(ISSN Volume 24 Winter 1990 Number 4 Contents Nest Defense by Male and Female Spanish Imperial Eagles. Miguel Ferrer , 77 The Identity of Pennant’s “Wapacuthu Owl” and the Subspecific Name of the Population of Bubo virginianus from West of Hudson Bay. M. Ralph Browning and Richard C. Banks 80 Habitat Use of the Northern Harrier in a Coastal Massachusetts Shrubland with Notes on Population Trends in Southeastern New ENGLAND. Dave A. Christiansen, Jr. and Steven E. Reinert 84 Molt Pattern and Duration in a Female Northern Goshawk (Accipiter gentilis). Christopher J. Reading 91 Raptor Road Surveys in South America. David h. Ellis, Richard l. Glinski and Dwight G. Smith 98 Response of Northern Goshawks to Taped Conspecific and Great HORNED Owl Calls. J. Timothy Kimmel and Richard H. Yahner 107 Clinical Hematology and Blood Chemistry Values for the Common BUZZARD ( Buteo buteo). Mauro Hernandez, Sonsoles- Martin and Paloma Fores 113 Is the Operational Use of Strychnine to Control Ground Squirrels Detrimental to Burrowing Owls? Paul c. James, Glen a. Fox and Thomas j. Ethier . . . ... 120 Short Communication Golden Eagles Take up Territories Abandoned by Bonelli’s Eagles in Northern Spain. Carmelo Fernandez and Jesus A. Insausti 124 Letters 126 Thesis Abstracts 127 News and Reviews 128 The Raptor Research Foundation, Inc. gratefully acknowledges a grant and logistical support provided by the University of Saskatchewan to assist in the publication of the journal. Persons interested in predatory birds are invited to join The Raptor Research Foundation, Inc. Send requests for information concerning membership, subscriptions, special publications, or change of address to Jim Fitzpatrick, Treasurer, 12805 St. Croix Trail, Hastings, Minnesota 55033, U.S.A. The Journal of Raptor Research (ISSN 0892-1016) is published quarterly for $18.00 per year by The Raptor Research Foundation, Inc., 12805 St. Croix Trail, Hastings, Minnesota 55033, U.S.A. Second class postage paid at Hastings, Minnesota, and additional mailing offices. Printed by Allen Press, Inc., Lawrence, Kansas, U.S.A. Copyright 1990 by The Raptor Research Foundation, Inc. Printed in U.S.A. THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER. THE JOURNAL OF RAPTOR RESEARCH A QUARTERLY PUBLICATION OF THE RAPTOR RESEARCH FOUNDATION, INC. Vol. 24 Winter 1990 No. 4 /. Raptor Res. 24(4):77-79 © 1990 The Raptor Research Foundation, Inc. NEST DEFENSE BY MALE AND FEMALE SPANISH IMPERIAL EAGLES Miguel Ferrer Estacion Biologica de Donana, Avd. Maria Luisa s/n, Pabellon del Peru, 41013 Sevilla, Spain Abstract. — Sexual differences in nest defense behavior in response to human intruders were studied in the Spanish Imperial Eagle ( Aquila adalberti). Females defended the nest more than males, and their defense increased as the breeding cycle progressed. The defensive behavior of the female was independent of male presence. Defensa del nido por progenitores de Aguila Imperial Iberica Extracto. — Se estudian las diferencias entre sexos en el comportamiento de defensa del nido, en respuesta a la presencia humana, en el Aguila Imperial Iberica ( Aquila adalberti). Las hembras defienden el nido mas que los machos, y su comportamiento defensivo aumenta al avanzar el ciclo reproductive. El com- portamiento defensivo de la hembra es independiente de la presencia del macho. Diurnal raptors exhibit three strategies of nest defense: principally by the female, by both sexes, or principally by the male (Mueller and Meyer 1985, Andersson and Wiklund 1987). There are also dif- ferences in the division of other roles associated with reproduction, such as incubation, provisioning of food, feeding the young and surveillance (Cramp and Sim- mons 1980, Mueller and Meyer 1985). In the Span- ish Imperial Eagle {Aquila adalberti ), incubation and care of chicks in the nest are principally carried out by the female (Cramp and Simmons 1980), but it is unknown whether or not there are sexual differences in offspring protection. Several authors have sug- gested that the general trend towards reversed sexual dimorphism in the Falconiformes may be at least partly an evolutionary response to sexual differences in nest defense (Snyder and Willey 1976, Andersson and Norberg 1981, Andersson and Wiklund 1987). Information on the division of nest defense between the sexes is scarce in raptors (Mueller and Meyer 1985), however. Recently Andersson and Wiklund (1987) studied nest defense in the Rough-legged Buzzard {Buteo lagopus ) and found that the smaller males undertake most of the defense. Although contrary to the pre- diction, this result does not refute the hypothesis. The evaluation of such an evolutionary hypothesis necessitates a comparative analysis seeking repeated patterns throughout the order (Pages and Harvey 1988). In this study, I examine the contribution of male and female Spanish Imperial Eagles to nest defense against human intruders. Methods The study was undertaken in Donana National Park in southwest Spain from 1974-1988. Observations were made during 201 visits to 51 nests of 15 different pairs. A visit involved one person climbing to the nest to record breeding stage and examine eggs or young. During the visit I recorded the sex, and distance from the nest and attack behavior of the adults. Even though the sexual dimorphism is slight (mean weight was 2613 g [N = 16] for males, and 3467 g [N = 21] for females, Ferrer unpubl. data), it was possible to assign sex to adults with confidence approximately 50% of the time, using individual differ- ences in moult and observations of the two members of the pair together. Nest defense was classified into three categories based on the distance the adults kept from the person investigating the nest: 1) >50 m when the adult remained more than this distance from the nest while the chicks were handled or the eggs examined, 2) <50 m when 77 78 Miguel Ferrer Vol. 24, No. 4 Table 1. Observations of nest defense by Spanish Im- perial Eagles according to distance classes and sex (0 m corresponds to direct attack). Distance MALE FEMALE Incuba- tion Nestling Stage Incuba- tion Nestling Stage >50 m 10 8 15 9 <50 m 2 2 9 17 0 m 0 1 2 10 Total 12 11 26 36 the adult remained closer than this distance and, 3) 0 m when the adult attacked the observer. Statistical analyses for evaluating the significance of nest defense variations were performed using N X M exact test (Wells and King 1980). Results In 85 observations the sex of attending adults was recorded. The female was always present, while the male was present on only 23 occasions (27%). The single versus paired ratio during visits in which the sex of the adults could not be identified (N = 116) was 41 .3%. Females remained closer and made more attacks than males (Table 1; x 2 = 10.65, P = 0.005). During incubation, females were not significantly more aggressive than males (% 2 = 2.66, P = 0.269). However, during the nestling stage the female was more aggressive than the male (x 2 = 8.32, P = 0.01 5). Males did not differ significantly in defense ac- tivity between incubation and nestling stages (x 2 = 1.18, P = 0.784). In contrast, females increased de- fense significantly in the later stage of the nesting cycle (x 2 = 7.89, P = 0.019). The responses of fe- males did not differ in the presence of males (Table 2, incubation x 2 = 0.35, P = 0.834, nestling x 2 = 0.04, P = 0.978). Discussion In the Spanish Imperial Eagle, incubation and the care of chicks in the nest are principally carried out by the female (Cramp and Simmons 1980). This was evident from our data, as we never observed the male alone at a nest, and we frequently observed females alone. During incubation, males provide food, while females are spending more time at the nest. Females defended more than males. This is in agree- ment with the hypothesis that the maintenance of reversed sexual dimorphism favors larger females which can better defend the nest against predators Table 2. Observations of female Spanish Imperial Ea- gles defending their nests in relation to presence or absence of the male (0 m corresponds to direct attack). Distance INCUBATION NESTLING STAGE Male Present Male Absent Male Present Male Absent >50 m 8 8 3 6 <50 m 3 5 5 12 0 m 1 1 3 7 Total 12 14 11 25 (Snyder and Willey 1976, Andersson and Norberg 1981). Recently, some authors (Wiklund and Stigh 1983, Andersson and Wiklund 1987) have presented an alternative hypothesis for the maintenance of re- versed sexual dimorphism in Falconiformes. Al- though still arguing that nest defense is an important selection pressure, they suggest that the selection would not be to increase female size, but rather to decrease male size. This would enhance flying agility and reduce the risk of defending the nest against predators of larger size, including human intruders (Andersson and Wiklund 1987). This hypothesis is not supported by our data because the smaller males did not defend the nest vigorously. Additionally, Pleasants and Pleasants (1988) have suggested that sexual dimorphism evolved through an increase in female size and not a decrease in male size, at least in diurnal raptors. My data show a rise in female nest defense be- havior as the breeding cycle progresses, a decrease in distances from the observers and an increase in the frequency of attacks. In contrast, males did not increase their defensive efforts along the breeding cycle. Similar results have been found in Merlins (. Falco columbarius ) by Wiklund (1990). He sug- gested that male investment in nest defense influ- enced mate selection by females. Consequently, later in the breeding cycle a greater investment is not necessary for males. The increase in female defensive behavior could be more related with the age of the chicks and accumulated investment (Andersson et al. 1980). Nevertheless, Ferrer et al. (in press), reported a rise in nest defense over the years in this eagle population, suggesting that this increase can be at- tributed to experienced adults whose defense behav- Winter 1990 Nest Defense by Spanish Imperial Eagles 79 ior has been positively reinforced upon not losing their offspring after a hypothetical predatory attack (Knight and Temple 1986). The fact that females spend more time at the nest and, consequently, they probably have more previous experiences with hu- man intruders than males, could explain, at least in part, the increase of aggressive behavior detected in this sex. Acknowledgments I wish to thank L. Garcia and R. Cadenas for their rigor and dedication to detail, without which this partic- ular study would not have been possible. Special thanks goes to E. Schupp for the revision of the English text and the improvement of the original manuscript. Literature Cited Andersson, M., C.G. Wiklund and H. Rundgren. 1980. Parental defense of offspring: a model and an example. Anim. Behav. 28:536-542. and R.A. Norberg. 1981. Evolution of reversed sexual size dimorphism and sex role partitioning among predatory birds with a size scaling of flight perfor- mance. Biol. J. Linn. Soc. 15:105-130. and C.G. Wiklund. 1987. Sex role partitioning during offspring protection in the Rough-legged Buz- zard ( Buteo lagopus). Ibis 129:103-107. Cramp, S. and K.E.L. Simmons. 1980. Handbook of birds of Europe, the Middle East, and North Africa. Vol. II. Oxford University Press, Oxford, U.K. Ferrer, M., L. Garcia and R. Cadenas. In press. Long-term changes in nest defense intensity of the Spanish Imperial Eagle. Ardea. Knight, R.L. and S.A. Temple. 1986. Why does in- tensity of avian nest defense increase during the nest- ling cycle? Auk 103:318-327. Mueller, H.C. and K. Meyer. 1985. Evolution of reversed sexual dimorphism in size. Pages 65-101 in J.F. Johnston [Ed.], Current Ornithology, Vol. II. Ple- num Press, New York. Pages, M.A. and P.H. Harvey. 1988. Recent devel- opments in the analysis of comparative data. Quart Rev. Biol. 63:413-440. Pleasants, J.M. and B.Y. Pleasants. 1988. Reversed size dimorphism in raptors: evidence for how it evolved. Oikos 52:129-135. Snyder, N. and J.W. Willey. 1976. Sexual size di- morphism in hawks and owls of North America. Or- nithol. Monogr. 20:1-96. Wells, H. and J.L. King. 1980. A general “exact test” for N X M contingency tables. Bulletin Southern Cal- ifornia Academy of Science 79:65-77. Wiklund, C.G. 1990. The adaptative significance of nest defence by merlin, ( Falco columbanus), males. Anim. Behav. 40:244-253. AND J. STIGH. 1983. Nest defense and evolution of reversed sexual dimorphism in Snowy Owls Nyctea scandiaca. Ornis. Scand. 14:58-62. Received 15 May 1990; accepted 21 October 1990 J. Raptor Res. 24(4):80-83 © 1990 The Raptor Research Foundation, Inc. THE IDENTITY OF PENNANT’S “WAPACUTHU OWL” AND THE SUBSPECIFIC NAME OF THE POPULATION OF Bubo virginianus FROM WEST OF HUDSON BAY M. Ralph Browning and Richard C. Banks Biological Survey Group, National Ecology Research Center, U.S. Fish and Wildlife Service, National Museum of Natural History, Washington, DC 20560 Abstract. — The name Strix wapacuthu Gmelin, often used for the subspecies of Bubo virginianus west of Hudson Bay, cannot be associated with certainty with either B. virginianus or Nyctea scandiaca. The subspecific name for the population of B. virginianus from Mackenzie to central-eastern British Columbia and northern Ontario should be B. v. subarcticus Hoy. Identidad del Buho Wapacuthu de Pennant, y el nombre dado a una poblacion de Bubo virginianus del oeste de la Bahia de Hudson Extracto. — El nombre Strix wapacuthu Gmelin, usado frecuentemente para una subespecie de Bubo virginianus del oeste de la Bahia de Hudson, no puede ser asociado con certeza ni con el Bubo virginianus ni con el Nyctea scandiaca. El nombre para esa subespecie, desde Mackenzie hasta el oeste central de Colombia Britanica, y el norte de Ontario debe de ser Bubo virginianus subarcticus Hoy. [Traduction de Eudoxio Paredes-Ruiz] Gmelin (1788:291) proposed the name Strix wa- pacuthu for a species separate from what are now Nyctea scandiaca (Snowy Owl) and Bubo virginianus (Great Horned Owl). Latham (1790) followed Gmelin (op. cit.). Swainson and Richardson (1832) likewise did not associate Strix wapacuthu with either N. scandiaca or B. virginianus. However, Swainson (in Swainson and Richardson 1832) described an- other taxon as Strix (Bubo) arctica, a name later used subspecifically for Great Horned Owls of much of western Canada (see Stone 1896). After Richmond (1902) showed that Swainson’s name was preoc- cupied by Bubo arcticus, proposed by Forster (1817) for the Snowy Owl, and therefore unavailable for any Great Horned Owl, the name was replaced by B. v. subarcticus proposed by Hoy (1852). The name wapacuthu was associated with N. scan- diaca by Coues (1874), Brewster (1906), Oberholser (1908, 1917), Manning (1952), Snyder (1961), and Godfrey (1986). On the other hand, Sharpe (1875) synonymized wapacuthu with B. virginianus, and the name was subsequently used for the subspecies of B. virginianus west of Hudson Bay by others (Ober- holser 1904, Ridgway 1914, Cory 1918, Peters 1940, American Ornithologists’ Union [A.O.U.] 1944, 1957, Snyder 1957). Todd (1963:454) recommended that the name wapacuthu be discarded altogether because it “cannot certainly be identified with any species.” Authors (e.g., Manning 1952, Godfrey 1986) who reject wapacuthu as applicable to B. vir- ginianus follow Richmond (1902), the A.O.U. (1910, 1931), and Taverner (1938) in their use of B. vir- ginianus subarcticus as the name for the pale sub- species of the Great Horned Owl from west of the Hudson Bay region. In recently published books on owls, Karalus and Eckert (1974) used wapacuthu as the name of a sub- species of B. virginianus distinct from subarcticus, whereas Johnsgard (1988) stated that it is “identi- cal” to subarcticus, and Voous (1988) referred to the western Hudson Bay population as “(. subarcticus or wapacuthu ).” McGillivray (1989) also indicated un- certainty by using ‘‘‘‘wapacuthu / subarcticus .” Karalus and Eckert’s (1974) treatment is taxonomically un- verified as well as conceptually faulty (the ranges of the two subspecies are shown to overlap). Johns- gard’s synonymy is incorrect even if his concept is correct, because he did not use the earliest available name for the subspecies. Because of the continuing various applications of wapacuthu in spite of earlier brief discussions of its description, a review of its use and identification follows. Gmelin (1788:291) cited the “Arctic Zoology” of Pennant (1785) as the basis for the description of 80 Winter 1990 Pennant’s “Wapacuthu Owl” 81 Strix wapacuthu. Pennant’s (1785) description was based on manuscript notes supplied by Thomas Hutchins. Hutchins, employed at York Factory by the Hudson’s Bay Company as surgeon and “Cor- responding Secretary,” copied a manuscript by (Wil- liams 1969, 1978) or collaborated with (C.S. Hous- ton, in litl.) Andrew Graham, an employee of the same company who worked mostly at Fort Severn. Hutchins (in Allen 1951:521) stated that “In per- suance of Mr. Graham’s advice, I have described the plumage of the Birds, but as my knowledge of the variety of colours is very small, consequently the description must be very imperfect.” In 1784 Hutch- ins convinced Pennant that he was the author of Graham’s observations (Williams 1978); both Pen- nant (1785), who knew of Graham’s work, and La- tham (1785:43) credited Hutchins for information on birds from what is now northwestern Ontario and northeastern Manitoba. Hutchins and Graham sent birds to the Royal Society in England (Forster 1772, Williams 1969) and to Latham (1821 :xii), but those specimens are no longer extant (Williams 1978). The fate of the name wapacuthu must rest on its written description. Pennant’s (1785:231-232) description is under the section heading “without ears” in a review of Arctic owls and is “no. 119. Wapacuthu [owl].” The de- scription is as follows: “With glossy black bill, and claws much incurvated: base of the bill beset with strong bristles: irides bright yellow: space between the eyes, cheeks, and throat, white: the ends of the feathers on the head black: scapulars, and all the coverts of the wings, white, elegantly barred with dusky reddish marks, pointing downwards: prima- ries, secondaries, and tail feathers, irregularly spot- ted and barred with pale red and black: back and coverts of the tail white, mixed with a few dusky spots: breast and belly dirty white, crossed with in- numerable reddish lines: vent white: legs feathered to the toes which are covered with hairs. Weight five pounds: length two feet: extent four.” Pennant (op. at.) also gave descriptions of several other species of owls. Among these he included detailed descriptions of what are now B. virginianus (pp. 228-229) under the heading “with ears,” and N. scandiaca (p. 233) under the heading “without ears.” Comments in the literature on Pennant’s descrip- tion of the Wapacuthu Owl are brief. Brewster (1906: 205) stated that wapacuthu referred to N. scandiaca because it was described as “earless.” Ridgway (1914) acknowledged that N. scandiaca has rudimentary ear tufts but associated the name wapacuthu with the Great Horned Owl, commenting that molting B virginianus “are often destitute of obvious ear-tufts . . . or the ear-tufts may have been plucked before the specimen came into his [Pennant’s] possession.” Peters (1940) used the name B. virginianus wapa- cuthu ; he believed that Pennant’s description was a composite that included characteristics of both B. virginianus and N. scandiaca. Manning (1952) stated that “there is nothing in this [Pennant’s] description which is not reconciliable [sic] with a Snowy Owl; while apart from there being no mention of horns or the fine vermiculation of a Horned Owl (its most obvious characteristics), there are several points which definitely separate it from any Horned Owl ...” but did not provide additional details. If the ear-tufts had been plucked or were absent through molting (Ridgway 1914) the combination of plumage char- acters could suggest B. virginianus ; if the bird was normally “earless” the Wapacuthu Owl can easily be associated with N. scandiaca. Pennant (1785) further stated that the Wapacu- thu Owl “makes a nest on the moss on the ground. The young are hatched in May, and fly in June, and are white for a long time after.” Nesting on the ground is consistent with N. scandiaca ; B. virginianus usually nests in trees, and only rarely on the ground (Bent 1938), cliff's (Peck and James 1983) and rock outcrops (Johnsgard 1988). The description of the young as white is consistent with both B. virginianus and N. scandiaca for about the first 10 days of the natal plumage (Godfrey, pers. coinm.; Johnsgard 1988). Older B. virginianus are buffy with the de- veloping flight feathers similar to those of the adults (Johnsgard op. cit.). Older N. scandiaca are chocolate brown with white specks (N.K. Johnson, in litt.), the facial disc is white, and the flight feathers are white with brown crossbars “and brown vermicu- lations in the form of speck-like marbling” (Mikkola 1983). Hutchins (in Latham 1787:49) stated that the eggs of the Wapacuthu are “from five to ten in number,” This exceeds the normal clutch size of B. virginianus (only one nest containing five eggs is cited in Bent [1938]) but is within the range of N. scan- diaca clutches (Portenko 1972). The name “Wapacuthu” of Pennant’s (1785) owl was from the Cree language, in which “wap” refers to white (C.S. Houston, in litt.). Swainson and Richardson (1832) and Brewster (1906) commented that “wapacuthu” meant “White Owl.” Graham (in Williams 1969:xxxv, 106, 107) used the names 82 M. Ralph Browning and Richard C. Banks Vol. 24, No. 4 “Wawpekatheu, the Spotted Owl” and “Wapaca- thew Omissew, The Snowy-Owl.” We agree with Glover (in Williams 1969:106) that Graham’s “Wawpekatheu” appears to be a heavily spotted example of N. scandiaca. Graham’s “Wapacathew Omissew,” merely described as smaller than the “Wawpekatheu,” probably also refers to N. scan- diaca and was so identified by Glover (in Williams 1969:107). We conclude that the description of the Wapa- cuthu Owl as lacking ear-tufts, information on the nesting and clutch size, and the meaning of the Cree name indicate that Pennant’s description was of Nyc- tea scandiaca. However, we agree with Todd (1963) that wapacuthu cannot be identified with certainty. The name Stnx wapacuthu Gmelin, 1788 should be regarded as a nomen dubium (a name of doubtful application), and the subspecific name for the pop- ulation of Bubo virginianus from Mackenzie to cen- tral-eastern British Columbia and northern Ontario (Godfrey 1986) should be subarcticus Hoy, 1852. Acknowledgments We thank I. Goddard of the Department of Anthro- pology (Handbook of North American Indians), Smith- sonian Institution, for information on the Cree names. W. E. Godfrey, N.K. Johnson, and J.T. Marshall read an early draft of the manuscript and offered many useful comments. We also thank R.J. O’Hara, S.L. Olson, and J.S. Weske who read the manuscript, and C.S. Houston and other reviewers for helpful comments. Literature Cited Allen, E.G. 1951 . The history of American ornithology before Audubon. Trans. American Philos. Soc. (new ser.) 41:387-591. American Ornithologists’ Union. 1910. Check-list of North American birds. Third edition. American Or- nithol. Union, Lancaster, PA. . 1931. Check-list of North American birds. Fourth edition. American Ornithol. Union, Lancaster, PA. . 1944. Nineteenth supplement to the American Ornithologists’ Union check-list of North American birds. Auk 61:441-464. . 1957. Check-list of North American birds. Fifth edition. American Ornithol. Union, Baltimore, MD. Bent, A. C. 1938. Life histories of North American birds of prey. Pt. 2. U.S. Natl. Mus. Bull. 170. Brewster, W. 1906. The birds of the Cambridge region of Massachusetts. Mem. Nuttall Ornithol. Club No. 4. Cory, C.B. 1918. Catalogue of birds of the Americas. Pt. 2, No. 1. Field Mus. Nat. Hist. Publ. 197, Zool. Ser. 13. Coues, E. 1874. Birds of the northwest. Dept. Interior, U.S. Geol. Surv. Terr. Misc. Publ. No. 3. Forster, J.R. 1772. An account of the birds from Hud- son’s Bay; with observations relative to their natural history; and Latin descriptions of some of the most uncommon. Phil. Trans. 62:382-433. . 1817. A synoptical catalogue of British birds. Nichols, Son, and Bentley, London, U.K. Godfrey, W.E. 1986. The birds of Canada. Rev. ed Natl. Mus. Nat. Sci., Ottawa, ON, Canada. Gmelin, J.F. 1788. Caroli Linne . . . Systema natur- ae. .. . Thirteenth edition. Vol. 1, Pt. 1. Hoy, P.R. 1852. Description of two species of owls, presumed to be new, which inhabit the state of Wis- consin. Proc. Acad. Nat. Sci. Philadelphia 6:210-211. Johnsgard, P.A. 1988. North American owls: biology and natural history. Smithsonian Institution Press, Washington, DC. Karalus, K.E. and A.W. Eckert. 1974. The owls of North America (north of Mexico). Doubleday and Co., Inc., New York. Latham, J. 1785. A general synopsis of birds. Vol. 1, Pt. 3. Leigh, Sotheby and Sons, London, U.K. . 1787. Supplement to the general synopsis of birds. Vol. 1. Leigh, Sotheby and Sons, London, U.K . 1790. Index ornithologicus. Vol. 1. Leigh and Sotheby, London, U.K. . 1821. General history of birds. Vol. 1. Jacob and Johnson, London, U.K. Manning, T.H. 1952. Birds of the west James Bay and southern Hudson Bay coasts. Natl. Mus. Canada Bull. No. 125 (Biol. Ser. 43). 114 pp. McGillivray, W.B. 1989. Geographic variation in size and reverse size dimorphism of the Great Horned Owl in North America. Condor 91:777-786. Mikkola, H. 1983. Owls of Europe. Buteo Books, Ver- million, SD. Oberholser, H.C. 1904. A revision of the American Great Horned Owls. Proc. U.S. Natl. Mus. 27:177-192. . 1908. A new Great Horned Owl from Vene- zuela, with notes on the names of the American forms. Mus. Brooklyn Inst. Arts and Sci. 1:371-373. . 1917. Notes on North American birds. III. Auk 34:465-470. Peck, G.K. and R.D. James. 1983. Breeding birds of Ontario: nidiology and distribution. Vol. 1. Nonpas- serines. Royal Ontario Museum, Toronto, ON, Can- ada. Pennant, T. 1785. Arctic zoology. Vol. 2. Henry Hughs, London, U.K. Peters, J.L. 1940. Check-list of birds of the world. Vol. 4. Harvard Univ. Press, Cambridge, MA. PORTENKO, L.A. 1972. Die Schnee-Eule, Nyctea scan- diaca. Neue Brehm-Bucherei, A. Ziemsen Verlag, Wit- tenberg, Germany. Winter 1990 Pennant’s “Wapacuthu Owl” 83 Richmond, C.W. 1902. The proper name for the Arctic Horned Owl. Proc. Biol. Soc. Washington 15:86. Ridgway, R. 1914. The birds of North and Middle America. U.S. Natl. Mus. Bull. No. 50, Pt. 6. SHARPE, R.B. 1875. Catalogue of the birds in the British Museum. Vol. 2. British Mus. (Nat. Hist.). Printed by order of the Trustees, London, U.K. Snyder, L.L. 1957. Arctic birds of Canada. University of Toronto Press, Toronto, ON, Canada. . 1961. On an unnamed population of the Great Horned Owl. Royal Ontario Mus. Life Sci. Div. Contr. 54. Stone, W. 1896. A revision of the North American horned owls with description of a new subspecies. Auk 13:153-156. Swainson, W. and J. Richardson. 1832. Fauna bo- reali-Americana. Pt. 2. The birds. John Murray, Lon- don, U.K. Taverner, P.A. 1938. Birds of Canada. The Musson Book Co. Ltd., Toronto, ON, Canada. Todd, W.E.C. 1963. Birds of the Labrador Peninsula and adjacent areas. University of Toronto Press, To- ronto, ON, Canada. Voous, K.H. 1988. Owls of the northern hemisphere. William Collins Sons and Co., Ltd., London, U.K. Williams, G. [Ed.] 1969. Andrew Graham’s observa- tions on Hudson’s Bay 1767-91. The Hudson’s Bay Rec. Soc., London, U.K. . 1978. Andrew Graham and Thomas Hutchins* collaboration and plagiarism in 18th-century natural history. Beaver 308:4-14. Received 17 April 1990; accepted 23 August 1990 J. Raptor Res. 24(4):84- 90 © 1990 The Raptor Research Foundation, Inc. HABITAT USE OF THE NORTHERN HARRIER IN A COASTAL MASSACHUSETTS SHRUBLAND WITH NOTES ON POPULATION TRENDS IN SOUTHEASTERN NEW ENGLAND Dave A. Christiansen, Jr - 1 and Steven E. Reinert Lloyd Center for Environmental Studies, P.O. Box 7037, South Dartmouth, MA 02748 Abstract. — Northern Harrier ( Circus cyaneus ) use of a 55.3 ha, shrub-dominated coastal peninsula in Massachusetts was monitored, year-round, from 19 February 1987 to 26 April 1988. In 1987, two harrier pairs established breeding territories there. Two nests with eggshells were found, and represent the only documented harrier nesting attempts on mainland Massachusetts (excluding Cape Cod) in more than a decade. Both nests were situated within patches of low (0.93 ± 0.28 m), dense shrubs dominated by Black Huckleberry ( Gaylussacia baccata). Each winter, most harriers roosted within the same two patches of dense Black Huckleberry ( X shrub height = 0.84 ±0.18 m). The maximum count of harriers at the winter roost site was 23 during February of 1988 (3 adult male, 20 brown). Though the number of breeding harriers in Southeastern New England has declined during this century, analysis of Christmas Bird Count data from 1962 to 1988 demonstrates an increase in the number of wintering harriers in the same region during that period. Uso del habitat por el Circus cyaneus en un terreno de arbustos de una costa de Massachusetts, con notas sobre las tendencias de poblacion en el sudeste de Nueva Inglaterra Extracto. — Desde el 19 de febrero de 1987 hasta el 26 de abril de 1988, se han hecho observaciones del uso que hacen las aves raptoras Circus cyaneus, de 55.3 hectareas pobladas de arbustos, en la costa de la peninsula de Massachusetts. En 1987, esta area fue el territorio nupcial de dos parejas de estas aves raptoras. Dos nidos con cascaras de huevos han sido hallados y representan los unicos indicios, en mas de una decada, de las tentativas de anidar que las raptoras de esta especie han dejado en tierra firme de Massachusetts (excluyendo Cape Cod). Los dos nidos estaban situados dentro de pequenas areas pobladas densamente por bajos arbustos (0.93 ± 0.28 m) en los que domina el arandano negro ( Gaylussacia baccata ) [“Arandano =. . . Planta de la familia de las ericaceas . , , con ramas angulosas, hojas alternas . . . frutos bayas negruzcas o azuladas, dulces y comestibles. . .”]. Cada invierno, la mayor parte de estas aves descansaban dentro de esas dos areas pobladas por densos arandanos ( X altura por planta = 0.84 ± 0.18). El maximo numero de estas aves en estos lugares de descanso invernal fue de 23 (3 adultos machos, 20 marrones), en febrero de 1988. Aunque el numero de estas aves de rapina durante el ciclo reproductive, en el sudeste de Nueva Inglaterra, ha declinado durante esta centuria, los analisis del computo de aves, realizado en las epocas navidenas, desde 1962 a 1988, demuestran un aumento en el numero de ellas, en los inviernos, en la misma region durante ese periodo. [Traduccion de Eudoxio Paredes-Ruiz] The Northern Harrier ( Circus cyaneus hudsonius ) is found year-round in southeastern New England. Although there is evidence to suggest that numbers of wintering harriers have increased in this region during recent years (Christmas Bird Count data in Audubon Field Notes 1962-1970, and American Birds 1971-1988), numbers of breeding harriers have declined during this century (Forbush 1927, Bent 1937, Griscom and Snyder 1955, Hill 1965, Ser- rentino and England 1989, Nikula and Holt, in 1 Present address: Box P-162, South Dartmouth, MA 02748. prep.). Breeding harriers have been nearly extir- pated from mainland habitats in Connecticut, Mas- sachusetts and Rhode Island (Serrentino and En- gland 1989). Breeding populations persist in Massachusetts on Cape Cod and offshore islands, and on Block Island, Rhode Island (Serrentino and England 1989, Nikula and Holt, in prep.). The harrier is an endangered species in Rhode Island, and is listed as threatened in Massachusetts (no sta- tus available for Connecticut). Serrentino and England (1989) reported on the diminishing number of nesting harriers in the Northeast, and suggested areas of research for de- 84 Winter 1990 Harrier Habitat Use 85 veloping effective management/recovery strategies for the species. Here, we provide detailed, year- round data on the habitat use of harriers at Barney’s Joy Point, a coastal shrubland in southeastern (mainland) Massachusetts. We monitored winter roosting activity beginning February 1987, and in April 1987 documented the establishment of breed- ing territories by two male harriers. A nest found in one of the territories represents the first docu- mented harrier nesting attempt in mainland Mas- sachusetts (excluding Cape Cod) in more than a decade, and a nest from the previous breeding season (with eggshell fragments) was also discovered. Study Area The study area, Barney’s Joy Point, is a shrub domi- nated coastal peninsula in South Dartmouth, Massachu- setts (Fig. 1). The 55.3 ha study area is bordered to the northwest by a coastal dune system and tidal pond/marsh complex. Cultivated fields lie to the north of the study area, and the remainder of the, point is surrounded by the waters of the Slocum’s River to the east, and Buzzard’s Bay to the south and southwest. Harriers hunted in hab- itats adjacent to the study site, however all breeding and most roosting occurred within the boundaries of the study area. Although cattle grazing in past years has left an extensive network of cowpaths throughout the area, it is presently used as pasture for less than twenty cattle in the summer. We classified the habitats of the study area into five types. Shrub/herb (14.8 ha, 26.8% total area) occupies two tracts of pasture land in the northern and northeastern zones of the study area (Fig. 1). Several small, seasonal freshwater pools are found within the northeastern tract. Shrub/herb is characterized by a complete ground cover of grasses and forbs occurring beneath a sparse 1- to 2-m tall shrub overstory. Shrubs are denser and taller (2-4 m) along remnant hedgerows and within a few small patches within this community. Dominant shrubs are Black Huck- leberry ( Gaylussacta baccata), Highbush Blueberry ( Vac - cinium corymbosum ), Dwarf Sumac ( Rhus copallina), and Northern Arrowwood ( Viburnum recognitum). Grassland (6.5 ha, 11.8% total area) covers two tracts that are interspersed with the shrub/herb tracts in the northern section of the study area (Fig. 1). This habitat is maintained by annual mowing and is dominated by Red Fescue ( Festuca rubra ) and Wavy Hair Grass {Deschamp- sia flexuosa). Various forbs are also present. Dense shrub/grassland (16.9 ha, 30.6% total area) oc- cupies a wide band in the center of the study area (Fig. 1). It is characterized by dense, 0.5- to 2-m tall patches of Black Pluckleberry (62% of total area) with pockets of short (<30 cm) grasses (38% total area) interspersed among the shrub patches. Subdominant shrubs include Bayberry (Myrica pensylvanica), Virginia Rose ( Rosa virginiana ) and Pasture Rose ( Rosa Carolina). Grassy areas are dominated by Red Fescue and Wavy Hair Grass. Several junipers ( Jumperus virginiana) , most less than 3 m tall, are scattered throughout the community. Dense shrub/dune (7.7 ha, 13.9% total area) is found in a band along the southwestern edge of the dense shrub,/ grassland habitat (Fig. 1). This habitat consists of low, rolling sand dunes with numerous, small ( < 50 m 2 ) patches of low (<0.5 m) shrubs interspersed among herbaceous dune vegetation and areas of unvegetated sand. Black Huckleberry and Seaside Rose ( Rosa rugosa) are the dom- inant shrubs; American Beachgrass ( Ammophila brevili- gulata) and Beach Heather ( Hudsonia tomentosa) dominate the herbaceous community. Beachgrass dune (2.6 ha, 4.7% total area) occupies the southwestern corner of the study area (Fig. 1). This habitat consists of primary and secondary sand dunes, and is cov- ered almost entirely by American Beachgrass. Other spe- cies include Dusty Miller ( Artemisia stelleriana). Seaside Goldenrod ( Solidago setftpervirens ), and Seaside Rose. Ar- eas of unvegetated sand are also present. A brackish pond (0.8 ha, 1.4% total area) is located adjacent to the shoreline on the eastern side of the study area. Unpaved roads and rocky shoreline comprise the remaining 0.3 ha (0.5%) and 5.7 ha (10.3%) of the study area, respectively. Methods Observations of harriers at Barney’s Joy were made from 19 February 1987 to 26 April 1988. That period encompassed one full breeding season, which we defined as 9 April (initiation of courtship behavior) to 31 July 1987. We made 93 hours of observations during that breeding season. We observed harriers for 57 hours during the 86-87 roost period and for 112 hours during the 87- 88 roost period. Most roost observations in 87-88 were made during evenings. However, between 12 January and 10 March 1988, weekly counts were made during the morning by two observers, each counting the number of harriers leaving roosts. Observers arrived approximately 30 minutes before sunrise and remained until dispersal activity ceased, up to 1 hour after sunrise. A cover map of the study area (Fig. 1) was made from 1:9600 scale black and white vertical aerial photographs Habitat patches >50 m 2 were delineated with the aid of a 4x lens stereoscope and an 8x monocular lens. We also delineated two areas where most wintering harriers roost- ed. We did not attempt to delineate breeding territory boundaries. The areas of each habitat type and of the two winter roosts were measured on an enlarged (1:4048 scale) cover map using a dot-grid and compensating polar pla- nimeter. After the 1987 breeding season, shrub heights were estimated within each of the two principal roost sites, and within the shrub/herb and dense shrub/grassland habi- tats. Within each area sampled, two transects perpendic- ular to one another, and crossing at an approximated center of the area, were established; the bearing of the first transect was determined randomly. Shrub height and spe- cies were noted at 10-m intervals (5 m within roost sites) across the full length of each transect. No data were taken if the sample point fell in an herbaceous patch. The percent cover of habitat types surrounding the two nest sites was estimated along four, 5-m transects extend- ing out from each nest (Holt and Melvin 1986, Brower Vol. 24, No. 4 86 Dave A. Christiansen, Jr. and Steven E. Reinert Figure 1 . Distribution of habitats at Barney’s Joy Point, South Dartmouth, Massachusetts. Delineated areas represent winter roosts 1 and 2. Circled “ + ” signs indicate the sites of nests a and b. Winter 1990 Harrier Habitat Use 87 and Zar 1977). The transects were then extended from the nest to the edge of the patch of shrubs within which it occurred to determine the mean distance from each nest to the edge of its patch (Holt and Melvin 1986). To assess population trends of wintering harriers in southeastern New England, we determined the mean num- ber of harriers counted per party hour (Raynor 1975) across 17 Christmas Bird Count (CBC) census-circles in Massachusetts (9 circles), Connecticut (6 circles) and Rhode Island (2 circles) for each year from 1962 to 1988. CBC census-circles with no reported harrier observations during that period were excluded from our analysis. Also excluded were census-circles that were not censused during each year of that period. We chose 1962 as a starting point since several CBC census-circles with substantial harrier counts were added in that year. Results and Discussion Year-round Activity. Six to ten harriers (in- cluding 2 adult males) roosted at Barney’s Joy from February through March 1987. Behavior suggesting defense of a breeding territory was seen for the first time on 1 April 1987, when one male was seen escorting (Bildstein and Collopy 1985) another for approximately 8 minutes in the evening. Immature harriers were seen for the last time on 9 April, co- inciding with the first day an adult male was seen performing aerial courtship displays. By 15 April, two male harriers had established breeding territo- ries, the first of which (Territory 1) was centered in the larger shrub/herb field, and the second (Terri- tory 2) in the dense shrub/grassland habitat tract of the study area (Fig. 1). The two males repeatedly engaged in escorting flights along a line where the two territories met. Female harriers were frequently seen within the study area at that time. Courtship behaviors, including skydancing and aerial food transfers (Hamerstrom 1986) peaked during the 3rd week in April, and by late April two breeding pairs were established. Copulation (Ter- ritory 1 pair) was first seen on 30 April, and on 27 May a female was nest-building. Occasional food transfers were observed until the end of June; how- ever, neither of the pair from Territory 1 was seen during July. A nest containing eggshell fragments w r as found in Territory 1 on 31 July; the eggs had apparently been destroyed by a mammalian pred- ator. The male from Territory 2 was seen carrying nesting material on 7 May. The female of this pair was never associated with a nest, and was last seen on 28 May. An incomplete nest was found in Ter- ritory 2 on 24 June; we believe that that nest was built by the male as observed by Hamerstrom (1986). A lone male roosted in this territory throughout the remainder of the breeding season. After the 1987 breeding season, harriers began roosting at Barney’s Joy in mid-September, and by mid-October, 4 to 6 birds were roosting regularly within the same shrub patches as those used the previous winter. Throughout the 87-88 winter roost period there were substantial fluctuations in the numbers of roosting harriers. A maximum count of 23 (including only 3 adult males) was made on 12 January 1988; numbers then decreased over the next two weeks and levelled at approximately 15 until 25 February, after which numbers steadily decreased until the last count of two brown birds on 14 April. A brush fire on 19 March 1988 consumed 25.7 ha (46.5%) of the study area, including most shrub/ herb, dense shrub/grassland, and grassland habitats. For the remainder of the study period, only a few female or immature harriers roosted in the area, and no breeding territories were established at Barney’s Joy during 1988. Roosting Behavior and Habitat. Harriers gen- erally left roosts between 20 min before and 40 min after sunrise, with most leaving approximately 10 min before sunrise. In the evening, single harriers began arriving at the roost approximately 40 min before sunset. Most had entered the roost area by 10 min past sunset, and the greatest number of har- riers could be seen at this time. These behaviors are consistent with the findings of Weller et al. (1955) in Missouri, and Bosakowski (1983) in New Jersey. Harrier roosting sites were concentrated within tw'O large patches of dense Black Huckleberry (with- in dense shrub/grassland) in both years (Fig. 1). Roost 1 (1.9 ha) had 74.9% dense shrub cover; Roost 2 (2.6 ha) had 73.3% dense shrub cover. The mean height of shrubs in Roost 1 was 68.7 cm (SD ± 21.8 cm) and in Roost 2, 83.7 cm (SD ± 17.6 cm). Roost- ing harriers avoided patches of shorter shrubs (<0.3 m), which were interspersed with the taller shrub patches in Roost 1. Harriers roosted in dune swales to the west of the study area for more than a month after fire removed shrub cover from most of Roost 1. Although shrub height in shrub/herb was not significantly different than that in dense shrub/grassland ( X = 84.8 cm and 82.2 cm, respectively), harriers did not roost in the former, suggesting that dense shrub patches are preferred over sparse stands as roosting sites. Within roost areas, harriers used three site-types for roosting: (1) small patches of flattened forbs with- 88 Dave A. Christiansen, Jr. and Steven E. Reinert Vol. 24, No. 4 in the shrubs; (2) completely or partially flattened shrubs; and (3) narrow, grass covered cowpaths within the shrub patches. Most individual roost sites were smaller than 0.25 m 2 , and biweekly pellet col- lection demonstrated that harriers were generally faithful to these sites for several days or weeks at a time. Occasionally, feces and pellets would be piled on opposite sides within a site, indicating that the harrier had spent several nights in the same position on the ground. Harriers infrequently roosted to- gether in the same opening within a shrub patch. Though the harriers used the same shrub patches for roosting during each winter period, the actual roost sites used within the patches were not the same in each year. Breeding Season Habitat. Both breeding terri- tories were centered in shrub-dominated habitats. The mean height of shrubs sampled in Territory 1 (shrub/herb) was 84.8 cm (SD ± 32.6), and in Territory 2 (dense shrub/grassland), 82.2 cm (SD ± 29.3). The two nests found with eggshell fragments were both within patches of dense shrubs of the same height (0.93 ± 0.28 m). The 1987 nest (Territory 1 ) was placed in a patch of dense huckleberry within the generally sparser, shrub/herb habitat, near the highest elevation of the study area (Fig. 1, nest a). Sumac, Northern Arrowwood and blueberry were subdominant shrubs at that site. The nest from the 1986 breeding season (found while sampling vege- tation characteristics within the boundaries of Ter- ritory 2) was situated on a low hummock, uniformly covered with very dense huckleberry, Virginia Rose and Pasture Rose (Fig. 1, nest b). The habitat used by breeding harriers at Barney’s Joy was similar to that of harriers nesting on Nan- tucket Island, Massachusetts in 1985 (Holt and Melvin 1986). There, as at our study site, nests were placed in dense shrub patches within shrub-domi- nated territories. The mean distance from the Nan- tucket nests to the edge of the shrub patch within which they occurred (i.e., nearest herbaceous vege- tation) was 14.3 m (range = 3.1-27.0 m) (Holt and Melvin 1986); the mean for the two nests at Barney’s Joy was 17.3 m (10.8 and 27.7 m). Thus, although historical accounts of harriers in southeastern New England do not list upland hab- itats as nesting sites (Forbush 1927, Bent 1937, Gris- com and Snyder 1955, Hill 1965), such sites are now relatively common. This suggests that harriers have moved into upland sites in recent decades, or alter- natively, that upland sites went unnoticed by pre- vious investigators. F. Hamerstrom (pers. comm.) theorizes that harriers have been forced to breed in shrub communities as more preferred nesting hab- itats have become scarcer. Regardless of the reason, coastal upland-shrub habitats should be considered as potential harrier nesting areas when conducting surveys of breeding harriers, or of potential breeding habitat. Harrier Population Trends in Southeastern New England. As early as 1927 Forbush (1927: 100) noted that harriers were “formerly much more common” in New England, and subsequent accounts provide evidence that a decline in the number of breeding harriers has continued throughout this cen- tury (Bent 1937, Griscom and Snyder 1955, Hill 1965, Serrentino and England 1989, Nikula and Holt, in prep.). Presently, only 8-10 pairs are believed to nest on Cape Cod, and the islands of Martha’s Vineyard and Nantucket each support about 25 pairs (Nikula and Holt, in prep.). The harrier has been extirpated from mainland Connecticut and Rhode Island, and Barney’s Joy represents the only known mainland (excluding Cape Cod) breeding site in Massachu- setts (T. French, pers. comm.). Nikula and Holt (in prep.) cite loss of habitat as the principal reason for the marked reduction in the number of breeding harriers in Massachusetts. Serrentino and England (1989) broaden that explanation to include all of the Northeast. Both cite loss of coastal wetlands and the reforestation of farmlands as the principal factors relating to habitat loss. Serrentino and England (1989:42) noted that “Harrier populations do not appear to have de- creased as drastically on their wintering grounds in the Northeast compared to the decline in the number of breeding birds. ...” That premise is supported for southeastern New England by our relatively high counts of wintering harriers at Barney’s Joy, and by Reinert (1984), who observed as many as five har- riers hunting simultaneously during the winter at Sachuest Point, Rhode Island, where harriers do not breed. Results of the Christmas Bird Count analysis indicate a trend of increasing numbers of wintering harriers in southeastern New England beginning in 1975, with a leveling off occurring from 1983-1988 (Fig. 2). The mean number of harriers per 100 party hours was 3.66 from 1962-1966, and 5.35 from Winter 1990 Harrier Habitat Use 89 YEAR Figure 2. Three-year moving average of number of har- riers seen per 100 party hours during Christmas Bird Counts, cumulatively for 17 census-circles in Massachu- setts, Connecticut and Rhode Island, for years 1962-1988. 1984-1988, which represents an increase of 46% during the 27-year period. We found no significant relationships between harrier numbers and mean temperature or snow cover during the same 27-year period. Thus, while the number of breeding harriers in southeastern New England has decreased during re- cent decades, there is substantial evidence to suggest that wintering numbers have increased. This indi- cates that (1) a pool of harriers, which represents potential nesting birds, departs from the region at the end of the winter roost period, and (2) factors which do not operate in the winter are acting to limit the number of harriers that remain in southeastern New England to breed. Serrentino and England (1989) suggest that harriers leave wintering areas with potentially suitable nesting habitat due to the increased use of coastal areas by humans during the breeding season months. This explanation seems plausible at areas such as Barney’s Joy and Sachuest Point, where nearby beaches attract many visitors and seasonal residents in the spring and summer seasons. However, it is premature to attribute the opposing seasonal trends in harrier populations sole- ly to this explanation. Other factors, such as migra- tion patterns, seasonal changes in prey availability, and attraction to natal breeding areas, may also play roles. An understanding of these factors is essential in developing effective management strategies for breeding harriers, and we urge wildlife researchers elsewhere in the Northeast to initiate investigations in this area. Of particular interest would be studies of (1) harrier movements as they disperse from their winter roosts (via marking or radiotelemetry), and (2) the relationships between the prey base and har- rier nesting and roosting habitats. Acknowledgments This research was funded in part by the Island Foun- dation. We thank R. Deegan, M. Buehler, M. Mello, J Lyons, R. Marshall, and W. DeRagon for assisting with harrier censuses, nest searches and other field work, C. Murchie provided invaluable assistance in compiling Christmas Bird Count records. Access to the study areas was graciously provided by Tom and Angelica Claggett. Finally, we thank P. Serrentino, M. England, M. Mello, and S. Melvin for reviewing early drafts of this manu- script. Literature Cited Bent, A. C. 1937. Life histories of North American birds of prey. Part I. U.S. Natl. Mus. Bull. 167. Washington, DC. Bildstein, K.L. and M.W. Collopy. 1985. Escorting flight and agonistic interactions in wintering Northern Harriers. Condor 87:398-401. Bosakowski, T. 1983. Density and roosting habitats of Northern Harriers wintering in the Hackensack Meadowlands. Records of New Jersey Birds 9:50-54. BROWER, J.E. AND J.H. Zar. 1977. Field and laboratory methods for general ecology. W.C. Brown Co. Publ , Dubuque, IA. Forbush, E.H. 1927. Birds of Massachusetts and other New England states. Vol. II. Massachusetts Depart- ment of Agriculture, Boston, MA. Griscom, L. and D. Snyder. 1955. The birds of Mas- sachusetts, an annotated and revised check list. Peabody Museum, Salem, MA. HameRSTrom, F. 1986. Harrier, hawk of the marshes Smithsonian Institution Press, Washington, DC. Hill, N.P. 1965. The birds of Cape Cod, Massachu- setts. William Morrow and Co., New York. Holt, D.W. and S.M. Melvin. 1986. Population dy- namics, habitat use, and management needs of the Short- eared Owl in Massachusetts: summary of 1985 research. Massachusetts Division of Fisheries and Wildlife, Natural Heritage Program, Boston, MA. Nikula, B. and D.W. Holt. In prep. Northern Har- rier. In T.W. French and J.E, Cardoza [Eds.], Endan- gered, threatened, and special concern vertebrates of Massachusetts. Massachusetts Division of Fisheries and Wildlife, Natural Heritage Program, Boston, MA. RAYNOR, G.S. 1975. Techniques for evaluating and an- alyzing Christmas Bird Count data. Am. Birds 29:626- 633. Reinert, S.E. 1984. Use of introduced perches by rap- 90 Dave A. Christiansen, Jr and Steven E. Reinert Vol. 24, No. 4 tors: experimental results and management implica- tions. Raptor Res. 18:25-29. Serrentino, P. and M. England. 1989. Northern Harrier. Pages 37-46 in Proc. Northeast Raptor Man- agement Symposium and Workshop. Natl. Wildl. Fed., Washington, DC. Weller, M.W., I.C. Adams, Jr., and B.J. Rose. 1955. Winter roosts of Marsh Hawks and Short-eared Owls in central Missouri. Wilson Bull. 69:189-193. Received 22 March 1990; accepted 22 August 1990 J Raptor Res. 24(4):91-97 © 1990 The Raptor Research Foundation, Inc. MOLT PATTERN AND DURATION IN A FEMALE NORTHERN GOSHAWK {Accipiter gentilis) Christopher J. Reading Institute of Terrestrial Ecology, Furzebrook Research Station, Wareham, Dorset BH20 5 AS U.K. Abstract. — The successive molt patterns (1983-1989) of the major flight feathers (alulas, primaries, secondaries and tail) were studied in a wild bred, captive female German goshawk ( Accipiter gentilis). Where possible, the date that each feather was shed was recorded, and the rachis length and thickness measured. The observed variation in the total number of feathers molted each year was due to annual differences in the number of tail and secondary feathers shed. With the exception of 1985, all alulas and primary feathers were replaced annually. The pattern of secondary and tail feather loss in any one year was largely dependent on the molt pattern of the previous year. Although the duration of the overall molt decreased from a maximum of 141 days in 1984 to a minimum of 98 days in 1988, the onset of each molt remained relatively constant between days 134 and 145. The length of 2nd and subsequent primaries was greater than that of the first primaries grown in the nest. A significant thickening of the rachis in some primaries was also found, indicating an increase in feather strength between juvenile and subsequent feathers. Patrones y duracion del cambio de las plumas en un Gavilan Azor hembra ( Accipter gentilis ) EXTRACTO. — Los patrones del cambio sucesivo (1983-1989) de las plumas necesarias para vuelos mayores (bastardas, primarias, secundarias, y de cola) han sido estudiadas en un Gavilan Azor ( Accipiter gentilis) hembra, cautiva pero silvestremente incubada y criada. En lo posible, se ha registrado la fecha de la caida de cada pluma, asi como se han medido el largo y grosor del canon. La variacion observada en el numero total de plumas cambiadas cada ano se debio a las diferencias en el numero de plumas de cola y secundarias caidas anualmente. Con la excepcion de 1985, todas las plumas bastardas y las primarias fueron reem- plazadas anualmente. El patron de caida de las plumas secundarias y de cola en un ano, dependio mayormente del patron de cambio del ano anterior. Aunque la duracion de la muda total de plumas decrecio desde un maximo de 141 dias en 1984 a un minimo de 98 dias en 1988, el comienzo de cada muda permanecio relativamente constante entre los 134 y los 145 dias del ano, contando desde enero. El largo de las segundas y subsecuentes plumas primarias fue mayor que el de las primeras primarias crecidas en el nido. Se encontro un aumento significativo del grosor del canon en algunas plumas primarias, lo que indica un aumento en el vigor de las plumas correspondiente con la edad del ave. [Traduccion de Eudoxio Paredes-Ruiz] The goshawk has recently been identified as one of a relatively small number of species in which the lead (Pb) and cadmium (Cd) contamination of cer- tain well defined feathers can be measured and used as an indicator of environmental contamination by heavy metals (Dietrich and Ellenberg 1986). It is therefore important to know when or how frequently certain feathers are molted so that the period of contamination can be determined. Similarly, radiotelemetry is being used, increas- ingly, in the study of wild raptors, including gos- hawks, and in these studies the radios are often at- tached to one or two tail feathers (Kenward 1978). In order to maximize the amount of data obtained from each radio-tagged bird it is important to attach the transmitters to those feathers which are likely to remain unmolted the longest. The need to under- stand molt patterns is therefore clear. The first seven annual molt patterns of a wild bred, captive female German goshawk are described in this paper. The bird was taken as a juvenile in the autumn of the year it hatched (1982) and there- fore the successive molts cover the change from ju- venile to full adult plumage. The first feathers, grown in the nest, did not start to be replaced until the spring of the following year (1983). I investigated the loss and replacement of the main flight feathers (alulas, primaries, secondaries and tail). This long- term study of an individual captive bird kept under conditions of natural light, temperature and excess 91 92 Christopher J. Reading Vol. 24, No. 4 9 B 54 34 56 8 9 8: 6 TAI L 6 5 4 3fe|l|l|2|3 4 5|6 LEFT WING 89 RIGHT WING 1983 un n mum n 1984 Mill II 1985 mu 1986 rmn 1987 I I I I I I I 1988 I I I II I I 1989 I I I II I I I ■ I ■ II I n n n mnn imriiiriiini 1 1 1 1 M i ■ M 1 1 1 ■ 1 1 1 1 1 1 n 1 1 nm ni iiiiiiiin rrmn I 1 1 1 1 1 ■ 1 1 ■ 1 1 ■ ii 1 1 1 ■ 1 1 1 1 1 1 1 1 1 1 1 ■ 1 1 n n Figure 1. Diagrammatic representation of the main flight feathers in the Goshawk. A = alulas, P = primaries, S = secondaries. The bar codes show which feathers were molted (open cells) or retained (solid cells) each year between 1983 and 1989. The stippled cells represent those feathers for which no molt data were obtained. food also enabled the changes in molt pattern be- tween years, and the interdependent effects of one molt on the next, to be investigated. Between each molt (October-March) the bird was regularly flown at quarry. It may therefore be considered as a base- line study of a bird, under minimal stress during molt, to which studies of wild birds under many different stresses (food shortage, breeding, weather etc.) can be compared. Although molt data of this type are much more difficult to obtain from studies of wild birds, some limited success was achieved by Briill (1984) studying wild German goshawks be- tween 1950 and 1959. Unfortunately his data are incomplete. In contrast the molt data obtained in this study are complete for each of the four main flight feather types for either 6 or 7 years and should be viewed as being complementary to those of Briill (1984). Methods With the exception of the first year of the study (1983) the date on which each alula, primary, secondary and tail feather was molted was recorded. In 1983 only the primary and tail feathers were recovered and the date on which each was molted was not recorded. The thickness of the rachis of each feather was measured to an accuracy of 0.01 mm using vernier calipers. With the blade of the feather in the horizontal plane, the thick- ness of the rachis was taken in the vertical plane at a point 5 mm proximal to the point where the feather barbs attach to the rachis. Flattened rachis length was taken as a mea- sure of feather length and was recorded to an accuracy of 1.0 mm. It was only possible to measure those feathers that were found intact. All the feathers were numbered in the following way Winter 1990 Goshawk Molt Pattern 93 89 - 88 • 87 - 86 - 85 - 84 - 89 88 87 86 85 84 CC < LU >- 88 87 86 85 84 130 I 10 May 150 170 I 19 Jun 190 210 I 29 July 230 250 I 7 Sept DAY OF THE YEAR I 1 1- - 1 - 1 1 . 1 1 1 a 1 1 I p 20 20 20 20 1 8 ,20 1 1 1 1 l 1 1 1 1 I 1 » 1 s 24 21 19 29 9 3 1 -1 - i 1 1 - 1- 1 I I 1 1 I t T 12 6 12 9 6 6 1 1 1 1 1 1 1 I l_ 1 1 > 270 290 I 1 7 Oct Figure 2. Start and duration of the molt in each of the main flight feather types in the Goshawk for each year from 1983 to 1989. A = alulas, P = primaries, S = secondaries, T = tail. The figures show the number of each feather type molted each year. The alulas were numbered from the outermost inwards (1-3); the primaries from the innermost outwards (1-10); the secondaries from the outermost inwards (1-16) and the tail from the innermost outwards (1-6). The date on which a particular feather was molted was recorded as the nth day of the year with January 1st being day 1. This convention overcame the potential problem of the extra day during leap years (1984, 1988). The data were analysed using either Student’s £-test or linear regression analysis. Results A summary showing which feathers were molted during each of the years for the period from 1983- 1989 is given in Fig. 1. Similarly the overall duration of each molt for each feather type, and the number of feathers shed, are shown in Fig. 2. Each of the four feather types collected will be dealt with sep- arately. Alulas. All three alulas on both wings were molted every year, and in sequence, beginning with feather 3 and ending with feather 1. Although the interval between equivalent feathers on each wing being molted ranged from 0-20 days there was an overall reduction in the length of the interval between suc- cessive feathers being shed between 1984 and 1986 (Table 1). Between 1986 and 1989 this interval re- mained relatively constant. The starting date of the alula molt changed no- ticeably between 1985 and 1986. In 1984 and 1985 the molt began on days 178 and 177, respectively, whilst between 1986 and 1989 the molt started be- tween days 151 and 156, approximately 24 days earlier. Concurrent with this earlier start there was a reduction in the overall time taken for all six feath- ers to be shed from between 86 and 71 days in 1984 94 Christopher J. Reading Vol. 24, No. 4 Table 1 . Mean number of days taken to molt each feath- er each year (1984-89) in a female goshawk. Year Alula Primary Sec- ond- ary Tail Over- all 1984 14.3 14.1 9.3 5.8 2.6 1985 11.8 13.2 11.3 7.2 3.1 1986 9.8 7.3 (6.6) a 6.9 6.8 1.7 1987 9.2 8.9 11.0 7.1 1.8 1988 9.8 9.1 8.4 3.2 1.9 1989 9.5 9.1 8.3 6.8 1.7 (1984-85) Mean 13.1 13.7 10.3 6.5 2.9 SD 1.8 0.6 1.4 1.0 0.4 N 2 2 2 2 2 (1986-89) Mean 9.6 8.6 8.7 6.0 1.8 SD 0.3 0.9 1.7 1.9 0.1 N 4 4 4 4 4 a In 1986 primary 10, which was not shed in 1985, was shed out of sequence thus resulting in an artificially low mean (6.6). The corrected mean was calculated excluding the feather 10 pair. and 1985 to between 55 and 59 days for the period 1986 to 1989 (Fig. 2). Primaries. All ten primaries on each wing were molted every year except 1985 when primary 10 on each wing was retained (Fig. 3). With the exception of 1986 they were also molted in sequence, starting with feather 1 and ending with feather 10. Although feather 10 was molted out of sequence, between feathers 7 and 8, in 1986 the intervals between the loss of feathers 1 to 9 were not affected. Equivalent primaries on each wing were usually molted on the same day or within 5 days of each other. Unlike the pattern shown by the alula molt there was no clear change, between 1984 and 1989, in the starting date of the primary molt (days 134-145). There was, however, a significant reduction it = 7.05, df = 4, P < 0.01), from 1985 to 1986, in the interval between the shedding of successive feathers from a mean of 13.7 days in 1984/1985 to 8.6 days between 1986 and 1989 (Table 1) resulting in a large reduction in the overall time taken to complete the primary molt (Fig. 2). Secondaries. The total number of secondary feathers molted each year was unpredictable and varied considerably from a minimum of 9 in 1985 to a maximum of 29 in 1986 (Fig. 1). The start of the secondary molt was also more variable than that of the alulas or primaries ranging between day 134 in 1987 and day 159 in 1985 (Fig. 2). No significant change was found in the interval between the molting of successive feathers during the course of the study it = 1.13, df = 4, P > 0.10; Table 1). There is some evidence, however, to suggest that feathers retained in a particular year were more likely to be molted the following year than those that were previously molted. In only one year, 1986, was an almost full DAY OF THE YEAR Figure 3. The mean molt date for each pair of primary feathers between 1984 and 1989. • = 1984, O = 1985, ■ = 1986, □ = 1987, ▲ = 1988, A = 1989. The No. 10 primary, molted out of sequence in 1986, is the primary that was not shed in 1985. Winter 1990 Goshawk Molt Pattern 95 DAY OF THE YEAR Figure 4. Start and duration of the complete molt for each year between 1984 and 1989. The figures show the total number of feathers molted each year. complement of secondaries molted and replaced (29/ 32 feathers). The sequence in which the feathers were molted varied between years and was, to some extent, also dependent on whether or not a particular feather had been molted the previous year. However, by scoring each molted feather, each year, according to when it was molted and by combining the scores for equivalent feathers on each wing a pattern was found. Overall, the secondary molt followed the sequence: feather 6, 15, 5, 3, 11, 10, 2, 7/14, 12, 13, 9, 8, 4, 1, 16. This pattern suggested the presence of four molt centers on each wing at feathers 3, 6, 11 and 15. Tail. As with the secondaries, the total number of feathers molted each year was largely unpredict- able ranging between 6 and the full complement of 12. There was, however, a tendency for feathers that had not been shed in one year to be shed the following year (e.g., 1984/85; Fig. 1). The starting date of the tail molt got progressively earlier between 1984 (day 192) and 1987 (day 148), and then stabilized at between days 148 and 154 over the last three years with the biggest difference occurring between 1985 and 1986 (Fig. 2). The sequence in which the feathers were molted varied between years and was dependent on whether or not a particular feather had been shed the previous year. However, by using the method already ex- plained for secondaries the overall molt pattern for either side of the tail followed the sequence: feather 5, 1, 2, 6, 3, 4. Exceptions to this usually started with the sequence: feather 1, 6, 3. Overall Timing and Duration of Molt. The total number of major flight feathers molted each year varied between 39 in 1985 and 64 in 1986, the full complement being 70. Of these the number of alulas and primaries molted annually remained al- most constant whilst the variation in the number of secondaries and tail feathers that were molted ac- counted for the annual differences in the total num- ber of feathers molted. The start of the annual molt occurred between days 134 and 145. The apparent trend towards an earlier start date with increasing age was not sig- nificant (r 2 = 0.47, N = 6, P > 0.05). The time taken to complete the annual molt (Fig. 4) decreased progressively from 141 days in 1984 (54 feathers) to a minimum of 98 days in 1988 (53 feathers). The largest difference between the mean time to molt each feather occurred between 1985 (3.1 days) and 1986 (1.7 days). Rachis Length and Thickness. Because the 1983 alulas and secondaries, though molted, were not col- lected it was impossible to compare their rachis lengths with those of feathers from subsequent molts. Similarly, though collected, the 1983 tail feathers all had damage to their tips and therefore it was im- possible to compare rachis length in these feathers during the change from juvenile to first adult plum- age. It was possible, however, to compare rachis lengths and thicknesses for primary feathers from all seven years. Both an increase in rachis length between first and subsequent years (Table 2) and an increase in rachis thickness were found (Table 3). The thickening of the rachis between 1983 and 1984 did not occur in all the primaries equally, being most noticeable in feathers 1-5. An analysis of tail feather rachis thickness showed that a significant thickening was only detected in 96 Christopher J. Reading Vol. 24, No. 4 Table 2. Comparison of mean feather/rachis length (mm) between 1st and subsequent molts for equivalent primaries on the left and right wing of a female goshawk. Feather Type Feather Number 1 2 3 4 5 6 7 8 9 10 Primary (1st) Mean 211 214 222 233 239 265 — 240 226 — SD 0 1.4 6.4 11.3 — 0 — 0 2.8 — N 2 2 2 2 1 2 — 2 2 — Primary (2nd+) Mean 214 222 230 245 275 289 229 159 SD 1.4 1.2 1.9 2.2 4.7 8.0 — — 2.9 2.3 N 11 12 12 12 7 a 4 — — 5 9 a Primaries 5-10 being both the outermost and including the largest feathers (5-8) were most prone to damage often making it impossible to measure their length. feathers 3 (t = 2.27 , df = 10, P = 0.05) and 6 (t = 5.31, df = 6, P = 0.002) between first and subsequent molts. Discussion A number of conclusions can be drawn from this study of molting in a captive female goshawk. Although all the alula and primary feathers were replaced during the first molt, some tail and possibly some secondary feathers remained and were not shed until the second molt. With the exception of one year, all the alulas and primaries were replaced ev- ery year. Previous studies of wild sparrowhawks in Scotland (Newton and Marquiss 1982) and both wild sparrowhawks and wild goshawks in the Federal Republic of Germany and the Netherlands (Briill 1984, Opdam and Miiskens 1976) have demonstrat- ed similar molt patterns for the primaries but none looked at the smaller alulas. Complete replacement of the secondary and tail feathers did not occur an- nually. Instead, it was found that in these two feather types the molt pattern in any particular year was determined to some extent by the molt pattern of the previous year. Table 3. Mean rachis thickness (mm) of 2nd and subsequent feathers compared with that of 1st feathers. Means calculated using the combined data for equivalent feathers from both wings. Feather Number Type 1 2 3 4 5 6 7 8 9 10 Alula (1st) Mean — — — Alula (2nd+) Mean 3.27 2.89 2.37 SD 0.07 0.09 0.06 N 12 12 12 Primary (1st) Mean 3.71 4.04 4.32 4.53 4.81 5.04 4.69 4.47 3.88 3.28 SD 0.04 0.06 0.08 0.13 0.13 0.06 0.01 0.04 0.06 0.03 N 2 2 2 2 2 2 2 2 2 2 Primary (2nd+) Mean 4.40 4.58 4.80 5.03 5.28 5.26 5.00 4.60 4.01 3.40 SD 0.10 0.09 0.05 0.07 0.11 0.15 0.15 0.13 0.13 0.10 N 12 12 12 12 12 12 12 12 12 10 Tail (1st) Mean 4.89 5.06 5.01 5.15 5.16 5.03 SD 0.27 0.06 0.01 0.01 0.08 0.04 N 2 2 2 2 2 2 Tail (2nd+) Mean 4.80 5.02 5.16 5.19 5.18 5.24 SD 0.17 0.16 0.09 0.08 0.09 0.05 N 10 7 10 8 7 6 Winter 1990 Goshawk Molt Pattern 97 The increases in feather length between juveniles and adults, seen by Opdam and Miiskens (1976) in wild goshawks, were also found in this study. In- creases in the thickness of the rachis of some primary (feathers 1-5) and tail (feathers 3 & 6) feathers were also detected. This occurred between the juvenile and first adult feathers and suggests an actual strengthening of the feathers rather than propor- tional growth. The reasons for this are unclear but may be a response to differences in food availability. Food shortage as a chick may well result in reduced feather growth, whereas in captivity feather growth would not have been constrained in this way, re- sulting in stronger feathers. In addition, and con- sistent with this hypothesis, is the fact that the feath- ers of the first plumage are all grown simultaneously whereas feather replacement following a molt is staggered. The other main changes occurred between the third and fourth molts. The fourth molt of both the alula and tail feathers started earlier and the du- ration of the alula and primary molts shortened sig- nificantly. In contrast to the findings of Newton and Marquiss (1982) for the sparrowhawk, the onset of the molt remained relatively constant during the sev- en years of the study and may even have occurred a little earlier as the bird aged. In a comparison between the molt patterns ob- served in wild German goshawks (1950-1959) by Briill (1984) and the captive German goshawk in this study two clear differences emerge. First, the start of the primary molt in the wild birds occurred, on average, about 50 days earlier than in the captive bird but proceeded at approximately the same rate (Briill’s data for the end of the primary molt are poor). Second, the interval between successive pri- maries being shed in the wild birds was initially very short: 2-7 days for primaries 1-4; but then increased to 16 days between primaries 4 and 5. Briill gives no reliable data for primaries 6-10. No such sudden increase in the molt interval was found in the captive bird. These differences may illustrate both the hor- monal control of molting and its known correlation with egg laying (Briill 1984) which in Germany usually occurs in late March or early April (approx day 90), and the effect of food stress on molting and feather growth. Once the female has eggs, she is dependent on the male for food. With respect to radio-tagging wild goshawks, the tail feather molt pattern determined in this study suggests 1) that feathers 2/3 or 3/4 should be used for the attachment of transmitters, thus largely sup- porting Kenward’s finding (1978) and 2) that new feathers should be used in preference to old ones since these have the highest chance of remaining unmolted the following year. This was a study of a single captive goshawk and therefore the data should be interpreted with care, particularly when extrapolating to the wild situa- tion. Nevertheless, the data are relevant for those studying wild goshawks/raptors in that they show both the changes that occur in molt pattern with age and the relationship between successive molt pat- terns. Acknowledgments The goshawk used in this study was imported into the UK from West Germany, for the purposes of falconry, under DoE license BP 000056 and held under DoE license UK 60834. I wish to thank R.E. Kenward for his com- ments on the manuscript. Literature Cited Brull, H. 1984. Das Leben europaischer Greifvogel 4th edition. Gustav Fischer, Stuttgart, F.R.G. Dietrich, J. and H. Ellenberg. 1986. Habicht-Mau- serfedern als hochintegrierende, standardisierte Um- weltproben. Verhandlungen der Gesellschaft fur Okologie (Hohenheim 1984), Band XIV, 413-426. Kenward, R.E. 1978. Radio transmitters tail-mounted on hawks. Ornis Scand. 9:220-223. Newton, I. and M. Marquiss. 1982. Moult in the sparrowhawk. Ardea 70:163-172. Opdam, P. and G. Muskens. 1976. Use of shed feathers in population studies of Accipiter hawks (Aves, Accip- itriformes, Accipitridae). Beaufortia 24:55-62. Received 25 July 1990; accepted 9 September 1990 J. Raptor Res. 24(4):98-106 © 1990 The Raptor Research Foundation, Inc. RAPTOR ROAD SURVEYS IN SOUTH AMERICA David H. Ellis Patuxent Wildlife Research Center, U.S. Fish and Wildlife Service, Laurel, MD 20708 Richard L. Glinski Arizona Game and Fish Department, 3333 West Greenway Road, Phoenix, AZ 85023 Dwight G. Smith Southern Connecticut State University, 501 Crescent Avenue, New Haven, CT 06515 Abstract. — Twenty-six (23 traveling and three point) raptor roadside surveys were conducted during a 29,000 km expedition through nine nations of South America. During roadside surveys, we tallied 41 of the 87 (47%) diurnal raptor species (including vultures) that occur in South America. The number of species observed per route varied from 17 in the wet savanna of Venezuela to only two species recorded in the harsh Atacama Desert and the dry montane grasslands of Chile and Peru. Raptor density (non- vultures) varied from 1 per 67 km in the Atacama Desert to more than 1 per km in agricultural areas where caracaras and other species that utilize disturbed habitats were common. Responses of raptor communities to deforestation and other habitat disturbances are discussed. While certain habitat modi- fications potentially increase raptor abundance and diversity, the alteration of primary forest has the opposite effect, at least on diversity. Indagaciones sobre aves raptoras, hechas en carreteras de America del Sur Extracto. — 26 (23 viajando y 3 estacionarias) ispecciones de aves raptoras, a lo largo de una carretera, fueron realizadas durante una expedition de 29,000 km a traves de 9 naciones de America del Sur. Durante las inspecciones de carretera hemos contado 41 de las 87 (47%) especies raptoras diurnas (incluyendo buitres) que se encuentran en America del Sur. El numero de las especies observadas en cada ruta vario de 17 en las praderas lluviosas de Venezuela, hasta solo 2 especies registradas en el desierto de Atacama y las secas lomas de Chile y Peru. La densidad de aves raptoras (no buitres) vario desde 1 por 67 km en el desierto de Atacama, hasta mas de 1 por km en areas agricolas, donde eran comunes las caracaras y otras especies que utilizan habitats alterados. Se discuten los resultados de la deforestation y otras alteraciones del habitat en las comunidades de aves raptoras. Mientras que ciertas modificaciones del habitat potencialmente aumentan la abundancia y diversidad de aves raptoras, la alteration de florestas naturales tiene el efecto opuesto, a lo menos en la diversidad. [Traduction de Eudoxio Paredes-Ruiz] Raptor survey methods have been reviewed by Fuller and Mosher (1981, 1987). Although the lim- itations and biases inherent in the road counts are well known (Verner 1985, Millsap and LeFranc 1988), road surveys unfortunately are the only prac- tical means now available for rapidly assessing rap- tor distribution and, to a degree, abundance over large areas. Roadside surveys have been used to de- termine species composition and to estimate relative abundance for diurnal raptor communities in Africa (Rowan 1964, Cade 1969), Europe (Meyburg 1973, Saurola 1976), North America (Nice 1934, Craig- head and Craighead 1954, Enderson 1965, Johnson and Enderson 1972, Wofhnden and Murphy 1977, and many others), and, to a very limited degree, in Latin America (Reichholf 1974, Ellis et al. 1983, Albuquerque et al. 1986, Wotzkow and Wiley 1988). Using road counts and point counts, we surveyed diurnal raptors in nine South American nations and related both species composition and relative abun- dance to gross features of the habitats. Methods The expedition was conducted from 1 2 January through 23 April 1979. During this time, we established 23 road- side survey routes and two point count locations (Fig. 1). Abbreviated descriptions of survey locations, habitat types, and other salient features are presented in Table 1. De- scriptions of the physical and biotic characteristics of the survey routes and photographs of survey route boundaries are available from the senior author. To facilitate relocating each route in future surveys, 98 Winter 1990 Raptor Surveys in South America 99 Table 1. Descriptions of South American raptor counts in 1979. No. a Nation: Zone Length (km) Date Day/Mo. Time Start Habitat TYPE b Disturbance 0 1 Venezuela: NC 90 26.01 0645 VDFor (montane), OFor (coastal swamp), AgC Low-Medium 2 Venezuela: NC 81 29.01 1615 DSav, OSav Low 3 Venezuela: NC 113 30.01 0910 DSav, OSav, AgP Medium 4 Venezuela: NE 80 31.01 1355 OFor, VDFor (montane), VDFor (lowland) Medium 5 Venezuela: E 96 01.02 1130 DFor (montane) Low-Medium 6 Brazil: N 119 03.02 0835 VDFor & SecG (road swath) Low-Medium 7 Brazil: N 119 05.02 0722 VDFor & SecG (road swath) Medium 8 Brazil: WC 144 13.02 0840 D -VDFor & AgC (road swath) Medium 9 Brazil: SC 124 24.02 1240 OSav & Wd, AgC & AgP Medium 10 Brazil: S 131 28.02 0825 Wd, AgC Medium-High 11 Paraguay: S 111 01.03 1050 Wd, AgC High 12 Argentina: NE 76 03.03 0805 OSav & RGFor Low 13 Argentina: SE 150 07.03 0945 AgC & AgP, DScrub High 14 Argentina: S 178 08.03 0920 DScrub Low 15 Argentina: S 162 12.03 0745 DFor, DScrub Low 16 Argentina: S 77 12.03 1445 DScrub Low 17 Argentina: SE 169 16.03 1000 DScrub Low 18 Chile: C 84 21.03 1100 AgC & AgP High 19 Chile: NC 134 23.03 1000 SDes Low 20 Peru: S 202 31.03 0715 MScrub, SDes Low 21 Ecuador: N 70 07.04 0655 MScrub, OFor, RGFor, AgC Medium 22 Venezuela: W 79 13.04 1035 DSav, RGFor, AgP, OSav Medium 23 Venezuela: N 116 15.04 0655 OSav, AgP Medium A Venezuela: NC N.A. 28.01 1050-1150 VDFor (montane), same location as C Low B Venezuela: NC N.A. 31.01 1535-1805 OSav (formerly DSav) Medium-High C Venezuela: NC N.A. 19.04 0835-0935 VDFor (montane), same location as A Low a Count type: Road Count (1-23), Point Count (A-C). b Habitat type abbreviations and overstory canopy cover (oscc) classes: Desert (Des), Semi-Desert (SDes), Desert Scrub (DScrub), and Montane Scrub (MScrub) 0-5% oscc; Open Savanna (OSav) 5-20% oscc; Dense Savanna (DSav) 20-50% oscc; Open Forest (OFor) 50- 70% oscc, Dense Forest (DFor) 70-90% oscc; Very Dense Forest (VDFor) 90-100% oscc; Second Growth (SecG), Woodlots (Wd), Agricultural Pastures (AgP), Agricultural Croplands (AgC), Riparian Gallery Forest (RGFor). c A gross evaluation of the degree of alteration of habitat from pristine form. where practical, we chose distinctive topographic features and road junctions to define beginning and ending points of survey routes. We located routes in one habitat type, or in a uniform interspersion of two habitat types. The imposition of these parameters resulted in transects of varying lengths. We also attempted to limit surveys to morning hours (five exceptions) during fair, calm weather. Driving speeds were 70-80 km/hr on paved roads and 50 km/hr or less on dirt roads, although road conditions were occasionally too variable to permit a uniform driving speed. Roadside counts (a form of Verner’s 1985 line transects without distance estimates) were conducted by two ex- perienced observers; a driver and a record keeper, both in the front seat of a Toyota Land Cruiser. A third person acted as a recorder for some of the routes. We identified and tallied most raptors while we were in transit. Occa- sionally, we stopped to confirm identification of an indi- vidual bird; raptors detected during these stops were not tallied unless, in our judgment, they would have been noted during uninterrupted travel. Although the new world vul- tures (family Cathartidae) are not now considered mem- bers of Order Falconiformes (Rea 1983), we included them in our counts. Along many survey routes, however, vul- tures were so abundant that to count them all would have diverted our attention inordinately from our search for true raptors. Therefore, only the first 20 individuals of the common vulture species were counted. Only rarely did we identify and tally raptors further than 1 km from the road. Three point counts (a form of Verner’s 1985 point counts Table 2. Raptor count summary, South America, 1979. 100 David H. Ellis et al. Vol. 24, No. 4 T“< u pi CM t-h u Pi cr> CM CM O O A SO O CM O CM A u e i o CM A o o p 4 LT) r- o CM A <0 g o Q pi oo O r- u et! M3 o O o CM A CM CO O O t-h t— i CM CM A A CM co o CM A to 00 LO o p£ cO NO M- U eel O CM A CM -<*■ CM co o p i o CM A o CM A CM u oi o CM A o CM A o pi o CM A Ln r- lO Table 2. Continued. Winter 1990 Raptor Surveys in South America 101 £ § U Q C? Pi co v pi CM U Pi o Pi u Pi Os u Pi oo u pi u Pi SO u Pi LO u Pi M- o Pi co u Pi cm o Pi u Pi V) w 5 a (X, CO o CM cm lo O o so CM \0 CM 00 co r~~ CM Os CM CM t— CO LO M- CM O' CM cO CM SO CO cO CM cO CM CO so r-~ CM OS r-~ 3 to 3 3 3 ■ H k a to • ^2 k £ t-o 3 g 3 a “2 Q M „ ■Ki 3 C -O O <3 ^ k ^ k flq k "*-3 S? c I tour) -£s to -2 ak« 3 ~o fc -q «o 3 o o o o to to to to “+~i •*-* 'Ka +»j 3 3 3 3 Qq q oq Qq § o -S' o k o to 3 s 3 g <-5 ■k «3 tj c to Q- C <3 3 VI -2 .§ £ flj < ■ i-^ -c C o in H to to <43 ■V-i ^ K b. S3 S5 « S3 S3 -3 -3 -fit -3 ~3 ■*<» "*<> Q S3 S3 S3 S3 o o c o o o to s r o <3 in' 1 ^ S 3 Sq 2 .3 R -g 3 S -g -Sr^ 3 tie g 2 ^ ;S Hi 3 3 ■ S c 3 X 2 to * C«a -c -I s © 5 Hi 3 ■*-3 Hi • o 3 3 Hi 3 V 3 u 3 cw ?v 3 © Hi HJ ■«, 3 ■' 3 3 Caq sB t*q -c> S3 © *to C^J >«© to to 2 to S3 to 3 50 ^ to ^ •£ 51 © S 3 2 to 3 K o © §o C to g ^ =s -s ■§ 3 3 ■ v hi 3 Oh ■“ H, . « F 3 a R r vi to ■*«* S3 S3 ^ ~ 15 * Sp *3 P O © G • «" ^ to 'to N 3 ~ “ Si 'to _^3 3 K «o 3 O U O ^ vj oq 03 Heterospizias meridionalis Busarellus nigricollis Geranoaetus melanoleucus Table 2. Continued. Winter 1990 Raptor Surveys in South America 103 ✓ H Z D O o H Z o Plh U O Oh 0Q o pH < o Oh co cO CM CM r-* o r- SO T — i O' CM t-~. oo r-i 00 tJ" cO r- oo d CM CM O co CM O (A co oo O' CM h O' N *-> M" 00 C~- w CN CO ^ O CM CM CM u (A LO CO CM O' 00 rO CM O ^ O N h w O CM r- CM o tA CM moo o o CM U (A co CM cO O CO CO t"-’ SO H Z o U a < o O' r — < o oo T—< O CO 00 SO CM oo o m C-; sd ✓-s O' CO CM CM SO lO so CM O O h- u (A r- SO O' CM r- CM — X lO t~h r- on r- r- ^ "t ao ^ © o T— < 0 0 1 CM _ LO O' ^ r- o 3 LO Tt M" O lo LO T — < u tA T-* SO cO CO co co CM ’M’ 'M" O- O' LO M- ' i— 1 M- CO r- od u tA lO O' O' CO 00 rO ld i - m o m- w W u w Ph CO £ 3 o 3 3 -O g J3 -§ 3 3 ’ r3 -2 d ^ $ X.g <3 <3 O t-T, O c .£ £ 3 1 Jg 5 3 - §> fTl -a ■ta S -2 g s o d K ^ -o <3 'Cl, n tq o CO OH 13 13 +j -*-* o o H H u 3 3 O' 3 > ■ G O G 3 > i G o G -G 3 s JA >> u u G G m -o G 2 3 CG -Q Q < 104 David H. Ellis et al. Vol. 24, No. 4 Figure 1. South America, showing count locations and route of travel. without distance estimation) were conducted to evaluate use of a stationary watch in finding and counting raptors and also for comparison with traveling counts. For two of these point counts, an elevated ridge was chosen to afford maximum visual advantage. The point counts were con- ducted in fair, calm weather. Two observers equipped with 7 x 50 binoculars and a 20 x spotting scope counted all raptors seen during at least a 60-min observation period. Raptor abundance for each survey route was calculated as the number of km per observed individual (excluding vultures). A gross measure of raptor diversity was obtained using the Shannon diversity index (Zar 1974). Because various species of raptors differ in observability in some habitat types and a single species is often more detectable in more open habitats than in forest, these indices should be used with caution. Results and Discussion Raptor diversity and abundance varied greatly among survey routes. The greatest variety (17 spe- cies) and the highest species diversity index (H' = 1.020) were found in a region of mixed riparian gallery forests, savanna, and pastures in western Venezuela. By contrast, only two species (present in low densities) were found in the Atacama Desert and the high elevation montane scrub habitats of Peru and Ecuador. In other, less harsh environ- ments, raptor density typically varied inversely with raptor diversity. Raptors were most abundant in modified or open habitats and in association with human activities such as animal husbandry. In such areas, vultures and Crested Caracaras ( Polyborus plancus) predominated. For example, a Chilean sur- vey route (No. 18, Table 2) through cropland and pasture habitat had the greatest concentration of rap- tors observed (0.65 km per individual), but only two species were represented (diversity index H' = 0.163). Similarly, certain raptors were observed in dense concentrations in agricultural habitats in Venezuela, Paraguay, Brazil, and Peru. Surveys with high rap- tor variety and abundance were typically in mixed habitat, ranging from highly stratified dense forest to relatively open savannas or second growth forest. Results of the three point counts also varied great- ly. Eight species were recorded on point counts B and C, but only two species on point count A, al- though A and C were both in dense montane forests. Three problems were encountered for point counts. First, we found it difficult to identify raptors readily observed at distances greater than 1 km. Second, we could not accurately tally the total number of indi- viduals because birds often repeatedly soared through the observation zone. Finally, it proved impractical to observe birds in dense forest. Because of these difficulties, we de-emphasized point counts in favor of road surveys. Roadside raptor surveys proved efficient in de- tecting and counting common raptors in non-forested habitats. For all counts combined, we recorded 41 (47%) of the 87 diurnal raptor species that occur in South America. This included all of the Cathartid vultures and most other species that soar regularly. Raptors of the deep forest are much more difficult to observe and our roadside counts in forest habitats certainly underestimate both diversity and abun- dance of raptor communities. Most of the raptors counted on routes through rainforest habitats in Venezuela and Brazil (RC 5-8) also normally occur in secondary forest habitat and were probably seen because of their association with the secondary growth forest of the roadside swath. By contrast, many spe- cies of obligate forest raptors (e.g., Harpia, Morph- nus, Micrastur ) were never recorded during counts nor seen while traveling between survey locations. We suggest that occurrence and abundance estimates Winter 1990 Raptor Surveys in South America 105 for these will ultimately come from species-specific counting techniques such as listening to early morn- ing calls or mark and recapture methods used for forest falcons ( Micrastur ; Klein and Bierregaard 1988). Other species will require mist nets, radio- telemetry, systematic searches, and other more time- consuming methods (Fuller and Mosher 1981, 1987, Thiollay 1989, Vannini 1989). As primary forests are converted to open agri- cultural areas, forest-dwelling raptors, especially the large eagles, disappear while migrants and savanna- dwelling species increase (Harris 1984). In southern Brazil, Reichholf (1974) found that as forests were cut and replaced by agricultural and grazing lands, the diversity of highly rapacious species decreased while scavengers (i.e., vultures and caracaras) in- creased. Thiollay (1985) reported similar changes in raptor diversity and abundance in a survey of seven habitat classes in the southern Ivory Coast of Africa and four habitat classes in the neotropics. By contrast, in arid regions, the introduction of irriga- tion may benefit some raptors. Fields and croplands provide foraging areas; trees provide nesting and roosting habitat. Sheep and cattle ranching can also result in local concentrations of caracaras and other scavengers that forage on carcasses and offal (e.g., road counts RC3, RC15, and RC17-18, Table 2). Conclusions and Summary Our survey work, and the studies of Reichholf (1974) and others, demonstrate the feasibility of roadside counts in estimating relative abundance and determining species composition of diurnal raptor communities in relatively open neotropical habitats. The efficiency of roadside surveys is, however, ex- tremely limited for owls (we detected only one spe- cies) and for other raptors that live primarily beneath the forest canopy. Our results suggest that total raptor diversity and relative abundance were frequently inversely cor- related. The most diverse raptor species assemblages observed during roadside counts were in mixed sa- vanna and dense forest habitats. By contrast, the greatest number of scavengers were found in asso- ciation with agricultural fields and rangelands. In addition, while diversity of forest raptor communi- ties often decreases as natural habitats are modified, agricultural development in arid environments tends to increase primary productivity, biotic diversity, and food supply. These changes, in turn, result in local increases in raptor abundance and diversity. Such localized increases, however, fail to compensate for the loss of raptor diversity that is the direct result of widespread neotropical deforestation. Acknowledgments Safari Club International financed travel expenses. The Institute for Raptor Studies provided field equipment. James R. Hunt participated as a recorder and interpreter from Caracas to Manaus. Cathy Ellis prepared the figures and assisted in manuscript preparation. The Patuxent Wildlife Research Center paid for some travel expenses We also express appreciation to the people and govern- mental agencies that hosted our visits. Literature Cited Albuquerque, J.L.B., A.J. Witech and A.M. Aldous 1986. A roadside count of diurnal raptors in Rio Grande do Sul, Brazil. Birds of Prey Bull. 3:82-87. Cade, T.J. 1969. The status of the peregrine and other Falconiformes in Africa. Pages 289-321 in J.J. Hickey [Ed.], Peregrine Falcon populations: their biology and decline. University of Wisconsin Press, Madison, WI Craighead, J.J. and F.C. Craighead, Jr. 1954. Hawks, owls and wildlife. Stackpole, Harrisburg, PA Ellis, D.H., R.L. Glinski, J.G. Goodwin, Jr. and W.H. Whaley. 1983. New World vulture counts in Mex- ico, Central America, and South America. Pages 124- 132 in S.R. Wilbur and J.A. Jackson [Eds.], Vulture biology and management. University of California Press, Berkeley, CA. Enderson, J.H. 1965. Roadside raptor counts in Col- orado. Wilson Bull. 77:82-83. Fuller, M.R. and J.A. Mosher. 1981. Methods of detecting and counting raptors: a review. Studies in Avian Biology 6:235-246. AND . 1987. Raptor survey techniques. National Wildl. Fed. Scientific and Tech. Series 10:37- 65. Harris, L.D. 1984. The fragmented forest. University of Chicago Press, Chicago, IL. Johnson, D. and J.J. Enderson. 1972. Roadside rap- tor census in Colorado — winter 1971-1972. Wilson Bull 84:489-490. Klein, B.C. and R.O. Beirregaard. 1988. Capture and telemetry techniques for the Lined Forest-falcon {Micrastur gilvicollis) . J. Raptor Res. 22:29. Meyburg, B.-U. 1973. Observations sur l’abondance relative des rapaces (Falconiformes) dans le nord et l’ouest de l’Espagne. Ardeola 19:129-150. Millsap, B.A. and M.N. LeFranc, Jr. 1988. Road transect counts for raptors: how reliable are they? J Raptor Res. 22:8-16. Nice, M.M. 1934. A hawk census from Arizona to Massachusetts. Wilson Bull. 46:93-95. Rea, A.M. 1983. Cathartid affinities: a brief overview. Pages 26-54 in S.R. Wilbur and J.A. Jackson [Eds.], 106 David H. Ellis et al. Vol. 24, No. 4 Vulture biology and management. University of Cal- ifornia Press, Berkeley, CA. Reichholf, J. 1974. Artenreichtum, Haufigkeit und Diversitat der Greifvogel in einigen Gebieten von Sii- damerika. /. Ornithol. 115:381-397. Rowan, M.K. 1964. Relative abundance of raptorial birds in the Cape Province. Ostrich 35:224-227. SAUROLA, P. 1976. Finnish raptor censuses. Orn is Fenn. 53:135-139. Thiollay, J.-M. 1985. Composition of Falconiform communities along successional gradients from pri- mary rain forest to secondary habitats. 1CBP Tech. Publ. 5:181-190. , 1 989. Censusing of diurnal raptors in a primary rain forest: comparative methods and species detect- ability. J. Raintor Res. 23:72-84. Vannini, J.P. 1989. Neotropical raptors and defores- tation: notes on diurnal raptors in Finca El Faro, Quetzaltenango, Guatemala. /. Raptor Res. 23:27-38 Verner, J. 1985. Assessment of counting techniques. Pages 247-302 in R.F. Johnson [Ed.], Current orni- thology. Vol. 2. Plenum Press, New York. Woffinden, N.D. and J.R. Murphy. 1977. A roadside raptor census in the eastern Great Basin — 1973-1974 Raptor Res. 11:62-66. WOTZKOW, C. AND J.W. WlLEY. 1988. Turkey Vulture surveys in Cuba. J. Raptor Res. 22:3-7. Zar, J.H. 1974. Biostatistical analysis. Prentice-Hall, Englewood Cliffs, NJ. Received 6 February 1990; accepted 10 August 1990 J. Raptor Res. 24(4):107-112 © 1990 The Raptor Research Foundation, Inc. RESPONSE OF NORTHERN GOSHAWKS TO TAPED CONSPECIFIC AND GREAT HORNED OWL GALLS J. Timothy Kimmel and Richard H. Yahner School of Forest Resources, Pennsylvania State University, University Park, PA 76802 Abstract. — We compared responses of Northern Goshawks (Accipiter gentilis) to conspecific “kakking” and Great Horned Owl ( Bubo virginianus ) “hooting” calls during the 1989 breeding season in Pennsyl- vania. Calls were played 150 and 300 m from active goshawk nests during nestling (7 nests, N = 27 trials) and fledgling (7 nests, N = 28 trials) periods. Five nests were tested during both nestling and fledgling periods. Response rate of goshawks to calls played at 150 m was highest for conspecific calls during the nestling period (0.71) and lowest for owl calls during the fledgling period (0.14). Goshawk response rate to conspecific calls at 300 m during nestling and fledgling periods was 0.29 each, and no response was detected to owl calls at 300 m during either period. Response rates of goshawks differed significantly both for type of call and broadcast distance, due largely to the lack of response by goshawks to owl calls at 300 m. During the nestling period, goshawks responded in significantly less time to conspecific (median =13 sec) than to owl calls (95 sec) at 150 m. There were no differences in response rates relative to time of day or period of breeding season, but adult goshawks were observed near nests more frequently during the nestling versus the fledgling period. Based on our findings, we recommend that conspecific “kakking” calls be used for censuses of Northern Goshawks during nestling and early fledgling periods, and that calls be played along transects that are spaced no more than 300 m apart. Respuestas de Gavilan Azor ( Accipiter gentilis) a reproducciones de grabaciones de las llamadas de su misma especie, 6, a las de las llamadas del buho de la especie Bubo virginianus Extracto. — Hemos comparado las respuestas de Gavilan Azor {Accipiter gentilis ) a las llamadas (“kak- kak”) grabadas de su misma especie, con las respuestas a las llamadas (“jut-jut”) grabadas del buho de la especie Bubo virginianus, durante la epoca de reproduccion en 1989, en Pennsylvania. Las grabaciones fueron reproducidas a distancias de 150 y 300 metros de nidos activos de los gavilanes tanto durante los periodos de cria de los polluelos (7 nidos, N = 27 pruebas), como durante los periodos de los primeros vuelos (7 nidos, N = 28 pruebas). Las pruebas se realizaron con cinco nidos durante esos dos periodos. La proporcion de las respuestas de los gavilanes, a las llamadas reproducidas a 150 metros, fue mas alta con llamadas de su misma especie durante el periodo de cria de los polluelos en el nido (0.71), y mas baja con llamadas de buhos {Bubo virginianus ) durante el periodo de los primeros vuelos (0.14). La proporcion de las respuestas a reproducciones de llamadas de su especie, a una distancia de 300 metros, durante el periodo de crianza y el de los primeros vuelos, fue de 0.29 cada una. No se detecto respuesta alguna a reproducciones de las llamadas (“jut-jut”) de buho, para ninguno de estos periodos y a la misma distancia. Las proporciones de respuesta de los gavilanes fueron, significativamente diferentes, en el caso de las pruebas con el tipo de llamada (de la misma especie o de especie diferente) y en el de las pruebas de distancia de la llamada, debido mayormente a la falta de respuestas a las llamadas de buhos, emitidas a 300 metros del nido. Durante el periodo de cria en el nido, a 150 metros de distancia, los gavilanes respondieron en un tiempo significativamente menor, a las emisiones de las llamadas de su misma especie (media =13 segundos), que a las emisiones de las llamadas de buhos (95 segundos). No se notaron diferencias en la proporcion de las respuestas, en relation con la hora del dia o el tiempo del periodo de reproduccion; pero si se observo que los gavilanes adultos estaban cerca a sus nidos mas frecuentemente durante el periodo de crianza, que durante el periodo de los primeros vuelos. Basados en nuestros resultados, recomendamos que para censar Gavilanes Azor {Accipiter gentilis) sean usadas grabaciones de las llamadas (“kak-kak”) emitidas por la misma especie, tanto durante el periodo de crianza de los polluelos como a principios del periodo de los primeros vuelos. Tambien recomendamos que las llamadas sean emitidas en secciones espaciadas a no mas de 300 metros entre ellas. [Traduction de Eudoxio Paredes-Ruiz] Taped calls of avian vocalizations have been used to detect a variety of raptor species (Fuller and Mo- sher 1981, Johnson et al. 1981). Red- shouldered {Buteo lineatus ), Broad- winged {B. platypterus). Red- tailed {B. jamaicensis ) , Sharp-shinned {Accipiter striatus), and Cooper’s Hawks {A. cooperii) respond to broadcasts of conspecific vocalizations (Balding and Dibble 1984, Fuller and Mosher 1981, 1987, 107 Vol. 24, No. 4 108 J. Timothy Kimmel and Richard H. Yahner TIME OF DRV Early a.m. Late a.m. Early p.m. Late p.m. 1 NEST n 60S-150 NEST B OHIL-150 NEST R OUIL-300 NEST B G0S-300 2 NEST B OIIIL-300 NEST R G CIS-300 NEST B E0S-1 50 NEST R 0I11L- 150 3 NEST C QUIL-150 NEST D EOS-1 50 NEST C G0S-3OO NEST D OIIIL-300 4 NEST D G0S-300 NEST C 0IDL-30D NEST D 0 111 L— 1 SO NEST C G0S-150 Figure 1. Modified Latin square used to schedule trials for testing response of Northern Goshawks to taped calls played near active nests. Conspecific (GOS) and Great Horned Owl (OWL) calls were played at 150 and 300 m from nests. Rosenfield et al. 1985, 1988). Fuller and Mosher (1987) reported that Red-shouldered and Cooper’s Hawks responded as readily to taped calls of the Great Horned Owl ( Bubo virginianus) as to conspe- cific calls. They also noted the value of using the call of a single species, such as that of the Great Horned Owl, to increase the efficiency of surveys intended for multiple raptor species. Despite the fact that Northern Goshawks ( Accip - iter gentilis) also are known to respond to taped calls (Hennessy 1978, Fuller and Mosher 1981), little information is available regarding the application of this technique for goshawk surveys or censuses. Our objectives were to (1) compare responses of goshawks to taped conspecific and Great Horned Owl calls played at two distances from active nests, and (2) evaluate the effects of time of day and period of breeding season on response rates of goshawks to these calls. Results of our study will be useful in developing a standard census technique for nesting goshawks. Materials and Methods The study was conducted at nine active nests of North- ern Goshawks located in six counties in central and north- ern Pennsylvania, USA, in 1989. Five of the nine nests were used for trials during both nestling and fledgling periods. A modified Latin square design was used to schedule trials at nests (Fig. 1). This design maximized indepen- dence of the four experimental factors (Sokal and Rohlf 1981:393) and helped control for possible variation in response rate due to sequential visitation of nests to conduct trials. Experimental factors considered for each trial were type of call (defensive “kakking” call of a Northern Gos- hawk or territorial “hooting” call of a Great Horned Owl), distance from nest (150 or 300 m), time of day (early morning 0800-1000 H, late morning 1001-1200 H, early afternoon 1201-1500 H, or late afternoon 1501-1800 H), and period of breeding season (nestling or fledgling pe- riod). A balanced design required grouping nests in sets of four (i.e., two pairs), with trials at each pair conducted twice daily for 2 consecutive days. Our goal was to test taped calls at two sets of four nests (i.e., eight nests) during both nestling and fledgling periods. However, the widely dispersed nature of goshawk nests combined with nest failure of several nests limited the number of nests for use in our study and resulted in trials being conducted at only seven nests during each period. Twenty-seven trials were conducted during the nestling period (30 May-16 June, mean estimated age of young = 28 d, range = 21-37 d), and 28 trials were conducted during the fledgling period (14 June-4 July; mean estimated age of young = 53 d, range = 46-64 d). Only three trials were conducted at the Warren Co. #4 nest during the nestling period. An owl call was not played at 300 m from this nest because an adult goshawk detected us and vocally responded as we approached the broadcast station; we were unable to com- plete this trial at a later date. Logistics of travelling between nests to conduct trials required the grouping of nests in pairs based on geographic proximity. Thus, complete randomization of nests within the experimental design was not feasible. However, pairs of nests were assigned randomly within the design. Recordings of Northern Goshawk and Great Horned Owl calls were obtained from the Cornell Library of Nat- ural Sounds (Laboratory of Ornithology, Cornell Univer- sity, Ithaca, New York) and were broadcast with a portable Realistic CRT-7 cassette tape player and Half-Mile Hail- er (Perma Power Electronics, Inc., 5615 West Howard Ave., Chicago, Illinois). Audio output was adjusted to 100- 110 db at 1 m in front of the speaker (after Fuller and Mosher 1987) using a Realistic sound level meter set on C-weighting and slow-response. Broadcast stations were established 1 50 and 300 m from active goshawk nests at least one week prior to experi- mental trials, and each was marked with vinyl flagging. Stations were situated so that the slope of a straight line between the station and the nest did not exceed 10% and the area between station and nest was continuous forest and unobstructed by terrain. None of the nests used for experimentation was located initially with taped calls, but all were reported to us by various sources (bird-watchers, foresters, etc.). We wore camouflage clothing during each trial to avoid detection by goshawks. After arriving at a station, we waited quietly for 5 min before beginning the trial. Each trial consisted of playing six bouts of goshawk calls (25- 30 “kaks” over 7 sec) or owl calls (seven “hoots” over 2 sec) spaced evenly over a 5-min period (Fuller and Mosher 1987) with the speaker oriented toward the nest. We re- corded type of response (approach but no vocalization, vocalization but no detectable approach, or approach and vocalization), time (sec) from initiation of playback to de- tection of response, and sex and age of responding bird(s) for all detectable goshawk responses. If no responses were detected, we waited quietly for an additional 5 min and approached the nest to determine the presence of adult or young goshawks. Winter 1990 Goshawk Response to Taped Calls 109 Table 1. Response rates of Northern Goshawks to taped calls (proportion of trials with detectable responses) at nine active nests in Pennsylvania, 1989. Trials during nestling and fledgling periods consisted of four call-dis- tance combinations (conspecific and Great Horned Owl calls at 1 50 and 300 m) played at each nest at four times of day. Results of trials when adults were known or pre- sumed to be near nests are shown in parentheses. Period of Breeding Season Nestling Fledgling Nest N = 27 (N = 20) N = 28 (N = 14) Elk Co. #2 0.50 (0.67) 0.25 (0.25) Elk Co. #6 a 0.50 (0.50) — (-) Forest Co. #3 b — (-) 0.00 (0.00) c McKean Co. #1 0.00 (0.00) c 0.25 (1.00) Potter Co. #3 0.50 (1.00) 0.00 (-) d Snyder Co. #l b — (-) 0.25 (1.00) Warren Co. #4 a 0.67 e (1.00) — (-) Warren Co. #6 0.50 (0.50) 0.75 (0.75) Warren Co. #7 0.25 (0.50) 0.25 (1.00) a No trials conducted during fledgling period due to nest failure. b Nest discovered during fledgling period. c Adults presumed to be present near nest during 3 trials. d Adult(s) not observed near nest during or following any trials. e Only 3 trials; no Great Horned Owl call played at 300 m. Response rate of goshawks to calls was defined as the proportion of trials for which goshawk responses were detected. Differences in response between or among levels of experimental factors were tested using Fisher’s exact test or a G-test of independence, depending on size of cells (Sokal and Rohlf 1981:735). Difference in median time for detectable goshawk response to owl versus conspecific calls was tested with a two-tailed Wilcoxon two-sample test (Sokal and Rohlf 1981:432) and reported as a Chi- square approximation. Statistical significance was P < 0.05 for all tests. We presumed that adult goshawks were not present near nests during some trials. Evidence for this was the apparent absence of adults at some nest sites when we approached nests after trials with no detectable responses. Because we were uncertain of the location of adults during trials that yielded no detectable responses (i.e., adults might have left before, or arrived shortly after, the end of a trial), response rates were evaluated both for all trials conducted (hereafter referred to as “total trials”) and for only those trials when adults were known or presumed to be near nests (“trials with adults present”). Results Goshawk responses to taped calls were detected for 18 of 55 (33%) total trials (Table 1). We at- tempted to determine the presence of adult goshawks near nests following 36 of the 37 trials for which no responses were detected. Adults were observed near 150 300 Broadcast Distance (m) Figure 2. Response rates of Northern Goshawks to con- specific (GOS) and Great Horned Owl (OWL) calls played at 150 and 300 m from active goshawk nests during nest- ling and fledgling periods (N = 55 trials). nests following 9 of 15 (60%) and 7 of 21 (33%) of these trials during nestling and fledgling periods, respectively. Thus, responses were detected for 18 of the 34 (53%) trials with adults present. Response rates by goshawks at any given nest per period of the breeding season (4 call-distance combinations combined) ranged from 0.0 to 0.75 for total trials and 0.0 to 1.0 for trials with adults present (Table D- Behavior of goshawks responding to taped calls ranged from silent approach to approach with vo- calization, and 15 of the 18 (83%) detectable re- sponses included vocalizations. Vocal responses to conspecific and owl calls, respectively, included five and one without approach and six and three with approach. Seventeen of the 18 (94%) detectable re- sponses were by single adult goshawks, presumed in most cases to be females of the breeding pairs. One response was a non-vocal approach by a fledgling 20 sec after termination of a conspecific broadcast at 150 m. Effects of Experimental Factors. Response rates for total trials generally were lower during the fledg- ling period, particularly for the owl call (Fig. 2). Also, adult goshawks were detected near nests more fre- quently during the nestling period than during the fledgling period ( G = 4.3, P = 0.04). Nonetheless, differences in response rates between nestling and 110 J. Timothy Kimmel and Richard H. Yahner Vol. 24, No. 4 Table 2. Frequencies of detectable responses by North- ern Goshawks to taped calls in Pennsylvania, 1989, in relation to four experimental factors (N — 55 trials). Fre- quencies of responses for trials when adults were known or presumed to be near nests (N = 34) are shown in parentheses. Experimental Goshawk Response Detected Factor Yes No Period of breeding season Nestling 11 (11) 16(9) Fledgling 7(7) 21 (7) Type of call Northern Goshawk 13 (13) 15(6) Great Horned Owl 5(5) 22 (10) Broadcast distance 150 m 14(14) 14(4) 300 m 4(4) 23 (12) Time of day Early a.m. 3(3) 11 (3) Late a.m. 5(5) 8(4) Early p.m. 4(4) 10(6) Late p.m. 6(6) 8(3) Sequence of broadcast trials Trial 1 4(4) 9(3) Trial 2 5(5) 9(4) Trial 3 4(4) 10(4) Trial 4 5(5) 9(5) fledgling periods (Table 2) were not significant for total trials (G = 1.6, P = 0.21) or for trials with adults present (G = 0-1, P = 0.77). Similarly, there were no significant differences in response rates be- tween nestling and fledgling periods (total trials) for conspecific ( G = 0.1, P = 0.71) or for owl calls (Fisher’s exact test, P = 0.17). There was no difference among times of day for adult goshawks to be observed near nests (G = 2.4, 3 df, P = 0.49). Response rates among times of day (Table 2) did not differ for total trials (Fisher’s exact test, P = 0.64) or for trials with adults present (Fish- er’s exact test, P — 0.73). During the nestling period, response rates to total trials were lowest in early morning and higher in late morning and late after- noon (Fig. 3); however, differences among times of day during the nestling period were not significant for total trials (Fisher’s exact test, P = 0.22) or for trials with adults present (Fisher’s exact test, P = 0.71). Response rates did not vary among four sequential 1.0 3 °’ 8