THE RAPTOR RESEARCH FOUNDATION, INC. (Founded 1966) http: //biology.boisestate.edu/ raptor/ OFFICERS SECRETARY: Judith Henckei. TREASURER: Jim Fitzpatrick BOARD OF DIRECTORS INTERNATIONAL DIRECTOR #3: Steve Redpatii DIRECTOR AT LARGE #1: Jemima Parry-Jones DIRECTOR AT LARGE #2: Eduardo Inigo-Elias DIRECTOR AT LARGE #3: Michael W. Collopy DIRECTOR AT LARGE #4: Carol McIntyre DIRECTOR AT LARGE #5: John A. Smallwood DIRECTOR AT LARGE #6: Daniel E. Varland Ruth Tingay NORTH AMERICAN DIRECTOR #1: Steve Hoffman NORTH AMERICAN DIRECTOR #2: Gary Santolo NORTH AMERICAN DIRECTOR #3: Ted Swem INTERNATIONAL DIRECTOR#!: Nick Mooney INTERNATIONAL DIRECTOR #2: PRESIDENT: Brian A. Millsap VICE-PRESIDENT: David M. Bird EDITORIAL STAFF EDITOR: James C. Bednarz, Department of Biological Sciences, RO. Box 599, Arkansas State University, State University, AR 72467 U.S.A. ASSOCIATE EDITORS James R. Belthofe Joan L. Morrison Clint W. Boat Fabrizio Sergio Chfryi. R. Dykstra Ian G. Warkentin Michael I. Goi dstein James W. Watson BOOK REVIEW EDITOR: Jeffrey S. Marks, Montana Cooperative Research Unit, University of Montana, Missoula, MT 59812 U.S.A. SPANISH EDITORS: Carlos Daniel Cadena, Lucio R. Malizia, Cintia Cornelius EDITORIAL ASSISTANTS: Autumn Earless, Joan Clark The Journal of Raptor Research is distributed quarterly to all current members. Original manuscripts dealing with the biology and conservation of diurnal and nocturnal birds of prey are welcomed from throughout the world, but must be written in English. Submissions can be in the form of research articles, short communications, letters to the editor, and book reviews. Contributors should submit a typewritten original and three copies to the Editor. All submissions must be typewritten and double-spaced on one side of 216 X 278 mm {SVi X 11 in.) or standard international, white, bond paper, with 25 mm (1 in.) mar- gins. The cover page should contain a title, the author’s full name(s) and address (es). Name and address should be centered on the cover page. If the current address is different, indicate this via a footnote. A short version of the title, not exceeding 35 characters, should be provided for a running head. An abstract of about 250 words should accompany all research articles on a separate page. Tables, one to a page, should be double-spaced throughout and be assigned consecutive Arabic numer- als. Collect all figure legends on a separate page. Each illustration should be centered on a single page and be no smaller than final size and no larger than twice final size. The name of the author(s) and figure number, assigned consecutively using Arabic numerals, should be pencilled on the back of each figure. Names for birds should follow the A.O.U. Checklist of North American Birds (7th ed., 1998) or another authoritative source for other regions. Subspecific identification should be cited only when pertinent to the material presented. Metric units should be used for all measurements. Use the 24-hour clock (e.g., 0830 H and 2030 H) and “continental” dating (e.g., 1 January 1999). Refer to a recent issue of the journal for details in format. Explicit instructions and publication policy are outlined in “Information for contributors,”/. Raptor Res., Vol. 38(4), and are available from the editor. Submit manuscripts to J. Bednarz at the address listed above. COVER; Northern Goshawk {Accipiter gentilis) . Painting done in gouache by Michael Demain; for more information and images, visit www.michaeldemainwildlifeart.co.uk Contents Introduction Preface: Proceedings oe the International Symposium on the Ecology and Management of Northern Goshawks. Clint W. Boal 189 In Memoriam: Suzanne MeridethJoy. Richard T. Reynolds 190 Status Technical Review of the Status of Northern Goshawks in the Western United States. David E. Andersen, Stephen DeStefano, Michael I. Goldstein, Kimberly Titus, Cole Crocker-Bedford, John J. Keane, Robert G. Anthony, and Robert N. Rosenfield 192 Biology Is Fledging Success a Reliable Index of Fitness in Northern Goshawks? J. David Wiens and Richard T. Reynolds 210 Productivity and Mortality of Northern Goshawks in Minnesota. Clint W. Boal, David E. Andersen, and Patricia L. Kennedy 222 Relationships Between Winter and Spring Weather and Northern Goshawk (Accipiter gentilis) Reproduction in Northern Nevada. Graham D. Fairhurst and Marc J. Bechard 229 Patterns of Temporal Variation in Goshawk Reproduction and Prey Resources. Susan R. Salafsky, Richard T. Reynolds, and Barry R. Noon 237 A Skewed Sex Ratio in Northern Goshawks: Is It a Sign of a Stressed Population? Michael E Ingraldi 247 Northern Goshawk {Accipiter gentilis iaingi) Post-fledging Areas on Vancouver Island, British Columbia. Erica L. McClaren, Patricia L. Kennedy, and Donald D. Doyle 253 Northern Goshawk Diet in Minnesota: An Analysis Using Video Recording Systems. Brett L. Smithers, Clint W. Boal, and David A. Andersen 264 Techniques Sampling Considerations for Demographic and Habitat Studies of Northern Goshawks. Richard T. Reynolds, J. David Wiens, Suzanne M. Joy, and Susan R. Salafsky 274 Population Genetics and Genotyping for Mark-Recapture Studies oe Northern Goshawks {Accipiter gentius) on the Kaibab Plateau, Arizona. Shelley Bayard de Volo, Richard T. Reynolds, J. Rick Topinka, Bernie May, and Michael F. Antolin 286 When Are Goshawks Not There? Is a Single Visit Enough to Infer Absence at Occupied Nest Areas? Douglas A. Boyce, Jr., Patricia L. Kennedy, Paul Beier, Michael F. Ingraldi, Susie R. MacVean, Melissa S. Siders, John R. Squires, and Brian Woodbridge 296 Quantifying Northern Gosfiawk Diets Using Remote Cameras and Observations FROM Blinds. Audi S. Rogers, Stephen DeStefano, and Micheal F. Ingraldi 303 Conservation Temporal Patterns of Northern Goshawk Nest Area Occupancy and Habitat: A Retrospective Analysis. Steven Desimone and Stephen DeStefano 310 Monitoring Results of Northern Goshawk Nesting Areas in the Greater Yellowstone Ecosystem: Is Decline in Occupancy Related to Habitat Change? Susan M. Pada 324 Effects of Timber Harvesting Near Nest Sites on the Reproductive Success of Northern Goshawks {Accipiter gentius). Todd Mahon and Frank I. Doyle 335 A Review of the Status and Distribution of Northern Goshawks in New Engiand. Stephen DeStefano 342 The Raptor Research Foundation, Inc. gratefully acknowledges funds and logistical support provided by Arkansas State University to assist in the publication of the journal. THE JOURNAL OF RAPTOR RESEARCH A QUARTERLY PUBLICATION OF THE RAPTOR RESEARCH FOUNDATION, INC. VoL. 39 September 2005 No. 3 J Raptor Res. 39(3):189 © 2005 The Raptor Research Foundation, Inc. PREFACE PROCEEDINGS OF THE INTERNATIONAL SYMPOSIUM ON THE ECOLOGY AND MANAGEMENT OF NORTHERN GOSHAWKS Clint W. Boal^ U.S. Geological Survey, Texas Cooperative Fish and Wildlife Research Unit, Department of Range, Wildlife, and Fisheries Management, Texas Tech University, Lubbock, TX 79409-2120 U.S. A. For almost 20 years, the Northern Goshawk {Ac- cipiter gentilis) has been the focus of considerable research effort, management and conservation planning, and litigation. This has been driven by conflict between conservation concerns and forest management practices. Due to increasing concern for the species, the symposium “The Biology and Management of Northern Goshawks,” was held in conjunction with the Cooper Ornithological Soci- ety in 1993, with the goal of assembling the most current information available at that time. Since that 1993 symposia, 10 years had passed during which a considerable amount of research had been conducted on goshawks, but a venue for sharing the information among researchers had not been made available. To provide an opportunity for researchers and managers to exchange information with which to assess the current state of knowledge on Northern Goshawks, a coordinating committee consisting of Steve DeStefano, Patricia Kennedy, Mi- chael Goldstein, John Keane, and myself organized the “International Symposium on the Ecology and Management of Northern Goshawks.” This sympo- sium was held 4—5 September 2003, in conjunction with the 2003 Raptor Research Foundation Annual Meeting in Anchorage, Alaska. The symposium was well attended, with 32 papers presented. Topics in- cluded status reports; improvements in survey and monitoring methodology; information on popula- tion demography, food habits, habitat assessment, and winter ecology; and the use of genetic ap- proaches to goshawk studies. Given the obvious interest in goshawk ecology ^ E-mail address: cboal@ttu.edu and management, as indicated by attendance and number of papers presented at the symposium, there was a need to make this information more broadly available. We have attempted to do so with this issue of The Journal of Raptor Research, which includes peer-reviewed versions of some of the pa- pers presented at the 2003 symposia. The lead pa- per is authored by a committee put together jointly by The Raptor Research Foundation and The Wildlife Society. This committee was charged with reviewing the status of Northern Goshawks in the western United States. This review provides a con- text for the rest of this issue in presenting an over- view of the state of knowledge leading up to the symposium. Manuscripts following the lead paper are grouped topologically (i.e.. Biology, Tech- niques, and Conservation). All manuscripts in this issue went through the same peer-review process as required for regular issues of the Journal. For this issue, I served as a Coordinating Editor, with the symposia committee members serving as Associate Editors. I, the Asso- ciate Editors, and Jim Bednarz (Editor-in-Chief) , would like to thank all of the individuals that served as peer-referees for the manuscripts in this issue. We would also like to acknowledge the U.S. Geological Survey Cooperative Research Units, the North American Falconer’s Association, and the U.S. Forest Service Alaska Region and the Chu- gach National Forest, for providing support for publication of this issue of the Journal. Finally, this issue of The Journal of Raptor Research is dedicated to the memory of Suzanne Joy, a friend and colleague to many of us working with goshawks. Please see the memoriam provided by Richard Reynolds on the following page. 189 J. Raptor Res. 39 (3); 190-1 91 © 2005 The Raptor Research Foundation, Inc. IN MEMORIAM Suzanne Merideth Joy 12 APRIL 1961-7 DECEMBER 2004 Richard T. Reynolds Rocky Mountain Research Station, USDA Forest Service, 2150 Centre Avenue, Building A, Suite 350, Fort Collins, CO 80526-1891 U.S.A. Suzanne Merideth Joy, microbiologist, photographer, animal ecologist, raptor biologist, tree climber extraordinaire, spatial analyst, and always a mentor, passed away on 7 December 2004 in Teramo, Italy. She was in Italy developing spatial models of the habitat of biting midge (Ceratopogomidae) vectors of bluetongue disease (Reoviridae) in cattle and sheep on behalf of the Italian government and the U.S. Department of Agriculture Animal and Plant Health Inspection Service (APHIS), her employer since 2003. Suzanne was born in Goose Bay, Labrador, but grew up in Virginia, Madrid (Spain), Minnesota, and Arizona. After graduating from Colorado State University with honors (Phi Beta Kappa, Phi Kappa Phi) and as the outstanding senior in Microbiology in 1983, she spent 2 yr with the Peace Corps in Kenya, East Africa. She was conversant in French, Swahili, and Kikuyu. Upon returning from Kenya, she worked as a radioimmunoassay technician for Hazleton Biotechnologies Company in Massachusetts and researched a pertussis vaccine for the Food and Drug Administration in Bethesda, Maryland. In 1987, wanting to move from the confines of research labs to the natural habitats of birds and mammals, Suzanne returned to Colorado State University for a Master of Science (1990) degree in Fishery and Wildlife Biology. Her thesis topic was the feeding ecology and nest habitat of the Sharp-shinned Hawk {Acdpiter striatus) in Colorado’s Rocky Mountain forests. The experience and knowledge she gained while sampling for nesting Acdpiter, collecting and identifying their food remains, and measuring the composition and structure of their forest nest sites was soon put to use in helping me establish in 1991 what would turn out to be an intensive and long-term study of the demography, genetics, and habitat of Northern Goshawks (A. gentilis) on the Kaibab Plateau, AZ. Before beginning a Ph.D. program at Colorado State University, Suzanne helped develop protocols for finding and trapping goshawks and managing budgets and large field crews through much of the Kaibab goshawk study. She loved being in the piney woods searching for and trapping goshawks, and making difficult tree climbs to goshawk nests to band their young. In 2002, Suzanne was awarded her Ph.D. by the Graduate Degree Program in Ecology at Colorado State University. Her dissertation, entitled “Northern Goshawk Habitat on the Kaibab National Forest in Arizona: Factors Affecting Nest Locations and Territory Quality,” included a dynamic spatial simulation model that described the spatial dependence of goshawk nest locations on both territoriality and the availability of suitable nest sites. She identified the correlates of habitat quality by quantifying the relationship between the long-term reproductive per- formance of breeding goshawks and the composition and structure of habitats within their territories. Within a few months of completing her dissertation, Suzanne moved to APHIS, where she worked as a spatial analyst, and where once again she quickly became a star. Suzanne Joy is survived by a son, Quinn, who shows his mother’s love of nature, and husband, Vern Thomas, both of whom brought much additional joy to her life. Suzanne will be greatly missed by all those whose lives she touched. I was privileged to have known Suzanne as a student, employee, colleague, mentor, and above all, a friend. . The editors and authors of these proceedings dedicate the following contributions on the biology and conservation of the Northern Goshawk to Suzanne Joy’s memory. 190 September 2005 In Memoriam 191 IN MEMORIAM Suzanne Merideth Joy 12 APRIL 1961-7 DECEMBER 2004 Suzanne Joy with captured Northern Goshawk in the Kaibab National Forest, Arizona. J. Raptor Res. 39(3):192-209 © 2005 The Raptor Research Foundation, Inc. TECHNICAL REVIEW OF THE STATUS OF NORTHERN GOSHAWKS IN THE WESTERN UNITED STATES^ David E. Andersen^ U.S. Geological Survey, Minnesota Cooperative Fish and Wildlife Research Unit, St. Paul, MN 35108 U.S.A. Stephen DeStefano U.S. Geological Survey, Massachusetts Cooperative Fish and Wildlife Research Unit, Amherst, MA 01003 U.S.A. Michael L Goldstein U.S. Forest Service, Chugach National Forest, Anchorage, AK 99503 U.S.A. Kimberly Titus Alaska Department of Fish and Game, funeau, AK 99802 U.S.A. Cole Crocker-Bedford U.S. Park Service, Grand Canyon National Park, Grand Canyon, AZ 86023 U.S.A. JohnJ. Keane US. Forest Service, Pacific Southwest Research Station, Sierra Nevada Research Center, Davis, CA 95616 U.S.A. Robert G. Anthony U.S. Geolo^cal Survey, Oregon Cooperative Fish and Wildlife Research Unit, Corvallis, OR 97331 U.S.A. Robert N. Rosenfield Department of Biology, University of Wisconsin— Stevens Point, Stevens Point, WI 54481 U.S.A. Abstract. — The U.S. Fish and Wildlife Service (FWS) was petitioned in 1997 to consider listing North- ern Goshawks (Accipiter gentilis atricapillus) under the Endangered Species Act of 1973, west of the 100* meridian of the contiguous United States. In their 12-mo finding issued in June 1998, the FWS deter- mined that listing this population as threatened or endangered was not warranted and based that decision on review of existing population and habitat information. Because the status of goshawks in the western U.S. continues to be contentious and the FWS finding has been challenged, the Raptor Research Foundation, Inc. and The Wildlife Society jointly formed a committee to review information regarding the status of the goshawk population in the contiguous U.S. west of the 100* meridian. The committee was requested to: (1) determine if there is evidence of a population trend in goshawks in the western U.S., excluding Alaska; (2) determine if there is evidence that goshawks nesting in the eastern and western U.S. represent distinctive, genetically unique populations; and (3) evaluate evidence for goshawk— habitat relations, including any association with large, mostly-unbroken tracts of old growth and mature forests. Based on existing information, the committee concluded: (1) existing data are not adequate to assess population trend in goshawks west of the 100* meridian; (2) existing analyses of phylogeography have not provided evidence of genetic differences among recognized {atricapillus, laingi) or putative {apache) subspecies, and the genetic distinctness of atricapillus goshawks in western and eastern North America is not known; and (3) at present, assessing the status of goshawks solely using distribution of late-successional forests is not appropriate, based on the current understanding of goshawk-habitat relations, although goshawks clearly use and often select late-successional forests for * Summary of a report prepared by the Technical Committee on the Status of Northern Goshawks in the Western United States sponsored jointly by the Raptor Research Foundation, Inc. and The Wildlife Society. A copy of the complete report can be obtained from The Wildlife Society: (http://www.wildlife.org). ^ Corresponding author’s email address: dea@umn.edu 192 September 2005 Status 193 nesting and foraging. We provide recommendations on information needs to assess status and popula- tion trend of goshawks in the western U.S. Keywords: Northern Goshawk, Accipiter gentilis atricapillus; western U.S.; status', review, population trend', genetic structure, habitat relations. REVISION TECNICA DEL ESTATUS DE ACCIPITER GENTILIS ATRICAPILLUS EN EL OESTE DE LOS ESTADOS UNIDOS Resumen. — El Servicio de Pesca y Vida Silvestre de los Estados Unidos (FWS, por sus siglas en ingles) recibio en 1997 la peticion de considerar a Accipiter gentilis atricapillus al oeste del meridiano 100 de los Estados Lfnidos (considerando solo los estados contiguos) como un ave amenazada de acuerdo al acta de 1973. Luego de 12 meses, en junio de 1998 el FWS dictamino que clasificar a esta poblacion como amenazada o en peligro no era justificable, y baso dicha decision en una revision de la informacion poblacional y de habitat existente. Debido a que el estatus de A. g. atricapillus en el oeste de los Estados Unidos es aun controversial y a que el hallazgo del FWS ha sido desafiado, la Raptor Research Foun- dation, Inc. y la Wildlife Society formaron un comite conjunto para revisar la informacion concerniente al estatus de la poblacion de esta ave en los estados contiguos de los Estados Unidos al oeste del meridiano 100. Al comite se le solicito que (1) determinara si existe evidencia de una tendencia de cambio en el tamaho poblacional de A. g. atricapillus en el oeste de E. U., excluyendo Alaska; (2) determinara si existe evidencia de que los individuos que nidifican en el este y el oeste de E. U. representan poblaciones distintivas, geneticamente unicas; y (3) evaluara la evidencia sobre las rela- ciones de A. g. atricapillus con el habitat, incluyendo cualquier asociacion con reductos grandes y no fragmentados de bosques maduros. Con base en la informacion existente, el comite concluyo que: (1) los datos disponibles son inadecuados para evaluar si existe una tendencia de cambio en el tamaho poblacional al oeste del meridiano 100; (2) los analisis de filogeografia existentes no han provisto evidencia que indique la existencia de diferencias entre las subespecies reconocidas {atricapillus, laingi) o putativas {apache), y no se conocen diferencias geneticas entre las poblaciones del oeste y el este de Norte America; y (3) en la actualidad, evaluar el estatus de A. g. atricapillus con base solo en la distri- bucion de bosques de estadios sucesionales tardios no es adecuado de acuerdo al conocimiento actual de sus relaciones con el habitat, aunque es claro que esta ave utiliza y a menudo selecciona bosques de sucesion avanzada para nidificar y forrayear. Ofrecemos recomendaciones en cuanto a la informacion necesaria para evaluar el estatus y las tendencias poblacionales de A. g. atricapillus en el oeste de Estados Unidos. [Traduccion del equipo editorial] In 1997, the U.S. Fish and Wildlife Service (FWS) received a petition to list the Northern Gos- hawk {Accipiter gentilis atricapillus; hereafter re- ferred to as goshawk) west of the 100‘^ meridian of the contiguous United States under the Endan- gered Species Act of 1973. In its 90-d finding issued in September 1997 (United States Department of Interior [USDI] 1997), the FWS found that the pe- tition “presented substantial information indicat- ing that the listing of the Northern Goshawk as a threatened or endangered species in the contigu- ous United States west of the 100* meridian may be warranted” (USDI 1998). The FWS at that time initiated a status review (FWS 1998) for the gos- hawk, and in June 1998 issued its 12-mo petition finding (USDI 1998) and indicated that after “. . . reviewing all available scientific and commercial in- formation, the Service finds that listing this popu- lation as endangered or threatened is not warrant- ed” (USDI 1998:35183). The FWS used data from recent survey and mon- itoring that suggested goshawks had generally been located where intensive survey and monitoring ef- forts were implemented, and that goshawks re- mained widely distributed throughout their histor- ical range. The FWS also reviewed existing habitat data and concluded that there was no evidence that habitat was currently limiting the goshawk population in the western U.S., and habitat was un- likely to limit this population in the foreseeable future. The petition for listing suggested that gos- hawks in the western U.S. were dependent upon large, unbroken tracts of late-successional forest, but the FWS concluded that there was little or no support for that assertion. Subsequent to release of the 12-mo finding by the FWS, several court chal- 194 Andersen et al. VoL. 39, No. 3 lenges were submitted, both to the finding itself and to the process used to arrive at the finding. Clearly, there is considerable concern for con- servation of goshawk populations and their habi- tats in western North America. As some of the fore- most professional societies concerned with conservation of wildlife in general, and raptors in particular, the Raptor Research Foundation, Inc. (RRF) and The Wildlife Society (TWS) formed a joint committee to review information regarding the status of the goshawk, population in the western contiguous U.S. The purpose behind forming this committee was to provide an independent techni- cal review of existing information related to gos- hawk status and to identify additional information necessary to assess population trend adequately. Specifically, the committee was requested to: (1) determine if there was evidence of a population trend in goshawks in the western U.S. west of the 100* meridian, excluding Alaska; (2) determine if there was evidence that goshawks nesting in the eastern and western U.S. represent genetically dis- tinct populations; and (3) evaluate evidence for goshawk-habitat relations, including any associa- tion with large, mostly unbroken tracts of old- growth and mature forests. In addition, the com- mittee was asked to evaluate existing information on population trend, genetic structure, and habitat relations and to identify types of information need- ed to assess the status of goshawks more conclu- sively in the western U.S., excluding Alaska. This manuscript summarizes the process used, infor- mation evaluated, and opinions of the Joint RRF- TWS Technical Committee on the status of North- ern Goshawks in the western United States. A copy of the complete report can be obtained from TWS (http://www.wildlife.org) . Methods The scope of the committee’s review and evaluation was restricted to pertinent technical information, com- prised of peer-reviewed primary literature, theses, or un- published technical information that the committee deemed credible and that related directly to the com- mittee’s charge. Information considered included that summarized in the FWS goshawk status review (USDI 1998) and related documents (e.g., FWS 1998), syntheses of the published literature (e.g., Squires and Reynolds 1997), and published and unpublished information not included in previous reviews and available to the com- mittee through completion of its charge in 2003. Where possible, the committee reviewed primary literature and data, rather than relying solely on published or unpub- lished syntheses. Committee deliberations focused on three mzyor areas: (1) population trends, (2) genetic structure, and (3) goshawk— habitat associations. In ad- dition, as a fourth area, the committee considered recent conservation efforts that focused on the possibility of us- ing goshawk-habitat associations and habitat monitoring as a surrogate for population monitoring. Results Population Trends. Migration counts. Migration counts have several mEyor drawbacks as an index to the population size of goshawks in western North America. First, there is a nearly complete lack of knowledge of the geographic origin (e.g., breeding grounds) of goshawks observed at count locations. Second, migration routes for goshawks in western North America are poorly known (Squires and Reynolds 1997). Third, a primary lim- itation of migration counts is that changes in counts (FWS 1998) have an unknown relation to changes in the size of the target population (Ken- nedy 1998). Fourth, many migration counting sta- tions, especially in western North America (FWS 1998), have small counts of migrating goshawks. Fifth, counting effort at some migration sites is var- iable through time and would need to be stan- dardized if counts were to be used as an index to population size (Mueller et al. 1977, Bednarz et al. 1990, Bildstein 1998). Finally, continental counts included in the FWS status review (FWS 1998) are comprised primarily of counts of migrating gos- hawks from a single, more eastern site — Hawk Ridge near the western end of Lake Superior. For these reasons, migration counts at present are not a reliable index of goshawk population size in west- ern North America. Trend data. Breeding Bird Survey (BBS) data are inadequate to estimate population trends for gos- hawks across the western U.S., both because the number of routes on which goshawks were detect- ed (<35) and the encounter rate of goshawks on these routes (mean detection rate <0.02 goshawks per route) were too low. Christmas Bird Count (CBC) data were also inadequate to estimate gos- hawk population trends at large scales because of low encounter rates. In addition, the CBC is con- ducted outside of the breeding season, thereby making the origin of observed birds uncertain. Thus, observed trends in CBC data cannot be re- lated to the population of goshawks breeding in the western U.S. Productivity. Existing data on goshawk reproduc- tion in the western U.S. suggest that annual pro- ductivity (e.g., FWS 1998, Ingraldi 1998, Reynolds and Joy 1998) and nest success (Squires and Reyn- September 2005 Status 195 olds 1997, FWS 1998, Ingraldi 1998, Reynolds and Joy 1998) are highly variable. Interpretation of studies of goshawk productivity is further con- founded by small sample sizes (e.g., FWS 1998) and biases in estimates of breeding area occupancy and nest success. High annual variability in repro- duction appears to be characteristic of all goshawk populations studied to date and is associated with annual variation in weather and prey (Kostrzewa and Kostrzewa 1990, Keane 1999, Doyle and Smith 2001). Finally, research on long-lived raptors sug- gests that some breeding areas consistently fledge more young than others, with the majority of young in the population being produced by a few females that occupy high-quality breeding areas (e.g., Newton 1989, 1991). Relations between and among productivity, habitat quality, population size, and trends in goshawks are not clear, and ob- served trends in productivity by themselves cannot be related to population status. As a result, it is difficult or impossible to discern any trends in gos- hawk reproductive success in the recent past over a wide geographic area. However, even if such tem- poral trends were discernable in the western U.S., such trends per se would not serve as an adequate foundation for concluding that similar trends would thereby exist in population size. Informa- tion on reproduction must be combined with sur- vival and immigration-emigration data at appro- priate scales to derive population growth rates (e.g., Maguire and Call 1993). To date, such infor- mation on goshawks in the western U.S. does not exist. Distribution. Squires and Reynolds (1997) provid- ed the most current delineation of known year- round and wintering ranges of goshawks in the western U.S. Contraction of historical breeding or wintering ranges could suggest a decline in popu- lation size (Kennedy 1997), but no historical or current evidence is available to suggest either a range contraction or expansion in the western U.S. Without reliable information on historical breed- ing and wintering ranges, knowledge of current ranges has limited utility to evaluate current pop- ulation size or trends. Encounter rates-detection surveys. Surveys for nest- ing goshawks in the western U.S. have been con- ducted in anticipation of proposed timber sales. While some land-management agencies adhere to established survey protocols (e.g., Kennedy and Stahlecker 1993, Joy et al. 1994), many have not, resulting in spatial and temporal variation in meth- odology. Techniques that do not detect all gos- hawks present have not been validated by estimat- ing density at multiple sites with known breeding densities (presumably all methods except complete searches of survey plots; even with complete searches, multiple years are probably necessary to detect all goshawk pairs present [DeStefano et al. 1994a, Reynolds and Joy 1998]). Thus, goshawk de- tection rates and estimated nest densities generally cannot be directly compared spatially, or even tem- porally at the same site. Nest density and detection rates from surveys are also influenced by how study areas are defined and located (Smallwood 1998). The primary purpose of most goshawk surveys is not to estimate breed- ing densities or population parameters, but to lo- cate nests for protection and to predict or mitigate the effects of proposed timber sales on goshawks. As a result, the locations of surveys for goshawks are generally not random with respect to potential goshawk habitat. Thus, the results from such sur- veys can appropriately be applied to the goshawk nests studied, but any inference beyond the sample is speculative. Comparing among studies is also in- appropriate in some cases because of differences in survey techniques, interpretation, and report- ing. Inconsistent definition and use of terms relat- ed to goshawk ecology (see the Appendix for pro- posed standard terminology) further confound comparisons among studies. These factors limit the utility of detection surveys as an index to goshawk densities and population trends in the western U.S. Existing data from detection surveys do not pro- vide insight into goshawk population status beyond documenting occurrence of breeding birds at sur- vey sites. Demographic data. Demographic studies often fo- cus on estimating lambda (the annual rate of pop- ulation growth) with Leslie-matrix projection mod- els from estimates of age-specific fecundity and survival. However, even at the scale of local study areas, data necessary to estimate lambda are gen- erally inadequate for goshawks (e.g., DeStefano et al. 1994b, Reynolds and Joy 1998). While consid- erable information exists on fecundity, there are few estimates of adult survival, and data on juvenile survival are lacking (but see DeStefano et al. 1994b, Kennedy 1997, Ingraldi 1998, Reynolds and Joy 1998). With the possible exception of the on- going long-term study on the Kaibab Plateau in Ar- izona (Reynolds and Joy 1998), studies have not been conducted for sufficient time periods with ad- 196 Andersen et al. VoL. 39, No. 3 equate sample sizes to understand temporal varia- tion in adult survival and fecundity. The propor- tion of adults attempting to breed has been estimated in only a few places (Reynolds and Joy 1998). Among-year movements, especially by adult female goshawks to different nesting areas, add complexity to estimating demographic parameters, because without radiotelemetry data, the fate of these birds will often be unknown (Flatten et al. 2001 ) . Production of young (to fledging) has been estimated in a number of studies, but only in a few locations have these data been coupled with sur- vival information. Finally, information regarding immigration and emigration of juvenile and adult goshawks is lacking. Thus, while demographic stud- ies have significantly increased understanding of goshawk population dynamics, no studies to date have generated adequate empirical stage-specific estimates of survival and fecundity for estimating lambda with matrix projection models at local scales, and demographic data are unavailable at larger scales, making it impractical to estimate lambda for the western U.S. Recent alternative models for estimating lambda (e.g., Pradel 1996), or models for assessing trends in adult survival, have not been applied to existing goshawk data, but they should be explored. Direct estimation of trends in breeding popula- tion size on local study areas has been hampered by problems associated with searching large areas for nests, difficulty in detecting pairs that are pres- ent but not nesting, edge effects, limited method- ology available to estimate density, and spatial and temporal variation in search effort and protocols. In addition, size and location of study areas can affect estimation of population size (Smallwood 1998) because study areas are seldom chosen ran- domly. Thus, similar to estimating population growth rate based on demographic rates, estimat- ing population trends on the scale of local study areas has had limited success. Trends in density. Breeding densities of goshawks vary considerably across their geographic range; densities in 10 published studies in North America ranged from 0.03-11.9 pairs or nests per 100 km^. In the western U.S., excluding Alaska, densities in seven published studies ranged from 1.4-11.9 pairs or nests per 100 km^ (Squires and Reynolds 1997, Reynolds and Joy 1998, FWS 1998, Bosakowski 1999) . Goshawk density (number of breeding pairs/area) reported in unpublished work sum- marized by the FWS (1998) fell within the same range. Comparison among existing estimates of breeding density are confounded by a number of factors, including variation among studies in defi- nitions of densities, territories, pairs, “active” nests, and occupied nest areas (see Appendix) . In addition, the small number of published studies of goshawk breeding density {N — 7) , the limited du- ration of most studies (median = 2.0 yr; Squires and Reynolds 1997), and high temporal variability in reproduction preclude reliable assessment of temporal trends in breeding densities of goshawks across the western U.S. The logistical problems of determining density in goshawks and possible methodological bias in selecting nest search areas for some studies (Kennedy 1997, Squires and Reyn- olds 1997, Smallwood 1998, Trexel et al. 1999) may further confound analyses of breeding densities as an index to population size. Moreover, densities of the nonbreeding segment of goshawk populations (floaters) and their demographic role are entirely unknown (Hunt 1998) . Theoretically, a population decline may occur without concurrent decline in nesting densities if floaters are available to fill va- cant breeding territories. Declines in nesting den- sity may only then become apparent after the float- er population has been exhausted (Franklin 1992). Currently, existing data on nesting and breeding densities are not adequate to assess goshawk pop- ulation trends across western North America. Historical records. There have been no systematic efforts to synthesize existing historical goshawk rec- ords across North America, and only limited infor- mation is available for portions of their range (e.g., Grinnell and Miller 1944); therefore, historical data were not available to the FWS for assessing change in goshawk distribution in the western U.S. Use of historical records for assessing distributional change has limitations because natural history col- lections are not random or systematic samples from across the historical range of a species (Shaf- fer et al. 1998). The number of historical goshawk records represented in museum collections is also limited because of the relative rarity of goshawks, their secretive behavior, and predominant occur- rence in remote locales. Because of these limita- tions, historical records are not available for as- sessing historical ranges and current changes in distribution for goshawks in all regions of North America. Data necessary to assess historical gos- hawk distribution across western North America have not been collected, and thus contrasts be- September 2005 Status 197 tween historical and current ranges of goshawks in the western U.S. are only possible for limited areas. Genetic Structure. Observed morphological patterns. Two subspecies of goshawks {A. g. atricapillus, A. g. laingi) were recognized in the western U.S. and southeast Alaska by the American Ornithologists’ Union in 1957 (AOU 1957). A. g. atricapillus occnv^ across nearly all forested regions of the western U.S., Canada, the western Great Lakes region, and the northeastern U.S. A. g. laingi occurs from Van- couver Island, insular British Columbia, to the Al- exander Archipelago of southeastern Alaska (Whaley and White 1994). A third, putative sub- species (e.g., Stresemann and Amadon 1979), A. g. apache, occurs in the mountains of southern Ari- zona, but was not recognized by the AOU (1957) and is currently not recognized by most taxono- mists (Whaley and White 1994) . Morphological dif- ferences between eastern and western A. g. atricap- illus have not been demonstrated in the literature (see Whaley and White 1994). Ridgway (in Baird et al. 1875) speculatively divided eastern {Astur atri- capillus atricapillus) and western (then termed Astur atricapillus striatulus) goshawks, but others, includ- ing Taverner (1940), have not made this distinc- tion. Sample sizes have been small in the analyses of eastern A. g. atricapillus, or the analyses were confounded by migrants (Mueller et al. 1976). Since Whaley and White (1994), there have not been any in-depth analyses of A. g. atricapillus across the continent using larger sample sizes. Genetic population structure. There are few publi- cations on the phylogeography of DNA in North American goshawks. In an unpublished report, Gavin and May (1996) did not detect genetic dif- ferences among goshawks representing A. g. atri- capillus, A. g. laingi, and A. g. apache. More recently, Sonsthagen et al. (2004) used eight microsatellite DNA loci and mitochondrial DNA control-region sequence data to assess population structure of goshawks breeding in Utah. Their pairwise com- parisons using microsatellite markers found no dif- ferentiation among the sampled sites {N = 49 birds) from northern to southern Utah. Overall, they found low levels of population structuring. During the 1990s, numerous goshawk tissue sam- ples were collected from Arizona to Alaska, and many of these samples have been analyzed to eval- uate genetic variation in North American gos- hawks. Preliminary data from markers assayed from goshawks nesting in Alaska (coastal and in- terior), British Columbia (coastal and interior). and Utah suggest that genetic differences in pop- ulations will be found as analyses are completed. Western goshawks as a discrete population. In the context of the Endangered Species Act, a discrete population of a vertebrate species is one that sat- isfies at least one of the following conditions: (1) it is markedly separated from other populations of the same taxon as a consequence of physical, phys- iological, ecological, or behavioral factors or (2) it is delimited by international boundaries within which differences in control of exploitation, man- agement of habitat, conservation status, or regu- latory mechanisms exist that are significant in light of section 4(a) (1) (D) of the Act (USDI and United States Department of Commerce 1996). Goshawks that breed in the western and eastern U.S. are part of a continuous population that extends across Canada and into interior Alaska but that is seg- mented by international boundaries (Squires and Reynolds 1997). It was beyond the scope of our charge to assess differences in management of gos- hawks in the U.S. and Canada, and there is cur- rently little evidence of biological differences be- tween goshawks in the eastern and western U.S. Therefore, it is unclear whether goshawks breed- ing in the western and eastern U.S. should be viewed as discrete population segments under fed- eral threatened and endangered species policy. Habitat Relations. Long-term forest-management patterns. It is likely that past and current forest management on public and private lands in the western U.S. has resulted in existing landscapes that are quite different from historical landscapes and their natural range of variation. It was beyond the scope of our charge to project the condition and attributes of future forested landscapes in the western U.S. Clearly, though, forested landscapes that contain goshawk habitat will be necessary to support goshawk populations in the future. In 1998, the FWS (USDI 1998) concluded that cur- rent and projected land-management practices in the review area would not result in landscapes in- capable of supporting goshawks. This conclusion was predicated on both an assessment of future landscape condition and goshawk response to that condition, both of which were speculative. Status of prey populations. Across western North America, goshawks feed on a variety of prey spe- cies, including birds and mammals from small to moderately large in size. Passerines (primarily corvids and thrushes [Turdidae]), woodpeckers (Picidae), Galliformes (grouse, ptarmigan, quail). 198 Andersen et al. VoL. 39, No. 3 tree (Sciurus spp.) and ground squirrels (Spermo- philus spp.), and lagomorphs (including snowshoe hares [Lepus americanus] and cottontail rabbits [5))Z- vtlagus spp.]) are the major prey species. Almost all information (but see Beier and Drennan 1997, Drennan and Beier 2003) regarding prey use of goshawks is derived from studies of nests during the breeding season, and it is based on observa- tions of prey delivered to nests, prey remains col- lected at nests, or pellets and remains collected at nests or plucking perches. These data may primar- ily reflect prey selection by male goshawks, which provide most of the food during pre-incubation through fledging. Further, most studies report on the frequency of prey species pooled across years. Only a few North American studies have assessed annual variation in diet and related it to variation in demographic parameters, such as reproduction (e.g., Keane 1999, Maurer 2000, Doyle and Smith 2001). Diets during winter may differ from diets during the breeding season (e.g.. Widen 1989) be- cause of prey hibernation, goshawk migration, or changes in use of vegetation types by prey species or goshawks in different bioregions. Little infor- mation exists on winter diets for goshawks in west- ern North America (Squires and Reynolds 1997). In the western U.S., most diet studies report that prey associated with late-successional forests are important (Reynolds and Meslow 1984, Kennedy 1991, Reynolds et al. 1992, Keane 1999, Maurer 2000, Lewis 2001), although species associated with other forest age classes or vegetation associations are also used (e.g., Reynolds et al. 1992, Boal and Mannan 1994, Doyle and Smith 1994, Younk and Bechard 1994, Patla 1997, Watson et al. 1998). Al- though a large number of species are usually re- corded in overall summaries of prey species, par- ticular species or a smaller suite of prey species make a relatively greater contribution to total bio- mass and have been associated with temporal var- iation in reproduction. Further, these important prey species, or suites of prey species, vary among bioregions or major vegetation types (Reynolds et al. 1992, Watson et al. 1998, Keane 1999, Doyle and Smith 2001). Although considerable information exists about diet of goshawks during the breeding season, the relations between goshawks and prey abundance, availability, and distribution in the landscape are difficult to study and will not be well understood in the near future, at least at the scale of the west- ern U.S. Considerable additional information re- garding the impacts of future forest conditions in the western U.S. on goshawk prey species is re- quired before goshawk population responses to trends in prey abundance resulting from forest- management practices can be assessed. Association of goshawks with habitat at multiple spa- tial scales. Goshawk-habitat relations have been in- vestigated at a number of spatial and temporal scales. There is general agreement among biolo- gists that goshawk breeding habitat can be dis- cussed in terms of three nested spatial scales: a nest stand (and alternative nest stands; 10-12 ha), a post-fledging area (PFA; 120-240 ha), and a for- aging area (1500-2100 ha; Reynolds et al. 1992). Considerable information exists regarding charac- teristics of nest trees, but comparatively fewer data exist on habitat use outside of the breeding season. Breeding Season. Nest tree. Goshawks build and use nests in a variety of conifer and hardwood tree species. They often use trees that are among the larger or largest in the stand (e.g., Keane 1999). Common nest-tree species include ponderosa pine {Pinus ponderosa) in the southwestern U.S,, Doug- las-fir (Pseudotsuga menziesii) and other conifers in the Rocky Mountains, Sierra Nevada, Pacific Northwest, and Alaska, and aspens (Populus spp.) in portions of the Rockies and interior Alaska. Squires and Reynolds (1997:6) concluded that gos- hawks “tend to nest in a relatively narrow range of vegetation structural conditions,” suggesting that tree species used for nesting is secondary to struc- tural characteristics of the tree and surrounding vegetation. Nest stand. A nest stand is that area covered by a forested patch consisting of trees often character- ized by similar size, species, and spacing, in which a goshawk nest is located. Studies of nests and nest stands have been widespread, covering much of the goshawk’s range in the western U.S. Stands where tree species such as ponderosa pine or lodgepole pine {P. contorta) predominate and stands of mixed conifer species are used for nest- ing. Aspen stands in mountain valleys and draws in the Great Basin of Nevada and Oregon are also used for nesting. Most studies of goshawk nest stands have focused on forest structure (e.g., Reyn- olds et al. 1982, Moore and Henny 1983, Hayward and Escano 1989, Daw et al. 1998) in the vicinity of the nest tree and indicate that large trees and well-developed canopies are important. The spe- cies of tree used for nesting or those that constitute the nest stand appear to be less critical. Goshawks September 2005 Status 199 usually nest in stands of late-successional forest, where trees are often larger than those of other forested stands nearby (e.g., Reynolds et al. 1982). Habitat composition within these nesting stands may include single canopy or multi-story layer components. Forest management that fragments and reduces the extent and area of stands suitable for nesting in a breeding area may result in its less consistent use for nesting over time (e.g., Wood- bridge and Detrich 1994, Desimone 1997). Across the western U.S. and Alaska, many studies have documented goshawks selecting nest stands that are more mature or consist of late-successional forest compared with random assessments of near- by forest habitat, irrespective of scale of analysis (e.g., Moore and Henny 1983, Crocker-Bedford and Chaney 1988, Desimone 1997, Keane 1999). Some studies have suggested that high-canopy clo- sure is one of the more uniform characteristics of goshawk nest stands (Hayward and Escano 1989, Keane 1999), and others have documented that a higher percent canopy closure was associated with a higher probability that goshawks would nest in a stand (Crocker-Bedford and Chaney 1988). Cano- py closure in nest stands is variable across North America, and in some regions of the western U.S. and Alaska mean canopy closure near the nest might be rather low (ca. 50% in parts of Oregon and Washington [McGrath 1997] and southeastern Alaska [Iverson et al. 1996]). Differences in sam- pling methods probably account for some of this apparent inconsistency because measurement of canopy closure has not been conducted consis- tently among studies (Crocker-Bedford and Cha- ney 1988). However, even where canopy closure around a nest area is apparently low, it is still gen- erally higher than the surrounding portions of the stand or other nearby stands. This suggests that high-canopy closure relative to the range of avail- able canopy closure might be more important than absolute canopy closure, at least above some min- imum threshold. Why goshawks select stands with relatively larger trees and higher canopy cover is not known. Po- tential hypotheses include; (1) increased protec- tion from predators, (2) increased food availability, (3) reduced exposure to cold temperatures and precipitation during the energetically stressful pre- laying period in late winter-early spring, (4) re- duced exposure to high temperatures during the summer nestling period, (5) reduced competition with raptor species that nest in more open envi- ronments (e.g., Red-tailed Hawk {Buteo jamaicen- ^A]), or (6) increased mobility because of reduced understory vegetation in mature stands. Use area-home range. How goshawks use habitats away from their nests during the nesting season is not well understood. Methods to evaluate gos- hawk-habitat associations at the home-range scale fall into a few different categories, including: (1) habitat evaluations based on circular areas cen- tered on the nest that are often made using aerial photography or other remote sensing methods and Geographic Information Systems, (2) habitat-selec- tion studies using radiotelemetry, (3) evaluating hunting habitat use with radiotelemetry and direct observation, and (4) evaluating patterns associated with habitat disturbance and logging versus fre- quency of nesting. Most studies of habitat use based on a nest-cen- tered evaluation have loosely linked the scale of measurement to a nest stand, PFA, or mean home- range size. In general, the preponderance of late- successional forest in the landscape decreases as the scale increases (i.e., as one moves from nest stand to PFA to foraging area; Iverson et al. 1996, Finn 2000, Daw and DeStefano 2001, Finn et al. 2002, McGrath et al. 2003). Radiotelemetry studies to evaluate habitat use within the home range during the nesting season have found that goshawks selected for late-succes- sional forests even beyond their nest stands (Widen 1989, Austin 1993, Bright-Smith and Mannan 1994, Hargis et al. 1994, Iverson et al. 1996, Beier and Drennan 1997). Goshawks used larger stands of late-successional forest than was available in south- eastern Alaska (Iverson et al. 1996, Pendleton et al. 1998) and Sweden (Widen 1989); in Arizona, some goshawks selected for late-successional forest >200 m from openings (Bright-Smith and Mannan 1994) . In California, goshawk locations had greater basal area, canopy cover, and more large trees than did random points (Austin 1993, Hargis et al. 1994). These results suggest a fine-scale selection for larger stands of mature forests within goshawk nesting-season home ranges. Presumably, vegetative characteristics associated with foraging sites influence prey availability. For example, Beier and Drennan (1997) concluded that goshawks in Arizona did not select foraging sites based on prey abundance; rather, they select- ed sites based on vegetation structure. Goshawk foraging locations had a higher canopy closure, greater tree density, more large trees, and fewer 200 Andersen ex al. VoL. 39, No. 3 shrubs and saplings than random reference plots. There was also selection for dense stands with high canopy closure that were rare on their study area. Widen (1989) had previously reported that in Eu- rope, hunting sites were associated with habitat structure and did not seem to be related to abso- lute prey abundance. A number of authors have noted that foraging sites typically are characterized by open space between the bottom of the canopy and the top of the shrub layer (e.g., Reynolds 1989, Widen 1989, Crocker-Bedford 1990, 1998, Beier and Drennan 1997) and have speculated that this space may increase prey vulnerability by providing a flight path for foraging goshawks. Results of several studies suggest that goshawks are more likely to reoccupy breeding areas within landscapes that have larger proportions of late-suc- cessional forest, compared with landscapes that have smaller proportions of these forests (Ward et al. 1992, Woodbridge and Detrich 1994, Daw 1997, Patla 1997, Finn 2000, Finn et al. 2002). Widen (1997) concluded that intensive forest manage- ment was the prime factor in reductions in gos- hawk breeding density across nine study areas in Scandinavian boreal forests. Assessing habitat use at the home-range-use area scale has several important limitations, includ- ing small sample sizes, variation in fecundity, and the small range of vegetation types in which these studies have been conducted. In addition, consid- erable variation likely exists among home range- use areas, with some use areas consistently produc- ing young, and others only occasionally producing young (Newton 1989, Joy 2002, McClaren et al. 2002). Thus, habitat evaluations that are not relat- ed to productivity and population dynamics might have limited utility. Including use areas that rarely produce young in these evaluations might make it difficult to identify characteristics of use areas as- sociated with high-quality habitat. Finally, habitat use at the home-range scale has been assessed in only a few vegetation types, limiting inference to scales below that of the western U.S. Clearly, ad- ditional information is necessary to better assess habitat use patterns at the scale of home range- use areas. Non-nesting season. There are few studies of gos- hawk-habitat associations during the non-nesting season in North America. Iverson et al. (1996) ex- amined year-round habitat selection by radio- tagged adult goshawks in southeastern Alaska with- in their seasonal use area and found no differences in habitat selection between the nesting season and non-nesting season. Adult goshawks selected for larger size classes of late-successional conifer- ous forest compared with other habitat cover types. Beier (1997) and Drennan and Beier (2003) ex- amined winter foraging habitat of adult goshawks in northern Arizona and found that goshawk lo- cations were in areas with a slightly higher medi- um-size tree density and higher canopy cover than contrast plots. Females remained in the ponderosa pine vegetation type, and most males moved to pinyon-juniper (Pinus-Juniperus) woodlands. Some goshawks move to open or scrub habitats in the winter (Squires and Ruggiero 1995), while others seem to remain in forested areas, making it diffi- cult to generalize across populations in terms of goshawk winter-habitat use. Summary of goshawk habitat use. Goshawks have broad geographic and elevational distributions in North America and can be found in many different forest types and forest stand conditions (Squires and Reynolds 1997). Goshawks have relatively large home ranges, are able to move great distances— es- pecially dnring times of low prey abundance, and use a wide variety of prey species across the range of landscapes in which they occur. Goshawks tend to nest in forest stands with specific structural char- acteristics, such as stands with large trees and mod- erate to high canopy closure that is high relative to the range of available canopy closure. Goshawks forage in a variety of habitats, ranging from early- successional forests, to mature forests, to open hab- itats adjacent to forested habitats. During the breeding season, late-successional forests appear to be used predominantly for foraging, although some of the prey taken by goshawks use young for- ests and open habitats. Goshawk breeding habitat can be discussed in terms of three nested, spatial scales: a nest stand (and stands containing alternative nests), within a PFA, and within a foraging area. At the nest-stand scale, late-successional forest characteristics are of- ten important determinants of where goshawks lo- cate their nests. The preponderance of late-succes- sional forest in the landscape decreases as the scale increases (e.g., as one moves from nest stand to PFA to foraging area), and existing data from te- lemetry and observational studies suggest that gos- hawks use late-successional forests within their home ranges for foraging, but use prey associated with both early- and late-successional forests, and in some cases, open habitats. Thus, goshawks ap- September 2005 Status 201 pear to be associated with late-successional forests for nesting and foraging, but clearly also use, and use prey associated with, other cover types. Gos- hawk breeding habitat has been studied much more intensively than nonbreeding habitat. In some landscapes, goshawks appear to remain near breeding areas throughout the year, although there is considerable annual variation and varia- tion between sexes in nonbreeding habitat use. In at least some landscapes, goshawks forage in late- successional forest habitats throughout the year. Conversely, some goshawks use landscapes during the nonbreeding season (e.g., pinyon^uniper and open sagebrush basins) that are quite different from landscapes used during the breeding season. In general, there appears to be a wider range of habitats used during the non-breeding season than during the breeding season. Habitat as a Surrogate for Population Trends. Context. The population status of goshawks and their association with late-successional forests in western North America has been debated for >10 yr. This debate has considerable bearing on the FWS decision that listing goshawks in the western U.S. under the Endangered Species Act was not warranted (USDI 1998). In 1990, Crocker-Bedford (1990) reported a relation between timber harvest and loss of goshawk territories on the Kaibab Pla- teau in Arizona and suggested that some forest- management practices might negatively affect gos- hawk populations. Considerable discussion of that conclusion and the evidence supporting it ensued. Kennedy (1997) later reviewed the status of gos- hawks and concluded that data were lacking to de- termine if populations of goshawks were increas- ing, decreasing, or stationary. She called for more in-depth demographic studies, including meta- analysis approaches, combining ongoing studies with marked goshawks. Smallwood (1998) and Crocker-Bedford (1998) both responded to Ken- nedy’s review paper. Smallwood (1998) suggested that in lieu of appropriate sampling and agree- ment among scientists regarding additional vari- ables that should be analyzed, evidence for a gos- hawk population decline should be based on availability and contiguity of habitat and migration counts. Crocker-Bedford (1998) hypothesized that distribution of foraging habitat across the land- scape influenced goshawk home-range size, which in turn influenced breeding pair density and re- productive success. He suggested further devel- opment of models of goshawk-habitat relations, inventory of current forest conditions, and assess- ment of population status based on habitat condi- tions at the landscape level. In their status review (FWS 1998), the FWS at- tempted to assess population status from popula- tion data and also by using the distribution and extent of habitat, particularly older forest (specifi- cally old-growth), as a surrogate for a direct mea- sure of population trends. This effort represented the largest concerted attempt to date to document goshawk locations and habitat in North America. The FWS concluded that it was evident that “there [are] inadequate data available which could be used to determine the population trend for gos- hawks throughout the review area. Furthermore, our knowledge of the factors that affect the size of goshawk populations at local and regional levels, or in the entire area is incomplete. A clearer un- derstanding of population size and factors affect- ing goshawk populations is needed. Much of what is known is currently applicable only to local pop- ulations and localized habitat conditions and ef- fects, and should not be extrapolated to the larger range of the species” (FWS 1998). The FWS also noted that few studies have focused on goshawk population dynamics over a sufficient period of time to provide the kinds of demographic data needed for a status review. With this realization, FWS attempted to identify trends in habitat. The FWS concluded that they could not directly tie changes in goshawk populations to changes in hab- itat over time because of a lack of data and little confidence regarding how goshawk populations re- spond to changes in their habitat. The FWS deci- sion that listing goshawks in the western U.S. un- der the Endangered Species Act was not warranted was based in large part on lack of evidence that habitat was currently limiting goshawks, and that habitat was unlikely to limit the goshawk popula- tion in the review area in the foreseeable future. Such an approach is clearly limited by how well the relations between goshawks and their habitat are understood, and how well existing vegetative con- ditions are known. Existing goshawk-habitat models. Warren et al. (1990) developed a goshawk-habitat model based on a review of published and unpublished litera- ture and expert opinion using the Delphi method. In their model, habitat suitability increased with increasing canopy cover, size of overstory trees, size of the nest stand, and decreasing slope. Suitability of foraging habitat was modeled in relation to prey 202 Andersen et al. VoL. 39, No. 3 availability, forest type, and tree species composi- tion. Reynolds et al. (1992) synthesized habitat as- sociations for goshawks and 14 prey species and silvicultural prescriptions designed to produce suit- able forest conditions for goshawks and their prin- cipal prey in the southwestern U.S. Such prescrip- tions were developed with the intent of (1) sustaining goshawk populations in the Southwest, (2) providing desired forest conditions for the gos- hawk and its prey, (3) using the natural, presettle- ment forest composition, structure, and landscape pattern of each forest type as a template for assem- bling, and assuring the sustainability of, goshawk and prey habitats in large landscapes, and (4) man- aging southwestern forests as an ecosystem (i.e., re- taining all of the parts) . For the goshawk, this is a conceptual model, but the recommendations that came from this model are being implemented on national forests throughout the Southwest while components of the model are being implemented throughout much of the western U.S. and in Brit- ish Columbia, Canada. The model of Reynolds et al. (1992) has served as the primary model for gos- hawk management in the southwestern U.S. (Reyn- olds et al. 1996, Long and Smith 2000) and has been the subject of considerable debate and eval- uation (e.g., Braun et al. 1996). In Utah, Johansson et al. (1994) used elevation and vegetation models to predict potential gos- hawk nesting sites. They found elevation to be a better predictor of goshawk nest locations than vegetation, although elevation, vegetation, and veg- etative characteristics of PFAs were the best predic- tors overall. In Idaho, Lilieholm et al. (1994) ap- plied a stand density index (SDI) — a measure of stand density that is based on mean tree size and density and is comparable among stands — to guide management practices intended to create forest conditions similar to those found in goshawk nest areas. Although this latter method was primarily intended to assist silviculturalists in managing for- est stands, mean tree size and density of stands rep- resenting goshawk habitat (e.g., goshawk nest areas) can be used as models of desired future con- ditions. Similarly, Graham et al. (1994) pointed out that the way forests regenerate, develop, and die is highly variable in time and space, and recom- mended managing large tracts of forests as sustain- able ecological units rather than managing smaller tracts as individual home ranges. DeStefano (1998) suggested that goshawk occurrence was related to characteristics associated with late-successional for- est, but that goshawks are found in a wide variety of forest conditions. Crocker-Bedford (1998) hy- pothesized that distribution of foraging habitat across the landscape influences goshawk home- range size, which in turn influences breeding pair density and reproductive success. Landscapes that contain a higher concentration of foraging habitat with adequate prey abundance should support higher densities of breeding goshawks. Joy (2002) developed spatial-simulation models to assess the relations between goshawk habitat composition and structure and the location of nests and use areas and the relations between the amount and arrangement of habitat components in high- and low-quality breeding areas. High- and low-quality breeding areas were distinguished based on long-term (10 yr) demographic data from 101 breeding areas in northern Arizona. Joy (2002) found that intraspecific territoriality plays a more significant role in nest location than avail- ability of nest area habitat on the Kaibab Plateau. In addition to using habitat models to identify spa- tial and compositional differences between gos- hawk nests and random locations, Joy (2002) and Reich et al. (2004) used these models to predict nest locations likely to have high reproductive out- put. McGrath et al. (2003) developed models relating habitat characteristics around goshawk nest sites at scales from 1-170 ha in eastern Oregon and Wash- ington. At the 1-ha scale, structural stage (i.e., late- seral), topographic position (i.e., lower slopes and drainage bottoms), and stand-basal area (i.e., high basal area) were associated with goshawk nests, with high basal area being the most important. At larger scales (10-170 ha), later serai stages, high understory growth, and high canopy closure were associated with nests and these associations were prevalent up to 83 ha. They concluded that: (1) there is a core area around goshawk nests where the forest is generally mid- to late-successional stage (large trees with high canopy closure) and (2) this core is surrounded by diverse types of for- est cover that are equally abundant (i.e., no one cover type dominates). In summary, most existing models of goshawk- habitat relations are limited to vegetative structure used for nesting. Other habitat variables (such as microclimatic conditions at nest, foraging, or roost sites) and other aspects of life history (such as ju- venile dispersal and territory establishment, non- breeding or failed breeding adults, and winter September 2005 Status 203 ecology) have received relatively little attention compared to vegetative structure around nests, largely because of the difficulties in working with goshawks in the field. Limitations on using current goshawk— habitat models for predicting population status. Currently, the rela- tions between goshawks and their habitat in the western U.S, are not understood well enough to use trends in habitat as a surrogate for trends in goshawk populations. Fundamentally, this is be- cause there are unknown functional relations among the amounts and distribution of goshawk habitat, the range of vegetation conditions that characterize goshawk habitat, and goshawk popu- lation densities and population dynamics. There- fore, it is not currently possible to predict how changes in habitat, or changes in specific types of vegetation such as old-growth forests, are related to changes in goshawk population densities or trends. The use of late-successional forests (specif- ically old-growth forest) as a surrogate for goshawk population status is limited because: (1) goshawks show a high degree of versatility in habitat use, and although late-successional forest is a commonly used habitat, other serai stages also are used; thus, reliance on distribution of late-successional forests alone for determining the status and distribution of goshawks in the western U.S. is not sufficient; (2) important prey species vary among bioregions and major vegetation types with late-successional forest associates (e.g., Douglas [Tamiasciurus doug- lasii\ and red squirrels [T hudsonicus^) important in some regions and early-seral species (e.g., snow- shoe hares) relatively more important in other re- gions; (3) there is currently no consistent defini- tion of old-growth forest as it pertains to goshawk habitat that can be applied across the entire west- ern U.S. or at the scale of major vegetation types; (4) habitat may not be occupied if factors other than old-growth vegetative structure (e.g., weather, prey availability) are limiting goshawk populations; and (5) large-scale, regional vegetation mapping efforts (e.g., msyor portions of the western U.S.) are not sufficiently precise or accurate to assess current or future conditions. Multiple factors influ- ence habitat use, especially on very large spatial or temporal scales, and relations between goshawks and habitats and goshawks and prey species, seem to be variable across vegetation types. Knowledge concerning the functional relation between the distribution and abundance of habitat and gos- hawk population densities and trends is required in order to draw scientifically defensible inferences regarding how changes in habitat, or specific hab- itat types such as old-growth, relate to changes in goshawk populations. Currently this relation is un- known. Recommendations To assess goshawk population status in the west- ern U.S. or any other portions of this bird’s range in North America, several improvements in exist- ing data-collection efforts and protocols are nec- essary. Additional data that do not currently exist will also need to be collected before adequate pop- ulation status assessment can take place in the west- ern U.S. Items we identified include: (1) Compilation and accessibility of existing data. We urge organization of existing data into a for- mat that would make it readily accessible to management agencies and other interested parties. Development of standardized proto- cols for future monitoring and inventory data collection will benefit from an assessment of the existing information. In addition, devel- opment of procedures to systematically and regularly capture new information to maintain a current database is necessary. (2) Sampling strategy. Outside of intensive research studies, most existing goshawk distributional or occurrence records are based on ad hoc sampling generally associated with manage- ment activities. If goshawk population trends are to be assessed, sampling must represent the target population and yield defensible trend estimates. Monitoring approaches should be based on sample designs that ad- dress the definition of the target population, appropriate response variable, definition of a sampling frame and primary sample units, is- sues of probability of detection, and estimates of necessary sample sizes required to detect a specific change. Monitoring strategies should also be designed to assess both population trend and habitats, as defined through devel- opment of empirical goshawk— habitat rela- tions models. Land managers and agency de- cision-makers should recognize that continued funding of uncoordinated, small-scale goshawk monitoring efforts will not yield useful results across a large land area. In addition, it may be fruitful to address population status at a scale smaller than that of the review area. Rather 204 Andersen et al. VoL. 39, No. 3 than evaluating goshawk population status for the entire western U.S., consideration should be given to monitoring trends in goshawk pop- ulations and habitat at the ecoregion or biome scale (e.g., Sierra Nevada forests; coastal tem- perate forests and rainforests of Oregon, Washington, and southern coastal British Co- lumbia; ponderosa pine forests of New Mexi- co, Arizona, and southern Colorado). (3) Relation of populations and subspecies. We rec- ommend that variation in DNA be used to as- sess the phylogenetic relations among eastern and western North American A. g. atricapillus, and atricapillus to A. g. laingi, and to the pu- tative A. g. apache. (4) Addressing current limitations of existing data sources. Potentially useful data are currently limited by a lack of knowledge about popula- tion affiliation (e.g., migration counts), small sample sizes (e.g.. Breeding Bird Survey data), or inadequate sampling strategies. Consider- ation should be given to addressing these lim- itations where possible. For example, in the case of migration counts, population affilia- tion of goshawks counted at migration sites needs to be determined, perhaps through con- servation genetic and stable isotope analyses (e.g., Meehan et al. 2001). (5) Standardization of terminology^ and protocols asso- ciated with estimating breeding status and produc- tivity. We recommend that researchers and land managers cooperate in developing stan- dardized protocols (including terminology and data-collection methods) based on peer- reviewed literature with the specific intention of performing pooled data analysis across the entire review area at a later date (e.g., Ander- son et al. 1999). If a single set of protocols cannot be used for the entire western U.S., then standardized protocols should be used for large areas (e.g., biomes or ecological hab- itat types, but not political boundaries) . (6) Research priorities. To assess demography and population trends adequately, goshawk-habitat relations and the effects of specific land-man- agement practices on goshawks in the western U.S., coordinated studies of habitat use (possi- bly using radiotelemetry) are necessary. Studies of demography and habitat use also need to address the nonbreeding season, when factors regulating populations may be important. In addition, land managers need to continue to work on remote-sensing applications so that broad-scale analysis of habitats such as late-suc- cessional forest and patch size can be evaluated. Finally, long-term experimental or quasi-exper- imental studies are necessary at the landscape scale to understand how forest management in- fluences goshawks. These studies will be most beneficial when accomplished using an inter- disciplinary approach in close collaboration with land managers. An integrated approach between research and management consisting of extensive population and habitat monitoring at the bioregional scale coupled with intensive, long-term demography studies in each of the major vegetation types will provide the data necessary to monitor goshawk populations and habitat and to generate a scientific understand- ing of goshawk ecology needed to improve management and conservation efforts. Finally, we emphasize that in addition to assess- ing population trends and status in the western U.S., it is also important to better understand gos- hawk-habitat relations and the influence of various human activities, especially forest-management practices, on goshawks. Much of the controversy regarding goshawk conservation in the western U.S. and elsewhere has to do with concerns about forest management and how forest management affects goshawks. Thus, it is not sufficient to simply assess goshawk population trends in the western U.S.; it is also necessary to better understand the relations between goshawks and their habitat and how human activities affect that habitat. Consid- erable information regarding population ecology and goshawk-habitat relations currently exists, but additional information is necessary. Individual gos- hawks or goshawk pairs exhibit landscape-level use of space, and thus, occur naturally at relatively low densities. They are highly mobile, and as such, have proved difficult to study. Thus, a long-term investment of resources in a coordinated effort di- rected at large spatial scales will be required to as- sess goshawk population trends adequately and un- derstand goshawk-habitat relations in the western U.S. and elsewhere. Acknowi.egments We express our thanks to Patricia L. Kennedy (Oregon State University) and Richard T. Reynolds (U.S. Forest Service) for critically reading a prior draft of our report and for offering substantive suggestions for improve- ment. Dan Edge (TWS), Winni Kessler (TWS), Len Car- September 2005 Status 205 penter (TWS), Jim Bednarz (RRF), and Brian Millsap (RRF) patiently worked with the committee to meet its charge. We also thank Board members of RRF, especially Ted Swem, and Council members of TWS, especially Mac Baughman, who read this report and offered constructive suggestions. Krista E.M, Galley provided copy editing of the final report published by TWS. Clayton M. White and R. William Mannan provided constructive reviews, and Patricia L. Kennedy served as an insightful editor for this manuscript. Literature Cited American Ornithologists’ Union (AOU). 1957. Check- list of North American birds. 5th Ed. American Or- nithologists’ Union, Washington, DC U.S.A. Anderson, D.R., K.P. Burnham, A.B. Franklin, RJ. Gu- tierrez, E.D. Forsman, R.G. Anthony, G.G. White, AND T.M. Shenk. 1999. A protocol for conflict reso- lution in analyzing empirical data related to natural resource controversies. Wildl. Soc, Bull, 27:1050-1058. Austin, K.K. 1993. Habitat use and home range size of breeding Northern Goshawks in the southern Cas- cades. M.S. thesis, Oregon State Univ., Corvallis, OR U.S.A. Baird, S.F., T.M. Brewer, and R. Ridgway. 1875. A history of North American birds. Vol. 3. Land birds. Little, Brown, Boston, MA U.S.A. Bednarz, J.C., D. Klem, Jr., LJ. Goodrich, and S.E. Sen- NER. 1990. Migration counts of raptors at Hawk Moun- tain, Pennsylvania, as indicators of population trends, 1934-86. Auk 107:96-109. Beier, P. 1997. Winter foraging habitat of Northern Gos- hawks in northern Arizona. Final report. Arizona Game and Fish Department Heritage Grant 195032, Pheonix, AZ U.S.A. and J.E. Drennan. 1997. Forest structure and prey abundance in foraging areas of Northern Gos- hawks. Ecol. Appl. 7:564-571. Bildstein, K.L. 1998. Long-term counts of migrating rap- tors: a role for volunteers in wildlife research./. Wildl. Manag. 62:435-445. Boal, C.W. and R.W. Mannan. 1994. Northern Goshawk diets in ponderosa pine forests on the Kaibab Plateau. Stud. Avian Biol. 16:97-102. Bosakowski, T. 1999. The Northern Goshawk: ecology, behavior, and management in North America. Han- cock House, Blaine, WA U.S.A. Braun, C.E., J.H. Enderson, M.R. Fuller, Y.B. Linhart, AND C.D. Marti. 1996. Northern Goshawk and forest management in the southwestern United States. The Wildlife Society, Bethesda, MD U.S.A. Bright-Smith, D.J. and R.W. Mannan. 1994. Habitat use by breeding male Northern Goshawks in northern Ar- izona. Stud. Avian Biol. 16:58—65. Crocker-Bedford, D.C. 1990. Goshawk reproduction and forest management. Wildl. Soc. Bull. 18:262-269. . 1998. The value of demographic and habitat studies in determining the status of Northern Gos- hawks {Accipiter gentilis atricapillus) with special refer- ence to Crocker-Bedford (1990) and Kennedy (1997). J. Raptor Res. 32:329-336. AND B. Chaney. 1988. Characteristics of goshawk nesting stands. Pages 210-217 in R. Glinski, B.G. Pen- dleton, M.B. Moss, B.A. Millsap, and S.W. Hoffman [Eds.], Southwest raptor management symposium and workshop. Natl. Wildl. Fed. Sci. Tech. Ser. 11. Daw, S.K. 1997. Northern Goshawk nest site selection and habitat associations at the post-fledging family area scale in Oregon. M.S. thesis, Oregon State Univ., Corvallis, OR U.S.A. and S. DeStefano. 2001. Forest characteristics of Northern Goshawk nest stands and post-fledging ar- eas in Oregon. J. Wildl. Manag. 65:59-65. , S. DeStefano, and RJ. Steidl. 1998. Does survey method bias the description of Northern Goshawk nest-site structure? J. Wildl. Manag. 62:1379-1384. Desimone, S.M. 1997. Occupancy rates and habitat rela- tionships of Northern Goshawks in historic nesting areas in Oregon. M.S. thesis, Oregon State Univ., Cor- vallis, OR U.S.A. DeStefano, S. 1998. Determining the status of Northern Goshawks in the west: is our conceptual model cor- rect? /. Raptor Res. 32:342-348. , S.K. Daw, S.M. Desimone, and E.C. Meslow 1994a. Density and productivity of Northern Gos- hawks: implications for monitoring and management. Stud. Avian Biol. 16:88-91. , B. Woodbridge, and P. Detrich. 1994b. Survival of Northern Goshawks in the southern Cascades of California. Stud. Avian Biol. 16:133-136. Doyle, F.I. and J.M.N. Smith. 1994. Population responses of Northern Goshawks to the 10-year cycle in num- bers of snowshoe hares. Stud. Avian Biol. 16:122-129. and . 2001. Raptors and scavengers. Pages 377-404 in C.J. Krebs, S. Boutin, and R. Boonstra [Eds.], Ecosystem dynamics of the boreal forest: the Kluane Project. Oxford Univ. Press, New York, NY U.S.A. Drennan, J.E, and P. Beier. 2003. Forest structure and prey abundance in winter habitat of Northern Gos- hawks./. Wildl. Manag. 67:177-185. Finn, S.P. 2000. Multi-scale habitat influences on North- ern Goshawk occupancy and reproduction on the Olympic Peninsula, Washington. M.S. thesis, Boise State Univ., Boise, ID U.S.A. , J.M. Marzluff, and D.E. Varland. 2002. Effects of landscape and local habitat attributes on Northern Goshawk site occupancy in western Washington. For Sci. 48:427-436. Flatten, C., K. Titus, and R. Lowell. 2001. Northern Goshawk monitoring, population ecology and diet on the Tongass National Forest: annual monitoring re- port. Alaska Department of Fish and Game, Juneau, AK U.S.A. Franklin, A.B. 1992. Population regulation in northern 206 Andersen et al. VoL. 39, No. 3 Spotted Owls: theoretical implications for manage- ment. Pages 815-827 in D.R. McCullough and R.H. Barrett [Eds.], Wildlife 2001: populations. Elsevier Applied Science, London, U.K. Gavin, T.A. and B. May. 1996. Genetic variation and tax- onomic status of Northern Goshawks in Arizona: im- plications for management. Unpublished report sub- mitted to Arizona Game and Fish Department. Cornell Univ., Ithaca, NY U.S.A. Graham, R.T., R.T. Reynolds, M.H. Reiser, R.L. Bassett, AND D.A. Boyce. 1994. Sustaining forest habitat for the Northern Goshawk; a question of scale. Stud. Avi- an Biol. 16:12-17. Grinnell, J. AND A.H. Miller. 1944. The distribution of the birds of California. Pacific Coast Avifauna 27. Hall, L.S., P.R. Krausman, and M.L. Morrison. 1997. The habitat concept and a plea for standard termi- nology. Wildl. Soc. Bull. 25:173-182. Hargis, C.D., C. McCarthy, and R.D. Perloff. 1994. Home ranges and habitats of Northern Goshawk in eastern California. Stud. Avian Biol. 16:66-74. Hayward, G.D. and R.E. Escano. 1989. Goshawk nest site characteristics in western Montana and northern Ida- ho. ConcJor 91:476-479. Hunt, W.G. 1998. Raptor floaters at Moffat’s equilibrium. Oikos 82:191-197. Ingraldi, M.F. 1998. Population biology of Northern Goshawks in east-central Arizona. Nongame and En- dangered Wildlife Program Technical Report 133. Ar- izona Game and Fish Department, Phoenix, AZ U.S.A. Iverson, G.C., G.D. Hayward, K. Titus, E. DeGavner, R.E. Lowell, D.C. Crocker-Bedford, P.F. Schempf, andJ. Lindell. 1996. Conservation assessment for the Northern Goshawk in southeast Alaska. General Tech- nical Report PNW-GTR-387. U.S. Department of Ag- riculture, Forest Service, Pacific Northwest Research Station, Portland, OR U.S.A. Johansson, C., P.J. Hardin, and C.M. White. 1994. Large-area goshawk habitat modeling in Dixie Nation- al Forest using vegetation and elevation data. Stud. Avian Biol. 16:50-57. Joy, S.M. 2002. Northern Goshawk habitat on the Kaibab National Forest in Arizona: factors affecting nest lo- cations and territory quality. Ph.D. dissertation, Col- orado State University, Fort Collins, CO U.S.A. , R.T. Reynolds, and D.G. Leslie. 1994. Northern Goshawk broadcast surveys: hawk response variables and survey cost. Stud. Avian Biol. 16:24-31. Keane, J.J. 1999. Ecology of the Northern Goshawk in the Sierra Nevada, California. Ph.D. dissertation, Univ. California, Davis, CA U.S.A. Kennedy, P.L. 1991. Reproductive strategies of Northern Goshawks and Cooper’s Hawks in north-central New Mexico. Ph.D. dissertation, Utah State Univ., Logan, UT U.S.A. . 1997. The Northern Goshawk {Accipiter gentilis atricapillus) : is there evidence of a population decline? J. Raptor Res. 31:95-106. . 1998. Evaluating Northern Goshawk {Accipiter gentilis atricapillus) population trends: a reply to Small- wood and Crocker-Bedford. J. Raptor Res. 32:336-342. AND D.W. Stahlecker. 1993. Responsiveness of nesting Northern Goshawks to taped broadcasts of three conspecific calls./. Wildl. Manag. 57:249-257. Kostrzewa, a. and R. Kostrzewa. 1990. The relationship of spring and summer weather with density and breeding performance of the buzzard Buteo buteo, gos- hawk Accipiter gentilis and kestrel Falco tinnunculus. Ibis 132:550-559. Lewis, S.B. 2001. Breeding season diet of Northern Gos- hawks in southeast Alaska with a comparison of tech- niques used to examine raptor diet. M.S. thesis, Boise State Univ., ID U.S.A. Lilieholm, RJ. J .N. Long, and S. Patia. 1994. Assess- ment of goshawk nest area habitat using stand density index. Stud. Avian Biol. 16:18-23. Long, J.N. and F.W. Smith. 2000. Restructuring the for- est: goshawks and the restoration of southwestern ponderosa pine. J. For. 98:25-30. Maguire, L. and D. Call. 1993. Population viability anal- ysis of Northern Goshawks {Accipiter gentilis) on the North Kaibab Ranger District, Arizona. Contract No. G10076-A. Arizona Game and Fish Department, Phoe- nix, AZ U.S.A. Maurer, J.R. 2000. Nesting habitat and prey relations of the Northern Goshawk in Yosemite National Park, California. M.S. thesis, Univ. California, Davis, CA U.S.A. McClaren, E.L., P.L. Kennedy, and S.R. Dewey. 2002. Do some Northern Goshawk nest areas consistently fledge more young than others? Condor 104:343-352. McGrath, M.T. 1997. Northern Goshawk habitat analysis in managed forest landscapes. M.S. thesis, Oregon State Univ., Corvallis, OR U.S.A. , S. DeStefano, R.A. Riggs, L.L. Irwin, and G.J. Roloff. 2003. Spatially explicit influences on North- ern Goshawk nesting habitat in the interior Pacific Northwest. Wildl. Monogr. 154:1-63. Meehan, T.C., C.A. Lott, Z.D. Sharp, R.B. Smith, R.N. Rosenfield, A.C. Stewart, and R.K. Murphy. 2001. Using hydrogen isotope geochemistry to estimate the natal latitudes of immature Cooper’s Hawks migrating through the Florida Keys. Condor 103:11—20. Moore, K.R. and C.J. Henny. 1983. Nest site character- istics of three coexisting accipiter hawks in northeast- ern Oregon. Raptor Res. 17:65-76. Mueller, H.C., D.D. Berger, and G. Allez. 1976. Age and sex variation in the size of goshawks. Bird-Banding 47:310-318. , , and . 1977. The periodic invasions of goshawks. Auk 94:652-663. Newton, I. 1989. Sparrowhawk. Pages 279-296 ini. New- September 2005 Status 207 ton [Ed.], Lifetime reproduction in birds. Academic Press, London, U.K. . 1991. Habitat variation and population regula- tion in sparrowhawks. Ibis 133 (Supplement I):76-88. Patla, S.M. 1997. Nesting ecology and habitat of the Northern Goshawk in undisturbed and timber harvest areas on the Targhee National Forest, Greater Yellow- stone Ecosystem. M.S. thesis, Idaho State Univ., Po- catello, ID U.S.A. Pendleton, G.W., K. Titus, E. DeGavner, C.J. Flatten, AND R.E. Lowell. 1998. Compositional analysis and GIS for study of habitat selection by goshawks in Southeast Alaska. J. Agric. Biol. Environ. Stat. 3:280— 295. POSTUPAI.SKY, S. 1974. Raptor reproductive success: some problems with methods, criteria, and terminology. Pages 21—31 mF.N. Hamerstrom Jr., B.E. Harrell, and R.R. Olendorff [Eds]. Raptor Research Report No. 2. Management of raptors. Raptor Research Founda- tion, Inc., Vermillion, SD U.S.A. Pradel, R. 1996. Utilization of capture-mark-recapture for the study of recruitment and population growth rate. Biometrics 52:703-709. Reich, R.M., S.M. Joy, and R.T. Reynolds. 2004. Predict- ing the location of Northern Goshawk nests: model- ing the spatial dependency between nest locations and forest structure. Ecol. Model. 176:109-133. Reynolds, R.T. 1989. Accipiters. Pages 92-101 in B.A. Pendleton, C.E. Ruibal, D.L. Krahe, K. Steenhof, M.N. Kockert, and M.N. LeFranc, Jr. [Eds.], Proceedings of the western raptor management symposium and workshop. National Wildlife Federation Scientific and Technical Series No. 12. , W.M. Block, and D.A. Boyce, Jr. 1996. Using ecological relationships of wildlife as templates for re- storing southwestern forests. Pages 35-43 in W. Cov- ington and P.K. Wagner, technical coordinators. Con- ference on adaptive ecosystem restoration and management: restoration of Cordilleran conifer land- scapes in North America. General Technical Report RM-GTR-278. U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Col- lins, CO U.S.A. , R.T. Graham, M.H. Reiser, R.L. Bassett, P.L. Kennedy, D.A. Boyce, Jr., G. Goodwin, R. Smith, and E.L. Fisher. 1992. Management recommendations for the Northern Goshawk in the southwestern United States. General Technical Report RM-217. U.S. De- partment of Agriculture, Forest Service, Rocky Moun- tain Forest and Range Experiment Station, Fort Col- lins, CO U.S.A. and S.M. Joy. 1998. Distribution, territory occu- pancy, dispersal, and demography of Northern Gos- hawks on the Kaibab Plateau, Arizona. Final report for Heritage Project No. 194045. Arizona Game and Fish Department, Phoenix, AZ U.S. A. and E.C. Meslow. 1984. Partitioning of food and niche characteristics of coexisting Accipiter during breeding. Auk 101:761-779. , E.C. Meslow, and H.M. Wight. 1982. Nesting habitat of coexisting Accipiter in Oregon. /. Wildl, Man- ag. 46:124-138. Shaffer, H.B., R.N. Fisher, and C. Davidson. 1998. The role of natural history collections in documenting species declines. Trends Ecol. Evol. 13:27—30. Smallwood, K.S. 1998. On the evidence needed for list- ing Northern Goshawks {Accipiter gentilis) under the Endangered Species Act: a reply to Kennedy. J. Raptor Res. 32:323-329. SONSTHAGEN, S.A., S.L. TALBOT, AND C.M. WHITE. 2004. Gene flow and genetic characterization of Northern Goshawks breeding in Utah. Condor 106:826— 836. Squires, J.R. and L.F. Ruggiero. 1995. Winter move- ments of adult Northern Goshawks that nested in southcentral Wyoming. J. Raptor Res. 29:5-8. AND R.T. Reynolds. 1997. Northern Goshawk {Ac- cipiter gentilis) . In A. Poole and F. Gill [Eds.], The birds of North America, No. 298. The Birds of North Amer- ica, Inc., Philadelphia, PA U.S. A. Stresemann, E. and D. Amadon. 1979. Falconiformes Pages 270-425 in E. Mayr and G.W. Cottrell [Eds.], Check-list of birds of the world: revision of the work of James L. Peters. Museum of Comparative Zoology, Cambridge, MA U.S.A. Taverner, P.A, 1940. Variation in the American goshawk Condor 42:157-160. Trexel, D.R., R.N. Rosenfield, J. Bielefeldt, and E.A. Jacobs. 1999. Comparative nest site habitats in Sharp- shinned and Cooper’s hawks in Wisconsin. Wilson Bull. 111:7-14. United States Department of Interior (USDI). 1997. Endangered and threatened wildlife and plants: 90- day finding for a petition to list the Northern Gos- hawk in the contiguous United States west of the 100* meridian. Federal Register 62:50892—50896. . 1998. Endangered and threatened wildlife and plants: notice of 12-month finding on a petition to list the Northern Goshawk in the contiguous United States west of the 100* meridian. Federal Register 63. 35183-35184. and United States Department of Commerce. 1996. Policy regarding the recognition of distinct ver- tebrate population segments under the Endangered Species Act. Federal Register 61:4721-4725. United States Fish and Wildlife Service (FWS). 1998. Northern Goshawk status review. U.S. Fish and Wild- life Service, Office of Technical Support, Portland, OR, U.S.A. Ward, L.Z., D.K. Ward, and TJ. Tibbits. 1992. Canopy density analysis at goshawk nesting territories on the North Kaibab Ranger District, Kaibab National Forest: final report. Arizona Game and Fish Department, Phoenix, AZ U.S.A. Warren, N.M., G.D. Hayward, R. Holland, R. Escano, 208 Andersen et al. VoL. 39, No. 3 D.C. Crocker-Bedford, T. Komberec, D. Sasse, L. Saunders-Ogg, and W.C. Shuster. 1990. Goshawk habitat relationships. Pages 12—27 in N.M. Warren [Ed.], Old-growth habitats and associated wildlife spe- cies in the northern Rocky Mountains. Publication Rl-90-42. U.S. Department of Agriculture, Forest Ser- vice, Northern Region, Missoula, MT U.S.A. Watson, J.W., D.W. Havs, S.P. Finn, and P. Meehan-Mar- TIN. 1998. Prey of breeding Northern Goshawks in Washington./. Raptor Res. 32:297-305. Whaley, W.H. and C.M. White. 1994. Trends in geo- graphic variation of Cooper’s Hawk and Northern Goshawk in North America: a multivariate analysis. Proc. West. Found. Vertebr. Zool. 5:161-209. Widen, P. 1989. The hunting habitats of goshawks Acdp- iter gentilis in boreal forests of central Sweden. 131: 205-213. . 1997. How, and why, is the goshawk {Accipiter gen- tilis) affected by modern forest management in Fen- noscandia? J. Raptor Res. 31:107-113. WOODBRIDGE, B. AND P.J. Detrich. 1994. Territory occu- pancy and habitat patch size of Northern Goshawks in the southern Cascades of California. Stud. Avian Biol. 16:83-87. Younk, J.V. AND M.J. Bechard. 1994. Breeding ecology of the Northern Goshawk in high-elevation aspen forests of northern Nevada. Stud. Avian Biol. 16:119-121. Received 30 January 2004; accepted 28 September 2004 Guest Editor: Patricia L. Kennedy Appendix. Definitions of ecological terms as they apply to Northern Goshawks. “Active” nest: The term “active” as applied to raptor nests was hrst dehned by Postupalsky (1974) to describe a nest where si egg was laid. However, the term has been used in different ways since then, and is probably best avoided. Breeding area: a nesting area used by goshawks in the pres- ent, past, or both. Breeding area occupancy: goshawks are thought to defend use areas from conspecifics (territories) during the breeding season, and these territories are often used m subsequent years. However, because it is generally impractical to assess territory occupancy, occupancy of breeding areas has been assessed in field studies of gos- hawks. Breeding areas are occupied when goshawks are present, and what constitutes presence has been vari- able across studies, or is undefined. We suggest that breeding areas are occupied when any of the following occur: (1) nesting, (2) one or more goshawks are ob- served in association with a nest with evidence of recent use (e.g., fresh greenery or other evidence of recent nest construction), (3) goshawks respond ag- gressively to humans or respond to conspecific call broadcasts during the breeding season, or (4) pre-dis- persal fledglings are located in the vicinity of a nest that has evidence of recently being used (e.g., fresh whitewash, goshawk feathers, prey remains, or pellets) . If none of these conditions exist, a breeding area can- not be assumed to be unoccupied without meeting additional criteria (e.g., no goshawk detection during systematic searching for nests or in response to con- specific call broadcasts) . Consistent, specific criteria for categorizing a breeding area as unoccupied need to be developed. Breeding density: the number of nests used by breeding goshawks per unit area; alternatively, the number of goshawk breeding areas through a specified time pe- riod per unit area. Breeding population: a group of goshawks that interact in space and time and that breed or potentially breed and for which it is reasonable to discuss emergent popula- tion properties, such as rate of growth, productivity, etc. Goshawk populations are delimited by spatial boundaries based on where they breed, but these boundaries may not be relevant throughout an annual period (e.g., goshawks that annually migrate from breeding areas) or from one year to the next (e.g., goshawks that migrate from breeding areas in only some years). Habitat: the collection of biotic and abiotic factors that produce occupancy by goshawks {sensu Hall et al. 1997). Nest(ing) area: the immediate area surrounding (a) nest(s) used by breeding goshawks. Nest(ing) attempt: a nest that has been used in any manner by goshawks during the breeding season. Goshawks can be observed at a nest, or there may be evidence of egg laying (e.g., eggs or egg fragments), nestlings, or fledglings. Other evidence is often used to iiifer that an egg has been laid or that a pair of goshawks is pre- paring to lay eggs, including observation of goshawks reconstructing an existing nest or building a new nest, observation of greenery added to existing nests, pres- ence of recently molted goshawk feathers in or be- neath a nest, etc. A nest attempt does not necessarily result in egg laying (i.e., nest failure can occur prior to egg laying). Nest stand: the area covered by a forested patch consisting of trees that are often characterized by having a similar size, species, and spacing and in which a goshawk nest occurs. Nest(ing) success: the proportion of nests in which eggs are laid that produce at least 1 fledgling. Nest tree: the tree in which a goshawk nest is placed. Occupied nest area: an area on which a pair of goshawks have established residency during the nesting season and includes ^1 nest. Post-fledging area: the area that is used by recently fledged goshawks before they become independent of adults (sensu Reynolds et al. 1992). Successful nest: a nesting attempt that results in S;1 young fledged. Territory: an area defended by goshawks from conspecifics during the breeding season that generally contains the September 2005 Status 209 nest, alternative nest(s), if any, nest stand(s), nesting area, post-fledging area, and at least some of the area used by adults for foraging. Use area-home range: area traversed by a goshawk or pair of goshawks during the course of normal, daily activi- ties. It is generally necessary to define specific time periods over which use areas or home ranges apply, as they can change in size and other attributes through time. Buteq Books Toll Free: 800-722-2460 phone: 434-263-8671 fax: 434-263-4842 Specializing in Ornithology Buteo Books is the largest retailer of Ornithology books in North America, with over 2,000 in-print titles, arid hundreds of out-of-print titles available. WAfCM UNPtfiMrtNOlKf OWLS NEW, IN-PRINT ORNITHOLOGY The most popular authors, cutting edge studies, and unusual subjects. We stock scientific texts, hard-to-find foreign books, and entertaining reading; plus audio, video, and software. RARE AND OUT-OF-PRINT ORNITHOLOGY Buteo Books has a wide selection of used, out-of-print, and scarce titles available, including classics on birds of prey and falconry. Our stock changes weekly, so call to check availability. BIRDS OF NORTH AMERICA SERIES This series consists of individual species accounts for each of the species which breed in the United States and Canada, including diurnal and nocturnal raptors. These illustrated reviews provide comprehensive summaries of the current knowledge of each species, with range maps and extensive lists of references. As the exclusive distributor of print copies in this series, Buteo Books is pleased to offer these accounts for $7.50 each. All 716 profiles are listed in taxonomic order on our website. Buteo Books; 3130 Laurel Road; Shipman, VA 22971; USA Visit our website for more information; www.buteobooks.com J Raptor Res. 39(3);210-221 © 2005 The Raptor Research Foundation, Inc. IS FLEDGING SUCCESS A RELIABLE INDEX OE EITNESS IN NORTHERN GOSHAWKS? J. David Wiens^ and Richard T. Reynolds USD A Forest Service, Rocky Mountain Research Station, 2150 Centre Avenue, Building A, Suite 350, Fort Collins, CO 80526 US. A. Abstract. — Fledging success is often assumed to be a reliable index of reproductive success (i.e., fitness) in the ornithological literature. We examined the validity of this assumption in a large population of Northern Goshawks {Accipiter gentilis) on the Kaibab Plateau, Arizona, at both the population and in- dividual levels. We used mark-recapture data from 558 fledglings produced at 494 nests over a 10-yr period to assess the hypothesis that the number of fledglings returning to breed from an annual fledg- ling cohort is positively correlated with the size of the cohort. Natal philopatry was low and recruitment was gradual: only 48 fledglings (8.6%) returned to breed between 2-8 yr of age (x = 3.5 yr). We found no evidence that the breeding population produced more local recruits in years of high fledgling production than in years of low fledgling production. At the individual level, however, fledgling pro- duction for 290 breeding adults was related to their contributions to the future breeding population. Variation in fitness potential among territorial adults was high, as only 20% of the breeding population produced nearly 50% of the fledglings and 84% of the local recruits during the study. Our results indicate that measures of annual productivity for a large breeding population were not reflective of reproductive success, whereas measures of individual productivity were. We conclude that fledging suc- cess of individual goshawks is a reliable index of fitness, but that population productivity is a poor predictor of local recruitment. Keywords: Northern Goshawk, Accipiter gentilis; individual heterogeneity, Kaibab Plateau', recruitment; reproductive success. iCONSTITUYE EL EXITO DE EMPLUMAMIENTO UN INDICE CONFIABLE DE ADECUACION BIOLOGICA EN ACCIPITER GENTILIS? Resumen. — En la literatura ornitologica, a menudo se asume que el exito de emplumamiento representa un indice confiable del exito reproductivo (i.e., adecuacion biologica) . Examinamos la validez de este supuesto en una poblacion de gran tamaho de Accipiter gentilis en Kaibab Plateau, Arizona, tan to a nivel de poblacion como de individuo. Usamos datos de marcado y recaptura de 558 volantones provenientes de 494 nidos a lo largo de un periodo de 10 ahos para evaluar la hipotesis de que el numero de volantones que regresan a reproducirse de una cohorte anual de volantones esta positivamente corre- lacionado con el tamaho de la cohorte. La filopatria natal fue baja y el reclutamiento fue gradual: solo 48 volantones (8.6%) regresaron a reproducirse con entre dos y ocho ahos de edad (x = 3.5 ahos). No encontramos evidencia de que la poblacion reproductiva produjo mas reclutamientos locales en ahos de alta produccion de volantones que en ahos de baja produccion. A nivel individual, sin embargo, la produccion de volantones correspondiente a 290 adultos reproductivos se relaciono con sus contribu- ciones a la futura poblacion reproductiva. La variacion en la adecuacion biologica entre los territorios de los adultos fue alta, ya que solo el 20% de la poblacion reproductiva produjo cerca del 50% de los volantones y el 84% de los reclutamientos locales durante el estudio. Nuestros resultados indican que las medidas de productividad anual de una poblacion reproductiva de gran tamaho no reflejaron el exito reproductivo, mientras que las medidas de productividad individual si lo hacen. Concluimos que el exito de emplumamiento individual de A. gentilis representa un indice confiable de adecuacion biol- ogica, pero que la productividad a nivel poblacional predice de modo inadecuado el reclutamiento local. [Traduccion del equipo editorial] ^ Gorresponding author’s email address: jdwiens@comcast.net 210 September 2005 Biology 211 In birds, a commonly measured reproductive variable used to assess population performance over time is fledging success (the number of young that fledge per nest) . A widely held assumption in avian studies is that fledging success is a reliable index of reproductive success (the number of off- spring that survive to become breeding adults) and thus, fitness (Endler 1986, Weatherhead and Du- four 2000, Keedwell 2003). However, the spatial and temporal scales over which many populations are studied may not correspond well with the spa- tial extent of natal dispersal or the time periods over which recruitment occurs. These limitations, superimposed on a variety of stochastic factors, could easily disrupt the relationship between fledg- ing success and fitness. Nevertheless, researchers often assume this relationship is positive, in part, due to the difficulties associated with estimating pre-breeding survival and emigration rates, even in long-term banding studies. Difficulties in attaining direct measures of reproductive success are there- fore exemplified in long-lived, wide-ranging spe- cies that occur at low densities, have low natal phil- opatry, elude detection when not breeding, and initiate breeding at a delayed age (Weatherhead and Dufour 2000). The Northern Goshawk (Accipiter gentilis) is a long-lived raptor that occupies mature temperate and boreal forests throughout the Holarctic (Squires and Reynolds 1997). The goshawk is a so- cially monogamous, territorial species that lays one clutch per year (Reynolds et al. 1994, Kennedy and Ward 2003). Although several studies have docu- mented extensive temporal variation in fledging success in goshawk populations (e.g., McClaren et al. 2002, Reynolds and Joy in press), none have addressed the relationship between fledging suc- cess and local recruitment. The strength of this re- lationship is particularly relevant when the dynam- ics of a local population are more heavily reliant on external recruitment (i.e., immigration) than internal productivity, which may occur when natal philopatry is low and adult site fidelity is high (Sta- cey and Taper 1992, Martin et al. 2000). While mate and site fidelity in adult goshawks is high (75-95%; Detrich and Woodbridge 1994, Reynolds and Joy in press) , the degree of natal philopatry is largely unknown due to low juvenile recapture probabilities, few recoveries of banded nestlings, and a general lack of information on the extent of natal dispersal (Kennedy and Ward 2003, but see Wiens 2004). However, molecular evidence has shown that gene flow among subpopulations of goshawks over large geographic areas is high (Son- sthagen 2004, Bayard de Volo et al. 2005), indicat- ing that juveniles may disperse over long distances because adults rarely disperse once they have set- tled on a breeding territory. Juvenile survival is one of the most difficult demographic parameters to estimate precisely in goshawks (Kennedy 1997, Wiens 2004), further emphasizing the need to as- sess the assumed relationship between fledging production and reproductive success. In this paper we evaluate whether fledgling suc- cess is a reliable predictor of reproductive success in goshawks. We examined this relationship at both the population and individual levels using a 13-yr mark-recapture (resight) data set obtained from a breeding population of goshawks exceeding 120 pairs on the Kaibab Plateau, Arizona. Our investi- gation was inspired by studies showing a positive relationship between fledging success and recruit- ment in bird species such as Red-winged Blackbirds (Agelaius phoeniceus) , Eurasian Sparrowhawks {Ac- cipiter nisus) , Ural Owls {Strix uralensis) , and Osprey {Pandion haliaetus; Weatherhead and Dufour 2000, Newton 1989a, Saurola 1989, Postupalsky 1989, re- spectively) . At the population level, we assessed the hypothesis that the number of fledglings produced annually is positively correlated with the number of individuals from annual cohorts that were even- tually recruited into the local breeding population. At the individual level, we anticipated that total fledgling production of color-marked male and fe- male adult (>2 yr old) goshawks would be posi- tively related to the number of their descendants that were recruited into the local breeding popu- lation. In examining our hypotheses, we also re- port on local recruitment and ages at first breeding for goshawks on the Kaibab Plateau. Methods Study Area and Field Procedures. The study area in- cluded all the coniferous forest above 2182 m elevation on the Kaibab Plateau of northern Arizona. This 1732 km^ area included the northern portion of the Kaibab National Forest and the Grand Canyon National Park (North Rim). The Kaibab Plateau is a large (95 X 55 km), oval-shaped plateau that rises from a shrub-steppe plain at 1750 m elevation to the highest point at 2800 m and is dissected by moderately sloping valleys (Rasmussen 1941). Forests of the Kaibab Plateau are isolated from similar forests by variable distances (35-250 km) of pin- yon-juniper {Pinus edulis-Juniperus spp.) woodland, and sagebrush {Artemisia spp.) plains. See Reynolds et al. (1994) and Reynolds et al. (2005) for further detail on the study area, its management history, and protocols 212 Wiens and Reynolds VoL. 39, No. 3 used to locate, survey, and monitor goshawk breeding areas. We defined a territory as a breeding area used, but not necessarily defended against conspecifics, by a single pair of goshawks during a breeding season (Reynolds et al. 2005). During 1991-2003, a high density of regularly- spaced goshawk breeding territories were identified on the Kaibab Plateau (Reich et al. 2004, Reynolds et al. 2005) . Territories contained multiple alternate nests that were used one or more times over the years by goshawks. We visited all nests in early spring of each year to estimate occupancy and reproductive status of territorial pairs. If goshawks were not using a known nest, a three-stage ter- ritory survey protocol (Reynolds et al. 2005) was used to determine territory status. We classified a territory as "ac- tive' when eggs were laid, “occupied' when adult gos- hawks were observed on two or more occasions in the vicinity of a nest (or a single observation of an adult in combination with observations of molted feathers, feces, and fresh nest construction), and “unknown" if eggs were not laid and no evidence of goshawk occupancy was found (Reynolds et al. 2005). We determined nest fate (successful = fledged ^1 young; failed = eggs laid but no fledglings produced), nest productivity (number of fledglings) , and identity of adults during weekly visits to active territories. For our purposes here, we defined the number of young fledged as the number of nestlings present at the time of banding (ca. 1 wk prior to fledg- ing) . Nesting adults were captured, measured, sexed, and aged during the mid-late nestling period following meth- ods described in Reynolds et al. (1994); nestlings were captured by climbing nest trees during the last week of the nestling period (mid-late June). Sex of nestlings was assigned on the basis of body mass, tarsometatarsal length, and toepad-length measurements (Wiens 2004). All captured hawks received a U.S. Geological Survey alu- minum leg band and an anodized colored leg band with a unique alpha-numeric code (Arcraft Sign and Plate Co., Edmonton, Canada) readable to 80 m with 40-60 X spot- ting scopes. Age at First Breeding and Local Recruitment, We con- sidered a goshawk to have been recruited locally if it was banded as a nestling on the study area and later observed breeding in the study population. Thus, hawks classified as “local recruits” had to have been recaptured or re- sighted at an active territory on the Kaibab Plateau. To attain an unbiased estimate of the age at first breeding, we included only banded (known-aged) hawks that we were confident had been detected on their first breeding attempt. This meant that banded recruits had to have been resighted on territories where the same-sex occu- pant during the prior year was known. Estimated ages at first breeding could have been biased high by breeding dispersal (i.e., undetected movement among territories between successive breeding attempts). However, this bias was likely to be small because <6% of adult goshawks moved to a different territory between successive breed- ing attempts (R. Reynolds unpubl. data). To attain a mean estimate of local recruitment, we subtracted the number of nestlings that were too young (based on the median age at first breeding) to have attained breeding territories by spring 2003 from the total number of nest- lings banded during 1991-2003. Hence, local recruit- ment was calculated as the total number of banded re- cruits detected during 1991-2003 divided by the total number of nestlings banded during 1991-2000. Population Productivity. We were unable to capture and mark all nestlings produced in some years because several nests were not located until after fledging or, rare- ly, a nest tree was unsafe to climb. Thus, the number of young banded represented a portion (69% during 1991- 2003) of the known number of young that actually fledged. For this reason, we defined “productivity” as the number of young banded. We assumed that banded young comprised a representative sample in terms of population productivity and local recruitment. Using the number of young banded annually, we assessed whether the size of each annual fledgling cohort was correlated with the number of local recruits originating from each cohort. We conducted analyses with sexes combined and then separately for males and females. Data were exam- ined by sex for two reasons. First, patterns of post-fledg- ing mortality have been found to differ for male and fe- male goshawks on the Kaibab Plateau (Wiens 2004). Second, natal dispersal distances may differ between sex- es (Greenwood and Harvey 1982, Byholm et al. 2003). One or both of these factors could result in sex-depen- dent local recruitment. For sex-specific analyses, the number of individuals of each sex that successfully re- cruited from a fledgling cohort was assessed relative to the total number of individuals of each sex that were banded as nestlings. Individual Productivity. To examine the relationship between total fledgling production of individual adult goshawks and recruitment success of their offspring, we calculated the number of banded fledglings produced by each color-marked adult that bred during 1991-2000. As in the population-level analysis, excluding fledgling pro- ductivity during 2001-03 ensured that all fledglings in our analysis had at least three years to acquire a breeding territory and be detected. We then related the number of banded fledglings produced by each adult to the total number of their offspring that had successfully recruited to the local breeding population by 2003, We used fledg- ling production by adults during the study period rather than lifetime productivity because our interest was in the reliability of individual fledging success as a measure of an individual’s fitness potential, regardless of how often they bred or how long they had occupied a breeding territory. Nonetheless, lifetime fledgling production was captured for nearly all goshawks included in our analysis because of the duration of our study relative to the num- ber of years goshawks successfully laid eggs (x = 2.1 yr, min. = 1 yr, max. = 10 yr, 1991-2004; R. Reynolds un- publ. data) . We assumed that adults captured or resight- ed at a nest were indeed the biological parents of the nestlings banded at that nest. In raptors, females are of- ten alone before and during egg laying while their mates forage, which could result in extra-pair fertilizations by conspecific males (Reynolds and Linkhart 1990, Negro et al. 1996, Gavin et al. 1998). Although extra-pair fertil- izations could confound the relationship between fledg- ling success and recruitment, genetic evidence demon- strated that extra-pair fertilizations are infrequent for goshawks on the Kaibab Plateau (Gavin et al. 1998). Data Analysis. At the population level, we used Spear- September 2005 Biology 213 Table 1. Northern Goshawk survey effort, banding activity, and local recruitment (banded nestlings from annual cohorts that eventually returned to breed) on the Kaibab Plateau, Arizona, 1991-2003. Year Territories Surveyed Used Nests (%)^ Adults Banded Nestlings Banded Nestlings Returned 1991 37 36 (97) 48 46 8 1992 64 59 (92) 43 32 2 1993 82 67 (82) 30 62 4 1994 88 21 (24) 13 18 2 1995 99 53 (54) 39 52 7 1996 105 46 (44) 18 41 1 1997 106 31 (29) 15 36 9 1998 109 58 (53) 37 84 9 1999 113 57 (50) 14 76 5 2000 120 66 (55) 33 111 1 2001 120 30 (25) 5 31 0 2002 121 21 (17) 5 16 0 2003 121 10 (8) 2 9 — ® Percent of total territories under study that contained used nests (eggs laid) . man correlation coefficients (r^) to characterize the strength of the relationship between the size of each an- nual fledgling cohort and the number of local recruits originating from each cohort. We estimated the expected number of fledglings recruited from each annual cohort by multiplying our overall estimate of local recruitment by cohort size. A chi-square analysis was then used to eval- uate possible deviations from the expected number of fledglings recruited from each cohort. At the individual level, we used generalized linear models (PROG GEN- MOD; SAS Institute 1999) to investigate our prediction that total fledging production by individual adults would be positively related to the number of local recruits each produced. Specifically, we used Poisson regression in a log-linear model in which a count of recruits produced per adult was the response variable and the number of banded fledglings produced per adult was assessed as an explanatory variable. Chi-square tests or a Fisher’s exact test were used to examine potential sex-related differ- ences in the age at first breeding and local recruitment. All analyses were conducted using SAS (ver. 8.2). Results are reported as means ± SE, with 95% confidence inter- vals (Cl). Results The number of territories surveyed increased from 37 in 1991 to 121 in 2003 (Table 1). All but two of the 121 territories contained nests with eggs or young in one or more years of the study. The exceptions were territories where goshawks occu- pied newly built or reconstructed old nests, but did not lay eggs. The percentage of territories under study that contained nests with eggs or young var- ied substantially among years, ranging from 8% in 2003 to 97% in 1991 (Table 1). This resulted in highly variable fledgling production among years. ranging from only 9 fledglings in 2003 to 111 in 2000. Of the 13-yr total of 555 nests with eggs, 447 (80.5%) adult pairs fledged 897 young successfully (614 of which were banded), and 108 (19.5%) pairs lost their clutch during the incubation or nestling stages. Age at First Breeding and Local Recruitment. In the 10 yr that we included for fledgling production (1991-2000), we banded 558 nestlings (278 fe- males and 280 males) at 494 nests (Table 1). Forty- eight (8.6%) of these nestlings returned to breed on the study area-26 (9.4%) females and 22 (7.9%) males. Nestling return rates were similar between the sexes (x^ = 0.40, df = 1, P = 0.53). Ages at first breeding could be determined with confidence for 28 of 48 fledglings that returned to breed (17 of 26 females and 11 of 22 males; Fig. 1). These hawks initiated breeding between 2-8 yr of age (3.5 ± 0.32 yr, 95% Cl = 2.84—4.16, median = 3 yr) . Only four hawks (all females) were older than 4 yr of age at first breeding. Differences be- tween sexes in the age at first breeding were not supported (Fisher’s Exact Test: P = 0.49). No hawks in first-year plumage were observed breed- ing at or occupying a nest during the 13-yr study. Two females banded as nestlings were later resight- ed in adult plumage occupying inactive breeding territories on the study area, and two locally band- ed nestlings were found breeding in forests beyond the Kaibab Plateau (Wiens 2004). Population Productivity. We found no evidence 214 Wiens and Reynolds VoL. 39, No. 3 Age (years) Figure 1. Distribution of ages at first breeding for 28 known-aged Northern Goshawks recaptured on their first breeding attempt on the Kaibab Plateau, Arizona, 1991-2003. that the number of fledglings recruited from an- nual cohorts was correlated with cohort size (r^ = 0.10, P = 0.79, N = 10; Fig. 2). Likewise, there was no evidence of a correlation between annual esti- mates of fledging production and recruitment when sexes were analyzed separately (females: = 0.24, P = 0.50; males: r, = 0.11, P = 0.76). The number of local recruits produced was significantly higher than expected for the 1991 (8 observed, 3.9 expected; ~ 4.13, df = \, P = 0.04) and 1997 (9 observed, 3.1 expected; — 11.25, df = 1, P< 0.01) fledgling cohorts, but much lower than ex- pected for the 2000 fledgling cohort (1 observed, 9.5 expected; x^ ~ 7.65, df = 1, P < 0.01), We found no other deviations in numbers of observed versus expected recruits in any other fledgling co- hort. Thus, the 2000 fledgling cohort was the only one of the 10 cohorts in which significantly fewer fledglings returned to breed than expected. To en- sure that including this year did not unduly bias Figure 2. Number of locally produced Northern Goshawk fledglings recruited into the breeding population on the Kaibab Plateau, Arizona, relative to the number of young fledged within each annual cohort, 1991-2000. Each point represents one year. September 2005 Biology 215 (0 3 ■o < 0 ) n E ■ Females □ Males n "I — 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Number of Fledglings Produced Figure 3. Fledgling production of 156 female and 134 male color-marked adult (>2-yr-old) Northern Goshawks breeding between 1991 and 2000 on the Kaibab Plateau, Arizona. our results, we reanalyzed the data without this co- hort. Although this improved the relationship slightly, the size of a fledgling cohort was still a poor predictor of the total number of goshawks that returned to breed from the cohort (sexes combined: = 0.46, P = 0.21, N — 9; females: = 0.52, P = 0.15; males: r, = 0.46, P = 0.21). Individual Productivity. Fledging success was cal- culated for 156 females and 134 males breeding during 1991—2000 (Table 1). During this period, females spent between 1-7 yr as breeders (2.18 ± 0.11 yr; 95% Cl = 1.96-2.40) and fledged between 0 and 15 young (3.39 ± 0.23 young; 95% Cl = 2.93-3.85). Similarly, males also spent between 1 and 7 years as breeders (1.96 ± 0.11 yr; 95% Cl = 1.73-2.18) and fledged between 0 and 15 young (2.98 ± 0.23 young; 95% Cl = 2.51-3.44). The dis- tribution of fledgling production among breeders was highly skewed, with most individuals producing few or no fledglings and only a few individuals hav- ing extremely high success (Fig. 3). In total, 156 adult females produced 529 fledglings and 134 adult males produced 399 fledglings (the remain- ing 29 and 159 banded nestlings were parented by unknown adult females and males, respectively). Thirty-six adults (19 females and 17 males) laid eggs but failed to produce fledglings. The recruitment success of an adult’s offspring was significantly related to the total number of fledglings the adult produced (sexes combined: re- gression slope coefficient, $ = 0.22 ± 0.03, Wald = 74.77, df = I, P< 0.01, N = 290; females: p = 0.22 ± 0.04, = 38.2, df = 1, P < 0.01, N = 156; males: 3 = 0.22 ± 0.04, x^ ^ 37.3, df = 1, P < 0.01, N = 134; Fig. 4). We had two concerns regarding this result. First, we suspected that fledg- lings produced during the 2000 breeding season might not have had sufficient time to be recruited by 2003 (as was our concern in the population- level analysis) . Second, there were 33 new unband- ed adults (17 females and 16 males) who first bred in the study population in 2000 (Table 1). Several of these individuals also bred in subsequent years, so our initial analysis may have not represented their contributions adequately to fledgling produc- tion. To address these concerns, we reanalyzed the data without individual productivity and recruit- ment data from the 2000 breeding season. Exclud- ing the 2000 breeding season had little effect on the strength of the relationship (sexes combined: p = 0.28 ± 0.03, x^ ^ 71.0, df = 1, P < 0.01, N - 257; females: g = 0.28 ± 0.05, x^ = 37.1, df = 1, P < 0.01, N = 139; males: p = 0.28 ± 0.05, x^ = 36.5, df = 1, P < 0.01, A = 118). Thus, fledging success of individual goshawks could be taken as a reasonable index of their contributions to the fu- ture breeding population. Both sexes exhibited similar patterns of variance in fledgling production and in the number of re- cruits they produced (Figs. 3, 4) . Moreover, fledg- ing success and contributions of offspring to the future breeding population varied extensively 216 Wiens and Reynolds VoL. 39, No. 3 Number of Young Fledged Per Adult Figure 4. Number of descendants of individual female {N = 156) and male (N = 134) adult (>2-yr-old) Northern Goshawks that successfully recruited into the breeding population in relation to the number of fledglings each adult produced between 1991 and 2000 on the Kaibab Plateau, Arizona. Each point represents an individual adult. Trend lines determined by Poisson regression with a log-link function (females: y = exp [—2.217 + 0.221x]; males: y = exp [-2.056 + 0.217x]). among territorial hawks; 20% of adults breeding during 1991-2000 contributed nearly 50% of the fledglings and 83% of the local recruits produced during this time (Fig. 5). Of 290 breeding gos- hawks, 73 (25.2%) produced at least one local re- cruit each, seven (2.4%) produced two recruits each, and three (1%) produced three recruits each. One male produced four local recruits, but no female produced more than three. In tracking lineages up to three generations, three females left one breeding descendant each who also produced young that were recruited, and two males left one breeding descendant each who also produced young that were recruited. Discussion When measured at the population level, our re- sults failed to support the hypothesis that fledging success is a reliable index of local recruitment in goshawks. When measured on an individual basis, however, adult fledging success was significantly re- lated to local recruitment, indicating that total fledgling production by individuals was a more re- liable index of reproductive success than a mea- Cumulative Proportion of Breeding Adults Figure 5. Individual variation in total fledgling production and reproductive success (number of offspring that were local recruits) among 290 color-marked adult (>2-yr-old) Northern Goshawks breeding between 1991 and 2000 on the Kaibab Plateau, Arizona. Whereas 254 (88%) adults successfully fledged young, only 73 (25%) made contributions to the future local breeding population. sure of annual productivity. Another important finding was that reproductive success varied widely among breeding adults. Thus, the relationship be- tween fledging success and recruitment appeared to be highly dependent upon the disproportionate contributions of fledglings and recruits made by a relatively small number of adults rather than over- all population productivity. Our results further demonstrate that recruitment of goshawks on the Kaibab Plateau was a gradual process (as shown by a wide range among individuals in the age at first breeding) and that local recruitment was relatively low (8.6%). Given that 25% of adults fail to return to reclaim their territory each year (Reynolds et al. 2004), external recruitment (i.e., immigration) would need to be ca. 16% to maintain a stable breeding population. Age at First Breeding and Local Recruitment. Goshawks are capable of breeding in their first year of life, and the proportion of breeders in first-year plumage has been reported to be as high as 35- 40% (McGowan 1975, Reynolds and Wight 1978, Speiser and Bosakowski 1991, Younk and Bechard 1994) . The proportion of young hawks in a breed- ing population may be reflective of density-depen- dent processes driven by the availability of nesting territories, food, or mates (McGowan 1975, New- ton 1989b, Ferrer et al. 2004). As breeder density and survival increases, fewer territories are avail- able to prospective breeders, forcing socially sub- ordinate younger hawks to wait, perhaps several years, to gain a breeding vacancy. Observations of individuals breeding in non-adult plumage may, therefore, reflect an important buffer mechanism compensating for an increased adult mortality rate (Ferrer et al. 2004) . On the Kaibab Plateau, several demographic features such as a temporally stable adult survival rate (Reynolds et al. 2004), a high density of breeding territories (8.6/100 km^; Reyn- olds et al. 2005), and strong adult fidelity to terri- tory and mate (R. Reynolds unpubl. data) suggest that young goshawks prospecting for breeding op- portunities are faced with the alternatives of delay- ing breeding altogether or dispersing to recruit elsewhere (Wiens 2004) . That some individuals re- main (or return) as floaters in the vicinity of their natal population on the Kaibab Plateau has been confirmed by radiotelemetry (Wiens 2004) and a quick replacement of territorial hawks following mortality (Reynolds and Joy in press). Collectively, these features could explain the relatively ad- vanced age of first-time breeders as well as the low 218 Wiens and Reynolds VoL. 39, No. 3 local recruitment level during our study. It is im- portant to note, however, that these characteristics (territory density, adult fidelity, breeding age) ap- pear to occur at higher levels on the Kaibab Pla- teau than reported for goshawks elsewhere. Population Productivity. We found that the gos- hawk population on the Kaibab Plateau was no more successful in producing local recruits in years of high fledgling production than in years of low fledgling production. Thus, contrary to predic- tions, the size of an annual fledgling cohort was a poor predictor of local recruitment. In contrast, Weatherhead and Dufour (2000) found that local recruitment increased disproportionately with the size of Red-winged Blackbird fledgling cohorts, even though the overall return rate of banded nest- lings in that study was small (2.4%) relative to our estimate for goshawks (8.6%). The strength of the relationship between population productivity and local recruitment, and our inability to detect such a relationship, may largely depend on the spatial extent of the study area relative to the extent of natal dispersal. For example, Red-winged Black- birds disperse up to 40 km from natal territories (Moore and Dolbeer 1989), whereas young gos- hawks disperse up to 440 km from the Kaibab Pla- teau (Wiens 2004). The lack of a population-level relationship between fledging success and recruit- ment in our study may, therefore, simply reflect the difference between observing local recruitment within a relatively small subpopulation relative to general recruitment taking place over a regionally fragmented population that is connected by natal dispersal. Several lines of evidence indicate that most juveniles disperse long distances beyond the Kaibab Plateau in their first fall (Wiens 2004) . Un- fortunately, with the exception of two known cases of juvenile emigration, no information on the ex- ternal recruitment success of locally-produced hawks exists for this study population. Aside from a potentially inappropriate scale of investigation relative to the extent of the goshawk recruitment process, our data may have failed to support our hypothesis at the population level be- cause of a lack of breeding opportunities during the later years of the study. During 2001-03, ex- treme drought conditions in northern AZ likely led to significant declines in goshawk prey populations (Salafsky 2004), and few adult pairs on the Kaibab Plateau attempted to breed. Under such poor breeding conditions, opportunities for inexperi- enced hawks attempting to breed for their first time were reduced. For the most part, adult gos- hawks could only be captured or resighted when breeding, so newly recruiting hawks that acquired a breeding territory but did not lay eggs may have gone undetected during 2001-03. Poor breeding conditions during the later years of our study could also explain the lower than expected recruitment success of fledglings produced in the 2000 breed- ing season. However, removing the 2000 cohort from the analysis failed to substantially improve the relationship between population productivity and local recruitment. Our chi-square analysis showed that fledglings produced in 1991 and 1997 were twice as likely to be recruited as expected. With the exception of the 2000 fledgling cohort, we found no other significant deviations in the number of local recruits expected. We suggest that high juve- nile survival, density-dependent effects, or changes in resource availability may have contributed to the disproportionately high recruitment success of fledglings produced during 1991 and 1997. Individual Productivity. If fitness is defined as the contribution of an individual’s genotype to sub- sequent generations proportional to that of other individuals (Lincoln et al. 1998), then our results clearly show a difference in fitness among gos- hawks on the Kaibab Plateau. Our results indicated that the likelihood of an adult contributing off- spring to the future breeding population on the Kaibab Plateau increased disproportionately with the number of fledglings it produced. Therefore, goshawks appear to be similar to other monoga- mous raptor species in that reproductive success varies widely among individuals (Newton 1989b, Hakkarainen et al. 1997, Marti 1997). As would be expected for a monogamous raptor in a popula- tion showing high fidelity to territory and mate and equal survival rates of males and females, the sexes showed similar patterns of variance in repro- ductive success. Indeed, a few territorial pairs ex- hibited disproportionately high fitness levels in terms of the number of years they bred, total fledg- ling production, and the number of their offspring that survived to become breeding adults. Factors not explored in this study, such as breeding age, habitat quality, individual quality (as determined by genetic forces), or climate may contribute to individual heterogeneity in goshawk reproductive success. On the Kaibab Plateau, the mean reproductive lifespan of adults (first breeding to disappearance) is 2 yr, but a few adults were found to breed for as September 2005 Biology 219 many as 10 yr (R. Reynolds unpubl. data). Adults who breed more often are likely to be more ex- perienced, produce more fledglings, and show the highest fitness potential in terms of survival and reproduction (Gam et al. 1998). Newton and Roth- ery (2002) , for example, demonstrated that almost all aspects of breeding performance in female Eur- asian Sparrowhawks improved with age, but that the degree of improvement lessened with each suc- cessive year of life. In Denmark, Nielsen and Drachmann (2003) showed that fledgling produc- tion of European goshawks increased with female age from 1-7 yr, but declined thereafter. Similarly, Reynolds et al. (1994) reported a difference in fledging production between young-adult (second- year) and full-adult goshawks. These authors attri- buted changes in reproductive success to age-relat- ed trends in foraging efficiency, which could be buffered by pairing with more experienced mates. Variation among individuals in fitness levels can have substantial effects on population growth, sta- bility, and persistence (Bj0rnstad and Hansen 1994, Conner and White 1999). Individual varia- tion within raptor populations may be generated by temporal or spatial variations in resource avail- ability and habitat quality (Hakkarainen et al. 1997, Franklin et al. 2000). Of particular interest is the relationship between individual fitness and habitat quality, as it is commonly supposed that birds pre- fer the habitat that will confer the greatest fitness (Fretwell and Lucas 1970). For goshawks, hetero- geneity in fitness levels among territorial adults could be caused by spatial or temporal variations in food abundance, resource availability (as mea- sured by qualitative differences in forest composi- tion and structure among breeding areas) , and in- dividual hawk quality. However, recent efforts to detect spatial heterogeneity in goshawk reproduc- tive parameters have had limited success. On the Kaibab Plateau, Joy (2002) ranked 101 goshawk territories as “high” or “low” quality based on dif- ferences in egg laying frequency and fledgling pro- duction. The length of time an adult female re- mained on a territory, breeder age, and the amount or spatial configuration of vegetative types within territories explained little of the variation (Joy 2002). Elsewhere, McClaren et al. (2002) found minimal spatial variation in the number of young fledged per nest among goshawk nest areas in three study sites in western North America. Giv- en that these studies focused on territories (Joy 2002) and nest areas (McClaren et al. 2002) rather than individual goshawks, we suggest that unde- tected turnovers between territory or nest occu- pants of different age, experience, or genetic qual- ity could mask existing spatial patterns in reproduction. For the purposes of identifying the determinants of habitat quality for goshawks, we believe our finding of individual variation in repro- ductive success highlights the need to partition the effects of individual hawk quality from habitat qual- ity. By controlling for individual hawk quality, re- searchers can more precisely estimate the effects of habitat versus non-habitat factors on compo- nents of goshawk fitness. Given the hypothesized role of forest composition and structure in gener- ating individual variation in goshawk reproductive performance, long-term demographic research based on color-marked individuals can provide a powerful tool to guide goshawk conservation and management. Acknowledgments Our research was supported by the USDA Forest Ser- vice (Southwest Region), Rocky Mountain Research Sta- tion, the North Kaibab Ranger District, and a grant from the Heritage Program, AZ Game and Fish Department. We are grateful to the many field technicians and vol- unteers who helped collect data on Northern Goshawks during 1991-2003. The North Kaibab Ranger District graciously provided housing and logistical support dur- ing this study. Special thanks to S.M. Joy, C.M. Erickson, J.S. Lambert, J.C. Seyfried, S.R. Salafsky, and S. Bayard de Volo for their help with data entry and summarization. R.M. King provided statistical advice. We thank B.R. Noon, P.J. Weatherhead, J.A. Smallwood, C. Crocker-Bed- ford, and C.W. Boal for providing constructive comments which improved this manuscript. Literature Cited Bayard de Volo, S., R.T. Reynolds, J.R. Topinka, B. May, AND M.F. Antolin. 2005. Population genetics and ge- notyping for mark-recapture studies of Northern Gos- hawks {Accipiter gentilis) on the Kaibab plateau, Ari- zona. J. Raptor Res. 39:286-295. Bjornstad, O.N. and T.F. Hansen. 1994. Individual vari- ation and population dynamics. Oikos 69:167-171. Byholm, P., P. Saurola, H. Linden, and M. Wikman. 2003. Causes of dispersal in Northern Goshawks {Ac- dpiter gentilis) in Finland. Auk 120:706-716. Cam, E., J.E. Hines, J. Monnat, J.D. Nichols, and E. Dan- chin. 1998. Are adult nonbreeders prudent parents? The Kittiwake model. Ecology 79:2917—2930. Conner, M.M. and G.C. White. 1999. Effects of individ- ual heterogeneity in estimating the persistence of small populations. Nat. Res. Model. 12:109—127. Detrich, P.J. and B. Woodbridge. 1994. Territory fidel- ity, mate fidelity, and movements of color-marked 220 Wiens and Reynolds VoL. 39, No. 3 Northern Goshawks in the southern Cascades of Cal- ifornia. Stud. Avian Biol, 16:130-132. Endler, J.A. 1986. Natural selection in the wild. Prince- ton Univ. Press, Princeton NJ U.S.A. Ferrer, M., F. Otalora, and J.M. Garcia-Ruiz. 2004. Density-dependent age of first reproduction as a buff- er affecting persistence of small populations. Ecol. Appl. 14:616-624. Franklin, A.B., D.R. Anderson, R.J. Gutierrez, and K.P. Burnham. 2000. Climate, habitat quality, and fitness in Northern Spotted Owl populations in northwest- ern California. Ecol. Monographs 70:539—590. Fretwell, S.D. and H.L. Lucas, Jr. 1970. On territorial behavior and other factors influencing habitat distri- bution in birds. I. Theoretical development. Acta Bio. 19:16-36. Gavin, T.A., R.T. Reynolds, S.M. Joy, D.G. Leslie, and B. May. 1998. Genetic evidence for low frequency of ex- tra-pair fertilizations in Northern Goshawks. Condor 100:556-560. Greenwood, P.J. and P.H. Harvey. 1982. The natal and breeding dispersal in birds. Annu. Rev. Ecol. Syst. 13: 1 - 21 . Hakkarainen, H., V. KorvxJVEN, and E Korpimaki. 1997. Reproductive success and parental effort of Teng- raalm’s Owls: effects of spatial and temporal variation in habitat quality. Ecoscience 4:35-42- Joy, S.M. 2002. Northern Goshawk habitat on the Kaibab National Forest in Arizona: factors affecting nest lo- cations and territory quality. Ph.D. dissertation, Col- orado State Univ., Fort Collins, CO U.S.A. Keedwell, RJ. 2003. Does fledging equal success? Post- fledging mortality in the Black-fronted Tern. J. Field Ornithol. 74:217-221. Kennedy, P.L. 1997. The Northern Goshawk {Accipiter gen- tilis atricapillus) : is there evidence of a population de- cline? J. Raptor Res. 31:95-106. AND J.M. Ward. 2003. Effects of experimental food supplementation on movements of juvenile Northern Goshawks {Accipiter gentilis atricapillus) . Oe- cologia 134:284—291. Lincoln, R., G. Boxshall, and P. Clark. 1998. A dictio- nary of ecology, evolution, and systematics, 2nd Ed. Cambridge Univ. Press, Cambridge, U.K. Marti, C.D. 1997. Lifetime reproductive success of Barn Owls near the limit of the species’ range. Auk 114: 581-592. Martin, K., P.B. Stacey, and C.E. Braun. 2000. Recruit- ment, dispersal, and demographic rescue in spatially- structured White-tailed Ptarmigan populations. Con- dor 102:503-516. McClaren, E.L., P.L. Kennedy, and S.R. Dewey 2002. Do some Northern Goshawk nest areas consistently fledge more young than others? Condor 104:343-352. McGowan, J.D. 1975. Distribution, density, and produc- tivity of goshawks in interior Alaska. Federal Aid Wild- life Restoration Project. Rep. W-17-4, W-17-5, W-17- 6, Job 10.6A. Alaska Department of Fish and Game. Anchorage, AK. Moore, W.S. and R.A. Dolbeer. 1989. The use of band- ing recovery data to estimate dispersal rates and gene flow in avian species: case studies in the Red-winged Blackbird and Common Grackle. Condor 91:242-253. Negro. JJ-. M. Villarroel, J.L. Tella, U. Kuhnlein, F. Hiralso, J.A. DonAzar, and D.M. Bird. 1996. DNA fingerprinting reveals low incidence of extra-pair fer- tilizations in the Lesser Kestrel. Anim. Behav. 51:935— 943. Newton, 1. 1989a. Sparrowhawk. Pages 279-296 in I. Newton [Ed..], Lifetime reproduction in birds. Aca- demic Press, New York, NY U.S.A. (Ed.). 1989b. Lifetime reproduction in birds. Ac- ademic Press, New York, NY U.S.A. AND P. Rothery. 2002. Age-related trends in dif- ferent aspects of the breeding performance of indi- vidual female Eurasian Sparrowhawks {Accipiter nisus) Auk 119:735-748. Nielsen, J.T. and J. Drachmann. 2003. Age-dependent reproductive performance in Northern Goshawks {Accipiter gentilis) . Ibis 145:1-8. Postupalsky, S. 1989. Osprey. Pages 297-313 in I. New- ton [Ed.], Lifetime reproduction in birds. Academic Press, New York, NY U.S.A. Rasmussen, D. I. 1941. Biotic communities of the Kaibab Plateau, Arizona. Ecol. Monographs 11:230—273. Reich, R.M, S.M. Joy, and R.T. Reynolds. 2004. Predict- ing the location of Northern Goshawk nests: model- ing the spatial dependency between nest locations and forest structure. Ecol. Modeling 176:109—133. Reynolds, R.T. and S.M. Joy. In press. Demography of Northern Goshawks in Northern Arizona, 1991-96. Stud. Avian Biol. , , AND D.G. Leslie. 1994. Nest productivity, fidelity, and spacing of Northern Goshawks in north- ern Arizona. Stud. Avian Biol. 16:106-113. AND B.D. Linkhart. 1990. Extra-pair copulation and extra-range movements in Flammulated Owls. Or- nis. Scand. 21:74—77. , G.C. White, S.M. Joy, and R.W. Mannan. 2004. Effects of radiotransmitters on Northern Goshawks, do tailmounts lower survival of breeding males? J. Wildl. Manag. 68:25-32. , J.D. Wiens, S.M. Joy, and S.R. Salafsky 2005. Sampling considerations for demographic and habitat studies of Northern Goshawks. J. Raptor Res. 39:274- 285. AND H.M. Wight. 1978. Distribution, density, and productivity of Accipiter hawks breeding in Oregon. Wilson Bull. 90:182-196. Salafsky, S.R. 2004. Covariation between prey abundance and Northern Goshawk fecundity on the Kaibab Pla- teau, Arizona. M.S. thesis, Colorado State Univ., Fort Collins, CO. U.S.A. SAS Institute, 1999. SAS/STAT User’s guide, version 8, September 2005 Biology 221 Volumes 1, 2, and 3. SAS Institute, Inc., Cary, NC U.S.A. Saurola, P. 1989. Ural Owl. Pages 327-345 in I. Newton [Ed.], Lifetime reproduction in birds. Academic Press, New York, NY U.S.A. SONSTHAGEN, S.A., S.L. TALBOT, AND C.M. WHITE. 2004, Gene flow and genetic characterization of Northern Goshawks breeding in Utah. Condor 106:826-836. Speiser, R. and T. Bosakowski. 1991. Nesting phenology, site fidelity, and defense behavior of Northern Gos- hawks in New York and New Jersey. J, Raptor Res. 25: 132-135. Stacey, P.B. and M. Taper. 1992. Environmental variation and the persistence of small populations. Ecol. Appl. 2:18-29. Squires, J.G. and R.T. Reynolds. 1997. Northern Gos- hawk {Accipiter gentilis) . In A. Poole and F. Gill [Eds.], The birds of North America, No. 298. The Birds of North America, Inc., Washington, DC U.S.A. Weatherhead, PJ AND K.W. Dufour. 2000. Fledging suc- cess as an index of recruitment in Red-winged Black- birds. Auk 117:627-633. Wiens, J.D. 2004. Post-fledging survival and natal dis- persal of Northern Goshawks in Arizona. M.S. thesis, Colorado State Univ., Fort Collins, CO U.S.A. Younk, J.V. and MJ. Bechard. 1994. Effect of gold min- ing activity on Northern Goshawks breeding in Ne- vada’s Independence and Bull Run mountains. Rap- tor Research Center, Department of Biology, Boise State University, Boise, ID U.S.A. Received 18 February 2004; accepted 5 April 2005 Associate Editor: Clint Boal J Raptor Res. 39 (3) :222-228 © 2005 The Raptor Research Foundation, Inc. PRODUCTIVITY AND MORTALITY OF NORTHERN GOSHAWKS IN MINNESOTA Clint W. Boal^ U.S. Geological Survey, Texas Cooperative Fish and Wildlife Research Unit, Texas Tech University, Lubbock, TX 79409 US. A. David E. Andersen U.S. Geological Survey, Minnesota Cooperative Fish and Wildlife Research Unit, University of Minnesota, St. Paul, MN 53108 U.S. A. Patricia L. Kennedy Eastern Oregon Agricultural Research Station and Department of Fisheries and Wildlife, Oregon State University, P.O. BoxE, Union, OR 97883 U.S. A. Abstract. — Compared to other regions of North America, little information exists regarding Northern Goshawk {Accipiter gentilis) ecology and population dynamics in the western Great Lakes Region. We examined productivity and nesting habitat characteristics of goshawks in Minnesota from 1998-2001. Apparent nesting success varied annually from as low as 38% to as high as 83%. The Mayfield estimate of daily survival for nests was 0.992 ± 0.002 (SE). The mean fledgling number across years was 1.85 ± 0.14 for successful nests and 1.14 ± O.D for all nesting attempts. Twenty-one percent of all nesting attempts failed, primarily due to predation or suspected predation (52%) and inclement weather (35%). Overall, productivity of goshawks in Minnesota was at the lower end of the range reported in other studies across western North America, which is not atypical for peripheral populations. During the 3-yr study, we recorded mortalities of nine (four males and five females; eight radio-marked and one un- marked) adult goshawks — causes of mortality were avian (33%) and mammalian (22%) predation, hu- man persecution (22%), and unknown causes (22%). Fifty-six percent of mortalities occurred during the breeding season, and 44% occurred during the winter. Based on radiotelemetry data, we estimated adult annual survival to be 74 ± 7.8%, which is similar to survival estimated using mark-recapture analysis in three western North America studies. Key Words: Northern Goshawk, Accipiter gentilis; breeding, Minnesota; mortality; productivity. PRODUCTIVIDAD YMORTALIDAD DE ACCIPITER GENTILIS YM MINNESOTA Resumen. — Existe poca informacion sobre la ecologia y la dinamica poblacional de Accipiter gentilis en el oeste de la region de Grandes Lagos comparado con otras regiones de America del Norte. Examinamos la productividad y las caracteristicas del ambiente de nidificacion de A. gentilis en Minnesota desde 1998 hasta 2001. El exito de nidificacion aparente vario anualmente de modo drastico, desde 38% a 83%. El estimado de Mayfield de la supervivencia diaria de los nidos fue 0.992 ± 0.002 (EE). El numero medio de volan tones a lo largo de los afios fue 1.85 ± 0.14 para los nidos exitosos y 1.14 ± 0.17 para todos los intentos de nidificacion. El 21% de todos los intentos de nidificacion fracaso, debido principalmente a la depredacion o a la supuesta depredacion (52%) y a las inclemencias del clima (35%). En total, la pro- ductividad de A. gentilis en Minnesota estuvo en el extremo inferior del rango reportado en otros estudios para el oeste de America del Norte, lo cual no es atfpico para poblaciones perifericas. Durante los tres anos de estudio, registramos la muerte de nueve adultos de A. gentilis (4 machos y 5 hembras; 8 marcados con transmisores y 1 sin marcar). Las causas de la mortalidad fueron depredacion por aves (33%) y mamiferos (22%), persecucion humana (22%) y causas desconocidas (22%). El 56% de las muertes ocu- rrieron durante la estacion reproductiva y el 44% durante el invierno. Basados en datos de radio telemetria, estimamos que la supervivencia anual de los adultos fue del 74 ± 7.8%, lo cual es similar a la supervi- vencia estimada usando analisis de captura-recaptura en tres estudios del oeste de America del Norte. [Traduccion del equipo editorial] ^ Email address: cboal@ttu.edu 222 September 2005 Biology 223 The Northern Goshawk {Accipiter gentilis) is a large, forest-dwelling raptor generally associated with mature deciduous, coniferous, or mixed for- ests. Possible conflicts between timber harvest practices and goshawk habitat requirements have led to concern for the species’ status (Kennedy 1997, United States Fish and Wildlife Service 1998). The goshawk has been proposed for listing several times under the U.S. Endangered Species Act and its status has been (and still is) the object of considerable litigation. In the western Great Lakes Region (WGLR) of North America, the gos- hawk is currently listed as a migratory non-game bird of management concern by the U.S. Fish and Wildlife Service (Region 3) and as a sensitive spe- cies by the U.S. Forest Service (Region 9). Few studies have examined goshawk productivity (Erd- man et al. 1998) and mortality in the WGLR. Re- gion-specific information on productivity and mor- tality factors is essential for development of sound management guidelines, but active management of the species in the WGLR has been hampered by the lack of data. In 1998, we initiated a broad-based ecological study of goshawks in Minnesota (Boal et al. 2001, Boal et al. 2003). Herein, we present the productivity and mortality data we collected on breeding goshawks in Minnesota, 1998-2000. Study Area The study area encompassed most of northern Min- nesota within the Laurentian Mixed-Forest Province (Minnesota Department of Natural Resources 2004; Fig. 1). Goshawks were distributed across the study area (Fig. 1), but a majority of goshawk nests were located on or near the Chippewa National Forest (47°23'N, 94°35'W). Study area elevation was ca. 200-400 m. Historical mean summer and winter temperatures were 18°C and — 11“C, respectively, with maximum and minimum temperature records of 40°C and — 46°C, respectively. Vegetation com- munities are described in Boal et al. (2003). Methods Study Population. We did not systematically survey for breeding goshawks, so known breeding pairs in a single year were likely a relatively small proportion of all gos- hawks breeding in the study area (Daw et al. 1998). How- ever, the goshawks monitored in this study were all known nesting goshawks in Minnesota during the study period of 1998-2000 (Boal et al. 2001, 2003). Nests in this study were from across the Laurentian Mixed-Forest Province of northern Minnesota and were likely repre- sentative of the Minnesota landscapes that goshawks use for nesting (Boal et al. 2001), but because our sample was not randomly selected, our inferences are limited to our sample. Before this study, few goshawk nesting areas were known in Minnesota. We searched known goshawk nest stands and areas where goshawks had been seen during previous breeding seasons. If a previous year’s nest was not occupied, we conducted tree-by-tree searches of the stand, up to 500 m from the old nest (if the stand was sufficiently large for this search pattern) . We also located new goshawk nest stands by searching likely areas or fol- lowing up on reports of probable goshawk nests located serendipitously by personnel from cooperating agencies and the timber industry. We considered an area to be occupied if one goshawk was observed in or near a known nest stand, radio-tagged hawks were located in the area, or other evidence of activity was observed (e.g., re- cent construction of a nest) . If an area was occupied by goshawks, we attempted to locate an occupied nest. An occupied nest was defined as a nest with eggs or young or the presence of an incubating goshawk. Productivity. Once an occupied nest was located, we made periodic visits to monitor reproductive success. We considered goshawks to be nesting if a female was ob- served in an incubation position on the nest or during later stages of the nesting period when young were ob- served in the nest. We considered nestlings to have sur- vived to fledge if they attained at least 80% of their first flight age (32 d old for goshawks; Boal 1994). We consid- ered a nesting attempt as successful if at least one young fledged. We estimated both apparent nest success (e.g., the proportion of monitored nests known to have fledged young) and nest success using the Mayfield esti- mate based on exposure days (Bart and Robson 1982). Because confidence intervals can be more informative than tests of statistical significance (Johnson 1999), we assessed differences in productivity by examining overlap of 95% confidence intervals. Nesting Failure. We attempted to determine cause of all nesting failures. In instances where dead adults or their remains were found at nests, we conducted in-field examinations of each carcass and location of death to attempt to identify the cause of death and, if depredated, the predator species. For example, claw marks ascending the nest tree and teeth marks on the carcass, feathers, and radio harness material of radio-tagged birds were in- dicative of mammalian predation (Einarsen 1956). In contrast, crimping plucks of feathers, stripped bones without tooth marks or evidence of mastication, single bill bite nips, and scrapes in bones indicated avian pred- ators (Einarsen 1956). Adult Mortality. There is little information on causes of adult mortality for raptors in general and goshawks m particular (Squires and Reynolds 1997). In addition to assessing causes of mortality of adult goshawks at nest sites, our sample of 32 radio-tagged goshawks (Boal et al. 2003) provided us with an opportunity to examine causes and timing of mortality among goshawks in Minnesota. We used telemetry to relocate all radio-tagged goshawks that died during the course of this study (Boal et al. 2001). We estimated the annual survival rate with the Kaplan-Meier survival model (Kaplan and Meier 1958) as modified by Pollock et al. (1989). Results We located 13, 19, and 21 areas occupied by gos- hawks in 1998, 1999, and 2000, respectively. Two 224 Boal et al. VoL. 39, No. 3 Minnesota Ecortglon Section Nam« K Minn«sota and Ontario Paatlanda K MInnaaota Df+tt and Lakt Plaint Northtm Su pari or Uplandt Southarn Suparlor Uplands Waatarn Suparlor Uplar>ds TO 140 210 280 S50 Km Figure 1. Study area and distribution of Northern Goshawk nests sites (open circles) included in this study, Min- nesota 1998-2000. Ecoregions are based on Minnesota Department of Natural Resources (2004), additional areas were located by cooperators in 1998, but were not reported to us until 1999. Al- though one of these nesting attempts was verified as successful, the two nests were not monitored to assess productivity. Thus, we only include the orig- inal 13 of 15 areas from 1998 for productivity as- sessment. Of the 15 breeding areas occupied in 1998, 11 (73%) were occupied in 1999. Of 23 known breeding areas occupied in 1998 and/or 1999, 13 (57%) were occupied in 2000. Although breeding did not occur in all occupied areas, 15 occupied areas were located in 1998, seven addi- tional areas in 1999, and nine additional areas in 2000, for a total of 31 areas occupied by goshawks at least 1 yr during the 3-yr study period. We did not monitor productivity at one nest, and two oth- ers failed in 1998. Sixteen (84%) pairs of goshawks from 19 occupied areas nested in 1999, and 15 (71%) pairs from 21 occupied areas nested in 2000. We observed that some areas were occupied by non-breeding goshawks. For example, a wid- owed female, radio-tagged in 1998, was tracked in her breeding area, but did not breed in 1999. Like- wise, in 2000 a widowed, non-breeding female roamed more widely than she had while breeding in 1999, but still occupied her 1999 breeding area. A pair that had been radio-tagged and successfully nested in 1999 occupied their breeding area, but did not nest in 2000. In contrast, after her mate died during the winter, one widowed female moved 15 km to pair with a male in a previously unknown breeding area the following spring. Productivity. We assessed success at 43 and pro- ductivity at 42 goshawk nests. Nesting success var- ied considerably among years, with a high of 83% in 1998 and a low of 37% in 1999. We observed that 67% of nesting attempts fledged young suc- cessfully in 2000 and the 3-yr mean for fledging success was 62 ± 23.4% (SE). Mayfield estimates for daily survival were 0.9998 ± 0.0006 (SE) in 1998, 0.985 ± 0.005 in 1999, and 0.993 ± 0.005 in 2000, with an overall daily survival rate of 0.992 ± 0.002. Based on a 32-d incubation period (Squires and Reynolds 1997) and a 32-d period to 80% of first flight age (Boal 1994), Mayfield estimates of nest success were 99% in 1998, 39% in 1999, 65% in 2000, and 59% over the 3-yr study period. Goshawk nests fledged a mean of 1.75 ± 1.05 September 2005 Biology 225 young in 1998 (N ~ 12), 0.81 ± 1.17 young in 1999 (N= 16), and 0.93 ± 0.80 {N— 15) young in 2000. Fledglings per nesting attempt were not statistically different between 1998 and 1999 {x difference = 0.938; 95% Cl = 0.057-1.820), between 1998 and 2000 {x difference = 0.817; 95% Cl = 0.082- 1.550), or between 1999 and 2000 (x difference = —0.121; 95% Cl = —0.861—0.619). In contrast, when examining only those nests that were suc- cessful, goshawks fledged a mean of 2.10 ± 0.74 young in 1998 {N — 10), 2.17 ± 0.75 young in 1999 {N = 6), and 1.40 ± 0.52 young in 2000 (A^ =10). Fledgling numbers at successful nests were not sta- tistically different between 1998 and 1999 (x dif- ference = —0.067; 95% Cl = —0.890-0.757), but were higher in 1998 (x difference = 0.700; 95% Cl = 0.102-1.300) and 1999 (x difference = 0.767; 95% Cl = 0.895—1.440) than in 2000. Mean num- ber of fledglings per nest for all years combined was 1.14 ± 1.07 for all nesting attempts and 1.85 ± 0.73 for successful nests only. Nesting Failure. Of the 43 goshawk nests moni- tored, two failed in 1998, 10 failed in 1999, and five failed in 2000. Of these 17 failures, 23% were due to mammalian predation, 18% were due to avian predation, and we suspected another 12% were due to predation, but we were unable to de- termine whether the predator was avian or mam- malian. Two of the mammalian predations resulted in mortalities of adult female goshawks (detailed below) . Weather contributed to 35% of nesting fail- ures, the majority of which occurred during the incubation stage of the nesting period in 1999 when the region experienced a 10-11 d period of almost constant rain. The cause of 12% of nesting failures was undetermined. Adult Mortality. Nine goshawks, eight of which were radio-tagged, died during this study. Five (56%; four females and one male) of these nine mortalities occurred during the breeding seasons. One female and one male were preyed upon by Great Horned Owls {Bubo virginianus) , two females were killed by mammals, and one female was con- sumed by a Red-tailed Hawk {Buteo jamaicensis) . The remaining four (44%) mortalities (one female and three males) occurred during the winter months. The female that died during the winter had been shot. The mortality of one male ap- peared to also be due to human actions; only the radio that had been attached to the male was re- covered and it had been obviously cut from the body of the goshawk. Furthermore, the radio lacked any mastication or pecking marks typical of those on goshawks that were depredated. The caus- es of death could not be verified for the other two male goshawks. We excluded the single, non-radio-marked fe- male that was killed in 1999 from estimates of sur- vival. The Kaplan-Meier estimate for annual surviv- al based on 32 radio-tagged goshawks was 74 ± 7.8%. Because our study was not originally de- signed to estimate adult survival, our survival esti- mate should be interpreted in the context of our sample of 32 radio-marked goshawks captured at breeding areas over the 3-yr study period. However, these data do provide some insight in the annual survival of goshawks in the study area. Discussion Productivity. We observed annual variability in fledglings produced per nest attempt (range = 0.87—1.85) and per successful nest (range = 1.40— 2. 1 7) during the 3 yr of this study. Such variability is typical of temporal patterns in reproductive suc- cess in goshawks (e.g., DeStefano et al. 1994, Ken- nedy 1997, McClaren et al. 2002). McClaren et al. (2002) found high temporal variability in produc- tivity among goshawk nests monitored 4—10 yr in three different populations in western North America. Within the WGLR, Erdman et al. (1998) reported fledgling numbers from Wisconsin gos- hawk nests over a 24-yr period, and found a mean of 1.7 fledglings per nesting attempt {N = 184) and 2.2 per successful nest {N = 138). However, Erdman et al. (1998) also indicated that fledging rates among nesting attempts decreased from the earlier years (1971-81) to later years (1982-92) of their study. We do not have historical data for this study area to evaluate temporal trends in produc- tivity, but the fledging success over the 3-yr period of monitoring did not vary statistically. Productivity in the Upper Peninsula of Michigan was similar to ours, with a reported 1.1 and 1.7 fledglings per occupied {N = 36) and successful {N — 24) nests, respectively (Lapinski 2000) . Fledgling rate among successful nests in our study and others conducted in the WGLR appear to be slightly lower than av- erage, but well within the range reported in studies from the western United States (e.g., Kennedy 1997, Boal and Mannan 1994, Bull and Hohmann 1994, Reynolds et al. 1994). Nesting Failure. The most common nest preda- tor of goshawk nests in North America appears to be Great Horned Owls (Kennedy 2003) , but wol- 226 Boal et al. VoL. 39, No. 3 verines (Gulo gulo; Doyle 1995) and fishers {Martes pennanti; Erdman et al. 1998) are known to prey upon goshawk nestlings, and raccoons {Procyon lo- tor) are also likely nest predators. Erdman et al. (1998) attributed predation by fishers as the pri- mary cause of nesting failure among goshawks in Wisconsin, but did not provide details for the basis of their conclusion, nor did they report the actual number of nesting failures due to fishers. Mammal depredation (suspected to be fishers) of nests in our study was comparatively low (9%), but collec- tive depredation (mammalian, avian, unknown) caused the failure of 21% of goshawk nests in Min- nesota. Weather can also influence productivity of gos- hawks. Cold weather and rain can reduce the num- ber of goshawk pairs attempting to nest (Kostrzewa and Kostrzewa 1990) and can lead to egg and chick (Zachel 1985) mortality. In our study, inclement weather accounted for failure of 12% of all nesting attempts. These failures occurred primarily during the incubation stage in 1999 when our study area experienced a 10-11 d period of almost constant rainfall. We suspect that some male goshawks may have been unable to provision their mates ade- quately during this period, eventually leading fe- males to either abandon their nests or temporarily leave their nests to forage, allowing the eggs to chill and die. Adult Mortality. Mortality data for goshawks in the WGLR are based almost solely on females found killed at or near their nests (Erdman et al. 1998). Thus, there are no data available prior to our study on causes of goshawk mortality away from their nests or during the non-breeding season. Our estimate of annual survival (74 ± 7.8%) based on telemetry was quite similar to mark-recapture estimates in Califor- nia (61-69%; DeStefano et al. 1994), New Mexico (60-96%; Kennedy 1997), and northern Arizona (69-87%; Reynolds and Joy 1998). All of these au- thors indicate imprecision in their studies due to a variety of reasons, and DeStefano et al. (1994) con- cluded accurate estimates of survival based on mark- resightings would require large numbers of marked birds, high resighting rates, and a minimum of 5 yr of data. This robust a data set has not been and is unlikely to be collected in the WGLR. In contrast, White and Garrott (1990) indicated survival estimates based on radiotelemetry requires smaller samples in general than mark-resighting estimates. Further- more, backpack radio attachments appear to have no significant effect on survival of goshawks (Reynolds et al. 2004) . Our data supported White and Garrott’s (1990) assertion; we were able to estimate survival rates with reasonable precision through marking con- siderably fewer birds than banding and resighting would require. However, we did not have a sufficient sample of radio-tagged birds to estimate temporal and gender differences in adult survival. One male that died during the winter of 1999- 2000 had been banded as a juvenile at Hawk Ridge, MN, during the fall migration of 1988 (D. Evans pers. comm.). This male and his mate had fledged two young successfully in 1999. To our knowledge this 11-yr male is the oldest known recorded breed- ing male goshawk reported for North America. In- terestingly, the oldest reported female goshawk (12 yr old) in North America was also reported from Minnesota (Evans 1981). The majority of information on causes of mor- tality among adult goshawks is anecdotal (Squires and Reynolds 1997). Goshawks succumb to several different diseases and parasites (Redig et al. 1980, Ward and Kennedy 1996, C. Boal unpubl. data). Accidents and injuries, such as flying into windows (C. Boal unpubl. data) or choking on prey (Blox- ton et al. 2002), also result in mortality. The pri- mary documented cause of mortality among free- ranging goshawks, however, appears to be predation (Squires and Kennedy in press). Known predators of adult goshawks include Great Horned Owls (Rohner and Doyle 1992, Boal and Mannan 1994, Erdman et al. 1998), Golden Eagles {Aquila chrysaetos; Squires and Ruggiero 1996), Pine Martens {Martes americana; Paragi and Wholecheese 1994), and fishers (Erdman et al. 1998). Of five adult goshawks taken by predators in our study, two were killed by Great Horned Owls. In the WGLR, fishers may also be an impor- tant predator of goshawks; predation by fishers was identified as the cause of mortality for four adult female goshawks in Wisconsin (Erdman et al. 1998) and two of five goshawk deaths in our study. We believe the goshawk killed in our study by a Red-tailed Hawk may be an exceptional incident, but the two species have been observed engaged in physical agonistic encounters (Crannell and DeStefano 1992, C. Boal unpubl. data). In areas of sympatry (La Sorte et al. 2004), Red-tailed Hawk predation may be more common. Most mortality data for goshawks is for the nest- ing season. We found that mortality occurred with equal frequency in the breeding and winter sea- sons, suggesting that survival outside of the breed- September 2005 Biology 227 ing season is an important aspect of goshawk pop- ulation dynamics. Our data also suggested that, despite legal protection, persecution was still a fac- tor affecting goshawk survival. Results from Wisconsin (Erdman et al. 1998) and our study suggested predators were a major cause of goshawk mortality in the WGLR. However, the influence of predators on goshawk population demography and whether current predation rates are similar to historic rates or have increased as a consequence of human activities (e.g., timber har- vest, reintroduction of fishers) in the WGLR, as suggested by Erdman et al. (1998), has yet to be assessed rigorously. The development and use of standardized field methods for evaluating causes of mortality of goshawks and publication of existing mortality data would be helpful in this regard. Without reliable survival data, rates of population growth or decline cannot be estimated accurately for the WGLR goshawk population. Acknowledgments We appreciate our dedicated field assistants that en- dured long hours of work: L. Belmonte, W. Estes, J. Sam- mler, R. Sandstrom, and A. Wester. A. Roberson and B. Smithers assisted with practically all aspects of this study, and M. Solensky assisted with trapping. D. Harrington and A. Weaver provided nest sites, and members of the Minnesota Falconers’ Association assisted with surveys. Personnel from the many cooperating agencies and or- ganizations provided an array of assistance and logistical support during the projects. They include R. Baker, J. Blanchard, J. Casson, M. Cole, J. Gallagher, L. Grover, M. Hamady, J. Hines, M. Houser, G. Kirk, B. Marty, R. Vora, and A. Williamson. Funding for this project was provided by the Chippewa National Forest, The National Council for Air and Stream Improvement, Minnesota Department of Natural Resources, Potlatch Corporation, Leech Lake Band of Ojibway, Superior National Forest, U.S. Fish and Wildlife Service, The Raptor Center at the University of Minnesota, The Minnesota Falconers’ Association, Voya- geurs National Park, the Minnesota Cooperative Fish and Wildlife Research Unit, and grants from the National Forest Foundation and the National Fish and Wildlife Foundation. Literature Cited Bart, J. and D.S. Robson. 1982. Estimating survivorship when the subjects are visited periodically. Ecology 63: 1078-1090. Bloxton, T.D., A. Rogers, M.F. Ingraldi, S. Rosen- stock, J.M. Marzluef, and S.P. Finn. 2002. Possible choking mortalities of adult Northern Goshawks. J. Raptor Res. 36:141—143. Boat, C.W. 1994. A photographic and behavioral guide to aging nestling Northern Goshawks. Stud. Avian Biol. 16:32-40. , D.E. Andersen, and P.L. Kennedy. 2001. Home range and habitat use of Northern Goshawks {Acciptter gentilis) in Minnesota. Minnesota Cooperative Fish and Wildlife Research Unit, St. Paul, MN U.S.A. , , AND . 2003. Home range and res- idency status of Northern Goshawks breeding in Min- nesota. Condor 105:811— 816. AND R.W. Mannan. 1994. Northern Goshawk diets in ponderosa pine forests on the Kaibab Plateau. Stud. Avian Biol. 16:97—102. Bull, E.L. andJ.H. Hohmann. 1994. Breeding biology of Northern Goshawks in northeastern Oregon. Stud Avian Biol. 16:103-105. Crannell, D. and S. DeStefano. 1992. An aggressive in- teraction between a Northern Goshawk and a Red- tailed Hawk. J. Raptor Res. 26:269-270. Daw, S.K., S. DeStefano, and R.J. Steidl. 1998. Does sur- vey method bias the description of Northern Goshawk nest-site selection? J. Wildl. Manag. 62:1379-1384. DeStefano, S.K. Daw, S.M. Desimone, and E.C. Meslow. 1994. Density and productivity of Northern Goshawks: implications for monitoring and management. Stud. Avian Biol. 1 6 : 88-9 1 . , S., B. WOODBRIDGE, AND PJ- Detrich. 1994. Sur- vival of Northern Goshawks in the southern Cascades of California. Stud. Avian Biol. 16:133—136. Doyle, F.I. 1995. Bald Eagle {Haliaeetus leucocephalus) and Northern Goshawk (Accipiter gentilis) nests apparently preyed upon by a wolverine (s) , (Gulo gulo), in the southwestern Yukon Territory. Canad. Field-Nat. 109: 115-116. Einarsen, A.S. 1956. Determination of some predator species by field signs. Oregon State Monogr. Stud. Zool 10 . Erdman, T.C., D.F. Brinker,J.P. Jacobs, J. Wilde, andT.O. Meyer. 1998. Productivity, population trend, and sta- tus of Northern Goshawks, Accipiter gentilis atricapillus, in northeastern Wisconsin. Canad. Field-Nat. 112:17- 27. Evans, D.L. 1981. Banding recoveries from Hawk Ridge. Prairie Nat. 32:18. Johnson, D.H. 1999. The insignificance of statistical sig- nificance testing. J. Wildl. Manag. 63:763—772. Kaplan, E.L. and P. Meier. 1958. Nonparametric esti- mation from incomplete observations./. Am. Statistical Assoc. 53:457-481. Kennedy, P.L. 1997. The Northern Goshawk {Accipiter gen- tilis atricapillus) : is there evidence of a population de- cline? J. Raptor Res. 31:95-106. . 2003. Northern Goshawk {Accipiter gentilis atricap- illus) : a technical conservation assessment. USDA For- est Service, Rocky Mountain Region, Species Conser- vation Project. Published as a web document by USFS at http://www.fs.fed.us/ r2/scp/species_assessments/ northern-goshawk/ northern_goshawk.pdf. Kostrzewa, a. and R. Kostrzewa. 1990. The relationship of spring and summer weather with density and 228 Boal et al. VoL. 39, No. 3 breeding performance of the buzzard Buteo buteo, gos- hawk Accipiter gentilis, and kestrel tinnunculus. Ibis 132:550-559. Lapinski, N.W. 2000. Habitat use and productivity of the Northern Goshawk in the upper peninsula of Michi- gan. M.S. thesis, Northern Michigan Univ., Mar- quette, MI U.S.A. La Sorte, F.A., R.W. Mannan, R.T. Reynolds, and T.G. Grubb. 2004. Habitat associations of sympatric Red- tailed Hawks and Northern Goshawks on the Kaibab Plateau./. Wildl Manag. 68:307-317. McClaren, E., P.L. Kennedy, and S.R. Dewey. 2002. Do some Northern Goshawk nest areas fledge more young than others? Condor 104:343-352. Minnesota Department of Natural Resources. 2004. Ecological classification system. Minnesota Depart- ment of Natural Resources, St. Paul, MN U.S.A. http:/ /www.dnr.state.mn.us/ecs/index.html Paragi, T.G. AND G.M. Wholecheese. 1994. Marten, Mar- ies americana, predation on a Northern Goshawk Ac- cipiter gentilis. Canad. Field-Natur. 108:81-82. Pollock, K.H., S.R. Winterstein, C.M. Bunck, and P.D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design./. Wildl. Manag. 53:7-15. Redig, P.T., M.R. Fuller, and D.L. Evans. 1980. Preva- lence of Aspergillus fumigatus in free-living goshawks {Accipiter gentilis atricapillus) . J. Wildl. Diseases 16:169— 174. Reynolds, R.T. and S.M. Joy. 1998. Distribution, territory occupancy, dispersal and demography of Northern Goshawks on the Kaibab Plateau, Arizona: final re- port. Heritage Project No. 194045. Arizona Game and Fish Department, Phoenix, AZ U.S.A. , , and D.G. Leslie. 1994. Nest productivity, fidelity, and spacing of Northern Goshawks in north- ern Arizona. Stud. Avian Biol. 16:106-113. , G.C. White, S.M. Joy, and R.W. Mannan. 2004. Effects of radiotransmitters on Northern Goshawks: do tailmounts lower survival of breeding males ? J. Wildl. Manag. 68:25-32. Rohner, C. and F.I. Doyle. 1992. Food-stressed Great Horned Owl kills adult goshawk: exceptional obser- vation or community process? /. Raptor Res. 26:261- 263. Squires, J.R. and P.L. Kennedy. In press. Northern Gos- hawk ecology: an assessment of current knowledge and information needs for conservation and manage- ment. Studies Avian Biology. AND R.T. Reynolds. 1997. Northern Goshawks {Accipiter gentilis) . In A. Poole and F. Gill [Eds.] , The birds of North America, No. 298. The Academy of Natural Sciences. Philadelphia, PA, and The Ameri- can Ornithologists’ Union, Washington, DC U.S.A. and L.F. Ruggiero. 1996. Nest-site preference of Northern Goshawk in south-central Wyoming. /. Wildl. Manag. 60:170-177. United States Fish and Wildlife Service. 1998. Notice of 12-month finding on a petition to list the Northern Goshawk in the contiguous United States west of the 100* meridian. Federal Register 63:35183-35184. Ward, J.M. and P.L. Kennedy. 1996. Effects of supple- mental food on size and survival of juvenile Northern Goshawks. Auk 113:200-208. White, G.C. and R.A. Garrott. 1990. Analysis of Wildlife Radio-tracking Data. Academic Press, San Diego, CA U.S.A. Zachel, C.R. 1985. Food habits, hunting activity, and post-fledging behavior of Northern Goshawks {Accip- iter gentilis) in interior Alaska. M.S. thesis, Univ. Alas- ka, Fairbanks, AK U.S.A. Received 8 February 2004; accepted 27 September 2004 Guest Editor: Stephen DeStefano /. Raptor Res. 39(3):229-236 © 2005 The Raptor Research Foundation, Inc. RELATIONSHIPS BETWEEN WINTER AND SPRING WEATHER AND NORTHERN GOSHAWK (ACCIPITER GENTILIS) REPRODUCTION IN NORTHERN NEVADA Graham D. Fairhurst^ and Marc J. Bechard Raptor Research Center, Department of Biology, Boise State University, Boise, ID 83725-1365 U.S.A. Abstract. — Ecological factors, such as weather, play important roles in raptor population dynamics. We used logistic and Poisson regression analyses to investigate relationships between late winter, spring, and early summer temperatures and precipitation and Northern Goshawk {Accipiter gentilis) breeding, failure, and productivity in northern Nevada from 1992-2002. We also examined weather data for possible patterns that could explain reported trends in goshawk reproduction. Declines in occupancy of nesting territories by breeding goshawks were related to colder February and March temperatures and increased April precipitation. Warmer April temperatures and decreased precipitation in April-July favored re- productive success. Of all significant weather variables, only February and March temperatures had significant temporal trends. Although adverse weather is known to affect goshawk reproduction by decreasing nestling growth and survival, it is unlikely that direct weather effects were responsible for reported reproductive trends in our study area. Weather may have operated indirectly, influencing reproduction through changes in goshawk hunting behavior or food supply. Key Words; Northern Goshawk, Accipiter gentilis; weather, breeding, population trends', Nevada. RELACIONES ENTRE EL CLIMA DE INVIERNO YDE PRIMAVERA Y LA REPRODUCCION DE AC- CIPITER GENTILIS EN EL NORTE DE NEVADA Resumen. — Factores ecologicos como el clima tienen un papel importante en la dinamica poblacional de las aves rapaces. Utilizamos analisis de regresion logistica y de Poisson para investigar las relaciones entre las temperaturas y las precipitaciones de fines del invierno, de la primavera y del comienzo del verano, y los fracasos o exitos reproductivos y la productividad de Accipiter gentilis en el norte de Nevada entre 1992 y 2002. Tambien examinamos los datos de clima para encontrar posibles tendencias que puedan explicar las tendencias documentadas de la reproduccion de estos halcones. La disminucion en la ocupacion de territorios de nidificacion por halcones reproductivos se relaciono con las tempe- raturas mas Mas de febrero y marzo y el aumento de las precipitaciones en abril. Las temperaturas mas calidas de abril y la disminucion de las precipitaciones en abril^ulio favorecieron el exito reproductivo. De todas las variables climaticas significativas, solo las temperaturas de febrero y marzo presentaron tendencias temporales significativas. A pesar de que es sabido que las condiciones climaticas adversas afectan la reproduccion de estos halcones al disminuir el crecimiento y la supervivencia de los polluelos, es poco probable que los efectos directos del clima fueran responsables de las tendencias reproductivas documentadas en nuestro sitio de estudio. Las condiciones climaticas pueden haber operado indirec- tamente, influenciando la reproduccion a traves de cambios en el comportamiento de caza de los halcones, o en la disponibilidad de alimento. [Traduccion del equipo editorial] Weather can directly influence Northern Gos- hawk {Accipiter gentilis) population dynamics by af- fecting survival (Zachel 1985, Squires and Reyn- olds 1997, Bloxton 2002), movements (Marcstrom and Kenward 1981, Squires and Ruggiero 1995), and nestling development (Kostrzewa and Kostrze- ^ Corresponding author’s email address: gdfair@gmail. com wa 1990). Temperature and precipitation may also indirectly affect prey populations (Van Horne et al. 1997, Bloxton 2002), foraging behavior (Zachel 1985), and other mortality factors (Newton 1979). Studies addressing the relationships between weather and goshawk reproduction (Kostrzewa and Kostrzewa 1990, 1991, Patla 1997, Penteriani 1997, Ingraldi 1998, Bloxton 2002), generally agree that colder and wetter spring weather nega- 229 230 Fairhurst and Bechard VoL. 39, No. 3 lively affects goshawk reproduction; however, the association between winter weather and goshawk reproduction is not well studied. These studies have also only considered time periods <6 yr (In- graldi 1998). We previously reported declines in goshawk nest- ing territory occupancy and increases in breeding failure in northern Nevada from 1992-2002 (Be- chard et al. in press). Determining the ecological factors responsible for these reproductive trends is difficult because any variable that showed a tem- poral trend from 1992-2002 will consequentially be correlated with reproduction. However, because of the known links between weather and goshawk reproduction and the abnormally low precipitation and drought conditions reported in the northern Great Basin from 1999-2002 (National Drought Mitigation Center 2003), we suspected that weath- er conditions affected goshawk reproduction in northern Nevada. Here, we address the associa- tions between late winter, spring, and early sum- mer temperature and precipitation and long-term trends in goshawk reproductive performance. Methods Study Area. We conducted the study in the Indepen- dence and Bull Run Mountain ranges of Elko County, northern Nevada, during 1992-2002. The study area ex- tended ca. 150 km north-to-south, 10-30 km east-to-west, and encompassed ca. 94 000 ha. The area is a mosaic of public lands administered by the United States Depart- ment of Agriculture Forest Service (Humboldt-Toiyabe National Forest) interspersed with private lands. Eleva- tions range from ca. 1700-3000 m on the highest peaks. A mixture of land uses occurred in the study area, in- cluding cattle ranching, gold mining, and outdoor rec- reation (hunting, camping, and off-road vehicle use). The sagebrush steppe was typified by big sagebrush {Artemisia tridentata) , bitterbrush {Purshia tridentata), and rabbitbrush {Chrysothamnus . Common native grass- es included native bluebunch wheatgrass {Pseudoroegneria spicata) and Idaho fescue {Festuca idahoensis), and intro- duced cheatgrass (Bromus tectorum) and medusahead wildrye (Taeniatherum caput-medusae) also occur (Loope 1969; P. Jelinek, Humboldt-Toiyabe National Forest, pers. comm.). In the study area, goshawks nested exclusively in quaking aspen (Populus tremuloides) , which occurred in naturally-fragmented stands where sufficient moisture was present. Subalpine fir {Abies lasiocarpa) replaces quak- ing aspen above elevations of 2500 m (Loope 1969). Dense willow {Salix spp.), thickets, and cottonwoods {Populus spp.) occurred in riparian areas at all elevations. Field Methods. In April of 1991, 1992, and 1994-96, we used helicopters to initially locate and subsequently survey all historical goshawk nesting territories in the study area. We defined a nesting territory as the area containing one or more nests occupied by a single pair of goshawks in any breeding season (Postupalsky 1974, Woodbridge and Detrich 1994, Reynolds and Joy 1998) We discontinued helicopter surveys after 1996. Beginning in mid-May of each year from 1992-2002, we conducted ground surveys of all historically-occupied nesting territories (i.e., previously located by helicopter and used by goshawks) to determine occupancy by gos- hawk breeding pairs. We conducted ground surveys on foot and with all-terrain vehicles by returning to nesting territories and thoroughly searching stands and adjacent stands for the presence of breeding goshawks. Because territories in our study area were relatively small and nest structures obvious, we were able to search each territory completely and, therefore, assume a uniform probability of detection. We were unable to reach all nesting terri- tories each year because roads throughout the study area were periodically snow-covered or washed-out, and pri- vate land was not always accessible. Beginning in mid-June of each year, we revisited oc- cupied nesting territories to determine productivity (Be- chard et al., in press). We climbed all occupied nest trees and counted and banded nestlings when they were ap- proximately 21-31 d old (age based on nestling plumage, Boal 1994). We considered a pair failed if there was no sign of goshawks (adults or nestlings) at or near an oc- cupied nest when it was revisited in June. Data Analyses. We determined nesting territory occu- pancy, productivity per breeding pair, breeding failure, and productivity per successful pair. Because we visited most nests only twice during the breeding season, there was as much as a 30-d interval between our nest visits. Therefore, we could not always determine the cause of a nest failure. There were no instances that clearly indicat- ed the nesting attempt had failed due to depredation or any other factor unrelated to weather; therefore, we in- cluded all failures in our analysis. We downloaded weather data for the study area from the Natural Resources Conservation Service/SNOTEL website (Jack’s Creek Upper weather station; Natural Re- sources Conservation Service 2003). Because SNOTEL considered temperature data for 2002 unreliable, we did not include those data. To analyze 11-yr trends in reproduction, we used logis- tic regression for binary outcomes (i.e., nesting territory occupancy and nesting failures) and Poisson regression for counted outcomes (i.e., productivity). To account for the fact that repeated measurements on a nesting terri- tory over time might not be statistically independent (Al- lison 1999), we used Generalized Estimating Equations (“GEE;” PROC GENMOD; SAS Institute, Inc., Cary, NC U.S.A.), which allowed us to cluster our data by nesting territory (Stokes et al. 2000). We modeled mean daily temperature (°C), cumulative monthly precipitation (cm; Fig. 1,2), and year as contin- uous explanatory variables. We used only weather data from January-July, as we felt that they were most biolog- ically relevant. We considered individual months and groups of months (i.e., January and February, February and March) in analyses, resulting in 27 possible explan- atory variables per reproductive outcome (13 for mean daily temperature, 13 for cumulative monthly precipita- tion, and 1 for year). Due to low sample sizes {N = 11 yr) , we did not consider more than one explanatory var- iable per model. Therefore, to avoid over specifying our September 2005 Biology 231 0 01 £ 3 £ 0) Q. E 0) to Q c cd <]} Year Figure 1. Mean daily temperature for January-July in the Independence and Bull Run mountains, Nevada for 1992- 2002. models, we ran separate univariate models for each ex- planatory weather variable and employed a pre-screening procedure to decide which models best explained the relationship between weather and goshawk reproduction. During pre-screening, we only considered weather var- iables that appeared to be biologically relevant. For ex- ample, July weather would not be biologically related to occupancy in the same year because occupancy occurs before July. In addition, for each of the four reproductive outcomes, we ran several competing univariate models: each competing model used weather data from a single month or group of months as the explanatory variable. For example, II competing occupancy models used mean daily temperature as the explanatory variable, and an additional II occupancy models used cumulative monthly precipitation. For each reproductive outcome, we then selected the models with the single most statis- tically significant temperature and precipitation variables. Thus, for each reproductive outcome, we presented three separate univariate models: one for temperature, one for precipitation, and one for year. To avoid inflated Type I error rates, we assessed significance of all models using a step-down Bonferroni correction (Holm 1979). Although our statistical method has the potential to pro- duce spurious results (Freedman 1983), the GEE has no other measure by which to assess multiple competing models. We evaluated the results of logistic regression models by exponentiation of the model coefficient to obtain odds ratios, and we evaluated the results of Poisson re- gression by exponentiation of the model coefficient to obtain percent increase in the mean values of dependent variables. In the analyses, we only included nesting ter- ritories for which we knew the reproductive outcome. To determine if any patterns existed in local weather variables, we used simple linear regression (JMP IN; SAS Institute, Inc., Cary, NC U.S.A.). For each weather vari- able that was significantly related to occupancy and fail- ure (one temperature variable and one precipitation var- iable per each reproductive outcome) , we regressed year against the temperature or precipitation variable. Results We initially located 27 nesting territories in 1992, and found five, five, and four additional nesting territories in 1993, 1994, and 1996, respectively. We monitored a mean of 32.5 ± 4.7 (x ± SD) nesting territories annually (Table 1). Goshawks occupied an average of 20.3 ± 6.7 of these nesting territories each year. The odds of nesting territory occupancy by breeding pairs increased by 55.8% with each 1°C increase in combined February and March mean daily temperature (odds ratio = 1.558, P = 0.018; Table 2). The odds of occupancy of a nest- ing territory increased by 7.7% for every cm in- crease in cumulative April precipitation (odds ratio = 1.077, P — 0.03). We detected a significant cool- ing trend in combined February and March mean daily temperature in our study area (r^ = 0.41, P — 0.04) , but we found no trend in cumulative April precipitation (i^ < 0.0002, P = 0.97). Goshawk breeding pairs fledged a mean of 2.27 ± 0.76 young annually. Mean productivity per breeding pair increased 10.5% for every 1°C in- crease in the mean daily April temperature (Table 3; percent change in mean = 1.105, P < 0.0001). Mean productivity per breeding pair decreased by 1.9% for every 1 cm increase in combined April and May precipitation (percent change in mean = 0.981, P< 0.0003). 232 Fairhurst and Bechard VoL. 39, No. 3 E o. c o '-4-» CO ■q. o CD k. Q. x: c o .1 3 E 3 o Jan Feb Mar Apr May Jun Jul • 1992 o 1993 Y 1994 V 1995 ■ 1996 □ 1997 ♦ 1998 O 1999 A 2000 A 2001 • 2002 Month Figure 2. Monthly precipitation in the Independence and Bull Run Mountains, Nevada, for 1992—2002, expressed as cumulative monthly totals for January-July. On average, 13.5% of breeding attempts failed. The odds of failure decreased by 40.5% with every 1°C increase in mean daily April temperature (Ta- ble 2; odds ratio = 0.595, P < 0.0007) and in- creased by 8.7% with each cumulative 1 cm in- crease in combined May and June cumulative precipitation (odds ratio = 1.087, P< 0.006). We found no trends in April (r^ = 0.094, P — 0.39) or combined May and June (r^ = 0.007, P = 0.80) temperatures. Successful pairs produced a mean of 2.64 ± 0.57 young. For each 1°C increase in mean daily April temperature, we found a 6.4% increase in mean productivity per successful pair (Table 3; percent change in mean = 1.064, P < 0.0001). Mean pro- ductivity per successful pair decreased by 3.2% for each cumulative cm of combined June and July precipitation (percent change in mean = 0.978, P = 0.042). Discussion Long-term trends in goshawk reproduction were significantly related to weather, with a stronger in- fluence of temperature than of precipitation. Al- though late winter temperatures decreased in the study area from 1992—2002, our results suggested that warmer late winter temperatures favored gos- hawk breeding. Decreased productivity has been related to colder and wetter spring weather (Kos- trzewa and Kostrzewa 1990, Patla 1997, Penteriani 1997, Bloxton 2002). Colder temperatures increase energetic stresses and increase dietary demands on raptors and can result in non-laying (Newton 1979). Studies correlating winter temperatures with North American goshawk reproduction are lacking, but in European goshawks winter temper- atures were not related to occupancy by breeding pairs (Kostrzewa and Kostrzewa 1991). However, the larger size of European goshawks may make them more robust to temperature and energetic demands early in the breeding season than the smaller North American subspecies (Kendleigh 1970). Our finding of increased April precipitation fa- voring occupancy by breeding pairs was unusual, and we found no previous studies to support this result. Moreover, increased precipitation early in the breeding season is typically associated with re- duced numbers of breeding pairs (Kostrzewa and Kostrzewa 1990, Ingraldi 1998, Bloxton 2002). Per- haps our finding of statistical significance does not necessarily relate to biological relevance, and the significant result is spurious. The temporal trend in the failure of breeding attempts was strongly related to April temperature and cumulative May and June precipitation. In- creased precipitation and decreased temperatures during the egg-laying and early nestling periods can increase egg and nestling mortality rates (Hog- lund 1964, Zachel 1985) and affect nestling devel- opment (Kostrzewa and Kostrzewa 1990). Table 1. Annual reproductive performance of Northern Goshawk nesting territories based on surveys conducted May-June 1992-2002 in the Indepi Biology 233 September 2005 c u cn CO OJ t-i 03 u <u P3 ii o <u ;-i CO 3 O Si CX B ti Tj > lU z co" !=1 a o a CP t 3 § (U u p <u T3 p 3 X O X o CM 00 CM (SO cc I> X 1> Sd P 1> 00 (J3 03 <o q q CM in q , Dh 5 CO d d d d d d d rH d rH d d 0 2 0 Z P CO M U U S +1 o +1 00 +1 CO +1 CO + 1 rH +1 +1 fr + 1 J> +1 03 + 1 o + 1 o 1 +1 P Oi oq I> in q CO rH q q m q |X tzs pel g 4 CM CM CM CM CM ci ci ob (ci (ci ci ps; w CM CM O in CO X m o X C7i CM 03 00 rH q q i> No. Pi o z p 0 z HH Q d +1 +1 00 I— s +1 nH +1 d +1 on d +1 m rH +1 CM rH + 1 CO d + 1 <03 rH + 1 CM rH + 1 GO 1 d +1 .0 w r^ o oq o CM q q q q q 3 P3 CN CM CM rH ci CM CM rH cb rH rH (ci X d 2 •o 0 00 rH X Z t-h o J> X <n 03 o CO rH m o X + 1 P X in -cf CO 'vh CM CO CM rH 1> P 0 'Cf^ q H GO S pi u X in in q G q O rH q q X CM d 00 <o in cb <03 d GO ci 1 GO t-H CM rH CM CM •rri rH (X X p X bh CO cb d c« 03 rH CM o 00 03 in CM X o GO +1 2 X o CM CM CM CM CM rH rH rH rH (03 U IX rH q P C/3 rH 0 rH z rC C/3 cb 0 Q p 3 CM (03 in o (03 00 i> X CO i> CM +1 2 CM CM X CM 00 rH i—S rH rH rH rH § CM q PQ <03 rH X U x' < in S-H on q CM q X q <o q CO 1 ^H Ph P u 1— ( 00 d (m (m (SO 'Cf f — ( ob CM 1 +1 00 J> 00 m m in -'f CM 'Cf u 0 ci CO o CO z W Q S X n cb C/3 w 0 X p u u CM in X i> o o 00 j> X X X +1 GO 2 CM CM CM CM eo CM rH I — s rH rH CM CM d X 0 d 2 H CM c/3 0 X s 0 s 1> z 0 P § x> CM 1> i> t-H CO CO X o rH X + 1 m ci 2 u 2 X p< p c/5 CM on on en •'f CM (30 CO eo oo GO m X H GO CM SSO in CO 1> X 03 O rH CM 23 u c rt CJ3 03 (03 <03 (03 (03 03 03 o o O r® 03 03 <03 03 (03 03 03 (03 o o o lH rH 1^ rH i-H CM CM CM T3 u a E Si a 11 T3 w O S5 bsD . s T3 H U 03 I £r- ;s <J zs T5 S a. II Si SU XI 2 U a, D "O I u 3 X 0 Si S3- X u 1 a Al o3 o- <U 'S - s -S (J +j c2 '? ^ I 8 a 3 ^ X a 3 SJ V X 3 bo c u 'c3 a Cl. bo eft .s T3 V qj G u CC Xi QJ ^H TJ o eft 4-* eft G Qj V U CJ U u C V CL, £ u 234 Fairhurst and Bechard VoL. 39, No. 3 Table 2. Odds ratios, confidence intervals (Cl), and significance of weather variables related to Northern Goshawk nesting territory occupancy in the Independence and Bull Run Mountains, Nevada, for 1992—2002. Model Term Odds Ratio 95% CP P-value^ Occupancy by breeding pairs Year‘s 0.785 0.701-0.880 <0.0002 Mean daily combined February and March temperature 1.558 1.135-2.138 0.018 Cumulative April precipitation 1.077 1.014-1.143 0.03 Failure Year 1.157 1.021-1.312 0.044 Mean daily April temperature 0.595 0.455-0.780 <0.0007 Cumulative combined May and June precipitation 1.087 1.031-1.146 <0.006 ^ Odds ratios are nonsignificant if confidence interval covers 1.0 (even odds). Significance of terms assessed using a step-down Bonferroni adjustment (Holm 1979). Data for year model terms taken from Bechard et al. (in press). Despite evidence that weather can directly affect goshawk breeding, it is unlikely that direct weather effects are solely responsible for our reported trends in reproduction (Newton 1998). We found no significant temporal trends in weather related to nest failure, suggesting changes in that repro- ductive variable were due to other factors such as reduced hunting and food provisioning due to continued rainfall (Zachel 1985, Bloxton 2002). Also, depredation of goshawk nests can result in nest failures. Because our nest visits were several weeks apart, and we could not determine the exact cause of nest failure in all cases, we included all failures in our analysis, possibly biasing our results. Nevertheless, we found no direct evidence indicat- ing that other factors unrelated to weather, such as depredation by Great Horned Owls {Bubo vir^ni- anus), played a significant role in the breeding fail- ures we observed. The confounding influences of unmeasured, but plausible, factors that may have changed during the study period complicated the analysis. Obvious among these was a possible trend in prey popula- tions. Goshawks respond numerically to changes in numbers of prey (McGowan 1975, Doyle and Smith 1994). In our study area, they relied heavily on Table 3. Percent change in mean, confidence intervals (Cl), and significance of weather variables related to North- ern Goshawk productivity in the Independence and Bull Run Mountains, Nevada, for 1992-2002. Model Term Percent Change IN Mean 95% CP P-value’^ Productivity per breeding pair Year‘s 0.985 0.963-1.008 0.20 Mean daily April temperature 1.105 1.070-1.140 <0.0001 Cumulative combined April and May precipitation 0.981 0.971-0.990 <0.0003 Productivity per successful pair Year 1.003 0.989-1.017 >0.90 Mean daily April temperature 1.064 1.040-1.089 <0.0001 Cumulative combined June and July precipitation 0.978 0.962-0.994 0.042 “‘Percent changes in the means are nonsignificant if confidence interval covers 1.0 (even odds). •’ Significance of terms assessed using a step-down Bonferroni adjustment (Holm 1979). Data for year model terms taken from Bechard et al. (in press). September 2005 Biology 235 Belding’s ground squirrels {Spermophilus beldingi) for food (Younk and Bechard 1994, Younk 1996), but because we did not census ground squirrels, we could not determine what effect change in ground squirrel populations had on goshawk re- production. Further, interactions of weather and prey abun- dance may affect raptor reproduction (Gargett et al. 1995, Steenhof et al. 1997, Bloxton 2002). Weather has been shown to affect ground squirrel populations in other parts of the northern Great Basin (Van Horne et al. 1997). Bloxton (2002) at- tributed increased breeding failure and signifi- cantly lower productivity to reduced abundances of goshawk prey species following the wet and cold winter and spring of a La Nina weather event in western Washington. He noted that goshawks did not breed if weather had affected prey popula- tions. Although the climate of the northern Great Basin differs markedly from western Washington, the interactive effects of weather and prey may not. Acknowledgments This project was supported by the Nevada Department of Wildlife, U.S. Department of Agriculture Forest Ser- vice (Humboldt-Toiyabe National Forest), Independence Mining Company (now AngloGold-Meridian Jerritt Can- yon Joint Venture), and Boise State University. We would like to thank P. Bradley of the Nevada Department of Wildlife for his foresight in initiating this much-needed study of goshawks in an unusual habitat. We are deeply grateful to Forest Service biologists W. Amy, S. Anderson, P. Jelinek, J. Warder, and B. Whalen for all their efforts through the years to help with this project. Without the support of Independence Mining Company this project would never have been possible. We would like to espe- cially thank J. Bokich, J. Campbell, L. Gionet-Sheffield, G. Goodrich, M. Jones, S. Lewis, K. McAdoo, andj. Parks at the mine for their support. G. Fairhurst was supported by a Dan Montgomery Graduate Research Fellowship while at Boise State University. We are indebted to J. Younk, M. Shipman, and H. Smith for their substantial contributions to this project, and to L. Bond for her sta- tistical help. We would like to thank all the field assistants who contributed to this project. We are grateful to re- viewers P. Beier, C. Boal, and M. Goldstein for providing valuable comments and suggestions. Finally, we would like to thank the organizers of the goshawk symposium, Editors C. Boal and J. Bednarz for their hard work. Literature Cited Allison, P.D. 1999. Logistic regression using the SAS® system; theory and application. SAS Institute, Inc., Cary, NC U.S.A. Bechard, M.J., G.D. Fairhurst, and G.S. Kaltenecker. In press. Occupancy, productivity, turnover, and dis- persal of Northern Goshawks in portions of the north- eastern Great Basin. Stud. Avian Biol.\\n press. Bloxton, T.D., Jr. 2002- Prey abundance, space use, de- mography, and foraging habitat of Northern Gos- hawks in western Washington. M.S. thesis, Univ. Wash- ington, Seattle, WA U.S.A. Boat, C.W. 1994. A photographic and behavioral guide to aging nestling Northern Goshawks {Acdpitenr gentil- is). Stud. Avian Biol. 16:32-40. Doyle, F.I. and J.M.N. Smith. 1994. Population responses of Northern Goshawks to the 10-year cycle in num- bers of snowshoe hares. Stud. Avian Biol. 16:122—129. Freedman, D.A. 1983. A note on screening regression equations. Am. Statistician Gargett, V., E. Gargett, and D. Damania. 1995. The influence of rainfall on Black Eagle breeding over 31 years in the Matopo Hills, Zimbabwe. 66:114— 12L Hoglund, N.H. 1964. The hawk Accipiter gentilislAnn^ in Fennos Kandia (English translation). Viltrevy 2:195- 269. Holm, S. 1979. A simple sequentially rejective multiple test procedure. Scand. J. Statist. 6:65-70. INGRAI.DI, M.F. 1998. Population biology of Northern Goshawks in east-central Arizona. Tech. Rep. 133, Nongame and Endangered Wildlife Program, Arizona Game and Fish Department, Phoenix, AZ U.S.A. Kendleigh, S.C. 1970. Energy requirements for existence in relation to size of bird. Condor 72:60—65. Kostrzewa, a. and R. Kostrzewa. 1990. The relationship of spring and summer weather with density and breeding performance of the buzzard Buteo buteo, gos- hawk Accipiter gentilis and kestrel Falco tinnunculus. Ibis 132:550-559. AND . 1991. Winter weather, spring and summer density, and subsequent breeding success of Eurasian Kestrels, Common Buzzards, and Northern Goshawks. Auk 108:342-347. Loope, L.L. 1969. Subalpine and alpine vegetation of northeastern Nevada. Ph.D. dissertation, Duke Univ., Durham, NC U.S.A. Marcstrom, V. AND R. Kenward. 1981. Movements of wintering goshawks in Sweden. Swed. Wildl. Res. 12:1- 36. McGowan, J.D. 1975. Distribution, density, and produc- tivity of goshawks in interior Alaska: final report. Alas- ka Department of Fish and Game, Juneau, AK U.S.A. National Drought Mitigation Center, [online]. 2003. Drought monitor archive (http://drought.unl.edu/ dm/archive. html) . Natural Resources Conservation Service, [online]. 2003. Nevada/Utah Snotel Sites (http://www.wrcc. dri.edu/snotel/snogtb.html) . Newton, I. 1979. Population ecology of raptors. Buteo Books, Vermillion, SD U.S.A. . 1998. Population limitation in birds. Academic Press, New York, NY U.S.A. 236 Fairhurst and Bechard VoL. 39, No. 3 Patla, S.M. 1997. Nesting ecology and habitat of the Northern Goshawk in undisturbed and timber harvest areas on the Targhee National Forest, greater Yellow- stone ecosystem. M.S. thesis, Idaho State Univ., Po- catello, ID U.S.A. Penteriani, V. 1997. Long-term study of a goshawk breeding population on a Mediterranean mountain (Abruzzi Apennines, central Italy) : density, breeding performance and diet./. Raptor Res. 31:308-312. PosTUPALSKY, S. 1974. Raptor reproductive success: some problems with methods, criteria, and terminology. In F.N. Hamerstrom, Jr., B.E. Harrell, and R.R. Olen- dorff [Eds.], Management of raptors. Raptor Re- search Foundation, Vermillion, SD U.S.A. Reynolds, R.T. and S.M. Joy. 1998. Distribution, territory occupancy, dispersal, and demography of Northern Goshawks on the Kaibab Plateau, Arizona. Final Rep. Arizona Game and Fish. Heritage Project No. 194045. Phoenix, AZ U.S.A. Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- hawk (Accipiter gentilis ) . In A. Poole and F. Gill [Eds.] , The birds of North America, No. 298. The Academy of Natural Sciences, Philadelphia, PA and The Amer- ican Ornithologists’ Union, Washington, DC U.S.A. and L.F. Ruggiero. 1995. Winter movements of adult Northern Goshawks that nested in southcentral Wyoming./. Raptor Res. 29:5-9. Steenhof, K., M.N. Kochert, and T.L. McDonald. 1997. Interactive effects of prey and weather on Golden Ea- gle reproduction./ Anim. Ecol. 66:350—362. Stokes, M.E., C.S. Davis, and G.G. Koch. 2000. Categor- ical data analysis using the SAS system. SAS Institute, Cary, NC U.S.A. Van Horne, B., G.S. Olson, R.L. Schooley, J.G. Corn, AND K.P. Burnham. 1997. Effects of drought and pro- longed winter on Townsend’s ground squirrel demog- raphy in shrubsteppe habitats. Ecol. Monogr. 67:295- 315. Woodbridge, B. and PJ. Detrich. 1994. Territory occu- pancy and habitat patch size of Northern Goshawks in the southern Cascades of California. Stud. Avian Biol. 16:83-87. Younk, J.V 1996. Breeding ecology of the Northern Gos- hawk in relation to surface gold mining in naturally- fragmented aspen forests of northern Nevada. M.S thesis, Boise State Univ., Boise, ID U.S.A. and M.J. Bechard. 1994. Breeding ecology of the Northern Goshawk in high-elevation aspen forests of northern Nevada. Stud. Avian Biol. 16:119-121. Zachel, C.R. 1985. Food habits, hunting activity, and post-fledging behavior of Northern Goshawks {Accip- iter gentilis) in interior Alaska. M.S. thesis, Univ. of Alaska Fairbanks, Fairbanks, AK U.S.A. Received 2 February 2004; accepted 9 March 2005 Associate Editor: Michael I. Goldstein /. Raptor Res. 39(3);237-246 © 2005 The Raptor Research Foundation, Inc. PATTERNS OF TEMPORAL VARIATION IN GOSPIAWK REPRODUCTION AND PREY RESOURCES Susan R. Salafsky^ and Richard T. Reynolds USDA Forest Service, Rocky Mountain Research Station, 2150 Centre Avenue, Building A, Suite 350, Fort Collins, CO 80526 U.S.A. Barry R. Noon Department of Fishery and Wildlife Biology, Colorado State University, Fort Collins, CO 80523 U.S.A. Abstract. — To investigate whether Northern Goshawk {Accipiter gentilis) reproduction is food-limited, we evaluated the reproductive output from 401 goshawk breeding opportunities on the Kaibab Plateau, Arizona during 1999-2002. Concurrently, we estimated densities of 10 goshawk prey species (seven birds, three mammals) using distance sampling. We then assessed the relationship between goshawk produc- tivity (number of fledglings produced) and prey density within and among years by relating the contri- bution of individual prey species and total prey density to goshawk productivity. We also estimated the proportion of total diet and biomass for each species that contributed &3% of all prey items. Total prey density was highly correlated with variation in goshawk productivity {F — 0.98, P = 0.012). Red squirrel ( Tamiasciurus hudsonicus) density explained more variation in goshawk productivity than any other spe- cies (r^ = 0.94, P — 0.031), but density could not be estimated for every predominant prey species in goshawk diets. However, only red squirrels had a positive and significant relationship to goshawk pro- ductivity in terms of frequency {F = 0.97, P — 0.014) and biomass = 0.95, P = 0.033). Northern Flickers {Colaptes auratus) and cottontail rabbits (Sylvilagus spp.), which contributed the greatest fre- quency and biomass, respectively, to goshawk diets, showed no relationship with goshawk productivity. Even though goshawks on the Kaibab Plateau have a diverse diet and will readily switch to alternate prey species, goshawk productivity showed significant interannual variation. Our results suggest that the magnitude of goshawk productivity was determined by total prey density and annual variation was driven by differences in the densities of critical prey species. Key Words; Northern Goshawk; Accipiter gentilis; diet, distance sampling, predator-prey dynamics; prey density; productivity. PATRONES DE VARIACION TEMPORAL DE LAS PRESAS Y DE LA REPRODUCCION DE ACCIPITER GENTILIS Resumen. — Para investigar si la reproduccion de Accipiter gentilis se encuentra limitada por la dispo- nibilidad de alimento, evaluamos el rendimiento reproductive de 401 oportunidades reproductivas de estos halcones en Kaibab Plateau, Arizona, entre 1999 y 2002. A1 mismo tiempo, estimamos las densi- dades de 10 especies de presas para los halcones (siete aves, tres mamiferos) utilizando el metodo de conteo con distancias variables. Luego determinamos la relacion entre la productividad de los halcones (numero de volantones producidos) y la densidad de presas dentro y entre anos, relacionando la con- tribucion de cada especie de presa y la densidad total de presas con la productividad de los halcones. Tambien estimamos la proporcion de la dieta total para cada especie que contribuyo mas del 3% de todas las presas en la dieta. La densidad total de las presas se correlaciono fuertemente con la variacion en la productividad de los halcones {F = 0.98, P = 0.012). La densidad de la ardilla Tamiasciurus hudsonicus explico la mayor parte de la variacion en la productividad de los halcones con relacion a las otras especies {F = 0.94, P = 0.031), pero no se pudo estimar la densidad de cada especie de presa predominante en la dieta de los halcones. Sin embargo, solo la ardilla T. hudsonicus presento una relacion positiva y significativa con la productividad de los halcones en terminos de frecuencia {F = 0.97, P = 0.014) y biomasa {F = 0.95, P — 0.033). Las aves del genero Colaptes y los conejos, que contribuyeron la mayor frecuencia y biomasa de la dieta de los halcones, respectivamente, no se cor- ^ Corresponding author’s email address: salafsky@lamar.colostate.edu 237 238 Salafsky et al. VoL. 39, No. 3 relacionaron con la productividad de los halcones. A pesar de que los halcones en Kaibab Plateau tienen una dieta diversa y pueden cambiar facilmente a especies de presas alternativas, su productividad mostro una variacion interanual significativa. Nuestros resultados sugieren que la magnitud de la productividad de los halcones fue determinada por la densidad total de presas y que la variacion anual fue producida por las diferencias en la densidad de especies de presas criticas. [Traduccion del equipo editorial] To understand temporal variation in population size, it is necessary to focus on the factors that limit demographic processes, such as reproduction and survival. Ultimately, the availability of essential re- sources within a habitat regulates population growth. Resource availability, specifically food, is hypothesized to be an important limiting factor of many raptor populations (Newton 1979). Varia- tions in food supply often result in extensive fluc- tuations in population demographic parameters (Gotelli 1998, Newton 1998), but the mechanisms of food-limitation are difficult to quantify, espe- cially in complex systems. Consequently, most in- formation on the influence of food resources on population dynamics comes from correlations be- tween reproduction and food abundance (Martin 1987). The magnitude of the effects of food-limitation on reproduction is poorly understood, especially for predators with broad diets, such as Northern Goshawks {Accipiter gentilis). Goshawks regularly consume a variety of prey including ground and tree squirrels, rabbits, medium to large passerines, woodpeckers, and gallinaceous birds (Squires and Reynolds 1997). The diversity of prey in their diets ultimately depends on the abundance and avail- ability of the local bird and mammal fauna, which varies geographically. In Canada, although gos- hawks regularly consumed several prey species (>5) , goshawk reproduction showed a strong func- tional response to only one species — snowshoe hare {Lepus amencanus; Doyle and Smith 2001). In contrast, 14 species of birds and mammals regular- ly contributed to goshawk diets in the southwestern United States (Reynolds et al. 1992). This diet di- versity may stabilize their breeding rates. When prey populations vary asynchronously, the ability of goshawks to switch between alternative prey species may result in less annual variation in reproduction than in areas where goshawks rely primarily on cy- clic populations of a single prey species (Newton 1979). Our objectives were to: (1) determine if prey re- sources limit the reproductive rates of goshawks with relatively diverse diets and (2) describe how changes in prey populations may influence gos- hawk productivity (number of fledglings pro- duced) . If food is a limiting factor of goshawk pro- ductivity, then variation in the number of fledglings produced should be associated with fluc- tuations in prey resources. However, if there is a difference in the contribution of individual prey species, then goshawk productivity should respond to fluctuations in the densities of individual prey species. Finally, if the densities of important prey species vary in synchrony, then goshawk productiv- ity should exhibit greater temporal variation. To explore these relationships we studied goshawk productivity and prey resources on the Kaibab Pla- teau, Arizona during 1999-2002. Study Area The Kaibab Plateau is a large (95 X 55 km) forested island, surrounded by shrub-steppe desert, in northern Arizona. Steep slopes and escarpments form the eastern, southern, and western edges of the Kaibab Plateau and create a distinct boundary between the shrub-steppe des- ert at 1750 m elevation above sea level and the plateau (maximum elevation 2800 m). The northern edge of the plateau gradually descends to sagebrush desert, forming an indistinct boundary between the two landforms. The study area (1285 km^) on the Kaibab Plateau in- cluded forests above 2182 m elevation on the North Kai- bab Ranger District of the Kaibab National Forest. Four forest types dominated the study area: Pinyon-juniper {Fi- nns edulis-Juniperus spp.) woodlands occupied 106 km^ at lower elevations, ponderosa pine {Finns ponderosa) forests occupied 714 km^ at mid-elevation zones, mixed conifer {Abies concolor, Finns ponderosa, Fseudotsuga menziesii, Ficea engelmannii) forests occupied 275 km^ at the highest el- evations, and quaking aspen {Fopulns tremuloides) forests occupied 112 km^ interspersed among the other forest types (Joy 2002). Methods Goshawk Productivity. We estimated annual goshawk productivity per territory in 1999-2002. A territory was defined as the area (approximately 1 1 km^) defended by a pair of goshawks during the breeding season (Reynolds et al. 2005). Because goshawks may use more than one nest within a territory among breeding years (Reynolds et al. 2005), all nest structures were visited annually in spring to determine the territory occupancy status. If an active nest (nest containing eggs or young) was not lo- cated within an existing territory, we conducted system- atic surveys until we found an active nest or thoroughly September 2005 Biology 239 searched the entire territory, which required a minimum effort of 10 person-days (Reynolds et al. 2004) . Each year, we also conducted surveys throughout the study area to locate territories not detected in previous years (Reyn- olds and Joy 2005). To determine nest status and fledg- ling production, all active nests were visited weekly throughout the breeding season. Goshawk offspring were counted in the nest 7-10 d prior to fledging or from the ground after fledging. Goshawk productivity was estimat- ed annually as the mean number of fledglings produced per territory under study. Prey Density. To obtain estimates of prey density, we conducted distance sampling (Buckland et al. 1993) along line transects from 1999-2002. Sixty 500-m tran- sects were placed randomly throughout the study area within two strata defined by the forest types (mixed co- nifer, ponderosa pine) that occupied most of the study area. We established 30 transects per stratum and char- acterized each transect by its elevation, tree species com- position, and tree density. Within a given year, we sam- pled each transect during three time periods that corresponded with specific goshawk reproductive stages: spring (28 May-24 June)-incubation/hatching stage, summer (25 June-22 July) -nestling stage, and late sum- mer (23 July-14 August) -fledging stage. To reduce travel time between transects and to increase sampling efficien- cy, transects were grouped by location. Transects were sampled in groups of four per day, and the sampling or- der of groups was determined using a random number table. Daily sampling began 0.5 hr after sunrise and was completed within 3 hr. All transects were sampled by one observer (Salafsky) during the 4 yr of the study. Sampling was not conducted during inclement weather (rain, winds >20 kph) due to reduced probability of prey de- tection. Prey seen or heard during sampling were iden- tified to species, and the perpendicular distance from the detected animal to the transect line was measured with a laser rangefinder (accurate to ±1 m). Data were collect- ed on 15 prey species common in goshawk diets on the Kaibab Plateau (S. Salafsky unpubl. data) and considered important components of goshawk diets in the south- western United States (Reynolds et al. 1992) . Goshawk Diet. The species composition of goshawk di- ets was determined from prey remains (pelage, plumage, skeletal parts) that were collected from active goshawk nest sites during weekly visits throughout the breeding season. Prey remains were pooled by territory and date collected, identified to species, and paired to assess the minimum number of individuals consumed (Reynolds and Meslow 1984) . The biomass contribution of individ- ual prey was based on the published mass of each avian (Dunning 1993) and mammal (Hoffmeister 1986) spe- cies. All methods for quantifying raptor diets have inher- ent biases (Marti 1987). However, Kennedy (1991) re- ported that estimates of prey use were similar for prey remain, pellet, and direct observation methods of diet analysis for goshawks in New Mexico. Data Analysis. We based goshawk productivity on the number of fledglings produced per territory under study. We classified territories based on ^1 attempt to breed on the territory, the identity of the adult birds, and the av- erage inter-territory distance (Reynolds et al. 2005). A high density of territories, a tendency of individuals to retain the same territory for life, and a delayed age at first breeding (Wiens and Reynolds 2005) suggests that the breeding habitat on the Kaibab Plateau was saturated. This evidence combined with the observed patterns of territory occupancy for individual adults over a 14-yr pe- riod (S. Salafsky unpubl. data) indicated that goshawks occupied the territories, even when we found little evi- dence of birds present. By including all territories rather than only those that were confirmed “active” or “occu- pied,” we accounted for all potential breeding opportu- nities and the full range of variability in the reproductive quality of territories. Variable distance sampling data were analyzed with program DISTANCE, Version 3.5 (Thomas et al. 1998) Reliable estimates of density from distance sampling de- pend on several critical assumptions: all individuals on the transect line were detected, all individuals were de- tected at their initial location, and all distances were mea- sured accurately (Buckland et al. 1993). Data collection methods were designed to meet these assumptions. Be- cause variable distance sampling uses a detection func- tion that compensates for differences in detection prob- abilities among species, habitats, and distances from transects (Emlen 1971, Buckland et al. 1993), density es- timates based on distance data are not confounded by factors affecting detectability and thus are representative of the true population size. Prey densities were estimated separately for mixed conifer and ponderosa pine to ac- count for differences in detection probabilities among forest types. These estimates were then multiplied by the proportion of each forest type within the study area and added together to calculate prey densities for the entire study area. Annual density estimates were computed only for species with sufficient sample sizes. Total prey density was calculated as the sum of the individual prey densities for species with a sufficient number of detections. We stratified total prey density by sampling period within each year to estimate prey densities associated with gos- hawk breeding phenology. We used the Tukey-Kramer adjustment for multiple comparisons of means to test for differences in goshawk productivity among years (PROG GLM, SAS Institute 1999). Z-statistics were used to test for differences in mean prey densities among years and sampling periods (Buckland et al. 1993). To control for Type I error, we only tested for differences in density between specific pairwise comparisons (e.g., years of highest and lowest density) . To assess the relationship between goshawk pro- ductivity and prey density, we used linear regression (PROG REG, SAS Institute 1999), where annual goshawk productivity was the dependent variable, and estimates for individual prey species and summed over prey species were used as explanatory variables. Linear regression was also used to assess the relationship between goshawk pro- ductivity and prey species in the diet. In these regres- sions, annual goshawk productivity was the dependent variable and percent of total diet or biomass contribution for individual prey species were assessed as explanatory variables. We used an information-theoretic approach (Burnham and Anderson 2002) to identify the prey var- iables that explained the most annual variation in gos- hawk productivity per territory. A priori candidate models were developed to represent the potential effects of prey 240 Salafsky et al. VoL. 39, No. 3 1.5 n 1999 2000 2001 2002 Year Figure 1. Mean number of Northern Goshawk fledglings produced per territory (±SE) on the Kaibab Plateau, Arizona, 1999-2002. density on goshawk productivity. We hypothesized that goshawk productivity would be most strongly related to prey densities that contributed the most to goshawk re- production. Competing models were ranked by their ad- equacy in explaining the variation in goshawk productiv- ity using Akaike Information Criterion (PROC MIXED, SAS Institute 1999). To compare the relative importance of each prey species, we also used cumulative Akaike weights, which were calculated by summing the weights across all models that included the variable of interest (Burnham and Anderson 2002) . Results Variation in Goshawk Productivity. The number of goshawk territories used to estimate productivity was 97 in 1999, 98 in 2000, and 103 in 2001 and 2002 {N = 401). The proportion of territories with active nests was 54% in 1999, 58% in 2000, 28% in 2001, and 18% in 2002. Goshawk productivity {x ± SE) varied among years = 26.78, P< 0.001) and ranged from 0.14 ± 0.04 fledglings produced per territory in 2002 to 1.23 ± 0.14 fledglings pro- duced per territory in 2000 (Fig. 1). There was a significant decline = 37.15, P < 0.001) in goshawk productivity between 2000 and 2001 (Fig. 1 ). Variation in Prey Density. Ten prey species had sufficient detections to estimate density: American Robin ( Turdus migratorius) , Clark’s Nutcracker (Nu- dfraga Columbiana), Downy Woodpecker (Picoides pubescens), golden-man tied ground squirrel (Sper- mophilus lateralis). Hairy Woodpecker {Picoides vil- losus), Kaibab squirrel {Sciurus aberti kaibabensis) , Northern Flicker {Colaptes auratus), red squirrel {Tamiasdurus hudsonicus), Steller’s Jay {Cyanodtta stelleri), and Williamson’s Sapsucker (Sphyrapicus thyroideus). We were unable to estimate densities for black-tailed jackrabbit {Lepus califomicus) , Blue Grouse {Dendragapus obscurus), chipmunk {Euta- mias spp.), cottontail rabbit (Sylvilagus spp.), and rock squirrel (Spermophilus variegatus) due to low numbers of detections. Detection probability plots showed little evidence of heaping, measurement errors, and evasive movement prior to detection. Total prey density (±SE) varied annually and ranged from 2.22 ± 0.08 individuals ha“^ in 2001 to 3.96 ± 0.14 individuals ha“^ in 2000 (z = 10.39, P < 0.001). Density also varied significantly among years for most individual prey species (Table 1) in- cluding golden-mantled ground squirrel (z = 2.18, P — 0.015), Hairy Woodpecker (z = —2.88, P = 0.002), Kaibab squirrel (z = 2.47, P = 0.007), Northern Flicker (z = 5.70, P < 0.001), red squir- rel (z = 8.32, P < 0.001), Steller’s Jay (z = 3.25, P < 0.001), and Williamson’s Sapsucker (z = —2.78, P = 0.003). Significant declines in prey densities were also observed between 2000 and 2001 for golden-mantled ground squirrel (z = 2.18, P = 0.015), Kaibab squirrel (z = 2.47, P = 0.007), Northern Flicker (z = 2.62, P = 0.005), and red squirrel (z = 8.32, P< 0.001), but only red squirrel September 2005 Biology 241 Table 1. Annual estimates of Northern Goshawk prey density ha~^ for American Robin (AMRO), Clark’s Nutcracker (CLNU), Downy Woodpecker (DOWO), golden-mantled ground squirrel (GMSQ), Hairy Woodpecker (HAWO), Kaibab squirrel (KASQ), Northern Flicker (NOFL), red squirrel (RESQ), Steller’sjay (STJA), Williamson’s Sapsucker (WISA), and all 10 prey species’ densities combined (Total) on the Kaibab Plateau, Arizona, 1999-2002. Species 1999 2000 2001 2002 X SE X SE X SE X SE AMRO 0.23 0.05 0.22 0.07 0.27 0.06 0.25 0.06 CLNU 0.05 0.02 0.05 0.02 0.04 0.01 0.11 0.04 DOWO 0.20 0.05 0.11 0.03 0.20 0.06 0.11 0.03 GMSQ 0.28 0.30 0.64 0.18 0.22 0.06 0.32 0.10 HAWO 0.09 0.03 0.05 0.02 0.23 0.06 0.17 0.04 KASQ 0.11 0.04 0.26 0.07 0.07 0.03 0.08 0.02 NOFL 0.58 0.08 0.77 0.09 0.48 0.06 0.20 0.04 RESQ 1.16 0.17 1.38 0.15 0.12 0.04 0.23 0.05 STJA 0.41 0.07 0.12 0.05 0.30 0.07 0.33 0.05 WISA 0.18 0.04 0.36 0.08 0.28 0.06 0.45 0.09 Total 3.29 0.19 3.96 0.14 2.22 0.08 2.24 0.09 density decreased by an order of magnitude (Table 1 ). Prey density also varied by sampling period (Fig. 2). However, there were too few observations to accurately estimate density by sampling period for most individual prey species, so we report only to- tal prey density by sampling period. Total prey den- sity in the spring sampling period was highest in 2000, followed by 1999, 2002, and 2001 (Fig. 2). However, the decrease in density was only statisti- cally significant between 1999 and 2002 (z = 1.74, P = 0.041), and 2002 and 2001 (z = 6.58, P = 0.005). The only significant decrease in total prey density between the late-summer sampling period of one year and the spring sampling period of the next occurred between 2000 and 2001 (z = 6.58, P< 0.001; Fig. 2). Goshawk Diets. Goshawks on the Kaibab Plateau captured and consumed a wide diversity of prey. A total of 710 individual prey items consisting of 30 Figure 2. Total Northern Goshawk prey density estimates ha~^ (±SE) by sampling period on the Kaibab Plateau, Arizona, 1999-2002. 242 Salafsky et al. VoL. 39, No. 3 Table 2. Prey species each contributing ^3% of all items (N = 710) to Northern Goshawk diets in terms of percent frequency and biomass (kg) , and their relationship to the number of fledglings produced per goshawk territory on the Kaibab Plateau, Arizona, during 1999—2002. Species No. Percent Frequency P Percent Biomass P Black-tailed jackrabbit 23 3 -0.32 0.43 24 -0.36 0.40 Clark’s Nutcracker 34 5 0.03 0.82 2 0.01 0.88 Cottontail rabbit 125 18 -0.13 0.64 42 -0.03 0.84 Kaibab squirrel 40 6 -0.05 0.78 13 0.02 0.87 Northern Flicker 141 20 0.87 0.07 8 0.87 0.07 Red squirrel 87 12 0.97 0.01 7 0.95 0.02 Steller’s Jay 88 12 -0.09 0.70 4 0.04 0.81 species were collected from nest areas during 1999-2002. Seven species each contributed ^3% of all prey items collected in terms of percent fre- quency (Table 2) . In descending order of percent of total diet, the most common prey items were Northern Flickers, cottontail rabbits, red squirrels, Steller’s Jays, Kaibab squirrels, Clark’s Nutcrackers, and black-tailed jackrabbits. The descending order of species biomass contribution to goshawk diets was: cottontail rabbits, black-tailed jackrabbits, Kai- bab squirrels, Northern Flickers, red squirrels, Steller’s Jays, and Clark’s Nutcrackers (Table 2). The mean number of prey items per fledgling was 1.8 in 1999, 2.3 in 2000, 5.1 in 2001, and 7.6 in 2002. In contrast the mean biomass of prey items per fledgling was 0.8 kg in 1999, 0.6 kg in 2000, 2.3 kg in 2001, and 2.8 kg in 2002. Goshawk Productivity and Prey Resources. We found a strong positive relationship (r^ = 0.98, P = 0.012) between total prey density and goshawk productivity from 1999-2002 (Fig. 3) . Although an- nual goshawk productivity was highly correlated with prey density in the spring sampling period (r^ = 0.70, P = 0.163), summer sampling period (r^ = 0.75, P = 0.131), and late-summer sampling pe- riod (r^ = 0.79, P = 0.112), annual prey density accounted for more of the variation in goshawk productivity. Based on regression models for each prey species, only red squirrel density had a signif- icant and positive relationship to goshawk produc- Figure 3. The relationship between total prey density ha ^ and the mean number of Northern Goshawk fledglings produced per territory on the Kaibab Plateau, Arizona, 1999-2002. September 2005 Biology 243 Table 3. Top 10 models for mean number of Northern Goshawk fledglings produced per territory on the Kaibab Plateau, Arizona, 1999-2002. Models are ranked based on Akaike’s Information Criteria (AIC) and include model covariates, number of parameters (K) , AIC differences (AAIC) and Akaike weights (wj) . Model AIC K AAIC Wi Total prey species 1134.80 3 0.00 0.70 Red squirrel 1137.70 3 2.90 0.16 Mammal prey species 1138.10 3 3.30 0.13 Northern Flicker 1144.20 3 9.40 0.01 Total prey in late-summer 1149.50 3 14.70 0.00 Hairy Woodpecker 1150.80 3 16.00 0.00 Total prey in summer 1152.20 3 17.40 0.00 Kaibab squirrel 1154.20 3 19.40 0.00 American Robin 1155.40 3 20.60 0.00 Total prey in spring 1156.90 3 22.10 0.00 tivity (r^ = 0.94, P = 0.031). Red squirrel was also the only species that had a significant and positive relationship to goshawk productivity for percent of diet {P- = 0.97, P — 0.014) and biomass (r^ = 0.95, P = 0.024; Table 2). The densities of mammal prey species {P = 0.94, P — 0.033) explained more of the variation in goshawk productivity than avian prey species < 0.01, P = 0.949). Our model selection results showed that total prey density was clearly the top model (Table 3) . This model, which included an annual summation of all prey species’ densities, received >70% of the Akaike weight across the model set (Table 3) and was more than four times as likely as the next best model. The only single species models with some weight of evidence included those for red squirrel and Northern Flicker (Table 3). However, the red squirrel density covariate had a higher cumulative Akaike weight (99%) than Northern Flicker (71%). All other models based on individual prey species, avian density, and models of total prey den- sity by sampling period had minimal support and failed to explain variation in goshawk productivity (Table 3). When we compared only the models with total prey density by sampling period in a sep- arate analysis, total prey density summed over all sampling periods was selected as the best model (AIC = 811.10, K = 3, AAIC = 0.00, Wj = 0.93). All other models, including the model with the dif- ference in prey density between late-summer and the successive spring (AIC = 873.00, K = 3, AAIC = 61.90, Wj = 0.00) and the lowest ranked model with late-summer prey density from the prior year (AIC = 883.60, K = 3, AAIC = 72.50, Wi = 0.00), were not supported by the data. Discussion A short-term observational study cannot provide a strong basis for estimating the causal relationship between prey resources and annual goshawk pro- ductivity. Thus, our study only established a strong association between variation in prey resources within the study area and goshawk productivity. Be- cause fluctuations in other limiting factors (e.g., climate) may have coincided with changes in prey resources, we cannot identify the factors ultimately responsible for variation in goshawk productivity. However, if the patterns we observed between prey resources and goshawk productivity were support- ed by experimental studies that established a rela- tionship between food-supply and goshawk repro- duction, then it would be reasonable to infer that prey resources may be an important limiting factor of goshawk reproduction on the Kaibab Plateau. During 1999-2002 we observed high temporal correlations between goshawk productivity and an- nual prey density; changes in goshawk productivity paralleled changes in prey density. Total prey den- sity, in addition to the proportion of active gos- hawk nests and mean number of fledglings pro- duced, was high in 1999 and 2000 and low in 2001 and 2002. Therefore, it appears that goshawk re- production on the Kaibab Plateau responded to inter-annual increases in prey density. Several oth- er studies have also found close ties between mea- sures of goshawk reproduction and the relative abundances of prey (Huhtala and Sulkava 1981, Doyle and Smith 1994, Keane 1999). Further, gos- hawk studies that experimentally manipulated food-supply found supplemental food may have in- 244 Salafsky et al. VoL. 39, No. 3 fluenced goshawk productivity by increasing nest- ling survival when background prey-levels were low (Ward and Kennedy 1996, Dewey and Kennedy 2001) . Thus, we suggest that the number of gos- hawk fledglings produced may be influenced by fluctuations in prey density. On an annual basis, the reproductive responses of goshawks depend on the abundance of prey dur- ing critical time periods. Low food resources may manifest through failure to lay eggs, smaller clutch- es, and reduced survival of young (Newton 1998). The abundance of prey may be an important de- terminant of the “decision” to breed. Goshawks initiate breeding before most prey species repro- duce, so the density of prey during the incubation period is likely similar to prey levels prior to egg- laying. On the Kaibab Plateau, prey densities dur- ing the incubation stage were similar in 1999 and 2002, yet goshawk productivity was six times higher in 1999. In addition, although there was a signifi- cant increase in prey density during the incubation period between 2001 and 2002, goshawk produc- tivity changed little between these years, suggesting that below a density of ca. 0.8 prey ha“^, fewer fledglings are produced. However, prey density lev- els prior to egg-laying may alter the threshold ef- fects of prey density on goshawk productivity through physiological constraints. Assuming our density estimates represented true population size, the difference in prey density between late-summer and the next spring should reflect prey density lev- els prior to egg-laying. The large decline we ob- served in prey density between August 2000 and May 2001 indicated that there was substantial over- winter mortality for prey species. The lower prey numbers prior to egg-laying may have affected the ability of females to accumulate sufficient reserves to produce eggs in 2001. Our results suggest there is a difference in the contribution of individual prey species to goshawk reproduction. Red squirrel density and their per- cent frequency and biomass contribution to gos- hawk diet accounted for more variation in goshawk productivity than any other species. Although rab- bits contributed the majority of biomass to gos- hawk diets (>66%), goshawk reproduction was lower in most years when rabbits contributed the greatest proportion of biomass to the diets. Fur- ther, in “poor” goshawk reproductive years (2001, 2002) , the number of prey items and total biomass per fledgling was twice as high as in “good” repro- ductive years (1999, 2000). The difference in the apparent influence of individual prey species is likely a result of encounter rates with goshawks. Goshawks are opportunists and will presumably at- tempt to capture whatever prey species are readily available. However, the limited distributions or dif- ferent activity patterns of some prey species de- creases the probability that diurnal goshawks will encounter them while foraging. Jackrabbits are less common in upper elevation forests, and although cottontails are widely distributed across the study area, they are crepuscular (Hoffmeister 1986). In contrast, red squirrels are among the heaviest of the diurnal prey species, with a wide distribution across the study area (Salafsky 2004) . Red squirrels do not hibernate, which likely increases their im- portance to goshawks, particularly prior to egg-lay- ing. However, the importance of other prey species may vary with the spatial distribution of goshawk territories relative to the spatial distribution of prey habitats. For example, goshawks with territories lo- cated primarily within lower elevation forests may rely more heavily upon jackrabbits. In our study, goshawk productivity on the Kaibab Plateau was more closely associated with variation in mammal density than in avian density. Goshawks may consume more mammals than birds in some areas due to the availability and sizes of local prey species (Zachel 1985, Widen 1987, Doyle and Smith 1994). Similar to our study, Boal and Man- nan (1994) and Reynolds et al. (1994) found that goshawks on the Kaibab Plateau consumed a high- er proportion of mammalian prey. Other goshawk studies conducted in northern latitudes identified a strong link between goshawk reproductive rates and cyclical variation in hare abundance (Mc- Gowan 1975, Doyle and Smith 2001). Because an- nual variations in predator reproductive rates are greatest among species with limited diets that are dominated by cyclic prey (Newton 1979), goshawks on the Kaibab Plateau may be subject to more marked variations in productivity due to their re- liance on prey species with fluctuating densities. Goshawks have the ability to switch to alternate prey when the densities of essential prey species are reduced (Doyle and Smith 1994). However, if different prey species’ populations decline simul- taneously, then the opportunities for goshawks to switch to alternative prey species are limited. The densities of golden-mantled ground squirrels, Kai- bab squirrels, Northern Flickers, and red squirrels declined significantly between 2000 and 2001. Fur- ther, these species contributed >39% of all prey September 2005 Biology 245 items to goshawk diets. Parallel fluctuations in an- nual densities of important prey species may result in potentially “poor” and “good” years of prey re- sources. Thus, it may be that the collective density of the entire prey community influences the mag- nitude of variation in goshawk productivity on the Kaibab Plateau. In summary, our results indicate that prey den- sity is an important limiting factor of goshawk pro- ductivity. Although the temporal correlations be- tween goshawk productivity and prey resources were consistent over time, other factors may have varied with prey density and limited goshawk re- production in our study. Synchronous declines in prey species’ densities suggests that landscape-level factors acting at broad spatial scales, such as cli- mate, may interact with prey abundance to limit goshawk productivity. Because unfavorable weather conditions may have a greater effect on goshawk productivity when prey resources are already low, it is important to study the relationship between goshawk productivity and prey density over long time periods and variable environmental condi- tions. Acknowledgments Many thanks go to the goshawk technicians that col- lected data for this project. We also thank Clint W. Boal, Curtis H. Flather, John J. Keane, Rudy M. King, Marc A. Snyder, J. David Wiens, and three anonymous referees for reviewing this manuscript. This project was made possible with funding provided by both the Rocky Mountain Re- search Station and the Southwestern Region of the U.S. Department of Agriculture Forest Service. Literature Cited Boal, C.W. and R.W. Mannan. 1994. Northern Goshawk diets in ponderosa pine forests on the Kaibab Plateau. Stud. Avian Biol. 16:97-102. Buckland, S.T., D.R. Anderson, K.P. Burnham, and J.L. Laake. 1993. Distance sampling: estimating abun- dance of biological populations. Chapman & Hall, London, U.K. Burnham, K.P. and D.R, Anderson. 2002. Model selec- tion and inference: a practical information-theoretic approach, 2nd Ed. Springer-Verlag, New York, NY U.S.A. Dewey, S.R. and P.L. Kennedy. 2001. Effects of supple- mental food on parental-care strategies and juvenile survival of Northern Goshawks. Auk 118:352-365. Doyle, F.l. and J.M.N. Smith. 1994. Population responses of Northern Goshawks to the 10-year cycle in num- bers of snowshoe hares. Stud. Avian Biol. 16:122-129. AND J.M.N. Smith. 2001. Raptors and scavengers. Pages 378—404 in C.J. Krebs, S. Boutin, and R. Boon- stra (Eds.), Ecosystem dynamics of the boreal forest Oxford University Press, New York, NYU.S.A. Dunning, J.B., Jr. 1993. GRC handbook of avian body masses. CRC Press, Inc., Boca Raton, FL U.S.A. Emlen, J.T. 1971. Population densities of birds derived from transect counts. Auk 88:332-334. Gotelli, N.J. 1998. A primer of ecology. Second edition. Sinauer Associates, Inc., Sunderland, MA U.S.A. Hoffmeister, D.F. 1986. Mammals of Arizona. The Uni- versity of Arizona Press, Tucson, AZ U.S.A. Huhtala, K. and S. Sulkava. 1981. Environmental influ- ences on goshawk breeding in Finland. Pages 89-104 in R.E. Kenward and I.M. Lindsay (Eds.), Understand- ing the goshawk. The International Association for Falconry and Conservation of Birds of Prey, Oxford, England. Joy, S.M. 2002. Northern Goshawk habitat on the Kaibab National Forest in Arizona: factors affecting nest lo- cations and territory quality. Ph.D. dissertation, Col- orado State Univ., Fort Collins, CO U.S.A. Keane, JJ- 1999. Ecology of Northern Goshawk in the Sierra Nevada, California. Ph.D. dissertation, Univ. of California, Davis, CA U.S.A. Kennedy, P.L. 1991. Reproductive strategies of Northern Goshawks and Cooper’s Hawks in north-central New Mexico. Ph.D. dissertation, Utah State Univ., Logan, UT U.S.A. Marti, C.D. 1987. Raptor food habitat studies. Pages 67- 80 in B.G. Pendleton, B.A. Millsap, K.W. Cline, and D.M. Bird, [Eds.], Raptor management techniques manual. National Wildlife Federation, Sci. Tech. Se- ries No. 10, Washington, DC U.S.A. Martin, T.E. 1987. Food as a limit on breeding birds: a life-history perspective. Ann. Rev. Ecol. Syst. 18:453- 487. McGowan, J.D. 1975. Distribution, density and produc- tivity of goshawks in interior Alaska. Final Report to Federal Aid in Wildlife Restoration, Project Report W17-3, 4, 5, 6, Alaska Department of Fish and Game, Juneau, AK U.S. A. Newton, I. 1979. Population ecology of raptors. Buteo Books, Vermillion, SD U.S. A. , 1998. Population limitation in birds. Academic Press, Inc., San Diego, CA U.S.A. Reynolds, R.T., R.T. Graham, M.H. Reiser, R.L. Bassett, P.L. Kennedy, D.A. Boyce, Jr., G. Goodwin, R. Smith, AND E.L. Fisher. 1992. Management recommenda- tions for the Northern Goshawk in the southwestern United States. Gen. Tech. Report RM-217. USDA For- est Service, Rocky Mountain Research Station, Fort Gollins, CO U.S.A. AND S.M. Joy. 2005. Demography of Northern Goshawks in northern Arizona, 1991-1996. Stud. Avi- an Biol.', in press. , S.M. Joy, and D.G. Leslie. 1994. Nest productiv- ity, fidelity, and spacing of Northern Goshawks m northern Arizona. Stud. Avian Biol. 16:106-113. 246 Salafsky et al. VoL. 39, No. 3 AND E.C. Meslow. 1984. Partitioning of food and niche characteristics of coexisting Accipiter during breeding. Auk 101:761-779. , G.C. White, S.M. Joy, and R.W. Mannen. 2004. Effects of radiotransmitters on Northern Goshawks: do tailmounts lower survival of breeding males? J. Wildl. Manage. 68:25—32. , J.D. Wiens, S.M. Joy, and S.R. Salafsky. 2005. Sampling considerations for demographic and habitat studies of Northern Goshawks. J. Raptor Res. 39:274— 285. , J.D. Wiens, and S.R. Salafsky. 2005. A review and evaluation of factors limiting Northern Goshawk pop- ulations. Stud. Avian Biol.: in press. Salafsky, S.R. 2004. Covariation between prey abundance and Northern Goshawk fecundity on the Kaibab Pla- teau, Arizona. M.S. thesis, Colorado State Univ., Fort Collins, CO U.S.A. SAS Institute. 1999. SAS/STAT User’s Guide, Version 8. SAS Institute, Inc., Cary, NC U.S.A. Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- hawk {Accipiter gentilis) . Pages 1-32 in A. Poole and F. Gill [Eds.], The Birds of North America, No. 298. The Academy of Natural Sciences, Philadelphia, PA U.S.A. Thomas, L., J.L. Laake, J.F. Derry, S.T. Buckland, D.L. Borchers, D.R. Anderson, K.P. Burnham, S. Strind- berg, S.L. Hedley, M.L. Burt, F. Marques, J.H. Pol- lard, AND R.M. Fewster. 1998. Distance 3.5. Research Unit for Wildlife Population Assessment, Univ. of St. Andrews, U.K. Ward, J.M. and P.L. Kennedy. 1996. Effects of supple- mental food on growth and survival of juvenile North- ern Goshawks. Auk 113:200—208. Widen, P. 1987. Goshawk predation during winter, spring, and summer in a boreal forest in central Sweden. Hol- arctic Ecol. 10:104-109. Wiens, J.D. and R.T. Reynolds. 2005. Is fledgling success a reliable index of fitness in Northern Goshawks? J. Raptor Res. 39:210—221. Zachel, C.R. 1985. Food habits, hunting activity, and post-fledgling behavior of Northern Goshawks {Accip- iter gentilis) in interior Alaska. M.S. thesis, Univ. Alas- ka, Fairbanks, AK U.S.A. Received 3 March 2004; accepted 15 May 2005 Associate Editor: Clint Boal J. Raptor Res. 39(3):247-252 © 2005 The Raptor Research Foundation, Inc. A SKEWED SEX RATIO IN NORTHERN GOSHAWKS: IS IT A SIGN OF A STRESSED POPULATION? Michael F. Ingraldi^ Arizona Game and Fish Department, Research Branch, 2221 W. Greenway Road, Phoenix, AZ 85023 US. A. Abstract. — I examined 6 yr (1993-98) of data on Northern Goshawk {Accipiter gentilis) reproductive performance in east-central Arizona. Forty-four breeding territories were monitored over the 6-yr period, yielding 109 nesting attempts and 141 fledglings produced from 76 successful broods. Among the 63 broods from which sex of nestlings could be determined by measurements, 29% fledged one young, 54% fledged two young, and 17% fledged three young. The mean sex ratio across years was 1.93 ± 0.70 (±95% Cl) males/females (annual range 1.1:1-3:1). When combining all fledglings of known sex (76 males and 43 females), a 1.77:1 male to female sex ratio was significantly different from 1:1 (P = 0.002). A male-biased fledgling sex ratio may be explained by one or more of the following hypotheses: (1) more males were produced due to nutritional stress resulting in minimization of investment in the larger sex (females) and (2) fewer females were produced because of differential mortality due to exposure to the elements during the nestling stage. I propose that environmental stress in the form of exposure to the elements during a critical life stage (e.g., rainfall during the nestling stage), combined with limited food availability, may be driving the skewed sex ratios observed in this local Northern Goshawk population. Key Words: Northern Goshawk; Accipiter gentilisr, nestling, sex ratio; sex allocation. iES EL COCIENTE DE SEXOS SESGADO DE ACCIPITER GENTILIS UNA SENAL DE UNA POBLA- CION ESTRESADA? Resumen. — En este estudio examine datos sobre el desempeno reproductivo de Accipiter gentilis reco- lectados a lo largo de seis ahos (1993-98) en el centro-este de Arizona. Un total de 44 territorios de nidificacion fueron monitoreados a traves de los seis anos, observando un total de 109 intentos de nidificacion y 141 volantones producidos en 76 nidadas exitosas. De las 63 nidadas en las que el sexo de los pichones pudo ser determinado mediante mediciones, el 29% produjeron un volanton, el 54% dos volantones y el 17% tres volantones. El cociente de sexos promedio a traves de los anos fue de 1.93 ± 0.70 (±95% IC) machos/hembras (rango anual 1.1:1-3:1). A1 combinar todos los volantones de sexo conocido (76 machos y 43 hembras), el cociente de machos a hembras resultante de 1.77:1 fue signi- ficativamente diferente de 1:1 (P = 0.002). Un cociente de sexos sesgado hacia los machos en los volantones podrfa explicarse por las siguientes hipotesis: (1) se produjeron mas machos como conse- cuencia de estres nutricional que llevo a minimizar la inversion en el sexo de mayor tamaho (hembras) y/o (2) se produjeron menos hembras como resultado de una mortalidad diferencial debida a la ex- posicion al ambiente durante la permanencia de los pichones en el nido. Propongo que el estres ambiental causado por la exposicion a las condiciones ambientales durante una etapa critica de la vida (e.g., la Iluvia durante la etapa de crianza en el nido), combinado con la disponibilidad limitada de alimento, podrian estar determinando el sesgo en el cociente de sexos observado en esta poblacion local de A. gentilis. [Traduccion del equipo editorial] Female Northern Goshawks {Accipiter gentilis) are approximately 20—30% heavier than males (New- ton 1979, Palmer 1988, Squires and Reynolds 1997). Larger offspring require more food to at- ^ Email address: mingraldi@cybertrails.com tain fledging age, and thus are more susceptible to food shortages (Glutton-Brock et al. 1985, Teather and Weatherhead 1988, Anderson et al. 1993). An underlying assumption of an adjustable sex ratio is that parents should bias the sex ratio of their off- spring toward the sex whose production will most increase their own fitness (Trivers and Willard 247 248 Ingraldi VoL. 39, No. 3 1973, Clark 1978, Charnov 1982, Glutton-Brock 1986). The sex ratio at fledging should be the in- verse of the ratio of the cost of producing each sex (Fisher 1930) . Consequently, under times of limit- ed prey resources the sex ratio should be biased toward the “cheaper sex” (in Northern Goshawks: the male; Howe 1977, Cronmiller and Thompson 1981, Teather and Weatherhead 1988). In sexually- dimorphic species, differential survival between the sexes is most likely due to differing nutritional requirements, with higher mortality incurred by the larger sex (Torres and Drummond 1997, Shel- don et al. 1998). However, this may be offset in Northern Goshawks by the tendency of the female nestlings to seize prey items more readily than males (Lee 1980). Sex allocation in sexually dimorphic raptors has provided mixed observations. Studies have shown prevalence toward male fledglings in Golden Ea- gles (Aquila chrysaetos; Edwards et al. 1988), Har- ris’s Hawks (Parabuteo unicinctus; Bednarz and Hay- den 1991), Montagu’s Harriers {Circus pygargus; Leroux and Bretagnolle 1996), and Cooper’s Hawks {Accipiter cooperii; Rosenfield et al. 1996). Other studies showed prevalence toward female fledglings in Northern Harriers {Circus cyaneus; Balfour and Cadbury 1979), Bald Eagles {Halieaee- tus leucocephalus; Bortolotti 1986), and Peregrine Falcons {Falco peregrinus; Olsen and Cockburn 1991). However, most studies on raptors have re- vealed equal sex ratios in offspring (Newton 1979). In some previous work on Northern Goshawks, the sex ratio for fledglings has been shown not to dif- fer from 1:1 (Newton 1979, Reynolds and Joy 1998). A male-biased fledgling sex ratio may be ex- plained by one or more of the following hypothe- ses: (1) more males were fledged due to nutritional stress resulting in minimization of investment in the larger sex (females) during the nestling stage (Trivers and Willard 1973, Torres and Drummond 1997) and (2) fewer females were produced be- cause of differential mortality due to exposure to the elements during the nestling stage with larger young exhibiting a higher mortality rate (Newton 1979:136-137). Hypothesis 1 above may be a direct product of adaptive selection, whereas hypothesis 2 may simply be a nonadaptive by-product of dif- ferential mortality. The purpose of this paper is to discuss these potential explanations for the sex ra- tio exhibited by Northern Goshawk fledglings in east-central Arizona. Study Area The Sitgreaves portion of the Apache-Sitgreaves Na- tional Forest is located on the Mogollon Plateau in east- central Arizona and encompasses approximately 330 300 ha. Elevation ranges from 1768-2417 m. To the south, the study area is bounded by the Mogollon Rim, a large escarpment extending east across central Arizona and into New Mexico. To the east, the study area is bounded by the Springerville Ranger District boundary on the Apache National Forest. A wide range of vegetation com- munities occurs within the study area (Brown 1994) . The Mogollon Rim edge is dominated by deep drainages with mixed-conifer communities of Douglas-fir {Pseudotsuga menziesii), white fir {Abies concolor), and ponderosa pine {Pinus ponderosa), with pockets of aspen {Populus tremu- hides). New Mexico locust {Robinia neomexicana) , and Gambel oak {Quercus gambelii). Ridgetops are generally dominated by ponderosa pine forest. To the north, as elevation decreases, a ponderosa pine/juniper-pinyon ecotone transitions to a juniper-pinyon woodland domi- nated by alligator juniper (Juniperus deppeana), Utah ju- niper (/. osteosperma) , and Rocky Mountain pinyon pine {P. edulis). As elevation decreases further, a plains grass- land community develops, dominated by blue grama {Bouteloua gracilis), sand dropseed {Sporobolus cryptan- drus) , and fourwing saltbush {Atriplex canescens ) . Methods I monitored demographic parameters of a Northern Goshawk population on the Apache-Sitgreaves National Forest from 1993-98 and specifically noted the sex of fledgling birds. I visited occupied nests periodically dur- ing the breeding season (ca. late April-early August) to monitor status and productivity. An occupied nest was one in which at least one egg was laid (usually inferred by observing a bird in incubation posture). I estimated the ages of nestlings using a photographic guide pro- duced by Boal (1994). Birds were deemed to have sur- vived to fledge when they were greater than 80% of fledg- ling age (31 d old; Steenhof 1987). When nestlings were between 30-40 d old, I banded them with U.S. Geological Survey aluminum leg bands and took the following stan- dard morphological measurements: tarsus dorsal-ventral (smallest front to back measurement of the tarsus to nearest 0.1 mm), tarsus lateral (smallest side to side mea- surement of the tarsus to nearest 0.1 mm), hallux (tip of talon to the beginning of the fleshy pad of the hind toe to nearest 0.1 mm), bill depth (perpendicular to the bill from the top in front of the cere to the bottom of the bill to nearest 0.1 mm), culmen (top portion of the bill in front of the cere to the tip of the bill to nearest 0.1 mm), mass (taken to nearest 10 g). I used a A-means cluster analysis (SPSS 1997) to determine if morpholog- ical measurements could adequately identify sexes of nestlings. Only birds from broods in which all the mea- surements were collected were included in the analysis. In 1999, I collected blood samples from 22 nestlings and had them genetically analyzed for sex identification (Avi- an Biotech International, Tallahassee, FL U.S.A.) using a polymerase-chain reaction to amplify CHD (chromo-he- licase-DNA binding) genes Z and W that are located on the avian sex chromosomes in birds. For detailed proce- dural information, see Griffiths et al. (1998), Fridolfsson September 2005 Biology 249 Table 1. Nestling sex-ratio and productivity of Northern Goshawks from territories monitored on the Apache-Sit- greaves National Forest, Arizona, (1993—98). Year No. Nests Monitored Nesting Occupancy Rate=^ No. Successful Broods No. Male Fledglings No. Female Fledglings No. Fledglings OF Unknown Sex’’ Sex Ratio‘s pd 1993 30 0.69 17 10 9 11 1.1 0.819 1994 33 0.33 8 9 3 4 3 0.083 1995 39 0.66 14 17 10 0 1.7 0.178 1996 42 0.52 16 16 8 4 2 0.102 1997 42 0.31 6 7 3 0 2.3 0.206 1998 44 0.43 15 17 10 3 1.7 0.178 Total 230 X = 0.49 76 76 43 22 1.7 0.003 “ The number of occupied nests (a nest where at least one egg was laid) /total number of nesting territories monitored. Fledglings not measured, therefore gender indeterminate. Male/female. Significance of Chi-square test for difference from a 1:1 sex ratio. and Ellegren (1999), and Avian Biotech International (2005) . I calculated the sex ratio as the number of male per female fledglings. I used a chi-square analysis to test if the sex ratio was significantly different from 1:1 (Zar 1984). I used Spearman’s correlation analysis to test for significance in relationships between demographic pa- rameters (e.g., sex ratio versus the number of territories with an occupied nest per the number of territories mon- itored). All statistical tests were deemed significant at P < 0.05, and all means were expressed ±95% Confidence Interval (Cl) . I compiled monthly summaries of total pre- cipitation from U.S. National Oceanic and Atmospheric Administration records collected at the Show Low, Ari- zona, municipal airport (elevation = 1950 m) located within the study area. Results Breeding territories were monitored over the 6- yr period (range = 30-44 per yr), yielding 109 nesting attempts and 141 fledglings produced from 76 successful broods (Table 1). Among the 63 broods from which gender could be determined, 29% fledged one young, 54% fledged two young, and 17% fledged three young (Table 2). Two rel- atively homogenous groups of nestlings were dis- cerned using a /5-means cluster analysis, and their morphological measurements showed minimal overlap (Table 3). The two groups are easily rec- ognized when the lateral tarsus and culmen length measurements are plotted (Fig. 1). I considered all members of the larger group as females. The mea- surements used above correctly classified to gender all 22 nestlings that were genetically analyzed in 1999. The mean sex ratio across years was 1.93 ± 0.70 (±95% Cl) males/female (range = 1. 1:1-3: 1). When combining all fledglings of known sex (76 males and 43 females), a 1.77:1 male to female sex ratio resulted that was significantly different from 1:1 (x^ — 9.15, df = 1, P = 0.002). Table 2. Observed brood size and sex ratio of Northern Goshawks fledged on the Apache-Sitgreaves National Forest, Arizona, (1993-98). Nesting Outcome Brood Size 1 2 3 All males 13 13 2 All females 5 7 0 1 male and 1 female — 14 — 2 males and 1 female — — 8 1 male and 2 females — — 1 Sex ratio (M/F) 13/5 = 2.6 40/28 = 1.43 23/10 = 2.3 P* 0.06 0.15 0.02 ^ Significance of Chi-square test for difference from a 1:1 sex ratio. 250 Ingraldi VoL. 39, No. 3 Table 3. Summary statistics of Northern Goshawk morphological measurements taken on the Apache-Sitgreaves National Forest, Arizona (1993-98), Group Measurement (x ± 2 SE, Range) a Fledgling Group 1 (Presumed Male) AT= 59 Fledgling Group 2 (Presumed Female) N= 34 Adult Females N= 25 Culmen (mm) 19.05 ± 0.27 21.36 ± 0.34 24.53 ± 0.34 16.2-21.7 18.5-23.4 23.0-26.3 Beak depth (mm) 14.63 ± 0.18 16.63 ± 0.21 17.93 ± 0.18 13.3-16.1 15.1-17.8 17.2-18.7 Tarsus D/V*’ (mm) 7.04 ± 0.13 8.38 ± 0.19 10.48 ± 0.36 6.0-8. 1 7.4-9.3 8.7-12.0 Tarsus (mm) 5.36 ± 0.07 6.50 ± 0.13 7.48 ± 0.15 4.8-6.0 5.9-7.4 6.9-S.5 Hallux (mm) 23.79 ± 0.36 27,02 ± 0.52 31.98 ± 0.39 19.6-27.0 24.1-31.1 30.2-33.8 Mass (g) 702 ± 14 894 ± 36 1026 ± 39 565-810 620-1085 845-1265 See text for detailed description of measurements, D/V = the smallest front to back measurement of the tarsus (dorsal/ventral). ' L = the smallest side to side measurement of the tarsus (lateral) . Discussion Hypothesis 1 — Nutritional Stress Would Mini- mize the Investment in Larger Sex. Meyers (1978) predicted that during periods of lower than aver- age resource abundance, offspring sex ratio in a population should shift toward the sex having the lower energy needs. In times of plentiful resources, adults should invest in the larger offspring, which benefit from greater size. I observed a negative re- lationship between the annual sex ratio and the 24.0 -1 'i 23.0 - E 22.0 - £ o> 21.0 - c V 20.0 - c 0) 19.0 - E 3 18.0 - u 17.0 - 16.0 - 4.0 • \, *X m ••• « ♦ ♦ * * t • ••• 5.0 6.0 7.0 Tarsus lateral (mm) 8.0 Figure 1. Plot of the culmen length and lateral tarsus measurements for Northern Goshawk nestlings mea- sured on the Apache-Sitgreaves National Forest, AZ (1993-98). The two groups (circles presumed to be males and the diamonds females) were distinguished us- ing a fe-means cluster analysis of all morphological mea- surements collected. nesting occupancy rate (Fig. 2A) . A low nesting oc- cupancy rate most likely reflects a lower than av- erage abundance of some resource such as prey availability. Conversely, in years with a high nesting occupancy rate, which may have reflected an above average resource availability, the number of female offspring was greater (Fig. 2B). Yet, for this hy- pothesis to be an adaptive selection strategy, I would expect to find a sex ratio skewed toward fe- males when the nesting occupancy was high. In- stead when the nesting activity rate was high the sex ratio approached 1:1. This result may lend cre- dence to the resource-shortage hypothesis being a nonadaptive by-product of differential mortality (Weatherhead and Teather 1991). Hypothesis 2 — Male Nestlings are Less Suscep- tible to Adverse Weather Conditions. Newton (1979) suggests that females would be more sus- ceptible to mortality from exposure than smaller males. When being brooded, smaller nestlings may be more sheltered from environmental elements than larger siblings; or during episodes of rain, larger drenched female nestlings may hold more water and require more time to dry. Thus, larger female nestlings would remain cooler longer and be more susceptible to hypothermia. In years when more rainfall occurred during the nestling period (May and June), I observed more male fledglings September 2005 Biology 251 3.5 n 3 - 2.5 - o CO 2 - DC X 1.5 - (D CO 1 J 0.2 0.4 0.6 0.8 Nesting Activity Rate Fecundity O O O O 3 Kd G) 00 1 1 1 1 1 • ♦ ♦ ♦ ♦ ♦ u n 0 “ 1 \ 1 2 0.4 0.6 0.8 Nesting Activity Rate Figure 2. (A) Sex ratio of fledgling Northern Goshawks plotted against the annual nesting occupancy rate (num- ber of occupied nests/total number of nests monitored) on the Apache-Sitgreaves National Forest, AZ, 1993-98 (r = —0.84, P = 0.04). (B) Fecundity rate (number of fe- male fledglings produced/ total number of breeding fe- males) plotted against the nesting occupancy rate of Northern Goshawks on the Apache-Sitgreaves National Forest, AZ, 1993-98 (r = 0.77, P = 0.07). produced (Fig. 3) . This phenomenon of induced mortality due to exposure may also be compound- ed by nutritional stress (i.e., shortage of prey). Wet weather conditions have been shown to prevent adult raptors from hunting efficiently (Hiraldo et al. 1990, Kostrzewa and Kostrzewa 1990). Times of food shortage may force the nesting adult female to leave the nest in search of prey, thereby increas- ing the risk of exposure to the nestlings. For ex- ample, Boal et al. (2005) reported the failure of several nests in Minnesota after a 10-d period of constant rain. Conclusion I suggest that limited food availability combined with environmental stress in the form of exposure to the elements during a critical life stage (e.g., rainfall during the nestling stage) may be driving the skewed sex ratios observed in this local North- ern Goshawk population. Of the two hypotheses 3.5 n II S 3 - o 2.5 - CO DC 2 - X 0) 1.5 1 -- 0 2 4 6 8 Combined Rainfall for May and June (cm) Figure 3. Relationship between the sex ratio of fledg- ling Northern Goshawks monitored from 1993-98 and rainfall (r = 0.81, P = 0.05) measured at the U.S. Na- tional Oceanic and Atmospheric Administration weather station located at the Show Low, Arizona, airport within the Apache-Sitgreaves National Forest, during the brood- rearing period (May and June) . presented, the available evidence supports that Northern Goshawks may exhibit selection toward the minimization of investment in the larger sex (i.e., support for the nutritional-stress hypothesis). But at this time the compounding phenomena of a potential increase in exposure time of nestlings because of the decrease in the adult female nest attentiveness due to a possible decrease in prey availability cannot be discarded. During the early nestling period, monitoring adult female nest at- tentiveness (i.e., her sheltering of the nestlings) and the amount of prey brought to the nestlings could help tease apart these two potential expla- nations that may be driving the skewed sex ratio (i.e., exposure or food shortage). Acknowledgments I would like to thank the 1993-98 field crews, especially Renee Wilcox, John Koloszar, and Nora Schubert, for their hard work and dedication over the years. A special gratitude goes to my wife. Merry, for putting up with all the field crews, crew living conditions, and time away. Thanks to Glenn Proudfoot for assisting in my under- standing of avian genetics and genetic analysis. Lastly, I appreciate all the time and energy spent by the reviewers and editors that worked on this manuscript. Literature Cited Anderson, D.J.,J. Reeve, J.E.M. Gomez, W.W. Weathers, S. Hutson, H.V. Cunningham, and D.M. Bird. 1993. Sexual size dimorphism and food requirements of nestling birds. Canadian J. Zoology 71:2541-2545. Avian Biotech International. 2005. Avian Biotech International, Tallahassee, FL U.S.A. <http://www. avianbiotech.com/sexing_center.htm. > Balfour, E. and CJ. Cadbury. 1979. Polygyny, spacing, and sex ratio among Hen Harriers Circus cyaneus in Orkney, Scotiand. Ornis Scandinavica 10:133—141. 252 Ingraldi VoL. 39, No. 3 Bednarz, J.C. AND TJ. Hayden. 1991. Skewed brood sex ratio and sex biased hatching sequence in Harris’s Hawks. Am. Nat. 137:116-132. Boal, C. 1994. A photographic and behavioral guide to aging nestling Northern Goshawks. Stud. Avian Biol. 16:32-40. AND R.W. Mannan. 1994. Northern Goshawk diets in ponderosa pine forests on the Kaibab Plateau. Stud. Avian Biol. 16:97-102, AND D.E. Andersen, and P.L. Kennedy. 2005. Pro- ductivity and mortality of Northern Goshawks in Min- nesota. J. Raptor Res. 39:222-228. Bortolotti, G.R. 1986. Influence of sibling competition on nestling sex ratios of sexually dimorphic birds. Am. Nat. 127:495-507. Brown, D.E. 1994. Biotic communities: southwestern United States and northwestern Mexico. Univ. Utah Press, Salt Lake City, UT U.S.A. Charnov, E.L. 1982. The theory of sex allocation. Prince- ton Univ. Press, Princeton, NJ U.S.A. Clark, A.B. 1978. Sex ratio and local resource competi- tion in a prosimian primate. Science 201 :1&5-165. Clutton-Brock, T.H. 1986. Sex ratio variation in birds. Ibis 128:317-329. , S. Albon, and re. Guinness. 1985. Parental in- vestment and sex differences in juvenile mortality in birds and mammals. 313:131-133. Cronmiller, J.R. and C.R. Thompson. 1981. Sex ratio adjustment in malnourished Red-winged Blackbird broods. J. Field Ornith. 52:65-67. Edwards, T.C., Jr., M.W. Collopy, K. Steenhof, and M.N. Kochert. 1988. Sex ratios of fledgling Golden Eagles. Auk 105:793-796, Fisher, R.A. 1930. The genetical theory of natural selec- tion. Oxford Univ. Press, Oxford, U.K. Fridolfsson, A.K. and H. Ellegren. 1999. A simple and universal method for molecular sexing of non-ratite birds./. Avian Bio. 30:116-121. Griffiths, R., M.C. Double, K. Orr, and RJ .G. Dawson. 1998. A DNA test to sex most birds. Molecular Ecology 7: 1071-1075. Hiraldo, R, J.R Veiga, and M. Manez. 1990. Growth of nestling Black Kites Milvus migrant, effects of hatching order, weather and season. T. Zool. Soc. Lond. 222:197- 214. Howe, H.R 1977. Sex-ratio adjustment in the Common Grackle. Science 198:744—746. Kostrzewa, a. and R. Kostrzewa. 1990. The relationship of spring and summer weather with density and breeding performance of the buzzard Buteo buteo, gos- hawk Accipiter gentilis and kestrel Falco tinnunculus. Ibis 132:550-559. Lee, J.A. 1980. Survival of the smallest nestling in gos- hawks. /. Raptor Res. 14:70-72. Leroux, a. and V. Bretagnolle. 1996. Sex ratio varia- tions in broods of Montagu’s Harriers Circus pygargus J. Avian Biol. 27:63-69. Meyers, J.H. 1978. Sex ratio adjustments under food stress: maximization of quality or numbers of off- spring? Am. Nat. 112:381-388. Newton, I. 1979. Population ecology of raptors. Buteo Books, Vermillion, SD U.S.A. Olsen, P.D. and A. Cockburn. 1991. Female-biased sex allocation in Peregrine Falcons and other raptors. Be- hav. Ecol. Sociobiol. 28:417-423. Palmer, R.S. 1988. Handbook of North American birds. Vol. 4. Diurnal raptors. Yale University Press, New Ha- ven, CT U.S.A. Reynolds, R.T. and S.M. Joy. 1998. Distribution, territory occupancy, dispersal, and demography of Northern Goshawks on the Kaibab plateau, Arizona. Final Rep. Heritage Project No, 194045, Arizona Game and Fish Dept., Phoenix, AZ U.S.A. Rosenfield, R.N., J. Bielefeldt, and S.M. Vos. 1996. Skewed sex ratios in Cooper’s Hawk offspring. Auk 113:957-960. Sheldon, B.C., J. Merila, G. Lindgren, and H. Elle- gren. 1998. Gender and environment sensitivity in nestling Collared Flycatchers. Ecology 79:1939-1948. Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- hawk {Accipiter gentilis) . In A. Poole and F. Gill [Eds.], The Birds of North America, No. 298. The Academy of Natural Sciences, Philadelphia, PA and The Amer- ican Ornithologists’ Union, Washington, DC U.S.A. SPSS. 1997. User’s guide. Base 7.5 for Windows, SPSS Inc., Chicago, IL U.S.A. Steenhof, K. 1987. Assessing raptor reproductive success and productivity. Pages 157-170 in B.A. Pendleton, B.A. Millsap, K.W. Kline, and D.M. Bird [Eds.], Rap- tor management techniques manual. Natl. Wildl. Fed. Sci. Tech. Ser. No. 10., Washington, DC U.S.A. Teather, K.L. and P.J. Weatherhead. 1988. Sex-specific energy requirements of Great-tailed Grackle (Quisca- lus mexicanus) nestlings. J. Anim. Ecol. 57:659 — 668. Torres, R. and H. Drummond. 1997. Female-biased mor- tality in nestlings of birds with size dimorphism. J. Anim. Ecol. 66:859-865. Trtvers, R.L. and D.E. Willard. 1973. Natural selection of parental ability to vary the sex ratio of offspring. Science 179:90—92. Weatherhead, PJ- AND K.L. Teather. 1991. Are skewed fledgling sex ratios in sexually dimorphic birds adap- tive? Am. Nat. 138:1159-1172. Zar, J.H. 1984. Biostatistical analysis. 2nd edition. Pren- tice-Hall, Inc., Englewood Cliffs, NJ U.S.A. Received 2 February 2004; accepted 9 March 2005 Associate Editor: Michael I. Goldstein J Raptor Res. 39(3) :253— 263 © 2005 The Raptor Research Foundation, Inc. NORTHERN GOSHAWK {ACCIPITER GENTILIS LAINGI) POST- FLEDGING AREAS ON VANCOUVER ISLAND, BRITISH COLUMBIA Erica L. McClaren^ BC Ministry of Environment, 2080 Labieux Road, Nanaimo, BC V9T 6J9 Canada Patricia L. Kennedy Eastern Oregon Agricultural Research Center, Oregon State University, RO. Box E, Union, OR 97883 US. A. Donald D. Doyle BC Ministry of Environment, 2080 Labieux Road, Nanaimo, BC V9T 6J9 Canada Abstract. — The area used by immature birds from the time they fledge until independence is the post- fledging area (PEA). Published estimates of PFA size (170 ha) are only available from a Northern Goshawk {Accipiter gentilis atricapillus) population in New Mexico and applicability of this estimate to other regions and habitat types is unknown. Our objectives were to estimate PFA size and length of the post-fledging period for Northern Goshawk (A. g. laingi) nests on Vancouver Island, British Columbia. We estimated PFA size from 95% adaptive kernel estimates of telemetry locations from 12 fledglings at 12 nests between 29 June and 2 September 2001-02 (N = 6, 2001; N = 6, 2002). Because our adaptive kernel estimates are based on a small number of locations, we also estimated the precision of these home range estimates using a smoothed bootstrap approach. Almost all (93%) fledgling locations were within 200 m of nests during the early fledgling-dependency period, but less than half (42.4%) of these locations were within this distance during the late fledgling-dependency period. Northern Goshawks departed from PFAs 45.9 ± 1.3 d post-fledging. Mean PFA size was 59.2 ± 16.1 ha, and the bootstrapped variance around PFA estimates ranged from 12.7-1820.8 ha. Our estimate for the mean size of one PFA per nest area for A. g. laingi fledglings on Vancouver Island was much smaller than the mean size estimate reported for A. g. atricapillus in New Mexico. However, management plans should consider nest areas and PFAs to be one functional component of Northern Goshawk breeding habitat and should include multiple alternative nest trees, each with an associated PFA. Keywords: Northern Goshawk, Accipiter gentilis laingi; activity centers', adaptive kernel', bootstrapping, fledg- ing-dependency period', immature movements', natal dispersal. Areas post-emplumamiento de accipiter gentilis laingi uis Vancouver island, Brit- ish COLUMBIA Resumen. — El area utilizada por las aves inmaduras desde que abandonan el nido hasta que alcanzan la independencia de sus padres es el area post-emplumamiento (APE). El unico estimado publicado del tamano del APE de Accipiter gentilis (170 ha) corresponde a una poblacion del estado de New Mexico (subespecie atricapillus), y la aplicabilidad de este estimado a otras regiones y tipos de habitats es des- conocida. Nuestros objetivos fueron estimar el tamano del APE y la longitud del periodo post-emplu- mamiento para nidos de A. g. laingi ubicados en Vancouver Island, British Columbia. Estimamos el tamano del APE a partir de estimados adaptativos de los kernels del 95% de ubicaciones obtenidas mediante telemetria para 12 volantones de 12 nidos, entre el 29 de junio y el 2 de septiembre de 2001- 02 {N = 6, 2001; N = 6, 2002). Debido a que nuestros estimados de los kernels estan basados en un numero pequeno de ubicaciones, tambien estimamos la precision de los estimados del rango de hogar empleando un metodo de bootstrap alisado. Casi todas las ubicaciones de los volantones (93%) estu- vieron a menos de 200 m de los nidos durante el periodo temprano de emplumamiento-dependencia, pero menos de la mitad de las ubicaciones (42.4%) tuvieron lugar a menos de 200 m durante la fase tardia de este periodo. Los individuos abandonaron sus APE 45.9 ± 1.3 d despues de abandonar el nido. El tamano promedio del APE fue 59.2 ±16.1 ha, y la varianza de los estimados obtenida mediante ^ Corresponding author’s email address: erica.mcclaren@gov.bc.ca 253 254 McClaren et al. VoL. 39, No. 3 el metodo de bootstrap estuvo entre 12.7 y 1820.8 ha. Nuestro estimado del tamano medio de un APE por area de nidificacion para los volantones de A. g. laingi en Vancouver Island es mucho menor que el estimado medio documentado para A. g. atricapillus en New Mexico. Sin embargo, los planes de manejo deben considerar que las areas de nidificacion y las APE son un componente importante del habitat de cria de A. gentilis, y deben incluir multiples arboles que puedan servir como sitios alternativos para nidificar, cada uno con su APE asociada. [Traduccion del equipo editorial] Suitable breeding habitat for avian species con- sists of adequate nest sites, roost sites, post-fledging areas (PFAs), and foraging areas. PFAs represent the habitat used by fledglings prior to indepen- dence and may be especially important for species with long post-fledging-parental-care periods, such as raptors. The survival of fledglings through the post-fledging period and their first year is likely in- fluenced by PFA quality, which may be reflected by PFA size and habitat characteristics. Several studies have reported areas around nests to be important for fledglings during the post- fledging period, before dispersal is initiated (Bald Eagles [Haliaeetus kucocephalus] , Wood et al. 1998; Great Tits [Parus major], Naef-Daenzer et al. 2001; Scarlet Macaws [Ara macao], Myers and Vaughan 2004). However, the PFA concept (originally re- ferred to as the post-fledging family area; Reynolds et al. 1992) and its integration with management prescriptions (Reynolds et al. 1992) have only been applied to Northern Goshawks {Accipiter gentilis) , a species of concern in North America (Kennedy 1997, Crocker-Bedford 1998, DeStefano 1998) and Europe (Widen 1997). Kennedy et al. (1994) esti- mated the size of goshawk PFAs in New Mexico to be ca. 168 ha based on movement patterns of ra- dio-tagged fledglings and adult female core-use ar- eas. Currently, the British Columbia (BC) govern- ment recommends managing a 200-ha PFA around designated goshawk {A. g. laingi) nest areas (BC Ministry of Water, Land, and Air Protection, 2004). This recommendation was a modification of the southwestern U.S. guidelines (Reynolds et al. 1992) because local data were unavailable. Our objective was to evaluate the applicability of PFA guidelines developed for goshawk populations in the southwestern U.S. to coastal BC, where hab- itat characteristics, harvest regimes, and goshawk subspecies differ. We provide the first PFA estimate for goshawks based on home-range estimates of fledglings prior to independence as well as preci- sion estimates of PFAs, which are rarely provided in home-range analyses (Kernohan et al. 2001). Local knowledge of goshawk PFA size on Vancou- ver Island is crucial for adequately managing the breeding habitat of A. g. laingi which is federally designated by the Committee on the Status of En- dangered Wildlife in Canada (COSEWIC) as Threatened in Canada (COSEWIC, in press) and is provincially Red-listed (BC Conservation Data Centre 2005). Methods Study Area. Goshawk nest areas were located on Van- couver Island, BC (Fig. 1) in forests dominated by west- ern hemlock (Tsuga heterophylla) and Douglas-fir (Pseu- dotsuga menziesii), although western red cedar {Thuja plicata) , amabalis fir {Abies amabalis) , and red alder {Alnus rubra) were also abundant. Nest stands ranged in age from 45 to >250 yr. See McClaren et al. (2002, 2003) for more study area details. Data Collection. Goshawk nests {N = 17) used in this study were located either through broadcast surveys of conspecific calls (McClaren et al. 2002, 2003), or inci- dentally by forest company personnel and the public, from 1994-2002. When nestlings were ca. 21 d, we climbed nest trees and lowered nestlings to the ground where they were weighed, measured, sexed, and aged by the senior author to maintain consistency in the data. Nestling gender was determined using tarsal width, rec- ommendations provided by Kenward et al. (1993a). We aged nestlings from a photographic and behavioral key (Boal 1994) and from our estimated hatch dates. Nest- lings were banded with U.S. Geological Survey bands and color-rivet bands (Acraft Sign and Nameplate Co. Ltd., Edmonton, AB Canada). With two exceptions, we fitted only the largest female nestling in each nest with a 9-g tarsal mount transmitter with a mortality switch (Ad- vanced Technology Services, Isanti, MN U.S.A.). Males were fitted with transmitters when: (1) all nestlings were male {N = 1); and (2) the sole female nestling’s trans- mitter battery died, and we fitted its male sibling with a new radiotransmitter. We chose the largest female to re- duce potential variation in fledgling movements caused by gender differences (Byholm et al. 2003, J. Wiens un- publ. data) and to lessen possible impact of transmitter mass on survival probability. Radiotransmitters were at- tached to tarsi with a leather Jesse (Ward and Kennedy 1996, Dewey and Kennedy 2001), so that fledglings could remove them after the 90 d transmitter battery expired. Trade name products are mentioned to provide com- plete descriptions of methods. The authors’ institutions neither endorse these products nor intend to discrimi- nate against products not mentioned. Prior to collecting fledgling location data, we centered a 600 X 600-m grid on nests with young with location September 2005 Biology 255 Map Insert AK. USA I BC. CANADA WA. USA 1:2,100,000 25 0 25 50 75 km Figure 1. Northern Goshawk nest areas on Vancouver Island, British Columbia, where chicks were captured and radiotagged. Small open circles represent nests where there were insufficient radio locations (N < 15) to estimate post-fledgling area (PEA) size and small closed circles represent nests where there were sufficient radio locations to estimate PFA size. Larger circles around nests represent the 20 km ground search area for fledglings after they began to leave the natal area. Figure 2. Reference station grid (600 m X 600 m) at Northern Goshawk nests on Vancouver Island, British Co- lumbia, used for geo-referencing radio-tagged fledgling locations from ground-based telemetry during the 2001- 02 breeding seasons. Each grid cell represents 100 m X 100 m. reference stations at 100-m intervals (Fig. 2). After nest- lings were radio-tagged, nests were visited weekly to mon- itor chick development, survivorship, and transmitter op- eration, Once tagged nestlings fledged, nest areas were visited every 1—3 d to collect fledgling location data. We rotated sampling equally among nest areas and sampling periods (<0800-1100 H, 1101-1400 H, 1401-1700 H, and >1700 H) to ensure all times of day were equally represented. Teams of two observers listened for radio signals prior to entering nest stands. Using a 3-element Yagi antenna (Telonics Inc., Mesa, AZ U.S.A.) and re- ceiver (Models SRX-1000, SRX-400, Lotek Engineering Inc., ON, Canada), observers quietly approached fledg- lings on foot to obtain visual locations and to prevent influencing their movements. We verified the observed fledgling was radio-tagged by identifying color bands. At one nest site where the radio-tagged fledgling was pre- dated 5 d prior to dispersal, we continued tracking the remaining two siblings through aural locations. We mea- sured distance and direction of the fledgling to the clos- est reference grid station using a meter-marked rope and compass. When fledglings became more mobile and 256 McClaren et al. VoL. 39, No. 3 moved outside the 600 X 600-m grid, we either expanded the grid or used triangulation from roads to estimate fledgling locations. We estimated date of departure from natal areas as mid-way between the last visit when fledg- lings were verified <1.6 km from nests and when no ra- dio signal was heard within this distance on two consec- utive telemetry sessions (Kenward et al. 1993a, Kennedy and Ward 2003). Immediately after young departed from natal areas, we conducted intensive road searches for ra- dio-tagged fledglings using a vehicle-mounted omni-di- rectional antenna (Telonics Inc., Mesa, AZ U.S.A.), and when a signal was detected, we used triangulation to ob- tain a location. Road searches were conducted within a 20-km^ area surrounding each nest (Fig. 1). We also searched a 30-km area around nests using a single aerial telemetry flight. Location Determination. We used the survey mapping extension of Road Engineering Software (Softree Tech- nical Systems Inc. 1998) to calculate UTM coordinates for visual fledgling locations based on the known coor- dinates of the nest tree location, grid reference station locations, and measured offsets from grid reference sta- tions to fledglings. Fledgling locations derived from tri- angulation and their associated error ellipses were esti- mated using Locate II, version 1.5 (Nams 2000), based on the number of bearings, angles between bearings, and the distance from bearing locations to radio-tagged birds. Post-fledging Area Estimation. We used Home Ranger (version 1.5, Ursus Software, Revelstoke, BC Canada) to estimate PFA size and a smoothed bootstrap (Worton 1995) with 1000 replications to estimate variance of each PFA. Bootstrapping is a common technique for numeri- cally estimating the precision of measurements for which sampling distributions and variances are unknown (Efron and Tibshirani 1991, Quinn and Keough 2002). Al- though PFAs are not equivalent to home ranges, we as- sumed the area used by fledglings prior to dispersal could be modeled using home-range estimation tech- niques. Ninety-five percent adaptive kernel estimates were used to estimate PFA size because kernel estimators were highest ranked in a recent review of the perfor- mance of home range estimators (Kernohan et al. 2001). Adaptive kernel estimates are non-parametric, indicate areas of concentrated use by fledglings (i.e., activity cen- ters or core areas), and are more conservative than min- imum convex polygon estimators because the home range boundaries are based on probability functions around bird locations rather than on linear connections between the outermost data points (Seaman et al. 1999, Kenward et al. 2001). We only included 95% of locations because we wanted to exclude exploratory or excursion behaviors that could artificially inflate PFA size (Walls and Kenward 1998, Kennedy and Ward 2003). We used the reference bandwidth ( ^^f) smoothing parameter be- cause it appeared to model most accurately the number of activity centers in our data. Triangulated locations with associated ellipsoid error polygons >1 km^ were not in- cluded in PFA estimates {N = 2). We did not estimate PFA size for fledglings with <15 locations (N = 3). A minimum of 15 locations per fledgling appears small rel- ative to the recommended minimum of 30 locations per individual for home-range estimation (Kernohan et al. 2001). However, the short post-fledging period limited our ability to collect >30 locations per fledgling that were not temporally correlated. In addition, area-observation curves (Gese et al. 1990 in Kernohan et al. 2001) sug- gested our sample sizes adequately represented maxi- mum distances moved from nests during the post-fledg- ing period, prior to departure from natal areas. Statistical Analyses. We considered fledgling location data collected from different nest sites within the same nest area in different years to be independent (N = 2). Mean and median hatching, fledging, and departure from natal area dates were estimated with Julian days. We used the correlation procedure in SAS (SAS Institute, Inc. 1997) to examine the relationship between fledging date and the number of days until dispersal was initiated. We used a Fisher’s exact test (PROG FREQ SAS Institute, Inc. 1997) to evaluate changes in the distance fledglings were from nests at three stages of maturity during the post-fledging period. Distances were calculated as the Eu- clidian distance from the nest tree to the fledgling loca- tion. We categorized the distance we observed fledglings (0-99 m, 100-199 m, 200-399 m, 400-799 m, >800 m) into three time intervals (1-3 wk, 4—5 wk, and 6-7 wk post-fledging). Distance categories were as fine scale as possible given the number of locations in each category required to run the analyses. We used Wilcoxon’s rank- sum tests (PROG NPARIWAY; SAS Institute, Inc. 1997) to compare hatching, fledging, dispersal dates, and PFA size between years. We used Wilcoxon’s rank-sum tests for these pair wise comparisons because small sample sizes and non-normal data may invalidate the results provided by Kests (Ott 1993). Results were considered statistically significant at P < 0.10. Means and standard errors are presented unless otherwise stated. Data from all radio- tagged fledglings were included in all analyses except PFA size estimates and comparisons in which we only in- cluded data from young with sufficient locations {N = 12 ). Results Forty-two goshawk nestlings from 17 nests were banded and measured. Mean age of young at banding was 20.2 ± 0.8 d. Tarsus width for male (N = 15) and female {N = 27) chicks averaged 6.1 ±0.1 mm and 7.3 ± 0.1 mm, respectively, and means were significantly different (^s “ 2.0, P < 0 . 001 ). Breeding Phenology. Median hatch date was 29 May (N = 17), median fledge date was 7 July (N = 17), and median date of departure from PFAs was 25 August (N = 15). Goshawk nestlings spent a mean of 40.4 ± 0.3 d (N = 17) in their nests before fledging and 45.9 ± 1.3 d {N = 15) in PFAs before departing from natal areas. Goshawks hatched (S = 65.0, P = 0.07) and left PFAs (5 = 24.5, P = 0.003) significantly earlier in 2002 than in 2001. In 2001, individuals that fledged later spent less time within PFAs than early fledged young (Fig. 3). Thus, hatching date (r = —0.8, P September 2005 Biology 257 60 1 < u. 55 - Q. E 50 o w Li. P 45 - tr 03 40 ■ Q. 0) Q 35 - 170 180 190 200 Julian Day of Fledging Figure 3. Relationship between fledging date and time spent within post-fledgling areas (PFAs) before dispersal initiation for Northern Goshawk fledglings on Vancouver Island, British Columbia, in 2001 and 2002. = 0.007) and fledging date (r = -0.8, P = 0.01; Fig. 3) were negatively correlated with the amount of time young spent in PFAs before initiating dis- persal. In 2002, relationships between hatching date (r = -0.3, P = 0.5) and fledging date (r = -0.3, P = 0.5; Fig. 3) with the amount of time young spent within PFAs were weak because both early and late fledged young spent nearly equal time periods within PFAs. In both years, fledgling departure from PFAs occurred over a 2-wk period (2001: 10-25 August; 2002: 20 August-2 Septem- ber) . Fledgling Location Data. We collected 236 radio- telemetry locations from 15 radio-tagged goshawk fledglings. Most of these were visual locations (93.2% N = 220). Triangulated locations (6.8%; N = 16) had a mean error ellipse of 0.029 ± 0.009 km^, equivalent to a circle around each location with a mean radius of 96.1 m. Ninety-three percent of fledgling locations were within 200 m of nests {x = 107.8 ± 8.9 m, iV = 105) during the first 3 wk post-fledging, but only 42.4% of locations were within this distance (x = 261 ± 17.5 m, N = 131) during the remaining 4 wk post-fledging (Fig. 4) . As fledglings matured, we generally located them farther from nests (2001: 52.5, P < 0.001; 2002: - 32.4, P< 0.001). However, we observed fledglings returning to nest trees throughout the post-fledging period, so fledglings did not contin- ue to expand their PFA size indefinitely until they departed from natal areas. Maximum movements were observed in the 2 wk after fledglings com- pleted feather growth and subsequent feather hardening, ca. 70-75 d of age (Fig. 4). We were 1400 1200 - Fledging Feather Hardening 00% Dispersal Complete -P 1000 800 600 - 400 - 200 35 95 Figure 4. Distance of Northern Goshawk fledglings {N = 15) from nests during the post-fledging period on Van- couver Island, British Columbia, 29 June-2 September, 2001-02. Vertical lines represent median fledge date and estimated median feather-hardening date. Horizontal grey bars represent mean distances fledglings were ob- served from nests pre-feather hardening and post-feather hardening. only able to relocate one fledgling 27 d after it left the PFA, and it was 82 km from its natal site. We gathered 17.8 ± 0.6 (range = 15—22) loca- tions per fledgling from the 12 fledglings for which we had sufficient locations {N > 15) to estimate PFA size (Table 1). We did not include locations from one male fledgling because he was sick and remained near the same perch tree until he was recovered dead 3 wk post-fledging. Overall, mean PFA size was 59.2 ± 16.1 ha {N = 12), and the range was 14.5-229.7 ha (Table 1). Variance around PFA estimates based on the bootstrap sam- ples ranged from 12.7—1820.7 ha (Table 1). PFA size did not significantly differ {S = 39.0, P = 1.0) between 2001 (71.1 ± 32.3 ha) and 2002 (47.4 ± 6.8 ha), although small sample sizes may have re- duced our power to detect annual differences. The size and shape of PFAs varied among fledg- lings, and one fledgling used >1 activity centers. In 75% of PFAs, fledgling activity centers included nest trees. The three fledglings with PFAs that did not include nest trees had activity centers that were ca. 100 m, 150 m, and 300 m from nest trees. Fledgling Fate. Brood reduction occurred at a minimum of seven nests prior to banding. At least three nestlings died post-banding, but pre-fledg- ing. In 2001, 100% of radio-tagged fledglings sur- vived through the post-fledging period (Table 2). 258 McClaren et al. VoL. 39, No. 3 Table 1. Post-fledging area estimates for Northern Goshawk nests (N = 12) on Vancouver Island, British Columbia, 29 June-2 September 2001-02. Nest Area Bootstrap Min., Max. Percent Visual Locations N Name War PFA Size (ha) Great Central Lake 2001 54.3 50.8, 173.8 69 17 Klaklakama No. 7 2001 229.7 182.7, 1820.8 53 20 Loon Lake 2001 34.7 26.4, 275.9 100 18 Paterson 2001 53.4 39.0, 446.3 95 21 Roberts Lake 2001 14.5 12.7, 101.0 95 23 Toad No. 2 2001 40.0 36.4, 107.6 100 20 Sutton 2002 26.7 24.5, 103.0 100 15 John Road 2002 77.8 36.2, 1120.0 100 17 Klaklakama No. 3 2002 46.6 32.0, 272.5 100 19 Toad No. 3 2002 47.9 23.4, 547.8 100 16 Surprise Lake 2002 40.9 34.0, 259.5 100 16 Pye Lake 2002 44.4 23.7, 372.1 100 17 However, in 2002, 37.5% {N = 8) of radio-tagged fledglings, and at least one untagged fledgling, died prior to leaving the natal area. We experi- enced one premature battery failure, and one fledgling removed its transmitter prior to initiating dispersal. Discussion Post-fledging Period Behavior. Goshawk fledg- lings on Vancouver Island exhibited similar move- ment patterns during the post-fledging period to goshawks in Sweden (Kenward et al. 1993a) and New Mexico (Kennedy et al. 1994, Kennedy and Ward 2003). Within the first 3 wk post-fledging, fledglings remained within 200-300 m (x = 107.8 ± 8.9 m, N = 105) of nests, and they were often on the ground or in the lower canopy. Immediately after this period, fledglings experienced a behav- ioral transition in which they were frequently lo- cated farther from nests (x = 261 ± 17.5 m, V = 131); they were adept fliers and perched in the upper canopy, often in treetops. These changes in fledgling behavior correspond with completion of primary and retrix feather growth and subsequent feather hardening (Kenward et al. 1993a). Inter- estingly, goshawks did not continue to expand their PFA size indefinitely until departing from nest areas. Instead, the farthest distance we ob- served fledglings from nests during the post-fledg- ing period peaked within 1-2 wk after they com- pleted feather growth and approximately 10 d prior to departing PFAs. Kenward et al. (1993a) and Minguez et al. (2001) described a similar pat- tern for goshawks in Sweden and for Bonelli’s Eagles (Hieraaetus fasciatus) in Spain, respectively. Such a pattern illustrates the importance of col- lecting fledgling locations uniformly throughout the post-fledging period when trying to character- ize fledgling movement patterns. In 2001, hatch and fledge dates were negatively correlated vdth the amount of time young spent within PFAs. Similar negative relationships between hatch date and age when dispersal was initiated were reported for goshawks in Sweden (Kenward et al. 1993b) and Finland (Byholm et al. 2003). However, in 2002 goshawks initiated breeding ear- lier on Vancouver Island than in 2001, and fledg- lings spent similar amounts of time within PFAs, regardless of their hatch and fledge dates. In sev- eral bird species, an early onset of breeding often indicates higher food availability within the nest area, which results in higher fledgling mass and survival (Dewey and Kennedy 2001, Naef-Daenzer et al. 2001, Aparicio and Bonal 2002). Because we did not manipulate any proximate factors that may have influenced the length of the post-fledging pe- riod, we can only speculate on what influenced the timing of fledgling departure from natal areas and length of post-fledging periods in our study. Food availability within home ranges, predator and com- petitor abundance, and weather are all possible in- fluential factors (Kenward et al. 1993a, Dewey and Kennedy 2001, Byholm et al, 2003). We found no evidence that parental aggression caused fledglings to disperse on Vancouver Island, which supports September 2005 Biology 259 Table 2. Fate of radio-tagged Northern Goshawk fledglings 29 June-11 September 2001-02 during post-fledgling area (PFA) estimation on Vancouver Island, British Columbia. Fledgling ID Year Tagged Date Last Obs. (Bird Age in d) Fate China 2001 27-Aug-Ol (84) Departed from PFA Cous 2001 22-Aug-Ol (88) Departed from PFA Great Central Lake 2001 27-Aug-Ol (79) Departed from PFA Mesachie 2001 16-Aug-Ol (89) Departed from PFA Klaklakama No. 7 2001 24-Aug-Ol (94) Departed from PFA Loon Lake No. 3 2001 24-Aug-Ol (88) Departed from PFA Paterson 2001 25-Aug-Ol (82) Departed from PFA Roberts Lake 2001 25-Aug-Ol (97) Departed from PFA Toad No. 2 2001 24-Aug-Ol (90) Departed from PFA Claud Elliot 2002 29-Jul-02 (63) Dead (unknown cause) Toad No. 3 2002 05-Aug-02 (77) Dead (predated) John Road a^ 2002 03-Aug-02 (75) Battery failed John Road b 2002 ll-Sep-02 (114) Mortality switch on Loon Lake No. 3 2002 14-July-02 (41) Dead (unknown cause) Sutton 2002 lO-Aug-02 (79) Departed from PFA Klaklakama No. 3 2002 19-Aug-02 (87) Departed from PFA Pye Lake 2002 21-Aug-02 (87) Departed from PFA Surprise Lake 2002 23-Aug-02 (83) Departed from PFA ® Two individuals were radiotagged at this nest because the first radiotransmitter battery died. We captured and radiotagged its sibling after a failed attempt to recapture the originally tagged individual. experimental results provided by Kenward et al. (1993a). Although the onset of fledgling dispersal varied by approximately 10 d between 2001 and 2002, fledglings departed natal areas abruptly between 80-96 d of age in both years. This seems to be a common pattern for goshawks (Kenward et al. 1993a, Dewey and Kennedy 2001) and for many other raptors (Spotted Owls [Strix occidentalism. Wil- ley and van Riper 2000; Bonelli’s Eagles: Minguez et al. 2001). In contrast, Walls and Kenward (1998) reported a bimodal pattern of departure from na- tal areas for Common Buzzards {Buteo buteo) and Kennedy and Ward (2003) observed supplemen- tally-fed goshawk fledglings returning to natal ar- eas after they initiated dispersal. We were unable to evaluate movement patterns for radio-tagged goshawks during their first year of life because our transmitter batteries expired when young were ca. 110 d of age. However, we searched for fledglings within 30 km of nest areas after they initiated dis- persal, and our inability to locate them suggested fledglings moved >30 km after departing PFAs. Initial departure distances were probably moder- ated by local food availability, whereby fledglings within food-rich areas moved shorter distances af- ter leaving natal areas than fledglings from food- poor areas (Kenward et al. 1993b, Kennedy and Ward 2003). Post-fledging Area Size. Most fledglings included nest trees within their activity centers throughout the post-fledging period. Similar patterns reported by Ward and Kennedy (1996: goshawks). Wood et al. (1998: Bald Eagles), and Belthoff and Ritchison (1989: Eastern Screech-Owls [Otus asio^) suggest that nest trees are important throughout the post- fledging period for raptors. Some goshawk man- agement guidelines recommend reduced distur- bance levels around goshawk nests until young fledge (e.g., BC Ministry of Water, Land, and Air Protection 2004). However, the vulnerability of young during the early fledging-dependency peri- od (Wiens 2004) and their continued use of the nest site throughout the post-fledging period, sug- gests there should be strict adherence to distur- bance recommendations until young leave PFAs. Disturbance near nest areas during the post-fledg- ing period may interfere with adult prey deliveries to young and development of juvenile hunting and flight skills (Kenward et al. 1993a, Kennedy et al. 1994, Wood et al. 1998). Our study reports the first published estimate of 260 McClaren et al. VoL. 39, No. 3 goshawk PFAs based on home-range estimates de- rived almost entirely from visual locations, with no location error. Our estimated mean PFA size of 59.2 ha is smaller than that reported by Kennedy et al. (1994). Kennedy et al. (1994) based their PFA size estimate on adult female core use areas which were corroborated with fledgling location data, rather than calculating PFA size directly from fledgling locations. Also, reanalysis of fledgling lo- cation data from Kennedy et al.’s (1994) study in- dicated the non-visual observations of fledglings frequently had a 500-m radius error (Kennedy and Ward 2003). Similarly, fledgling distances provided by Kenward et al. (1993a) were likely inflated be- cause their telemetry locations were accurate with- in only 100 m. Our PFA estimates may have been slightly inflated because they included one fledg- ling (Klaklakama No. 7) from which 47% of loca- tions were collected using triangulation with an as- sociated 104.4-m radius error (Table 1). However, our second largest PFA estimate was for a fledgling (John Road) for which 100% of locations were vi- suals. Additionally, post-fledging movement patterns may be influenced by fledgling gender (Byholm et al. 2003, J. Wiens unpubl. data) and by landscape habitat characteristics surrounding nests. Our PFA estimates may be smaller than those reported by Kennedy et al. (1994) because all but one PFA es- timate were for females that, in one northern Ar- izona study, were smaller than male PFAs (J. Wiens unpubl. data). PFA size on Vancouver Island may also be smaller than in New Mexico because the definitive forest edges of nest stands in coastal for- est ecosystems may act as barriers to fledgling movements more than the less defined ecotones that occur between southwestern forest types (Sid- ers and Kennedy 1996). Because PFA size can only be estimated from lo- cation data, providing variance estimates for these and other types of home range data is extremely informative, but rarely done (Worton 1995, Ker- nohan et al. 2001). Our variance estimates of the PFA estimates include the 169 ± 129 ha PFA size reported by Kennedy et al. (1994) and the dis- tances (100-1000 m) that Kenward et al. (1993a) observed fledglings from nests. Our PFA estimate was closer to the minimum bootstrapped estimate than the maximum bootstrapped estimate because a greater proportion of our location data were clos- er to nests. Few fledgling locations far from nests created more variability in the maximum boot- strapped estimate, although maximum estimates provide important information (F. Hovey pers. comm.). Home-range estimates also vary depending on the techniques used to collect location data and on the home-range estimation program used to calculate home range size (Lawson and Rodgers 1997, Seaman et al. 1999, Kenward et al. 2001). For example, Kennedy et al. (1994) used a harmonic mean estimator to calculate female core use areas, whereas we used an adaptive kernel estimate. Therefore, comparison of PFA size estimates among studies that use different data collection and size estimation techniques is difficult. Management Implications. Most goshawk man- agement guidelines in North America are based on Reynolds et al. (1992), which suggest managing for three hierarchical levels of goshawk home ranges: (1) nest area, (2) PFA, and (3) foraging area. How- ever, Reynolds et al. (1992) also recommended managing for alternative nests within goshawk nest areas and they assumed that all alternative nests were within PFAs. Therefore, the biological func- tionality of a nest area independent of a PFA is questionable, and managing these habitat compo- nents in isolation may reduce the effectiveness of management plans. Recent studies comparing hab- itat characteristics around goshawk nests to ran- dom sites (areas assumed not to contain goshawk nests) at multiple spatial scales concluded that gos- hawk habitat could be discriminated from random sites by a larger proportion of large-diameter, late- seral, closed-canopy forests (Ethier 1999) at scales between 83 ha (McGrath et al. 2003) and 170 ha (Daw and DeStefano 2001). Additionally, Finn et al. (2002) reported occupied historic goshawk nests had a greater proportion of late-seral forest with high canopy closure, less stand initiation cov- er, and reduced landscape heterogeneity at 177 ha and 1886 ha scales, than at similar scales around unoccupied historic nests. These studies suggest goshawk PFAs may be characterized by unique hab- itat characteristics at spatial scales within the size range we have reported for PFAs as well as the size range reported by Kennedy et al. (1994). Mean PFA size estimates on Vancouver Island were smaller than the 200-ha area currently rec- ommended for managing the area around gos- hawk nests in coastal BC (BC Ministry of Water, Land, and Air Protection 2004). However, our re- sults represent only one nest, and therefore, one PFA per nest area, within a given year. We moni- September 2005 Biology 261 tored fledglings from two different nest sites within two nest areas in 2001 and 2002, and there was minimal overlap between PFAs in different years. This suggests that each alternative nest site may have a unique PFA. Therefore, a more meaningful approach to managing goshawk breeding home ranges is to manage for areas that include multiple nests and associated PFAs. Our bootstrapping re- sults suggest this PFA size is highly variable and likely depends upon methods used to estimate PFA size as well as environmental factors such as topog- raphy, habitat characteristics around nests, prey availability, and fledgling gender (Dewey and Ken- nedy 2001, Byholm et al. 2003, Kennedy and Ward 2003, J. Wiens, unpubl. data). We developed a simplistic graphical depiction of how our information could be applied to develop management scenarios for A. g. laingi nest areas throughout coastal BC (Fig. 5) . This figure is based on Vancouver Island data with a mean number of 3.0 ± 0.2 (N = 34 nest areas) alternative nests/ nest area and a mean distance of 274 ± 37.2 m (N — 65) between alternative nest trees (E. McClaren unpubl. data) . The total area to be managed would vary by nest area and depends on the juxtaposition of alternative nests and PFAs (Fig. 5). For example, Figures 5a and 5b depict areas that are 104.8 ha and 96.3 ha in size, respectively. In areas where the inter-alternate distance is larger (inter-alternate distances >1.0 km are not uncommon; Dewey et al. 2003, Squires and Kennedy in press), the total management area would be larger. In the absence of fledgling radiotelemetry data and information on fledgling habitat selection patterns, multiple PFAs within one goshawk home range should be managed to create an area that maintains connec- tivity among alternative nests and to adjacent stands of similar habitat (i.e., reduce stand isola- tion) to minimize possible edge effects, facilitate food transfers from adults, and provide dispersal corridors. Although our results suggested the area used by goshawk fledglings on Vancouver Island, and pos- sibly elsewhere, was smaller than estimated in New Mexico (Kennedy et al. 1994), PFA habitat was not the only habitat necessary for goshawks to success- fully reproduce. Prey availability in habitats outside of, but in proximity to, PFAs was also essential for adults to rear young (Reynolds et al. 1992, Ken- nedy and Ward 2003, Wiens 2004). For example, Bloxton (2002) reported radio-tagged adult gos- hawks in the Olympic Peninsula, WA, to concen- Figure 5. Conceptual representation of managing three alternative Northern Goshawk nests and their associated post- fledgling areas (PFAs) . Two possible configurations, (a) all nests adjacent or (b) two adjacent and one below, are shown with 274 m as the mean distance between al- ternative nests and a mean PFA size of 59 ha on Vancou- ver Island, British Columbia. The diagram is drawn to scale. trate their foraging efforts within 5 km of occupied nests during the breeding season. Current goshawk management guidelines in BC (BC Ministry of Wa- ter, Land and Air Protection 2004) do not include explicitly managing for goshawk foraging areas, and the effect of not managing goshawk foraging areas in landscapes actively managed for timber harvest is unknown. Research Recommendations. In future PFA stud- ies, we recommend increasing the minimum num- ber of locations/ fledgling to a minimum of 30 to improve the precision of PFA estimates (Kernohan et al. 2001). Goshawk post-fledging periods are ex- tremely short, and the timeframe for data collec- tion is limited. Obtaining reasonable samples of lo- cations will require collecting either daily locations 262 McClaren et al. VoL. 39, No. 3 after young fledge or collecting multiple locations per sample day and relaxing guidelines around in- dependence of locations. A sampling regime that spaces location data collection evenly throughout the duration of the study, enabling a reasonable amount of time for animals to relocate, may be more important than concerns about autocorrela- tion of data (Kernohan et al. 2001), Additionally, PFA size and habitat studies should be conducted across a diversity of ecosystems, so that manage- ment recommendations may be fine-tuned to re- flect similarities and differences across broad geo- graphic areas. This information may assist with designating suitable PFAs around nests when it is not feasible to collect radiotelemetry data. Acknowledgments We appreciate Forest Renewal BC and the Forest In- vestment Account funding allocated by Canadian Forest Products, Weyerhaeuser, TimberWest and Western Forest Products, as well as the cooperation of these companies, enabling us to work on their tree farm licenses and pri- vate lands. Pacific Northwest Canadian Studies Consor- tium provided assistance with travel for authors to collab- orate. This project would not have been possible without the labor-intensive data collection provided by; J. Bond, N. Davey, R. Dickson, I. Jacobs, M. Krkosek, P. Levesque, J. Malt, C. Pendergast, J. Strain, and A. Zeeman. We ex- tend our gratitude to I. Jacobs for fearlessly climbing gos- hawk nest trees and to C. Pendergast for assisting with data organization and the figures and tables in this re- port. T. Dunlop assisted with the CIS work and calcula- tions for the conceptual PFA diagram. We thank D. Cator for keeping her eyes and ears cued for goshawks while canoeing. S. Dewey, F. Doyle and J. Ward provided valu- able technical advice for this project. Discussions with J. Horne, F. Hovey, J. Millspaugh, and A. Rodgers clarified our understanding of home-range estimation programs and conducting bootstrapping within home-range pro- grams. J. Morrison and two anonymous reviewers im- proved the quality of this manuscript with their editorial suggestions. Literature Cited Aparicio, J.M. and R. Bonal. 2002. Effects of food sup- plementation and habitat selection on timing of Less- er Kestrel breeding. Ecology 83:873—877. Belthoff, J.R. and G. Ritchison. 1989. Natal dispersal of Eastern Screech-Owls. Condor 91:254— 265. Bloxton, T.D., Jr. 2002. Prey abundance, space use, de- mography, and foraging habitat of Northern Gos- hawks in western Washington. M.S. thesis, Univ. Wash- ington, Seattle, WA U.S.A. Boat. C.W. 1994. A photographic and behavioral guide to aging nestling Northern Goshawks. Stud. Avian Biol. 16:32-40. British Columbia Ministry of Water, Land, and Air Protection. 2004. Identified wildlife management strategy accounts and measures for managing identi- fied wildlife. Available: http://wlapwww.gov.bc.ca/ wld/identified/accounts.html (27 June 2005). British Columbia Conservation Data Centre. 2005 British Columbia species and ecosystem explorer. Brit- ish Columbia Ministry of Sustainable Resource Man- agement. Victoria, BC Canada. Available: <http:// srmapps.gov.bc.ca/apps/eswp/ (June 27, 2005). Byholm, R, P. Saurola, H. Linden, and M. Wikman. 2003. Causes of dispersal in Northern Goshawks {Ac- dpiter gentilis) banded in Finland. Auk 120:706-716. Committee on the Status of Endangered Wildlife in Canada. In press. COSEWIC assessment and update status report on the Northern Goshawk lain^ subspe- cies Accipiter gentilis laingi in Canada. Committee on the Status of Endangered Wildlife in Canada, Ottawa, ON Canada. Crocker-Bedford, D.C. 1998. The value of demographic and habitat studies in determining the status of Northern Goshawks {Acdpiter gentilis atricapillus) with special reference to Crocker-Bedford (1990) and Ken- nedy (1997)./. Raptor Res. 32:329-335. Daw, S.K. and S. DeStefano. 2001. Forest characteristics of Northern Goshawk nest stands and post-fledging family areas in Oregon. J. Wildl. Manag. 65:46-58. DeStefano, S. 1998. Determining the status of Northern Goshawks in the west: is our conceptual model cor- rect? J. Raptor Res. 32:342-348. Dewey, S.R. and P.L. Kennedy. 2001. Effects of supple- mental food on parental-care strategies and juvenile survival of Northern Goshawks. Auk 118:352-365. -, , and R.M. Stephens. 2003. Are dawn vo- calization surveys effective for monitoring goshawk territory occupancy? J. Wildl. Manag. 67:390-397. Efron, B. and R. Tibshirani. 1991. Statistical data anal- ysis in the computer age. Sdence 253:390-395. Ethier, T. 1999. Breeding habitat of Northern Goshawks {Acdpiter gentilis laingi) on Vancouver Island: a hier- archical approach. M.S. thesis, Univ. Victoria, Victoria, BC, Canada. Finn, S.R, D.E. Varland, and J.M. Ma rzl uff. 2002. Does Northern Goshawk breeding occupancy vary with nest-stand characteristics on the Olympic Peninsula, Washington? /. Raptor Res. 36:265-279. Kennedy, P.L. 1997. The Northern Goshawk {Acdpiter gen- tilis atricapillus) : is there evidence of a population de- cline? J. Raptor Res. 31:95-106. and J.M. Ward. 2003. Effects of experimental food supplementation on movements of juvenile Northern Goshawk {Acdpiter gentilis atricapillus). Oec- ologia 134:284-291. , , G.A. Rinker, and J.A. Gessaman. 1994. Post-fledging areas in Northern Goshawk home rang- es. Stud. Avian Biol. 16:75-82. Kenward, R.E., R.T. Clarke, K.H. Hodder, and S. Walls. 2001. Density and linkage estimators of home range. September 2005 Biology 263 nearest-neighbor clustering defines multinuclear cores. Ecology 82:1905-1920. , V. Marcstrom, and M. Karlbom. 1993a. Post- nestling behaviour in goshawks, Accipiter gentilis: the causes of dispersal. Anim. Behav. 46:365-370. , , AND . 1993b. Post-nestling behav- iour in goshawks, Accipiter gentilis: II. Sex differences in sociality and nest-switching. Anim. Behav. 46:371- 378. Kernohan, B.J., R.A. Gitzen, and J.J. Millspaugh. 2001. Analysis of animal space use and movements. Pages 125-166 wJ.J. Millspaugh and J.M. Marzluff [Eds.], Radio tracking and animal populations. Academic Press, San Diego, CA U.S.A. Lawson, E.J.G. and A.R. Rodgers. 1997. Differences in home-range size computed in commonly used soft- ware programs. Wildl. Soc. Bull. 25:721-729. McClaren, E.L., P.L. Kennedy, and P.L. Chapman. 2003. Efficacy of male goshawk food-delivery calls in broad- cast surveys on Vancouver Island. J. Raptor Res. 37: 198-208. , , AND S.R. Dewey. 2002. Do some North- ern Goshawk nest areas consistently fledge more young than others? Condor 104:343—352. McGrath, M.T., S. DeStefano, R.A. Riggs, L.L. Irwin, AND G.J. Roloff. 2003. Spatially explicit influences on Northern Goshawk nesting habitat in the interior Pa- cific northwest. Wildl. Monogr. 154:1-63. Minguez, E., E. Angulo, and V. Siebering. 2001. Factors influencing length of the post-fledging period and timing of dispersal in Bonelli’s Eagle {Hieraaetus fas- ciatus) in southwestern Spain. 2001. f. Raptor Res. 35: 228-234. Myers, M.C. and C. Vaughan. 2004. Movement and be- havior of Scarlet Macaws {Ara macao) during the post- fledging dependence period: implications for in situ versus ex situ management. Biol. Conserv. 118:411- 420. Naef-Daenzer, B., F. Widmer, and M. Nuber. 2001. Dif- ferential post-fledging survival of Great and Coal Tits in relation to their condition and fledging date. J. Anim. Ecol. 70:730-738. Nams, VO. 2000. Locate II user’s guide. Pacer, Truro, Nova Scotia, Canada. Ott, R.L. 1993. An introduction to statistical methods and data analysis, 4th Ed. Duxbury Press, Belmont, CA U.S.A. Quinn, G.P. and M.J. Keough. 2002. Experimental de- sign and data analysis for biologists. Cambridge Univ. Press, Cambridge U.K. Reynolds, R.T., R.T. Graham, M.H. Reiser, R.L. Bassett, P.L. Kennedy, D.A. Boyce, G. Goodwin, R. Smith, and E.L. Fisher. 1992. Management recommendations for the Northern Goshawk in the southwestern United States. USDA Forest Service General Technical Re- port RM-GTR-217, Fort Collins, CO U.S.A. SAS Institute, Inc. 1997. SAS/STAT Software: changes and enhancements through release of 6.12. SAS In- stitute, Inc., Cary, NC USA. Seaman, D.E., JJ- MILI.SPAUGH, B.J. Kernohan, G.C. Brundige, K.J. Raedeke, and R.A. Gitzen. 1999. Ef- fects of sample size on kernel home range estimates. J. Wildl. Manag. 63:739-747. SiDERS, M.S. AND P.L. Kennedy. 1996. Forest structural characteristics of Accipiter nesting habitat: is there an allometric relationship? Corerfor 98:123-132. SoFTREE Technical Systems Inc. 1998. Road Eng forest engineer vs. 3.0. Softree Technical Systems, Inc., West Vancouver, BC Canada. Squires, J. and P.L. Kennedy. In press. Northern Gos- hawk ecology: an assessment of current knowledge and information needs for conservation and manage- ment. Stud. Avian Biol. Walls, S.S. and R.E. Kenward. 1998. Movements of ra- diotagged buzzards Buteo buteo in early life. Ibis 140' 561-568. Ward, J.M. and P.L. Kennedy. 1996. Effects of supple- mental food on size and survival of juvenile Northern Goshawks. Auk 113:200-208. Widen, P. 1997. How, and why, is the goshawk {Accipiter gentilis) affected by modern forest management in Fennoscandia? J. Raptor Res. 31:107-113. Wiens, J.D. 2004. Post-fledging survival and natal dis- persal of Northern Goshawks in Arizona. M.S. thesis, Colorado State University, Fort Collins, CO U.S.A. Willey, D.W. and C. van Riper, III. 2000. First-year move- ments by juvenile Mexican Spotted Owls in the Can- yonlands of Utah./. Raptor Res. 34:1-7. Wood, P.B., M.W. Collopy, and C.M. Sekerak. 1998. Postfledging nest dependence period for Bald Eagles in Florida./. Wildl. Manag. 62:333-339. WORTON, BJ- 1995. Using Monte Carlo simulation to evaluate kernel-based home range estimators./. Wildl. Manag. 59:794—800. Received 2 February 2004; accepted 2 September 2004 Guest Editor: Stephen DeStefano J. Raptor Res. 39(3):264— 273 © 2005 The Raptor Research Foundation, Inc. NORTHERN GOSHAWK DIET IN MINNESOTA: AN ANALYSIS USING VIDEO RECORDING SYSTEMS Brett L. Smithers^ Department of Range, Wildlife and Fisheries Management, Texas Tech University, Lubbock, TX 79409-2120 US. A. Clint W. Boat us. Geological Survey, Texas Cooperative Fish and Wildlife Research Unit, Texas Tech University, Lubbock, TX 79409-2120 U.S.A. David E. Andersen U.S. Geological Survey, Minnesota Cooperative Fish and Wildlife Research Unit, University of Minnesota, St. Paul, MN 55108-6124 U.S.A. Abstract. — ^We used video-recording systems to collect diet information at 13 Northern Goshawk (Ac- cipiter gentilis) nests in Minnesota during the 2000, 2001, and 2002 breeding seasons. We collected 4871 hr of video footage, from which 652 prey deliveries were recorded. The m^ority of prey deliveries identified were mammals (62%), whereas birds (38%) composed a smaller proportion of diet. Mammals accounted for 61% of biomass delivered, and avian prey items accounted for 39% of prey biomass. Sciurids and leporids accounted for 70% of the identified prey. Red squirrel ( Tamiasciurus hudsonicus ) , eastern chipmunk ( Tamias striatus) , and snowshoe hare (Lepus americanus) were the dominant mammals identified in the diet, while American Crow {Corvus brachyrhynchos) and Ruffed Grouse {Bonasa umbellus) were the dominant avian prey delivered to nests. On average, breeding goshawks delivered 2.12 prey items/d, and each delivery averaged 275 g for a total of 551 g delivered/d. However, daily {P< 0.001) and hourly {P — 0.01) delivery rates varied among nests. Delivery rates (P = 0.01) and biomass delivered {P = 0.038) increased with brood size. Diversity and equitability of prey used was similar among nests and was low throughout the study area, most likely due to the dominance of red squirrel in the diet. EIey Words: Northern Goshawk, Accipiter gentilis; diet, Minnesota', prey diversity, red squirrel, Tamiasciurus hudsonicus. DIETA DE ACCIPITER GENTILIS EN MINNESOTA: UN ANALISIS BASADO EN SISTEMAS DE GRA- BACION EN VIDEO Resumen. — Empleamos sistemas de grabacion en video para recolectar informacion sobre la dieta de Accipiter gentilis en 13 nidos ubicados en Minnesota durante las temporadas reproductivas de 2000, 2001 y 2002. Obtuvimos 4871 hr de grabacion, a partir de las cuales registramos 652 entregas de presas. La mayoria de las presas entregadas que identificamos fueron mamiferos (62%), mientras que las aves (38%) representaron una proporcion menor de la dieta. Los mamiferos y las aves representaron el 61% y el 39% de la biomasa entregada, respectivamente. Los sciuridos y leporidos representaron el 70% de las presas identificadas. Los mamiferos predominantes identificados en la dieta fueron Tamiasciurus hudsonicus, Tamias striatus y Lepus americanus, mientras que las aves llevadas a los nidos predominante- mente fueron Corvus brachyrhynchos y Bonasa umbellus. En promedio, los individuos nidificantes entre- garon 2.12 presas/d, y cada entrega tuvo un promedio de 275 g, para un total de 551 g entregados/d. Sin embargo, las tasas diarias {P < 0.001) y horarias {P = 0.01) de entrega de presas variaron entre nidos. Las tasas de entrega (P = 0.01) y la biomasa entregada (P = 0.038) incrementaron con el tamano de la nidada. La diversidad y equitabilidad de las presas consumidas fueron similares entre nidos y bajas a traves del area de estudio, probablemente debido a la dominancia de T. hudsonicus en la dieta. [Traduccion del equipo editorial] ^ Present address and corresponding author: P.O. Box 1363, Meeker, CO 81641 U.S.A.; email: brett^mithers@blm.gov 264 September 2005 Biology 265 The Northern Goshawk {Accipiter gentilis) is a large, forest-dwelling raptor generally associated with mature deciduous, coniferous, or mixed for- ests (e.g., Bright-Smith and Mannan 1994, Siders and Kennedy 1996, Beier and Drennan 1997, Squires and Reynolds 1997). Goshawk research in North America has been conducted primarily in the western half of the continent (Boal et al. 2003). Consequently, there is little published literature describing ecology of the species in the Western Great Lakes Region (WGLR) of North America, where it is currently listed as a Migratory Non- game Bird of Management Concern by the U.S. Fish and Wildlife Service (Region 3) and as a sen- sitive species by the U.S. Forest Service (Region 9) due to loss of habitat (Reynolds et al. 1992). Depending on region, season, and availability, goshawks capture a wide variety of prey and are considered prey generalists (Squires and Reynolds 1997, Squires and Kennedy 2005). Although breed- ing-season diet composition has been studied for many populations (e.g., Meng 1959, Grzybowski and Eaton 1976, Boal and Mannan 1994, Younk and Bechard 1994, Lewis 2001), site-specific studies of diet are necessary for developing management strategies for goshawk populations at regional and local levels (e.g., Reynolds et al. 1992). A number of records exist of prey items collected opportu- nistically at goshawk nests in the WGLR (Eng and Gullion 1962, Apfelbaum and Haney 1984, Mar tell and Dick 1996), but these reports are anecdotal and provide a prey list rather than a quantitative assessment of food habits (Roberson et al. 2003) . Methods used in goshawk food habits research have included indirect (i.e., identification of prey remains or contents of regurgitated pellets) and direct observations of prey deliveries to nests (Meng 1959, Grzybowski and Eaton 1976, Bosa- kowski and Smith 1992, Boal and Mannan 1994). Indirect methods of assessing raptor diet can lead to biased results (e.g., Bielefeldt et al. 1992), whereas direct methods should provide the least- biased results (Collopy 1983, Marti 1987, Boal and Mannan 1994). During the breeding seasons of 2000-02, we used videography as a modified meth- od of direct observation of prey deliveries to ex- amine diet of Northern Goshawks in northern Minnesota. Methods Study Area. The study area was located in the Lauren- tian Mixed-Forest Province of north-central and north- eastern Minnesota (46°50'N, 92°11'W) as described by Boal et al. (2001) and Roberson (2001; Fig. 1). The study area elevation ranged from ca. 200-400 m. Mean summer and winter temperatures were 18°C and — 11°C, respec- tively, and maximum and minimum temperature records for the region were 40°C and — 46°C, respectively (Daniel and Sullivan 1981). Annual precipitation averaged 60-70 cm. The study area was dominated by pine, mixed-hard- wood, boreal, and second-growth forests with wetland community types interspersed among forest stands (Tes- ter 1995). Goshawk Nests. Nests included in this study were con- sidered as sampling units and were selected from all known occupied nests in the study area (Boal et al. 2001) . With the exception of one nest, where few data were col- lected during 2000, diet information was not collected at any nest for more than one breeding season. Nests were selected randomly within the constraints of accessibility and to include different land ownerships. Thus, our sam- ple is not truly random and may not be representative of the goshawk population of our study area. However, to examine the applicability of our diet data to the goshawk population as a whole, we examined prey diversity and overlap among nests. High overlap and low diversity would suggest prey use was similar among goshawk pairs and that our data were representative of the population in general. Video Recording. We used VHS (Model SL 800, Se- curity Labs®, Noblesville, IN U.S.A.) and 8-mm video re- cording systems (Sony® Model M-350, Fuhrman Diversi- fied, Inc., Seabrook, TX U.S.A.) with color or black-and-white cameras (Model CCM-660W, Clover Elec- tronics®, Los Alamitos, CA U.S.A.). Cameras were in- stalled on nest trees within 0.6 m of the nest or, for cam- eras with zoom lenses, on an adjacent tree up to 9 m from the nest. Video recorders were placed in weather- proof cases ca. 30 m from the base of each camera tree. Coaxial-video cables were used to convey power to and transmit images from the cameras. Recorders were pro- grammed to record from 0530-2100 H (15.5 hr of foot- age) at the 48-hr (1.3 frames/sec) or the 72-hr (0 8 frames/sec) setting to optimize the amount of tape used per sampling session and battery life. We replaced tapes and batteries every 3-4 d. Prey Identification. To identify prey delivered to nests, we reviewed video footage until a prey delivery occurred, then advanced frame by frame and freeze-framed to fa- cilitate prey identification. We identified avian and mam- malian prey by morphological features and developed a list of prey species delivered by goshawks to all nests (Ta- ble 1). Goshawks may cache prey and retrieve cached prey items (Boal and Mannan 1994), which could bias estimates of delivery rates and proportional use of species in the diets. We attempted to identify cached prey on basis of a successive, iterative process that included com- paring prey items using flesh color, pelage or feather con- dition, and time of delivery from review of video footage, and then remove those items thought to be cached from analysis. Age and Biomass Estimation. We assigned avian prey to age categories (e.g., adult, juvenile, or nestling) based on plumage (e.g., feathers and down) and amount of sheathing on flight feathers (Reynolds and Meslow 266 Smithers et al. VoL. 39, No. 3 Figure 1. Study area and distribution of Northern Goshawk nests in Minnesota where food habits information was collected during the 2000-02 breeding seasons. The three-letter designations indicate individual nests. Breeding season diet information collected at the DTK breeding area was omitted from all analyses because of nest failure. Breeding areas with similar prey composition are indicated with the same superscripts. Superscripts indicate cluster number (see Fig. 2). 1984). We categorized mammalian prey as adults or ju- veniles based on size (Bielefeldt et al. 1992). Because of difficulty in estimating age of small mammals, we consid- ered all mammals smaller than chipmunks to be adults. Biomass for partial prey items was calculated using the proportion of prey delivered to nests, and proportions were estimated qualitatively (e.g., 50% of adult size). We estimated biomass for prey identified to family, ge- nus, or species and used the mean mass of both sexes (Reynolds and Meslow 1984, Lewis 2001). Biomass esti- mates were based on published information on mam- malian and avian species occurring in the study area (Burt and Grossenheider 1980, Jones and Birney 1988, Dunning 1993, Dunn and Garrett 1997, Dunn 1999, Sib- ley 2000). We calculated mass for nestlings following Bielefeldt et al. (1992) using 100% of the adult mass for warbler-sized species, 65% of the adult mass for robin and jay-sized species, and 55% of the adult mass for large birds such as grouse. We calculated mass of juvenile red squirrel ( Tamiasciurus hudsonicus ) , eastern chipmunk September 2005 Biology 267 Table 1. Number, percent occurrence, and biomass of mammalian and avian prey delivered to Northern Goshawk nests {N = 13) in Minnesota, 2000—02. Values represent pooled number of prey identified at nests during the 2000, 2001, and 2002 breeding seasons. Prey Category Common Name N Percent Biomass (g) Percent Mammals Tamiasdurus hudsonicus red squirrel 202 31.0 38046 23.6 Tamias striatus eastern chipmunk 95 14.6 8108 5.0 Lepus americanus snowshoe hare 31 4.8 41027 25.5 Sylvilagus floridanus eastern cottontail 7 1.1 7654 4.8 Sdurus carolinensis eastern gray squirrel 3 0.5 1679 1.0 Peromyscus spp. 2 0.3 47 0.0 Family: Muridae 1 0.2 18 0.0 Mustela frenata long-tailed weasel 1 0.2 210 0.1 Unknown mammal (MSC1)“ 8 1.2 186 0.1 Unknown mammal (MSC2)® 9 1.4 1720 1.1 Birds Corvus brachyrhynchos American Crow 37 5.7 14515 9.0 Bonasa umbellus Ruffed Grouse 33 5.1 18448 11.5 Ay thy a spp. diving duck 12 1.8 11360 7.1 Cyanodtta cristata Blue Jay 8 1.2 664 0.4 Fulica americana American Coot 6 0.9 3338 2.1 Turdus migratorius American Robin 3 0.5 205 0.1 Quiscalus quiscula Common Crackle 3 0.5 341 0.2 Family: Icteridae blackbird 3 0.5 189 0.1 Picoides spp. woodpecker 3 0.5 199 0.1 Dryocopus pileatus Pileated Woodpecker 3 0.5 861 0.5 Unknown duckling 4 0.6 400 0.2 Butorides virescens Green Heron 2 0.3 420 0.3 Perisoreus canadensis Gray Jay 2 0.3 142 0.1 Agelaius phoeniceus Red-winged Blackbird 2 0.3 105 0.1 Strix varia Barred Owl 1 0.2 394 0.2 Buteo platypterus Broad-winged Hawk 1 0.2 455 0.3 Genus: Calidris 1 0.2 73 0.0 Bucephala clangula Common Goldeneye 1 0.2 900 0.6 Accipiter cooperii Cooper’s Hawk 1 0.2 439 0.3 Gallus spp. domestic chicken*’ 1 0.2 Coccothraustes vespertinus Evening Grosbeak 1 0.2 59 0.0 Pipilo erythrophthalmus Eastern Towhee 1 0.2 41 0.0 Genus: Euphagus 1 0.2 63 0.0 Acdpiter gentilis Northern Goshawk 1 0.2 820 0.5 Picoides villosus Hairy Woodpecker 1 0.2 66 0.0 Charadrius vodferus Killdeer 1 0.2 97 0.1 Anas platyrhynchos Mallard 1 0.2 1082 0.7 Sitta canadensis Red-breasted Nuthatch 1 0.2 10 0.0 Sdurus aurocapillus Ovenbird 1 0.2 19 0.0 Catharus fuscescens Veery 1 0.2 31 0.0 Unknown nestling 33 5.1 1190 0.7 Unknown bird (ASCl)^ 18 2.8 173 0.1 Unknown bird (ASC2)® 23 3.5 1778 1.1 Unknown bird (ASC3)^ 6 0.9 3459 2.1 Items not identified to class Mammalia or Aves 76 11.7 ■* MSCl = mouse-sized prey item; MSC2 = red squirrel-sized prey item; ASCI = warbler-sized prey item: ASC2 = robin-sized prey Item; ASC3 = Ruffed Grouse-sized prey item. ^ Omitted from analysis. 268 Smithers et al. VoL. 39, No. 3 {Tamias striatus), snowshoe hare {Lepus americanus), and eastern cottontail (Sylvilagus floridanus) using 95% of the adult mass; if ages could not be determined reliably, we assigned juvenile masses to these species. To estimate biomass of unidentified prey, we pooled unidentified birds into three a priori size classes (SC) fol- lowing Storer (1966) and Kennedy and Johnson (1986) that represented average mass of common species in our study area: SCI = 10 g (e.g., warbler-sized), SC2 = 77 g (e.g., robin-sized), and SC3 = 576 g (e.g., Ruffed Grouse [Bonasa umbellus]-sized) , Similarly, we pooled unidenti- fied mammal prey into two a priori size classes: SCI = 23 g (e.g., mouse-sized) and SC2 = 192 g (e.g., squirrel- sized) . Prey and Biomass Delivery Rates. We calculated deliv- ery rates on the basis of number of prey delivered per day, number of prey delivered per nestling per day, and number of prey delivered per day at nests with one, two, and three nestlings. We calculated biomass estimates in the same manner. We calculated mean delivery rates over 5-d intervals from hatching to 5 d post-fledging (i.e., from 0-45 d). Prey Diversity and Overlap. We calculated prey diver- sity for the study area using ungrouped prey categories (i.e , using each prey category identified to family, genus, or species separately) . Because samples were smaller when examining individual nests, we generalized prey into similar species categories (Lewis 2001) to calculate prey diversity for individual nests. The generalized prey categories for among-nest diversity assessment were: (1) Sciurids, (2) blackbirds and Corvids, (3) Leporids, (4) Ruffed Grouse, (5) diving ducks (Aythya spp.) , (6) water and shore birds, (7) passerines, (8) Picidae, (9) Falcon- iforms, (10) miscellaneous mammals (e.g., long-tailed weasel [Mustela frenata ] ) . We calculated prey diversity using Williams (1964) and MacArthur’s (1972) modified form of the Simpson’s in- dex (Simpson 1949) and diet equitability using Smith and Wilson’s index of evenness (Smith and Wilson 1996) . We used prey identified to family, genus, or species to estimate diet overlap among nests with the Simplified Morisita’s Index of Overlap (Krebs 1999). Overlap mea- sures are designed to measure the degree that two spe- cies share a set of common resources or utilize the same parts of the environment (Lawlor 1980). Overlap mea- sures are scaled from zero to one, where zero overlap indicates dissimilarity in resource use, and one indicates complete overlap (Krebs 1999). We also assessed similar- ity in prey use among nests with cluster analysis using average linkage clustering (Romesburg 1984, Krebs 1999, McGarigal et al. 2000). As suggested by Romesburg (1984), we used the un-weighted pair-group method us- ing arithmetic averages (UPGMA). Statistical Analysis. We used analysis of variance (AN- OVA) to examine relationships between delivery rate var- iables and brood size using log-transformed data (Zar 1999). Biomass of prey delivered per day per nest was transformed by taking the logarithm of biomass delivered per day and adding 1.0 (Zar 1999). Normality of exper- imental error was tested using the Shapiro-Wilk test pro- cedure, and assumptions regarding homogenous varianc- es were tested using Levene’s test (Zar 1999). We examined differences in the number of mammals and birds delivered among nests over 5-d intervals, because of missing data among sampled days, with a Kruskal-Wal- lis single-factor ANOVA (Zar 1999). Because observations within breeding areas were not independent, we exam- ined differences in provisioning rates among breeding areas with multivariate repeated measures ANOVA. We used the General Linear Model (GLM) module of STA- TISTICA (Version 6.0, StatSoft, Inc., Tulsa, OK U.S.A.) for all statistical analyses except calculation of diet over- lap and similarity, for which we used Ecological Meth- odology 6.1 (Exeter Software, Setauket, NYU.S.A.). An alpha level of P = 0.05 was used for all statistical tests, and we present means and standard errors. Results Video Recording and Prey Identification. We in- stalled video monitoring systems at three, five, and seven occupied goshawk nests during the 2000, 2001, and 2002 field seasons, respectively. We placed cameras at nests when nestlings were ca. 8 d old (±1.18; range = 1-18 d). One of the 15 nests failed within 3 d of camera placement and was re- moved from analysis. Due to camera malfunctions, we were only able to collect 16 hr of footage at one of the nests in 2000. We placed a camera at the 2002 nest of the same pair, but pooled data from both years as one nest area for analysis. Thus, our sample of 4801 hr (x = 320 ± 42 hr/nest) of video footage is derived from 13 nesting pairs of gos- hawks. We identified 59 (8.3%) of 711 prey deliveries as being retrievals of cached items. Of the 652 fresh prey deliveries, we identified 451 (69%) to the spe- cies level, 20 (3%) to genus, four to family (1%), and four (1%) as unidentifiable ducklings (Table 1). Eighty (12%) birds and 17 (3%) mammals were unidentifiable beyond class, and we were unable to identify 76 (12%) deliveries. The majority of prey deliveries identified to at least class {N = 576) were mammals (62%), whereas birds (38%) comprised a smaller proportion of diet. When considering only those deliveries identi- fied to family or finer resolution (i.e., to genus or species; N = 476), the dominant prey species were red squirrels (41.2%), eastern chipmunks (19.8%), American crows (7.7%), Ruffed Grouse (6.9%), and snowshoe hares (6.5%). No other individual species accounted for >5% of identified prey. As a group, Sciurids and Leporids {N = 338) accounted for 70% of the identified prey. Among mammals, 51.8% were adults, 25.4% were Juveniles, and we were unable to estimate age for 22.8%. Of the birds, 36.7% were adults, 9.6% were Juveniles, September 2005 Biology 269 27 . 5 % were nestlings, and we could not reliably estimate age for 26 . 2 %. Biomass. In context of the prey species and bio- mass proportion used by goshawks in our study, the delivery of one domestic chicken {Gallus spp.) was unusual and the mass would dramatically influence biomass estimates for avian prey. We therefore con- sidered it an outlier and deleted it from biomass estimates. We estimated the total biomass of all prey deliv- eries at nests as 161 kg. The mean mass for both avian and mammalian prey was 281 g (± 13 . 7 , 95 % confidence interval = 254—308 g). Although aver- age mass of avian prey {x = 292 g; range = 10 - 1082 g) was similar to that for mammalian prey (275 g; range = 18-1361 g), avian prey accounted for only 39 % of biomass delivered whereas mam- mals accounted for 61 % of biomass delivered. Snowshoe hare ( 25 %), red squirrel ( 24 %), Ruffed Grouse ( 11 %), American Crow ( 9 %), diving ducks ( 7 %), chipmunk ( 5 %), and eastern cottontail ( 5 %) accounted for 86 % of biomass used by gos- hawks. No other species accounted for >5% of bio- mass. Delivery Rates. Breeding goshawks delivered 2.12 (± 0 . 14 ) prey per day (i.e., 0.14 deliveries/hr), each delivery had a mean mass of 275 g (±20 g), for a total of 551 g (±50 g) delivered per day. How- ever, daily ( 7 ^ 13,253 = 3 . 44 , P < 0 . 001 ) and hourly ( 7 ^ 13,250 = 2 . 31 , P = 0 . 01 ) delivery rates varied among nests. 1.3 (± 0 . 1 ) prey items were delivered per nestling per day, but delivery rates increased with brood size ( 7 ^ 2,271 “ 5 . 23 , P = 0 . 01 ). Daily prey delivery rates were 1.8 (± 0 . 1 ) at nests with one nestling, 2.3 (± 0 . 1 ) at nests with two nestlings, and 2.5 (± 0 . 2 ) at nests with three nestlings. Despite the increase in prey deliveries among nests with larger broods, there was an inverse relationship between brood size and the number of prey delivered per nestling per day (r = — 0 . 43 , P < 0 . 05 ). Each nestling in single broods received a mean of 1.8 (± 0 . 1 ) prey items per day, whereas each nestling in broods of two received only 1.2 (± 0 . 1 ) prey items per day, and each nestling in broods of three received only 0.9 (± 0 . 1 ) prey items per day. 322 g (±32 g) of biomass were delivered per nestling. However, we observed a pattern of bio- mass delivered to broods of different sizes that was similar to that of number of prey delivered to broods of different sizes; biomass delivered per nestling per day (Eg 5 = 5 . 96 , P = 0 . 038 ) varied with brood size. On average, daily biomass delivered was 509 g (±84 g) to nests with one nestling, 555 g (±42 g) to broods of two, and 756 g (±107 g) to broods of three. Despite greater amounts of bio- mass being provided to larger broods, this resulted in nestlings in single broods receiving 509 g (±84 g) of biomass per day, whereas nestlings in broods of two each received 278 g (±3 g) of biomass per day and nestlings in broods of three each receiving 252 g (±36 g) per day. Dietary Overlap. The diversity and equitability of prey delivered to nests was low for the study area, as indicated by a reciprocal of the Simpson diver- sity index ( 1 / 7 )) of 4.28 and a Smith and Wilson evenness index (i^ar) of 0 . 30 . Similarly, diversity among nests was low, with a mean value of l/D = 3.77 (± 0 . 41 , range = 2 - 09 - 7 . 35 ). The mean value of for all nests was 0.56 (± 0 . 04 , range = 0 . 36 - 0 . 80 ). Low prey diversity and evenness values may be attributable to goshawk diet being dominated by red squirrels and chipmunks in our study. Sim- ilarly, there was high dietary overlap (> 0 . 8 ) among breeding pairs of goshawks in our study (Table 2 ) , although one nesting area (LSP; Table 2 ) ap- peared to be measurably different from the rest. Cluster analysis indicated there were two groups of breeding goshawk diets that exhibited similar prey composition and proportion of use (Fig. 2 ) al- though, again, one nest (LSP; Fig. 2 ) appears to be an outlier. There was no apparent relationship between overlap measures and spatial distribution of nests across the study area (Fig. 1 , 2 ). Discussion Mammals were the dominant prey of breeding goshawks in Minnesota, with red squirrels and east- ern chipmunks appearing to be the most impor- tant species in terms of both number delivered and biomass. These two species alone accounted for 62 % of all prey identified to at least family and 51 % of prey identified to at least class. Several stud- ies have documented red squirrels as important prey for goshawks (Squires and Kennedy 2005 ) throughout their range. They may be especially im- portant during the winter when other prey may be less available (Widen et al. 1987 ). Squirrels domi- nated goshawk diets in Sweden in terms of number ( 79 %) and biomass ( 56 %) during winters of both high and low squirrel abundance (Widen et al. 1987 ). Diet information for winter goshawks in the WGLR is not available, but the extensive use of red squirrels during the summer and the patterns of Table 2. Dietary overlap values using the Simplified Morisita’s Index of Overlap. Values range from 0 (no overlap) to 1 (complete overlap). The data presented were generated from prey frequency data collected at Northern Goshawk nests {N = 13) in Minnesota during the 2000, 2001, and 2002 breeding seasons. The three letter codes designate specific goshawk nests. VOL. 39, No. 3 270 Smithers et al. CO 05 xf5 CO so t- xf5 J> rH 00 xO 00 ’sh rH 05 r-H XO i> O] 05 05 xO 00 05 05 05 05 00 00 o d d d d d d d d d d d < 00 05 I-H 00 00 XO so 05 CM 00 xn 00 cO cO 05 cn 00 00 CJ5 cO 05 05 05 00 05 00 d d d d d d d d d d d CM 05 I— c (M CM 05 05 ^H 50 o so r- 00 o rH CO 00 00 00 05 05 r- 00 CJ5 05 00 00 d d d d d d d d d d 00 so 00 r- C xn o so IC so C05 00 05 CM H CO 05 05 CO C05 05 05 05 (Z) d d d d d d d d d i> 00 05 05 so CO CM f-H 00 00 xO CO 05 XO SO |*q a> 05 05 xq 00 C05 05 03 d d d d d d d d H m 00 05 05 so CJ5 I-H ID o CO CM 05 00 s o> 05 05 05 CO 05 05 Cu d d d d d d d Q U 05 05 so CM so J> so r-- <y> 05 05 05 cO 05 d d d d d d <1 tJD CM r*H r- T-^ CM (Z) cn 00 05 05 CD hJ d d d d d to CM CM CO CM 05 so (Z) CO xq xO CO d d d d CO CM 00 r-f w >-n 00 XO 00 00 05 05 d d d o vr> CO 00 I— 1 00 05 d d X I-H Q d w w p DEE DIX HAG z w LSP LSA MCD PMT SHA STE WAG WFA WRI squirrel use during winter in other areas (Widen et al. 1987) suggest this species may be of year- round importance to goshawks in the region. In terms of biomass, snowshoe hares also appear to be important for goshawks in our study area, ac- counting for 25% of the biomass delivered to nests. Rabbits and hares are also used extensively by gos- hawks throughout their range (Squires and Ken- nedy 2005). Ruffed Grouse comprised 5% of prey deliveries and 11% of biomass delivered to goshawk nests during a 3-yr period of relatively low grouse abun- dance (Smithers 2003). There is anecdotal evi- dence that at least some goshawks in Minnesota may rely more heavily on Ruffed Grouse than oth- er prey during some time periods (Eng and Gul- lion 1962, Apfelbaum and Haney 1984). Eng and Gullion (1962) focused on Ruffed Grouse mortal- ity and did not assess proportional use of grouse in the diet of goshawks, and Apfelbaum and Haney (1984) reported on prey remains collected at a sin- gle nest in northern Minnesota. Because of the dif- ficulties in accurately quantifying the extent of grouse predation by goshawks (Eng and Gullion 1962) and the biases associated with determining raptor diets based on prey remains (Smithers 2003), the results of these studies need to be in- terpreted cautiously. We suspect that the previous research on goshawk diet for our study area, all collected by indirect methods (Eng and Gullion 1962, Apfelbaum and Haney 1984, Martell and Dick 1996), may overestimate the proportion of birds, especially large birds such as grouse, and un- derestimate the proportion of mammals in gos- hawk diets. Qualitative review of the data suggests the mean delivery rate of 0.14 deliveries/hr to nests in our study was less than that observed in Arizona (0.25 deliveries/hr; Boal and Mannan 1994), Nevada (0.31 deliveries/hr; Younk and Bechard 1994) and two areas of southeast Alaska (0.30 and 0.23 deliv- eries/hr; Lewis 2001). However, although mean biomass per delivery in our study (275 g) was less than that in Arizona (307 g/delivery) where Le- porids and Sciurids were the dominant prey (Boal and Mannan 1994), it was greater than the two ar- eas of Alaska (214 g and 173 g/delivery), where birds were the dominant prey (Lewis 2001). Our study indicates that goshawks with larger broods provision with greater delivery rates and biomass. Biomass per nestling was similar between broods of two and three (16.3-18.0 g/hr), but only September 2005 Biology 271 Figure 2. Cluster analysis dendrogram for food habits data collected at Northern Goshawk nests in Minnesota during the 2000, 2001, and 2002 breeding seasons. Parentheses indicate cluster number (see Fig. 1). The LSP and WAG breeding areas exhibited the least similarity of diet composition among breeding areas. about half as much as that received by nestlings in broods of one (33.0 g/hr). This poses an interest- ing question regarding energetic aspects of gos- hawk productivity; what is the minimum biomass/ hr necessary to fledge young successfully? The similarity between broods of two and broods of three suggests that, at least in our study area, and at nests with similar prey composition, a minimum of 16—18 g of biomass per hr may be required for successful nesting. However, a finer assessment of nestling energetics would likely require experi- mentation in a laboratory setting. Given our prey use and delivery rate data, one can make a generalized prediction of the relative impact of a breeding pair of goshawks in our study area during the 45-d nestling period. With an ex- pected delivery rate of 2.1 prey/ d over a 45-d nest- ling period, ca. 94 prey deliveries can be expected. Based on observed frequencies of prey use, this would translate to the average breeding goshawk pair capturing 29 red squirrels, 14 eastern chip- munks, six American Crows, five snowshoe hares, five Ruffed Grouse, two diving ducks, one cotton- tail, one Blue Jay, and 31 miscellaneous small birds and mammals. To put this level of predation in context, all of these prey captures would occur within a home range of 6376 ha for a goshawk pair in the study area (Boal et al. 2003) . Composition and richness of prey delivered to nests was similar across the study area, and esti- mates of prey diversity and equitability were gen- erally low among nests. We suspect the high dietary overlap and similarity of prey use among breeding areas was most likely attributable to the dominance of red squirrels and chipmunks in goshawk diets. However, goshawk diets were dominated by red 272 Smithers et al. VoL. 39, No. 3 squirrels and chipmunks, but snowshoe hare, Ruffed Grouse, and American Crow were also im- portant in terms of biomass. As pointed out by Reynolds et al. (1992), raptor populations are often limited by prey availability and their choice of foraging habitat is predicated on conditions in which prey are abundant and available. Thus, an understanding of goshawk prey species used and the relative importance of those prey species is an important step toward develop- ing management plans for goshawks. By identifying key prey species, as we have done here, forest man- agers can develop a set of desirable conditions that fosters presence of those species while incorporat- ing structural aspects of known goshawk foraging habitat (e.g., Boal et al. 2001). Those desirable for- est conditions can be incorporated into goshawk management plans as one factor of foraging habi- tat (e.g., Reynolds et al. 1992) and facilitate con- servation of the species. Acknowledgments A. Bellman, F. Nicoletti, J. Ridelbauer, A. Roberson, M. Solensky, L. Smithers, W. Steffans, C. Trembath, and A. Wester assisted with various aspects of this study. Person- nel from the many cooperating agencies and organiza- tions provided assistance and logistical support during this project. They included R. Baker, J. Casson, J. Gal- lagher, L. Grover, M. Hamady, J. Hines, M. Houser, E. Lindquist, C. Mortensen, S. Mortensen, B. Ohlander, W. Russ, R. Vora, and A. Williamson. Video equipment for the 2000 and 2001 field seasons was provided by Alaska Department of Fish and Game and the Wisconsin De- partment of Natural Resources. Funding and logistical support for this project was provided by the Minnesota Department of Natural Resources, Minnesota Forest In- dustries, The National Council for Air and Steam Im- provement, Potlatch Corporation, the U.S. Forest Service Chippewa and Superior National Forests, and the U.S. Fish and Wildlife Service. Literature Cited Apfelbaum, S.I. and A. Haney. 1984. Note on foraging and nesting habitats of goshawks. Loon 56:132-133. Beier, P. and J.E. Drennan. 1997. Forest structure and prey abundance in foraging areas of Northern Gos- hawks. Ecol. Appl. 7:564—571. Bielefeldt, J., R.N. Rosenfield, and J.M. Papp. 1992. Un- founded assumptions about diet of the Cooper’s Hawk. Condor 94:427-436. Boal, C.W., D.E. Andersen, and P.L. Kennedy. 2001. Home range and habitat use of Northern Goshawks in Minnesota. Final Report. Minnesota Cooperative Fish and Wildlife Research Unit, University of Min- nesota, St. Paul, MN U.S.A. , , AND . 2003. Home range and res- idency status of Northern Goshawks breeding in Min- nesota. Condor 105:811-816. AND R.W. Mannan. 1994. Northern Goshawk diets in ponderosa pine forests on the Kaibab Plateau. Stud. Avian Biol. 16:97-102. Bosakowski, T. AND D.G. Smith. 1992. Comparative diets of sympatric nesting raptors in the eastern deciduous forest biome. Can. J. Zool. 70:984-992. Bright-Smith, D.J. and R.W. Mannan. 1994. Habitat use by breeding male Northern Goshawks in Northern Arizona. Stud. Avian Biol. 16:58—65. Burt, WH. and R.P. Grossenheider, 1980. A field guide to the mammals. 3rd Ed. Houghton Mifflin Company, Boston, MA U.S.A. COLLOPY, M.W. 1983. A comparison of direct observation and collections of prey remains in determining the diet of Golden Eagles./. Wildl. Manag. 47:360-368. Daniel, G. and Sullivan, J. 1981. A Sierra Club natural- ist’s guide to the north woods of Michigan, Wisconsin, Minnesota and southern Ontario. Sierra Club Books, San Franciso, CA U.S.A. Dunn, J.L. 1999. Field guide to the birds of North Amer- ica, 3rd Ed. National Geographic Society, Mary B. Dickinson, Ed., Washington, DC U.S.A. AND K. Garrett. 1997. A field guide to warblers of North America. Houghton Mifflin Company, Bos- ton, MA U.S.A. Dunning, J.B., Jr. 1993. CRC handbook of avian body masses. CRC Press, Inc., Boca Raton, FL U.S.A. Eng, R.L. and G.W. Gullion. 1962. The predation of goshawks upon Ruffed Grouse on the Cloquet Forest Research Center, Minnesota. Wilson Bull. 74:227-241. Grzybowski, J.A. AND S.W. Eaton. 1976. Prey items of goshawks in southwestern New York. Wilson Bull. 88: 669-670. Jones, J.K., Jr. and E.C. Birney. 1988. Handbook of mam- mals of the north-central states. Univ. Minnesota Press, Minneapolis, MN U.S.A. Kennedy, P.L. and D.R. Johnson. 1986. Prey size selec- tion in nestling male and female Cooper’s Hawks. Wil- son Bull. 98:110-115. Krebs, C.J. 1999. Ecological methodology, 2nd Ed. Ad- dison-Wesley Educational Publishers, Inc., Reading, MA U.S.A. Lawlor, L.R. 1980. Overlap, similarity and competition coefficients. Ecology 61:245-251. Lewis, S.B. 2001. Breeding season diet of Northern Gos- hawks in southeast Alaska with a comparison of tech- niques used to examine raptor diet. M.S. thesis, Boise State Univ., Boise, ID U.S.A. McGarigal, K., S. Cushman, and S. Stafford. 2000. Mul- tivariate statistics for wildlife and ecology research. Springer, New York, NY U.S. A. MacArthur, R.H. 1972. Geographical ecology. Harper and Row, New York, NY U.S.A. Marti, C.D. 1987. Raptor food habits studies. Pages 67- 80 in B.A. Giron Pendleton, B.A. Millsap, K.W. Cline, September 2005 Biology 273 and D.M. Bird [Eds.], Raptor management tech- niques manual. National Wildlife Federation, Wash- ington, DC U.S.A. Martell, M. and T. Dick. 1996. Nesting habitat charac- teristics of the Northern Goshawk {Acdpiter gentilis) in Minnesota. Final Report to Minnesota Department of Natural Resources, Nongame Wildlife Program, Proj- ect No. 9407382., Minnesota Department of Natural Resources, St. Paul, MN U.S.A. Meng, H. 1959. Food habits of nesting Cooper’s Hawks and goshawks in New York and Pennsylvania. Wilson Bull. 71:169-174. Reynolds, R.T. and E.C. Meslow. 1984. Partitioning of food and niche characteristics of coexisting Acdpiter during breeding. Auk 101:761-779. , R.T. Graham, M.H. Reiser, R.L. Bassett, P.L. Kennedy, D.A. Boyce, Jr., G. Goodwin, R. Smith, and E.L. Fisher. 1992. Management recommendations for the Northern Goshawk in the southwestern United States. USDA Forest Service, Gen. Tech. Rep. RM-217, Ft. Collins, CO U.S.A. Roberson, A.M. 2001. Evaluating and developing survey techniques using broadcast conspecific calls for Northern Goshawks in Minnesota. M.S. thesis, Univ. Minnesota, St. Paul, MN U.S.A. , D.E. Andersen, and P.L. Kennedy. 2003. The Northern Goshawk {Acdpiter gentilis atricapillus) in the western Great Lakes region: a technical conservation assessment. Minnesota Cooperative Fish and Wildlife Research Unit, Univ. Minnesota, St. Paul, MN U.S.A. Romesburg, H.C. 1984. Cluster analysis for researchers. Lifetime Learning Publications, Belmont, CA U.S.A. Sibley, D.A. 2000. The Sibley guide to birds. National Audubon Society, Alfred A. Knopf, Inc., New York, NY U.S.A. SiDERS, M.S. and P.L. Kennedy. 1996. Forest structural characteristics of accipiter nesting habitat; is there an allometric relationship? ContJor 98:123-132. Simpson, E.H. 1949. Measurement of diversity. Nature 163:688. Smith, B. and J.B. Wilson. 1996. A consumer’s guide to evenness indices. Oikos 76:70-82. Smithers, B.L. 2003. Northern Goshawk food habits in Minnesota: an analysis using time-lapse recording sys- tems. M.S. thesis, Texas Tech University, Lubbock, TX U.S.A. Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- hawk {Acdpiter gentilis) . In A. Poole and F. Gill [Eds.], The birds of North America, No. 298. The Birds of North America Inc., Philadelphia, PA U.S.A. AND P.L. Kennedy 2005. Northern Goshawk ecol- ogy: an assessment of current knowledge and infor- mation needs for conservation and management. Stud. Avian Biol. 31:in press. StatSoft, Inc. 2001. STATISTICA; the small book. StatSoft, Inc., Tulsa, OK U.S.A. Stoker, R.W. 1966. Sexual dimorphism and food habits in three North American accipiters. Auk 83:423-246. Tester, J.R. 1995. Minnesota’s natural heritage: an eco- logical perspective. Univ. Minnesota, Minneapolis, MN U.S.A. Widen, R, H. Andren, P. Angelstam, and E. Lindstrom. 1987. The effect of prey vulnerability: goshawk pre- dation and population fluctuations of small game. Oi- kos 49:233-235. Williams, C.B. 1964. Patterns in the balance of nature. Academic Press, London, U.K. Younk, J.V. and MJ. Bechard. 1994. Breeding ecology of the Northern Goshawk in high-elevation aspen forest of northern Nevada. Stud. Avian Biol. 16:119-121. Zar, J.H. 1999. Biostatistical analysis, 4th Ed. Prentice- Hall, Inc., Englewood Cliffs, NJ U.S.A. Received 22 March 2004; accepted 28 September 2004 Guest Editor; Stephen DeStefano J. Raptor Res. 39(3):274-285 © 2005 The Raptor Research Foundation, Inc. SAMPLING CONSIDERATIONS FOR DEMOGRAPHIC AND HABITAT STUDIES OF NORTHERN GOSHAWKS Richard T. Reynolds, ^ J. David Wiens, Suzanne M. Joy, and Susan R. Salafsky Rocky Mountain Research Station, 2150 Centre Avenue, Building A, Suite 350, Fort Collins, CO 80526-1891 U.S.A. Abstract. — ^We used mark-recapture methods to monitor Northern Goshawks {Accipiter gentilis) and their nests over 12 yr in an increasing sample of breeding territories (37 in 1991 to 121 in 2002) in northern Arizona. As many as 8 yr of repeated nest searching were required to identify the population of breeders, as individuals skipped egg-laying on territories for up to 7 consecutive yr. Extensive temporal (within territory) and spatial (among territory) variation in reproduction and a high annual frequency of movements among dispersed alternate nests in territories made finding and monitoring goshawks problematic. Low detectability of nonbreeding goshawks (combined with uncertainties stemming from variations in breeding and use of alternate nests) made it difficult to categorize territories unequivocally as “unoccupied” by goshawks in non-egg-laying years. Temporal and spatial variation in reproduction required large numbers of territories to attain reliable estimates of reproduction of goshawks; such estimates were achieved only when samples approached or exceeded 60—100 territories. Our within- territory goshawk searching protocol, designed to increase the low and variable detectability of goshawks, required extensive sampling efforts to detect among-alternate nests movements. In lieu of such efforts, samples of territories occupied by goshawks may “decay” over time and lead to false inferences of population declines. Low detectability, variations in breeding, and large samples require that demo- graphic and habitat studies of goshawks employ intensive and repeated searches for goshawks in large study areas over at least 8 yr. Key Words: Northern Goshawk, Accipiter gentilis; alternate nest, Arizona; bootstrap; detectability; monitoring, reproduction; sampling, territory occupancy. GONSIDERACIONES SOBRE EL MUESTREO EN ESTUDIOS DEMOGRAFICOS Y DE HABITAT DE ACCIPITER GENTILIS Resumen. — Usamos tecnicas de captura-recaptura para evaluar las actividades de Accipiter gentilis y dc sus nidos a lo largo de 12 ahos en una muestra creciente de territorios de nidificacion (37 en 1991 a 121 en 2002) en el norte de Arizona. Para poder identificar la poblacion reproductiva de Accipiter gentilis, requerimos hasta 8 ahos de busqueda repetida de nidos, ya que esta especie evito poner huevos en territorios por periodos de hasta 7 ahos consecutivos. La gran variacion temporal (dentro de los terri- torios) y espacial (entre territorios) en la reproduccion y una alta frecuencia anual de movimientos entre nidos alternos disperses en los territorios dificulto encontrar y evaluar las actividades de A. gentilis. La baja detectabilidad de los individuos no-reproductivos de A. gentilis (combinado con incertidumbres provenientes de las variaciones en la reproduccion y en el uso alterno de nidos) hizo dificil categorizar los territorios de modo inequivoco como “desocupados” por A. gentilis en los ahos en que no pusieron huevos. Esta dificultad se manifesto por la presencia de nidos activos de los mismos individuos anillados de A. gentilis luego de mas de un aho sin presencia reproductiva en los territorios. La variacion temporal y espacial en la reproduccion requirio grandes numeros de territorios para alcanzar estimaciones con- fiables de reproduccion en A. gentilis. Estas estimaciones fueron obtenidas solo cuando las muestras alcanzaron o excedieron los 60-100 territorios. Nuestro protocolo de busqueda de A. gentilis denXxo de los territorios, disehado para incrementar la detectabilidad baja y variable de A. gentilis, requirio es- fuerzos de muestreo amplios para detectar movimientos entre nidos alternos. Sin estos esfuerzos, las muestras de los territorios ocupados por A. gentilis podrian “disminuir” a lo largo del tiempo y llevarnos a inferencias falsas sobre disminuciones poblacionales. La baja detectabilidad, las variaciones reprod- uctivas y la necesidad de muestras de gran tamaho requieren que los estudios demograficos y de habitat ^ E-mail address: rreynolds@fs.fed.us 274 September 2005 Techniques 275 de A. gentilis empleen busquedas intensivas y repetidas de esta especie en grandes areas de estudio durante al menos ocho anos. [Traduccion del equipo editorial] The distribution, abundance, vital rates, and habitat occupancy of Northern Goshawks (Accipiter gentilis) are difficult to determine because of their elusive behavior in structurally-complex habitats, their low breeding densities, and annually variable breeding rates (DeStefano et al. 1994, Reynolds et al. 1994, Kennedy 1997, Reynolds and Joy in press). While locating and monitoring nests are common approaches in studies of avian demography and habitat, making valid inferences to a target popu- lation depends on reliable (unbiased and precise) estimates of the distribution and abundance of nests, demographic rates at nests, and habitat oc- cupancy. In such studies, it is often too costly to detect all individuals and to sample all areas, mak- ing a census (complete count) impractical, espe- cially in difficult-to-detect species. For such species, population parameters and habitat occupancy are often estimated using sampling methods. Making inferences about a species’ distribution or habitat occupancy from samples requires inferences about the species’ detection probability (probability that an individual is included in a sample when pres- ent). Biologists attempt to minimize influences of incomplete observations on estimates of a species’ distribution, demographics, and habitats with sam- pling frameworks that increase the detection rates of the species (Peterson and Bayley 2004, Mc- Donald 2004) . The problem is to understand how detectability varies within and among individuals, both temporally (year-to-year) and spatially (among territories) , and to develop sampling pro- tocols and efforts that increase detection rates of all individuals. We used mark-recapture methods from 1991- 2002 to determine the distribution, abundance, vi- tal rates, fidelity to mate and territory, natal and breeding dispersal, and habitat occupancy of gos- hawks breeding on territories that increased in number from 37 in 1991 to 121 in 2002 (Reynolds et al. 1994, Reich et al. 2004, Reynolds et al. 2004, Wiens 2004). Because these objectives required a census of breeding goshawks, we attempted to find all breeding goshawks in our study area. In this paper, we first describe the sampling protocols we used to initially locate and monitor breeding gos- hawks on the Kaibab Plateau. We then describe the abundance and dispersion of breeding territories, the dispersion of alternate nests within territories, reproductive rates, and behaviors effecting gos- hawk detectability that resulted from 12 yr of im- plementing our protocols. Finally, we present boot- strap subsampling of our full samples of territories to estimate the number of breeding territories needed for precise estimates of the reproductive status and production of young by goshawks. Our purpose is to provide a framework for developing sampling protocols, sampling efforts, and sample sizes for demographic and habitat studies of gos- hawks in other populations. Study Area The study area (1728 km^) was all of the E^ibab Pla- teau above 2182 m elevation above sea level, and con- tained ca. 122 400 ha of ponderosa pine {Pinus ponderosa) forests between 2075-2450 m elevation, ca. 51 600 ha of mixed-conifer forests between 2450-2650 m elevation, and ca. 30 600 ha of spruce {Picea engelmannii) fir {Abies lasiocarpa) forests between 2650-2800 m elevation (Ras- mussen 1941, White and Vankat 1993). Pinyon {Pinus ed- ulis) juniper {Juniperus spp.) woodlands occurred below the study area between 1830-2075 m elevation and shrub-steppe occurred below 1830 m. With the exception of several narrow (<1 km) meadows, several areas burned by wildfire, and numerous tree harvest areas, for- ests on the study area were contiguous (Reynolds et al. 1994, Joy et al. 2003). The southern one-third of the study area included the Grand Canyon National Park- North Rim (GCNP) and the northern two-thirds, the E.ai- bab National Forest (KNF) . Forests on the Kaibab Plateau are isolated from other forests by varying distances of shrub-steppe; the nearest forest to the north, 97 km; to the east, 250 km; to the west, 80 km; and to the south, 89 km, with the exception of a small area of ponderosa pine forest on the south rim of the Grand Canyon at 18 km (Reynolds et al. 2004). Methods Field Procedures. We defined a breeding territory as an area exclusively occupied by a pair of goshawks during a breeding season. This definition implied that territories were defended by resident goshawks, and the dispersion of breeding pairs was constrained by territoriality. While uncertain if or how territories were defended by gos- hawks, we estimated their size on the Kaibab Plateau as the area whose radius was half the mean distance among neighboring pairs. Recapture of marked goshawks showed that territorial owners, as well as their replace- ments over time, had strong life-time fidelity to their ter- ritory (Reynolds and Joy in press, R. Reynolds unpubl. data) , and territories on the Kaibab Plateau appeared to be spatially fixed over years. 276 Reynolds et al. VoL. 39, No. 3 We located goshawk territories using two protocols: sys- tematic foot-searches for goshawks and their nests in ar- eas <1600 ha and broadcasts of goshawk vocalizations from stations on transects (Kennedy and Stahlecker 1993, Joy et al. 1994) in areas >2400 ha. Both nest- searching procedures were used each breeding season (April-August). A new territory was identified when a used goshawk nest (or, in rare cases, an occupied-only nest area; see below) was discovered in an area not al- ready in a known territory and when the new nest (or nest area) was used by unbanded goshawks. Once a ter- ritory was found, it was added to that year’s cohort of territories and assessed in all subsequent years for gos- hawk occupancy. Because we were unable to search our study area completely in a single year, we extended our nest searching into previously unsearched areas each year; hence, the number of territories under study in- creased over years. In addition to expanded nest search- es, we annually re-searched areas (using both foot and broadcast searches) suspected of having territorial gos- hawks based on goshawk nest spacing (Reynolds and Wight 1978, Reynolds et al. 1994). Goshawk territories often contain one or more alter- nate nests that are used by the goshawks over several years (Squires and Reynolds 1997, Reynolds and Joy in press). To prevent misclassifying the reproductive status of goshawks that may have moved to an alternate nest, we used a within-territory nest-searching protocol con- sisting of three sequential steps (Reynolds et al. 2004). Each year, beginning 3 wk before egg laying, we con- ducted “initial visits” to all known alternate nests and historical nest structures (existing nests with unknown histories of use) to determine if goshawks were present. Searches for goshawks, their feces, molted feathers, and nests refurbished with green twigs (Reynolds and Wight 1982) were conducted within 100-m radii of each alter- nate and historical nest. Initial visits to nests were com- pleted in all territories by 2 wk after egg-laying. If a used nest in a territory was not found in an initial visit, a “foot search” was conducted within a 500-m radius circle cen- tered on the last-used nest or the centroid of the territory (determined subsequent to discovery of >1 alternate nests in a territory). Territory centroids were the geo- metric means of coordinates of alternate nests weighted by the number of yr each alternate nest was used during our study (Reynolds et al. 2004, Reynolds and Joy in press) . A foot search involved systematically walking the 500-m radius circle looking for goshawks or signs of their presence (see above). Foot searches were conducted from egg laying to about 15—20 d after egg-hatching. In territories where used nests were not located in foot searches, a “broadcast search” was conducted in a 1600- m radius circle centered on either the last-used nest or the territory centroid. Broadcasting of goshawk vocaliza- tions were conducted from stations on transects arranged as described by Joy et al. (1994). Broadcast searches were conducted from about 10 d after egg hatching to the end of the post-fledging dependency period (late August or early September). All nest trees were mapped to the nearest 3 m with a global positioning system. Nests were “used” if goshawks laid eggs, and territories were “occupied-only” if eggs were not laid but evidence (goshawks observed, molted feathers, feces, reconstruct- ed nest) of goshawk presence was found in association with a nest structure, or “unknown” if insufficient evi- dence of occupancy was found. All used nests were visited weekly to count numbers of nestlings and fledglings and to determine the approximate timing and causes of nest failure. Goshawk nestlings were banded in the 10 d be- fore fledgling, and numbers of nestlings present at the time of banding was considered the number of young produced. Nesting adults were captured with dho-gaza nets placed in nest areas and baited with live Great Horned Owls {Bubo virginianus) during the nestling pe- riod (Reynolds et al. 1994). All goshawks received a U.S Geological Survey leg band and a colored-aluminum leg band with a unique alpha-numeric code readable from 80 ra with 40-60X telescopes (Reynolds et al. 1994). An- nual field efforts of crews consisting of 15-23 persons were focused on finding new territories, finding nests within known territories, and capturing and recapturing (or resighting) goshawks on the study area. Data Analysis. We used Dirichlet tessellation and De- launey triangulation (Cressie 1991) to estimate the dis- tances between the centroids of first-order neighboring goshawk territories. To estimate the dispersion of alter- nate nests within territories, we measured the within-ter- ritory map distances between each alternate nest (inter- alternate nest distance) and the within-territory centroid to each alternate nest (centroid-to-alternate nest dis- tance; excludes territories with only one nest). To test for differences in the spacing of goshawk territories in the KNF and the GCNP, we used a two-sample ^-test. To char- acterize the strength of the relationship between the numbers of new territories found in a year and the pro- portion of territories used in a year, we used a Spear- man’s correlation coefficient (rj. The annual proportion of territories with used nests was calculated as the pro- portion of those territories under study in the previous year (prior-year’s cohort of territories) that had used nests in the current year (Reynolds and Joy in press). We did this because the number of territories under study increased annually, and we included only territories that were monitored from before egg-laying to minimize bias associated with missed failed nests. We defined nest suc- cess as the proportion of used nests in a prior year’s co- hort of territories that produced >1 fledgling. To ex- amine annual differences between the proportion of territories with used nests and the mean number of young produced per used nest, we calculated 95% Con- fidence Intervals (Cl) for these parameters and visually assessed the degree of Cl overlap among estimates. We used the bootstrap method (Efron and Tibshiram 1993) to estimate the number of goshawk territories that needed monitoring to attain precise estimates of the pro- portion of territories with used nests, nesting success, and number of young fledged per used nest. Our objective was to display variability in these parameters for different- sized samples given the full sample estimate. We con- ducted, with replacement, 1000 bootstrap iterations with sample sizes of 20, 40, 60, 80, 100, and 120 territories. We present the bootstrap results in box and whisker plots for only 2000 and 2002 because numbers of territories under study during those years were similar (120 and 121), and 2000 was a relatively good breeding year (55% of territories had used nests) , while 2002 was a relatively September 2005 Techniques 277 Table 1. Total territories, number of used nests (eggs laid), and number and percent of territories from previous year’s cohort of territories with used nests on the Kaibab Plateau, Arizona, 1991-2002. Previous-year’s territory cohorts were used because all territories in that cohort were monitored from before egg-laying in a current year, minimizing bias created by missing used nests due to early nest failure. Year 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Total territories 37 64 82 88 99 105 106 109 113 120 120 121 Total used nests Used nests in 36 59 67 21 53 46 31 58 57 66 30 21 previous year’s cohort 32 49 18 44 40 31 55 56 60 30 21 Percent used in cohort 86 77 22 50 40 30 52 51 53 25 17 poor breeding year (17% of territories had used nests). We plotted the medians of the estimates of the bootstrap subsampling for the proportions of territories with used nests and for nesting success. For mean young fledged per territory, we plotted the medians and coefficients of variation of the bootstrap estimates. We used a CV of 20% as a target criterion for identifying a level of sampling needed to attain adequate estimates of numbers of young per territory (see Pollock et al. 1990). Results The Study Population. Numbers of goshawk ter- ritories under study increased from 37 in 1991 to 121 in 2002 as searches for new territories were annually extended into unsearched areas and as previously searched, but unoccupied areas, were searched again (Table 1). By 2002, about 95% of the KNF and 60% of the GCNP had been searched for nests. A total of 121 territories were discovered, and goshawks laid eggs in 1 or more years on all but six of these. Exceptions (two KNF, four GCNP territories) included territories occupied in ^2 yr by goshawks that built new, or reconstructed old, nests but did not lay eggs during the study. Terri- tory centroids were regularly spaced (Reich et al. 2004, Reynolds and Joy in press) . The mean De- launay triangle distance between 120 territory cen- troids (1 territory not included due to inadequate search for surrounding territories) was 3.8 km (SD = 1.3 km, min = 1.3 km, max = 8.1 km, N— 302 first-order neighbor distances; inter-centroid dis- tances that crossed unsearched areas in the ex- treme southeast of the study area were not includ- ed; Fig. 1). We estimated the total number of breeding ter- ritories on the study area by calculating an “exclu- sive” area for each pair of goshawks using one-half the mean distance between territory centroids (3.8 km) as the radius and dividing the study area (173 200 ha) by the exclusive area (1134 ha; Reyn- olds and Joy in press). This should result in an accurate estimate of the total number of territories because of the regular spacing of territories (known for 80% of our study area) and because forests on the study area were nearly contiguous (Reynolds and Joy in press). The study area was large enough for there to be approximately 150 territories, five territories more than our 1996 es- timate (Reynolds and Joy in press) . This increase reflected the discovery of 17 new territories be- tween 1997 and 2002 and a subsequent 0.1 km re- duction in the mean inter-centroid distance. Therefore, our sample of 121 known territories represented about 80% of the potential total num- ber of goshawk territories in our study area. Temporal and spatial variation in the frequency of egg-laying by goshawks on the study area was extensive. Temporal variation reflected periods of years with increasing or decreasing proportions of goshawks that laid eggs (Table 1, Fig. 2), and spa- tial variation reflected differences in the frequen- cies of egg-laying among territories (Table 2). In the 12 yr during which the 37 territories in the 1991 cohort were monitored, 13 territories (36%) had used nests in <6 yr and 23 (64%) had used nests in >7 yr (1 territory never had a used nest), and of the 27 new territories found in 1992, 17 (63%) had used nests in ^5 yr and 10 (37%) had eggs in >6 of the 11 yr they were monitored (Table 2). Overall, 75% (86 of 115 territories that had used nests in ^1 yr) of territories had used nests in ^3 yr. Most (87%) territories in which egg lay- ing was skipped in ^1 yr had used nests or were occupied-only in subsequent years, often by the 278 Reynolds et al. VoL. 39, No. 3 Grand Canyon National Park /V: : f • • A Jl.A Thiessen Polygons Delaunay Triangles Territory Centroids # i- • - . _ 1» ‘ mr '‘ > * « ■w • a N Kaibab k '•. * • National Forest ^ r ^ 10 20 I km Figure 1. Thiessen polygons and Delaunay triangles used to calculate first-order nearest neighbor distances between Northern Goshawk territory centroids on the Kaibab Plateau, Arizona, 1991-2002 {N = 120, see text). Mean inter- centroid distance was 3.8 km (SD =1.3 km, min = 1.3 km, max = 8.2 km, N = 302 triangle legs). same banded goshawks that had previously laid eggs on the territory (R. Reynolds unpubl. data) . Of a combined total of 435 used nests in all 11 prior-year cohorts of territories, 341 (63%) were successful (Table 3). Of 94 nest failures, 59 (63%) failed during the incubation period, and 35 (37%) failed during the nestling period. There was mini- mal among-year variation in nesting success (Table 3, Fig. 2). In 459 broods (not limited to nests in prior-year cohorts) with accurate counts of young, brood sizes ranged from 1-4 nestlings (median = 2; Table 2); 102 (22%) broods had one young, 219 (48%) had two young, 133 (29%) had three young, and five (1%) had four young. The mean annual number of fledglings produced per used nest was only moderately variable compared to the annual variation in the proportion of territories with used nests; the CV of the number of young produced September 2005 Techniques Year Figure 2. Annual variation in (a) the proportion of ter- ritories under study containing active (eggs laid) North- ern Goshawk nests and (b) in the mean numbers of young produced per used nest in the previous year’s co- hort of goshawk nests (see text) on the Kaibab Plateau, Arizona, 1991-2002. Error bars represent ±95% Cl. per used nest was 28%, while the CV of the pro- portion of territories with used nests was 114% (Fig. 2) . Likewise, the among-year variation in total young produced by the 1991 cohort of territories {N = 37) over 12 yr was also higher than the among-year variation of the means of young pro- duced per used nest for the same territories and years. Total young produced ranged from 16 in 2002 to 65 in 1992 and had an among-year coeffi- cient of variation (CV) of 68%, and mean number of young produced per used nest ranged from 0.6 in 2002 to 2.4 in 2000 and had a CV of 37%. Thus, both the annual proportion of territories with used nests and total young produced per year provide a more sensitive measure of the variable reproduc- tive output of goshawks than the annual mean number of young produced per used nest. Goshawk Behavior and Sources of Error. How well an estimate represents the true spatial distri- bution, density, or habitat occupancy of a species a 1 1 1— H 00 i-H rH rH o S' 1 1 CM 1 1 CM cn CM 1 ■' — ^ 1 00 rH H CTl o> y«— V r V 00 1 1 o' 5 rH rH 8 s rH I— ( rH rH 1 1 1 1 CO CO o u IZ) D ^ ^ kO o CO rH CD m 1 ^ ' 1 CM rH oo 1 0 V OO 00 1 rH oo CV 1 1 1 CV CV CM CO ^ o CQ § P CV CO o 1 CV CV 1 1 MO CD m OO in X V. 00 on t-H rH CD 1 rH 1 ' — ^ '•' — ^ ' 1 ^ 1 CO rH CV ,, — ^ rH s CM 00 rH rH 1 oo ‘ — ■ ' 1 ' — ' CO in CV rH CM s CV 0? 1 CV CV 1— ( 1 '' — ^ CM i-H CO I-H c/i 2 Q 0 .0 w H p <o j> 00 to ^H to 2 3 0 CO CV ^H rH 3 Ph H cd ?— 1 CV CO m to 02 Oi 02 Oi 02 02 <Ji 02 O) 02 t-h rH rH V u biD g 'C 2 3 o 6 bo bo u <ri (-M o o (J tu Xi "0 I o a CV o o CV O) u a; £ c 0 1 & 2 a u u o Vi u 2 u <v o Table 3. Number of used (eggs laid) and successful (fledged >1 young) nests, mean and standard deviation (SD) of fledglings per used nest, and proportion of used nests within the prior year’s cohort of Northern Goshawk territories that fledged young on the Kaibab Plateau, Arizona, 1991-2002. 280 Reynolds et al. VoL. 39, No. 3 (M O o CM o o o o o CM CJi CTS 00 C7i CJi as CO CJi as CJS os TP as CJS vn CJS os CM CJS CJS (M O CO o CO CO lO If) o 00 o tP tP C20 O) CO 00 00 d d so Csf O os 00 J o c^i 1— I 00 If) CM q CM CO CO tP so sf) oq i—i d so CM 00 OS rH O oO CM 00 Os O tP OO CM !-h Os I— I d CO Tp 00 q 1 — i ,-H CV tP (M CM SO OO CO 00 00 00 tP 00 00 00 00 CM CO X CM CM X CO CM iS w 4-1 1) G (/) W cA t/3 bo V 3 V 1) c Cm 1/3 (/3 w c Ti V V3 1 — , <u c/3 Q U 3 <u OJ p X X (1^ X - 5 V u y :3 c/3 depends on the error associated with the estimate (Thompson et al. 1998). A potential source of sam- pling variation is an incomplete count of breeding goshawks. Counts are related to the actual size of the territorial goshawk population by the proba- bility of detection, which may vary systematically. Because of their defensive behavior at nests, the detectability of breeding goshawks (still relatively low due to their elusiveness and complex habitats) is much higher than the detectability of non-nest- ing goshawks (including those whose nests failed). Low detectability of nonbreeders combined with the large annual variation in numbers of goshawks breeding can produce large sampling variation. To account for the low detectability of nonbreeders, we repeatedly searched areas suspected of contain- ing goshawks. Eleven yr of repeated searching for nests showed that the KNF was nearly saturated with breeding territories (Fig. 1 ) . We do not know if the GCNP was similarly saturated with territories because only ca. 60% of the GCNP was searched for goshawks. However, the mean distance between centroids of known territories in the GCNP was not significantly different from the mean distance be- tween KNF centroids (KNF x = 3722 m, SD - 1221 m, N — 271 triangle legs; GCNP x = 4028 m, SD -- 1477 m, N = 22 triangle legs; t — —1.1, df = 291, P = 0.27), suggesting that the density of gos- hawk territories in the GCNP was similar to terri- tory density in the KNF. The success of finding new territories in a year was positively related to the proportion of territories with used nests in that year (r^ = 0.73, P = 0.011, N= 11); we found more new territories in good breeding years (1991-93, 1998-2000) than in poor breeding years (1994, 2002 ). Another source of measurement error is mis- classification of the breeding status of territorial goshawks. Detecting movements of goshawks among alternate nests required considerable sam- pling effort, the level of which depends on the number and distribution of alternates within ter- ritories and frequency of movement among the al- ternates. Because numbers of known alternates de- pends on years of monitoring, we only report the numbers of alternate nests in the 1991 and 1992 cohorts of territories. Mean number of alternates in these territories was 3.2 (SD = 1.5 nests, min = 1, max = 6, A = 36) and 2.9 (SD = 1.4 nests, min = 1, max = 6, A= 27), respectively. The frequency distribution of inter-alternate nest distances within all territories with >2 alternates {N = 91 territo- September 2005 Techniques 281 Distance (m) Figure 3. Cumulative percent of alternate nests within territories with increasing distance (m) between territory centroids to alternate nests on the Kaibab Plateau, Ari- zona. Maximum centroid-to-alternate nest distance was 1452 m. ries) was right-skewed with a median of 402 m (x = 612 m, SD = 569 m, min = 9 m, max = 2426 m, = 308 alternate nests) . When measured from territory centroids, the median centroid-to-alter- nate distance was 228 m (x = 334 m, SD = 298 m, min = 6 m, max = 1452 m, = 273 alternates in 91 territories), about half of the median inter-al- ternate nest distance. The cumulative proportion of alternates captured with distance from centroids showed that about 75% occurred within 0.5 km, and about 95% occurred within 1 km of centroids (Fig. 3). Thus, our territory-focused broadcast searching protocol in areas of 1.6-km radius around centroids exceeded the maximum known centroid-to-alternate distance (1452 m). The fre- quency of movement of egg-laying goshawks to al- ternates was high; an annual mean of 64% of breeding goshawks moved to an alternate, and 42% of these movements were to new (unknown to us) alternates (Table 4). Sample Size. Bootstrap subsampling showed that samples of ca. 60—80 territories in good breeding years and 80-100 territories in poor breeding years were needed for precise estimates of the full sam- ple means of the proportion of territories with used nests and nesting success on the Kaibab Pla- teau (Fig. 4). Coefficient of Variation plots of the mean young per used nest in good breeding years showed that subsamples of >80 territories had bootstrap estimates entirely below a CV of 20%, although many of the estimates from subsamples of 60 territories were below 20% (Fig. 5) . In poor breeding years, subsamples of 100 territories were insufficient to achieve a CV of less than 20%, re- flecting the few (21) territories that were occcu- pied in 2002. How temporal and spatial variation in reproduction on the Kaibab Plateau compares to other goshawk populations is unknown because other studies typically reported reproduction at only used or successful nests (e.g., Reynolds and Wight 1978, DeStefano et al. 1994, Doyle and Smith 1994, Younk and Bechard 1994); only Keane et al. (in press) and Reynolds and Joy (in press) Table 4. Number (%) of breeding Northern Goshawks that stayed in the previous year’s nest or moved to a new or previously-used alternate nest within their territory on the Kaibab Plateau, Arizona, 1991-2002. Movement War Staved To New Alternate To Prior Alternate Total Percent Moving 1992 14 (45) 17 (55) — 55 1993 17 (35) 26 (53) 6 (12) 65 1994 7 (39) 7 (39) 4 (22) 61 1995 18 (43) 17 (40) 7 (17) 57 1996 9 (24) 16 (43) 12 (32) 76 1997 9 (30) 14 (47) 7 (23) 70 1998 19 (35) 27 (50) 8 (15) 65 1999 21 (38) 18 (32) 17 (30) 63 2000 18 (30) 20 (33) 22 (37) 70 2001 13 (43) 10 (33) 7 (23) 57 2002 7 (33) 8 (38) 6 (29) 67 Total 152 (36) 180 (42) 96 (22) 64 282 Reynolds et al. VoL. 39, No. 3 2000 2002 a. b. .2 J3 A V) O 0) C u S-i I* 20 40 60 80 100 120 20 40 60 80 100 120 Number of Territories in Subsamples 0 > 20 - £ 3 O >- -S 1-5- « .a E c n d) 1 . 0 - 0.5 2000 Be 20 40 60 80 100 120 0.5 - 0.4 ■ 0.3 0.2 -\ 0.1 0.0 H 2002 bb- _ - ^ 4- 20 40 60 80 100 120 b Number Of Territories in Subsamples Figure 4. Box plots of bootstrap subsamples estimating the effects of sample size in good (2000) and poor (2002) breeding years on estimates of the proportion of Northern Goshawk territories with used nests (eggs laid) (a) and nesting success (b) on the Kaibab Plateau, Ari- zona, 1991-2002. Dotted vertical lines are numbers of territories (120 in 2000, 121 in 2002) used to estimate the true sample means (solid horizontal lines). Box plot whiskers extend to the maximum and minimum esti- mates, boxes represent the first and third quartiles of estimates, and the horizontal lines within boxes represent the medians of estimates. Figure 5. Box plots of bootstrap subsamples estimating the effects of sample size in good (2000) and poor (2002) breeding years on estimates of (a) the mean, and (b) the coefficient of variation (CV), of young produced per used (eggs laid) Northern Goshawk nest on the Kai- bab Plateau, Arizona, 1991-2002. Dotted vertical lines are numbers of territories (120 in 2000, 121 in 2002) used to estimate the true sample means (solid horizontal lines). Box plot whiskers extend to the maximum and minimum estimates, boxes represent the first and third quartiles of estimates, and the horizontal lines within boxes represent the medians of estimates. reported the extent of temporal variation in the proportion of pairs breeding. Discussion Goshatvk populations are difficult to enumerate and monitor because of their elusive behavior, rel- atively low densities, and their structur ally-complex forest habitats. Nonetheless, goshawk detectability increases during breeding (a 6-mo period) because of their aggressive nest defense. However, detect- ability of goshawks was highly variable among in- dividuals because of extensive temporal (year-to- year) and spatial (among territory) variation in breeding. Within a year, nonbreeding territorial in- dividuals have lower detectability than breeders, and among years, low-quality individuals (Wiens and Reynolds 2005) or individuals on low-quality territories have lower detectabilities than higher- quality individuals or those on higher-quality ter- ritories because they breed less often. Detectability within and among individuals can also be variable from year-to-year because of the number and dis- persion of alternate nests, and the frequency of movement among them. Finally, breeders whose nests fail have lower detectability than successful breeders. Therefore, determining the distribution, abundance, and habitat of a population of terri- torial goshawks and their annual breeding status requires sampling protocols and efforts that pro- vide for the detection of both breeding and non- breeding goshawks. Repeated nest searching of areas suspected of having breeding goshawks (“holes” based on territory spacing) eventually showed that our study area was saturated with breeding territories. Repeated searching was re- quired because as many as 8 yr elapsed on some territories between egg-laying. Not surprisingly, our success in locating territories depended on the quality of the breeding year; more new territories September 2005 Techniques 283 were found in years when larger proportions of goshawks laid eggs. Nest searching did not cease with the discovery of a territory. Annually, between 50-75% of egg- laying goshawks moved to alternate nests within their territories, and in some years, more than half of these moves were to alternates unknown to us, some of which were more than 1 .4 km apart. Such movements have long been recognized as making the monitoring of breeding goshawks difficult (Woodbridge and Detrich 1994, Reynolds et al. 1994, Kennedy 1997). In attempts to locate gos- hawks that may have changed nests, Kennedy (1997) and Woodbridge and Detrich (1994) searched 0. 7-1.0 km and 1.6 km around the pre- viously-used nest in a territory, respectively. If the distribution of alternates within territories on the Kaibab Plateau is representative of the distribution of alternates elsewhere, then these radii would con- tain 95 and 100% of alternate nests, respectively, but only if the nest last used was close to the center of the territory. However, the farther the last-used nest was from the center of a territory, the higher the probability of missing alternates with these ra- dii. This suggests that in the early years of a mon- itoring study, longer search radii should be used, at least until centroids of territories can be esti- mated. In studies of goshawk demography (e.g., Reyn- olds and Wight 1978, DeStefano et al. 1994, Reyn- olds et al. 1994, Kennedy 1997, Reynolds and Joy in press) and habitat (e.g., Bosakowski et al. 1999, Daw and DeStefano 2001, Finn et al. 2002, Joy 2002, McGrath et al. 2003, La Sorte et al. 2004), valid inferences to the target population depend on an adequate temporal and spatial sampling. Our study showed that, because breeding is tem- porally and spatially variable and the detectability of nonbreeders is low, accurate estimates of the number and location of nests and territories de- pends on constancy in annual sampling efforts and numbers of years over which surveys are conduct- ed. Insufficient sampling for territories results in underestimates of breeding densities and habitat occupancy, and insufficient searches for nests with- in territories results in underestimates of annual proportions of pairs breeding and production of young. Because of large variation in the frequency of breeding, high rates of movement among nests, and low detectability of nonbreeders, it is particu- larly difficult to demonstrate unequivocally that territories are unoccupied in a year in which a used nest is not found. These factors, especially when combined with insufficient sampling, may result in an apparent decrease in territory occupancy and, ultimately, a population decline. The difficulty of confirming that a territory is actually unoccupied is the basis for our assigning territories with insuf- ficient evidence of occupancy as “unknown.” That territories continue to be occupied during non- breeding years was demonstrated by the fact that in many cases, the same color-marked goshawks were found to nest on the same territory before and after up to a 7 yr break in egg-laying (R. Reyn- olds unpubl. data) . Because of this, we suggest that “territory occupancy rate” (proportion of known territories occupied), a commonly used reproduc- tive parameter for goshawks (Crocker-Bedford 1990, Kennedy 1997), may be a biased estimator of the number of breeders in a population. Finally, the frequency of movements among alternate nests suggests that the scale of measurement for deter- mining the breeding status and reproduction of goshawks should be at the territory level and not at the nest area. An objective of population monitoring is to ob- tain reliable estimates from samples to infer chang- es in a target population. Our bootstrapping re- sults showed that large samples of territorial goshawks (often larger than attained in many gos- hawk studies) were needed for precise estimates of the proportion of territorial goshawks breeding and their nesting success and reproduction. Large samples are needed because of the extensive an- nual variation in the proportion of territories with reproductive goshawks. Whether equally large sam- ples of territories or pairs of goshawks are needed for reliable estimates of these parameters in other populations will likely depend on whether these populations are as temporally and spatially variable in reproduction as the Kaibab Plateau population. DeStefano et al. (1994) in Oregon, Doyle and Smith (1994) in northwestern Canada, Wood- bridge and Detrich (1994) in northern California, Kennedy (1997) in New Mexico, and Keane et al. (in press) in central California, all reported mod- erate to extensive temporal variation in goshawk reproduction. Both the proportion of territories with egg-laying goshawks and total young produced on the Kaibab Plateau were more variable among years than mean numbers of young produced per used nest per year, the most commonly reported goshawk reproductive parameter (Kennedy 1997 and references therein). Because the proportions 284 Reynolds et al. VoL. 39, No. 3 of goshawks breeding and total young produced in a year more accurately portrayed the extent of an- nual variation in reproduction of the Kaibab gos- hawk population, both are likely to better describe a population’s response to fluctuations in resourc- es (e.g., food abundance; Salafsky 2004, Salafsky et al. 2005) and habitat quality than numbers of young produced per used nest. Conclusion Stratification of a study area, protocols for de- tecting species, and sampling efforts in studies are based on subjective and previous information (Morrison et al. 2001). Our nearly complete census of breeding goshawks on the Kaibab Plateau pro- vides information on the distribution, density, var- iation in reproduction, and breeding behavior of territorial goshawks in one population. Our intent in presenting these data was to provide a frame- work for developing sampling protocols and iden- tifying sampling efforts that may be needed to re- liably estimate the distribution, density, vital rates, and habitats of breeding goshawks in other popu- lations. Extensive temporal and spatial variation in reproduction on the Kaibab Plateau required as many as 8 yr of repeated nest searching to identify a population of breeders and annual searches of areas of 1 .4-km radius around territory centers for reliable estimates of the reproductive status of ter- ritorial pairs. Further, as many as 60—80 goshawk territories were needed for precise estimates of the annual production of young by a population. The specific sampling protocols and efforts used in our study, and the samples of territories identified in this paper, demonstrate that demography and hab- itat studies of goshawks may have to employ inten- sive and repeated searches for goshawks in large study areas over at least 8 yr. Acknowledgments This study was supported by the USDA Forest Service Southwestern Region, Kaibab National Forest, North Kai- bab Ranger District, and Rocky Mountain Research Sta- tion, and a Heritage Program grant from the Arizona Game and Fish Department. Many helped find, trap, and mark goshawks during this study. We especially thank J. Seyfried, C. Erickson, J. Lambert, D. Laing, M. Gavin, D. Leslie, R. Hadwin, S. Bayard de Volo, J. Burns, J. Fein- stein, A. Gillen, B. Hunt, L. Hunt, and C. Van Cleve for three or more years of help. We thank M. Bevers, C. Flather, R. King, and P. Lukacs and R. Steidl for helpful reviews of the manuscript. Literature Cited Bosakowski, T., B. McCullough, F.J. Lapsansky, and M.E. Vaugn. 1999. Northern Goshawks nesting on a private industrial forest in western Washington./. Rap- tor Res. 33:240-244. Cressie, N. 1991. Statistics for spatial data. John Wiley and Sons, Inc., New York, NYU.S.A. Crocker-Bedeord, D.C. 1990. Goshawk reproduction and forest management. Wildl. Soc. Bull. 18:262-269. Daw, S.K. and S. DeStefano. 2001. Forest characteristics of Northern Goshawk nest stands and post-fledging areas in Oregon. J. Wildl. Manag. 65:59-65. DeStefano, S., S.K. Daw, S.M. Desimone, and E.C. Mes- LOW. 1994. Density and productivity of Northern Gos- hawks: implication for monitoring and management. Stud. Avian Biol. 16:88—91. Doyle, F.I. and J.M.N. Smith. 1994. Population responses of Northern Goshawks to the 10-year cycle in num- bers of snowshoe hares. Stud. Avian Biol. 16:122-129. Efron, B. and RJ. Tibshirani. 1993. An introduction to the bootstrap. Monographs on statistics and applied probability 57. Chapman and Hall Press, London, U.K. Finn, S.P., J.M. Marzluff, and D.E. Varland. 2002. Ef- fects of landscape and local habitat attributes on Northern Goshawk site occupancy in western Wash- ington. Forest Sci. 48:427-436. Goodman, L.A. 1964. Simultaneous confidence intervals for contrasts among multinomial populations. Annals Math. Stat. 35:716-725. Joy, S.M. 2002. Northern Goshawk habitat on the Kaibab National Forest in Arizona: factors affecting nest lo- cations and territory quality. Ph.D. disseration, Colo- rado State Univ., Fort Collins, CO U.S.A. , R.M. Reich, and R.T. Reynolds. 2003. A non- parametric, supervised classification of forest types on the Kaibab National Forest using decision trees. Int J. Rem. Sens. 24:1835-1852. , R.T. Reynolds, and D.G. Leslie. 1994. Northern Goshawk broadcast surveys: hawk response variables and survey costs. Stud. Avian Biol. 16:24^30. Keane, J.J., M.L. Morrison, and M. Fry. In press. Tem- poral variation in Northern Goshawk reproduction in the Sierra Nevada, California. Stud. Avian Biol. Kennedy, P.L. 1997. The Northern Goshawk {Accipiter gen- tilis atricapillus) : is there evidence of a population de- cline? J. Raptor Res. 31:95-106. AND D.W. Stahlecker. 1993. Responsiveness of nesting Northern Goshawks to taped broadcasts of three conspecific calls./. Wildl. Manag. 57:249-257. La Sorte, F.A., R.W. Mannan, R.T. Reynolds, and TG. Grubb. 2004. Habitat associations of sympatric Red- tailed Hawks and Northern Goshawks on the Kaibab plateau./. Wildl. Manag. 68:307-317. McClaren, E.L., P.L. Kennedy, and S.R. Dewey. 2002. Do September 2005 Techniques 285 some Northern Goshawk nest areas consistently fledge more young than others? Condor 104:343-352. McDonald, L.L. 2004, Sampling rare populations. Pages 11-42 in W.L. Thompson [Ed.], Sampling rare or elu- sive species. Island Press, Washington, DC. U.S.A. McGrath, M.T., S. DeStephano, R.A. Riggs, L.L. Irwin, AND G.J. Roloff. 2003. Spatially explicit influences on Northern Goshawk nesting habitat in the interior Pa- cific Northwest. Wildl. Monogr. 154. Morrison, M.L., W.M. Block, M.D. Strickland, and W.L. Kendall. 2001. Wildlife study design. Springer, New York, NY U.S.A. Peterson, J.T. and P.B. Bayley. 2004. A Bayesian ap- proach to estimating presence when a species is un- detected. Pages 173-188 in W.L. Thompson [Ed.], Sampling rare or elusive species. Island Press, Wash- ington, DC U.S.A. Pollock, K.H., J.D. Nichols, C. Brownie, and J.E. Hines. 1990. Statistical inference for capture-recap- ture experiments. Wildl. Monogr. 104:1-97. Rasmussen, D.I. 1941. Biotic communities of Kaibab Pla- teau, Arizona. Ecol. Monogr. 11:229-275. Reich, R.M., S.M. Joy, and R.T. Reynolds. 2004. Predict- ing the location of Northern Goshawk nests: model- ing the spatial dependency between nest locations and forest structure. Ecol. Model. 176:109-133. Reynolds, R.T. and H.M. Wight. 1978. Distribution, den- sity, and productivity of accipiter hawks breeding in Oregon. Wilson Bull. 90:182—196. and . 1982. North American accipiter hawks. Pages 288-289 in D.E. Davis [Ed.], Handbook of census methods for terrestrial vertebrates. CRC Press, Boca Raton, FL U.S.A. , S.M. Joy, and D.G. Leslie. 1994. Nest productiv- ity, fidelity, and spacing of Northern Goshawks in Ar- izona. Stud. Avian Biol. 16:106—113. , G.C. White, S.M. Joy, and R.W. Mannan. 2004. Effects of radiotransmitters on Northern Goshawks: do tailmounts lower snrvival of breeding males? J. Wildl. Manag. 68:25-32. and S.M. Joy. In press. Demography of Northern Goshawks in northern Arizona, 1991-1996. Stud. Avi- an Biol. Salafsky, S.R. 2004. Covariation between prey abundance and Northern Goshawk reproduction on the Kaibab plateau, Arizona. M.S. thesis, Colorado State Univ., Fort Collins, CO U.S.A. , R.T. Reynolds, and B.R. Noon. 2005. Patterns of temporal variation in goshawk reproduction and prey resources. J. Raptor Res. 39:237-246. Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- hawk {Accipiter gentilis). Pages 1—32 in A. Poole, and F. Gill [Eds.], The birds of North America. The Birds of North America, Inc., Philadelphia, PA U.S.A. Thompson, W.L., G.C. White, and C. Gowan. 1998. Mon- itoring vertebrate populations. Academic Press, Inc., San Diego, CA U.S.A. White, M.A. and J.L. Vankat. 1993. Middle and high el- evation coniferous forest communities of the north rim region of Grand Canyon National Park, Arizona, USA. Vegetatio 109:161-174. Wiens, J.D. 2004. Post-fledging survival and natal dis- persal of Northern Goshawks in Arizona. M.S. thesis, Colorado State Univ., Fort Collins, CO U.S.A. AND R.T. Reynolds. 2005. Is fledging success a re- liable index of fitness in Northern Goshawks 1 J. Raptor Res. 39:210-221. Woodbridge, B. and PJ- Detrich. 1994. Territory occu- pancy and habitat patch size of Northern Goshawks in the southern Cascades of California. Stud. Avian Biol. 16:83-87. Younk, J.V. AND MJ. Bechard. 1994. Breeding ecology of the Northern Goshawk in high-elevation aspen forests of northern Nevada. Stud. Avian Biol. 16:119-121. Received 26 February 2004; accepted 13 June 2005 Guest Editor: Patricia L. Kennedy Associate Editor: Clint Boal J Raptor Res, 39 {S):2S6~295 © 2005 The Raptor Research Foundation, Inc. POPULATION GENETICS AND GENOTYPING FOR MARK- RECAPTURE STUDIES OF NORTHERN GOSHAWKS (ACCIPITER GENTILIS) ON THE KAIBAB PLATEAU, ARIZONA Shelley Bayard de Volqi Graduate Degree Program in Ecology, Department of Biology, Colorado State University, Fort Collins, CO 80523 US. A. and Rocky Mountain Research Station, USDA Forest Service 2150 Centre Avenue, Building A, Suite 350, Fort Collins, CO 80526 US. A. Richard T. Reynolds Rocky Mountain Research Station, USDA Forest Service, NRRC 2150 Centre Avenue, Building A, Suite 350, Fart Collins, CO 80526 US. A. J. Rick Topinka Genomic Variation Laboratory, Department of Animal Science, Meyer Hall, University of California, Davis, One Shields Avenue, Davis CA 95616 US. A. Bernie May Genomic Variation Laboratory, Department of Animal Science, Meyer Hall, University of California, Davis, One Shields Avenue, Davis CA 95616 US. A. Michael F. Antolin Department of Biology, Colorado State University, Fort Collins, CO 80523 US. A. Abstract. — Advances in molecular techniques have facilitated use of genetic data in demographic wild- life studies. An important first step in genetic mark-recapture is selecting markers that uniquely “mark” and reliably “recapture” individuals. Markers should be tested on reliable DNAfrom known individuals (blood) before being used on non-invasively sampled DNA (hair, scat, or molted feathers) . To evaluate whether Northern Goshawks {Accipiter gentilis) can be uniquely identified by geno typing, 113 known (banded, sexed) goshawks from the Kaibab Plateau, Arizona, were genotyped using DNA from blood and five microsatellite markers and a sex-linked gene. We used mean relatedness to test whether adults in the population were related and probability of identity (P(id) = probability that two random individ- uals from the population have the same genotype) to test the ability of multi-locus genotyping for uniquely identifying goshawks. We used genetic data to assess inbreeding and demographic data to estimate the effective population size. Sixty-nine adult goshawks were sexed correctly and genotyped. Expected heterozygosity was high (H^ = 0.81), and relatedness among adults was low (r = —0.017). All individuals sampled (69 adults, 44 nestlings) had unique five-locus genotypes, the overall probability of identity was low (P(id) unbiased ~ ^7.03 X 10“^), and the observed P^jd) was <0.0001. Thus, Kaibab goshawks were uniquely “marked” by genotyping. Despite a small effective population size = 37 individuals), goshawks on the Kaibab Plateau functioned as a large breeding population with no inbreeding (Fjs = —0.001). We hypothesized that genetic diversity is maintained by gene flow via immigration of individuals from distant forests. Key Words: Northern Goshawk, Accipiter gentilis; capture-recapture, genetic tagging, individual identification', molecular sexing, probability of identity. genetica poblacional y genotipificacion para estudios de marcado-recaptura DE ACCIPITER GENTIUS EN KAIBAB PLATEAU, ARIZONA Resumen. — Los avances en las tecnicas moleculares han facilitado el uso de informacion genetica en estudios demograficos de fauna silvestre. Un primer paso importante en estudios geneticos de marcado * Corresponding author’s email address: sbdv@cnr.colostate.edu 286 September 2005 Techniques 287 y recaptura es seleccionar marcadores que “marquen” inequivocamente y que permitan “recapturar” confiablemente a los individuos. Los marcadores deben ser probados en ADN confiable de individuos conocidos (sangre) antes de ser usados en muestreos no invasivos de ADN (pelo, excremento, o plumas mudadas) . Para evaluar si los individuos de la especie {Accipiter gentilis) pueden ser identificados por genotipificacion, estudiamos 113 gavilanes conocidos (sexados y anillados) de la meseta ELaibab, Arizona, usando ADN de la sangre y cinco marcadores microsatelites y un gen ligado al sexo. Usamos el par- entesco promedio para probar si los adultos en la poblacion estaban relacionados y la probabilidad de identidad (P(id) = la probabilidad que dos individuos al azar de la poblacion tengan el mismo genotipo) para probar la habilidad de la genotipificacion mediante multiples loci para identificar individualmente a los gavilanes. Utilizamos datos geneticos para determinar el grado de endogamia e informacion de- mografica para estimar el tamafio efectivo de la poblacion. Sesenta y nueve gavilanes adultos fueron correctamente sexados y genotipificados. La heterocigocidad esperada fue alta (H^ = 0.81) y el par- entesco entre adultos fue bajo (r = —0.017). Todos los individuos analizados (69 adultos, 44 polluelos) tuvieron genotipos unices con cinco loci, la probabilidad de identidad total fue baja (P(id) unbiased ~ 7.03 X 10“^), y la P(id) observada fue <0.0001. Asi, los gavilanes de Kaibab fueron “marcados” singu- larmente por genotipificacion. A pesar de un tamano efectivo de la poblacion pequeno {N^ = 37 individuos) , los gavilanes en el Kaibab Plateau funcionaron como una poblacion reproductiva grande y sin endogamia (Fjs = —0.001). Nuestra hipotesis es que la diversidad genetica es mantenida por flujo genetico a traves de la inmigracion de individuos de los bosques lejanos. [Traduccion de Mauricio Cotera] Northern Goshawks {Accipiter gentilis) are highly secretive and are most easily detected during the breeding season when they aggressively defend their nests and young. While their defensive be- havior at nests facilitates capture-recapture studies of breeding individuals, population monitoring is difficult because individuals often forgo breeding, or their nests fail early in a breeding season. Even in years of high productivity, mark-recapture stud- ies can be prohibitively expensive because popu- lation sampling requires large field crews and mul- tiple nest visits to many breeding territories to capture and recapture breeding goshawks (Reyn- olds et al. 2005). Because of recent improvements in molecular techniques (Haig 1998, Parkeuet al. 1998), genetic capture-recapture may be a viable alternative to tra- ditional capture-recapture methods for goshawks. Collecting molted feathers requires fewer nest vis- its than traditional capture-recapture methods. Breeding goshawks begin an annual molt during spring (Squires and Reynolds 1997) and because they spend much of the breeding season near their nests, they drop many of their molts within their nest areas, including years when nesting attempts fail. Thus, goshawk feathers are readily collected from nest areas and may provide an efficient means to non-invasively sample their populations. Several factors influence the success of genetic capture-recapture studies. An appropriate number of highly variable genetic markers for identifying individuals are required, and potential biases must be identified. “Shadow effects” (lack of discrimi- nation of individuals because of low variability or sampling too few markers) can negatively bias es- timates of population abundance and positively bias estimates of survival (Mills et al. 2000) . On the other hand, when more markers than necessary are used, population abundance may be overesti- mated and survival underestimated if genotyping errors add unique “genotypes,” and thus individ- uals, to population samples (Lukacs and Burnham 2005). Both biases will inflate variance and lower precision of parameter estimates (Lukacs and Burnham 2005). Microsatellites are currently a preferred molec- ular marker for identifying individuals because they are easily interpreted (i.e., heterozygous ge- notypes are easily distinguished from homozygous genotypes) , highly variable, bi-parentally inherited, and generally appear to be selectively neutral. Fur- ther, a large body of literature exists on microsat- ellite evolution (Jarne and Lagoda 1996, Goldstein and Pollock 1997, Estoup et al. 2002), which has facilitated the development of much statistical the- ory and analytical software (Hedrick 2005). How- ever, microsatellites are expensive and time con- suming to develop for each newly-studied species. Occasionally primers used to amplify microsatellite markers in one species can be used in related spe- cies (Ellegren 1992, Primmer etal. 1996, Galbusera et al. 2000, Martinez-Cruz et al. 2002) . Prior to starting a non-invasive genetic study, es- tablishing intrapopulation genetic structure (i.e., 288 Bayard de Volo et al. VoL. 39, No. 3 levels of inbreeding and relatedness) and the fre- quency of null alleles (alleles that fail to amplify) is necessary for providing baselines against which feather samples can be compared (Mills et al. 2000). Likewise, it is important to establish statis- tical power of multi-locus genotyping for identify- ing individuals with an independent population sample. We present results from a pilot study where we assessed the feasibility of implementing a non- invasive genetic capture-recapture study on a pop- ulation of Northern Goshawks on the Raibab Pla- teau, Arizona. Before assessing the utility of molted feathers as a viable source of DNA, we established a dependable genotyping marker set using DNA derived from blood (Taberlet and Luikart 1999). Our objectives were to: (1) screen species-specif- ic and cross-specific (among species) microsatellite markers, (2) test a sex-linked gene in goshawks known to distinguish males and females in other raptors (Kahn et al. 1998), (3) assess the power of multi-locus genotyping to uniquely identify individ- uals using probability of identity analysis (P(id) 5 probability that two individuals drawn at random from the same population share the same multi- locus genotype), and (4) estimate average relat- edness, inbreeding, and effective population size for the goshawk population on the Kaibab Plateau. Methods Field Collection. The goshawk study population is lo- cated on the Kaibab Plateau in northern Arizona, an area that includes the North Kaibab Ranger District of the Kaibab National Forest and the North Rim of the Grand Canyon National Park (for descriptions of the study area see Reich et al. 2004, Reynolds and Joy 2005). It is a forested plateau surrounded by shrub-steppe habitat — the nearest forests being 97 km to the north, 250 km to the east, 80 km to the west, and 89 km to the south — except for a small patch of forest 18 km south on the south rim of the Grand Canyon. Sampled nests were well distributed across the study area. We captured 69 adult, breeding goshawks (1991-93, 2000-02) and 44 of their nestlings (Reynolds et al. 1994). We sexed adult goshawks using morphometries (mass, tarsus length) and behavior. Blood was sampled from the brachial vein with 22-gauge needles and non-heparinized capillary tubes (volume ^ 0.10 ml). Blood was transferred into STE (Sodium Chlo- ride-Tris-EDTA) buffer-filled storage tubes kept cool in insulated containers with frozen cold-packs until crews returned to the field station, where samples were subse- quently frozen (— 20°C). At the close of the field season, blood was transferred to and stored at — 80°C at Colorado State University, Fort Collins, CO. Laboratory Methods. We extracted DNA using QIAamp mini blood kits (Qiagen, Inc., Valencia, CA U.S.A.) following the manufacture’s protocol. To find mi- crosatellites, we screened published and unpublished primer sets that included microsatellites originally isolat- ed from Northern Goshawks (Topinka and May 2004), European goshawks {Aedpiter gentilis gentilis) , Golden Ea- gles {Aquila chrysaetos), and Red Kites {Milvus milvus, Peck 2000) . We also tested primers that amplify an intron within the avian CHD (chromo-helicase-DNA binding), which was used to determine gender in Red-tailed Hawks (Buteo jamaicensis) and Great Horned Owls {Bubo virgini- anus; Kahn et al. 1998). The CHD gene is located on the Z and W sex chromosomes. We expected males to be homozygous (ZZ genotype) and females to be heterozy- gous (ZW genotype). We used PCR (Polymerase-Chain-Reaction) to amplify microsatellites in 25 |xl reactions using 0.5 |xl (AGE la) or 1.0 [xl (all other markers) of template DNA, 2.5 pi 10 X buffer containing 15 mM MggCL (Promega Corp. Madison, WI U.S.A.; for markers AGE la, AGE 2 and AGE 4 an additional 3 mM Mg 2 Cl 3 was added), 20 mM dNTPs, 25 pM each primer, lU Taq polymerase, and one drop mineral oil to prevent evaporation. Negative con- trols (reactions that include all reagents except template DNA) were included in every set of reactions, and we used “cold start” PCR where tubes (in racks) were kept on ice to prevent premature non-specific priming. We used MJR PTC-100 thermocyclers programmed for the following protocol: denature at 94°C for 4 min, 31 cycles of denature at 94°C for 40 sec, annealing at 58°C for 40 sec, and chain extension at 72°C for 40 sec, with a final extension at 72°C for 5 min. We used PCR to amplify the CHD sex-linked gene in 25pl reactions using l.Opl template DNA and the same reaction buffer described above. The PCR protocol in- cluded an initial 5 min at 95°C denature, 11 cycles of denature at 94°C for 30 sec, annealing at 52°C for 35 sec, and chain extension at 72°C for 2.0 min, 31 cycles of denature at 92°C for 30 sec, annealing at 56°C for 35 sec, and chain extension at 72°C for 2.0 min, with a final extension at 72°C for 7 min. We used gel electrophoresis to separate alleles. For mi- crosatellites, we used 8% polyacrylamide (Long Ranger, Cambrex Corp., Rockland, MA U.S.A.) denaturing gels (55 cm long) that were run at 45 watts for 4-5 hr, de- pending on allele size. For the CHD sex-linked gene, we used single-strand-conformation-polymorphism (SSCP) methods (Hiss et al. 1994) and electrophoresed alleles on non-denaturing gels at 5 watts for 15 hr. For microsatellites, we established allele standards us- ing representative samples from our first gel and then standardized all other gels using those same samples. Gels were scored visually, and allele standards were run on both sides of a gel to account for gel ambiguities that cause slight variations in migration distances. Further, a subset of individuals {N = 23) was genotyped a second time to validate scores for microsatellite markers. For the CHD marker, we ran all known females together {N = 40) and all known males {N = 29) together to familiarize ourselves with allele morphology. Although not se- quenced, the fragments were ca. 240-260 base pairs, and Z and W alleles were similar in size, but were differenti- ated by the SSCP analysis (Hiss et al. 1994). Population Genetic Analysis. Population substructure, inbreeding, and genetic drift can reduce heterozygosity in populations. However, low yield and degraded DNA September 2005 Techniques 289 sampled from sources such as molted feathers can arti- ficially reduce population heterozygosity if allelic drop- out (ADO; one of two alleles in a heterozygous individual fails to amplify) at one or more markers occurs. It is, therefore, important to use high yield sources of DNA (typically blood) from a known reference population to determine frequency of ADO (or null alleles) and true levels of heterozygosity (Taberlet et al. 1999). We used Cervus 2.0 (Marshall et al. 1998) to estimate observed (Hq) and expected (He) population heterozy- gosity and null allele frequencies. Cervus provides esti- mates of null allele frequencies with an iterative algo- rithm based on differences between observed and expected homozygote frequencies. We used Genepop 3.4 (Raymond and Rousset 1995) to test for departures from Hardy-Weinberg equilibrium (random mating) and GDA 1.0 (Lewis and Zaykin 2001) to test for linkage-disequi- librium (genotypes at one marker are independent from genotypes at other markers) and to estimate Fjg, an in- dicator of population substructure and inbreeding. For a review of F-statistics and microsatellite genetic markers, see Balloux and Lugon-Moulin (2002). To test our assumption that our sample of adult gos- hawks was not comprised of closely-related individuals, we used Identix 1.1 (Belkhir et al. 2002) to estimate mean pairwise relatedness. We used Queller and Good- night’s (1989) estimator option, and tested the null hy- pothesis of no relatedness by comparing our estimate to a distribution of coefficients derived through convention- al Monte Carlo resampling procedures (1000 permuta- tions) . Probability of Identity. The uniqueness of an individ- ual’s genotype depends on the number and polymor- phism (heterozygous) of the markers. Multi-locus geno- types based on few highly-variable markers can be as powerful as those based on many less variable markers (Waits et al. 2001). Mills et al. (2000) suggested for stud- ies of genetic demography that profiles should be based on multi-locus genotypes capable of discriminating indi- viduals with 99% certainty. Estimating probability of iden- tity (P(iD)) is one way to establish this certainty when it is expressed as 1 — P(id)' P(id) is similar to the match prob- ability used in human forensics (Evett and Weir 1998, Avise 2004, Hedrick 2005), but is less susceptible to vio- lations of linkage-disequilibrium and Hardy-Weinberg equilibrium, both of which can be prevalent in small, iso- lated, or substructured populations (Waits et al. 2001). P(iD) analysis includes two steps. First, two theoretical P(iD)’s, one for unrelated individuals (P(id) unbiased)) one for siblings (P(iD)sibs)> ^.re estimated (for equations see Waits et al. 2001). Both estimators use population allele frequency data, and P(iD)unbiased corrected for bias in small samples. The two estimators provide lower and upper confidence bounds on the number of markers needed to discriminate individuals accurately. If the study population is composed of many related individuals, then resolving those individuals requires more markers. Step two involves calculating an observed P(m)obs based on ac- tual multi-locus genotypes from a known population sam- ple and is simply the proportion of all possible pairs of individuals with identical multi-locus genotypes (Waits et al. 2001). To estimate both theoretical P(id)’s and to quantify P(iD)obs (the proportion of individuals that share geno- types), we used PROB-ID5 (Waits et al. 2001). We used multi-locus genotypes derived from 69 adult goshawks, which we assume to he unrelated (see below), and 44 of their nestlings (sibling groups of 2-4 nestlings) . We first analyzed the adults and then added the offspring/ sibling groups. We used all five microsatellite markers and the CHD sex-linked gene and added markers sequentially starting with those having the highest number of alleles. Effective Population Size. To evaluate whether immi- gration and gene flow influenced genetic structure of the Kaibab population we estimated its effective size (N^) Effective population size is the idealized number of in- dividuals in a population measured either demographi- cally, in terms how many individuals actually contribute to breeding (i.e., variance in productivity), or genetically, using F-statistics and measures of inbreeding where the assumption of non-overlapping generations exists (Bar- ton and Whitlock 1997, Hedrick 2005). Goshawks sam- pled on the Kaibab Plateau during the study period likely represent at least three overlapping generations; thus, we relied on reproductive data to estimate N^. Effective population size is generally smaller than the censused population {N). Counts of breeding pairs of goshawks can be used to index N^, but not all goshawk pairs on the Kaibab produced an equal number of off- spring during the study (Wiens and Reynolds 2005). We therefore estimated annual N^’s (equation 6.8a in Hed- rick 2005) as; where k is mean productivity measured as the number of young fledged per used nest, eggs laid, subsequent young fledged, or the nest failed (1991-2003; Reynolds et al. 2005), 14 is the variance in annual mean productivity, and N is the annual count of breeding pairs for the year. We then calculated a 13-yr harmonic mean of annual A(.’s (equation 6.12b in Hedrick 2005) for our final size esti- mate. Results Genetic Markers. Of nine cross-specific and sub- specific markers tested, two did not resolve alleles successfully, six amplified successfully but lacked variability, and one both amplified and was poly- morphic (AGE la, Table 1). All four microsatellites originally isolated from Northern Goshawks ampli- fied and were polymorphic (Table 1). The CHD sex-linked gene amplified and SSCP genotypes were consistent within the sexes (females, N = 40; males N = 29) , making it useful for distinguishing between male and female goshawks. We validated our amplification and scoring of microsatellite markers after all individuals were genotyped and scored the first time. We genotyped the 23 individ- uals used as standards a second time using DNA that was archived and remained untouched in our Table 1. Statistics for microsatellites tested on blood-derived DNA sampled from female {N = 40) and male (N — 29) Northern Goshawks {Acapiter gentilis) , Kaibab Plateau, AZ (1991-1993, 2000-2002). 290 Bayard de Volo et al. VoL. 39, No. 3 u o K 1/2 i-l U c/2 W O M Cm cn CD eo CD eo O d d ID CD 1> O ^ ID ID (M o 9 . . . o o o o I I 0 o 1 I o o O 02 O ID ID O O O O O ^ d d d d T— I 00 O CD CT5 CD CM O O p p d d d d d Tf 00 ID ID O 1—1 OO 00 00 00 I> C30 d d d d d d 1> eo 00 D ^ 1> 00 eo 00 J> Jt> d> <6 d> <6 d> d: ID CD ID CM CD CM O O CD O O 00 O ID P O 00 O O d d d d d iD 1“^ O iD Ocf T— 1 i-H rH iD M) lo M, s S3 52 « •Ci, -Cl, -2 ^ ^ ^ U Cj Cj Co .2 P 'C 'S d s <«i*» cs c3 c e fciiO Csb bio bio fcuo roj 'tJ -ecj ■o IM CD c/2 <u <ri w w w w w O O O O O = < <! <3 ^ 0. O CM Z O B O ■D ce Mh <u o o o T5 B rt cn <U D u MJ cO D O o CM B O N 'u 0 B o S (L) T3 O O O M3 T3 0 D E ■a D ? O M u rt be E T3 V ’ii u T3 CO I u X M IjJ cn PS u E 0 ‘D a ■E ^ B £f tci CJ II «5 4; 4j ^ ert ^ '^3 ^ - E 1 S E o C*H (h O ^ "c Sr B E P . Q ^ M M5 K2 i C u a V u o o o o u <u M % bC c a Oh u 4-1 i2 O O Xi be a T3 OP IS u freezers from the time of original DNA extraction. We found only a single scoring error out of 230 opportunities (23 samples genotyped twice for five markers). This was a recording error; the sample had actually genotyped correctly. In comparisons of expected (H^) and observed (Ho) heterozygosity at each marker, four of the five microsatellites were similar. However, one marker (AGE 6) significantly departed from Hardy-Wein- berg expectations (P < 0.01) due to a deficiency of heterozygote genotypes (Table 1). Based on our data, we suspected this marker was sex-linked, as we found strong linkage-disequilibrium between AGE 6 and the sex-linked CHD marker (P< 0.001, Fisher’s method, 3200 runs) indicating that the two markers segregate together. It appeared that the marker was located on the Z sex-chromosome because all females (ZW) had only a single allele (homozygous), while most males (ZZ) were hetero- zygous. We hypothesize that alleles on the female’s W-chromosome are non-amplifiable (null) because of mutations in the priming sequences flanking the marker (Scribner and Pearce 2000) or because the marker on the Z-chromosome simply has no ho- mologous region on the W-chromosome. We found no evidence for linkage-disequilibrium be- tween the other four microsatellite markers (Table 1 ) when AGE 6 and CHD were excluded from the analysis. We found no evidence of null alleles, which is important for future assessments of ge- notyping error when using feathers as a source of DNA. Population Genetics. Fjs measures departures of observed and expected heterozygosity under as- sumptions of random mating and indicates either inbreeding (Fjs > zero) or inbreeding avoidance (Fis < zero). Thus, highly structured or isolated populations that experience genetic drift generally exhibit positive Fjs values. Alternatively, large pop- ulations or those experiencing high gene flow gen- erally exhibit nonsignificant or negative Fjs values. We found no evidence of inbreeding or inbreed- ing avoidance (Fjg = —0.001; 95% Cl = —0.070- 0.063; AGE 6 excluded; Table 1), suggesting that Kaibab goshawks mate randomly. Lack of inbreed- ing could result from large population size, gene flow by immigrants, or both. However, our demo- graphically derived estimate of effective population size (N^ = 37 individuals; range = 10-86) indicated that the population was demographically small, thus making gene flow a more likely source of ge- netic variability. This is consistent with our estimate September 2005 Techniques 291 Mean Coefficient of Relatedness Figure 1. Observed mean relatedness relative to 1000 randomized populations assumed to lack relatedness. The observed mean falls below that which is expected at random, occurring with a probability of 3.7%, indicating that Northern Goshawks {Accipiter gentilis) on the Kaibab Plateau, Arizona, are less related than expected at ran- dom. of relatedness among adults goshawks, where mean relatedness (r^ = —0.017) was less than expected by random (Fig. 1). Probability of Identity. Certainty of individual identification is equal to 1 -P(id)j and therefore the goal in estimating probability of identity is to ob- tain small values of P(id)- Waits et al. (2001) sug- gested a value :^0.0001 for forensic investigations where estimates of demographic parameters are needed. This threshold is interpreted as a 1:10 000 chance that two individuals sampled from the same population will have the same multi-locus geno- type. We found that all 69 unrelated adults had unique multi-locus genotypes with the inclusion of the first three markers (P(iD)obs 0.0001), and like- wise the estimated P(id) met the 0.0001 threshold (P(iD)unbiased = 1-13 X 10“^; Fig. 2a). With five markers, the same sample had a P(iD)unbiased ~ *7.03 X 10“^. However, based on demographic data, we know that siblings and parent-offspring nested si- multaneously in the Kaibab population (R. Reyn- olds unpubl. data). To model this effect we added 44 nestlings-siblings to the sample. While the two theoretical P(id)’s did not change, all five markers were required to differentiate individuals (P(iD)obs < 0.0001, Fig. 2b). In both cases (adult only and adults with off- spring-sibling groups) , our sample of markers was insufficiently large to bring the P(iD)sibs to the 0.0001 threshold. Thus, we were not able to esti- mate an upper number of markers needed for this (a) g '+-( o (b) Q Number of Markers Figure 2. Relationship between theoretical, observed, and sib probability of identity (P(id)) for Northern Gos- hawks {Accipiter gentilis) on the Kaibab Plateau, Arizona. The first five markers are microsatellites and the sixth is the CHD sex-linked gene. Observed data closely tracked that of the theoretical estimator; however, while (a) all 69 unrelated adult goshawks were resolved after the first three markers (observed P(jd) < 0.0001), (b) it took an additional two markers to resolve sibling and parent-off- spring goshawks when 44 nestlings were added to the sample. In both cases, the theoretical P(id) met our 0.0001 threshold (a 1:10 000 chance that two individuals sampled from the same population have identical multi- locus genotypes). However, we did not analyze enough markers to bring the sibling P(id) to the threshold level. resolution. Nonetheless, with five markers P(iD)sibs = 6.17 X 10 which translates into a six in a 1000 chance of drawing two identical genotypes. Be- cause we sampled many parent-offspring pairs that we could nevertheless distinguish, we are confident the five combined markers provide unique genetic marks. Discussion Our intent in this study was to develop a set of genetic markers that uniquely identified individual goshawks. We desired to establish this marker set using high-yield DNA (blood) sampled from a 292 Bayard de Volo ex al. Vox. 39, No. 3 known reference population. While most microsat- ellites tested did not amplify or were monomor- phic (most A. g. gentilis markers) , we did find a set of highly variable markers that consistently ampli- fied DNA from blood. Elsewhere (S. Bayard de Volo unpubl. data) , we found that the same genetic markers consistently and reliably amplified DNA from molted feathers. We note that because AGE 6 is probably on the Z sex chromosome, its utility for estimating within population relatedness and levels of inbreeding is limited to samples from males. However, differences in allele frequencies between populations will still be useful for larger- scale studies comparing populations. In a study of goshawks in Utah, Sonsthagen (2002) used a dif- ferent set of microsatellite markers than ours. Of eight markers, only one of theirs exhibited the same number of alleles (11 alleles, Hq = 0.73, Hjg = 0.74) as AGE 6 did in our study. This alternative marker (BV 20; Gautschi et al. 2000) would be use- ful if it exhibited similar levels of heterozygosity and allelic diversity in the Kaibab population. Re- placing AGE 6 with a less variable marker would result in having to add more markers to the entire genetic profile, which would introduce more op- portunities for genotyping error. We are currently testing BV 20 to see if it is an effective replacement for AGE 6. With the five microsatellite markers tested (Ta- ble 1), all 113 goshawks sampled had unique multi- locus genotypes resulting in a P(iD)obs 0.0001 (Fig. 2b) and a P(iD)unbiased = 7.03 X 10-'^. This was a powerful result considering that our sample in- cluded many parent-offspring and sibling pairs from the same nest. Likewise, the five microsatel- lites showed a high level of expected heterozygosity (He = 0.81). Others have shown that marker sets composed of five markers that result in Hg s 0.80 will have a theoretical P(id) ^ 0.0001 (Waits et al. 2001). In Paetkau’s (2003) retrospective analysis of 21 non-invasive genetic studies in bears {Ursus americanus, U. arctos), the number of markers used was determined by whether the first five most-var- iable microsatellite markers together had Hg > 0.80. He found that for some black bear popula- tions He was >0.80 for five markers; however, for others, and for all grizzly bear populations, that Hg was <0.80, requiring the marker set to be in- creased to six or seven loci in order to discriminate among individuals. We note that our estimates of P(id) are specific to the Kaibab goshawk population; we cannot pre- dict with complete certainty that these same mark- ers will uniquely mark goshawks from other pop- ulations. Power of discrimination depends on population-specific levels of genetic variability (het- erozygosity) ; goshawk populations that are less var- iable because of geographic isolation or habitat fragmentation may require more markers to uniquely genotype individuals (Paetkau and Stro- beck 1994). However, goshawks are highly vagile, and we suspect gene flow is high among popula- tions. These goshawk populations will probably ex- hibit similar heterozygosity, and the marker set test- ed here should prove useful for other studies. The Kaibab goshawk population exhibits high genetic variability (Table 1 ) , despite its geographic isolation and small effective population size (based on demographic data; 13-yr x = 37 individuals). Several explanations may account for this. First, it is possible that the markers used in this study are under selective sweeps with genes that are affected by balancing selection for heterozygous genotypes. Such selection has been found for the genes of the MHC (major histocompatability complex) in mam- mals (Avise 2004), in which heterozygous individ- uals experience a fitness advantage. However, we suggest it is unlikely that all four nonsex-linked markers would be under the same selective pres- sures, given that they exhibit independent segre- gation (no evidence of linkage disequilibrium; see Black et al. 2001). A second and more likely explanation is that ac- tual for this goshawk population is much larger because geographically distant populations in the region are connected by migration and gene flow. While adult goshawks are mostly sedentary on breeding territories (Detrich and Woodbridge 1994, Squires and Ruggiero 1995, Reynolds and Joy 2005), band recoveries of first-year goshawks from the Kaibab Plateau indicate dispersal distances of up to 440 km (Wiens 2004) . In addition, telemetry data show that juvenile goshawks disperse from the Kaibab Plateau in their first year, with the majority moving beyond the 80 km detection distance (Wiens 2004). Further, Wiens (2004) showed that only 11% of 614 banded nestlings returned to be recruited into the Kaibab breeding population, in- dicating high first-year mortality or low natal-site fidelity. Evidence for the latter is indicated by the lack of population genetic structure for goshawks in Utah (Sonsthagen et al. 2004). To better assess actual effective population size for goshawks in western North America, we are expanding our September 2005 Techniques 293 studies to include populations in the western por- tion of the species range. Data from these studies should allow a more comprehensive evaluation of the genetic structure and effective population size for goshawks in the West. Conclusions Genetic marking of Northern Goshawks on the Kaibab Plateau is both feasible and reliable. Like- wise, non-invasive genetic sampling will provide an alternative method for demographic and genetic data collection, as we have found that molted feathers are as reliable a source of DNA as blood (S. Bayard de Volo unpubl. data). Because gos- hawks show high territory fidelity (e.g., Detrich and Woodbridge 1994, Reynolds and Joy 2005), they are particularly well suited for non-invasive ge- netic sampling. We recommend that monitoring programs implement rigorous field collection of molted feathers. As with any demographic study, valid inferences to the population depend on ap- propriate spatial and temporal sampling from that population. Researchers and managers interested in implementing non-invasive genetic mark-recap- ture to study goshawks should contact the corre- sponding author or refer to Bayard de Volo (2005), Acknowledgments Laboratory research was funded through the Small Grant Research Fund, Grand Canyon National Park Foundation; Graduate Degree Program in Ecology Small Research Grant; Colorado State University Graduate Re- search Grant; and the 2002 Stephen R. Tully Memorial Grant, Raptor Research Foundation. Field sampling was funded by the USDA Forest Service Southwest Region and the Rocky Mountain Research Station. This work would have not been possible without the many field crew members who worked tirelessly to locate nests, and we especially thank M. Gavin, S. Joy, D. Laing, J. Seyfried, and D. Wiens for their undaunted efforts in trapping gos- hawks and collecting blood samples. L. Savage and D. Tripp provided invaluable expertise in laboratory proce- dures, and N. Stifani assisted with collecting genetic data. We thank R. King for statistical review and K. Burnham, M. Douglas, and R. Lopez, as well as two anonymous re- viewers for comments on early drafts of the manuscript. Literature Cited Avise, J.C. 2004. Molecular markers, natural history and evolution. Sinauer Associates, Inc., Sunderland, MA U.S.A. Balloux, F. and N. Lugon-Moulin. 2002. The estima- tion of population differentiation with microsatellite markers. Mol. Ecol. 11:155-165. Barton, N.H. and M.C. Whitlock. 1997. The evolution of metapopulations. Pages 183—209 in I. Hanski and M.E. Gilpin [Eds.], Metapopulation biology: ecology, genetics and evolution. Academic Press, San Diego, CA U.S.A. Bayard de Volo, S. 2005. Molt collection protocol, http:/ /lamar. colostate.edu/~sbdv/protocols_page/ protocols.htm. Belkhir, K., V. Gastric, and F. Bonhomme. 2002. Iden- Tix, a software to test for relatedness in a population using permutation methods. Mole. Ecol. Notes 2:611- 614. Black, W.C., IV, C.F. Baer, M.E Antoun, and N.M. DuTeau. 2001. Population genomics: genome-wide sampling of insect populations. Ann. Rev. of Entomology 46:441-469. Block, W.M., M.L. Morrison, and M.H. Reiser. 1994. Symposium overview: introduction. Stud. Avian Biol. 16:1-2. Detrich, P. J. and B. Woodbridge. 1994. Territory fidel- ity, mate fidelity, and movements of color-marked Northern Goshawks in the southern Cascades of Cal- ifornia. Stud. Avian Biol. 16:130—132. Ellegren, H. 1992. Polymerase-chain-reaction (PCR) analysis of microsatellites — a new approach to studies of genetic relationships in birds. Auk 109:886—895. Estoup, a., P. Jarne, and J.-M. Cornuet. 2002. Homo- plasy and mutation model at microsatellite loci and their consequences for populations genetics analysis. Mole. Ecol. 11:1591-1604. Evett, I.W. and B.S. Weir. 1998. Interpreting DNA evi- dence: statistical genetics for forensic scientists. Sin- auer Associates, Inc. Sunderland, MA U.S.A. Galbusera, R, S. van Dongen, and E. Matthysen. 2000. Cross-species amplification of microsatellite primers in passerine birds. Conserv. Genetics 1:163—168. Gautschi, B., I. Tenzer, J.P. Muller, and B. Schmid. 2000. Isolation and characterization of microsatellite loci in the bearded vulture {Gypaetus barbatus) and cross-amplification in three old world vultures. Mol. Ecol. 9:2193-2195. Goldstein, D.B. and D.D. Pollock. 1997. Launching mi- crosatellites: a review of mutation processes and meth- ods of phylogenetic inference. J! Heredity 88:335-342. Haig, S.M. 1998. Molecular contributions to conserva- tion. Ecol. 79:413-425. Hedrick, P.W. 2005. Genetics of populations. Jones and Bartlett Publishers, Inc., Sudbury, MA U.S.A. Hiss, R.H., D.E. Norris, C.H. Dietrich, R.F. Whitcomb, D.F. West, C.F. Bosto, S. Kambhampati, J. Piesman, M.F. Antolin, and W.C. Black, IV. 1994. Molecular taxonomy using single-strand conformation polymor- phism (SSCP) analysis of mitochondrial ribosomal genes. Insect Mol. Biol. 3:171—182- Jarne, P. and P.J.L. Lagoda. 1996. Microsatellites, from molecules to populations and back. Tree 11:424-429. Kahn, N.W., J. St. John, and T.W. Quinn. 1998. Chro- mosome-specific intron size differences in the avian 294 Bayard de Volo et al. VoL. 39, No. 3 CHD gene provide an efficient method for sex iden- tification in birds. Auk 115:1074-1078. Lewis, P.O. and D. Zaykin. 2001. Genetic data analysis: computer program for the analysis of allelic data. Ver- sion 1.0 (dl6c). Lukacs, P.M. and K.P. Burnham. 2005. Estimating pop- ulation size from DNA-based capture-recapture data incorporating genotyping error. J. Wildi Manag. 69: 396-403. Luikart, G. and P.R. England. 1999. Statistical analysis of microsatellite DNA data. Tree 14:253-256. Marshall, T.C., J. Slate, L.E.B. Kruuk, and J.M. Pem- berton. 1998. Statistical confidence for likelihood- based paternity inference in natural populations. Mol. Ecol. 7:639-655. Martinez-Cruz, B., V.A. David, J.A. Godoy, J.J. Negro, S.J. O’Brian, and W.E. Johnson. 2002. Eighteen poly- morphic microsatellite markers for the highly endan- gered Spanish Imperial Eagle (Aquila adalberti) and related species. Mol. Ecol. Notes 2:323-326. Mills, L.S., JJ- CiTTA, K.P. Lair, M.K. Schwartz, and D.A. Tallmon. 2000. Estimating animal abundance using non-invasive DNA sampling: promise and pit- falls. Ecol. Appl. 10:283-294. Nesje, M. and K.H. R0ED. 2000. Microsatellite DNA mark- ers from the Gyrfalcon (Falco rusticolus) and their use in other raptor species. Mol. Ecol. 9:1433-1449, Paetkau, D. 2003. An empirical exploration of data qual- ity in DNA-based population inventories. Mol. Ecol. 12: 1375-1387. AND C. Strobeck. 1994. Microsatellite analysis of genetic variation in black bear populations. Mol. Ecol. 3:489-495. Parker, P.G., A.A. Snow, M.D. Schug, G.G. Booton, and O.A. Euerst. 1998. What molecules can tell us about populations: choosing and using a molecular marker. Ecology 79:361-382. Peck, N. 2000. DNA forensics of raptors and the isolation and characterization of microsatellite markers in Ac- cipitridae. Ph.D. dissertation, Univ. of Nottingham, Nottingham, U.K. Primmer, C.R., A.P. Moller, and H. Ellegren. 1996. A wide-range survey of cross-species microsatellite am- plification in birds. Mol. Ecol. 5:365-378. Queller, D. and K. Goodnight. 1989. Estimating relat- edness using genetic markers. Evolution 43:258-275. Raymond, M. and E. Rousset. 1995. GENEPOP Version 3. Id; population genetics software for exact tests and ecumenicism. y. Heredity 86:248-249. Reich, R.M., S.M. Joy, and R.T. Reynolds. 2004. Predict- ing the location of Northern Goshawk nests: model- ing the spatial dependency between nest locations and forest structure. Ecol. Modeling 176:109—133. Reynolds, R.T. and S.M. Joy. 2005. Demography of Northern Goshawks in northern Arizona, 1991-1996. Stud. Avian Biol. In press. , , AND D.G. Leslie. 1994. Nest productivity, fidelity, and spacing of Northern Goshawks in north- ern Arizona. Stud. Avian Biol. 16:106-113. , J.D. Wiens, S.M. Joy, and S.R. Salafsky. 2005. Sampling considerations for demographic and habitat studies of goshawks. J. Raptor Res. 39:274-285. Scribner, K.T. and J.M. Pearce. 2000. Microsatellites: evolutionary and methodological background and empirical applications at individual, population and phylogenetic levels. Pages 235—273 mA.J. Baker [Ed.], Molecular methods in ecology. Blackwell Science Ltd., Oxford, U.K. Sonsthagen, S.A. 2002. Gene flow and genetic charac- terization of Northern Goshawks {Accipiter gentilis) breeding in Utah, USA. M.S. thesis, Brigham Young Univ., Provo, UT U.S.A. , S.L. Talbot, and C.M. White. 2004. Gene flow and genetic characterization of Northern Goshawks {Accipiter gentilis) breeding in Utah. Condor 106:826— 836. Squires, J.R. and R.T. Reynolds. 1997. Northern Gos- hawk {Acdpiter gentilis) . In A. Poole and F. Gill [Eds.], The birds of North America, No. 298. The Birds of North America, Inc., Philadelphia, PA U.S.A. AND L.F. Ruggiero. 1995. Winter movements of adult Northern Goshawks that nested in south-central Wyoming./. Raptor Res. 29:5-9. Taberlet, P. and G. Luikart. 1999. Non-invasive genetic sampling and individual identification. Biol. J. Linn. Soc. 68:41-55. , L.P. Waits, and G. Luikart. 1999. Noninvasive genetic sampling: look before you leap. TREE 14:323- 327. Topinka, J.R. AND B. May. 2004. Development of poly- morphic loci in the Northern Goshawk (Accipiter gentilis) and cross-amplification in other raptor spe- cies. Conserv. Genetics 5:861—864. Waits, L.P., G. Luikart, and P. Taberlet. 2001. Estimat- ing the probability of identity among genotypes in natural populations: cautions and guidelines. Mol. Ecol. 10:249-256. Wiens, J.D. 2004. Post-fledging survival and natal dis- persal of juvenile Northern Goshawks in Arizona. M.S. thesis, Colorado State University, Fort Collins, CO U.S.A. AND R.T. Reynolds. 2005. Is fledging success a re- liable index of recruitment in Northern Goshawks? J. Raptor Res. 39:210-221. Received 26 February 2004; accepted 15 May 2005 Guest Editor: John Keane September 2005 Techniques 295 Appendix. Source, repeat qualities and accession numbers or primer sequences for microsatellites found to be useful for genotyping Northern Goshawks {Accipiter gentilis) on the Kaibab Plateau, Arizona. Marker Repeat Motif^ Allele Size IN Base Pairs Accession No. OR Primer Sequences 5' TO 3' Author AGE 1 (gggaalg. . (gaga)g. . (gagaala 216 AY312451 AGE 2 (gagaa) io(ga)4 170 AY312452 Topinka and AGE 6 ( gagaa ) 4 . . ( gagaa ) 2 . • ( gagaa ) 5 259 AY312456 May 2004 AGE 4 (gagaa) 275 AY312454 AGE la (ggat) s 208*^ f acaactgggctgtgctttgc r cttcccggtggctgaggctt Peck 2000 ® Sequenced by authors. Mean allele size in European goshawk (A. g. gentilis) . J. Raptor Res. 39(3):296-302 © 2005 The Raptor Research Foundation, Inc. WHEN ARE GOSHAWKS NOT THERE? IS A SINGLE VISIT ENOUGH TO INFER ABSENCE AT OCCUPIED NEST AREAS? Douglas A. Boyce, W USDA Forest Service, Pacific Northwest Research Station, 2770 Sherwood Lane, Juneau, AK 99801 U.S.A. Patricia L. Kennedy Eastern Oregon Agricultural Research Center & Department of Fisheries and Wildlife, Oregon State University, P.O. BoxE, Union, OR 97883 US. A. Paul Beier School of Forestry, Northern Arizona University, Flagstaff, AZ 86011-5018 U.S.A. Micheal F. Ingraldi Arizona Game and Fish Department, Research Branch, 2221 W. Greenway Road, Phoenix, AZ 85023 U.S.A. Susie R. MacVean Arizona Game and Fish Department, 3500 South Lake Mary Road, Flagstaff, AZ 86004 U.S.A. Melissa S. Siders USDI Bureau of Land Management, Grand Staircase-Escalante National Monument, 190 East Center Street, Kanab, UT 84741 U.S.A. John R. Squires University of Montana Forestry Sciences Laboratory, 800 East Beckwith, Missoula, MT 59807 U.S.A. Brian Woodbridge USDI Eish and Wildlife Service, 1829 South Oregon Street, Yreka, CA 96097 U.S.A. Abstract. — ^We tested the efficacy of three methods (historical nest search, broadcast search, and tree transect search) for detecting presence of the Northern Goshawk (Accipiter gentilis) at occupied nest areas during the 1994 breeding season using only a single visit to a previously known nest area. We used detection rates in a probability model to determine how many visits are required to have confidence in reporting absence of goshawks. The purpose of this study is to understand if the three methods for detecting goshawks are robust enough for managers to rely on them for making land management decisions that may impact goshawk nest areas. Blind tests were conducted throughout the western United States. Results were similar among methods with goshawk presence going undetected at 36-42% of the occupied nest areas after a single visit. These results indicate that a single visit to a nest area is inadequate to provide reliable information on nest area occupation. Our probability of detection model showed that if each detection method is repeated three (historical or tree transect) or four (broadcast) times, goshawk absence can be inferred with a high level of confidence. Conclusions regarding nest area occupation using a single visit sampling method should be made with utmost caution. Classifying a nest area as vacant, when in fact goshawks are present, is a serious concern and leads to spurious conclusions. Land managers making habitat-altering decisions should not rely on a single visit to nest areas to establish the absence of goshawks. Possibilities for improving the detection of nesting goshawks include multiple independent visits using the same method, using a sequence of techniques in combi- nation to yield an improved cumulative probability of detection, or developing a new method yielding a higher probability of detection. The historical nest search obtained the best results, followed by the tree transect and broadcast search. ^ Email address: daboyce@fs.fed. us 296 September 2005 Techniques 297 Key Words: Northern Goshawk, Accipiter gentilis; detection rates; forest management; nest area; occupancy; repeated sampling <:CUANDO ESTA AUSENTE ACCIPITER GENTILIS? ^ES SUFICIENTE UNA SOLA VISITA PARA IN- FERIR AUSENCIA EN AREAS DE NIDIFICACION OCUPADAS? Resumen. — Probamos la eficiencia de tres metodos (busqueda de nidos historica, busqueda por medio reproduccion de grabaciones, busqueda a lo largo de transectos de arboles) para detectar la presencia del halcon Accipiter gentilis en areas de nidificacion activas durante la epoca reproductiva de 1994, utilizando una sola visita a un area de nidificacion previamente conocida. Utilizamos las tasas de detec- cion en un modelo de probabilidad para determinar cuantas visitas se requieren para tener certeza al reportar una ausencia de esta especie de halcon. El proposito de este estudio es entender si los tres metodos para detectar a esta especie son suficientemente robustos para confiar en ellos al tomar deci- siones de manejo de tierras que pueden afectar areas de nidificacion. Realizamos pruebas ciegas a traves del oeste de los Estados Unidos. Los resultados fueron similares entre los metodos; la presencia de los halcones no fue detectada en el 36-42% de las areas de nidificacion activas luego de una sola visita, Estos resultados indican que una sola visita a un area de nidificacion no es adecuada para obtener informacion confiable sobre la actividad de nidificacion en el area. Nuestro modelo de probabilidad de deteccion mostro que si cada metodo es repetido tres (historico o transecto de arboles) o cuatro (re- produccion de grabaciones) veces, la ausencia de halcones puede ser inferida con un alto grado de confianza. Las conclusiones con respecto a la actividad de las areas de nidificacion utilizando el metodo de muestreo de una sola visita deben tomarse con gran precaucion. La clasificacion de un sitio de nidificacion como vacio, cuando de hecho los halcones estan presentes, es una preocupacion seria y puede llevar a conclusiones falsas. Las personas encargadas de manejar las tierras y tomar decisiones con relacion a la alteracion de los habitats no deberian confiar en una sola visita a los sitios de nidifi- cacion para determinar la ausencia de estos halcones. Algunas de las posibilidades para mejorar la deteccion de halcones que se encuentran nidificando incluyen realizar visitas multiples e independientes utilizando la misma metodologia, utilizar conjuntamente una secuencia de tecnicas para producir me- jores probabilidades de deteccion acumulativas o desarrollar un metodo nuevo que pueda proveer de una probabilidad de deteccion mayor. La metodologia de busqueda de nidos historica obtuvo los me- jores resultados, seguida por la de los transectos de arboles y la busqueda por medio de reproduccion de grabaciones. [Traduccion del equipo editorial] The U.S. Department of Interior Fish and Wild- life Service (FWS) reviewed the status of the North- ern Goshawk {Accipiter gentilis atricapillus) for Fed- eral protection (i.e., listed as threatened or endangered under provisions of the Endangered Species Act) three times since 1991. In each case the FWS ruled that listing was unwarranted. Pop- ulation trend is one of five factors used by the FWS for determining whether to list a species as threat- ened or endangered. The majority of nesting gos- hawks in the western United States are located on lands managed by the U.S. Department of Agri- culture Forest Service (FS). Since the FS is re- quired by the National Forest Management Act (NFMA) to maintain species population viability, monitoring the occupancy of goshawk nest areas is necessary to evaluate population trends. Lacking a formal national goshawk monitoring program, the FS management approach to pro- tecting goshawks in the southwestern United States is to locate goshawk nest trees and post-fledging family areas (Kennedy et al. 1994) prior to habitat alterations and then to apply goshawk manage- ment recommendations (varying from region to region) to conserve the nest area, manage the post-fledging family area, and manage the foraging area (Reynolds et al. 1992) . After implementation of habitat management prescriptions, follow-up management practices should include monitoring the effect of habitat changes on species; however, this is rarely done. The untested assumption is that the management program will work as designed. Finding and monitoring nesting goshawks is a critical component of successful adaptive land management practices if goshawks are to persist in managed landscapes. Goshawks exhibit strong fi- delity to nest areas (Reynolds and Joy in press) , but have fluctuating population numbers and nesting success year to year. Goshawks also frequently change nest locations within a nest area or between 298 Boyce et al. VoL. 39, No. 3 nest areas within a territory. Because a proportion of the local population of goshawks moves to al- ternate nest areas on an annual basis, sampling only the historical nest areas over time without finding the alternate nest areas will result in fewer and fewer occupied nest areas (i.e., the unwar- ranted appearance of a declining population) . Counting, sampling, and detecting birds are im- portant concerns of avian researchers (Bart and Earnst 2002, Farnsworth et al. 2002, Rosenstock et al. 2002, Thompson 2002). Developing techniques to find goshawks efficiently has been an ongoing process (Kimmel and Yahner 1990, Kennedy and Stahlecker 1993, Joy et al. 1994, Watson et al. 1999, Penteriani 1999, Roberson et al. in press). Biolo- gists have yet to develop an accurate, cost-effective method that will detect goshawks throughout the nesting period. This is because the species is secre- tive, difficult to find and study, and their behavior changes during the breeding season. Kennedy and Stahlecker (1993) tested a technique for broad- casting goshawk vocalizations from calling stations positioned on parallel transects that were placed tangential to the occupied nest. Their tests were conducted in the southwestern U.S. during the nestling to fledging stage. They found that the probability of detecting a goshawk, when within 100 m of a nest, averaged 70% throughout the sea- son using multiple visits. The median detection dis- tance was 141 m. On control transects, without broadcasting, detection rates dropped to between 30% (courtship) and 60% (fledgling). In Washington, Watson et al. (1999) tested Ken- nedy and Stahlecker’s (1993) broadcast method us- ing three stations (400 m, 250 m, and 100 m) on a single transect that passed tangential to the nest at 100 m at its closest point. They found five visits at 100 m from the nest, eight visits at 250 m from the nest, and 10 visits at 400 m attained a 90% or higher detection rate. In another study using the broadcast technique from courtship to fledgling dependency, only 52% of goshawks were detected (McClaren et al. 2003) ; but, detections were lower during courtship (40%) and highest during fledg- ling dependency (75%). Kennedy and Stahlecker (1993), Watson et al. (1999), and McClaren et al. (2003) are examples of experienced goshawk bi- ologists evaluating goshawk survey techniques. Their prior experience with goshawks and knowl- edge of nest locations may have positively influ- enced experimental results (i.e., their detection rates probably represent maximum rates under test conditions) . A problem with past goshawk inventory and monitoring efforts has been a reliance on meth- odologies whose bias, probability of detection, and magnitude of detection error were unknown. There has always been uncertainty associated with misclassifying a goshawk territory as unoccupied when it may be occupied (i.e., error of omission). In 1994, the FS identified the need to test the ef- ficacy of techniques for finding goshawks. This was driven by the FS desire to implement specific hab- itat altering management actions designed to pro- tect goshawk nest areas, post-fledging family areas, and the surrounding foraging area from harm (Reynolds et al. 1992), or to allow for flexible man- agement options if goshawks were not present. Three commonly used detection methods available at that time were identified as needing testing (his- torical nest tree search, broadcast search, and tree search within potential nest areas). No investiga- tors had compared the potential errors associated with the three typical inventory techniques. Our objectives were to: (1) document the error associated with each of these three detection tech- niques and (2) use the error rates to estimate the number of nest area visits needed to infer absence of goshawks with different levels of confidence. We conducted a blind test of these methods for de- tecting breeding goshawks to reveal the magnitude of error associated with each technique. The rea- son we conducted blind tests was to control the variability introduced in previous tests conducted by experienced goshawk biologists that had prior knowledge of the nest area and its status (Kennedy and Stahlecker 1993, Joy et al. 1994); possibly in- fluencing their results. We then input our results into a probability model to conceptually explore various combinations of detection rates, errors as- sociated with these detection rates, and predict the number of sampling visits needed to have confi- dence in the information collected. Methods We tested the efficacy of revisiting historical nest trees, broadcasting goshawk vocalizations in nest areas, and scanning all trees along transects established throughout nest areas within an 800 m diameter area centered on occupied (nest with eggs/young) nest areas. The size of our sampling unit was 1/35 the estimated size of the ter- ritory (2400 ha) (Reynolds et al. 1992) and was selected to account for alternate nest locations within a single nest area. Field tests were conducted from June to early mid- July 1994 during nestling and fledgling dependency pe- September 2005 Techniques 299 riods (Squires and Reynolds 199'7). Experienced field bi- ologists determined that each nest area tested had nesting goshawks present prior to the test. During the testing period, occupancy was determined by observing goshawks incubating eggs, adults brooding young, or ot> serving young at the nest. The same criteria were used at all study areas. Personnel naive to the presence and location of occupied nests were used to test the three methods. Only one method was tested, and only one visit was made, at each occupied nest area. The three meth- ods were randomly assigned to active nest areas. To sim- ulate normal field conditions, experience was allowed to vary among field members; no effort was made to ran- domize field crew experience among the three detection methods. Results from each state were pooled to improve sample size. Study Areas. Tests were conducted in Arizona, Califor- nia, New Mexico, and Wyoming. In Arizona {N = 44), tests were conducted in the Apache/Sitgreaves, Coconi- no and Kaibab National Forests. Forests in Arizona were dominated by ponderosa pine {Pinus ponderosa ) , white fir {Abies concolor), and Douglas-fir {Pseudotsuga menziesii). In California {N =10), tests were conducted in the Klamath National Forest, where at higher elevations, forests were dominated by red fir {Abies magnified), white fir, ponde- rosa pine, lodgepole pine {Pinus contorta), Douglas-fir, and incense cedar {Calocedrus decurrens), and lower ele- vation forests by ponderosa pine and white fir (Kuchler 1977). In New Mexico {N = 11) tests were conducted in the Santa Fe National Forest where forests contained ponderosa pine, Douglas-fir, white fir, and quaking aspen {Populus tremuloides) and at higher elevations subalpine fir {Abies lasiocarpa) and Englemann spruce {Picea engel- mannii). In Wyoming {N = 12), tests were conducted in the Medicine Bow National Forest where lower elevation forests contained lodge pole pine with scattered quaking aspen, and higher elevation forests contained subalpine fir and Engelmann spruce (Alexander et al. 1986, Mar- ston and Clarendon 1988). Historical Nest Search. The most common goshawk search technique used prior to 1990 was to visit historical nest areas and relocate previously used nest trees to de- termine occupancy. Typically, little effort was spent in a broader search of a nest area if goshawks were not found. To simulate this method, biologists were given 1:24000 scale maps marked with the approximate locations of nest trees within a nesting area where goshawks had pre- viously nested. Biologists were instructed to relocate the nest trees and determine if goshawks were present and nesting. The strength of this method relies on goshawk fidelity to nest areas (Reynolds et al. 1994) and that field personnel often detect goshawk presence by observing the defensive behavior of goshawks near their nests. Oth- er clues to goshawk nest area occupancy with this method included observing fecal material, prey remains, or molt- ed goshawk feathers in the vicinity of nests. When these clues were found, the area was searched further to find the occupied nest. Broadcast Surveys. This goshawk detection technique was developed in the early 1990s and involved broadcast- ing taped goshawk calls (alarm and juvenile food beg- ging) to elicit a response. Field crews followed the pro- cedure of Kennedy and Stahlecker (1993), as modified by Joy et al. (1994). Recorded calls of goshawks were broadcast from stations located at 300-m intervals, on parallel transects, in an 800 m radius area. A search was initiated to locate visually the nest once a goshawk re- sponded. The broadcast method is a means of systemat- ically searching the landscape for goshawks. This method is also useful for locating nesting pairs that move to al- ternate areas within their territory. A problem with the technique is that goshawks do not always respond to the broadcast call when they are present, may respond with a silent approach, or may respond to broadcast calls when they are far away from their nest areas and, thus, confound results. Additional confounding factors in- clude seasonal effects and misidentification of calls such as Steller’s Jay {Cyanocitta stellen) mimicking goshawks (Kennedy and Stahlecker 1993). Tree Transect. The tree transect technique is a system- atic visual search of a forested area centered on the oc- cupied nest. This method involved field crews walking along parallel transects spaced 50 m apart while exam- ining individual trees along either side of and directly along the transect path for goshawk nests in tree crowns (Squires and Reynolds 1997). At 50 m, the probability of eliciting goshawk defensive behavior was assumed to be high because they could presumably hear or see the field crew. Crews also looked for prey plucking posts, fecal ma- terial or stains, and scattered prey remains that would provide evidence of a potential occupied nest nearby. The Model. To address our second objective, we input the estimates of detection obtained from each search method above into a probability model (McArdle 1990). This allowed an estimation of the sample size needed to have confidence that goshawks were absent. In other words, how many revisits to the nest area are necessary to conclude goshawks are absent? Guynn et al. (1985) and Reed (1996) used probability models to retrospec- tively estimate confidence in detecting a species. Kery (2002) applied their model prospectively to infer how many visits were needed to be statistically confident that the species being sampled was absent. McArdle’s (1990) probability model includes: (1) the number of sampling visits {N) to an area, (2) the species probability of detection (p) during any visit, (3) and con- fidence (a) levels acceptable to the investigator (usually 95%, and therefore a = 0.05). Assuming all visits to gos- hawk nest areas are similar and independent, the prob- ability of not detecting nesting goshawks after N visits (Kery 2002) is: Probability =„=(!-„)» (n {N unsuccessful visits) We can solve for N and get: log (a) = NX log (1 - p) (2) N= log (a)/log (1 - p) (3) The minimum number of visits, needed to conclude that a 800-m radius circle containing a previously used nest area is unoccupied within a 95% confidence interval (a = 0.05) can be estimated by substituting the proba- bility of detection values (historical = 0.64, broadcast = 0.58, transect = 0.62; see Results for details) into Equa- tion 4. 300 Boyce et al. VoL. 39, No. 3 N^in = log (0.05) /log (1 - p) (4) Results The results were similar for each method tested; between 58-64% of the occupied nest areas were found (historical nest search [16/25], tree transect [16/26], broadcast surveys [15/26]). Conversely, between 36-42% of the occupied goshawk nest ar- eas were missed. The broadcast result for a single visit is identical to what Kennedy and Stahlecker (1993) reported. We did not test for temporal dif- ferences in the methods due to limited sample siz- es. Despite the poor performance of each method for detecting goshawks using a single visit to a nest area, each method may be repeated several times to increase the probability of detection (Kennedy and Stahlecker 1993, Watson et al. 1999, McClaren et al. 2003). Using the detection results, we esti- mated the number of visits (N^m) needed to infer goshawk absence at nest areas at the 95% confi- dence level (a = 0.05) as 2.9 for the historical nest search, 3.1 for the tree transect, and 3.5 for the broadcast survey. These detection results are only relevant to active nest areas. Increasing the confidence level while maintain- ing a consistent detection rate quickly increases the number of visits needed to infer goshawk absence at nest areas and renders the sampling effort un- realistic (Table 1). For example, if we set the con- fidence level to 0.95, and want to limit the number of visits to two, then the probability of detection required for a method to be effective would have to be nearly 80%. Given this scenario, the goal for developing new or improved detection techniques should be to achieve a probability of detection lev- el of at least 80%. If the confidence level is in- creased to 0.99 (a = 0.01) to further reduce the misclassification error while retaining the detection probability at 80%, then the number of required visits to nest areas is three and is still a feasible management option (i.e., not cost prohibitive). McKelvey and Pearson (2001) examined a series of simulations for measuring small mammal popula- tions with different detection probabilities and their results revealed the same general pattern as ours in that low detection probabilities require a large number of sampling sessions to attain confi- dence in the findings. Discussion Our results were from occupied nest areas only. Although we controlled as much variation as pos- sible, there were many sources of variation we did not control. We did not test for false positive de- tections at unoccupied sites (Kennedy and Stah- lecker 1993), which are needed for a broader de- scription of detection probabilities. Detection frequencies of goshawks at nest areas may vary for any number of reasons, but perhaps most impor- tant are changes in goshawk behavior as breeding season progresses (Squires and Reynolds 1997). Breeding goshawks become more defensive at nest areas later in the nesting season and generally are easier to detect (Squires and Reynolds 1997). Young goshawks also are easier to detect later in the breeding season as they grow and become more active (McClaren et al. 2003). Because detec- tion methods may be temporally sensitive, manag- ers must interpret the results cautiously (McClaren et al. 2003, Roberson et al. in press). As the breeding season progresses from March through July, goshawk nest failures continue for a host of reasons. A difficult sampling problem is to account for these nest failures. Sampling after re- productive failure occurs may lead to misclassifi- cation of nest areas as inactive. In addition, nesting areas are occupied by adults that do not breed ev- Table 1. Theoretical number of visits to Northern Goshawk nest areas to infer goshawk absence using different detection probabilities (p) and confidence levels (a). Probability of Detection a 60 65 70 75 80 85 90 95 0.25 1.51 1.32 1.15 1.00 0.86 0.73 0.62 0.46 0.20 1.76 1.53 1.34 1.16 1.00 0.85 0.70 0.54 0 15 2.07 1.81 1.58 1.37 1.18 1.00 0.82 0.63 0.10 2.51 2.19 1.91 1.66 1.43 1.21 1.00 0.77 0.05 3.26 2.85 2.49 2.16 1.86 1.58 1.30 1.00 0.01 5.03 4.39 3.83 3.32 2.86 2.43 2.00 1.54 September 2005 Techniques 301 ery year and thus, detection probabilities at indi- vidual nest areas are likely to vary temporally (Boal et al. 2005, R. Reynolds pers. comm.). We did not test the ability to detect nonbreeding pairs occu- pying nest areas; we tested only the breeding por- tion of the population (i.e,, actively nesting in pairs) . This has important ramifications for under- standing the population’s status (Kennedy 1997) and for managers making decisions based on re- sults for years when few pairs are breeding. The ability to detect nonbreeding goshawks and breed- ing goshawks that have failed are likely to be dif- ferent. Improved probabilities of detection may be possible by regulating the timing of when different methods are used. Another source of variation that affects detec- tion probabilities is the timing of egg-laying by fe- males within and between populations: variation in the timing of egg-laying introduces inherent error to detection rate estimates. Thus, there will likely be differential success in detecting goshawks be- cause the detection method used will not be per- fectly sequenced to the breeding phenology of all pairs within or between populations. We recom- mend that managers determine the breeding phe- nology of their target population before imple- menting goshawk surveys (see Dewey et al. 2003) . Variation also exists in the experience of field crews and therefore, accuracy and reliability of sur- vey data. In addition, goshawks may move to alter- nate nesting areas within a territory; this constant shifting among alternate nests may result in a per- ceived decay in the number of occupied nests and a fallacious conclusion of population decline if only the historical nest areas are visited (R. Reyn- olds pers. comm.). Given that multiple factors in- fluence detection probabilities, the implication for monitoring populations at regional scales is that detection protocols should consider these sources of variation so that data sets from different loca- tions and times are comparable for later use in an- alyzing large-scale population trends. None of the goshawk detection methods tested in this study, when applied once, were adequate to conclude goshawks were absent at nest areas. The usefulness of new detection methods is dependent on understanding the associated detection proba- bilities and error rates for different spatial and temporal scales. Future approaches might include combining several different methods in a temporal sequence that improves the cumulative probability of detection throughout the breeding season (Dewey et al. 2003). Highly accurate methods ap- propriate early in the breeding season (e.g., listen- ing stations; Dewey et al. 2003) may be ineffective late in the breeding period. However, by combin- ing methods and taking advantage of their strengths, improved results may be obtained, but this remains to be tested. Another approach is to test the detection probability of successive appli- cations of the historical and tree-transect methods (i.e., multiple visits) and determine if the results match the outcome reported for the broadcast method (70%; Kennedy and Stahlacker 1993). The predictions in this paper related to cumulative de- tection probabilities from multiple applications of one technique should be tested. If these predic- tions are supported empirically, then managers could design a monitoring program that relies on multiple applications of a single technique (e.g., tree transects). Detection success may be optimized by using lis- tening stations prior to egg-laying (March and April; Penteriani 1999, Dewey et al. 2003), tree searches on parallel transects during incubation and the nestling stage (May-June) , and broadcast calling (wail and food begging) during the post- fledging dependency period (Kennedy and Stah- lecker 1993, McClaren et al. 2003). Although broadcast surveys are frequently used during the nestling stage, recent tests of this approach by Rob- erson et al. (in press) in Minnesota suggest broad- cast surveys may not be an effective tool during this stage. Roberson et al. (in press) report high detec- tion rates with broadcast surveys during courtship (70%) and fledgling-dependency phases (68%). Detection rates were lowest during the nestling phase (28%), when there appeared to be higher variation in likelihood of detecting individuals. Acknowledgments We thank all the field assistants who participated in this study. We particularly thank Drs. Richard T. Reynolds, Winston Smith, Clayton M. White, and Curtis Flather for their review of the manuscript. Bob Nelson, former na- tional director of wil d life and fisheries management for the Forest Service, funded and encouraged this study. We also thank Clint Boal, Vincenzo Penteriani, and two anonymous reviewers for their helpful comments. Literature Cited Alexander, R.R., G.R. Hoffman, and J.M. Wirsing. 1986. Forest vegetation of the Medicine Bow National For- est in southeastern Wyoming: a habitat type classifi- cation. USDA, Forest Service Research Paper, RM-271, 302 Boyce et ai.. VoL. 39, No. 3 Rocky Mountain Forest and Range Experiment Sta- tion, Fort Collins, CO U.S.A. Bart, J. and S. Earnst. 2002. Double sampling to esti- mate density and population trends in birds. Auk 119: 36-45. Boat, C.W., D.E, Andersen, and RL. Kennedy. 2005. Pro- ductivity and mortality of Northern Goshawks in Min- nesota. J. Raptor Res. 39:222-228. Dewey, S.R., RL. Kennedy, and R.M. Stephens. 2003. Are dawn vocalization surveys effective for monitoring goshawk nest-area occupancy? J. Wildl. Manag. 67: 390-397. Farnsworth, G.L., K.H. Pollock, J.D. Nichols, T.R. Si- mons, J.E. Hines, and J.R. Sauer. 2002. A removal model for estimating detection probabilities from point-count surveys. Auk 119:414—425. Guynn, D.C., Jr., R.L. Downing, and G.R. Askew. 1985. Estimating the probability of non-detection of low density populations. Cryptozoology 4:55-60. Joy, S.M., R.T. Reynolds, and D.G. Leslie. 1994. North- ern Goshawk broadcast surveys: hawk response vari- ables and survey cost. Pages 24-30 in Block, W.M., M.S. Morrison, and M.H Reiser (Eds.), The Northern Goshawk: ecology and management. Stud. Avian Biol. 16:24-30. Kennedy, P.L. 1997. The Northern Goshawk {Accipiter gen- tilis atricapillus) : is there evidence of a population de- cline? Special issue on responses of forest raptors to management: a holarctic perspective, y. Raptor Res. 31: 95-106. and D.W. Stahlecker. 1993. Responsiveness of nesting Northern Goshawks to taped broadcasts of three conspecific calls./. Wildl. Manag. 57:249-257. , J.M. Ward, G.A. Rinker, and J.A. Gessaman. 1994. Post-Hedging areas in Northern Goshawk home ranges. Stud. Avian Biol. 16:75-82. Kery, M. 2002. Inferring the absence of a species-a case study of snakes./. Wildl. Manag. 66:330-338. Kimmel, J.T., AND R.H. Yahner. 1990. Response of North- ern Goshawks to taped conspecific and Great Horned Owl calls. / Raptor Res. 24:107-112. Kuchler, A.W. 1977. The map of the natural vegetation of California. Pages 909-938 m M.G. Barbour and J. Major (Ed-S.), Terrestrial vegetation of California. John Wiley, New York, NY. Marston, R.A., AND D.T. Clarendon. 1988. Land system inventory of the Medicine Bow Mountains and Sierra Madre Mountains, Medicine Bow National Forest, Wy- oming. USDA Forest Service Research Paper MBNF- 88-01. Medicine Bow National Forest, Laramie, WY U.S.A. McArdle, B.H. 1990. When are rare species not there? Oikos 57:276-277. McClaren, E.L., P.L. Kennedy, and P.L. Chapman. 2003 Efficacy of male goshawk food-delivery calls in broad- cast surveys on Vancouver Island. /. Raptor Res. 37: 198-208. McKelvey, K.S. and D.E. Pearson. 2001. Population es- timation with sparse data: the role of estimators versus indices revisited. Canadian J. Zool. 79:1754—1765. Penteriani, V. 1999. Dawn and morning goshawk court- ship vocalizations as a method for detecting nest sites. /. Wildl. Manag. 63:511-516. Reed, J.M. 1996. Using statistical probability to increase confidence of inferring species extinction. Conserv Biol. 10:1283-1285. Reynolds, R.T, R.T. Graham, H.M. Reiser, R.L. Bassett, P.L. Kennedy, D.A. Boyce, Jr., G. Goodwin, R. Smith, AND E.L. Fisher. 1992. Management recommenda- tions for the Northern Goshawk in the southwestern United States. USDA For. Serv. Gen. Tech. Rep. RM- 217. Rocky Mountain Research Station, Fort Collins, CO U.S.A. AND S.M. Joy. In Press. Demography of northern goshawks in northern Arizona, 1991-1996. Stud. Avi- an Biol.: In press. , D.G. Leslie, and S.M. Joy. 1994. Nest productiv- ity, fidelity, and spacing of Northern Goshawks in Ar- izona. Stud. Avian Biol. 16:106-113. Roberson, A.M., D.E. Andersen and P.L. Kennedy. In press. Effectiveness of broadcast surveys for Northern Goshawks: considerations of breeding phase, detec- tion distance, and effective area surveyed. /. Wildl. Manag.: In press. Rosenstock, S.S., D.R. Anderson, K.M. Giesen, T. Leu- KERING, and M.F. Carter. 2002. Landbird counting techniques: current practices and an alternative. Auk 119:46-53. Squires, J.R. and R.T. Reynolds. 1997. The Northern Goshawk {Accipiter gentilis). In A. Poole and F. Gill, [Eds.], The birds of North America, No. 298. The Birds of North America, Inc., Philadelphia, PA U.S.A. Thompson, W.L. 2002. Towards reliable bird surveys: ac- counting for individuals present but not detected. Auk 119:18-25. Watson, J.W., D.W Hays, and DJ. Pierce. 1999. Efficacy of Northern Goshawk broadcast surveys in Washing- ton State. J. Wildl. Manag. 63:98-106. Received 28 January 2004; accepted 15 May 2005 Associate Editor: Clint W. Boal J Rapt&rRes. 39(3):303-309 © 2005 The Raptor Research Foundation, Inc. QUANTIFYING NORTHERN GOSHAWK DIETS USING REMOTE CAMERAS AND OBSERVATIONS FROM BLINDS Andi S. Rogers^ Arizona Game and Fish Department, 3500 South Lake Mary Road, Flagstaff, AZ 86001 U.S.A. Stephen DeStefano U.S. Geological Survey, Massachusetts Cooperative Fish and Wildlife Research Unit, Holdsworth Natural Resource Center, University of Massachusetts, Amherst, MA 01003 U.S.A. Michael F. Ingraldi Arizona Game and Fish Department, Research Branch, 2221 West Greenway Road, Phoenix, AZ 85023 U.S.A. Abstract. — Raptor diet is most commonly measured indirectly, by analyzing castings and prey remains, or directly, by observing prey deliveries from blinds. Indirect methods are not only time consuming, but there is evidence to suggest these methods may overestimate certain prey taxa within raptor diet. Remote video surveillance systems have been developed to aid in monitoring and data collection, but their use in field situations can be challenging and is often untested. To investigate diet and prey delivery rates of Northern Goshawks {Accipiter gentilis) , we operated 10 remote camera systems at occupied nests during the breeding seasons of 1999 and 2000 in east-central Arizona. We collected 2458 hr of useable video and successfully identified 627 (93%) prey items at least to Class (Aves, Mammalia, or Reptilia). Of prey items identified to genus, we identified 344 (81%) mammals, 62 (15%) birds, and 16 (4%) reptiles. During camera operation, we also conducted observations from blinds at a subset of five nests to compare the relative efficiency and precision of both methods. Limited observations from blinds yielded fewer prey deliveries, and therefore, lower delivery rates (0.16 items/hr) than simultaneous video footage (0.28 items/hr) . Observations from blinds resulted in fewer prey identified to the genus and species levels, when compared to data collected by remote cameras. Cameras provided a detailed and close view of nests, allowed for simultaneous recording at multiple nests, decreased observer bias and fatigue, and provided a permanent archive of data. Key Words; Northern Goshawk, Accipiter gentilis; prey-delivery rate, diet, remote camera; video surveillance. CUANTIFICACION DE LA DIETA DE ACCIPITER GENTILIS UTILIZANDO CAMARAS DE VIDEO CON SISTEMA REMOTO DE VIGILANCIA Y OBSERVACIONES DESDE UN ESCONDITE Resumen. — Comunmente la dieta de las aves rapaces es medida indirectamente por medio de analisis de egagropilas y restos de presas, o directamente por medio de observaciones de entregas de presa desde un escondite de observacion. Los metodos indirectos no solo toman mucho mas tiempo sino que tambien existe evidencia que sugiere que estos metodos pueden sobre-estimar la importancia de ciertos taxa de presa en la dieta de las rapaces. Se han desarrollado sistemas remotos de vigilancia con camaras de video para ayudar con la observacion y la recoleccion de datos, pero su uso en situaciones de campo puede ser dificil y en muchos casos no es un metodo probado. Para investigar la dieta y las tasas de entrega de presa de Accipiter gentilis, utilizamos 10 sistemas de camaras remotas en nidos activos durante las epocas reproductivas de 1999 y 2000 en el centro oriente de Arizona. Recolectamos 2,458 horas de video util y logramos identificar 627 (93%) restos de presa hasta Clase (Aves, Mammalia o Reptilia). Entre los restos de presa identificados a nivel de genero, identificamos 344 (81%) mamiferos, 62 (15%) aves y 16 (4%) reptiles. Durante la operacion de las camaras tambien hicimos observaciones desde escondites de un subgrupo de cinco nidos para comparar la eficiencia relativa y precision de los dos metodos. Las observaciones limitadas desde escondites rindieron menos entregas de presa y por lo tanto rindieron tasas de entrega mas bajas que la documentada simultaneamente con camaras. Los datos obtenidos mediante observaciones desde escondites indicaron una habilidad reducida de este metodo ^ Corresponding author’s email address: arogers@azfd.gov 303 304 Rogers et al. VoL. 39, No. 3 para identificar presas a nivel de genero y especie al ser comparados con los dates colectados de los videos de las camaras remotas. Las camaras produjeron una vista detallada y cercana de los nidos, permitieron la grabacion simultanea de varies nidos, redujeron el sesgo y la fatiga del observador y produjeron un archive permanente de dates. [Traduccion del Equipo Editorial] Information on diet is important in understand- ing aspects of avian ecology such as diet overlap among species, predation, and prey selection (Ro- senberg and Cooper 1990, Redpath et al. 2001). Diet assessment in raptors is usually done indirect- ly, by recovering pellets and prey remains, or di- rectly, by observing prey deliveries from blinds; however, accurate estimates of raptor diet can vary depending on the technique employed (Marti 1987). Indirect diet assessment can provide quantitative and qualitative information because raptors often leave behind undigested remnants of bones, feath- ers, and keratinous material as pellets, or as prey remains (Reynolds and Meslow 1984, MacLaren et al. 1988, Steenhof and Kochert 1988, Boal and Mannan 1994). However, prey remains and pellets may bias the representation of certain prey items (e.g., bird feathers are more easily detected than small bones); therefore, avian prey may be over- represented in raptor diet (Simmons et al. 1991, Bielefeldt et al. 1992). Marti (1987) suggested that pellet analysis is accurate only for raptor species that swallow their prey whole. Loss of prey remains to scavengers, investigator disturbance in the nest- ing area, and miscounting of remnant and incom- plete remains may also bias or limit results. Direct observation of raptors is a more accurate method for investigating diet in species that do not swallow their prey whole. Observations can be made from a blind within the nesting area; how- ever, observations near nests can disturb hawks, are labor intensive and require dawn to dusk obser- vations to obtain complete samples. In addition, direct observation requires positioning of the blind so that a view inside the nest bowl is possible (Col- lopy 1983). A more recent technology for studying diet in- volves remote cameras at raptor nests (Ouchley et al. 1994, Booms and Fuller 2003, Lewis 2004a). Ad- vantages of video surveillance for measuring diet include a reduction in observer bias and fatigue, minimal impact on an animal’s behavior, detailed information on diet composition, and an archival record of footage (Kristan et al. 1996, Stewart et al. 1997, Delaney et al. 1999). Lewis et al. (2004b) compared three methods for assessing raptor diet: video recording, pellet anal- ysis, and prey remain analysis. They found that quantifying prey using either prey remains or pel- let analysis did not provide as complete a descrip- tion of diet when compared to remote cameras. They did not, however, compare observations from blinds to remote cameras. In this paper, we de- scribe a camera system, monitoring, and data col- lection using remote video technology, and discuss advantages and disadvantages. In addition, we con- ducted limited observations from blinds at five nests and simultaneously collected data with re- mote cameras to compare the two methods. Study Area We conducted this study on the Sitgreaves portion of the Apache-Sitgreaves National Forest in east-central Ar- izona. The Sitgreaves portion encompasses ca. 350 800 ha (elevation = 1768-2417 m) and is located atop the Mo- gollon Rim on the southern edge of the Colorado Pla- teau. The Mogollon Rim is a large glacial escarpment that extends east across central Arizona into New Mexico. The Mogollon Rim edge has deep drainages with mixed- conifer communities of Douglas-fir {Pseudotsuga menzie- sii), white fir {Abies concolor), trembling aspen (Populus tremuloides) , ponderosa pine {Pinus ponderosa) , New Mex- ican locust {Robinia neomexicana) , and Gambel oak (Quer- cus gambelii; Brown 1994). Ridgetops are dominated by ponderosa pine forest. Methods We chose video monitoring as the primary method to quantify the diet of breeding Northern Goshawks {Acap- iter gentilis) in east-central Arizona (Rogers et al. in press). During the breeding seasons of 1999 and 2000, we ran- domly selected 10 nests (four in 1999 and six in 2000) from a pool of known territories {N =48). During June 1999 and 2000, we mounted EOD-1000 Electro-optics® remote cameras (Electro-Optics, St, Louis, MO U.S.A.) when nestlings were between 4-7 d old (nestlings were shaded during camera installation). Cameras ran from 22 June-18 July 1999 and 6 June-31 July 2000. We needed a minimum of three people for camera placement with a mean setup time of 110 min per nest (range = 80-132 min). Nest trees were ponderosa pine or Douglas-6r, and nest heights were ca. 20 m above ground. Cameras were 3.5 X 12 cm and equipped with 3.6 mm lenses. Each camera had 380 lines of resolution and a one-lux digital color system. During installation, the ground crew viewed the nest using a Broksonic D.C. TV/ VCR combination (Broksonic Corporation of America, September 2005 Techniques 305 Figure 1. Schematic of camera system for monitoring Northern Goshawk nests in east-central Arizona in 1999 and 2000. Table 1. Cost of video surveillance equipment used for a diet study of Northern Goshawks in Arizona during breeding seasons 1999 and 2000. Prices based on 1999 retail costs associated with assembly of one system. Component Approximate Cost ($ US) VHS time-lapse recorder® 675 Remote video camera 250 DC television monitor 190 Rechargeable battery 180 2-amp battery charger‘s 85 Coaxial and power cables and connectors 80 50 caliber ammunition can/locks and cables 50 Total 1470 “ Cost of Panasonic and Sony VHS recorder were averaged (Pan- asonic = $810. 00/Sony = $520.00). ^ TV/VCR combo used for multiple units. New York, NY U.S.A.) while a person in the nest tree positioned the camera. Once positioned, we secured the camera to the trunk of the tree or an overhanging branch. The goal in camera placement was a field of view that contained the entire nest structure and focused on the nest bowl. This was achieved by positioning cameras about 3 m away from nests at about a 45° angle to the nest structure (Fig. 1 ) . Cameras were connected to 75 m of durable telephone-power cord and coaxial video cable (copper coated RG-59) tacked along the trunk of the tree. Camera cords were attached to a Panasonic® AG- 1070 DC (Panasonic, Secaucus, NJ U.S.A.) or Sony® SVT- DL224 (Sony, Park Ridge, NJ U.S.A.) time-lapse VHS re- corder, which were placed at least 50 m from the nest tree. Both time-lapse VCR models were industrial grade, 12-volt DC models with time-lapse programming capabil- ity. VCRs were housed in military ammunition cans (20 mm) and powered by one 12-volt, 64 amp-hr sealed Op- tima® (Optima, Denver, CO U.S.A.) rechargeable lead acid battery (22 kg each). Batteries were kept dry under a plastic bin. Ammunition cans were locked, and all ground equipment was secured to a tree. Finally, we cov- ered equipment with forest litter for shade and camou- flage. Cost for one complete system was about $1470 US (Table 1). We programmed VCRs to record 5 frames/ sec, which provided up to 24 hr of footage per videotape. We re- corded activity at each nest in a 2-d sequence (12 hr/d), and cameras recorded 6 of 7 d of the week. We recorded from 0450-1650 H on day one and 0800-2000 H on day two. During 1999, all batteries and tapes were changed at night to reduce disturbance to hawks. In 2000, battery and tape changes occasionally occurred before nightfall, but only if ground equipment was located out of sight of nests. No more than 5 min were spent within nest stands changing batteries and tapes every other night or day. We continued to record video at nests until fledgling Northern Goshawks did not receive prey deliveries for 2 consecutive days. We viewed video footage on a 19 in Toshiba® television (Toshiba, Irvine, GA U.S.A.) with a JVC® SuperVHS VCR (JVC, Wayne, NJ U.S.A.). Prey items were identified to Class (Mammalia, Aves, or Reptilia), genus and species when possible, or classified as unknown. Prey items iden- tified to class only were characterized as small, medium, or large based on a priori size class categories from Cock- rum and Petryszyn (1992) and Dunning (1993). To compare methods, we observed goshawks (2000 breeding season) at a subset of five nests from blinds constructed 25—40 m from nests. We erected blinds on ground prior to sunrise before each observation period. Blinds were constructed of camouflage heavy-duty canvas with screen vtindows in all directions, which allowed for observation of hawks within the immediate nesting area. We used 8 X 32 binoculars and a 20 X 60 spotting scope to count and identify prey items delivered to nests. Items were identified to Class (Mammalia, Aves, or Reptilia) and to genus and species when possible. We initiated blind observations when nestlings were between 8-12 d old and continued observing prey deliveries until young fledged. We conducted observations in 3-4 hr blocks each day starting at sunrise, and blinds were disassembled upon completion of each observation period. We allowed a 20-min acclimation period for adults and young before beginning observation. Observations from blinds were done in conjunction with video monitoring to compare accuracy of the two methods. In addition, observations from blinds and video reviewing were done by one per- son (Rogers) to minimize bias. Results Camera Results. Adult Northern Goshawks ac- tively defended the nest while we placed cameras in nest trees. However, adult females returned to the nest within 10-20 min after we vacated terri- 306 Rogers et al. VoL. 39, No. 3 tories, as documented from video. In addition, we had no nest abandonment due to camera pres- ence, and eight of 10 nests were successful (i.e., fledged young). One nesting attempt failed due to adult female mortality (Bloxton et al. 2002) , and one was depredated by a Great Horned Owl {Bubo virginianus) . Adults did not flush during nighttime battery and tape changes. Adults flushed infrequently during daytime changes because we were out of sight of the nest tree. We collected 2458 hr of usable video footage, and about 500 hr were spent viewing tapes to iden- tify and quantify prey items. Approximately 50 hr were spent changing batteries and tapes, excluding travel time. We documented 676 prey deliveries from camera footage. Of these, we identified 627 (93%) prey items to Class (Aves, Mammalia, or Reptilia) and observed a mean delivery rate of 0.28 (SE = 0.02) prey items/hr. We were able to iden- tify, at least to genus, 422 (62%) of all prey items. Of items identified to at least genus, 344 (81%) were mammals, 62 (15%) were birds, and 16 (4%) were lizards. Direct Observations. Because blinds were con- structed before sunrise, adults infrequently flushed from nests. However, during disassembly and exit- ing nest territories, adults actively defended nests. When adults did flush from nests prior to an ob- servation period, they returned to the nest within about 10 min. We observed goshawks for a total of 43 hr at five nests. We viewed seven prey deliveries, all of which were identifiable to Class. Mean prey delivery rate observed from blinds was 0.16 (SE = 0.06) items/hr. Camera Versus Direct Observation. Camera foot- age yielded a higher total number of prey items delivered; therefore, our estimated prey delivery rate derived from video footage was higher than that derived from direct observation. An important result was that the camera footage revealed 12 de- liveries during our observation period in which we visually documented only seven deliveries. Accura- cy of prey identification to class was 100% using both methods, but we were able to document 58% of all prey to genus and/or species from the video footage compared to 0% from direct observations. Discussion Use of remote camera systems is becoming a popular technique in wildlife studies, especially as equipment costs decrease. Eor example, the video surveillance equipment used in Lewis et al. (2004a) was over $2000 US and did not include batteries and chargers. Our equipment was similar to Lewis’s, but the cost was $1470 US, which in- cluded batteries and chargers. Cameras have been used to monitor diet, pre- dation events, and various behaviors of many spe- cies of wildlife (e.g., Wisniewski 1983, Sykes et al. 1995, Hughes and Shorrock 1998, King et al. 2001). Responses to camera installation may vary by species and individuals, timing of camera place- ment during the nesting season, and length of time needed for camera installation. Several work- ers reported no sign of nest abandonment due to cameras (Estes and Mannan 2003, Booms and Ful- ler 2003, Lewis et al. 2004a). However, Cain (1985) reported nest abandonment by Bald Eagles (Hal- iaeetus leucocephalus) due to camera installation. During our study, goshawks were distressed when cameras were installed, but did not seem to be af- fected by camera presence. In videos, adult and juvenile goshawks occasionally could be seen look- ing up at cameras, and there were several occasions when hawks perched directly below or on cameras. Goshawks were also distressed during our obser- vations from blinds, especially during our exit from territories, which suggested that direct observa- tions were more stressful to the nesting hawks than use of cameras. Remote cameras can facilitate sampling for ex- tended periods of time with a reduction in observ- er bias (Delaney et al. 1999) . In contrast, observa- tions from blinds are often done by more than one person, which increases the risk of observer bias (Boal and Mannan 1994). Video monitoring can also increase daily coverage because unmanned units can operate continuously. For example, we were able to collect nearly 2500 hr of observations in 2 yr with 10 cameras, whereas Boal (1993) and Boal and Mannan (1994) collected 1500 hr of ob- servation from blinds, which required three field assistants per year for 3 yr. Observer fatigue could also bias the results based on direct observation, and the cost of labor would be high. Using remote cameras, we were able to record prey deliveries at the nest for up to 1 mo after Northern Goshawk young fledged (Rogers et al. in press). In contrast, observations from blinds and pellet and prey remains collection are often dis- continued shortly after young fledge (MacLaren et al. 1988, Seguin et al. 1998). Although prey deliv- ered to nests during branching and fledgling stag- es often occurred out of camera range, allowing September 2005 Techniques 307 Figure 2. Images of Northern Goshawk nests taken from video footage in east-central Arizona: (a) Female goshawk feeding 5-d-old young, (b) Golden-mantled ground squirrel and plucked Stellar’s Jay in the nest with 25-30 d old goshawk young. 308 Rogers et al. VoL. 39, No. 3 cameras to operate longer provided additional qualitative information on post-fledgling diet. Ad- ditional advantages of cameras include decreased frequencies of observer entrances and exits within territories. We spent no more than 5 min in nest stands every other day and usually did not flush adults from nests. We strongly recommend chang- ing batteries and tapes at night, or alternatively lo- cating ground equipment 50 m or more from nest trees. Most importantly, using remote cameras greatly increased our ability^ to identify genus and species of prey delivered to nests. With cameras, we were able to see shape and color of most prey items (Fig. 2). Mammals were easiest to identify to genus and species due to their size and distinctive pelage, as well as the ability to see feet and tails. Small birds were the most difficult to identify, but the ability to see feathers, and hence make an identification, distinctly increased if the adults plucked avian prey in the nest. Our data indicated that observations from blinds resulted in underestimates of prey numbers and delivery rates, but this needs further investigation. Prey items were missed in two ways during obser- vations from blinds. First, on some occasions we failed to notice small prey items brought by the female because we were focusing on identifying an item brought previously by the male. Second, we missed some prey items that were delivered to nests after dark or prior to daybreak. We did not use these items in calculating prey delivery rates, but included them in total prey deliveries. Without the ability to play back the videotape, we would not have noticed these prey items. A final advantage of video monitoring was the ability to record infrequent behavioral events. For example, during the 1999 nesting season, we doc- umented an attempted predation by a Red-tailed Hawk (Buteo jamaicensis), and in 2000, we recorded a bobcat (Lynx rufus) scavenging prey from a nest that had already fledged young. There are some limitations and constraints to us- ing camera systems. We experienced technical dif- ficulties including rodent and ungulate damage to cords, battery failure, loose connections, and water damage. In addition, when cameras’ angles were >45° to the nest, the view was often obstructed by the adult female’s back. To alleviate this problem, we searched for alternative branches or nearby trees that allowed for a 45° angle to the nest bowl. We recommend placing cameras opposite the di- rection of the adult flight pathway to the nest. Therefore, observing adult movement patterns near nests before camera placement is recommended. Video monitoring involves a relatively high initial cost. Also, as of 2000, no audio capability was avail- able within a waterproof system. Thus, collecting data on vocalizations during prey deliveries was not possible. One additional disadvantage of video tech- nology is the additional time required to transcribe video data. Even though tapes were fast forwarded during non-prey delivery times, it took ca. 1 hr of viewing to transcribe data for every 5 hr of video footage collected. We suggest viewing collected tap- es daily to minimize backlog and to allow research- ers to become aware of system problems before data collection is complete. In conclusion, we think remote cameras allowed us to collect more accurate diet data than if we would have solely used blind observations. Camera- monitoring systems are efficient, relatively nonin- vasive tools for quantifying diet and behavior of raptors. Acknowledgments This study was funded by the Arizona Game and Fish Department and the Arizona Cooperative Fish and Wild- life Research Unit at the University of Arizona. We thank Patricia Kennedy, Michael Collopy, and Robert Lehman for their earlier reviews of this manuscript. We also thank the numerous field assistants associated with this research and Susan MacVean for the Spanish translation of the abstract. Literature Cited Bielefeldt, J., R.N. Rosenfield, andJ.M. Papp. 1992. Un- founded assumptions about diet of the Cooper’s Hawk. Condor 94:427—439. Bloxton, T.D., A.S. Rogers, M.F. Ingraldi, S. Rosen- stock, J.M. Marzluff, and S. Finn. 2002. Possible choking mortalities of adult Northern Goshawks. J Raptor Res. 36:141-143. Boat, C.W. and R.W. Mannan. 1994. Northern Goshawk diets in ponderosa pine forests on the Kaibab plateau. Stud. Avian Biol. 16:97-102. Booms, T.L. and M.R. Fuller. 2003. Time-lapse video sys- tem used to study nesting Gyrfalcons. f. Field Ornithol 74:416-422. Brown, D.E. 1994. Biotic communities: southwestern United States and northwestern Mexico. University of Utah Press, Salt Lake City, UT U.S.A. Cain, S.L. 1985. Nesting activity time budgets of Bald Ea- gles in southeast Alaska. M.S. thesis, Univ. Montana, Missoula, MT U.S.A. CocKRUM, E.L. and Y. Petrvszyn. 1992. Mammals of the southwestern United States and northwestern Mexico Treasure Chest Publications, Tucson, AZ U.S.A. September 2005 Techniques 309 COLLOPY, M.W. 1983. A comparison of direct observa- tions and collections of prey remains in determining the diet of Golden Eagles. J. Wildl. Manag. 47:360- 368. Delaney, D.K., T.G. Grubb, and D.K. Garcelon. 1999. An infrared video camera system for monitoring di- urnal and nocturnal raptors./. Raptor Res. 32:290-296. Dunning, J.B., (Ed.). 1993. CRC handbook of avian body masses. CRC Press Inc., Boca Raton, FL U.S.A. Estes, W.A. and R.W. Mannan. 2003. Feeding behavior of Cooper’s Hawks at urban and rural nests in south- eastern Arizona. Condor 105:107—116. Hughes, A.G. and G. Shorrock. 1998. Design of a du- rable event detector and automated video surveil- lance unit. /. Field Ornithol. 69:549—556. King, D.I., R.M. DeGraaf, RJ. Champlin, and T.B. Cham- PLIN. 2001. A new method for wireless video monitor- ing of bird nests. Wildl. Soc. Bull. 29:349—353. Kristan, D.M., R.T. Golightly, Jr., and S.M. Tomkiew- ICZ, Jr. 1996. A solar-powered transmitting video cam- era for monitoring raptor nests. Wildl. Soc. Bull. 24: 284-290. Lewis, S.B., P. DeSimone, M.R. Fuller, and K. Titus. 2004a. A video surveillance system for monitoring raptor nests in a temperate rainforest environment. Northwest Sci. 78:70-74. , M.R. Fuller, and K. Titus. 2004b. A comparison of 3 methods for assessing raptor diet during the breeding season. Wildl. Soc. Bull. 32:373-385. MacLaren, P.A., S.H. Anderson, and D.E. Runde. 1988. Food habits and nest characteristics of breeding rap- tors in southwestern Wyoming. Great Basin Nat. 48: 548-553. Marti, C.D. 1987. Raptor food habits studies. Pages 67- 80 in B.A. Giron Pendleton, B.A. Millsap, K.W. Cline, and D.M. Bird [Eds.], Raptor management tech- niques manual. National Wildlife Federation, Wash- ington, DC U.S.A. OucHLEY, K., R.B. Hamilton, and S. Wilson. 1994. Nest monitoring using a micro-video camera. J. Field Orni- thol. 65:410-412. Redpath, S.M., R. Clarke, M. Madders, and SJ. Thir- GOOD. 2001. Assessing raptor diet comparing pellets, prey remains, and observational data at hen harrier nests. Cowdor 103:184-188. Reynolds, R.T. and E.C. Meslow. 1984. Partitioning of food and niche characteristics of coexisting Accipiter during breeding. Auk 101:761-779. Rogers, A.S., S. DeStefano, and M.F. Ingraldi. 2005. Diet, prey delivery rates, and prey biomass of North- ern Goshawks in east-central Arizona. Stud. Avian Biol. 31: In press. Rosenberg, K.V. and R.J. Cooper. 1990. Approaches to avian diet analysis. Stud. Avian Biol. 13:80-90. Seguin, J.F., P. Bayle, J.C. Thibault, J. Torre, and J.D. Vigne. 1998. A comparison of methods to evaluate the diet of Golden Eagles in Corsica./. Raptor Res. 32:314— 318. Simmons, R.E., D.M. Avery, and G. Avery. 1991. Biases in diets determined from pellets and remains: correc- tion factors for a mammal and a bird-eating raptor./. Raptor Res. 25:63-67. Steenhof, K. and M.N. Kochert. 1988. Dietary respons- es of three raptor species to changing prey densities in a natural environment./. Anim. Ecol. 57:37-46. Stewart, P.D., S.A. Ellwood, and D.W. MacDonald. 1997. Remote video-surveillance of wildlife — an intro- duction from experience with the European badger Meles meles. Mammal Review 27:185-204. Sykes, P.W., W.E. Ryman, C.B. Kepler, and J.W. Hardy. 1995. A 2-hour remote surveillance system for terres- trial wildlife studies./. Field Ornithol. 66:199-211. Wisniewski, L. 1983. Remote video monitoring of eagles and falcons. Pages 58—61 m T.M. Ingram [Ed.], Bald Eagle restoration. Eagle Valley Environmentalists, Ap- ple River, IL U.S.A. Received 26 March 2004; accepted 15 December 2004 Guest Editor: Patricia L. Kennedy / Raptor Res. 39(3) :310-323 © 2005 The Raptor Research Foundation, Inc. TEMPORAL PATTERNS OF NORTHERN GOSHAWK NEST AREA OCCUPANCY AND HABITAT: A RETROSPECTIVE ANALYSIS Steven M. Desimone ^ and Stephen DeStefano^ U.S. Geological Survey, Oregon Cooperative Wildlife Research Unit, 104 Nash Hall, Oregon State University, Corvallis, OR 97331 U.S. A Abstract. — We studied occupancy and habitat associations of Northern Goshawks {Accipiter gentilis) at nest areas in south-central Oregon in 1992-94. We surveyed 51 pre-1992 nest areas (i.e., historical breeding areas first discovered during 1973-91) for goshawks and used aerial-photograph interpretation to document forest cover conditions and changes over time between areas that were occupied by gos- hawks and those where we did not detect goshawks (no-response sites). We also surveyed for new nests during 1992-94. Of 38 occupied nests first found in 1992-94 (i.e., post-1992 nest areas), 86% (33/38) were in mid-aged (mean stand DBH 23-53 cm, <15 trees/ha >53 cm DBH) or late (>15 trees/ha >53 cm DBH; mean stand DBH >53 cm) closed (>50% canopy closure) structural-stage forest. Occupancy of historical (pre-1992) nest areas by goshawks was 29% (15/51). Of 46 pre-1992 nest areas that we examined for habitat change, 15 were occupied by goshawks in 1994 and had more mid-aged closed and late closed forest in 12-, 2T, 52-, 120-, and 1 70-ha circular areas centered on nest locations than did 31 no-response areas. There was no difference in the amount of late closed and mid-aged closed forest in pre-1992 nest areas compared with occupied pre-1992 nest areas. A logistic regression model for all occupied nest areas confirmed that late closed and mid-aged closed forest variables were impor- tant indicators of forest conditions that supported breeding pairs. Goshawks were more likely to persist in the historical nest areas that had about 50% of mature and older closed-canopy forest within the 52- ha scale. We recommend retaining existing late closed, late open, and mid closed structure within 52- ha scale of the nest site. Moreover, late closed and mid closed structure combined should not fall below 50% within the 52-ha scale and should exceed 40% within the 1 70-ha scale surrounding the nest site. To optimize conditions for breeding goshawks, we recommend retaining large trees (>53 cm DBH) to help preserve stand integrity, maintain closed canopies, and provide connectivity to alternative nest sites within nest areas. Key Words; Northern Goshawk, Accipiter gentilis; habitat, historical nest areas', landscape change, Oregon. PATRONES TEMPORALES DE OCUPACION DE areas DE NIDIEICACION Y HABITAT DE ACCIP- ITER GENTILIS: UN ANALISIS RETROSPECTD/O Resumen. — Estudiamos la ocupacion y las asociaciones de habitat de Accipiter gentilis en areas de nidifi- cacion del centro-sur de Oregon entre 1992 y 1994. Tambien censamos 15 areas de cria historicas descubiertas entre 1973 y 1991 (i.e., nidificacion pre-1992), y usamos fotografias aereas para documen tar las condiciones de cobertura de bosque y cambios en el tiempo entre areas que estaban ocupadas por esta especie y areas en las que no la detectamos (sitios sin respuesta). Tambien realizamos censos para buscar nidos nuevos entre 1992 y 1994. De 38 nidos activos encontrados por primera vez entre 1992 y 1994 (i.e. nidificacion post-1992), el 86% (33/38) se encontro en bosques de sucesion media (promedio de DAP 23-53 cm, <15 arboles/ha >53 cm DAP) o bosques cerrados antiguos (>15 arboles/ha >53 cm DAP; promedio de DAP >53 cm; >50% de cobertura del dosel). La ocupacion de las areas de nidificacion historicas (pre-1992) por parte de A. gentilis ine del 29% (15/51). De 46 sitios de crfa pre- 1992 para los cuales evaluamos los cambios en el habitat, 15 estuvieron ocupados en 1994 y presentaron mayor cantidad de bosques cerrados de edad media y bosques antiguos en areas circulares de 12, 24, 52, 120 y 170 ha centradas en sitios en donde se ubicaban nidos, que 31 sitios sin respuesta. No existio ^ Corresponding author’s present address; Washington Department of Fish and Wildlife, 600 Capitol Way North, Olympia, WA 98501-1091 U.S.A.; email address; desimsrad@dfw.wa.gov ^ Present address: U.S.G.S. Massachusetts Cooperative Fish and Wildlife Research Unit, Holdsworth Natural Resources Center, University of Massachusetts, Amherst, MA 01003 U.S.A. 310 September 2005 Conservation 311 diferencia en la cantidad de bosques cerrados de edad media y bosques antiguos entre areas de nidi- ficacion pre-1992 en comparacion con las areas ocupadas pre-1992. Un modelo de regresion logistica que incluyo todas las areas de nidificacion ocupadas confirmo que las variables de los bosques cerrados de edad media y sucesion tardia fueron indicadoras importantes de las condiciones del bosque propicias para las parejas reproductivas. Las aves presentaron mayor probabilidad de persistir en las areas de nidificacion historica que presentaban aproximadamente el 50% de bosques maduros antiguos de dosel cerrado a la escala de 52 ha. Recomendamos mantener la estructura de bosques antiguos cerrados y abiertos y bosques de edad media cerrados en las 52 ha circundantes a los sitios de nidificacion. Ademas, la estructura combinada de bosques cerrados antiguos y de edad media no debe caer por debajo del 50% a la escala de 52 ha y no debe exceder el 40% en la escala de las 170 ha circundantes a los sitios de nidificacion. Para optimizar las condiciones para la nidificacion de A. gentilis, recomendamos man- tener arboles grandes (>53 cm DAP) para ayudar a preservar la integridad de los bosques, mantener doseles cerrados y proveer conectividad entre sitios de nidificacion alternatives ubicados en las mismas areas de cria. [Traduccion del equipo editorial] The ability of breeding pairs of Northern Gos- hawks (Accipiter gentilis; hereafter, goshawks) to per- sist in intensively managed and selectively harvest- ed forests over time is largely unknown. Evidence suggests tree harvest impacts nest site selection (Crocker-Bedford 1990, Penteriani and Faivre 2001), use (Woodbridge and Detrich 1994), and ultimately nesting persistence (Crocker-Bedford 1995). Penteriani and Faivre (2001) examined log- ging disturbance and habitat change over a limited time (6-11 yr) in a European shelterwood harvest regime, but the effects of habitat alteration in west- ern North American forests are not fully under- stood. Mature forest, consisting of large trees (di- ameter at breast height [DBH] >50 cm) and closed canopy cover (>50%), was demonstrated to be preferred by breeding goshawks for nest sites in western North America (e.g., Hayward and Escano 1989, Bull and Hohmann 1994, Squires and Rug- giero 1996, Daw and DeStefano 2001, McGrath et al. 2003). There has been concern and debate that gos- hawk populations in western North America may be declining in response to habitat alteration and loss of these forests (Kennedy 1997, DeStefano 1998, Smallwood 1998, Crocker-Bedford 1998). Us- ing aerial photographic records of timber harvest areas (Reutebuch and Gall 1990) on the Fremont National Forest and adjacent private forest lands dating from 1969-92, we evaluated temporal changes to forest structure around goshawk nests during 1992-94. Our objectives were to: (1) deter- mine if a random sample of historical goshawk nest areas (i.e., nests first found in 1973-91) were oc- cupied in 1994, (2) document post-1992 forest con- ditions and quantify change in forest cover on his- torical nest areas, and (3) compare 1994 forest cover between historical nest areas that were oc- cupied by goshawks between 1992 and 1994 to those historical nest areas where presence of gos- hawks was not detected. Study Area Research took place on the Fremont National Forest (FNF) and the Klamath Tree Farm of the Weyerhaeuser Company in south-central Oregon, encompassing >5000 km^. Elevations ranged from 1200-2200 m. Ponderosa pine {Pinus ponderosa) , white fir {Abies concolor) , and lod- gepole pine {P. contorta) were the dominant commercial tree species. Generally, large expanses of lodgepole pine interspersed with small stands of pure ponderosa pine on higher ground dominated the northern half of the study area; dry mixed-conifer stands (white fir, incense cedar [Libocedrus decurrens], ponderosa pine, and sugar pine [R lambertiana] ) dominated the southern half of the study area. Douglas-fir {Pseudotsuga menziesit) was rarely encountered or absent, and most adjacent private lands had extensive ponderosa pine plantations. Natural forest openings consisted of xeric rocky flats, which contained sagebrush {Artemisia spp.) and bitterbrush {Purshia triden- tata) near ponderosa pine and mixed-conifer stands, and moist meadows, which were typically associated with lod- gepole pine and had a vegetative cover of sedges ( Carex spp.), sagebrush, and willow {Salix spp.) next to peren- nial streams or springs. The landscape was a mosaic of forest cover types, containing two large burned areas from the 1950s and 1992, natural openings, and human- created openings. Dominant silvicultural practices on Forest Service lands were partial harvest, selective remov- al, and shelterwood treatments in mixed-conifer and ponderosa pine. All forest management terms used in this paper follow Helms (1998). Regeneration (clearcut) harvest was more typical in lodgepole pine habitat, al- though observational data and Forest Service records (Fremont National Forest Supervisor’s Office, Lakeview, OR U.S.A.) documented that regeneration harvest oc- curred in mixed conifer and mature ponderosa pine types. Private land management was dominated by mosdy early serai and some mid-seral plantations of P. ponderosa 312 Desimone and DeStefano VoL. 39, No. 3 in large clearcuts or past overstory removal with few scat- tered large seed trees. Forest Service management, reg- ulated timber harvest and aggressive fire suppression dates back >50 yr; selective railroad logging took place around 1920 (Hopkins 1979, Laudenslayer et al. 1989). Regional historical accounts state that ponderosa pine stands were typically composed of large trees with a mean DBH of 40-70 cm and basal area (BA) ranging from 13 to 23 m^/ha (Munger 1917), stands rarely encountered in managed forests during our study. Methods We defined nest site as the tree containing the occupied nest or the mapped location of the historically occupied nests and ^1 ha around the location. A nest area for this study was the area that we surveyed out to 1000 m (about 300 ha), centered on a nest site. We defined post-1992 nest areas as occupied (breeding) nest areas first discov- ered during our study, which was conducted during 1992-94. Historical nest areas were defined as pre-1992 nest areas if they were first discovered occupied 1973-91 by Forest Service or Weyerhaeuser personnel, or other researchers. Occupied nest areas were areas we surveyed during 1992-94, where at least one adult goshawk was present or actively nesting. For purposes of habitat- change comparisons in 1994, occupied nest areas were a subset of the historical pre-1992 nest areas that were found occupied in 1994. No-response areas were a subset of pre-1992 nest areas surveyed in 1994 that had no de- tections. Goshawk Nest Area Occupancy Surveys. We compiled a list of historical goshawk nest area locations from orig- inal data collected by Reynolds (1975, 1978), U.S. Forest Service (unpublished data, Fremont National Forest, Lakeview, OR U.S.A.), and Weyerhaeuser Company (un- published data, Klamath Tree Farm, Klamath Falls, OR U.S. A.) and evaluated each dataset based on quality of documentation (e.g., written reports, legal and area de- scriptions, mapped locations), observer reliability (e.g,, biologist or experienced observer) , and number of years the nest area was documented as occupied. Nest records were included if there was adequate documentation of a goshawk attending a nest structure, incubating, or if fledglings or nestlings were present at the nest site. Lo- cations meeting the above criteria were mapped, and for- est cover type was validated by aerial photograph or field examination before surveys commenced. We stratified sites into one of three forest cover types: dry mixed-co- nifer, ponderosa pine (<20% other tree species), and lodgepole pine (<20% other tree species). We broadcast conspecific vocalizations to elicit respons- es from nesting goshawks or fledglings from late May to early August 1992-94 (Kennedy and Stahlecker 1993, Joy et al. 1994). Surveys were centered on the last known occupied historical nest location, with at least 35-40 call- ing stations per survey area (see below), spaced 320 m apart and staggered on adjacent and parallel transect belts. To ensure coverage of potential nest areas, we ex- amined the literature for estimates of inter-nest distances between alternative nest sites (273 m in the Klamath NF, California [Woodbridge and Detrich 1994]; 266 m [Reyn- olds et al. 1994], and 489 m [Reynolds and Joy 1998] in Arizona; and 432 m in Utah [Dewey et al. 2003]), sizes of post-fledging family areas (PFA; ca. 168 ha, Kennedy et al. 1994), and the effective auditory range of the mega- phone (^150 m; Joy et al. 1994; S. Desimone unpubl. data) . Based on this information, we established our sur- vey area size as a circle with an approximate 1000-m ra- dius centered around the nest location (ca. 40 stations). This resulted in a search area of about 300 ha, nearly twice the area of mean PFA size reported by Kennedy et al. (1994). If a response was detected, we immediately searched the vicinity for an occupied nest. For those sur- veyed areas where there were no detections during the nestiing period (first survey), we resurveyed the area at least once in July— August during the fledgling period us- ing the same stations so that each “no-response” area was visited and surveyed at least twice in a season. We also conducted systematic and opportunistic searches (De- Stefano et al. 1994a, Daw et al. 1998) for new goshawk nests (i.e., post-1992) during May-August 1992-94. When surveying a known occupied nest area from the previous survey season (i.e,, 1992 or 1993), we used multiple ob- servers to conduct a silent search at the last known oc- cupied site to minimize disturbance. If there were no de- tections, we extended the search pattern by radiating out from the nest tree while using a combination of inter- mittent taped broadcast calls near the nest and regularly spaced calling stations. These areas had the same level of survey effort as the systematic searches: about 300 ha around the last known occupied nest. Vegetation Sampling. We used 1:12000 and 1:15 800 scale aerial photographs to describe and classify historical (1969-91) forest vegetation conditions and post-1992 conditions, obtained from the U.S. Forest Service and Weyerhaeuser Company for years 1969, 1972, 1976, 1978, 1980, 1983, 1988, and 1992 (the most recent available) for reference stands. Harvest inventory data from the Fremont National Forest were used to update 1988 and 1992 photos to 1994 conditions. We used a 3X Dietzgen stereoscope to delineate cover in an 11% random sample (25 of 227) of reference stand polygons representing the range of forest conditions and habitats on the 1992 photographs. The variable-plot veg- etation sampling method (Bell and Dilworth 1988) was used to verify the condition of these reference polygons on the ground. We sampled 7-11 plots (x = 8.1 plots/ reference polygon, SD = 1.8), 160 m apart, on a transect located through the longest axis of the habitat polygon or in parallel transects if the polygon was >300 m wide. Plots were measured for basal area (BA) using a 20-factor (ft^/acre, later converted to m^/ha) wedge prism at plot center to sort trees into diameter classes. We recorded DBH for all count trees by combining plots within a stand to determine trees/ha (TPH) and BA for each forest structure class. The stem count per sample point multi- plied by the BA factor equaled the total BA occupied by tree stems on a per ha basis (Bell and Dilworth 1988). We followed the U.S. Forest Service Region 6 Vegetation Structural Stage (VSS) guidelines for general forest cover type descriptions in eastern Oregon (U.S. Department of Agriculture 1994). We used two non-forest categories (open wet [moist meadows] and open dry [xeric flats]) and four forest structure categories (late, mid-age, early, very-early), combined with two canopy closure classes (<50% or September 2005 Conservation 313 Table 1. Forest structure classification for aerial photograph interpretation on the Fremont National Forest and adjacent private lands in Oregon U.S.A., based on mean tree diameter at breast height (DBH), mean canopy closure, and trees per ha (TPH) S;53 cm DBH (USDA 1994). Very early stage was forest regeneration or clearcut. Forest Structure DBH (cm) Canopy Closure (%) TPH >53 cm Late closed >53 >50 >15 Late open >53 <50 >15 Mid-aged closed 23-53 >50 <15 Mid-aged open 23-53 <50 <15 Early closed 12-23 >50 none Early open 12-23 <50 none Very early <12 <50 none >50%), to identify and delineate vegetative cover on ae- rial photographs. Stands were then typed into forest veg- etative cover classes based on total BA of trees per di- ameter class and TPH >53 cm (Table 1). We defined canopy closure as the amount of sky obscured by tree foliage and branches as measured by a Lemmon spheri- cal densitometer (Vales and Bunnell 1985). Canopy mea- surements were taken 5 m from plot center in four car- dinal directions, averaged, and mean percent canopy closure was calculated from all plots for the polygon. Following reference plot validation, all remaining hab- itat on photographs within a 170-ha circle around nest locations was delineated into vegetative cover polygons based on the validated reference plots and assigned veg- etation structure categories. When 1994 photographs were not available, the 1994 Fremont National Forest Harvest Inventory (U.S. Forest Service, Fremont NF, Lakeview, OR U.S.A.) was used to update the habitat con- dition. A 19% (AT — 102) random sample of polygons (N = 546), stratified by general forest cover type, was ground-verified using the same variable-plot sampling method outlined for reference stands. We assessed stand- typing accuracy by constructing an error matrix to deter- mine the accuracy of our photograph interpretation (De- simone 1997). To delineate historical forest conditions, we used U.S. Forest Service and Weyerhaeuser Company aerial pho- tographs (1:12000, 1:15 800, and 1:24000 scales) that represented stand conditions present in the year of the last known occupied nest. We extrapolated our reference set results to type stands into vegetative structure classes on the remaining historical photographs. All completed polygons were transferred to 1:24000 scale U.S. Geolog- ical Survey (USGS) quadrangle maps using a zoom trans- fer scope (Bausch and Lomb Corporation, Rochester, NY U.S.A.) and digitized into a Geographic Information Sys- tem, where area was calculated for each habitat polygon. Annual Variation in Occupancy. We estimated annual variation in occupancy by resident pairs of western North American goshawks by examining data from five other study areas (Table 2). We then compared the mean an- nual occupancy rates of goshawks from these five study areas to our findings for post-1992 nest areas and pre- 1992 (historical) nest areas assessed in 1994. Annual oc- cupancy was defined as the mean (SE) annual percent of occupied areas. We assumed (1) territory occupancy was determined using similar survey techniques with equal effort (Joy et al. 1994, Reynolds et al. 1994, Wood- bridge and Detrich 1994, Kennedy 1997) and (2) littie or no major stand disturbance or habitat alteration oc- curred within territories since discovery (S. Dewey, R Kennedy, R. Reynolds, and B. Woodbridge pers. comm.). An occupied territory was defined by all researchers as >:! adult goshawk present in or near the nest on >2 separate occasions during the breeding season and in- Table 2. Mean occupancy rates of Northern Goshawk nest areas (NA) among six concurrent studies in the western U.S., including occupied nest areas found on the Fremont National Forest (NF) and adjacent private lands 1992—94, Oregon, U.S.A. (this study). Occupancy is defined as a territory used regularly by at least one adult goshawk during the breeding season. The first year a nest was discovered is not included in the calculations. Study Area Source NA Mean SE Wars® Kaibab NF, AZ R.T. Reynolds pers. comm. 32 0.72 0.05 4-5 New Mexico Kennedy 1997 22 0.74 0.07 4-11 Klamath NF, CA Woodbridge and Detrich 1994 26 0.74 0.01 5-9 Utah Kennedy 1997 26 0.75 0.06 4-7 Malheur NF, OR S.K. Daw pers. comm. 33 0.66 0.02 2-4 Fremont NF, OR This study 20 0.79 0.04 2 ^ Number of years of occupancy data for known nests in the study area. 314 Desimone and DeStefano VoL. 39, No. 3 eluded pairs attempting to nest (Reynolds et al. 1994, Woodbridge and Detrich 1994, Kennedy 1997). Habitat Chaise Analysis. Using nest tree locations as nest area centers, we established five different radius cat- egories of 12, 24, 52, 120, and 170 ha. These areas had biological or managerial significance: 12 ha was recom- mended as a minimum nest area size for goshawks (Reyn- olds 1983, Reynolds et al. 1992); 24 ha was the size of goshawk habitat areas designated on the Fremont NF to protect nesting stands (U.S. Department of Agriculture [USDA] 1989); 52 ha was about the mean size of the ag- gregate of alternative nest areas associated with the pri- mary nest area of goshawks nesting in the Klamath NF in northern CA (Woodbridge and Detrich 1994); 120 ha was the area of old-growth habitat allocated for management of Pileated Woodpeckers (Dryocopus pileatus) on the Fre- mont NF (USDA 1989); and 170 ha was the size of the goshawk PEA (Reynolds et al. 1992, Kennedy et al. 1994). We conducted comparisons at both “disk” (12, 24, 52, 120, and 170 ha) and “ring” (the area between the 12- 24, 24-52, 52-120, and 120-170 ha disks) scales. Disks represent cumulative effects as scale increases, since smaller disks are included within the larger disks. Rings were tested individually so that influence of inner disks was removed (Ramsey et al. 1994, McGrath et al. 2003). We examined how forest structure around historical nest sites changed over time by calculating the % change for each vegetation cover (forest and non-forest) variable (equation 1): % Change = - Area^j^Tonic) / (1) AreafjisTORic^ X 100 where is the area of a cover category for 1994, and AreaffisTORic is the area of the same cover category in the year the site was last known to be occupied. This calculation was made for each of the paired sites for all scales of disks and rings. We used Wilcoxon signed-rank test for paired comparisons to test for changes in forest cover between pre-1992 and post-1992 conditions and Kruskal-Wallis one-way analysis of variance by ranks fol- lowed by multiple comparison tests for least significant difference (LSD) to test for differences in forest cover among pre-1992, pre-1992 occupied, and pre-1992 no- response nest areas (Conover 1980; JMP Statistical Soft- ware version 3.1, SAS Institute, Cary, North Carolina, U.S.A). Logistic Regression Model. We wanted to know the likelihood of predicting the suitability of historical nest areas by considering the amount of area of each vegeta- tion structure category (forest and non-forest) around occupied and no-response sites. Thus, we constructed a logistic regression model (Hosmer and Lemeshow 1989, Ramsey et al. 1994) using the binary response variable of occupied (F = 1) or no-response (F = 0) by goshawks in a historical nest area in 1994. The importance of a particular habitat variable was determined by a stepwise analysis (PROC LOGISTIC; SAS Institute, Inc., 1992). The alpha for entry level (p^) of the variable to be con- sidered for the model was 0.15 because we wanted to detect possible trends in the event of a nonsignificant P- value. Models were run for each of the five disks and four rings. The full model included all explanatory habitat variables (equation 2); logit P (Y) = -I- X VeryEarly + B 2 X EarlyClosed + X EarlyOpen + B4 X MidClosed + B^ X MidOpen + B^ X LateClosed -h By X LateOpen + Bg X OpenWet + B^ X OpenDry (2) where Bq is constant, and By through B^ are the coeffi- cients. The model was run in logit P(l) mode (stepwise descending) to calculate odds ratios for significant vari- able (s) associated with a nest area being occupied (i.e., F = 1). Interaction terms were evaluated in the final model. Results Nest Area Surveys. During 1992-94, we found 38 occupied goshawk nest areas (15 pre-1992 and 23 post-1992) that composed our sample of nests for vegetation analysis. Of the 51 pre-1992 nest areas we reexamined, 10 had evidence of nesting and five more had goshawks present, for a total of 15 historical occupied nest areas. Twelve of the 15 oc- cupied nest areas were contained wholly on Forest Service ownership, two were on Weyerhaeuser land, and one was mixed ownership. Of 36 no-re- sponse areas, 23 were on Forest Service, 1 1 on Wey- erhaeuser, and two were mixed ownership. We re- moved five nests from the historical sample for our vegetation analysis because of inadequate photo- graphic records. Therefore, of the remaining 46 pre-1992 nest areas surveyed to protocol, 15 were occupied, and 31 were no-response areas. Number of nestlings per nest was similar for pre-1992 areas occupied in 1994 and post-1992 nests (1.5 ± 1.2 [N = 10] and 1.4 ± 1.0 [N = 18] young/nest, re- spectively) . Annual Variation in Territory Occupancy. Of 38 occupied nest areas, we were able to consistently survey 20 for at least two seasons from 1992-94; these had a mean annual occupancy rate of 79% (SE = 4; Table 2). This was similar to the mean annual occupancy rates from five other concurrent studies in the western U.S. (73%, SE = 2, for Ari- zona, California, New Mexico, Utah, and Oregon; Table 2). Occupancy of all historical (pre-1992) nest areas surveyed in 1994 was 29% (15/51), which was significantly different from the occupan- September 2005 Conservation 315 cy rate for post-1992 areas (x^ = 12.4, 1 df, P = 0.0004) and substantially lower than reported in the literature (Table 2). For post-1992 nest areas, mean inter-alternative nest distance was 245 m (SE = 48, N = 23; no data for private lands). This was comparable to the in- ter-alternative nest distances reported in the liter- ature (Reynolds et al. 1994, Woodbridge and De- trich 1994, Dewey et al. 2003). Habitat Typing. Overall typing accuracy based on ground verification of reference polygons was 80%. Mid-age and late categories were 80-90% for reference polygons (Desimone 1997); we thought this was an acceptable rate to proceed with the analysis (Lillesand and Kiefer 1994). Our highest classification accuracies were for dry open and wet open non-forest categories (100% each) from ref- erence polygons, followed by late open and late closed forest structure (90% each), mid-aged open (84%), and mid-aged closed forest (80%). Early open and early closed forest structure was least ac- curately classified(67% and 69%, respectively). Forest Cover Distribution. For post-1992 nest ar- eas, 25 of 42 (60%) occupied nest trees were within late closed forest structure, and 11 of 42 (26%) were in mid-aged closed structure. Distribution of post-1992 and pre-1992 nest sites was similar among the three forest cover types: 56% versus 47% in mixed conifer, 24% versus 28% in lodge- pole pine, and 20% versus 25% in ponderosa pine, respectively. Habitat Change Analysis. Mean percent change of the seven forest-structure categories (Table 1) for pre-1992 nest areas over time occurred in all scales (i.e., five disks and four rings) (see Desi- mone 1997: Tables 11 and 12 for details). For disks, the largest increases were in the amounts for very- early (642%, SE = 93%) and early open (238%, SE = 17%) categories. The largest decreases over time were in the late open (—54%, SE = 3%), late closed (—49%, SE = 1%), and mid closed ( — 30%, SE = 3%) categories. The magnitude of the per- cent change decreased with increasing scale; for example, increase in very-early cover went from 742% to 640% to 435% at 12-, 52-, and l70-ha disk scales, respectively, while decreases in late open cover went from —58% to —56% to —47% for 12-, 52- and 170-ha scales, respectively. Similar results were noted for rings, although at lower magni- tudes. In Figure 1, we presented late closed, early open, and very-early structural stages because they rep- 40 35 ■; Late Closed 30 i 25 ■{ 20 -I $ 15 ■ ♦ Pre-'i99.' □ Occupied A No response ♦ 0 A. I ? T I •f t 12 24 52 120 170 Scale (ha) Figure 1, Mean (SE) area of late closed, early open, and very early structural stages among 5 circular analysis scales surrounding 46 historical (pre-1992; first discov- ered during 1973-91) goshawk nest sites in south-central Oregon, U.S.A.; 15 were occupied by goshawks and 31 had no evidence of occupancy in 1994. See text and Ta- ble 1 for further description of forest structure catego- ries. resented most confidence in correctly classifying habitat types, and therefore most confidence in de- tecting a decrease in area of highest suitable hab- itat (late closed) and an increase in area of known non-nesting habitat (early open and very early). For the 12-, 24- and 52-ha scales, mean percent late closed forest at all occupied nest areas in 1994 re- mained nearly the same as at pre-1992 areas (i.e., no significant difference) . However, mean percent late closed forest at no-response areas was about 316 Desimone and DeStefano VoL. 39, No. 3 80 ■ 70 ■ 60 50 40 .! 30 .| 20 ! 12h* • k d k. u I 10 i 0 60 50 • 40 30 ■ 20 . 10 ■ 0 • 60 ■ Very Early Early Open Mid-age Late Open Late Closed Mid Closed Closed + Late Closed Forest Structural Stage Figure 2. Mean (SE) distribution of forest structural stage categories (plus a combination of mid-aged and late closed canopy forest) at 12-, 52 -, and l70-ha scales sur- rounding 46 historical (pre-1992 — first discovered during 1973—91) goshawk nest sites in south-central Oregon, U.S.A.; 15 were occupied by goshawks and 31 had no evidence of occupancy in 1994. We omitted the 52- and 120-ha data because results were similar. See text and Ta- ble 1 for further description of forest structure catego- ries. Difference in grouped means assessed by Kruskal- Wallis test (a = 0.05) . Within each group, Fisher’s test of least significant difference for multiple comparisons was used; pairs within each forest structure stage not signifi- candy different share common letters. one-fourth to one-fifth the amount of late closed at pre-1992 and occupied pre-1992 areas (Fig. 1). With increasing scale, the mean proportion of ear- ly open structure at no-response nest areas was 4- 5 times greater than pre-1992 nest areas, and over twice that of occupied pre-1992 areas (Fig. 1). The mean proportion of very early stage increased with increasing scale and was 4-6 times greater at 12-, 24- and 52-ha scales for no-response areas than pre- 1992 and occupied pre-1992 areas (Fig. 1). For 120- and 170-ha scales, mean proportion of very early was about half that in occupied compared to no-response areas; pre-1992 areas had about 1/8*^ that of no-response areas. No-response areas (N = 31) showed significant changes in the general distribution of forest struc- ture compared to all pre-1992 areas and also dif- fered significantly from occupied pre-1992 areas for 12-, 52- and 1 70-ha disk scales (Kruskal-Wallis test, all P < 0.0454; Fig. 2). Histograms for 24- and 120-ha scales were not presented, as patterns in re- sults were similar but intermediate in values be- tween their adjacent scales. Mean proportion of area was significantly different for late closed for- est, mid-aged closed forest, early open forest, and very early among pre-1992, occupied pre-1992, and no-response areas (LSD for pairs of means, P < 0.05) . The greatest single change in a category was in the mean amount of late closed forest in pre- 1992 (x = 22-27% among all disk scales) and no- response areas (x = 6—8%, all disks; Kruskal-Wallis, P< 0.0003). At the 52-ha scale, the mean percent area of late closed forest (20%) for occupied pre-1992 areas remained similar to historical pre-1992 areas (24%; Fig. 2). Less than half of the mean area of mid- aged closed forest that once existed in pre-1992 areas (25%) occurred in no-response areas (12%). This corresponded with an increase in mean per- cent mid-aged open forest in no-response (24%) compared to the occupied pre-1992 areas (15%). In no-response areas, mean percent area of early open canopy forest was >4 times the historical mean (pre-1992) amount, and more than twice that of occupied pre-1992 areas (LSD test of means, P < 0.05). Very early mean percent area was significantly greater in no-response than oc- cupied pre-1992 areas (LSD test of means, P < 0.05; Fig. 2). We point out comparisons at the 52- ha scale because it represents, in theory, the ag- gregate of alternative nest sites for a nesting area, and the persistence of goshawk use or occupancy appears to be correlated with higher amounts of mature forest at about this scale (Woodbridge and Detrich 1994). Logistic Regression Model of Forest Structure Association. For post-1992 occupied nest areas, both late and mid-aged closed variables were as- sociated with the 52-ha disk model (drop in Devi- ance = 9-5; 1 df; P < 0.01) and the 24—52 ha ring model (drop in Deviance = 20.7; 1 df; P < 0.01), as described by the reduced model (Equa- tion 3): September 2005 Conservation 317 Table 3. Results of stepwise logistic regression analysis for occupied (F = 1) goshawk nest areas (A^ = 15), Fremont National Forest and adjacent private lands, Oregon, U.S.A., 1994. Stepwise entry level was at a = 0.15. Scales emanate from territory centers; ring size is the area between two concentric disk areas. Parameter estimates are natural log (In) of odds ratios. The interaction term (late closed X mid-aged closed) was not significant (P = 0.23). Disk Size® (ha) Variable Parameter Estimate SE Wald P-value Odds Ratio Estimate 95% Confidence Interval, of Odds Ratio 12 Intercept -83.9333 24.4576 11.7771 0.0006 — — Late closed 0.4771 0.1650 8.3594 0.0038 1.611 0.1537-0.8005 (1.166-2.227) Mid-aged closed 0.3344 0.1157 8.3554 0.0038 1.397 0.1076-0.5612 (1.114-1.753) 24 Intercept Late -46.5816 13.664 11.6205 0.0007 — 0.0804-0.4516 closed Mid-aged 0.2660 0.0947 7.8850 0.005 1.305 (1.084-1.571) 0.0522-0.2936 closed 0.1729 0.0616 7.8829 0.005 1.189 (1.054-1.341) 52 Intercept -21.9700 6.3879 11.8290 0.0006 — — Late closed 0.1131 0.0401 7.9426 0.0048 1.120 0.0345-0.1917 (1.035-1.211) Ring Size^ (ha) Mid-aged closed 0.0818 0.0307 7.1046 0.0077 1.085 0.0216-0.1420 (1.022-1.155) 12-24 Intercept -85.3932 25.1893 11.4925 0.0007 — — Late closed 0.5126 0.1798 8.1303 0.0044 1.670 0.2109-0.9366 (1.235-2.551) Mid-aged closed 0.3175 0.1215 6.8264 0.009 1.374 0.0984-0.5913 (1.103-1.806) 24-52 Intercept -33.9116 10.7673 9.9193 0.0016 — — Late closed 0.1754 0.0691 6.4489 0.0111 1.192 0.0533-0.3301 (1.055-1.391) Mid-aged closed 0.1423 0.0579 6.0437 0.0140 1.153 0.0376-0.2719 (1.038-1.313) ■^Forest structure classes were not significantly associated with the 120- and l70-ha disk scales. ’’Forest structure classes were not significantly associated with the 52-120 and 120-170 ring scales. logit (1) — Bq + (late closed) + B 2 (mid-aged closed) There was a strong association between nest area occupancy and both late closed and mid-aged closed forest at the 12, 24, and 52 ha scales (Table 3) . At the 12-ha nest area scale, the odds that a site was occupied increased by 61% (odds ratio 1.61) for each unit ( 1 ha) increase of late closed forest habitat, while holding the mid-aged closed forest variable constant. For each unit increase of mid- aged closed forest habitat, the odds that a site was occupied increased by 37% (odds ratio 1.37), while holding the late closed forest variable constant. The reduced model was also significant for the 12—24 and 24—52 ha rings. The stepwise descend- ing model procedure did not yield a significant model for any variables associated with occupied sites for 120 or 170 ha disks, or for 52-120 and 120-170 ha rings. The interaction term {late closed X mid-aged closed) of the reduced model for disks was not significant (x^ = 43.1; 1 df; P = 0.23). Discussion Not all goshawk territories may be occupied in all years (Detrich and Woodbridge 1994, Reynolds et al. 1994), and even in the absence of human- caused habitat alteration, some territories can be 318 Desimone and DeStefano VoL. 39, No. 3 expected to be lost due to natural (e.g., stand se- nescence, disease, fire) changes in the forest over time (Graham et al. 1994). Mean annual occupan- cy of goshawk nesting areas in six studies across the western U.S. (concurrent with this study) were con- sistently in the 65-80% range over 2-11 yr of study. An occupancy rate of 29% (15/51) of the historical nest areas on our study area in 1994 is low by com- parison. The low occupancy rate may be due in part to attrition of some suitable nest areas due to natural disturbance over time; one fire partially burned two nest areas in our study. It is also pos- sible that our goshawk surveys did not extend out far enough to include some of the alternative nests used. Recent data from Arizona suggests that about 67% of goshawks move to alternative nest locations every year and that a 1000 m broadcast calling ra- dius accounted for about 95% of the alternative nest attempts (R. Reynolds unpubl. data). If these findings are applied to our study, we likely missed about 5% of all alternative nests in our study. Like many raptors, individual goshawk pairs may not breed every year, and determining trends in territory occupancy using 2 yr survey data is ten- tative (DeStefano et al. 1994a). Pairs not nesting in a given year, but still occupying the nest area, are difficult to hnd when surveys are conducted after courtship (Dewey et al. 2003). However, we searched large areas (>300 ha) multiple times around each historical nest location during a pe- riod when local weather conditions were not par- ticularly inclement for the region and when gos- hawk productivity was relatively high: 74% (17/23) of nests on the Fremont NF and 91% (20/22) of nests on the nearby Malheur National Forest suc- cessfully fledged young in 1994 (S. Rickabaugh, S. Danver, and S. Daw unpubl. data). Other studies m eastern Oregon and Washington reported simi- lar high occupancy and nest success levels for 1994 (McGrath et al. 2003; S. Finn unpubl. data) . In ad- dition, Kostrzewa and Kostrzewa (1990) reported that weather did not affect the density of territorial goshawk pairs over an 8-yr period in Europe, but was an influential limiting factor to breeding suc- cess. Thus, we concluded that the low occupancy rates of the historical nest areas were not attributed to low detectability, although we could not com- pletely rule this out as a possibility. The difference in forest structure between post- 1992 occupied and no-response nest areas was compelling. Late structural stage forest, especially with canopy cover >50%, was much more preva- lent around occupied than no-response nest areas. Conversely, very early and early structural stage for- est was much more prevalent in no-response than occupied nest areas. Our results indicated that late forest structure declined by 20-50%, and very early and early forest structure increased by >400% around no-response nests. These trends were de- tectable at all scales, but were strongest at the smaller scales (12 and 52 ha) and decreased with increasing scale. Although we do not have detailed history of stand management for all cases, the ob- served difference in habitat is attributed to levels of timber harvest, which we verified by photo- graphic evidence and field examination. The loss of large trees (>53 cm DBH) and a reduction in canopy cover to <50% appeared to influence nest area occupancy. Penteriani and Faivre (2001) and Penteriani et al. (2002) found that nest sites (ca. 0.8 ha) around the nest tree altered by more than 30%, either by selective tree harvest (shelter wood) or windthrow, caused goshawk pairs to change lo- cations to new nest stands. The general conclu- sions reached by Penteriani and Faivre (2001) and Penteriani et al. (2002) on habitat disturbance were consistent with our results: goshawks were ab- sent from nest areas where there was ^30% mean decrease in late and mid closed forest (12-ha scale) compared to the pre-1992 condition. Our data showed that this pattern was consistent at larger (12-52 ha) scales as well. Our results suggested that nest area habitat al- teration (loss of nesting habitat) was the most likely reason for the low occupancy rates of historical nest areas in 1994. The habitat alteration was likely the result of timber harvest (documented by aerial photographs), which reduced the proportion of late and mid-aged forest with high canopy closure and increased the proportion of very early and ear- ly open forest conditions within 52 ha (scale of lo- gistic model significance) of goshawk nests. Management practices for nesting habitat pro- tection on the Fremont NF were limited during 1973-91, ranging from no protection (unrestricted harvest) of nest areas to 12-ha no-harvest buffers around nests during the breeding season (Reyn- olds 1983, USDA 1993). In 1983, the Fremont National Forest Plan established several 24-ha gos- hawk habitat management areas. However, condi- tions on most of these management areas ranged from early successional forests (unsuitable to mar- ginal for nesting habitat) to mid-aged forest with only small patches of late-successional forest. Some September 2005 Conservation 319 goshawk management areas were reassigned or re- located in subsequent years to achieve timber har- vest objectives (K. Palermo and W. Watkins, Fre- mont NF, pers. comm., S. Desimone unpubl. data). The photographic record revealed that little or no long-term habitat protection was implemented for the 31 no-response areas as of 1994. All were historical sites that had some portions within 52 ha of the nest site harvested during or after the his- torical nesting season. In contrast, most goshawk territories in the western U.S. study areas we re- viewed (Table 2) had little or no habitat loss from timber harvest practices since discovery by the re- searchers and had yearly monitoring programs that documented relatively high occupancy rates (B. Woodbridge, P. Kennedy, R. Reynolds, and S. Dew- ey pers. comm.). This further supported our con- clusion that timber harvest was a determining fac- tor leading to significantly lower occupancy rates in the no-response nest areas compared to the oc- cupied areas. Nest area fidelity (as indexed by occupancy rates) is likely to be higher in good quality habitats as compared to poor quality habitats. This may be advantageous because there is an increased likeli- hood of nesting success where they may have been successful before (Newton 1979, Newton and Wyl- lie 1992, Rosenfield and Bielefeldt 1996). Our re- sults suggest nest areas with >50% proportion of older and larger structural classes may be higher quality nest areas than areas dominated by younger serai stages (Woodbridge and Detrich 1994, Finn et al. 2002). Detrich and Woodbridge (1994) and