains, in part, to estimate prey biomass (Millsap remains were drawn from different sets of nests 198 1, Toland 1985). Both Millsap and Toland and in part from different years. Results might also assumed that all identifiable birds (88% of stem from annual or local contrastsin prey avail- total avian items in both instances) were adults, ability, or differing predilections for prey types and calculated avian biomass on that basis while among individual hawks, instead of a bias to- assigning sub-adult mass to most mammals. As ward avian items among prey remains. Both discussedlater, our resultsand other studiesshow samples, however, come from the same parts of that a large fraction of birds taken by nesting Wisconsin. We collected small numbers of prey Cooper’s Hawks are sub-adults. Inappropriate remains at each of many nests over 10 years, assignment of adult mass to birds when calcu- while nest deliveries, aside from blind observa- lating biomass will accentuate the bias toward tions, include a sub-samplefrom 35 “other” nests avian items among prey remains. over nine years (see Methods). It remains pos- It thus seems probable that Toland’s (1985) sible that mammals (i.e., chipmunks) were un- report of 6 5% avian biomass in the prey of breed- usually prominent in the regional prey base during Cooper’s Hawks is an overestimate. Millsap ing blindwork on nest deliveries in 1986-l 987. (198 1) inadvertently mis-calculated the mass of If so, more mammals might also appear among Gambel’s Quail (Callipepla gambelii) at 287 prey remains in those years, but this sub-sample rather than 187 g per adult (B. Millsap, pers. of remains (n = 23 items at 16 nests) was exclu- comm.), and his sum for mammalian biomass sively birds. omitted one black-tailed jackrabbit (Lepus cal- We conclude that the methodological bias in- ifornicus)(Millsap 198 1: Table 14). Recalculated dicated by our results is real. Previous studies avian biomasswould thus be 50% (34,020/67,862 relying partly or mainly on prey remains (Craig- g), not 54% as given in Millsap (1981). Because head and Craighead 1956, Meng 1959, Millsap of biasestoward birds, discussedabove, even this 198 1, Reynolds and Meslow 1984, Toland 1985) revised figure seemslikely to overestimate avian may have overestimated the relative frequency biomass in his breeding seasonprey sample. of avian prey in the diet of Cooper’s Hawks. Our results for Wisconsin nest deliveries as Although careful pellet analyses (Reynolds and well as our methodological critique thus indicate Meslow 1984) might alleviate the bias of relying that the diet of breeding Cooper’s Hawks is more on prey remains, only Janik and Mosher (1982) variable with respect to class of prey than prior have presentedresults separatelyor specifiedthe studies and reviews have suggested.The degree numbers of items identified by pelletal remains to which studiesusing indirect means have over- vs. other techniques. They found avian items to stated the importance of avian prey is unknown. be more frequent in pellets (54%) than in nest Eight studies using direct observation of crop deliveries (30%, see Table 4) but the sample of contents or nest deliveries (Table 4) have shown pelletal items was very small (n = 13). that birds may indeed be strongly predominant Even though birds formed the majority of prey in diets in some cases. The same studies have items among Wisconsin nest deliveries, mam- nonetheless reported a wide range in the shares mals outranked birds in terms of prey biomass of avian, mammalian, and reptilian prey pro- in two of our three delivery pools. Precise provided to nestling hawks, both among studies(26- portions of biomass (Table 2) depend heavily on 90% avian) and within study areas(see especially our estimates of mass for chipmunks and un- Snyder et al. 1973). Four of these eight studies identified SC2 birds, but the conclusion that avi- found that birds accounted for 550% of prey an biomass was in the minority is robust to si- items (Fitch et al. 1946, Janik and Mosher 1982, multaneous overestimates of mass (for Snyder et al. 1973) or prey biomass (this study) chipmunks) and underestimates of mass (for at most observed nests. birds) of 10% at “other” nests and 20% at Kettle Potential sourcesof dietary variation include Moraine nests, respectively. individual, age-related,or sexual propensitiesfor Among previous studies, only Reynolds and certain sizes and/or classesof prey as well as Meslow (1984: Table 1) had reported a majority regional, local, annual, or seasonaldifferences in of mammalian biomass. However, a bias toward the availability or vulnerability of prey types. avian items among prey remains could obviously Although Cooper’s Hawks breed or winter be compounded in studies that have used re- throughout the conterminous U.S. and southern DIET ASSUMPTIONS ABOUT COOPERS HAWKS 433

TABLE 4. Proportionsof avian prey in studiesusing direct observations of nestdeliveries or cropcontents.

State No. nests No. items % Avian (range) Fitch et al. 1946 CA 26’ Janik and Mosher 1982 MD (at 3857 3@ Snyderet al. 1973 AZ-NM 11 473 56 (8-93) This study WI 5 329 61 (50-72) Petersonand Murphy in press ND 7@ Enington 1933 WI ; 7424 Hamerstromand Hamerstrom195 1 MI 4 262 i&8&90) Kennedy 1980 WA 6 240 90 (76-95”)

= Not available or not applicable. b Five successful nests.

Canada, direct observations of diet-all for the 1959, Nelson 1968) can be regardedas unexcep- breeding season-have come mainly from a few tional. Young of the year have also provided the states in southwestern and midwestem regions majority of avian prey items in direct observa- (Table 4). Using unstated methods presumably tions of nest deliveries among various other fal- involving prey remains, Millsap (198 1) reported coniforms: the congeneric Sharp-shinned Hawk that birds provided 5 1% of prey items in two (A. striutus) in Alberta (Quinn 1991), Broad- southwestern states in winter. Except for stom- winged Hawks (Buteoplutypterus)in several U.S. ach analyses pooling regions, ages, seasons,and states (Fitch 1974, Mosher and Matray 1974, sometimes sexes, other data for non-breeding Rosenfield et al. 1984) and the Peregrine Falcon seasonsare lacking. (F&o peregrinus) in Greenland (pers. observ., In view ofthe methodological, geographic,sea- RNR). sonal, and other limitations of existing dietary Among identifiable avian prey with an adult data for the Cooper’s Hawk, generalizationsabout mass of 28-132 g, sub-adult items averaged 82- a predominant class of prey are unwarranted. 86% of adult mass in our Wisconsin and Mich- Indeed, further research may show that varia- igan analyses. This figure is somewhat less than tions in its diet hold more ecologicalinterest than the 90% assigned to juvenile birds of all adult an attempt to describe a “typical” diet (Wiens size classesby Kennedy (1980) but substantially 1989). greater than the 50% assumed by Reynolds and Meslow (1984). These differences do not seem PREY AGES AND FORAGING HEIGHTS wholly attributable to the varying taxa, ages, or Although they gave no systematic data on prey adult sizes encountered or considered in these ages,Craighead and Craighead (1956) and Meng respectiveanalyses. Future studiesprobably need (1959) remarked that young birds formed a large to consider such variables when calculating bio- share of the avian prey provided to nestling mass for sub-adult birds, rather than relying on Cooper’s Hawks. Our results and other quanti- an arbitrary estimate. For example, proportional tative studies support their comments. In both mass of sub-adult prey items may be relatively Wisconsin nest deliveries and Michigan crop low in larger or precocial speciessuch as quail content analyses,a conservatively estimated 56- (Colinus virginianus, Callipepla spp.). Quail are 57% of avian items were young ofthe year. About frequent prey elsewhere (e.g., Kennedy 1980, half of theseyoung birds were nestlings or newly- Millsap 198 1, Toland 1985) but were lacking in fledged individuals in Wisconsin, and 10% were our Wisconsin and Michigan samples. nestlings in Michigan. Data on the age and corresponding mass of In addition, Reynolds and Meslow (1984:766, mammalian prey are less satisfactory. Juveniles Appendix) reported that “nestlings and fledg- are common prey-ca. 50% of ageable mam- lings” accounted for 33% of avian items (n = malian items in Wisconsin nest deliveries and 267) among prey remains, while P. Kennedy (pers. Michigan crop content analyses-but better comm.) has judged that 35% of ageable avian methods of determining age and mass are desir- prey items in New Mexico were nestlings or able, especiallyfor lepids and larger sciuridswhere fledglings. It seems that incidental observations undocumented assumptions about age can have of nestling birds as prey (Linduska 1943, Mew strong effects on calculations of prey biomass. 434 J. BIELEFELDT, R. N. ROSENFIELD AND J. M. PAPP

This effect was not a problem in Wisconsin nest TABLE 5. Genera (no. studies) providing ~5.0% of deliveries, which included only one larger mam- identifiable prey items in one or more of 12 dietary studies. mal. There appear to be no reports of the ages of reptilian prey. Wisconsin and Michigan results furthermore Birds suggestthat nearly all mammalian prey and large Turdus (8) Junco (1) Cyanocitta (5) Melospiza (1) majorities of avian items (2 82% excluding nest- Sturnus (4) Bonasa (1) lings) can be classed as speciesthat forage pri- Colaptes (3) Mammals marily or frequently on the ground. Among prior Melanerpes (3) Callipepla (3) Sylvilagus (3) studies, only Reynolds and Meslow (1984: Ap- Tamias (3) pendix) attempted to quantify the distribution of Colinus (2) Phasianus (2) Eutamias (2) foraging heights in prey species taken by Coo- Zenaida (2) Spermophilus (2) per’s Hawks. Their five-part classification iden- Pipilo (2) Microtus (1) tified only 23% of Oregon prey items as “ground- Passer (2) Reptiles shrub” foragers. However, of the 33 prey taxa Dolichonyx (1) Cnemidophorus (1) Agelaius (1) Sceloporus (1) they treated as “shrub-canopy” or “generalist” Quiscalus (1) foragers, we would class 24 as primary or fre- Sturnella (1) quent ground foragers; the latter would include Sources (no. genera, cumulative % of items): Errington 1933 (6, 93); such “shrub-canopy” species as Rufous-sided Fitch et al. 1946 (5, 91); Hamerstrom and Hamerstrom 1951 (8, 64); Craighead and Craghead 1956 (4! 51); Meng 1959 (5, 80); Storer 1966 Towhee (Pipilo erythrophthalmus)and Fox Spar- (4, 34); Kennedy 1980 (4, 56); Mlllsap 1981 (4, 46); Janik and Masher 1982 (3, 92); Reynolds and Me+v 1984 (6, 59); Toland 1985 (6, 55); row (Passerellailiaca) and such “generalists” as this study (4, 72). American Robin, Dark-eyed Junco (Bunco hye- malis), and chipmunks (Eutamias spp.). Under our classification, 73% of Oregon prey items ground foraging individuals (especially mam- would qualify as primary or frequent ground for- mals) whose avenues of predator detection and agers. escape are more limited than those of arboreal Also, a list of the 25 genera providing 2 5.0% speciesand individuals (Reynolds and Meslow of identifiable prey items in one or more of 12 1984). dietary studies (Table 5) includes 24 genera, ex- In Michigan crop content analyses and Wis- cepting only Melanerpes, that we would classas consin nest deliveries, the aggregatenumbers of primary or frequent ground foragers. Although mammals, sub-adult birds, and ground foraging many of these data may be subject to the meth- avian taxa suggestthat prey of relatively low agil- odological problems discussed earlier, the lack ity and relatively high vulnerability is strongly of such widespread bark, shrub, or canopy for- predominant in the diet of nestling Cooper’s aging genera as Picoides,Pheuticus, Piranga, or Hawks. Highly agile prey (i.e., adult birds) of Zcterusseems notable. In the absenceof unbiased other foraging strata provided 5 10% of identi- observations of actual prey captures, interpre- fiable items in both states. tations of prey foraging heights are obviously Some theories of reversed sexual size dimor- subjective. Nevertheless, available evidence sug- phism (RSSD) in the genus Accipiter (Reynolds geststhat many or most prey items may be taken 1972) and other raptors (Earhart and Johnson from the ground. 1970, Snyder and Wiley 1976) have noted a pos- As Meng (1959) suggests,Cooper ’s Hawks itive correlation between the degreeof a species’ probably take whatever prey is easiest to catch. RSSD and the agility of its principal classof prey; Sub-adult and ground foraging animals may be birds are assumed to be the most agile prey cat- particularly vulnerable to the hunting tactics of egory. For the Cooper’s Hawk, one of the most Cooper’s Hawks, which typically employ a series strongly dimorphic falconiforms, this correlation of brief perch-and-scan episodesto locate poten- restsupon an assumedpredominance of birds in tial prey (Fischer 1986, Kennedy 1991) and the diet (as inferred from studies conducted al- probably take most prey by surpriseattack rather most entirely in the breeding season) as well as than active aerial pursuit (Meng 195 1). Such tac- the presumed agility of avian prey. We have tics should be most effective and economical shown, however, that birds are not necessarily against nestling birds, inagile fledglings, and oth- the dominant classof prey in the breeding season er inexperienced juvenile animals as well as in terms of frequency (Table 4 and earlier dis- DIET ASSUMPTIONS ABOUT COOPER’S HAWKS 435 cussion) or biomass (this study). We have also dimorphism and food habits of North American shown that most birds taken by Cooper’s Hawks owls. Condor 72~251-264. at this seasonare inagile or inexperienced young ERRINGTON,P. L. 1932. Technique of raptor food habits study. Condor 34~75-86. of the year. We have furthermore suggestedthat ERRINGTON,P. L. 1933. Food habits of southernWis- most breeding seasonprey items, avian or oth- consin raptors. Part II. Hawks. Condor 35: 19-29. erwise, are vulnerable ground foraging species FISCHER,D. L. 1986. Daily activity patternsand hab- and individuals captured by tactics that allow itat use of coexisting Accipiter hawks in Utah. Ph.D.diss., Brigham Young Univ., Provo, UT. them little chance to exercise whatever agility FISHER,A. K. 1893. Hawks and owls of the United they may possess. Statesin their relation to aariculture.USDA Bull. The assumptions underlying a correlation be- No. 3. tween prey agility and the pronounced RSSD of FITCH, H. S. 1974. Observations on the food and the Cooper’s Hawk might indeed be valid in non- nesting of the Broad-winged Hawk (Buteoplutyp- term) in northeasternKansas. Condor 76~331-333. breeding seasonsbut adequate dietary data are FITCH,H. S., B. GLAD~NG,AND V. HOUSE. 1946. Ob- not available for those seasons,and we doubt servations on Cooper’s Hawk nesting and preda- that studies from the breeding season provide tion. Calif. Fish Game 32: 144-154. sufficient support for such assumptions. The pre- HAMERSTROM,F. N., JR., ANDF. HAMERSTROM.195 1. Food of young raptors on the Edwin S. George dominance of young birds in the avian compo- Reserve. Wilson Bull. 63: 16-25. nent of breeding seasondiets of some other fal- JACKSON,H.H.T. 1961. Mammals of Wisconsin. coniforms, as noted earlier, suggests that Univ. Wisconsin Press. Madison. assumptions about prey agility may also need re- JANIK, C. A., AND J. A. MOSHER. 1982. Breeding evaluation in other cases,to the extent that di- biology of raptors in the central Appalachians. Raptor Res. 16:18-24. etary data are derived from studies of breeding JONES,S. 1979. Habitat managementseries for unique hawks. or endangeredspecies. Report No. 17, the Accip- iters. USDI-BLM Tech. Note 335. ACKNOWLEDGMENTS KENNEDY, P. L. 1980. Prey size selectionpatterns of We dedicatethis paper to the late Frederick N. Hamer- nesting male and female Cooper’s Hawks (Accip- Strom and the late Mary H. Nelson, without whose iter coouerii).M.S.thesis. Univ. ofIdaho. Moscow. example, support, and friendship our work would be ID. - much poorer. Fieldwork was financed in part by the KENNEDY, P. L., AND D. R. JOHNSON. 1986. Prey- Madison. Milwaukee. and Lakeland Audubon Soci- size selectionin nestingmale and female Cooper’s eties;C. and M. Nelson; ConsolidatedPapers, Inc.; the Hawks. Wilson Bull. 98: 110-l 15. Wisconsin Department of Natural Resources;and the KENNEDY, P. L., ANDJ. A. GESSAMAN.1991. Diurnal Wisconsin Societv for Omitholoav. We thank these resting metabolic rates of accipiters. Wilson Bull. organizations and individuals, as-well as the staff of 103:101-105. the Kettle Moraine State Forest and Walter and Pearl LINDUSKA,J. P. 1943. Cooper’s Hawk carrvina a nest Olsen for permission to erect blinds on their lands and of young Goldfinches. Auk 60:597. . - other help. S. K. Sherrod provided helpful comments MARTI, C. D. 1987. Raptor food habits studies, p. on the manuscript. We are also grateful to Randy Ju- 67-79. In B. G. Pendleton, B. A. Millsap, K. W. rewicz and especiallyto Paula Barsamian and Colleen Kline, and D. A. Bird reds.], Raptor management Rosenfield for their longstandingsupport and encour- techniaues manual. Natl. Wildl. Fed. Sci. Tech. agement. The University Personnel Development Ser. No. 10. Committee at the University of Wisconsin-Stevens MENG,H. K. 1951. Cooper’s Hawk Accipitercooperii Point provided support for publication. (Bonaparte).Ph.D.diss., Cornell Univ., Ithaca, NY. MENG, H. K. 1959. Food habits of nesting Cooper’s LITERATURE CITED Hawks and Goshawks in New York and Penn- CLENCH,M. H., ANDR. C. LIEBERMAN.1978. Weights sylvania. Wilson Bull. 7 1:169-l 74. of 151 speciesof Pennsylvania birds analyzed by MILLSAP,B. A. 1981. Distributional status of Fal- month, age, and sex. Bull. Carnegie Mus. Nat. coniformes in west central Arizona with notes on Hist. No. 5, Pittsburgh. ecology, reproductive success,and management. CRAIGHEAD, J. J., AND F. C. CRAIGHEAD,JR. 1956. USDI-BLM Tech. Note 355. Hawks, owls and wildlife. Stackpole Co., Harris- MOSHER,J. A., AND P. F. MATRAY. 1974. Size di- burg, PA. morphism: a factor in energy savings for Broad- DUNCAN,S. 1966. An analysis of the stomach con- winged Hawks. Auk 9 1:325-34 1. tents of some Cooper’s Hawks (Accipitercooperii). NELSON,R. W. 1968. Nest-robbing by Cooper’s Auk 83:308. Hawks. Auk 85696-697. DUNNING,J. B., JR. 1984. Body weights of 686 spe- PETERSON,D. J., AND R. K. MURPHY. In press. Prey cies of North American birds. Western Bird Band- delivered to two Cooper’s Hawk, Accipiter coo- ing Assoc. Mono. No. 1. perii, nests in northern mixed grassprairie. Can. EARHART,C. M., AND M. K. JOHNSON. 1970. Size Field Naturalist. 436 J. BIELEFELDT, R. N. ROSENFIELD AND J. M. PAPP

QUINN, M. S. 1991. Nest site and prey of a pair of Rosntnnn~~,R. N., M. W. GRATSON,AND L. B. CARSON. Sharp-shinned Hawks in Alberta. J. Raptor Res. 1984. Food brought by Broad-winged Hawks to 25:18-19. a Wisconsin nest. J. Field Omithol. 55:246-247. REYNOLDS,R. T. 1972. Sexual size dimorphism in SHERROD,S. K. 1978. Diets of North American Fal- accipiter hawks: a new hypothesis. Condor 74: coniformes. Raptor Res. 12:49-l 2 1. 191-197. SNYDER,N.F.R., H. A. SNYDER,J. L. LINGER,AND R. REYNOLDS,R. T., AND E. C. MESLOW. 1984. Parti- T. REYNOLDS. 1973. Organo-chlorines, heavy tioning of food and niche characteristicsof coex- metals, and the biology ofNorth American AC- istingAccipiter during breeding.Auk 101~76l-779. cipiters. Bioscience23:300-305. ROSENBERG,K. V., AND R. J. COOPER. 1990. Ap- SNYDA, N.F.R., ANDJ. W. WILEY. 1976. Sexualsize proaches to avian diet analysis, p. 80-90. In M. dimorphism in hawksand owls of North America. L. Morrison, C. J. Ralph, J. Vemer, and J. R. Jehl, American Ornithologists Union Mono. No. 20. Jr. [eds.], Avian foraging: theory, methodology, STORER,R.W. 1966. Sexual dimorphism and food and applications. Stud. Avian Biol. No. 13. habits in three North American Accipiters. Auk ROSENFIELD,R. N. 1988. Cooper’s Hawk: food, p. 83:423436. 353-354. In R. S. Palmer [ed.], Handbook ofNorth TOLAND,B. 1985. Food habits and hunting success American birds, Vol. 4. Yale Univ. Press, New of Cooper’s Hawks in Missouri. J. Field Omith. Haven, CT. 56:419422. ROSENFIELD,R. N., J. BIELEFELDT,AND J. CARY. 1991. WIENS,J. A. 1989. Ecologyofbird communities. Vol. Copulatory and other pre-incubation behaviors of 1: Foundations and patterns. Cambridge Univ. Cooper’s Hawks. Wilson Bull. 103:656-660. Press, New York.