[fiVS L -1 ■ A ^SSSM 1 n^ 1 J jl bb y- r: ^r ^ ► * i* ' THE RAPTOR RESEARCH EOUNDATION, INC. (Founded 1966) http://biology.boisestate.edu/ raptor/ OFFICERS SECRETARY: Judith Henckel TREASURER: Jim Fitzpatrick BOARD OF DIRECTORS INTERNATIONAL DIRECTOR #3: Stew Redpath DIRECTOR AT LARGE #1: Jemima ParryJones 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 Hofeman NORTH AMERICAN DIRECTOR #2: Gary Santoi.o NORTH AMERICAN DIRECTOR #3: Ted Swem INTERNATIONA!. DIRECTOR #1: Nick Mooney INTERNATIONAL DIRECTOR #2: PRESIDENT; Brian A. Millsap VICE-PRESIDENT; David M. Bird EDITORIAL STAFF EDITOR: James C. Bednarz, Department of Biological Sciences, P.O. Box 599, Arkansas State University, State University, AR 7246'7 U.S.A. ASSISTANT EDITOR: Jennifer L. Norris ASSOCIATE EDITORS James R. Belthofe Joan L. Morrison Clint W. Boat Fabrizio Sergio Cheryl R. Dykstra Ian G. Warkentin Michael I. Goldstein 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, Cintlv Cornelius EDITORIAL ASSISTANT; Blake Grisham, Joan Ciark 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 (8% X 1 1 in.) or standard international, white, bond paper, with 2.5 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; Great Horned Owl {Bubo virginianus) . Oil painting by Eva van Rijn; for more information and wildlife art visit: www.natureartists.com/artists/artist.asp?ArtistID=36 Contents The Effect of Broadcasting Great Horned Owl Vocalizations on Spotted Owl Vocal Responsiveness. Michelle L. Crozier, Mark E. Seamans, and RJ. Gutierrez Ill Home Range and Habitat Use by Great Horned Owls {Bubo virginianus) in Southern California. Jason R. Bennett and Peter H. Bloom 119 Evaluation of Methods for Gender Determination of Lesser Kestrel Nestlings. Carlos Rodriguez, Javier Bustamante, Begoha Martinez-Cruz, and Juan Jose Negro 12V Using Vocal Individuality to Monitor Queen Charlotte Saw-whet Owls {AEGOLIUS ACADICUS BROOKS !) . Carmen 1. Holschuh and Ken A. Otter 134 Partitioning of Genetic (RAPD) Variability among Sexes and Populations OF THE Barn Owl ( TyTO alba) in Europe. Robert Maries, Sandor Varga, Balazs Opper, Akos Klein, Gyozo Horvath, Alexandre Roulin, Peter Putnoky, and Gyula Hoffmann 142 Breeding Biology and Food Habits of the Madagascar Kestrel {Falco newtoni) IN Northeastern Madagascar. Lily-Arison Rene de Roland, JeanneneyRabearivony, Harilalaina Robenarimangason, Gilbert Razafimanjato, and Russell Thorstrom 149 Short Communications Interspecific Aggression and Nest-site Competition in a European Owl Community. Inigo Zuberogoiria, Jose Antonio Martinez, Jabi Zabala, and Jose Enrique Martinez 156 Prey Partitioning between Mates in Breeding Booted Eagles {Hieraaetus pennatus). Jose E. Martinez and Jose F. Calvo 159 Predation of Small Mammals by Rufous-legged Owl, Barn Owl, and Magellanic Horned Owl in Argentinean Patagonia Forests. Daniel E. Udrizar Sauthier, Analia Andrade, and Ulyses F. J. Pardihas 163 Changes in Site Occupancy and Nesting Performance of Peregrine Falcons in Colorado, 1963-2004. James H. Enderson 166 Analysis of Reservoir Selection by Wintering Ospreys {Pandion haliaetus hauaetus) in Andalusia, Spain: A Potential Tool FOR Reintroduction. Eva Casado and Miguel Ferrer 168 Introduced Animals in the Diets of the Ogasawara Buzzard, an Endemic Insular Raptor in THE Pacific Ocean. Yuka Kato and Tadashi Suzuki 173 The Diet of Eurasian Griffons {Gyps fulvus) in Crete. Stavros M. Xirouchakis 179 Letters Are Earlier Estimates of Accipitriformes Crossing the Channei. of Sicily (Central Mediterranean) During Spring Migration Accurate? Nicolantonio Agostini 184 Ground Nesting by Egyptian Vultures {Neophron percnopterus) in the Canary Islands. Laura Gangoso and Cesarjavier Palacios 186 First Summer Records of Ospreys {Pandion haliaetus) Along the Coast of Oaxaca, Mexico. Juan Meraz and Betzabeth Gonzalez-Bravo 187 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 June 2005 No. 2 J Raptor Res. 39(2) :1 11-1 18 © 2005 The Raptor Research Foundation, Inc. THE EFFECT OF BROADCASTING GREAT HORNED OWL VOCALIZATIONS ON SPOTTED OWL VOCAL RESPONSIVENESS Michelle L. Crozier, Mark E. Seamans, and RJ. Gutierrez^ University of Minnesota, Department of Fisheries, Wildlife, and Conservation Biology, 200 Hodson Hall, 1980 Folwell Avenue, St. Paul, MN 53108 U.S.A. Abstract. — One hypothesis advanced for the association of Spotted Owls {Strix occidentalis) with mature forest has been avoidance of competitors and predators such as Great Horned Owls {Bubo virginianus) . Great Horned Owls also have been identified as an issue of concern for the conservation of Spotted Owls. Thus, knowledge of Great Horned Owl presence in Spotted Owl territories could be valuable when evaluating trends in Spotted Owl survival. If Spotted Owls avoid Great Horned Owls because of risk of predation, we hypothesized that Great Horned Owl vocalizations should affect Spotted Owl calling behavior. Therefore, we experimentally examined vocal responsiveness of male Spotted Owls after Great Horned Owl vocalizations were played in their territories. We found little evidence that broadcasting Great Horned Owl vocalizations in Spotted Owl territories affected relatively short-term (24 hr) respon- siveness of male Spotted Owls. Heterospecific response rates were also low for both species. Thus, our prediction that the presence of Great Horned Owls (i.e., simulated calling by Great Horned Owls) would affect Spotted Owl responsiveness was not supported, at least on the temporal scale at which we conducted the experiment. Our results suggested that surveys to estimate Great Horned Owl presence on Spotted Owl study areas would not confound surveys for Spotted Owls in those areas if at least 24 hr passed between surveys for each species. Key Words: Great Horned Owl, Bubo virginianus; California Spotted Owl, Strix occidentalis occidentalis; cross-over experiment, heterospecific response, territoriality; auditory survey. EL EEECTO DE EMITIR VOCALIZACIONES DE bubo virginianus SOmE LA RESPUESTA VOCAL DE STRIX OCCIDENTALIS Resumen. — Se ha hipotetizado que la asociacion de Strix occidentalis con el bosque maduro se da para evitar competidores y grandes depredadores como Bubo virginianus. B. virginianus tambien ha sido identificado como un tema de preocupacion para la conservacion de S. occidentalis. De este modo, el conocimiento de la presencia de B. virginianus en los territories de S. occidentalis podna ser valioso al momento de evaluar las tendencias en la supervivencia de S. occidentalis. Si 5. occidentalis evita a B. virginianus por el riesgo de depredacion, hipotetizamos que las vocalizaciones de B. virginianus de- berian afectar el comportamiento de llamada de S. occidentalis. En consecuencia, examinamos de modo experimental la respuesta vocal de los machos de S. occidentalis luego de emitir en sus territorios vocalizaciones de B. virginianus. No encontramos evidencia sustancial de que emitir vocalizaciones de B. virginianus en los territorios de S. occidentalis afecto la respuesta de corto plazo (24 hr) de los machos de S. occidentalis. Las tasas heteroespecificas de respuesta fueron tambien bajas para ambas especies. De este modo, nuestra prediccion de que la presencia de B. virginianus (i.e., la simulacion del llamado de B. virginianus) afectaria la respuesta de S. occidentalis no fue avalada, al menos a la escala temporal a la cual condujimos el experimento. Nuestros resultados sugieren que los releva- mientos para estimar la presencia de B. virginianus en las areas de estudio de S. occidentalis no afec- ^ Email address: gutie012@umn.edu 111 112 Crozier et al. VoL. 39, No. 2 tarian los relevamientos de S. occidentalis en estas areas si ban pasado al menos 24 hr entre los rele- vamientos para cada especie. [Traduccion del equipo editorial] One hypothesis posed for Spotted Owl (Strix oc- cidentalis) habitat selection is that they use mature forest to avoid Great Horned Owls {Bubo virgini- anus) , which are considered competitors and pred- ators of Spotted Owls (Carey 1985, Gutierrez 1985). Forsman et al. (1984, 2002) also hypothe- sized that Spotted Owls avoided open areas to re- duce the risk of predation by Great Horned Owls. Although little evidence exists to support this hy- pothesis, spatial segregation and differences in habitat use may occur between the two species (Phillips et al. 1964, Johnson 1993, Ganey et al. 1997). In addition, Ganey et al. (1997) found con- siderable overlap in home ranges between Great Horned Owls and Mexican Spotted Owls (5. o. lu- cida) , but they noted that overlap within individual forest stands was limited. Spatial segregation also has been observed between Great Horned Owls and other owls. For example, Baumgartner (1939) hypothesized that the presence of Great Horned Owls, which are competitors and predators of both Barn Owls {Tyto alba) and Barred Owls (Strix va- ria), restricted these latter owls to “less favorable” home ranges. Other studies have supported Baum- gartner’s hypothesis (Barn Owl, Rudolph 1978; Barred Owl, McGarigal and Fraser 1984). Despite some evidence for spatial segregation. Great Horned Owls are often found near Spotted Owls and the two species can have overlapping home ranges (Forsman et al. 1984, Johnson 1993, Ganey et al. 1997, pers. obs.). Despite the potential for interspecific interac- tions, the effect of Great Horned Owl calling activ- ity on the subsequent responsiveness of Spotted Owls remains unexplored. The listing decision for the northern Spotted Owl (S. o. caurina) identified Great Horned Owls as a threat of unknown mag- nitude to the Spotted Owl (USDI 1990). Subse- quently, concerns for the conservation of the Spot- ted Owl have led to conservative Spotted Owl survey protocols. For example, current U.S. Forest Service survey protocol recommends skipping sur- vey stations where known predators are active, in- cluding Great Horned Owls (USDA Forest Service 1993). Because the Great Horned Owl is consid- ered a potential threat to the Spotted Owl, it would be desirable to monitor Great Horned Owl distri- bution and abundance within Spotted Owl demo- graphic study areas. However, we do not know how surveys for Great Horned Owls might affect Spot- ted Owl detection probabilities during subsequent Spotted Owl surveys. Thus, such critical informa- tion is needed before attempting simultaneous sur- veys of these species in the same area. In general, interspecific territoriality has been inferred from observation of agonistic behavior and response to song between two or more species (e.g., Orians and Willson 1964, M0ller 1992). Be- cause Great Horned Owls are predators of Spotted Owls (Forsman et al. 1984, Miller and Meslow 1985, Johnson 1993, Gutierrez et al. 1995), we hy- pothesized that Spotted Owls should actively avoid them. Therefore, we predicted Great Horned Owl vocalizations would suppress Spotted Owl vocal re- sponsiveness, including their subsequent vocal re- sponsiveness to conspecific calls. We tested this prediction experimentally by exposing male Cali- fornia Spotted Owls (S. 0 . occidentalis) to calls of Great Horned Owls in order to evaluate whether we could conduct surveys of Great Horned Owls on our Spotted Owl study area while not lowering subsequent detection probability of Spotted Owls. Methods Study Area. Our study was located in the central Sierra Nevada, California U.S.A. The owls we studied in this ex- periment were adjacent to the Eldorado Density Study Area (EDSA), the site of a long-term Spotted Owl pop- ulation study (Seamans et al. 2001). Elevation at Spotted Owl territories ranged from 930-1855 m, and vegetation was typical of middle elevation Sierran Montane Forest (Kiichler 1977). Prior to conducting the experiment, we established the presence of Spotted Owls within treat- ment and control areas by conducting surveys using stan- dard methods (Forsman 1983, Franklin et al. 1996). Dur- ing these surveys, we detected nine pairs and one single male defending territories at the experimental territory sites. Experimental Design. We used a 2 X 2 binary cross- over experimental design (Senn 1993) to test the short- term effect of Great Horned Owl calls (i.e., simulated presence) on male Spotted Owl responsiveness. To sim- ulate Great Horned Owl calling, we broadcast recorded calls of Great Horned Owls in 10 occupied Spotted Owl territories (see Morrell et al. 1991). Cross-over design. We considered each male Spotted Owl as an experimental unit. Great Horned Owl calling as the treatment, and male Spotted Owl responsiveness as the dependent variable. Our cross-over design consisted of June 2005 Interspecific Owl Calling 118 (1) randomly applying treatments during the first time period of the experiment to half the experimental units and using the other half as controls, then (2) switching treatments and controls for the second time period (Senn 1993). Thus, after the treatment of Great Horned Owl calling was applied (Treatment Day I), we measured responsiveness of Spotted Owls to conspecific calls (Treatment Day II), compared to a control Spotted Owl survey, to test if simulated Great Horned Owl presence (i.e., broadcast calls) had an effect on Spotted Owl re- sponsiveness. This design allowed us to use individual owls as their own controls, thus controlling for variation among experimental units (Ratkowski et al. 1993, Senn 1993). We only used Spotted Owl males because they are more vocally responsive than females (Reid et al. 1999). We defined “short-term effect” as the effect of waiting for a 24-hr interval between Great Horned and Spotted Owl surveys. We selected a 24-hr interval to assess the efficacy of surveying for Great Horned Owls in Spotted Owl territories and, secondarily, to assess interspecific in- teraction at a temporal scale we felt would minimize pre- dation risk to Spotted Owls (see below). First, we were interested in determining whether surveying for Great Horned Owls would bias the results of Spotted Owl sur- veys. Because Great Horned Owls have been considered a potential threat to northern Spotted Owls (USDI 1990), we wanted to estimate the distribution and abundance of Great Horned Owls on our study area, but not at the cost of disrupting our long-term Spotted Owl study. Thus, the question we tested was — do call surveys for Great Horned Owls cause Spotted Owls to reduce their responsiveness to standard survey protocol for the latter species? Con- cern about these species’ interactions is clearly expressed in current U.S. Forest Service survey protocol for Spotted Owls, which instructs observers to note predators when detected, including Great Horned Owls, and to skip sur- vey stations where predators are detected (USDA Forest Service 1993). Therefore, because we did not know enough about the ecological interactions between these two species to predict accurately what might be an ap- propriate stimulus-response interval and because the Spotted Owl is of great conservation concern, we selected a conservative 24-hr lag period to evaluate the response. That is, we did not want to follow a Great Horned Owl broadcast immediately with a Spotted Owl survey because of the potential predation risk to the latter species. In addition, an immediate progression of both species’ calls may have had a confounding effect on estimated re- sponse rates (i.e., we would not know whether the Spot- ted Owl was responding to the Great Horned Owl call or the subsequent Spotted Owl call) . We deliberated the is- sue of the appropriate stimulus-response period at great length prior to executing the experiment. Thus, we rec- ognized that inferences about behavioral responses per se (i.e., suppression of Spotted Owl calling activity) would be limited. However, we felt that this period would be appropriate to answer our most important question re- garding conducting surveys for both species on the same study area. Logistic constraints dictated the order in which terri- tories were visited during a survey period (i.e., territories m the same area were surveyed on the same night) . How- ever, all territories were randomly assigned initially to treatment or control groups. All territories were surveyed over nine consecutive nights during the first experimen- tal period, followed by an 8-d pause. Treatments were reversed and territories were surveyed again within nine consecutive survey nights during the second period. The survey order that we established for the first experimen- tal period was followed in the second experimental pe- riod so that an approximately equal amount of time (17 d) elapsed between complete treatment and control sur- veys in each territory. Territories took one extra day to survey in second round due to field conditions. Broadcast call experiment. We used methods outlined by Forsman (1983) and Morrell et al. (1991) to survey for Spotted and Great Horned Owls, respectively. Within each Spotted Owl territory, we established six call points 0.4-0. 6 km apart to attain complete coverage (Forsman 1983) of the area in which we had first detected each owl. We defined a complete survey as the combined re- sults of all individual call points from one survey period within a territory (Forsman 1983). At each point, we broadcast a Great Horned Owl call or imitated a Spotted Owl call for 10 min and recorded respo nses by species. .. Spotted Owl calls were produced vocally to be consistent with methods used for the demography study. Complete surveys were conducted from 2000-0100 H PST to limit within-night variation in responsiveness (Forsman 1983). We did not conduct surveys if wind was >12 km/hr or it was raining (Forsman 1983, Morrell et al. 1991). We structured Great Horned Owl surveys to be similar to Spotted Owl surveys. During Great Horned Owl treat- ments (Treatment Day I), we broadcast a recording of a male and female Great Horned Owl engaged in a calling bout (Stokes et al. 1999). For Great Horned Owl treat- ments only, observers listened for the first min and the last 3 min for unsolicited calls (Morrell et al. 1991, John- son 1993). For the remainder of the survey (min 2-7), we played six Great Horned Owl broadcasts, consisting of six sets of 20-sec, 4—7 note calls by a pair of Great Horned Owls separated by a 40-sec interval. The first 20- sec broadcast was made with the speaker perpendicular to the road, then rotated 180° following each 20-sec broadcast. During Spotted Owl treatments (Control and Treatment Day II), observers vocally produced Spotted Owl calls for the entire 10 min, imitating 3-5 four-note location calls every 15 sec (Forsman 1983). The 20-sec and 15-sec intervals of silence between Spotted and Great Horned Owl calls represented the frequency of unsolic- ited calls observed in the field for each species (Spotted Owls: Forsman et al. 1984, Johnson 1993; Great Horned Owls; Houston et al. 1998). A positive treatment response included any complete survey (i.e., calling at six survey points) in which a male Spotted Owl was detected during Spotted Owl broadcasts (Treatment Day II or Control). If a Great Horned or Spotted owl of either sex was detected at any survey point, observers noted time of detection, owl species, sex (based on pitch of call; Great Horned Owl: Miller 1930; Spotted Owl: Forsman 1983), response type (visual or vo- cal), compass estimated direction and distance to the owl, and whether the response occurred during pre- broadcast, broadcast, or post-broadcast time periods (Morrell et al. 1991). We considered a Great Horned Owl 114 Crozier et al. VoL. 39, No. 2 Table 1. A. priori models used to evaluate the effects of treatment (broadcasting Great Horned Owl calls) and presence of Great Horned Owls (GHOW) on short-term responsiveness of male Spotted Owls (SPOW). All models are Generalized Linear Mixed Models (GLMMs) in which individual owl (SPOW) has been blocked as a random variable and all other variables have fixed effects. T, P, C, and GHOW indicate Treatment, Period, Carryover, and Great Horned Owl covariates, respectively. Intercept is included as a parameter in each model. Model Model Strucuture Model Description M() Po SPOW (Random) 2 IVl'y Po + Pi (T) Treatment(pixed) 8POW(R^adom) 3 Mx+p Po + Pi (T) + P 2 (P) Treatment(Fi,,ed) + Period(Fi,,ed) -t SPOW(Random) 4 Mx+c Po + Pi (T) + P 2 (C) Treatment(Fi,,ed) + Carryover + SPOW(Random) 4 Mghow Po + Pi (GHOW) Great Horned Owljpi^ed) + SPOW(Random) 3 Mx+GHOW Po -f pi(T) + p2 (GHOW) Treatment(Fixed) + Great Horned Owl (Fixed) + SPOW(Randonj) 4 MxxGHOW Po + Pi (T) + Pa (GHOW) + p3 (T X GHOW) Treatment(Fixed) + Great Horned Owl (Fixed) G W ( j 4 ® K = number of parameters in model. present within a Spotted Owl territory if we detected it at any point during the study. Statistical Analysis. During our experimental design phase, we developed six a priori hypotheses (models) to explain how broadcasting Great Horned Owl calls might affect short-term responsiveness of male Spotted Owls (Table 1). During the data collection phase, we detected more Great Horned Owls than we expected, so we de- veloped another model prior to analysis that included a covariate representing the detection of a Great Horned Owl(s) at a territory during a survey (i.e., a Great Horned Owl was actually present, not just simulated) . We considered our seven models as competing hypotheses (Burnham and Anderson 1998). We included the indi- vidual owl as a random effect in all models. We consid- ered treatment (T), detection of a Great Horned Owl(s) during broadcasts (GHOW), and structural components of the study design as fixed effects. Structural compo- nents of the study design included a carryover and pe- riod effect. We analyzed data within a maximum likeli- hood framework using a Generalized Linear Mixed Model (%GLIMMIX; SAS 8.02, SAS Institute 2001) with a logit link and binomial error because our response var- iable was binary (no response = 0, male Spotted Owl vocal response = 1). We used maximum likelihood esti- mators (MLEs) to determine parameter estimates of fixed variables (Littell et al. 1996). We objectively ranked models using a bias-corrected version of Akaike’s Information Criterion (AIC^ and AAICc: Akaike 1973, Burnham and Anderson 1998). All models were compared to a means-only model (no fixed effects). We used Akaike weights {w^ to estimate the like- lihood of each model relative to competing models, given the data (Akaike 1973, Burnham and Anderson 1998). An Akaike weight {w) is the weight of a specific model, defined as EXP {—0.5 ZLAICc) of that specific model di- vided by the sum of (EXP{— 0.5 AAICcl) for all models (Burnham and Anderson 1998). We calculated an intercept, parameter estimates, and associated standard errors for fixed effects used in each of the models (Littell et al. 1996). The sign of the esti- mate indicated whether the variable had a positive or negative effect on Spotted Owl responsiveness. If the 90% confidence interval for a parameter estimate did not include zero, we concluded that the parameter estimate was different from zero. Therefore, if this result occurred for the treatment parameter, we inferred that Great Horned calling had an effect on Spotted Owl responsive- ness. Results We conducted our experimental study from 16 July-8 August 2003 following preliminary surveys that occurred in late June to locate occupied ter- ritories. Spotted Owl responsiveness was similar be- tween treatment (Treatment Day II) and control surveys (Fig. 1). Although we only included male Spotted Owl response in our models, we recorded responses from both species for both sexes. Spot- ted Owl response rates (control = 70%, treatment — 60%) were similar to the response rate (57.3%, SE = 7.5) of Spotted Owls occupying established territories on the EDSA (R.J. Gutierrez unpubl. data). Of 30 complete surveys (10 Control, 10 Treatment Day I, 10 Treatment Day II), Spotted and Great Horned owls were detected together during the same survey only once. Overall, there was little evidence that broadcast- ing Great Horned Owl calls affected male Spotted Owl responsiveness at the temporal scale we eval- uated for the experiment (M^: Fj g = 0.22, P = 0.651; Mqhow- ~ 1-87, P = 0.201). The means model was the top-ranked model based on AIC^ (Table 2). The second-best model was a treatment- only model, followed by a model with Great Horned Owl presence only. However, both of these June 2005 Interspecific Owl Calling 115 10 - • UM SPOM — V — reniaWSPOW 6 -j ■ UaM GMOW .. - -•''w fZHnM T-1 (SHOW CALL) T-2 (SPOWCALL) CONTROL Treatment Type Figure 1. Number of owl responses following broadcast of Great Horned Owl calls (Treatment Day I = Tl) and Spotted Owl calls (Treatment Day II = T2 and Control = CONTROL) . Responses of both male and female Spot- ted Owls (SPOW) and Great Horned Owls (GHOW) were noted during all surveys. One Great Horned Owl detection where gender could not be determined was included as probable male. models were >3 AICc units from the means model, indicating that support for these effects was weak. The specific estimates of fixed parameters for these latter models indicated that the slope estimates were not different from zero (Table 3). For the treatment-only model (M^), the 90% confidence interval of the treatment estimate included zero ((3x = 0.442 ± 1.55). The 90% confidence interval of the parameter estimate for the presence of a Great Horned Owl (Mghow) ^ilso included zero (Pghow = -1-38 ± 1.67). Discussion We found no evidence that the simulated pres- ence of Great Horned Owls had an effect on male Spotted Owl vocal responsiveness at the temporal scale of our evaluation. This suggests that con- ducting surveys for both species can be conducted on the same study area without biasing surveys for Spotted Owls given a reasonable lag (at least 24 hr) between surveys of each species. This result also weakens, but does not entirely refute (see below) the hypothesis that Spotted Owls select territories (habitat) to avoid Great Horned Owls (Carey 1985, Gutierrez 1985). However, the low heterospecific response rates suggest that these species are not interspecifically territorial. Other studies of heter- ospecific and conspecific avian responsiveness have measured the response of one species to another Table 2. Ranking of a priori models to assess the short- term responsiveness of male California Spotted Owls to simulated Great Horned Owl presence in the central Si- erra Nevada, California. Ranking is based on AICc values; Wi values are Akaike weights. Model Log- likelihood AICc AAICc u >^ M(.) 2 243.2 91.1 0.0 0.725 Mt 3 43.3 94.1 3.0 0.162 Mghow 3 44.5 96.4 5.3 0.051 Mx+p 4 43.4 97.5 6.4 0.030 Mx+c 4 43.7 98.0 6.9 0.023 Mx+ghow 4 44.8 100.2 9.1 0.008 Mx=>GHOW 5 44.7 103.6 12.5 0.001 “ K = number of parameters in model. = Akaike weight = (EXP~°® ^ aaicc [specific model])/ (2 of (EXP-o-5 X aaicc [all models])). species (Bosakowski and Smith 1998, Boal and Bi- bles 2001). Our study differed in one fundamental way from these studies because we broadcast calls of two species, a Great Horned Owl call followed, after a latent period, by a Spotted Owl call to assess whether the first species affected the response of the second species to conspecific calls. Given some level of background exposure (i.e., Spotted Owls may normally hear Great Horned Owls), conduct- ing surveys of Great Horned Owls within Spotted Owl territories does not appear to alter detection of Spotted Owls or increase predation risk to Spot- ted Owls, at least at the temporal scale of our ex- periment. We noted that response to heterospecif- ic calls by both species was low. This was consistent with other studies comparing responsiveness of raptor species to conspecific and Great Horned Owl calls (Johnson 1993, Bosakowski and Smith 1998, Boal and Bibles 2001). The low response rate to heterospecific calls and the high response rate of Spotted Owls to conspecific calls following ex- posure to Great Horned Owl calls also implies that Spotted Owls may not vocally defend their terri- tories against Great Horned Owls, which we had expected to detect if interspecific competition was present. In addition. Spotted Owl response rates were very similar to response rates of Spotted Owls on a nearby study area, which suggested the ex- perimental effect did not result in changes of the patterns of calling by Spotted Owls. We did not measure immediate or long-term ef- fects of broadcasting Great Horned Owl calls in 116 Crozier et al. VoL. 39, No. 2 Table 3. Estimates of fixed parameters with associated standard errors, F, P-values, and degrees of freedom for hypothesized models explaining short-term responsiveness of male Spotted Owls to Spotted Owl calls after exposure to Great Horned Owl calls in the central Sierra Nevada, California. Models are presented according to rank based on AIC(;. For all parameters, estimates represent probability the Treatment (T) = control, Period (P) = 1st, Carryover (C) = no, and Great Horned Owl Present (CHOW) = no. Model Parameter Parameter Estimate (SE) P-value P-value'' df M() fA 0.619 (0.469) 0.219 Mt 0.406 (0.646) T 0.442 (0.945) 0.651 0.22 1, 9 MghOW 1.39 (0.791) 0.118 CHOW -1.39 (1.01) 0.201 1.87 1, 10 Mx+p P- 0.187 (0.791) 0.818 T 0.447 (0.949) 0.650 0.22 1, 8 P 0.447 (0.949) 0.650 0.22 1,8 Mx+c P- -0.575 (1.58) 0.724 T 0.990 (1.29) 0.469 0.58 1, 8 C 0.990 (1.44) 0.516 0.46 1, 8 Mx+GHOW P 1.16 (0.906) 0.237 T 0.491 (1.00) 0.636 0.24 1, 9 CHOW -1.40 (1.03) 0.205 1.87 1, 9 ® T-value calculated from Type III sums of squares. Spotted Owl territories. Although we chose to be conservative when selecting an appropriate time interval, 24 hr may not have been the most appro- priate stimulus-response interval to detect a differ- ence in responsiveness (i.e., call suppression). If the primary goal of a study was to examine explic- itly the behavioral interactions of the two species, we would recommend employing a shorter stimu- lus-response interval with appropriate consider- ation for increasing potential for predation of Spotted Owls. We also did not evaluate biological factors that might stimulate territorial defense such as brood defense. Although we did not assess re- production of Spotted Owls in this study, Spotted Owl reproduction on the nearby EDSA was very low in 2003 (R.J. Gutierrez unpubl. data), and re- production is highly correlated among regional populations of Spotted Owls in the Sierra Nevada (Franklin et al. 2004) . We are not certain what ef- fect breeding status might have on Spotted Owl responsiveness. Great Horned Owl detections within Spotted Owl territories were common following Great Horned Owl broadcasts. When Great Horned Owls were detected following the broadcast of Great Horned Owl calls (Treatment Day 1), 6 of 7 (85.7%) Great Horned Owls flew to within 50 m of the broadcast location and continued to call. Thus, surveying for Great Horned Owls caused movement of these predators into the survey area, which was within an occupied Spotted Owl terri- tory; this validated some of our initial concern re- garding risk to Spotted Owls. However, we noted no discernable effect on subsequent Spotted Owl vocal responsiveness after 24 hr. Because we de- tected Great Horned Owls in half of the Spotted Owl territories that we surveyed, it was likely that Spotted Owls were exposed regularly to Great Horned Owl calling. This might explain why we did not see a treatment effect. However, the model for Great Horned Owl presence was not a signifi- cant predicator of male Spotted Owl responsive- ness. Thus, it appeared that neither artificial ex- posure nor live exposure to Great Horned Owls affected detection rates of Spotted Owls following our lag period. Lack of vocal interaction between these two owl species suggests that (1) Great Horned and Spot- ted owls may not be strong competitors, (2) Great Horned Owls may prey on Spotted Owls only in an opportunistic manner, (3) these species segregate habitat on a fine scale even when apparently oc- cupying the same general areas, or (4) some other mechanism has evolved to maintain ecological or spatial separation between these two species. From a conservation perspective, it appears that Great Horned Owl surveys may not have a con- founding effect on Spotted Owl population stud- June 2005 Interspecific Owl Calling 117 ies. Given the interest in the interspecific interac- tions of these owl species, our results suggest that surveying for Great Horned Owls would not affect the detection probability of Spotted Owls if surveys for the former species are conducted at least 24 hr apart from surveys of the latter species. Because Great Horned Owls occupy more open habitats and if their numbers increase in response to hab- itat fragmentation induced by logging, then the opportunity for Great Horned Owl predation on Spotted Owls may increase (Thomas et al. 1990). Such a numerical response might also suggest al- ternative silvicultural practices to reduce the im- pact of changing forest habitat that favors Great Horned Owls. Thus, Great Horned Owl surveys may be important to include in studies designed to monitor the effects of logging on Spotted Owl survival rates because timber removal may create more habitat suitable for this potential predator. Acknowledgments We thank Foresthill, Georgetown, and Pacific U.S. For- est Service Ranger Districts and Forest Biologists Charis Parker, Matthew Triggs, and Don Yasuda for providing us with the location of historical Spotted Owl territories. We also thank our field technicians, J. Anfinson, B. Bloemke, A. Chatfield, M. Chatfield, J. Corcoran, S. Dallmann, and A. Rex for their assistance. Finally, we appreciate the thoughtful comments of E.D. Forsman and two anony- mous reviewers. This study was funded by the U.S. Forest Service (contract FS53-91S8-00-EC14 to R. Gutierrez) and the University of Minnesota. Literature Cited Akaike, H. 1973. Information theory as an extension of the maximum likelihood principle. Pages 267-281 in B.N. Petrov and F. Csaki [Eds.], Second international symposium on information theory. Akademiai Kiado, Budapest, Hungary. Baumgartner, F.M. 1939. Territory and population in the Great Horned Owl. Auk 56:274-282. Boat, C.W. and B.D. Bibles. 2001. Responsiveness of Elf Owls to conspecific and Great Horned Owl calls. J. Field Ornithol. 72:66-71. Bosakowski, T. and D.G. Smith. 1998. Response of a for- est raptor community to broadcasts of heterospecific and conspecific calls during the breeding season. Can. Field-Nat. 112:198-203. Burnham, K.P. and D.R. Anderson. 1998. Model selec- tion and inference: a practical information-theoretic approach. Springer-Verlag, New York, NYU.S.A. Carey, A.B. 1985. The scientific basis for Spotted Owl management. Pages 110-114 in RJ. Gutierrez and A.B. Carey [Eds.], Ecology and management of the Spotted Owl in the Pacific northwest. Gen. Tech. Rep. PNW-185, USDA, Forest Service Pacific Northwest Re- search Station, Portland, OR U.S.A. Forsman, E.D. 1983. Methods and materials for locating and studying Spotted Owls in Oregon. Gen. Tech. Rep. PNW-162, USDA, Forest Service Pacific North- west Research Station, Portland, OR U.S.A. , E.C. Meslow, and H.M. Wright. 1984. Distri- bution and biology of the Spotted Owl in Oregon. Wildl. Monogr. 87:1-64. , R.G. Anthony, J.A. Reid, PJ- Loschl, S.G. Sov- ern, M. Taylor, B.L. Biswell, A. Ellingson, E.C. Meslow, G.S. Miller, K.A. Swindle, J.A. Thrailkill, F.F. Wagner, and D.E. Seaman. 2002. Natal and breed- ing dispersal of Northern Spotted Owls. Wildl. Monogr 149:1-35. Franklin, A.B., D.R. Anderson, E.D. Forsman, K.P. Burnham, and F.W. Wagner. 1996. Methods for col- lecting and analyzing demographic data on the Northern Spotted Owl. Stud. Avian Biol. 17:12-20. , RJ. Gutierrez, J.D. Nichols, M.E. Seamans, G C. White, G.S. Zimmerman, J.E. Hines, T.E. Munton, W.S. LaHaye, J.A. Biakesley, G.N. Steger, B.R. Noon, D.W.H. Shaw, JJ. Keane, T.L. McDonald, and S. Britting. 2004. Population dynamics of the California Spotted Owl {Strix occidentalis occidentalis): a meta-anal- ysis. Ornithol. Monogr. 54. Ganey, J.L., W.M. Block, J.S. Jenness, and R.A. Wilson 1997. Comparative habitat use of sympatric Mexican Spotted and Great Horned owls./. Wildl. Res. 2:115- 123. Gutierrez, RJ. 1985. An overview of recent research on the Spotted Owl. Pages 39-49 in R.J. Gutierrez and A.B. Carey [Eds.], Ecology and management of the Spotted Owl in the Pacific northwest. Gen. Tech. Rep. PNW-185, USDA, Forest Service Pacific Northwest Re- search Station, Portland, OR U.S.A. , A.B. Franklin, and W.S. LaHaye. 1995. Spotted Owl {Strix occidentalis). In K. Poole and F. Gill [Eds ], The birds of North America, No. 179. The Birds of North America, Inc., Philadelphia, PA U.S.A. Houston, C.S., D.G. Smith, and C. Rohner. 1998. Great Horned Owl {Bubo virginianus) . In A. Poole and F. Gill [Eds.], The birds of North America, No. 372. The Birds of North America, Inc., Philadelphia, PA U.S.A. Johnson, D.H. 1993. Spotted Owls, Great Horned Owls, and forest fragmentation in the central Oregon Cas- cades. M.S. thesis, Oregon State Univ., Corvallis, OR U.S.A. Kuchler, A.W. 1977. The map of natural vegetation of California. Pages 909-938 in M.G. Barbour and J. Ma- jor [Eds.], Terrestrial vegetation of California. J. Wiley and Sons, New York, NY U.S.A. Littell, R.C., G.A. Milliken, W.W. Stroup, and R.D. Wolfinger. 1996. SAS system for mixed models. SAS Institute, Cary, NC U.S.A. McGarigal, K. and J.D. Fraser. 1984. The effect of forest 118 Crozier et al. VoL. 39, No. 2 stand age on owl distribution in southwestern Virgin- ia. Wildl. Manag. 48:1393—1398. Miller, L. 1930. The territorial concept in the horned owls. Condor 32:290-291. Mii.ler, G.S. and E.C. Meslow. 1985. Dispersal data for Spotted Owls: the problem of small sample size. Pages 69-73 mR.J. Gutierrez and A.B. Garey [Eds.], Ecology and management of the Spotted Owl in the Pacific northwest. Gen. Tech. Rep. PNW-185, USDA, Forest Service Pacific Northwest Research Station, Portland, OR U.S.A. M0LLER, A.P. 1992. Interspecific response to playback of bird song. Ethology 90:315-320. Morrell, T.E., R.H. Yahner, and W.L. Hajrkness. 1991. Factors affecting detection of Great Horned Owls by using broadcast vocalizations. Wildl. Soc. Bull. 19:481- 488. Orians, G.H. and M.F. Willson. 1964. Interspecific ter- ritories of birds. Ecology 45:736—745. Phillips, A.R., J. Marshall, and G. Monson. 1964. The birds of Arizona. Univ. Arizona Press, Tucson, AZ U.S.A. Ratkowski, D.A., M.A. Evans, and J.R. Alldredge. 1993. Cross-over experiments: design, analysis, and appli- cation. Marcel Dekker, Inc., New York, NYU.S.A. Reid, J.A., R.B. Horn, and E.D. Forsman. 1999. Detec- tion rates of Spotted Owls based on acoustic-lure and live-lure surveys. Wildl. Soc. Bull. 27: 986—990. Rudolph, S.G. 1978. Predation ecology of coexisting Great Horned and Barn owls. Wilson Bull. 90:134-137. SAS Institute, Inc. 2001. SAS software version 8.02 for Windows. SAS Institute, Inc., Cary, NC U.S.A. Seamans, M.E., R.J. Gutierrez, C.A. Moen, and M.Z Peery. 2001. Spotted Owl demography in the central Sierra Nevada./. Wildl. Manag. 65:425-431. Senn, S. 1993. Cross-over trials in clinical research. John Wiley and Sons, Ltd., New York, NYU.S.A. Stokes, D., L. Stokes, and K.J. Colver. 1999. Stokes field guide to bird songs: western region. Time Warner Au- diobooks, New York, NY U.S.A. Thomas, J.W., E.D. Forsman, J.B. Lint, E.C. Meslow, B.R. Noon, and J. Verner. 1990. A conservation strat- egy for the northern Spotted Owl. Report of the In- teragency Scientific Committee to address the conser- vation of the northern Spotted Owl. U.S. Department of Agriculture, Forest Service, U.S. Department of In- terior, Bureau of Land Management, Fish and Wild- life Service, and National Park Service. U.S. Govern- ment Printing Office 791-171/20026, Washington, DC U.S.A. USDA Forest Service. 1993. Protocol for surveying for Spotted Owls in proposed management activity areas and habitat conservation areas. Pacific Southwest Re- gion, San Francisco, GA U.S.A. USDI. 1990. Endangered and threatened wildlife and plants; determination of threatened status for the northern Spotted Owl. Federal Register 55:26114- 26194, U.S. Department of Interior, Washington, DC U.S.A. Received 17 March 2004; accepted 11 December 2004 J Raptor Res. 39(2):119-126 © 2005 The Raptor Research Foundation, Inc. HOME RANGE AND HABITAT USE BY GREAT HORNED OWLS {BUBO VIRGINIANUS) IN SOUTHERN CALIFORNIA Jason R. Bennett^ Department of Biological Sciences, California State University, Long Beach, CA 90840 U.S.A. Peter H. Bloom Western Foundation of Vertebrate Zoology, 439 Calk San Pablo, Camarillo, CA 93010 U.S.A. Abstract. — Great Horned Owls {Bubo virginianus) are a common, widespread species that can be found in a variety of habitats across most of North America, but little is known about their space and habitat requirements. Using radiotelemetry, location data were collected on nine male and five female Great Horned Owls to determine home range and habitat use in southern California. Owls were tracked between Januciry 1997 and September 1998 for periods ranging from 5-17 mo. Seven owls were also followed during 13 all-night observation periods. The mean 95% adaptive kernel home-range size for females was 180 ha (range = 88-282, SE = 36) and that for males was 425 ha (range = 147-1115 ha, SE = 105). Core areas estimated by the 50% adaptive kernel averaged 27 ha (range = 7-44, SE = 7) for females and 61 ha (range = 15-187, SE = 18) for males. Owls were located in areas with varying degrees of human disturbance ranging from almost entirely urban to native oak ( Quercus agrifolia) woodland. Oak/ sycamore {Quercus agrifolia/ Platanus racemosa) woodland and ruderal grassland (Bromus spp., Avena spp., and various other non-native invasives), were used more often than expected based on availability, but we found no correlation between home-range size and any single habitat type or habitat groups. Key Words: Oreat Horned Owl, Bubo virginianus; home range; habitat use, southern California. AMBITO DE HOGAR Y USO DE habitat DE bubo virginianus EN EL SUR DE CALIFORNIA Resumen. — Bubo virginianus es una especie comun y ampliamente distribuida que puede ser encontrada en una gran variedad de habitats a traves de gran parte de America del Norte. Sin embargo, se sabe poco sobre sus requerimientos de espacio y habitat. Se recolectaron datos de localizacion de nueve machos y cinco hembras de B. virgnianus utilizando radio-telemetria, con el fin de determinar el ambito de hogar y la utilizacion del habitat de esta especie en el sur de California. Los buhos fueron seguidos entre enero de 1997 y septiembre de 1998 durante periodos que variaron entre 5-17 meses. Siete buhos tambien fueron seguidos durante 13 periodos de observacion que duraron toda la noche. El tamano promedio del ambito de hogar identificado por el metodo de kernel adaptativo del 95% fue de 180 ha para las hembras (rango = 88-282, SE = 36) y de 425 ha para los machos (rango = 147-1115 ha, SE = 105). Las areas nucleo estimadas por el kernel adaptativo del 50% fueron en promedio de 27 ha para las hembras (rango = 7-44, SE = 7) y de 61 ha para los machos (rango = 15-187, SE = 18). Los buhos se localizaron en areas con distintos grados de perturbacion humana, variando desde areas totalmente urbanas hasta bosques natives de Quercus agrifolia. Los bosques de Q. agrifolia y Platanus racemosa y las praderas ruderales con Bromus spp., Avena spp. y varias otras especies invasivas no nativas fueron utilizadas con mayor fre- cuencia de lo esperado segun la disponibilidad de estos habitats, pero no encontramos una correlacion entre el tamano del ambito de hogar y un habitat en particular o grupos de habitats. [Traduccion del equipo editorial] The Great Horned Owl {Bubo virginianus) is one of the most widespread birds of prey in the Amer- icas (Houston et al. 1998). They are able to pop- ulate a wide range of habitats because they are gen- ^ Corresponding author’s present address; USGS/BRD, Kilauea Field Station, Hawaii National Park, HI 96718, U.S.A.; Email address: j_bob_bennett@yahoo.com eralist predators with one of the most diverse prey profiles of all North American raptors and can use a diverse range of nest sites (Bent 1938, Houston et al. 1998). In California, Great Horned Owls nest from sea level to at least 2500 m in elevation within a diverse range of both natural and human altered habitats and play an important role as a top pred- ator in southern California’s wildlife communities. 119 120 Bennett and Bloom VoL. 39, No. 2 Although Great Horned Owls are in little danger of vanishing from southern California, diverse na- tive wildlife communities are threatened by ram- pant urban development. In general, conservation efforts in southern California have focused on sin- gle-species management of state and federally list- ed threatened and endangered species. An alter- native approach is to focus conservation efforts on upper-trophic-level predators, such as large rap- tors, whose spatial and ecological requirements are likely to encompass those of many other species (Bednarz et al. 1990, Bloom et al. 1993). Further- more, top-level predators play important ecological roles in maintaining biological diversity in human- altered landscapes by keeping mesopredator num- bers in check (Soule et al. 1988, Litvaitis and Vil- lafuerte 1995, Crooks and Soule 1999). Great Horned Owls are of particular interest be- cause they are one of the largest raptors in south- ern California, are likely to have large space re- quirements, and can adapt to and expand into areas altered or disturbed by humans. To our knowledge no quantitative information has been published on home-range size, habitat composi- tion, and response of Great Horned Owls to land development in southern California. Study Area The study area consisted of urbanized and “natural” areas of coastal foothills extending from Rancho Mission Viejo in the south, north to Huntington Beach in Orange County, California. Topography consisted of low eleva- tion rolling hills and plains with seasonal streams and small rivers bisecting the landscape. We studied nesting pairs of Great Horned Owls in the cities of Huntington Beach, Lake Forest, Irvine, and Mission Viejo, as well as the more natural area of Rancho Mission Viejo (20 km east of Mission Viejo) and Ronald W. Caspers Regional Park. Elevation varied between 30-300 m above sea level. Principal land uses in urban areas included city and regional parks, agriculture, housing, and industry. Land uses on Rancho Mission Viejo were cattle ranching and agriculture, but the area also contained large tracts of native vegetation communities. Permanent or intermit- tent water sources within owl home ranges included streams, channelized waterways, and artificial ponds. The region’s climate is Mediterranean, typically arid with most rain occurring in February. Methods Great Horned Owls were captured using bal-chatri traps (Berger and Mueller 1959, Bloom 1987) baited with live mice {Mus musculus), or with a dho-gaza trap using a live Great Horned Owl as a lure (Hamerstrom 1963, Bloom 1987, Bloom et al. 1992). Gender was determined by the presence or absence of a brood patch, by body size and mass, and age (i.e., hatch year, second year, and after hatch year) was determined by molt characteristics. Each owl was banded with a U.S. Geological Survey alu- minum band and equipped with a radiotransmitter (Communications Specialists, Orange, CA U.S.A.) in a backpack configuration, fitting the radio between the wings with Teflon straps joining at the breast (Dunstan 1972). The combined mass of the transmitter and har- ness (28 g) was less than 3% of the mean body mass of the owls. Transmitters had an estimated battery life of 2 yr and a range of ca. 3 km. Radiotagged owls were relocated using a hand-held ra- dio receiver with a three-element yagi antenna. After a bearing was obtained, a precise location was ascertained by a visual sighting 44% of the time. Street lamps and urban glow often facilitated the sightings of owls. Move- ments were detected visually or by a change in radio sig- nal strength, usually followed by a change in signal di- rection. When an owl could not be located visually, locations were determined by triangulation. At least three compass bearings were taken sequentially within 15 min, usually at a distance of <150 m, and care was taken to minimize disturbance. We were often able to encircle an owl’s po- sition, and thus, could reliably infer the owl’s location. A location determined by triangulation was used only if the resulting error polygon was <2 ha. Owl locations were plotted on a U.S. Geological Survey 7.5-min quadrangle map or on a road map. Owls were located at all hours between sunset and sun- rise and were followed through the battery life of the transmitter or the termination of the study (September 1998). Each owl was located ca. once per week and was tracked for up to 5 hr after initial detection. Additionally, six owls were tracked continuously throughout two entire nights and one owl was tracked a single night. Spatial autocorrelation results from sampling station- ary animals at short, regularly spaced time intervals, and Great Horned Owls often remain on a single perch for many hours. In order to reduce the degree of autocor- relation in our data set, yet maintain an adequate sample size, we recorded owl successive locations only when a perch change occurred. Also, we removed from the anal- ysis all point locations recorded within 30 min of each other. Because location points were not collected at reg- ularly spaced time intervals, “time to independence” of successive location points (Swihart and Slade 1985) was not applicable. The same locations were used for both home-range estimation and habitat-use analysis. We digitized location points using Geographic Infor- mation System (GIS) software (ESRI 1995). The adaptive kernel (AK) estimate of home-range size (Worton 1989) was calculated for each owl using the program CALHO- ME (Kie et al. 1994). The AK method is less biased by the scale or grid density and can produce more consis- tent results than many other home-range estimators (Kie et al. 1994, Worton 1995, Seaman and Powell 1996, Hans- teen et al. 1997, Lawson and Rodgers 1997). The grid cell option for the AK was set at a density of 50 X 50 cells for all home-range estimations. We used the 95% AK utilization contours to delineate home-range boundaries for each owl, but if a location was used only once and it increased the home-range size by >10%, we removed it from our calculation (Bloom June 2005 Great Horned Owl Home Range 121 1989). We used CALHOME’s estimated optimum band- width (smoothing parameter; Worton 1989) for each data set. For comparison with other studies the 100% minimum convex polygon (MCP; Mohr 1947) and the 95% harmonic mean (HM; Dixon and Chapman 1980) estimations were also calculated. We chose the 50% AK contour to represent core areas within the home range of each owl. We categorized habitat as belonging to one of eight common vegetation communities of southern California: oak woodland {Quercus spp.), oak/ sycamore woodland {Quercus agrifolia/Platanus racemosd), exotic woodland, coastal sage scrub, riparian scrub, agriculture, urban, and ruderal grassland. We based these habitat types on dom- inant vegetation and physiognomic features. All areas in- cluded roads, utility poles, and buildings to varying de- grees. Oak woodland was characterized by a closed or nearly closed canopy of coast live oak {Quercus agrifolia), with a relatively open understory, was relatively rare, found pri- marily in linear groves along the bottoms and on north facing canyon slopes. Oak/ sycamore woodland was more closely associated with intermittent or perennial streams. This habitat contained ca. equal proportions of coast live oak and sycamore {Platanus racemosa) 10-20 m in height with a broken canopy. We classified various nonnative woodland habitats as exotic woodland. Included were parks and golf courses, which contained pine {Pinus spp.) and gum {Eucalyptus spp.) stands with an open un- derstory of turf grass. Gum trees were common in urban and ranch areas and were included in this habitat type when stands exceeded ca. 1 ha. Of the non-woodland habitat, coastal sage scrub was found on exposed hillsides with a diverse array of drought tolerant shrubs predominating. Dominant shrubs included lemonade berry {Rhus integrifolia) , laurel sumac {Malosma laurina), and California sagebrush {Ar- temisia californica) . Riparian scrub consisted of young and mature willow {Salix spp.), mulefat {Baccharis salicifolia) , and other shrubs found along open-stream washes and creeks. Agricultural areas included citrus, corn, strawber- ry, and potted-ornamental plant production often with open patches of exposed soil. Urban habitats consisted of housing, industrial parks and buildings, small areas of associated landscaped vegetation, pavement, and the sur- rounding road system. Ruderal grassland was characterized by large open fields of nonnative grasses {Bromus spp., Avena spp., Hor- deum spp.), black mustard {Brassica nigra), and other weedy plant species occasionally interspersed with trees or small patches of coastal sage scrub. Presence of these areas was mainly a result of cattle ranching and the in- vasion of nonnative weeds into disturbed native habitats. These same weedy species were found to some extent in all of the habitats. Native grasses, such as perennial nee- dle grass {Nassella pulchra) , were present but were mainly restricted to small patches in the coastal sage scrub com- munity. Habiat boundaries were digitized using GIS software (ESRI 1995). Total area of each habitat type within an owl’s home range was determined by clipping the habitat polygon layer with the home-range boundary layer. The percentage of owl locations within each habitat was com- pared with the percent of each available habitat within the owl’s home range. Often an owl was found on an ecotone between two habitat types. In these instances one-half of a location was recorded for each habitat. We applied the two-tailed Mann-Whitney U-test to com- pare male and female home-range size and statistics are presented with standard error (SE). We used the Fried- man method (Friedman 1937, Alldredge and Ratti 1986) to test if Great Horned Owls used certain habitats pro- portionately more than the availability of that habitat within their home range. We analyzed the relationship between home-range size and percent of each habitat type found within an owl’s home range using the Spear- man’s rank correlation (r,). Results Five female and 10 male territorial-adult Great Horned Owls were fitted with radio transmitters and tracked during time periods ranging from 5- 17 mo (Table 1). Nine owls were caught near their occupied nests and six were caught outside the breeding season. All tracking periods were be- tween January 1997 and September 1998. The nesting success of one male was unknown, but all other owls fledged young successfully during at least one breeding season. We collected 1069 location points for 15 owls. Area-observation curves (Odum and Kuenzler 1955) using both 95% AK and 100% MGP were produced for each owl to ensure that enough lo- cation points were obtained to describe the home ranges adequately. The area estimated by both the AK and MCP approached an asymptote for most owls after ca. 50 ± 4.5 locations were obtained. Area-observation curves for four owls indicated that enough location points might not have been obtained to describe these home ranges adequate- ly. The signal for one owl was lost early in the study and the individual was removed from the analysis (M02; Table 1). Females FOG and F14 were tracked for ca. 6 mo starting from the late nestling stage through the fledging stage of their young. These locations may not be representative of a full year’s movements, but F14 had the largest home range of all females and her home-range size was unlikely underestimated. Because F14 and FOG were tracked over the same time period, both were included m home-range comparisons. The area-observation curves for Ml 8 were level from 10-35 location points, but toward the end of the study its home- range size more than doubled when it started using a new area outside its previous range. We used the home-range estimated before it moved for home- range analysis. 122 Bennett and Bloom VoL. 39, No. 2 Table 1. Home-range and core-area sizes (ha) for Great Horned Owls in southern California radiotracked between January 1997 and September 1998. Home ranges determined by adaptive kernel (AK), minimum convex polygon (MCP), and harmonic mean (HM) methods. Owl ID^ Months Tracked Number of Locations AK95% AK 50% Percent Core Area MCP 100% HM 95% FO 31 17 83 133 16 12 123 140 FO 52 11 76 152 31 20 146 130 F063 5.5 51 88 7 8 80 74 F08 9 68 244 44 18 212 275 FI44 5 61 282 37 13 256 271 Female Means 10 68 180 27 14 163 178 MOO 15 89 659 79 12 510 602 MOl 13 80 1115 187 17 1066 1195 M02i‘^ 3.5 16 285 46 16 160 145 M04 17 97 409 57 14 ' 465 451 MO 73 15 97 147 15 10 163 171 MOO 14 98 211 38 18 260 247 M12^ 6 70 589 96 16 484 698 MI 52 6 68 179 40 22 166 159 M17 6 68 257 21 8 188 382 M18 6 47 257 18 7 172 237 Male Means 11 79 425 61 14 386 460 ® Superscript number indicates pairs and F = female and M = male. Percent core area = (AK 50%)/(AK 95%) X 100. Removed from analysis and calculation of means. Female Home-Range Size. The 95% AK home range of five female Great Horned Owls averaged 180 ± 36 ha (Table 1). The largest female home range (282 ha) was more than three times larger than the smallest (88 ha). Both of these females fledged young successfully and were tracked dur- ing ca. the same 5-mo period. Male Home-Range Size. The 95% AK home range of nine male Great Horned Owls averaged 425 ±105 ha. The largest male home range (1115 ha) was more than seven times larger than the smallest male home range (147 ha). Both of these owls were tracked for 15 mo and nested. There was high variation among home-range siz- es. Although home-range size of males averaged more than twice that of females, it was not signifi- cantly different (G = 36, P — 0.04). The discrep- ancy in mean home range size between the sexes was due primarily to the large size of MOl’s home range (1115 ha), which was 1.7 times larger than the next largest male home range (659 ha). When we compared home ranges of four mated pairs the male’s home range encompassed most of the fe- male’s and was, on average, 36% larger. Core Areas. Core areas estimated by the 50% AK averaged 27 ± 7 ha for females and 61 ± 18 ha for males. The percentage of the core area aver- aged 14% ± 1.3% of the total home range and was nearly identical for both males and females. Core areas were centered on a few frequently-used perches at which the owls could be regularly found throughout the study. Nightly Home Range. Three males and three fe- males were tracked continuously over two entire nights and one male was tracked for one entire night. Owls often returned to the same perch after short visits to nearby perches and many were highly sedentary. Mean number of perch changes from the day roost through the night for all owls was 10.0 ± 1.3 (Table 2). The mean 95% AKarea used nightly was 46.2 ± 9.8 ha. This averaged 21.3% (range 0.9-39 ha) of the entire home-range size. There was no difference between the size of home range used {U = 19, P = 0.47) or the percentage of the home range used nightly {U — 25, P = 0.15) between males and females. Habitat Use, Not all habitat types were found in each owl’s home range and owls in the most ur- banized areas had the fewest habitat types within their home range. We found no correlation be- tween home-range size and any of the habitat types (r, < 0.4, N= 14, P> 0.05). June 2005 Great Horned Owi. Home Range 123 Table 2. Nightly 95% adaptive kernel (AK) home range (ha) and perch changes by seven Great Horned Owls in southern California radiotracked continuously from sunset to sunrise. Owl id 1998 Date Perch Changes Nightly Home Range (ha) Percent of Total Home Range F05 Sept 4 7 53 34.9 Sept 10 4 53 34.9 F06 June 19 12 34 38.6 July 20 17 12 13.6 F14 July 23 9 33 5.6 Aug 17 10 134 33.3 M04 Feb 22 15 49 12.0 M12 July? 5 a a Aug 17 10 37 6.3 M15 Aug 28 13 66 36.9 Sept 4 5 61 34.1 M17 Aug 14 6 2.3 0.9 Sept 7 17 20 7.8 Means 10 46 22 ^ Only two perches close in proximity were used, hence home range could not be estimated. Relative to availability, oak/ sycamore and ruden al grassland were used by Great Horned Owls to a greater extent than agriculture, exotic forest, coast- al sage scrub, and urban habitats {P = 0.03). Ru- deral grassland was found within each owl’s home range, but oak/sycamore woodland was absent from home ranges of five of the 14 owls (those in mostly urban areas) . Table 3. Percent of Great Horned Owl locations within respective home range (right). Discussion Few studies have attempted to determine home- range size and habitat use of Great Horned Owls in North America. Early estimates of home-range size relied on resighting unmarked individuals in Wyoming and Utah and ranged between 70-300 ha (Craighead and Craighead 1956, Smith 1969). Home-range estimates (cumulative grid square) of each habitat type (left) /percent of habitat type within Owl id Oak/ Sycamore Oak Exotic Forest Sage Scrub Riparian Scrub Ruderal Grassland Agriculture Urban MOO 47/12 0/2 a 51/81 1/3 1/0 0/1 a MOl 31/13 a 3/4 26/17 a 24/20 a 16/45 F03 a — 31/35 a — 62/47 — 7/18 M04 17/8 6/8 — 38/25 6/3 6/2 — 27/54 F05 1/3 31/1 A 1/1 19/38 25/16 20/20 3/4 — M15 0/4 o 20/1 A 1/3 34/36 26/16 17/21 2/6 — F06 22/5 — 3/9 — 16/21 0/7 59/58 M07 13/5 — — 4/14 — 19/21 2/4 62/56 F08 32/30 — — 10/14 — 28/20 — 30/35 MOO — — — — 8/9 13/7 — 79/84 M12 — — — — — 30/8 26/9 44/83 F14 — — — — — 13/1 14/13 73/86 M17 — — — — — 67/40 10/14 23/46 M18 1/2 — 6/5 4/9 — 58/44 31/40 — Habitat not found in home range. 124 Bennett and Bloom VoL. 39, No. 2 one radio-tagged female and two radio-tagged males in Minnesota were 71, 148, and 495 ha re- spectively (Fuller 1979). In the Yukon Territory, home-range size of 16 pairs of owls observed while hooting ranged from 230-883 ha = 483 ± 40 ha; Rohner 1997). Home-range sizes in this study were consistent with these observations — 337 ± 75 ha, range = 88-1115). As with studies of many other raptors, home- range size of females was smaller on average than that of males (Brown and Amadon 1968, Newton 1979, Bloom et al. 1993); perhaps, due to the fe- male’s responsibilities at the nest during the early and middle parts of nesting period. During incu- bation and brooding (January-March) , three fe- male owls with radiotransmitters rarely left their nests. Females found away from their nests when young were present were typically within 0.5 km of the nest structure. Male home-range sizes were consistently larger than female home ranges from nesding through the late post-fledging stages of breeding. Our sample size was too small to make seasonal home-range comparisons. Home-range sizes of both male and female Great Horned Owls varied substantially among individu- als. The largest home range (1115 ha) for males was 7.5 times larger than the smallest (147 ha) and 12 5 times larger than the smallest female home range (88 ha). This was not caused by differences in tracking periods as the largest and smallest home ranges for both males and females were from individuals tracked over ca. the same time periods. Rohner and Krebs (1998) found that home- range size of Great Horned Owls was related to owl density rather than to prey availability. Owl density was not measured here, but there was no home range overlap between three territorial males that held territories adjacent to one another. Also, vocal exchanges between radio-tagged owls and un- tagged adjacent owls were infrequent, suggesting relatively low densities of Great Horned Owls in the study area. Variation in home-range size may be due to such factors as prey abundance and availability across each owl’s territory. Although some studies have shown a correlation between home-range size and prey availability or preferred habitat (e.g., Carey et al. 1990, Bloom et al. 1993, Babcock 1995, Zabel et al. 1995, Mazur et al. 1998), we found no cor- relation between any single-habitat type or habitat groups and home-range size. Owls were located more often in oak/ sycamore woodland and ruderal grassland when available. Both of these habitats were ideally suited to a perch-and-wait predator, such as the Great Horned Owl, having numerous elevated perches with sparse or open ground cover. The lack of correla- tion between home-range size and habitat type may be due to differential prey availability within habi- tat classifications. Habitat types under the same classification were not homogeneous throughout the study area and prey availability may have varied between sites. Although prey abundance was not measured in this study. Great Horned Owls likely respond to prey availability by ranging more widely where availability is low (Newton 1979). During continuous all-night observations owls used a mean of 21% of their home range, but per- cent of total home-range use varied considerably (Table 2). Interestingly, one owl moved so little that the 95% AK home range could not be esti- mated and four owls used less than 8% of their total home range. During the fall, one female was observed on the same perch for eight continuous hours and only changed perches four times the en- tire night; there were no observed interactions with its mate. Anecdotal evidence suggests that hunting suc- cess may play a role in activity level. On several occasions owls were observed overlooking numer- ous cottontail rabbits with no apparent interest. Owls observed making unsuccessful attempts at prey capture usually continued to be active. Many factors may affect the space use by Great Horned Owls in southern California. In general, home-range sizes of birds of prey are strongly in- fluenced by the interactions of habitat availability, prey abundance and distribution, energetics, and territoriality (Newton 1979, Forsman et al. 1984, Bloom et al. 1993, Babcock 1995). Some of these factors may become more complex in areas of in- tense land development, where space, prey abun- dance, and prey vulnerability change rapidly. Great Horned Owls are an important upper-tro- phic level component of southern California’s wildlife communities with a wide prey base. These prey include mesopredators, such as the striped skunk {Mephitis mephitis), California ground squir- rel {Spermophilus beecheyi; R Bloom unpubl. data) and possibly young house cats (Felis domesticus) . An increase in mesopredator numbers due to the ex- clusion of top predators has been shown to be det- rimental to avian populations in fragmented habi- June 2005 Great Horned Owl Home Range 125 tats of southern California and elsewhere (Soule et al. 1988, Litvaitis and Villafuerte 1995, Rodgers and Caro 1998, Crooks and Soule 1999). Maintaining and managing Great Horned Owls, particularly on the interface between urban and natural areas, may act to reduce the threat of mesopredator re- lease and maintain greater biological diversity. Par- adoxically, Great Horned Owls are known to prey upon White-tailed Kites {Elanus leucurus;]. Bennett and P. Bloom pers. obs.) and Peregrine Falcons {Falco peregnnus] Walton and Thelander 1988), and they pose a threat to other sensitive species in the region. Although Great Horned Owls are successful in some human-altered landscapes, they are typically absent from most urban and suburban areas (P. Bloom unpubl. data) . As development continues to remove natural wildlife habitat, land-use planners and wildlife biologists need information on what is required to sustain healthy wildlife communities in the surrounding landscape. Urban and rural parks and preserves can encourage the presence of Great Horned Owls if they are provided an area of at least 425 ha in size with appropriate habitats to sustain an adequate prey base. Acknowledgments We would like to thank Charles Collins for direction, support and encouragement. We are indebted to Spence Porter of Communications Specialists, Inc. for donating the radio-tracking equipment-without his support this study would have been impossible. For assistance with trapping and tracking owls we would like to thank: Scott and Cheryl Thomas, Jeff Kidd, Michael VanHattem, Don- na Krucki, Greg Nicholson, Shoshana Mauzey, Sheri As- cari. Rich Beck, and Scott and Dawn Smithson. For di- rection and assistance with statistical analysis, GIS expertise, and other support, we would like to thank Mar- ni Koopman, Stewart Warter, Chris Nations, Nate Nib- belink, Rodd Kelsey, Greg Hayward, Rich Russell, and Amanda Hale. We thank Donna and Richard O’Niell of Rancho Mission Viejo for access to their land. Also, this project received donations from the Laguna Hills, South Coast, El Dorado and Sea, and Sage Audubon Societies. Lastly, we would like to thank Stuart Houston, Kurt Ma- zur, and Tom Morrell for important comments and sug- gestions that greatly improved the final manuscript. Literature Cited Alldredge, J.R. and J.T. Ratti. 1986. Comparison of some statistical techniques for analysis of resource se- lection./. Wildl. Manag. 50:157-165. Babcock, K.W. 1995. Home range and habitat use of breeding Swainson’s Hawks in the Sacramento Valley of California./. Raptor Res. 29:193-197. Bednarz, J.C., T. Hayden, and T. Fischer. 1990. The rap- tor and raven community of the Los Medanos area in southeastern New Mexico: a unique and significant resource. Pages 92-101 in R.S. Mitchell, C.J. Sheviak, and D.J. Leopold [Eds.], Proceedings of ecosystem management of rare species and significant habitats New York State Museum, Albany, NY U.S.A. Bent, A.C. 1938. Life histories of North American birds of prey, part 2. U.S. Natl. Mus. Bull. 170:295-357. Berger, D.D. and H.C. Mueller. 1959. The bal-chatri. a trap for the birds of prey. 30:18-26. Bloom, P. H. 1987. Capturing and handling raptors. Pag- es 99-123 in B.G. Pendleton, B.A. Millsap, K.W. Cline, and D.A. Bird [Eds.], Raptor management techniques manual. National Wildlife Federation, Washington, DC U.S.A. . 1989. Red-shouldered Hawk home range and habitat use in southern California. M.S. thesis, Cali- fornia State Univ., Long Beach, CA U.S. A. , J.L. Henckel, E.H. Henkel, J.K. Schmutz, B. WOODBRIDGE, J.R. BrYAN, R.L. AnDERSON, PJ. De- trich, T.L. Maechtle, J.O. McKinley, M.D. McCrary, K. Titus, and P.F. Schempf. 1992. The Dho-gaza with Great Horned Owl as lure: an analysis of its effective- ness in capturing raptors./. Raptor Res. 26:167-178. , M.D. McCrary, and MJ. Gibson. 1993. Red- shouldered Hawk home range and habitat use in southern California./ Wildl. Manag. 57:258-265. Brown, L. and D. Amadon. 1968. Eagles, hawks, and fal- cons of the world. McGraw-Hill Book Co., New York, NY U.S.A. Carey, A.B., J.A. Reid, and S.P. Horton. 1990. Spotted Owl home range and habitat use in southern Oregon coast ranges./. Wildl. Manag. 54:11-17. Craighead, JJ- and F.C. Craighead. 1956. Hawks, owls, and wildlife. Stackpole Co., Harrisburg, PA U.S.A. Crooks, K.R. and M.E. Soule. 1999. Mesopredator re- lease and avifaunal extinctions in a fragmented sys- tem. Nature 400:563-566. Dixon, K.R. and J.A. Chapman. 1980. Harmonic mean measure of animal activity areas. Ecology 61:1040- 1044. Dunstan, T.C. 1972. Radio-tagging Falconiform and Stri- giform birds. Raptor Res. 6:93-102. ESRI. 1995. Arcinfo. Environmental Systems Research In- stitute Inc., Redlands, CA U.S. A. Forsman, E.D., E.C. Meslow, and H.M. Wight. 1984. Dis- tribution and biology of the Spotted Owl in Oregon. Wildl. Monogr. 87. Friedman, M. 1937. The use of ranks to avoid the as- sumption of normality implicit in the analysis of var- iance./ Am. Stat. Assoc. 32:675-701. Fuller, M.R. 1979. Spatiotemporal ecology of four sym- patric species. Ph.D. dissertation, Univ. Minnesota, Minneapolis-St. Paul, MN U.S.A. Hamerstrom, F. 1963. The use of Great Horned Owls in catching marshhawks. Proc. Internat. Ornithol. Congr. 13:866-869. 126 Bennett and Bloom VoL. 39, No. 2 Hansteen, T.L., H.P. Andreassen, and R.A. Ims. 1997. Effects of spatiotemporal scale on autocorrelation and home range estimators./. Wildl. Manag. 61:280-290. Houston, C.S., D.G. Smith, and C. Rohner. 1998. Great Horned Owl {Bubo virginianus) . In A. Poole and F. Gill [Eds.], The birds of North America, No. 372. The Birds of North America, Inc., Philadelphia, PA U.S.A. Kie, J.G., J.A. Baldwin, and C.J. Evans. 1994. CALHOME: home range analysis program electronic user’s man- ual. U.S. Forest Service, Pacific Southwest Research Station, Fresno, 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. Litvaitis, J.A. and R. Villafuerte. 1995. Intraguild pre- dation, mesopredator release, and prey stability. Con- serv. Biol. 10:676-677. Mazur, K.M., S.D. Frith, and PC. James. 1998. Barred Owl home range and habitat selection in the boreal forests of central Saskatchewan. Auk 115:746—754. Mohr, C.O. 1947. Table of equivalent populations of North American small mammals. Am. Midi Nat. 37: 223-249. Newton, I. 1979. Population ecology of raptors. Buteo Books, Vermillion, SD U.S.A. Odum, E.P and EJ. Kuenzler. 1955. Measurement of ter- ritory and home range size in birds. Auk 72:128—137. Rodgers, C.M. and M.J. Caro. 1998. Song Sparrows, top carnivores and nest predation: a test of the mesopre- dator release hypothesis. Oecologia 116:227-233. Rohner, C. 1997. Non-territorial ‘floaters’ in Great Horned Owls: space use during a cyclic peak of snow- shoe hares. Anim. Behav. 53:901-912. AND C.J. Krebs. 1998. Responses of Great Horned Owls to experimental ‘hot spots’ of snowshoe hare density. Auk 115:694—705. Seaman, D.E. and R.A. Powell. 1996. An evaluation of the accuracy of kernel density estimators for home range analysis. Ecology 77:2075—2085. Smith, D.G. 1969. Nesting ecology of the Great Horned Owl, Bubo virginianus. Brigham Young Univ. Sci. Bull. Biol. Sen 10:16-25. Soule, M.E., D.T. Bolger, A.C. Alberts, J. Wright, M. Sorice, AiND S. Hill. 1988. Reconstructed dynamics of rapid extinctions of chaparral-requiring birds in ur- ban habitat islands. Conserv. Biol. 2:75-92. SwiHART, R.K. AND N.A. Slade. 1985. Testing for inde- pendence of observations in animal movements. Ecol- ogy 66:1176-1184. Walton, B.J. AND C.G. Thelander. 1988. Peregrine Fal- con management efforts in California, Oregon, Wash- ington, and Nevada. Pages 587-597 in T.J Cade, J.H. Enderson, C.G. Thelander, and C.M. White [Eds.], Peregrine Falcon populations: their management and recovery. The Peregrine Fund, Inc., Boise, ID U.S.A WORTON, B.J. 1989. Kernel methods for estimating the utilization distribution in home range studies. Ecology 70:164-168. . 1995. Using Monte Carlo simulation to evaluate kernel-based home-range estimators. J. Wildl. Manag 59:794-800. Zabel, C.J., K.S. McKelvey, and J.P Ward, Jr. 1995. In- fluence of primary prey on home-range size and hab- itat use patterns of Spotted Owls {Strix occidentalis) . Can. J. Zool. 73:433-439. Received 28 June 2004; accepted 1 1 March 2005 Associate Editor: Ian G. Warkentin J. Raptor R£s. 39(2) : 127-1 33 © 2005 The Raptor Research Foundation, Inc. EVALUATION OF METHODS FOR GENDER DETERMINATION OF LESSER KESTREL NESTLINGS Carlos Rodriguez, M^vier Bustamante, Begona Martinez-Cruz, and Juan Jose Negro Department of Applied Biology, Estacion Biologica de Donana(CSIC), Avda Maria Luisa s/n. Pabelldn del Peru 41013 Sevilla, Spain Abstract. — The traditional method of determining gender of Lesser Kestrel {Falco naumanni) nestlings by visual assessment was tested for accuracy by using data from birds banded as nestlings and recaptured as adults. Concordance between gender assignment by different observers, and between visual and molecular gender determination was also evaluated. We tested whether color measurement of rumps and tails could improve gender determination. Based on recaptured kestrels, gender determination by eye had a 9.7% error, and was significantly greater for males than for females. Observers mostly relied on rump and tail color to assign gender to nestlings. Assessment of head, shoulders, tail, and rump patterns did not provide additional information that could improve gender determination in nestlings at the time of banding. Gender assignment based on color measurement on digital photos of rumps and tails did not improve determination by eye, but color measurement from a scanned rump feather approached 100% accuracy. We provide a discriminant function equation based on red, green, and blue brightness values (RGB) of a scanned rump feather and propose this as an efficient and effective method for gender determination in Lesser Kestrel nestlings. Key Words: Lesser Kestrel, Falco naumanni; digital image analysis', gender determination', nestlings', RGB values. evaluacion de metodos para la determinacion del sexo en pollos de FALCO NAUMANNI Resumen. — Evaluamos la forma tradicional de determinar visualmente el sexo de los pollos de Falco naumanni mediante las recapturas de individuos adultos anillados, cuyo sexo habia sido determinado en la etapa de pollos. Se calculo la concordancia en la determinacion del sexo entre diferentes obser- vadores, asi como entre la determinacion del sexo de modo molecular y visual. Ademas, se investigo si medidas del color de la cola y la rabadilla determinadas a partir de fotografias digitales o de plumas de la rabadilla escaneadas aumentaban el porcentaje de acierto en la determinacion del sexo de los pollos. El porcentaje de error en la determinacion visual del sexo fue de 9.7%, y fue significativamente mayor en el caso de los machos. Los observadores se basaron mayormente en el color de la cola y la rabadilla para asignar el sexo a los pollos. Aunque fue dimorfico, el patron de manchas de la cabeza, los hombros, la cola y la rabadilla no aporto informacion adicional para mejorar la determinacion del sexo de los pollos en el momento del anillado. La determinacion del sexo a partir de las medidas de color tomadas de fotos digitales de la cola y la rabadilla ofrecio peores resultados que la determinacion visual tradi- cional. Sin embargo, la medida de color de la pluma de la rabadilla escaneada ofrecio un porcentaje de acierto en la determinacion del sexo cercano al 100%. Se ofrece una funcion discriminante, basada en los valores de brillo del rojo, verde y azul de las plumas de la rabadilla escaneadas, como un metodo eficaz para determinar mas confiablemente el sexo de los pollos de Falco naumanni. [Traduccion del equipo editorial] The Lesser Kestrel {Falco naumanni) is a small colonial falcon that exhibits a dichromatic plum- age. Adult males have an unspotted chestnut back, a mostly blue-gray inner wing, and a blue-gray hood. Adult females are brownish with dark bars on the head, back, and tail (Cramp and Simmons 1980). Juvenile plumage of both sexes resembles ^ Email address: carlos_r@ebd.csic.es that of adult females, but shows dichromatism in the rump and tail (blue-grayish in males versus brownish in females; Bijlsma et al. 1988, Negro and Hiraldo 1992, Telia et al. 1996b, Palumbo 1997). Tail and rump color can be determined when feathers start growing, and this happens when chicks are 2 wk old (pers. observ.). These charac- ters have been used traditionally to determine gen- der in Lesser Kestrel nestlings at the time of band- 127 128 Rodriguez et al. VoL. 39, No. 2 Table 1 . Recognized pattern categories and colors of the four characters used to assign gender of Lesser Kestrel nestlings in southwestern Spain. Head Plumage Shoulder Plumage Rump Plumage Tail Colors Down Unspotted Unstriped Unstriped with thin subterminal bar Brownish Unstriped Thinly spotted Striped Unstriped with thick subterminal bar Non-uniform gray Thinly striped Heavily striped Heavily spotted Thinly striped Striped Heavily striped Uniform gray ing, and it has been assumed that this visual gender assignment is accurate (Negro and Hiraldo 1992). However, based on our own experience of 13 yr working with the species, there were nestlings with intermediate coloration that were difficult to as- sign gender. Observers may differ in their assign- ment, and male features in adult females (Telia et al. 1997) and mosaic plumages have both been de- scribed for this species (Telia et al. 1996a). All this suggests that errors in gender assignment of nest- lings have been underestimated. There are characters in addition to rump and tail color (e.g., marking pattern of head, shoulders, rump, and tail) that show variability among nest- lings, and these seem to be associated with nestling gender. Not all banders seem to be aware of these differences, and it is unclear if consideration of other characters could improve gender determi- nation in the field. Molecular techniques could be used as a 100% accurate standard for other techniques (Ellegren and Sheldon 1997, but see Dawson et al. 2001), but require access to a genetics lab and have an eco- nomic cost. Ideally, methods of gender determi- nation in wildlife species should be inexpensive, produce an immediate result, and require a mini- mal amount of handling stress on birds. These techniques should also be accurate for all age groups and populations (Eason et al. 2001). For these reasons, field methods, which are based on differences in size or color between sexes (e.g., Borras et al. 1993, Martin et al. 2000, Balbontin et al. 2001), are advantageous. Nonetheless, methods based on plumage features need some develop- ment, and they have not been adequate for deter- mining the sex ratio at the time of hatching. The goal of this study is to increase the accuracy of gender determination in Lesser Kestrel nest- lings. Our objectives are: (1) to test the accuracy of the traditional visual gender determination em- ployed for Lesser Kestrel nestlings at the time of banding and (2) to evaluate alternative gender de- termination methods. For this purpose, we first tested if visual gender determination by banders is accurate by using data on birds banded as nestlings and recaptured as adults. Second, we determined gender in a sample of nestlings with molecular methods and considered this the reference gender assignment to test the accuracy of visual assign- ment by three observers. By using categorized col- or and plumage pattern in key areas of bird phys- iognomy, we compared the discrimination ability of each of these characters. Finally, we tried to im- prove the traditional visnal gender determination by building discriminant function models using color measurements of rump and tail from digital photos taken in the field and from color measure- ments of rump feathers in the lab. Methods Gender Determination by Banders. From 1988 to date. Lesser Kestrel nestlings have been banded in several col- onies in southwestern Spain. Gender determination was done by different banders following a visual assessment based on published differences in plumage (Bijlsma et al. 1988, Negro and Hiraldo 1992, Telia et al. 1996b, Pal- umbo 1997). We recaptured 476 nestlings as adults, which allowed us to evaluate the accuracy of the gender determination by banders. During the 2000 breeding season (early June to mid- July), 62 Lesser Kestrel nestlings from 18—33 d old were visually assigned to gender by three banders. Nestlings were classified according to marking pattern and color of four areas: head, shoulders, rump, and tail (Table 1) These features were evaluated independently, and a final gender determination was made by each observer con- sidering all the characteristics together. Capturing Images with a Digital Camera. We used a Kodak DC40 (Rochester, NYU.S.A.) flash enabled digital camera (DC) with a 756 X 504 pixel matrix and 24-bit color. The color value of each pixel is characterized by June 2005 Gender of Lesser Kestrel Nestlings 129 Table 2. Contingency table where the influence of family on gender determination was evaluated. Where P is the probability of assigning gender correctly, (1— P) was the complementary probability, and Nwas the number of pairs (Sokal and Rohlf 1995). Observed Expected Number of pairs with both siblings assigned to gender correctly or incorrectly. Number of pairs with one sibling assigned to gender correctly and the other incorrectly. N (P2 -h (1-P)2) V(2P (1-P)) brightness values of red, green, and blue (RGB) scaled m a range from 0-255. To reduce environmental variability, we photographed nestlings on a copy stand baseboard with flash illumina- tion. Moreover, as the same individual may still show some variation in its RGB values from photograph to photograph (Villafuerte and Negro 1998), we used two control chips (standards), provided by a gray scale card (Smithe 1975) along with the object to be photographed to further standardize the images. Photos were taken at similar distances to objects from directly overhead and using oblique views to provide further analysis of color from critical gender-determination areas. From each nestling, a rump feather was removed and later scanned with a desktop scanner (Hewlett-Packard Scanjet 5200c, Palo Alto, CA U.S.A.), setting the resolu- tion at 150 dpi. Analysis of color was made from the dig- ital image created using the same procedure as with the digital photos. The Software. Portions of the image to be analyzed (e.g., portion of tail between dark stripes) and portions of standard chips were respectively selected with the “las- so” and “Rectangle marquee” tools of Adobe Photo- shop® (San Jose, CA U.S.A.) for Windows® (Redmond, WA U.S.A.). Following the procedure used by Villafuerte and Negro (1998) to analyze digital images, color from each rump and each tail was separated into RGB values. The theoretical and observed values of the standard chips were used to calculate linear regressions for each primary color. Observed red, green, and blue values of the standard gray chips were used to correct the ob- served values in the rump and in the tail. This procedure makes RGB values from photos made under different il- lumination conditions comparable (Villafuerte and Ne- gro 1998). We did not follow this procedure with scanned feathers because the distance from the lens and the il- lumination source were always the same and, therefore, images could be compared directly. Molecular Gender Determination. A drop of blood was taken by venipuncture of the brachial vein and stored in 1-ml ethanol. Crude DNA extracts were prepared by boil- ing 5 |xl of the blood in 100 gl of a 100 mM NaOH solution for 10 min, then 0.5 gl of the supernatant was used directly as the template for PCR. The CHDIW and CHDIZ genes were amplified using primers 291 7F and 3088R (Ellegren 1996). Sexes can be discriminated in an agarose electrophorectic gel, as males display a single PCR product of around 550 bp, while females display also an additional product of 450 bp. PCR was performed in a final volume of 25 pi con- taining 16 mM (NH 4 ) 2 S 04 , 3.5 mM MgC^, 0.01% gelatin, 0.2 mM each dNTP, 0.2 pM each primer, and 0.04 U/pl of Taq DNA polymerase. The thermal profile comprised an initial denaturation step of 94°C for 2 min, followed by a single cycle of 2 min at 94°C, 30 sec at 55°C, and 1 min at 72°C, and 34 cycles of 30 sec at 92°C, 30 sec at 50°C, 45 sec at 72°C. A final extension step of 72°C for 5 min was added after the last cycle. The same cycling pa- rameters were used with all primer sets. Twenty pi of the PCR reaction were analyzed by electrophoresis in a 2% agarose gel containing 0.3 pg/ml ethidium bromide. Known male and female blood samples were used as pos- itive controls. PCR products were visualized and photo- graphed under UV light. Statistical Analysis. Because the probability of assigning gender correctly could covary among brood mates, their presence in the data set of recaptured birds can be con- sidered a source of pseudoreplication. Therefore, the probability of correctly assigning gender for a nestling was not independent from the gender assignment of his brood mate. For this reason, we tested whether this effect could influence our results. We selected pairs of nestlings of the same gender (25 pairs of male siblings and 19 pairs of female siblings) from the data set of resighted birds We subdivided these pairs into two groups; the first group comprised pairs in which gender determination for both siblings was either correct or incorrect (20 pairs of male siblings and 17 pairs of female siblings), and the second comprised those in which the gender of one member was assigned correctly while the other was assigned incor- rectly (five pairs of male siblings and two pairs of female siblings). We compared the distribution of these cases versus that expected by chance considering the proba- bility (P) of making a correct assignment (Table 2). In the few nests with more than two siblings of the same gender, two birds were selected at random. We also tested if siblings were more similar in color by testing for a brood effect on color values of scanned rump feathers with a generalized linear model (GLM, McCullagh and Nelder 1983). We used a GLM to test if different factors like age at the time of banding, nestling body condition (see Rod- riguez and Bustamante 2003) , and true gender could in- fluence the probability of determining the gender of a nestling successfully. The response variable in the model was correct gender determination (true/false), using a binomial error and a logit link. The statistical signifi- cance of each predictor (factor or continuous variable) was tested by sequentially removing all predictors from the complete model, starting from the one producing the smaller increase in the model deviance (Crawley 1993). Models were fitted using the GLM procedure of S-plus 2000 (Professional, Release 2. 1988-99 MathSoft, Inc., Seattle, WA U.S.A.). 130 Rodriguez et al. VoL. 39, No. 2 Table 3. General linear model built to test the explan- atory ability of age, nestling body condition, and sex on the probability of assigning gender correctly to the nest- lings. Each row represents the change in degrees of free- dom and deviance when the variable was removed from the model. Chi and P values are also shown. The null deviance = 278.5746 with 446 df and residual deviance = 269.1751 with 443 df Explanatory Variables A Chi P Percent Total Deviance Body condition 1 -1.33 0.18 0.6 Age 1 1.71 0.09 1 Sex 1 2.32 0.02 2 To test the concordance between molecular and visual gender determination (both character by character and the final gender evaluation for each observer), we cal- culated the Kappa value (percent of agreement corrected for chance agreement; Titus et al. 1984), then we tested the concordance between observers calculating the Kap- pa value from a contingency table in which each row rep- resented an individual classified as male, femcile, or un- known (the three categories of the columns). Cell entries were the number of observers agreeing on each category (Siegel and Castellan 1988). Finally, we built several discriminant functions through a forward stepwise variable selection procedure (F to en- ter = 3.0, Fto remove = 2.0, Tolerance = 0.01) in Sta- tistica 99 (StatSoft 1999). We built a discriminant func- tion for each set of predictive variables: (1) color and pattern recorded visually by each observer, (2) color from rump and tail measured on digital photos, and (3) color of scanned rump feathers. For the first one, we used as possible predictors the recorded category of col- or and marking pattern of rump and tail, and the cate- gory of the marking pattern of head and shoulders ob- taining a discriminant function for each one of the observers. For the last two, mean, minimum, maximum, and standard deviation values of red, green, and blue brightness values were used as potential predictors in the analyses. Results Brood Effect. The probability of correctly assign- ing gender for a bird within a brood was indepen- dent from the probability of success in the gender determination of his brood mate, both for males (P = 0.35, Fisher’s exact test) and for females (P = 0.33). Brood did not explain the variability in the brightness values of red (Pi , 55 = 0.85, P = 0.36), green (Pi , 55 = 1.5, P = 0.23), and blue (Pi,s 5 = 0.64, P = 0.43), which allowed us to use nestlings as independent sample units even when more than Pool Head Shotidets Tail Rimp Detefmjnalion Figure 1. Kappa values for the concordance between molecular and visual gender determination by both sin- gle characters and the pooled evaluation inferred from all the characters. Bar colors represent the three differ- ent observers (empty bars for observer 1, shaded bars for observer 2, and black bars for observer 3). Non-signifi- cant concordances were denoted as NS. one bird from the same brood was present in the sample. Success of Gender Determination by Banders. On average, 90.3% of kestrels recaptured as adults {N = 476) were assigned correctly to gender at the time of banding (k = 0.81, Z = 17.99, P< 0.01). This indicated that the method was in general ad- equate, but the error in gender determination was significantly greater than 0 (95% Confidence In- terval [Cl] = 8.0-14.0%). A significantly greater fraction of males than females were determined in- correctly (31/243 versus 15/233, respectively; P = 0.008, Fisher’s exact test) . Mean error rate in gen- der determination according to recaptures is 14.6% for males (95% Cl 10.0-20.0%) and 6.4% for females (95% Cl 4.0-10.0%). According to the GLM model, the success in gender determination was only related to the gen- der of the bird (Table 3) , which indicated a higher probability of assigning gender correctly for fe- males. The body condition of the nestling had no explanatory ability on its gender determination, and although there was a slight trend for increas- ing determination success with nesding age, this trend was not significant (Table 3) . Visual Gender Determination Characters. All characters evaluated to classify gender in Lesser Kestrel nesdings visually showed some degree of sexual dimorphism. For two of the observers, head and shoulder patterns were used with high accu- racy in classification of gender when the pattern was clear (Fig. 1), but many of the birds were un- June 2005 Gender of Lesser Kestrel Nestlings 131 Deleminafion Figure 2. Percent of birds for which gender could not be determined. Bar colors represent different observers (empty bars for observer 1, shaded bars for observer 2, and black bars for observer 3) . Exact values are provided above bars. determined based on this character (Fig. 2). The contrary pattern was found for the remaining ob- server (black bars in Figs. 1, 2), who classified more birds, but made more errors. Considering undetermined birds as assigned gender incorrectly, none of the observers achieved a significant agree- ment between molecular gender determination and head pattern (k = 0.14, k = —0.37, and k = “0.11 for the three observers, respectively) or shoulder pattern (k = —0.2, k = —0.2, and k = 0.03, respectively). Both tail and rump characters showed high agreement between molecular and vi- sual determinations for the three observers with a low number of unknown individuals. Among birds classified erroneously, observers were not consis- tent in agreement with their gender assignment (k = —0.17, Z— —1.072, P = 0.142), suggesting that these were individuals with intermediate character- istics. Differences Between Observers. Of 62 nestlings, 47 were evaluated by all three observers. Each ob- server evaluated 61, 57, and 48 nestlings, respec- tively (Table 4). There was a high agreement between observers whether we considered unde- termined birds as a third category (k = 0.77, Z = 4.5, P < 0.01) or as errors (k = 0.8, Z = 8.9, P < 0.01). The gender assignment by the three observ- ers had a high and significant agreement with mo- lecular gender determination. Percentages of cor- rect gender determination for each observer were: 97% (k = 0.93, P < 0.01), 96% (k = 0.92, P < 0.01), and 87% (k = 0.73, P< 0.01). The observers did not determine gender for 3%, 5%, and 0% of the nestlings, respectively. Including the undeter- mined birds as errors, the accuracy level of the ob- servers was similar to results from the recaptures (error rate = 7, 9, and 13% for each observer, re- spectively) . The small sample size did not allow us to test if males were misclassified more frequently than females. Discriminant Analyses. By building a discrimi- nant function of the color and pattern categories (Table 1) recorded by each observer, we obtained a different discriminant function for each observ- er. For the first observer the discriminant function included only rump color and resulted in an error frequency of 7%. For the second observer, the dis- criminant function included two variables: rump and tail color, and also had a 7% error. The dis- criminant function for the third observer used the shoulder pattern (plus tail and rump color), and produced a classification with 8% error. By using RGB values from digital photos of in- dividuals to build a discriminant function, we had an error frequency of 21% when using only rump color values, a 19% error when using only tail color values, and a 17% error using both tail and rump values. The best discriminant function included standard deviation of blue from tail, and standard deviation of red from rump (83% correct classifi- cation, N = 53 nestlings). Kappa value from the classification matrix of this discriminant function indicated an agreement with molecular gender de- termination significantly greater than chance (k = 0.66, Z = 4.74, P< 0.01). The color of scanned rump feathers that isolated the red, green, and blue brightness (RGB) com- Table 4. Number of birds for which we assigned gender by molecular and visual determination. The number of misclassifications is indicated between parentheses. Gender Molecular Technique Observer 1 Observer 2 Observer 3 Male 28 27 (1) 24 (3) 20 (3) Female 34 34 (3) 33 (1) 28 (1) Non-evaluated 0 1 5 14 132 Rodriguez et al. VoL. 39, No. 2 ponents resulted in the method with greatest ac- curacy. Males and females could be separated by mean blue (B) value, mean green (G) value, and standard deviation of green value. This discrimi- nant function correctly classified 98.2% of the in- dividuals (1.8% error, N — 57). The agreement be- tween the discriminant function classification and molecular gender determination was significantly greater than chance (k = 0.96, Z — 7.1, P< 0.01). The method provides a lower error rate than gen- der determination by banders according to recap- tures (Yates corrected Chi-square — 3.04, one- tailed P = 0.04), an error rate similar to those obtained by two of the banders, but lower than that obtained by the remaining one (P = 0.034, Fisher exact test) . The discriminant function equation to separate males (positive values) from females (neg- ative values) based on RGB values of rump’s scanned feather was; D = 25.2931 + 1.0002 (x5) - 1.046(xG) - 0.2298 (SD of G). Discussion Previous works (Negro and Hiraldo 1992, Apar- icio and Cordero 2001) with a limited sample of birds (A = 45 and A = 14, respectively) suggested that visual gender assignment based on plumage characteristics in Lesser Kestrel nestlings was 100% accurate. Our analysis involving a larger sample in- dicated that 9.7% of nestlings were incorrectly as- signed to gender by banders and those errors in males were twice as frequent as in females. Al- though all the visual characters evaluated showed a certain sexual dimorphism, the color of rump and tail were clearly the characters most useful in the gender determination of Lesser Kestrel nest- lings (Fig. 1). Marking patterns of rump and tail did not provide any useful extra information for gender determination. On the other hand, rump and tail color can be evaluated as soon as the rump and tail feathers start growing, while the marking patterns require a more developed feather before an accurate assessment can be made. The head and shoulder patterns showed a certain amount of dimorphism, visible only when the nestlings were close to fledging and the down had disappeared. For this reason, these characters tend to give a high percentage of birds classified as unknown. The discriminant functions built with the color and marking pattern categories as recorded by each ob- server supported this conclusion. Rump color was entered into the best discriminant function of all three observers, while shoulder pattern was en- tered only in the function of one of them. The color of the rump measured on a scanned rump feather was able to correctly determine the gender of 98.2% of the birds. This result was similar to the visual determination. The fact that color measure- ment on digital photos taken in the field per- formed worse than visual assignment suggests that our standardization of photographs was not ade- quate, and that differing illumination conditions had a strong influence on the result. This also in- dicated that a higher resolution camera should be used for this kind of analysis. In addition, field ob- servers can compare nestlings of the same or dif- ferent broods. This seems to be a useful advantage (Bijlsma et al. 1988) in distinguishing between males and females with intermediate characters. The discriminant function built from scanned rump feather color offers an inexpensive, relatively efficient, and objective way to classify gender of Lesser Kestrel nestlings, although a scanner reso- lution of 300 dpi is recommended (S. Talbot pers. comm.). It is measurably more accurate and objec- tive than the traditional visual method for observ- ers with variable level of experience (1.8% error rate versus 13% for the observer with less experi- ence) and circumvents potential biases due to var- iance within humans regarding the perception of color (McMahon et al. 2004) . To remove a feather from a nestling at the time of banding is a simple and relatively nonintrusive task. Because access to a genetic laboratory is not available to all research- ers, the utility of this method to improve gender determination accuracy for field biologists is obvi- ous, especially when errors related to visual gender assignment are likely skewed toward one gender. Nonetheless, different questions require different levels of accuracy in terms of gender determina- tion, and studies that need the maximum accuracy or focus on the primary sex ratio should always use molecular techniques. Also, feathers plucked for gender determination may be used to address oth- er behavioral and reproductive genetic questions (e.g.. Alcaide et al. in press). Acknowledgments We thank Manuel de la Riva, Yolanda Menor, and Jose Maria Bermudez for their collaboration gathering histor- ical recapture data. Javier Seoane and Manuel Calvo helped us with the fieldwork. Enrique Collado assisted us with the computer analysis. Africa Dominguez, Monica Gutierrez, and Maria Gonzalez collaborated in the mo- June 2005 Gender of Lesser Kestrel Nestlings 133 lecular gender determination procedures. Andrea Kral- jevic helped with corrections of the English text. Roger Jovani, R. Grippo, S. Talbot, and J. Loutsch provided crit- icism and helpful suggestions to an earlier version of this manuscript. This research has been funded by projects PB97-1154 and REN2001-2134-GLO of the Ministry of Science and Technology and Fondo Europeo para el DeSarrollo Regional funds from the European Union. C. Rodriguez was supported by a pre-doctoral fellowship from the Spanish Ministry of Science and Technology. B. Martinez-Cruz was supported by a predoctoral fellowship from the La Rioja government. Literature Cited Alcaide, M., JJ- Negro, D. Serrano, J.L. Telia, and C. Rodriguez. In press. Extra-pair paternity in the Lesser Kestrel: a re-evaluation using microsatellite markers. Ibis. Aparicio, J.M. AND R Cordero. 2001. The effects of the minimum threshold condition for breeding on off- spring sex-ratio adjustment in the Lesser Kestrel. Evo- lution 55:1188-1197. Balbontin, J., M. Ferrer, and E. Casado. 2001. Sex de- termination in booted eagles {Hieraaetus pennatus) us- ing molecular procedures and discriminant function analysis./. Raptor Res. 35:20-23. Bijlsma, S., E.J.M. Hagemeijer, G.J.M. Verkley, and R. Zollinger. 1988. Ecological aspects of the Lesser Kes- trel Falco naumanni in Extremadura (Spain). Werk- groep Dieroecologie, Vakgroep Experimentele Zool- ogie, Katholieke Univ., Nijmegen, Netherlands. Borras, a., j. Cabrera, X. Colome, and J.C. Senar. 1993. Sexing fledglings of Cardueline finches by plumage color and morphometric variables./. Field Ornithol. 64: 199-204. Cramp, S. and K.E.L. Simmons. 1980. Handbook of the birds of Europe, the Middle East and North Africa. Oxford Univ. Press, Oxford, U.K. Crawley, M.J. 1993. GLM for ecologists. Blackwell Sci- entific Publications, Oxford, U.K. Dawson, D.A., S. Darby, F.M. Hunter, A.P. Krupa, I.L. Jones, and T. Burke. 2001. A critique of avian CHD- based molecular sexing protocols illustrated by a z- chromosome polymorphism detected in auklets. Mol. Ecol. Notes 1:201-204. Eason, D., C.D. Millar, A, Cree, J. Halverson, and D.M. Lambert. 2001. A comparison of five methods for as- signment of sex in the Takahe (Aves: Porphyrio man- telli).J Zool. Fond. 253:281-292. Ellegren, H. 1996. First gene on the avian W chromo- some (CHD) provides a tag for universal sexing of non-ratite birds. Proc, R. Soc. Fond. B. 263:1635-1641 AND B.C. Sheldon. 1997. New tools for sex iden- tification and the study of sex allocation in birds. Trends Ecol. Evol. 12:255-259. Martin, C.A., J.C. Alonso, J.A. Alonso, M.B. Morales, AND C. PiTRA. 2000. An approach to sexing young Great Bustards Otis tarda using discriminant analysis and molecular techniques. Bird Study 47:147-153. McCullagh, P. and J.A. Nelder. 1983. Generalized lin- ear modeling. Chapman and Hall, London, U.K. McMahon, C., J. Neitz, and M. Neitz. 2004. Evaluating the human X-chromomsome pigment gene promoter sequences as predictors of L:M cone ratio variation./ Vision 4:203-208. Negro, JJ- and F. Hiraldo. 1992. Sex ratios in broods of the Lesser Kestrel. Ibis 134:190-191. Palumbo, G. 1997. II Grillaio. A1 trimedia, Matera, Italy Rodriguez, C. and J. Bustamante. 2003. The effect of weather on Lesser Kestrel breeding success: can cli- mate change explain historical population declines? /. Anim. Ecol. 72:793-810. Siegel, S. and NJ. Castellan. 1988. Nonparametric sta- tistics for the behavioral sciences. McGraw-Hill Inter- national Editions, Singapore, Singapore. Smithe, F.B. 1975. Naturalist’s color guide. American Mu- seum of Natural History, New York, NY U.S.A. SOKAL, R.R. AND F.J. Rohlf. 1995. Biometry: the princi- ples and practice of statistics in biological research W.H. Freeman and Company, New York, NY U.S.A. StatSoft, Inc. 1999. STATISTICA for Windows. StatSoft, Inc., Tulsa, OK U.S.A. Telia, J.L., J.A. Donazar, and E. Hiraldo. 1996a. Vari- able expression of sexually mosaic plumage in female Lesser Kestrels. Condor 98:643-644. . . J-J- Negro, and F. Hiraldo. 1996b. Sea- sonal and interannual variations in the sex-ratio of Lesser Kestrel broods. Ibis 138:342-345. , M.G. Forero, J.A. Donazar, and F. Hiraldo. 1997. Is the expression of male traits in female Lesser Kestrels related to sexual selection? Ethology 103:72- 81. Titus, K., J.A. Mosher, and B.K. Williams. 1984. Chance- corrected classification for use in discriminant analy- sis: Ecological applications. Am. Midi. Nat. 111:1—7. Villaeuerte, R. andJ.J. Negro. 1998. Digital imaging for colour measurement in ecological research. Ecol. Fett. 1:151-154. Received 14 June 2002; accepted 1 March 2005 J Raptor Res. 39(2): 134-1 41 © 2005 The Raptor Research Foundation, Inc. USING VOCAL INDIVIDUALITY TO MONITOR QUEEN CHARLOTTE SAW-WHET OWLS {AEGOLIUS ACADICUS BROOKSI) Carmen I. Holschuh^ and Ken A. Otter Ecosystem Science & Management Program, University of Northern British Columbia, 3333 University Way, Prince George, BC V2N 4Z9 Canada Abstract.^ — Our goal was to assess whether male Northern Saw-whet Owls (Aegolius acadicus brooksi) in the Queen Charlotte Islands have sufficient variation in the notes of their territorial/ advertisement call to identify individuals within and between breeding seasons. Recordings of calling bouts were collected from 24 sites over the 2-yr study period and analyzed using discriminant function analysis. Using re- cordings of eight sites in 2002 and 16 sites in 2003, we correctly identified novel notes collected during a single recording session for 69% and 75% of the samples, respectively. Of the 13 sites for which we had multiple nights of recordings, 73% of notes measured from a novel night were classified to the correct site. The data also suggested that vocal individuality can be used to monitor site fidelity and rates of territory turnover of male owls across years, as correct classification of males re-occupying specific sites across years (>60%) was greater than rates of random classification (x = 3.3%, maximum = 19%). Vocal individuality, thus, appears to have promise as a useful tool for monitoring and studying life history parameters of Northern Saw-whet Owls both within and between years. Keywords: Northern Saw-whet Owl, Aegolius acadicus brooksi; site fidelity, vocal individuality. USO DE la INDfVIDUALIDAD VOCAL PARA MONITOREAR BUHOS AEGOUUS ACADICUS BROOK- SI EN IAS ISLAS QUEEN CHARLOTTE Resumen. — Nuestro objetivo fue determinar si los machos de Aegolius acadicus brooksi de las islas Queen Charlotte presentan suficiente variacion en las notas de sus llamados para permitir la identificacion de individuos dentro de una epoca reproductiva y entre epocas reproductivas. Durante un periodo de dos anos, recolectamos grabaciones de vocalizaciones en un total de 24 sitios y las sometimos a analisis discriminantes. Usando grabaciones de ocho sitios en 2002 y de 16 sitios en 2003, identificamos correcta- mente las notas distintas obtenidas durante una sola sesion de grabacion en el 69% y 75% de las muestras, respectivamente. En los 13 sitios para los que teniamos grabaciones hechas durante varias noches, el 73% de las notas medidas en grabaciones de una noche distinta fueron clasificadas correc- tamente de acuerdo al sitio. Los datos tambien sugirieron que la individualidad vocal puede emplearse para monitorear la fidelidad al sitio y las tasas de recambio de los territorios de los machos a traves de los anos, ya que la clasificacion correcta de los machos que ocuparon sitios especificos repetidamente en distintos anos (>60%) fue mayor que las tasas de clasificacion aleatoria (x = 3.3%, maximo = 19%). Asi, la individualidad vocal parece ser una herramienta promisoria para estudiar parametros de la his- toria de vida de A. a. brooksi tanto en un ano como entre anos. [Traduccion del equipo editorial] Avian studies that examine life-history parame- ters of focal species often require investigators to differentiate among individuals in a population; this is usually accomplished via the application of colored leg bands or radiotelemetry transmitters (Bibby et al. 2000). However, when working on cryptic species, the utility of visual identifiers, such ^ Corresponding author’s present address: CIH Environ- mental Consulting, #3-2506 Prior Street, Victoria, BC VST 3X6 Canada; Email address: holschuh@telus.net as leg bands, decreases. Similarly, the invasive na- ture of radiotelemetry (involving potentially mul- tiple captures of the bird as well as the added mass imposed by the transmitters) may be deemed counterproductive in sensitive species (e.g., Foster et al. 1992). Thus, any cues that study animals in- herently possess that distinguish them individually should be explored as a noninvasive method of study. One such cue, vocal individuality in birds, has recently gained popularity as a technique to study visually-cryptic species. Numerous species 134 June 2005 Monitoring Owls through Vocalizations 135 have now been shown to possess sufficient varia- tion in their vocalizations to allow individual iden- tification, including: Eurasian Pygmy Owls {Glau- ddium passerinum; Galeotti et al. 1993), European Nightjars {Caprimulgus europaeus] Reebeck et al. 2001) , Great Bitterns (Botaurus stellaris; McGregor and Byle 1992), Tawny Owls {Strix aluco; Galeotti and Pavan 1991), Bald Eagles (Haliaeetus leucoce- phalus; Eakle et al. 1989), Corncrakes {Crex crex, Peake et al. 1998) and Northern Saw-whet Owls (Aegolius acadicus] Otter 1996). An ability to identify male owls accurately within years can yield information about population size, by increasing census accuracy (e.g., Gilbert et al. 1994, Peake and McGregor 2001), and site-specific territory turnovers. Conversely, across year com- parisons using vocal characteristics can also indi- cate site fidelity and between-year territory turn- over (e.g., Galeotti and Sacchi 2001, Delport et al. 2002) . For both types of studies, analysis requires not only being able to distinguish between individ- uals based solely on their vocalizations, but also to classify correctly, and hence recognize, known in- dividuals (McGregor et al. 2000) . In this study, we determined the feasibility of monitoring individual male saw-whet owls using their vocalizations, within and between nights during a single year and be- tween years, to determine levels of site fidelity and territory turnover. There are two distinct subspecies of Northern Saw-whet Owl in North America; Aegolius acadicus acadicus occupies most of North America, while A. a. brooksi is isolated on an archipelago in north- western British Columbia, the Queen Charlotte Is- lands. A. a. acadicus is known to be partially migra- tory and displays low levels of site fidelity and, hence, high territory turnover between years (Can- nings 1993). Conversely, A. a. brooksi is nonmigra- tory, yet little is known about other differences in life history of this subspecies, such as prevalence of site fidelity and rates of turnover. The nocturnal nature of these owls coupled with difficult field conditions, makes monitoring individual owls by traditional means, such as banding or color mark- ing, difficult. However, males are highly vocal dur- ing the pre-nesting period (Cannings 1993), mak- ing male Queen Charlotte Islands saw-whet owls well suited for monitoring via vocal-individuality techniques. Study Area Saw-whet owl recordings were collected in the Queen Charlotte Islands of northwestern British Columbia, Can- Figure 1. Sonogram of saw-whet owl male advertisement call, illustrating measures taken. Fs, Fm, and Fe refer to measures of frequency at peak amplitude taken from spectral slices at 10%, 50%, and 90% of the note’s du- ration, respectively. Temporal measures were the length of the note (T note) and the internote length (T inter- note), referring to the time from the start of one note to the start of the next note. ada (53°N, 131°W). This area is a coastal temperate rain- forest with a cool mesothermal climate (cool summers and mild winters) with annual precipitation exceeding 130 cm (Environment Canada). The most common tree species are western hemlock ( Tsuga heterophylla) , western redcedar {Thuja plicata), Sitka spruce {Picea sitchensis), yellow cedar ( Chamaecyparis nootkatensis) , mountain hem- lock ( Tsuga mertensiana) , shore pine {Pinus contorta van contorta) , and red alder {Alnus rubra) . The research, with exception of one site, focused on the southern half of Graham Island, the largest island in the archipelago. Parts of the study area have been extensively modified by logging, creating a patchwork of remnant mature and old forest among regenerating stands. Methods We collected recordings of vocal male A. a. brooksi dur- ing the pre-nesting period (March-May) of 2002 and 2003 within the first 5 hr after sunset. Calls of owls were induced using a 1-min recording of typical saw-whet owl advertisement calling, which is a series of monotonous notes (Fig. 1) repeated at a rate of ca. 2/sec (Cannings 1993) broadcast through a portable CD player and am- plified powerhorn (RadioShack 32-2037). All calling ac- tivity was recorded using either Sennheiser MKH 70 or ME 67 directional, shotgun microphones (Sennheiser Electronic GmbH & Co. KG, Wedemark, Germany) in combination with Marantz PMD 430 tape recorders (Ni- hon Marantz Kabushiki Kaisha, Tokyo, Japan). Record- ings were restricted to nights with little or no precipita- tion and low winds. We attempted to approach vocal birds as closely as possible to obtain high-quality record- ings, although the structure of the landscape often re- stricted our approach. Most recordings were made within 100 m of individual birds, with a range up to 300 m in some cases. Distances to the bird were estimated in the field by compass triangulation on the calling owl in ad- dition to estimating direct-line distance. Considering that unobstructed saw-whet owl advertisement vocalizations can travel as far as 1000 m (Cannings 1993) and that the frequency and structural characteristics of this call ap- pear nearly optimal for minimizing degradation in dense 136 Holschuh and Otter VoL. 39, No. 2 habitat (Wiley and Richards 1982), using recordings col- lected at a range of up to 300 m should still allow stan- dardized measures to be taken on the individual call notes. When possible, we obtained recordings that min- imized environmental background noise (e.g., stream noise, ocean surf). Once a site was deemed occupied, it was visited a minimum of once per wk for the rest of the season in order to monitor the activity and to obtain re- peat-recordings of the males. In total, we recorded calling male saw-whet owls from at least one night at 24 different sites, eight during 2002 and 16 during 2003. Logistics suggested we were record- ing the same male owls at a particular site, as (1) densi- ties of owls were generally very low due to the fragment- ed nature of the habitat, and thus, in most cases, it was unlikely there was a second adjacent territory if the focal male had shifted call posts; (2) when there were multiple territories at higher density, simultaneous calling allowed us to identify the males; and (3) the birds were consistent in their calling location, generally calling from within 150 m of previous detection locations. This justification par- allels past studies with similar objectives (Galeotti and Pa- van 1991, Galeotti et al. 1993, Galeotti and Sacchi 2001). To confirm this assumption, in 2003 we fitted a subset of three male saw-whet owls with small (0.5 g, 3-wk battery life) tail-mount radiotransmitters (Model LB-2, Holohil Electronics, Carp, Ontario, Canada), thereby assuring us that we were recording the same male over multiple nights. Sonographic Analysis. For each site, we measured a minimum of 20 representative notes of advertisement calling if we only had a single recording session, and up to 56 notes if we had multiple recording sessions, leading to a mean of 35.1 ± 11.2 (SD) notes measured per site. We had collected a mean of 2.2 ±1.2 recordings per site (range = 1-5). When first screening the recordings, we saved sections of calling that were of high quality (mini- mal background noise and closest possible proximity to the bird). From those saved files, we randomly chose notes to analyze. Because recordings came in response to a series of 1-min playbacks of the advertisement call, we analyzed only the typical advertisement call and avoid- ed measuring notes given during the more rapid and quiet call that some males seem to include between ad- vertisement calling bouts when highly agitated (also called “introductory notes” [Otter 1996]). All notes were analyzed using Avisoft SAS Lab Pro 3.8 (Specht 2000). The temporal measures taken were note length (T-note) and internote length (T-internote), which were measured on the waveform screen at a reso- lution of 2.9 msec. As the frequency range of male saw- whet owl advertisement calls were between 1100-1350 Hz, we filtered out all noise below 900 Hz and above 1500 Hz to focus in on the bird’s signal itself. We defined note length as being from the start of the note at which point the amplitude first begins to increase to the end of the note at which point the amplitude is at one third of its maximum (Fig. 1). This technique was used to account for any differences in the degradation of the notes and to reduce measuring reverberation noise as being part of the note length. This allowed us to measure notes from calls collected at variable distances (25-300 m) from birds without fear of including signal degradation into the temporal variation between males. The internote length spans from the start of the focal note to the start of the following note. The frequency parameters we mea- sured were the frequency at the beginning (Fs = spectral slice at point 10% into note, as defined by temporal mea- sures), middle (Fm = point at 50% of note length), and end (Fe = point at 90% of note length) of the note. We measured frequency parameters at a frequency resolu- tion of 20 Hz, a temporal resolution of 2.9 ms, and a bandwidth of 56 Hz. Data Analysis and Individual Recognition. We ran pre- liminary statistics to assess the variability within each of the five parameters used to classify calls (Galeotti and Sacchi 2001). Analysis of variance (ANOVA) tests for each of the variables, using site identity as the grouping factor, allowed us to measure whether there were appre- ciable differences in the notes of individual birds at each of the variables. To classify males successfully, it is nec- essary for the variation in the individual variables to be less within individual sites than the variation between sites. We confirmed this by calculating coefficients of var- iation of each call measured across all sites (mean across the population), and compared these to mean coeffi- cients of variation within sites (mean within males). To test the ability to classify individuals correctly based on their vocalizations, we used the general discriminant analysis function (forward stepwise, P to enter/ remove = 0.05) in STATISTICA (StatSoft, Inc. 2003). We analyzed calls from a single night of recordings for each of the eight sites from 2002 and 16 sites from 2003 separately to determine levels of correct classification within a sin- gle night (i.e., known to be recordings from the same male, as notes were taken from a continuous bout of call- ing) . The discriminant function was built from a learning set of 66% of the calls. A test set contains the remaining calls, which are not used in the discriminant function itself, but are introduced as novel cases to be classified to site by the model. By using this approach of classifying novel notes, the estimation of correct classification rates should be more representative than when using more common analysis techniques in which the same notes that were used to build the discriminant function are lat- er classified (Terry and McGregor 2002). To test the ability of the model to classify males to the correct sites across nights, we used 13 sites from both years for which we had multiple detections. We used notes that were collected from one night to build a learn- ing set (from which the discriminant function was devel- oped), and notes recorded from a separate night during the same season as a test-set. Using the cross-validation function, we classified test notes to specific sites using the discriminant function built from the learning set. If the classification accuracy for a single male between nights was comparable to within nights (see above analysis), it is likely the same male, whereas if the between night com- parison has lower classification success, a territory turn- over event may have occurred. Site Fidelity/Territory Turnover. To determine wheth- er territories were being occupied for both field seasons, we repeatedly revisited all sites during the 2003 field sea- son that had been occupied in 2002. A site was deemed occupied in both years if we found calling activity on any night during the 2003 visits. We monitored occupied sites June 2005 Monitoring Owls through Vocalizations 137 Table 1. Parameters characterizing the notes of advertisement/ territorial calling of male saw-whet owls. Notes mea- sured for each of 24 males (8 from 2002 and 16 from 2003) were used to derive a coefficient of variation for measured parameters for each male; mean coefficients of variation for individuals (CV ind.) were then calculated. Means, standard deviations (SD), and range (maximum and minimum values) and coefficients of variation across males in the population (CV group) were derived from the mean value of each note parameter across the 24 males (N), F values reflect a comparison between CV group compared to CV individual. N Mean ± SD Minimum Maximum CV (Group) CV (Ind.) T-ratio df P T-note 24 112 ± 13 ms 72 ms 142 ms 12.03 6.64 85 23 <0.001 T-internote 24 431 ± 64 ms 273 ms 697 ms 14.80 9.47 46 23 <0.001 F (s) 24 1173 ± 41 Hz 1070 Hz 1310 Hz 3.50 1.37 191 23 <0.001 F (m) 24 1194 ± 48 Hz 1090 Hz 1350 Hz 3.98 1.41 210 23 <0.001 F (e) 24 1197 ± 51 Hz 1090 Hz 1370 Hz 4.37 1.49 224 23 <0.001 regularly (1/wk) and attempted to re-record the resident male. For sites with 2 yr of recordings, we ran a discrim- inant function analysis comparing recordings from the site in either year using the cross-validation function in the general discriminant analysis module. Recordings from 2002 were used as a learning set to develop the discriminant model; with this model, we attempted to classify recordings from the same site in 2003, which con- stituted the test set. To determine how the model would classify songs from two different males (such as would occur if the resident male differed between years), we also included seven sites from 2002 in the learning set that were not reoccupied in 2003, and randomly paired these with seven different sites (and presumably different males) from 2003. This control set served two functions: (1) it allowed us to determine how frequently calls from two different males would be cross-classified by chance, and (2) by increasing the total number of sites in the learning set, we decreased the probability that our focal sites would be cross-classified by chance alone. If the level of cross-classification of the focal sites was greater than the level of classification among sites vrith known differ- ent individuals (control set), we considered the site to be occupied by the same male. Conversely, if reclassification between years was low and similar to the control set, it suggested that a territorial turnover had occurred. Results The coefficients of variation in call measures were significantly greater across males (CV groups) in the population than the mean variation within an individual male (CV ind; Table 1), Within Year Vocal Individuality. In measuring the ability to correctly classify notes collected with- in a single recording session, the discriminant function analysis had an 84% ability to classify notes correctly used to build the discriminant func- tion, and a 69% ability to classify novel notes cor- rectly introduced into the analysis (test set) for the eight sites in 2002. In 2003, when 16 sites were used, the discriminant function correctly classified 84% of notes used in the learning set of the dis- criminant function and 75% of the novel notes used in the test set. We were able to obtain multiple nights of within- year recordings for 13 males. On average, we ana- lyzed 46 (± 9 SD, range 31-56) notes for each of these males. We used a mean of 30 (± 9 SD, range = 15-39) of these notes from recordings on a sin- gle night as the learning set to create the discrim- inant function and 15 (± 4 SD, range = 9-22) notes from a separate night as the test set to test the discriminant model in cross-validation analysis. We were able to classify notes correctly within the learning set a mean of 79% of the time, while clas- sification of notes from a different night (test set) had a mean of 74% correct classification (Table 2). Correct classification of recordings across different nights from males fitted with radiotransmitters ranged from 38—100%, indicating that there was variation between nights in the note structure of known males. One site had a disproportionately lower level of correct classification than the others, which may have indicated that a different male took over the territory between the two recordings (separated by 5 wk) . When classification accuracy was examined within night, 70% of novel notes introduced into the discriminant analysis were classified correctly, whereas only 7% of notes were classified correctly from a different night. All variables significantly contributed to the dis- criminant function (Table 3), although the vari- ables that loaded most heavily on building the function were note length and frequency at the end of the note (indicated by highest T-ratio val- ues). We explored the misclassification of particu- 138 Holschuh and Otter VoL. 39, No. 2 Table 2. Percent of correct classification of notes in male saw-whet owls with multiple recordings within sea- sons. The learning set was composed from recordings collected during one or more detections at each site. The test set calls originated from a separate recording session at the same site. Hence, the test set notes were not used m building the discriminant function; they were used in- stead for testing correct classification of novel notes. Site Within Night Year Classification Between Night Classification 1 2002 100 87 2 2002 100 87 3 2002 85 60 4 2002 100 87 5 2003 40 78 6 2003 85 81 7 2003 58 95 8 2003 100 85 9 2003 70 7 10 2003 85 73 IH 2003 100 60 12a 2003 16 100 13^ 2003 80 39 Males had transmitters that allowed for the confirmation of identity. lar males by comparing the distances to the nearest neighbor and to the site to which a focal male was most often misclassified. We found that males were usually misclassified not with adjacent sites, but with sites that were significantly further away than mean inter-male spacing {t = —7.531, df = 19, P < 0.0001; Fig. 2). Site Fidelity and Territory Turnover in Saw-whet Owls. Of the 13 sites from 2002 that we monitored regularly for both field seasons, seven sites were occupied again in 2003, indicating a 53% reoccu- pancy rate of sites. We were able to collect record- ings of sufficient quality at three of these sites to analyze whether the same males were occupying those territories (Fig. 3). By examining the cross- classification rate of control sites, we determined a random cross-classification of calls from two differ- ent males averaged 3%, with a maximum of 19%. Two of the three focal sites occupied both years had much greater levels of correct classification (>60%) than this, suggesting that the same male was occupying the sites during both breeding sea- sons. However, one site had 0% correct classifica- tion of notes, indicating a turnover had likely oc- curred (Table 4). Table 3. Results of multivariate analysis (Wilks’s test) to deterimine differences between calling males at each of the contributing parameters measured on the notes. F-ratio df P T-note 47.7656 23 <0.0001 T-internote 25.6395 23 <0.00001 F (s) 21.9992 23 <0.00001 F (m) 24.9339 23 <0.00001 F (e) 37.6732 23 <0.00001 Discussion Our results suggest that male Northern Saw-whet Owls may have sufficient variation in their adver- tisement calls to allow successful discrimination be- tween individuals within and between nights, and that it may be possible to use this technique to monitor these owls across years. However, some limitations exist that affect the reliability of iden- tifying some males correctly. Inherent variation in male owl advertisement calls of some individuals still present challenges in classifying some males to the correct sites. Further, our study site was char- acterized by a rugged landscape, which often pre- vented close approach to vocal owls, thereby some- times compromising recording quality. Although we took steps to limit measuring reverberation in calls, such degradation could still introduce vari- ability that can result in lowered discriminant abil- ity to identify birds. Also, as it is primarily the male saw-whet owls who employ the advertisement call, 24 1 . 1 ^ 1 22 ■ 20 ■ 1 10 ■ ** 16 ■ ? ^ ■ o> o c 4S 12 • G 10 - 0 ■ 6 - A ■ 2 I ■ ■ ' ■ Nearest Neighbor Misclassified Male Figure 2. Comparison of mean distances between near- est neighbor to each male saw-whet owl and the mean distance to the site any given site was most often misclas- sified as {N = 20). June 2005 Monitoring Owls through Vocauzations 139 A B C 1 sec 1 sec 1 sec 2003 1 sec 1 sec 1 sec Time (sec) Figure 3. Spectrograms of three male Queen Charlotte saw-whet owls occupying the same sites in 2002 and 2003. Site location is indicated as A, B, or C. Recordings of males occupying the territories in either 2002 (upper spectro- gram) or 2003 (lower spectrogram) are shown. Discriminate analysis on vocal characteristics indicated that the same males reoccupied sites B and C across the two years, but discrepancies in vocal signals between years in site A suggest a territorial turnover. this technique is not as applicable in learning about female territoriality and site fidelity. Some of the lack of reliability can be overcome by interpreting the results in context of the loca- tion of calling males across the landscape. When the model misclassified males, it usually attributed the note to a male distant from the singer. Neigh- boring males seem to not converge on similar call patterns, either by overlapping or frequency matching each other in call structure, the converse Table 4. Classification of novel notes recorded in 2003 (test set) using discriminant functions based on the learning set of recordings collected at sites in 2002. The numbered sites (1— V) refer to control sites, where the test set (2003) recordings were from different males than in the learning set (2002). The lettered sites (A-C) were sites that were occupied both years of the study. The numbers in the matrix refer to the classification of individual notes recorded in 2003 (test set) to individual sites. Of the sites occupied both years, sites B and C appeared to be defended by the same male, whereas site A likely experienced a territory turnover. Site Percent Correct No. Notes . Learning Set (2002 Recordings) Classified 1 2 3 4 5 6 7 A B C Test set (2003 recordings) 1 0.00 38 0 4 16 3 14 1 2 0.00 32 1 0 5 8 1 12 4 1 3 0.00 24 13 0 7 4 4 4.26 47 4 27 2 1 13 5 0.00 30 27 0 3 6 0.00 15 0 10 5 7 18.75 48 4 12 9 2 2 19 Control 3.29 A 0.00 17 1 3 3 5 0 5 B 64.91 57 5 2 5 5 2 1 37 C 61.29 31 9 3 19 140 Holschuh and Otter VoL. 39, No. 2 of which occurs in passerines (e.g., Morton and Young 1986, Otter et al. 2002). Thus, a combina- tion of vocal cues and knowledge of male location should provide some certainty of identifying resi- dent males correctly. The utility of this technique may be limited for comparing calls between years if males taking over a territory were to converge on the calling patterns of previous residents, with the note structure some- how being linked to location (e.g., possibly to in- crease transmission in different microhabitats). This seems highly unlikely; however, because the particular call type we measured is already stereo- typed among males for maximum transmission through forest habitat and has a sufficiently narrow range of variation between individuals that it would likely be of no benefit for increasing sound trans- mission. Our data also provided evidence that vo- cal characteristics were not linked to a particular site, as one of the test sites occupied both years had a classification rate (0%) akin to random. It therefore appears that variation in the advertise- ment call between males is due to inherent indi- vidual differences, thereby providing a useful tool for monitoring. Recently, some studies have started exploring the technique of vocal individuality to monitor population trends and to learn about site fidelity and territory turnover. For example, Delport et al. (2002) had great success monitoring individual male and female African Wood Owls (Strix woodfor- dii), determining turnover rates over a 12-yr period and thereby providing opportunities to measure as- pects of population turnover along with life history traits. Turnover rates between two years in Com- mon Scops Owls {Otus scops) have also been deter- mined using similar analyses to ours (Galeotti and Sacchi 2001). Although our sample is small, the preliminary analysis of site fidelity leads us to be- lieve that this technique may have utility in moni- toring the Queen Charlotte Islands’ subspecies of Northern Saw-whet Owl. Based on our results, we suggest that there is some level of site fidelity at occupied sites between years, a behavior which al- though occurring at low rates, is largely uncom- mon in the A. a. acadicus subspecies (Cannings 1993). Our preliminary results with this analysis suggested that further study may be warranted to determine the levels of site fidelity in the A. a. brooksi subspecies, given that high site fidelity is of- ten associated with increased reproductive success (Newton 1993) as well as a steady food supply (Lof- gren et al. 1986). As this subspecies is not migra- tory, it is less likely to be as nomadic as the A. a. acadicus counterpart, and hence, we would expect the breeding behavior to potentially differ in re- gard to site fidelity. Because a lack of site fidelity is often associated with reproductive failure in the previous year (Bried and Jouvetin 1999), and high breeding success is often associated with high qual- ity habitats, monitoring of levels of site fidelity may give insight into the habitat structure or type that facilitates increased reproductive success. Acknowledgments Funding for this project came from research grants to K. Otter from the following organizations: Natural Sci- ence and Engineering Research Council of Canada (NSERC), Vancouver Foundation, Canada Foundation of Innovation and the British Columbia Knowledge Devel- opment Fund. As well, we received monetary and in-kind support from the British Columbia Ministry of Water Land and Air Protection and the British Columbia Forest Service. C. Holschuh was funded through an NSERC In- dustrial Postgraduate Scholarship I and a Science Coun- cil of British Columbia Great Scholarship, both in coop- eration with Weyerhaeuser Canada. We are grateful to Alvin Cober, Frank Doyle, Dave Trim, Anne Hethering- ton, and Jared Hobbs for their logistical support. We thank Luke Hyatt for his excellent field assistance. Final- ly, we express thanks to Kathy Parker, Brent Murray, Dick Cannings, Jim Bednarz, David Evans, W.L. Eakle, and an anonymous reviewer for providing valuable comments on earlier drafts of this manuscript. Literature Cited Bibby, C.J., N.D. Burgess, D.A. Hill, and S.H. Mustoe. 2000. Bird census techniques, 2nd Ed. Academic Press, San Diego, CA U.S.A. Bried, J. and P. Jouvetin. 1999. Influence of breeding success on fidelity in long-lived birds: an experimental study. J. Avian Biol. 30:392-398. Cannings, R.J. 1993. Northern Saw-whet Owl. In A. Poole and E. Gill [Eds.], The birds of North America, No. 42. The Birds of North America, Inc., Philadelphia, PA U.S.A. Delport, W., A.C. Kemp, and J.W.H. Ferguson. 2002. Vo- cal identification of individual African Wood Owls Strix tooodfordii: a technique to monitor long-term adult turnover and residency. Ibis 144:30-39. Eakle, W.L., R.W. Mannan, and T.G. Grubb. 1989. Iden- tification of individual breeding Bald Eagles by voice analy.sis. /. Wildl. Manag. 52:450-455. Foster, C.C., E.D. Forsman, E.C. Meslow, G.S. Miller, J.A. Reid, F.F. Wagner, A.B. Carey, and J. Lint. 1992 Survival and reproduction of radio-marked adult Spotted Owls. y. Wildl. Manag. 56:91—95. Galeotti, R, M. Paladin, and G. Pavan. 1993. Individ- ually distinct hooting in male Pygmy Owls Glaucidium June 2005 Monitoring Owls through Vocaijzations 141 passerinum: a multivariate approach. Ornis Scand. 25: 15-20. AND G. Pavan. 1991. Individual recognition of male Tawny Owls {Strix aluco) using spectrograms of their territorial calls. Ethol. Ecol. Evol 3:113-126. AND R. Sacchi. 2001. Turnover of territorial Scops Owls Otus scops as estimated by spectrographic analy- ses of male hoots./. Avian BioL 32:256-262. Gilbert, G., P.K. McGregor, and G. Tyler. 1994. Vocal individuality as a census tool; practical considerations illustrated by a study of two rare species. /. Field Or- nithol 65:335-348. Lofgren, O., B. Hornfeld, and B.G. Carlsson. 1986. Site tenacity and nomadism in Tengmalm’s Owl {Ae- golius funereus) in relation to cyclic food production. Oecologia 69:321-326. McGregor, P.K. and P. Byle. 1992. Individually distinc- tive bittern booms: potential as a census tool. Bioacous- tics 4:93—109. , T.M. Peake, and G. Gilbert. 2000. Communica- tion behavior and conservation. Pages 261-280 in L.M. Gosling and W.J. Sutherland [Eds.], Behaviour and conservation. Cambridge Univ. Press, Cambridge, U.K. Morton, E.S. and K. Young. 1986. A previously unde- scribed method of song matching in a species vdth a single song “type”, the Kentucky Warbler {Oporornis formosus). Ethology 73:334—342. Newton, I. 1993. Age and site fidelity in female sparrow- hawks, Accipiter nisus. Anim. Behav. 46:161-168. Otter, K. 1996. Individual variation in the advertisement call of male Northern Saw-whet Owls. /. Field Ornithol 67:398-405. , L. Ratcliffe, M. Njegovan, and J. Fothering- ham. 2002. Importance of frequency and temporal song matching in Black-capped Chickadees; evidence from interactive playback. Ethology 108:181-191. Peake, T.M., P.K. McGregor, K.W. Smith, G. Tyler, G Gilbert, and R.E. Green. 1998. Individuality in Corn- crake Crex vocalizations. Ibis 140:121-127. and P.K. McGregor. 2001. Corncrake Crex crex census estimates: a conservation application of vocal individuality. An. Biodiv. Conserv. 24:81—90. Reebeck, M., R. Corrick, B. Eaglestone, and C. Stain- ton. 2001. Recognition of individual European Night- jars Caprimulgus europaeus from their song. Ibis 143 468-475. Specht, R. 2000. Avisoft SAS Lab Pro version 3.82b. Avi- soft Bioacoustics, Berlin, Germany. StatSoft, Inc. 2003. STATISTICA version 6. Stat Soft Inc., Tulsa, OK U.S.A. Terry, A.M.R. and P.K. McGregor. 2002. Census and monitoring based on individually identifiable vocali- zations: the role of neural networks. An. Conserv. 5. 103-111. Wiley, R.H. and D.G. Richards. 1982. Adaptations for acoustic communication in birds: sound transmission and signal detection. Academic Press, New York, NY U.S.A. Received 6 April 2004; accepted 27 February 2005 J Raptor Res. 39(2): 142-148 © 2005 The Raptor Research Foundation, Inc. PARTITIONING OF GENETIC (RAPD) VARIABILITY AMONG SEXES AND POPULATIONS OF THE BARN OWL ( TYTO ALBA) IN EUROPE Robert Matics,^ SAndor Varga, and Balazs Offer University of Pecs, Faculty of Sciences, Department of Genetics and Molecular Biology, Ijjusdgu. 6., H-7601 Pecs, Hungary Akos Klein Eotvos Lorand University, Department of Systematic Zoology and Ecology, Behavioural Ecology Group, Pazmany Peter setany 1/C, H-1117, Budapest, Hungary Gyozo Horvath University of Pecs, Faculty of Sciences, Department of Zootaxonomy and Synzoology, Ifjusdg u. 6., H-7601 Pecs, Hungary Alexandre Roulin University of Lausanne, Department of Ecology and Evolution, Biology Building, CH-1015 Lausanne, Switzerland Peter Putnoky and Gyula Hoffmann University of Pecs, Faculty of Sciences, Department of Genetics and Molecular Biology, Ijjusdg u. 6., H-7601 Pecs, Hungary Abstract. — The white Barn Owl subspecies ( Tyto alba alba) is found in southern Europe and the reddish- brown subspecies {T. a. guttata) in northern and eastern Europe. In central Europe, the two subspecies interbreed producing a large range of phenotypic variants. Because of the different ratios of the sub- species in different geographic regions, we predict that genetic variation should be greater in Switzer- land than in Hungary. We tested this hypothesis by measuring genetic variation with the RAFD method. As predicted, the genetic differentiation within a Swiss population of Barn Owls was significantly greater than the variation within a Hungarian population. This suggests that gene flow is greater in central Europe than at the eastern limit of the Barn Owl distribution in Hungary. In both countries genetic variation was more pronounced in females than in males. As in other birds, this is probably because female Barn Owls are less philopatric than males. The number of migrants between Hungary and Switzerland is ca. 1 individual per generation; if calculated separately for the sexes, then 0.525 for males and ca. 1 for females (Nm values). The difference in the number of migrants between genders again is likely a consequence of higher male philopatry. The sexual differentiation is greater in the Swiss population than in the Hungarian and the genetic substructuring of the populations of the species is substantial. The reason for the considerable population substructuring could be the nonmigratory be- havior and socially monogamous pairing of the species, as well as the geographical barriers (Alps) between the populations examined. Key Words: Barn Owl, Tyto alba; genetic variability; introgressiow, RAPD; subspecies. VARIABILIDAD GENETICA ESTIMADA MEDIANTE RAFD ENTRE SEXOS Y FOBLACIONES DE TYTO ALBA EN EUROFA Resumen. — La subespecie Tyto alba alba se encuentra en el sur de Europa, y T. a. guttata en el norte y oriente de este continente. En el centro de Europa, las dos subespecies se entrecruzan y producen una amplia gama de variantes fenotipicas. Debido a la variacion geografica en la proporcion de individuos pertenecientes a las distintas subespecies, predijimos que la variabilidad genetica deberia ser mayor en Suiza que en Hungria. Fusimos a prueba esta hipotesis midiendo la diferenciacion genetica usando el ^ Email address: bobmatix@freemail.hu 142 June 2005 Barn Owi. Genetic Variability 143 metodo de ADN polimorfico amplificado aleatoriamente (RAPD, por sus siglas en ingles). Como lo habiamos predicho, la diferenciacion genetica dentro de una poblacion suiza de T. alba fue significati- vamente mayor que la variacion observada dentro de una poblacion bungara. Esto sugiere que el flujo genetico es mayor en el centro de Europa que en el limite oriental de la distribucion de T. alba en Hungria. En ambos paises, la variacion genetica fue mas pronunciada en las bembras que en los macbos. Como en otras aves, esto probablemente se debe a que las bembras son menos filopatricas que los macbos. El numero de migrantes entre Hungria y Suiza es cercano a un individuo por generacion, y los valores de Nm calculados por separado para machos y bembras son de 0.525 y aproximadamente 1, respectivamente. Es probable que la diferencia en el numero de migrantes se deba nuevamente a la mayor filopatria de los machos. La diferenciacion sexual es mas pronunciada en la poblacion suiza que en la bungara, y estas poblaciones estan sustancialmente subestructuradas geneticamente. Posibles ra- zones para explicar la considerable subestructuracion podrian ser el comportamiento no migratorio y el apareamiento monogamo de la especie, ademas de la separacion de las poblaciones estudiadas por barreras geograficas, como los Alpes. [Traduccion del equipo editorial] In a number of birds, color polymorphism has evolved in allopatry. Under this scenario, a geo- graphical barrier has separated a population in two, a process that facilitates genetic differentia- tion as gene flow is physically reduced. In some cases, during the time of allopatric separation, sto- chastic processes or natural selection promoted the evolution of alternative color morphs for which the expression is under genetic control and not sensitive to the environment. After the geographi- cal barrier was removed, the two subpopulations could mix in a zone of secondary contact. If hy- brids are viable, color polymorphism allows re- searchers to trace back the origin of individuals. For example, in the Barn Owl {Tyto alba) the white subspecies, T. a. alba, is found mainly in southern Europe, whereas the reddish-brown subspecies, T. a. guttata, occupies northern and eastern Europe. In central Europe, the two subspecies seemingly pair randomly with respect to coloration (Baudvin 1975, Roulin 1999, Matics et al. 2002), which im- plies that hybrids may not be selected against. Following Voous (1950), these two subspecies may have evolved because during the last ice age, Barn Owls subsisted in two refugia located in southwestern Europe ( T. a. alba) and southeastern Europe {T. a. guttata). After the ice retreated, the two subspecies invaded Europe via Spain and France {T. a. alba) and the Balkans (T. a. guttata) to meet in a zone of secondary contact in central Europe. The existence of a color polymorphism is consistent with the hypothesis that Barn Owl pop- ulations located in central Europe involve two sub- species. From a genetic point of view, we expect that genetic variation should be more pronounced in the zone of hybridization in central Europe than to the east or west. The various extant populations show differential degrees of introgression suggest- ed by the phenotypic distribution of individuals (Matics et al. 2002, Roulin et al. 2001). In this case, one would expect disequilibria of neutral markers but this was not tested with our methodology. The transition zone appears to be very wide in compar- ison with other bird species, ranging from west- central France to eastern Hungary and northeast- ern Poland. On the eastern side of the zone, the reddish-brown form is more frequent comprising 84—92% of the individuals in Hungary (Matics and Hoffmann 2002) and 90% in southeastern Ger- many (Schonfeld 1974), whereas the white subspe- cies prevails on the western side of the zone (Glutz von Blotzheim and Bauer 1980; e.g., 75% in cen- tral France [Baudvin 1975] and 50% in Switzerland [Roulin et al. 2001]). The closer the distribution of the subspecies is to the 1:1 ratio in a population the greater variability of the population due to the differences in genetic constitution of the subspe- cies (Harrison 1993, Arnold 1997). In addition, be- cause the species is socially monogamous and non- migratory, we predict a relatively high genetic substructuring of the populations. Because in this species, as in other birds (Greenwood 1980), males are more philopatric than females, we expect that in both Swiss and Hungarian populations females are genetically more diverse than males as shown in the Black-billed Magpies {Pica hudsonia; Wang and Trost 2001) . We tested these two predictions by quantifying genetic variation using random amplified polymor- phic DNA (RAPD) technique. This technique is based on the amplification of unknown DNA se- quences using single, short, random oligonucleo- 144 MAtics et al. VoL. 39, No. 2 Czech Republik Germany Slovakia Hungary Austria /Slownia Croatia Figure 1. Location of the Barn Owl populations examined for genetic variability. The BIOTA© Association (Pecs, Hungary) provided the base map for this figure. tide primers. RAPDs provide an unbiased sample of DNA variation along the entire genome (Hwang et al. 2001) including non-nuclear genomes. The sensitivity of RAPDs to population divergence may be derived from rapid evolution of non-coding, re- petitive DNA sequences detected by RAPDs (Plom- ion et al. 1995). Therefore, the RAPD technique is able to detect variation within and among wild populations (Haig et al. 1994, Horn et al. 1996) and among sexes within a population (Wang and Trost 2001). In birds, mainly species with insular distribution (Zwar^es 1999) or living in fragment- ed habitats (Bouzat 2001) have been analyzed. However, this technique has also been used for other purposes, including sex determination (Park et al. 1997), analysis of wild versus captively-reared populations (Bagliacca et al. 1997), and detection and eradication of hybrids (Negro et al. 2001). In- formation on the amount of genetic variation with- in a species and its distribution within and between populations aids conservation planning as well (Hwang et al. 2001). Study Area The study areas were located in Hungary (between 47°02'N, 17^33'E and 45“50'N, 18°29'E) and Switzerland (between 46°56'N, 7°03'E and 46°44'N, 6°41'E; Fig. 1). The Swiss study area is located in the middle of the Eu- ropean distribution of Barn Owls and has an elevation between 400-650 m above sea level (masl); the Hungar- ian study area is near the distribution limit of the species with an elevation between 50-500 masl. Methods Sample Collecrtioiis. Blood samples from breeding Barn Owls were collected between May and August 1998. First, the skin was cleaned with ethanol over the point where the brachial vein crossed the elbow and insulin needles (diameter = 0.4 mm) were used to collect 100- |xL blood. Samples were stored on ice in sterile 1.5 mL Eppendorf tubes until they were carried to the laboratory and frozen at — 20°C. The birds were held for at least 5 min after blood collection and then were released. June 2005 Barn Owt^ Genetic Variability 145 Table 1, Values of Nei-Li genetic distance among individuals within the groups given. Mean, SE, and Confidence Intervals (CT) are estimated by jackknife procedure (H = Hungary, CH = Switzerland). Mean ± SE 95% Cl 99% Cl N H females 0.358 ± 0.006 0.346-0.370 0.342-0.374 13 H males 0.261 ± 0.010 0.241-0.280 0.235-0.287 7 CH females 0.501 ± 0.021 0.460-0.541 0.447-0.554 17 CH males 0.368 ± 0.010 0.349-0.387 0.342-0.394 15 H pooled 0.338 ± 0.004 0.331-0.344 0.329-0.347 20 CH pooled 0.471 ± 0.008 0.455-0.487 0.451-0.492 32 DNA Extraction. Standard phenol-chlorophorm-isoa- myl alcohol methods (Sambrook et al. 1989) were used to isolate total DNA from blood samples. Fifty p,L-blood samples were suspended in 200-(xL PBS buffer and cells were sedimented. The resulting pellet was suspended in extraction buffer containing 20 [xg/mL RNAse and was incubated at 37°C for 1 hr. Proteinase K treatment was then applied (to a final concentration of 100 pg/mL) and incubated at 50°C for 3 hr. Samples were extracted three times with equal volumes of phenol followed by ethanol precipitation. DNA was washed with 70% ethanol and resuspended in 100-200 [xL sterile-distilled water. The concentration of the extracted samples was quanti- fied using the photometric device GeneQuant (Pharma- cia, Cambridge, U.K.). RAPD PCR Procedure. To find the optimal concentra- tion, where the reactions gave the most consistent and clearest products, DNA was diluted in the range of 10- 80 ng/ fxL (by doubling dilutions) . In standard experi- ments 20-ng DNA in 25 |xL reaction volume was used in the presence of 1.5 mM MgCl 2 . Amplifications were car- ried out using PTC-150 Minicycler (MJ Research, San Francisco, CA U.S.A.) with the following program: after a first cycle (2.5 min at 94°C, 1 min at 35°C, and 2 min at 72°C), additional 35 cycles were done (40 sec at 94°C, 40 sec at 35°C, and 1 min at 72°C) . The products were run on 2% agarose gels. Gels were photographed using the digital-gel documentation system BioDocIt (UVP, Cambridge, U.K.). Twenty-one different primers were tested (QUIAGEN, Budapest, Hungary) and those giving the most variable patterns were used for further analyses (OPW-17, OPW-08, OPT-20, OPN-05, OPO-05, OPH-14, and OPJ-12). Scoring and Data Analysis. Fragments were visualized by adding 0.1 |xg/mL ethidium bromide to the agarose gels. Gel photographs were scored for the presence or absence of RAPD bands. A pair-wise matrix of the genetic distances between individuals was obtained using a eu- clidean distance measure (Huff et al. 1993), calculated from presence or absence data using RAPDistance (Armstrong et al. 1994). Components of variance parti- tioned into within- and between-populations were esti- mated from this matrix using AMOVA program version 1.55 (Analysis of Molecular Variance; Excoffier et al. 1992). The number of permutations for significance test- ing was set at 1000. Where Ost values differed signifi- cantly from zero, the number of migrants per generation was calculated using Nm = 0.25 (st'^‘ 1) (Wright 1951). AMOVA variance components were used as estimates of the genetic diversity within and between populations. Be- cause the genetic distance values were not independent data points, we calculated Nei-Li (Nei and Li 1979) ge- netic distances as well, followed by jackknife analysis (Shao and Tu 1995), to find differences between popu- lations and genders. The RAPD technique has some limitations. The most important might be that the banding pattern produced represents nuclear (both sex-linked and autosomal) and mitochondrial genomes. The fact that mitochondrial genes are exclusively maternally inherited, whereas au- tosomal ones are biparental markers, may obscure inter- pretation of results, particularly in connection with sex- ual differentiation and consequences of sex-biased dispersal (Goudet et al. 2002, Prugnolle and de Meeus 2002, Vitalis 2002). Results The amplifications with seven primers produced 106 reproducible bands in both populations (^ = 15.1 bands per primer). In the Hungarian popu- lation, 80 of these bands were present and five of them were invariant, whereas in the Swiss samples, 104 were present and no invariant band was found. There were six singletons (bands with one inci- dent; i.e., occurring only in one individual). The 99% confidence intervals of the two sexes did not overlap. In both populations the jackknife estimates of mean Nei-Li genetic distances of fe- males were greater than those of males (Table 1). The Swiss population showed higher Jackknife es- timate of mean genetic distance than the Hungar- ian population because there was no overlap in the 99% confidence intervals (Table 1). Because the probability that random distance (Ost) was greater than the observed distance was less than 0.05 in all cases, the ^sT"Values were sig- nificantly different from zero. The AMOVA indi- cates 7% of the variance is partitioned between the genders within the pooled sample. The Hungarian sample gave similar results (7.2%), whereas the Swiss population was of higher value (14%; Table 2). The among-coun tries differentiation reaches 146 Matics et al. VoL. 39, No. 2 Table 2. Results of the analysis of molecular variance with a between-gender and among-countries arrangement. df SS MS Variance Component (Fst Hungary Between genders 1 21.15 21.15 00.966 (7.2%) 0.07 <0.001 Among individuals/within genders 18 222.55 12.36 12.364 (92.8%) Switzerland Between genders 1 57.91 57.91 2.637 (14%) 0.14 <0.001 Among individuals/within genders 30 476.28 15.88 15.876 (86%) Pooled (all females vs. all males) Between gender 1 49.30 49.30 1.278 (7%) 0.07 <0.001 Among individuals/within gender 50 843.50 16.87 6.870 (93%) Pooled (all Hungarian vs. all Swiss) Among countries 1 114.92 114.92 4.04 (21%) 0.21 <0.001 Among individuals/within countries 50 777.89 15.56 15.56 (79%) ^ Nonparametric randomization test with 1000 permutations. 21% (Table 2) and the estimated number of mi- grant individuals between the Swiss and Hungarian populations is 0.96 per generation. The pair-wise OsT values gave the same results as the between- sexes arrangement (Table 3). Moreover, the among-males (0.32) and the among-females values (0.20) suggest a higher genetic difference of males between populations. From these data we estimat- ed that 0.53 males and 1.01 females migrate be- tween populations each generation (Nm values) . Discussion Genetic differentiation among breeding females was greater than among breeding males in both Hungary and Switzerland. The differential dispers- al distances of the sexes, males being more philo- patric than females, may explain this differentia- tion. Therefore, males are genetically more similar Table 3. Pairwise genetic distances of the populations (4 >st between pairs of populations) . Above diagonal is the probability that random distance (4>st) > observed dis- tance; (number of iterations = 1000; H = Hungary, CH = Switzerland). CH Male CH Female H Mai.e H Female CH male — 0 0 0 CH female 0.14 — 0 0 H male 0.32 0.29 — 0 H female 0.25 0.20 0.07 — to each other within a population. This scenario, based on sex-specific dispersal, may also explain the greater among-males differentiation between populations (about twice as many females as males appear to emigrate). Because males do not emi- grate as much as females, they preserve the genetic features of their population to a greater degree. This is further supported by the fact that the esti- mate of migrants per generation (based on genetic data) gave the same result as the ringing data: greater male philopatry (Taylor 1994). Previous studies suggested that, from Hungary, a greater percentage of individuals move toward central Eu- rope than in the opposite direction (Matics 2003); i.e., the Hungarian population was more of a “source” than a “sink” population. As a conse- quence, gender- and population-differentiation is greater in Switzerland than in Hungary, because the exchange of individuals is guttata- and female- biased. These results were concordant with the fact that both phenotypic and genotypic variability of individuals were greater in the middle of a transi- tion zone than on the edges (Arnold 1997, Roulin 2003). This proposal could be tested with data from other localities such as from western Euro- pean Barn Owl populations. The between-gender differentiation of the spe- cies detected using RAPDs (7%, 14%) seem to be disproportionally high. Using the random priming technique the sexual differentiation was expected to be between 1-2%, as the Barn Owl has 92 chro- June 2005 Barn Owi. Genetic Variability 147 mosomes (Belterman and De Boer 1984). This higher value of gender differentiation could be due to the relatively large size of sexual chromo- somes in birds (Stevens 1991). The sexual differ- entiation of autosomal markers caused by sex- biased dispersal is instantaneous because the next generation receives a random set of alleles from both parents. On the other hand, sex-chromosom- al markers preserve the differentiation for longer time and when gene flow occurs continuously, this differentiation could be detected permanently. Dif- ferentiation detected in this study therefore may be associated with markers located on sex-chro- mosomes. Although only two populations were analyzed, the value of 0.21 indicated a substantial genet- ic substructuring among our study populations. The results of another RAPD study gave values of 0.048 and 0.103 for island species (two popula- tions of both the Puerto Rican Vireo [ Vireo latimeri] and Jamaican Vireo [V. modestus], respectively), and 0.015 for a migratory continental species (three populations of the White-eyed Vireo [V. gri- seus\\ Zwartjes 1999). For the Greater Rhea {Rhea americana) a value of 0.0637 was found among four wild and a captive population (Bouzat 2001), which is a low value given that this species is flight- less. The special mating system of the Greater Rhea could play an important role in producing this low genetic substructuring. The male rhea establishes a territory and builds a nest. He will then attract a group of about three to six females with whom he mates and they lay ca. 20-30 eggs in his nest. While the females go off to mate with other males, the male will incubate the eggs and look after the chicks on his own. The relatively strong substruc- turing of the Barn Owl populations could be ex- plained by at least three factors: (1) nonmigratory behaviour, (2) the socially monogamous mating system of the species, and (3) the presence of a geographic barrier (Alps) between the populations analyzed. The Nm value of roughly one is the min- imum amount of gene flow that prevents differ- entiation at neutral loci among populations by ge- netic drift (Wright 1931). The conversion of Fs^-related values into Nm is problematic for several reasons, including that it is based on isolation-by— distance models. In this study, we could not test for correlations between genetic and geographic distance. Furthermore, population size and dispersal rates are not constant over time and space and assumptions of demo- graphic and genetic equilibrium and uniformity are unrealistic (Whitlock and McCauley 1999). Many assumptions of the models used are violated, so that results cannot be interpreted as direct mea- sures. In addition, other evolutionary forces con- tribute in establishing differentiation (Bossart and Prowell 1998). Finally, we suggest that drift should play an important role in the microevolution of the Barn Owl as this species of tropical origin (Voous 1988) goes frequently through bottlenecks in the suboptimal European area in hard winters (Taylor 1994). Acknowledgments Tamara Ilonczai helped in data processing. E. Pearls- tine, JJ. Purger, S. Talbot, and X. (Lucy) Wang provided comments to earlier versions of the manuscript. The re- search was supported by the Hungarian Scientific Re- search Fund Grants No. T025822 and T038377. A. Rouhn was supported by the Swiss Science Foundation (grant No. PPOOA-102913). Literature Cited Armstrong, J., A. Gibbs, R. Peakall, and G.F. Weiller. 1994. RAPDistance: random amplihed polymorphic DNA analysis, http://life.anu.edu.au/molecular/ software/ rapid.html. Arnold, M.L. 1997. Natural hybridisation and evolution Oxford Univ. Press, Oxford, U.K. Bagliacca, M., a. Valentine and I. Cappuccio. 1997. RAPD molecular markers to study populations of Black Grouse Tetrao tetrix in the Alps. Wildl. Biol. 3; 269. Baudvin, H. 1975. Biologic de reproduction de la chouet- te effraie {Tyto alba) en Cote d’Or: premiers resultats. Jean. Blanc 14:1—51. Belterman, R.H.R. and L.E.M. De Boer. 1984. A karyo- logical study of 55 species of birds, including karyo- types of 39 species new to cytology. Genetica 65:39—82. Bossart, J.L. and D.P. Prowell. 1998. Genetic estimates of structure and gene flow: limitations, lessons, and new directions. Trends EcoL Evol. 13:202-206. Bouzat, J.L. 2001. The population genetic structure of the Greater Rhea {Rhea americana) in an agricultural landscape. Biol. Conserv. 99:274—284. Excoffier, L., re. Smouse, and J.M. Quattro. 1992 Analysis of molecular variance inferred from metric distances among DNA haplogroups: applications to human mitochondrial DNA restriction data. Genetics 131:479-491. Glutz von Blotzheim, U.N. and K.M. Bauer. 1980 Handbuch der Vogel Mitteleuropas. Vol. 9. Akadem- ische Verlagsgesellschaft, Wiesbaden, Germany. Goudet, J., N. Perrin, and P. Waer. 2002. Test for sex- biased dispersal using b-parentally inherited genetic markers. Mol. Ecol. 11:1103-1114. 148 Matics et al. VoL. 39, No. 2 Greenwood, PJ- 1980. Mating systems, philopatry, and dispersal in birds and mammals. Anim. Behav. 28: 1140-1162. Haig, S.M., J.M. Rhymer, and D.G. Heckel. 1994. Popu- lation differentiation in randomly amplified polymor- phic DNA of Red-cockaded Woodpeckers Picoides bo- realis. Mol. Ecol. 3:581—595. Harrison, R.G. (Ed.). 1993. Hybrid zones and the evo- lutionary process. Oxford Univ. Press, New York, NY U.S.A. Horn, P.L., J.A. Rafai.ski, and P.J. Whitehead. 1996. Mo- lecular genetic (RAPD) analysis of breeding Magpie Geese. Auk 113:552-557. Huff, D.R., R. Peakall, and RE. Smouse. 1993. RAPD variation within and among natural populations of out crossing buffalograss (Buchloe dactyloides Nutt. En- gelm.). Theor. Appl. Genet. 86:927-934. Hwang, S.Y., H.W Lin, Y.-S. Kuo, and T.-P. Lin. 2001. RAPD variation in relation to population differentia- tion of Chamaecyparis formosensis and Chamaecyparis tai- wanensis. Bot. Bull. Acad. Sin. (Taipei) 42:173-179. Matics, R. 2003. Direction of movements in Hungarian Barn Owls {Tyto alba): gene flow and barriers. Diver. Distr. 9:261- -268. and G. Hoffmann. 2002. Location of the transi- tion zone of the Barn Owl subspecies Tyto alba alba and Tyto alba guttata {Stringiformes: Tytonidae). Acta Zool. Cracov. 45:242-250. , , T. Nagy, and A. Roulin. 2002. Random pairing with respect to plumage coloration in Hun- garian Barn Owls./. Ornithol. 143:493-495. Negro, J.J., M.J. Torres, and J.A. Godoy. 2001. RAPD analysis for detection and eradication of hybrid par- tridges {Alectoris rufa X A. graeca) in Spain. Biol. Con- serv. 98:19-24. Nei, M. and W. Li. 1979. Mathematical model for study- ing genetic variance in terms of restriction endonu- cleases. Proc. Natl. Acad. Sci. U.S.A. 76:5269-5273. Park, R.L., S.S. Xiong, J.O. Richards, J.C. Shaw, R.P. Evans, and R.N. Thwaits. 1997. Identification of RAPD markers linked to the W-chromosome in the Ostrich using segregant bulks. Scherago International Inc., San Diego, CA U.S.A. Plomion, C., N. Bahrman, C.E. Durel, and D.M. O’Malley. 1995. Genomic mapping in Pinus pinaster (Maritime pine) using RAPD and protein markers. Heredity 74:661-668. Prugnolle, F. and T. de Meeus. 2002. Inferring sex-bi- ased dispersal from population genetic tools: a review. Heredity 88:161-165. Roulin, A. 1999. Nonrandom pairing by male Barn Owls ( Tyto alba) with respect to a female plumage trait. Be- hav. Ecol. 10:688-695. . 2003. Geographic variation in sexually selected traits: a role for direct selection or genetic correla- tion? J. Avian Biol. 34:251-258. , C. Dijkstra, C. Riols, and A.L. Ducrest. 2001 Female- and male-specific signals of quality in the Barn Owl./. Evol. Biol. 14:255-266. Sambrook, J., E.F Fritsch, and T. Maniatis. 1989. Mo- lecular cloning: a laboratory manual. Cold Spring Harbor Laboratory Press, Cold Springs, NY U.S.A. Schonfeld, M. 1974. Ringfundauswertung der 1964—72 in der DDR beringter Schleiereulen Tyto alba guttata Brehm. Jahresber. Vogelwar. Hidden. 4:90-122- Shao, j. and D. Tu. 1995. The jackknife and bootstrap. Springer-Verlag, New York, NY U.S.A. Stevens, L. 1991. Genetics and evolution of the domestic fowl. Cambridge Univ. Press, Cambridge, U.K. Taylor, I. 1994. Barn Owls. Cambridge Univ. Press, Cam- bridge, U.K. Vitalis, R. 2002. Sex-specific genetic differentiation and coalescence times: estimating sex-biased dispersal rates. Mol. Ecol. 11:125-138. Voous, K.M. 1950. On the distributional and genetical origin of the intermediate populations of the Barn Owl in Europe. Verlagsgesellschaft Geest Portig, Leip- zig, Germany. . 1988. Owls of the northern hemisphere. W. Col- lins Sons & Co. Ltd., London, U.K. Wang, X.H. and C.H. Trost. 2001. Dispersal pattern of Black-billed Magpies {Pica hudsonia) measured by mo- lecular genetic (RAPD) analysis. Auk 118:137-146. Whitlock, M.C. and D.E. McCauley. 1999. Indirect mea- sures of gene flow and migration: Fst7^1/(4Nm + l). Heredity 82:117-125. Wright, S. 1931. Evolution in Mendelian populations. Genetics 16:97-159. . 1951. The genetical structure of populations. An- nals. Eugenics 15:323-354, Zwartjes, P.W. 1999: Genetic variability in the endemic vireos of Puerto Rico and Jamaica contrasted with the continental White-eyed Vireo. Auk 116:964—975. Received: 12 November 2003; accepted 16 March 2005 J. Raptor Res. 39(2) :149-155 © 2005 The Raptor Research Foundation, Inc. BREEDING BIOLOGY AND FOOD HABITS OF THE MADAGASCAR KESTREL {FALCO NEWTONl) IN NORTHEASTERN MADAGASCAR Lily-Arison Rene de Roland, Jeanneney Rabearivony, Hariiajajna Robenarimangason, and Gilbert Razafimanjato The Peregrine Fund’s Project in Madagascar, B.P 4113, Antananarivo (101), Madagascar Russell Thorstrom^ The Peregrine Fund, 3668 West F'lying Hawk Lane, Boise, ID 83709 U.S.A. Abstract. — ^We studied Madagascar Kestrels (Falco newtoni) on Masoala Peninsula, northeastern Mada- gascar during the 1997 and 1998 breeding seasons. We located five nest sites and observed eight nesting attempts during the two breeding seasons. All nests were in tree cavities and averaged 13.8 ± 2.0 m (SE) above the ground in trees averaging 22.8 ± 0.8 m (SE) in height {N = 5 nests) . Egg laying took place from mid-September to the first week of October. The modal clutch size was 4 ± 0.9 (N = 6 nests, range 3-5 eggs). The incubation period averaged 28 d, varying from 27-29 d (A^ = 5 nests). Hatching occurred from the middle of October to the first week of November with young fledging in late No- vember. Of 24 eggs laid in six nests, 13 (54%) hatched, and seven (54%) of those hatchlings fledged; thus, a total of 1.2 young fledged per breeding attempt were produced and overall nest success was 50%. The Madagascar Kestrel diet of 338 identified prey was composed of 93.8% lizards (N = 317), 2.6% insects (N = 9), 2.4% amphibians (A^ = 8), and 1.2% birds (A" = 4). Key Words: Madagascar Kestrel, Falco newtoni; nesting biology, food habits', nests-, productivity. BIOLOGIA REPRODUCTIVA Y HABITOS ALIMENTICIOS DE FALCO NEWTONI EN EL NORESTE DE MADAGASCAR Resumen. — Estudiamos individuos de la especie Falco newtoni en la peninsula Masoala en el noreste de Madagascar durante las estaciones reproductivas de 1997 y 1998. Localizamos cinco nidos y observamos ocho intentos de nidificacion durante los dos periodos reproductivos. Todos los nidos se encontraron en cavidades en arboles a una altura promedio de 13.8 ± 2.0 m (EE) sobre el suelo y en arboles con una altura promedio de 22.8 ± 0.8 m (EE) {N = 5 nidos) . La puesta de huevos ocurrio desde mediados de septiembre hasta la primera semana de octubre. El tamano modal de la nidada fue de 4 ± 0.9 huevos {N = 6 nidos, rango 3-5 huevos) . El periodo promedio de incubacion fue de 28 dias, variando entre 27-29 dias (N = 5 nidos) . La eclosion ocurrio desde mediados de octubre hasta la primera semana de noviembre y el abandono del nido por parte de los polluelos ocurrio a fines de noviembre. De los 24 huevos puestos en seis nidos, 13 (54%) eclosionaron, y siete (54%) de los polluelos que eclosionaron abandonaron el nido. Por lo tanto, se produjeron un total de 1.2 juveniles que abandonaron el nido por intento de nidificacion, y el exito de nidificacion general fue del 50%. La dieta de F. newtoni, basada en 338 presas identificadas, estuvo compuesta en un 93.8% por lagartijas (N = 317), en un 2.6% por insectos (N = 9), en un 2.4% por anfibios {N = 8) y en un 1.2% por aves (N = 4). [Traduccion del equipo editorial] Madagascar has three resident species of falcons: Madagascar Kestrel (Falco newtoni newtoni) , Banded Kestrel (F. zoniventris) , and Peregrine Falcon (F. pe- regrinus radamo), and two wintering species, Eleo- nora’s Falcon (F. eleonorae) and Sooty Falcon (F. concolor). Although the Madagascar Kestrel is the ^ Email address: rthorstrom@peregrinefund.org most common raptor in Madagascar (Siegfried and Frost 1970), detailed information on its biology and natural history are lacking (Langrand and Meyburg 1984, Langrand 1990). This species is widely distributed throughout Madagascar and is found in open grasslands, savannah habitat, de- graded forests, and in the vicinity of villages and towns. This falcon has two distinctive color morphs: pale and rufous (Siegfried and Frost 1970, 149 150 Rene de Roland et al. VoL. 39, No. 2 Cade 1982, Langrand 1990). In this paper, we re- port basic information on the breeding biology and food habits of the Madagascar Kestrel. Study Area and Methods This study was conducted within the secondary habitat and cultivated terrain surrounding Ambanizana Village (49°57'E, 15°37'S), situated at the western boundary of the Masoala National Park (MNP; Robenarimangason 1999). Most of the peripheral zones of the MNP, are com- posed of degraded forest, secondary forest, and fallow land intermixed with cultivated areas (crops include cloves, coffee, vanilla, and bananas) . The MNP itself con- sists of 230 000 ha of primary forest with a typical canopy height of 25 m and with some emergent trees exceeding 30 m (Guillaumet 1984). The forested terrain is moun- tainous, and lacks roads and a trail system. The altitude in the park ranges from sea level to 1230 m (Nicoll and Langrand 1989). The current study was undertaken in the lower altitude and cultivated areas ranging from sea level to 200 m above sea level. The annual mean rainfall at Andranobe Field Station, 7 km south of the study site, IS 6049 mm (Thorstrom et al. 1997). The dry season is from October to mid-December (Rene de Roland 2000) . The climate is mild and the annual mean temperature varies from 18-31°C. The study period was from September 1997^anuary 1998 and September 1998-January 1999, which coincides with the kestrel’s breeding season. During September and October, we searched for nesting pairs by following vocalizing and flying adults. Observations were made at least 50 m from nests, and five and three nests were mon- itored during the 1997 and 1998 breeding seasons, re- spectively. Nests were observed from courtship up to post-fledg- ling period, and 521 hr 22 min of observation were com- pleted. We recorded copulation frequency (per hr and d) and duration (sec), clutch size (number), incubation period (d), nest attendance (percent of observation time), dispersal of young (d), and productivity. Egg di- mensions (length and breadth), and egg and nestling mass were measured using vernier calipers (with 0.01 mm of precision) and Pesola spring-balance scales (0.1 g pre- cision), respectively. Daily nest observations were made from 0600-1800 H with lOX binoculars and a zoom spot- ting scope. While making observations at nests prey items were identified during prey deliveries. Kestrels were trapped with a bal-chatri, a noose carpet fixed over the nest entrance, or a “wire hoop trap” (Ber- ger and Mueller 1959, Thorstrom 1996). Morphometric measurements taken were: wing chord (mm), tail and tarsi length (mm), and body mass (g). One breeding pair of adult kestrels was radiotagged to estimate their rang- ing area. The kestrels were radio-tracked from October- December 1998 by homing on foot, and locational fixes were recorded using an Eagle Explorer Global Position System (GPS; Eagle Electronics, Catoosa, OKU.S.A.) with 30-m of accuracy. The home range was estimated by the minimum convex polygon (MCP) of the locations using Ranges IV software (Kenward 1990). Several nest site pa- rameters (i.e., nest height, nest tree height, nest dimen- sions, nest depth, and internest distance) were measured after the young had fledged. Results One pair of Madagascar Kestrels was trapped and marked during this study. The adult female measurements were wing chord, 200 mm; tail length, 129 mm; tarsus length, 38.2 mm; and body mass, 122 g. The adult male measurements were wing chord, 195 mm; tail length, 110.5 mm; tarsus length, 37.0 mm; and body mass, 110 g. Five nesting attempts were observed during the first season, 1997—98, and three were documented during the second field session, 1998-99. Of the five nesting pairs observed during the two breed- ing seasons, three pairs were composed of a pale- morph male and a rufous-morph female, one pair was of a rufous-morph male paired with a pale- morph female, and one pair included both rufous- morph individuals. Nest Characteristics. During this study, all nests observed for Madagascar Kestrels were placed in natural-tree cavities, with a decayed wood substrate, in secondary and human-modified habitat. The trees identified for nesting were mandrorofo ( Tra- chylobium rerrucosum) , hintsia (Afaelia bijuga) , lalona (Weinmania sp.), and dead unidentified snags. Madagascar Kestrel nests averaged 13.8 ± 4.5 m {N = 5) above the ground in trees averaging 22.8 ± 1.9 m {N = 5) in height. Nest cavities were oval- shaped and measured 61 ± 55.1 cm (N = 2) by 33.5 ± 16.2 cm (N = 2) with an interior depth of 22.5 ± 3.5 cm (N = 2). The distance between two neighboring nests averaged 675 ± 386.2 m {N = 4, range 300-1200 m). Nesting Biology. Courtship activity involved vo- calizations, nest site visits with food deliveries to the female, and copulations. The courtship period was marked by flights (e.g., in open areas and over trees) and accompanied by moderate fluttering of wings. During periods of inactivity, the kestrel pair perched together in the top of dead branches of trees. Courtship behavior, either flights or periods of inactivity, was associated with loud calls "‘itsi, kit- si, kitsi, kitsi."' During this period, the male’s pri- mary role seemed to be showing the female poten- tial nesting sites. Copulations usually occurred after prey deliver- ies from the male to the female. Copulations oc- curred on the top of dead branches. During cop- ulations, the male emitted a “iiitsi, kitsi, kitsi, kitsi. . .” that continued to the end of this activity. June 2005 Breeding Biology of the Madagasgar Kestrel 151 Table 1. Number of prey items delivered by male and female Madagascar Kestrels at Masoala Peninsula, Madagascar, during the breeding seasons of 1997 and 1998. Sex Courtship Incubation Nestling Post-fledging Total Male 13 110 1.30 18 271 Female 0 0 28 39 67 Total 13 110 158 57 338 The highest frequency of copulations during a 1 hr period occurred between 0600-0700 H (N ^ 13 of 91 observed copulations). Copulations were also frequent during the 0800-1000 H (A^ = 28) and 1300-1500 H {N = 28). Ten days prior to the onset of incubation, the frequency of copulations aver- aged 9.1 ± 0.4 (SE) times per day (N = 91 copu- lations, range 7-11). Copulations averaged 4.9 ± 1.6 sec in duration {N = 44, range 3-8 sec). Three copulation attempts were also observed during the late incubation and nestling period. Egg laying took place from mid-September to the first week of October. The earliest egg was laid on 18 September 1997 and the latest was on 5 Oc- tober 1997. The mean clutch size was 4.0 ± 0.9 (N = 6 nests, range 3-5 eggs). Eggs measured 33.9 mm ± 0.9 mm X 28 mm ± 0.6 mm and their mass was 14.3 ± 0.6 g {N = 5 eggs) during the first week of incubation. Eggs were pale white with dark brown spots. Constant incubation seemed to be initiated fol- lowing the laying of the second egg. The incuba- tion period averaged 28 d and varied from 27-29 d (A^ = 5 nests: 27 d [A^ = 3], 29 d [A^ = 2]). Incubation was done by females only, and they in- cubated for 88.5% of the observation time and were absent for 11.5% of the time {N — 153.1 hr). The male’s responsibility during this period was providing food to the incubating female. Prey ex- changes occurred on branches of trees neighbor- ing the nest tree. Nestling and Fledgling Period. Up to 6 d of age, the young were constantly brooded by the female (96.5% of the observation time [A^ = 20.7 hr]). Brooding progressively decreased to 38.7% just pri- or to the time of fledging. Young were fed solely by the female until about 15 d of age when they were able to feed themselves, were active in the nest, and completely covered with contour feath- ers. After about 15 d of age, prey delivered by the adults was dropped into the nest cavity from the entrance. Both the male and female provisioned the young with food during the nestling period (Table 1 ) . During this period, the adult female was very aggressive toward other cavity-nesting birds (e.g., Madagascar Starlings [Hartlaubius aurata] and Broad-billed Rollers [Eurystomus glaucurus]), and occasionally attacked them. Nestlings were observed wing exercising at the cavity entrance at 3 wk of age. Two to three days before fledging, two males weighed 112 g and 118 g, and one female was 128 g. Young fledged at 23- 24 d of age (N = 7 young). After fledging, the young were observed perching together in trees close to the nest tree. During the second week after fledging, they were observed catching insects and attacking prey, and the prey delivery rate by adults decreased progressively. Young dispersed from their natal areas at 44—45 d of age ( A^ = 7 young) . Reproductive Success. In six fully-documented nests containing 24 eggs, 13 (54%) eggs hatched and 7 (54%) of the hatched young fledged. In to- tal, seven young fledged from three successful breeding attempts, for a productivity of 2.3 young fledged per successful nest. Overall productivity was 1.2 (7/6 pairs) young per nesting pair. Nest success for the 2 yr of this study was 50% (3/6 pairs; Table 2). In the six nesting attempts during the two breed- ing seasons, one nest failure occurred during in- cubation and two during the nestling period. In 1997, we believe one nest failed during incubation due to nest competition with a pair of Broad-billed Rollers for the cavity, and the nest was abandoned by the kestrel pair prior to hatching. In another instance in 1997, three nestlings were killed and partially eaten by an unknown predator. In 1998, one nest failed when the eggs were eaten and an- other nest failed during an intense 5-d rain storm when the nestlings were found dead in and below the nest, possibly due to lack of sufficient brooding by the adult female or due to water in the nest cavity. Food Habits. Madagascar Kestrels used different hunting strategies, including hovering flight and perch hunting. Prey on the ground was hunted 152 Rene de Roland et al. VoL. 39, No. 2 Table 2. Reproductive success of six fully-documented breeding attempts of the Madagascar Kestrel {Falco newtoni) during 199V and 1998 at Masoala Peninsula, Madagascar. Breeding Attempts Number of Eggs Mean Clutgh Size Number Eggs Hatched (%) Number of Young Fledged (%) Fledglings/ Breeding Attempts Nest Success Percent (AO 1997 4 16 4 10 (63) 4 (40) 1.0 (4/4) 50 (4) 1998 2 8 4 3 (38) 3 (100) 1.5 (3/2) 50 (2) Total 6 24 4 13 (54) 7 (54) 1.2 (7/6) 50 (6) from the air by stationary flight or “hovering,” ending in the kestrel plunging down and capturing the prey. The second method was scanning for prey from a high lookout, usually from a tree. Once prey was located, kestrels bobbed their heads several times as if sighting in on the prey before gliding down to seize their quarry. During the two study seasons, 1997 and 1998, we recorded 370 Madagascar Kestrel prey items, mainly those brought to nests; 338 were identified. Plated lizards {Zonosaurus brygooi and Z. madagascariensis) com- prised 93.8% (N = 317), insects 2.6% (N = 9), amphibians 2.4% {N = 8), and birds 1.2% {N = 4). Adult males and females delivered different quantities of prey during the different breeding periods (Table 2) . In total, of 338 identified prey, 80.2% (N= 271) and 19.8% {N = 67) were deliv- ered by males and females, respectively. During the 1998 breeding season, one nesting pair of radio-tagged kestrels had an estimated MCP home range of 25.6 ha (N = 30 locational fixes). Discussion The Madagascar Kestrel is considered to be one of the most common and widespread raptor spe- cies throughout Madagascar (Rand 1936, Siegfried and Frost 1970, Langrand and Meyburg 1984). Fer- guson-Lees and Christie (2001) stated “. . .this is probably the only Malagasy raptor to have gained from deforestation, and is clearly under no threat.” This species appears to be benefiting from deforestation (Langrand and Meyburg 1984), and on the Masoala Peninsula, kestrels are occupying openings, secondary and human-modified habitats adjacent to intact forest fragments and the primary forest (Rene de Roland 1994). Cade (1982) commented that Madagascar Kes- trels show more pronounced sexual-size dimor- phism in comparison to most kestrel species based on work by Siegfried and Frost (1970), who re- ported males averaging 72.9% of female mass (males averaged 105 g [90-117 g, A ^ 4] and fe- males averaged 144 [131-153 g, N = 7]). We were not able to document the extent of sexual dimor- phism, as we measured only one pair, and in this single instance the male was 90% of the female’s mass. All nests found in this study were situated in nat- ural-tree cavities. All cavities seemed to have devel- oped through decay, where a limb had broken or the tree’s heartwood had a rotten opening the in- terior. Langrand (1990) reported that Madagascar Kestrel nests can also be found in cliff holes, under house roofs, and infrequently in old nests of other avian species. The Seychelles Kestrel {F. araea) and Mauritius Kestrels {F. punctatus) are known to use tree cavities for nesting and also potholes in cliff faces (Temple 1977, Cade 1982, Watson 1992). On Masoala Peninsula, rocky and potholed cliff faces are extremely rare and are unavailable for nesting sites for Madagascar Kestrels. In the central plateau region of Madagascar, this species is known to occupy buildings for nesting as is the Seychelles Kestrel (Watson 1992). However, on the Masoala Peninsula humans are recent in- habitants in this region and large suitable build- ings for nesting kestrels are nonexistent. In south- western Madagascar, kestrels occupy abandoned nests of other birds (e.g., Pied Crow [Corvus alhus\', Rene de Roland pers. obs.). In the northeastern region, old stick nests are built (and usually occu- pied) by larger raptors inside forested habitat, and this tends to exclude kestrels from using this type of nesting structure. Consequently, the human degradation of forested habitat on Masoala Pen- insula has left isolated trees for nesting habitat for Madagascar Kestrels. Madagascar Kestrels do not seem to be highly selective in regard to tree species or nest height they use, but are dependent on the availability of suitable tree cavities within the hu- man-modified habitat. On Masoala Peninsula, cav- June 2005 Breeding Biology of the Madagasgar Kestrel 153 ity nesting by Madagascar Kestrels differed mark- edly from sympatric Banded Kestrels, which placed their nests only inside arboreal-epiphytic ferns (Thorstrom 1999, Rene de Roland et al. 2005). During research conducted on raptors on Ma- soala Peninsula from 1991-97, no Madagascar Kes- trels were found nesting inside the intact primary forest (Rene de Roland 1994, Robenarimangason 1999, Thorstrom and Rene de Roland 2000). In contrast, the Mauritius Kestrel and Seychelles Kes- trel did occupy and nest in dense forests, forest fragments and secondary forest patches (Cade 1982, Watson 1992, Cade and Jones 1993). Most of the breeding biology of this species fol- lows the usual kestrel pattern and behavior, but there are some exceptions. For instance, the court- ship period is marked by a distinct flight consisting of a series of climbs and dives with continuous pow- erful wing beats as seen in American Kestrels {F. sparverius; Willoughby and Cade 1964) and Euro- pean Kestrels (F. tinnunculus; Brown and Amadon 1968), but we did not observe such courtship flights for Madagascar Kestrels or the sympatric Banded Kestrels in this region (Rene de Roland et al. 2005). Seychelles Kestrels do not display this specialized courtship flight either (Vesey-Fitzgerald 1940). Both sexes of Madagascar Kestrels, Banded Kestrels, as well as the Seychelles Kestrel do emit specific vocalizations during the courtship period (Watson 1993, Rene de Roland et al. 2005). The copulation duration of 5 sec (range 3—8 sec) for the Madagascar Kestrel was slightly shorter than the sympatric Banded Kestrel with a mean near 8 sec (range of 5-10 sec; Rene de Roland et al. 2005). Newton (1979) noted that the eggs of smaller raptor species are laid every couple of days, but in this study we observed Madagascar Kestrels laying eggs daily. At one nest in 1997, five eggs were laid on five consecutive days. The fresh egg masses for Seychelles Kestrels, 12.4 g or 14% of mean female body mass {N =10 eggs; Watson 1993) and Amer- ican Kestrels, 13.8 g or 11.2% of mean female body mass {N = 53 eggs; Balgooyen 1976) are compa- rable to those of Madagascar Kestrels (14.3 g or 10% of female body mass) from this study. Cade (1960) noted that the genus Falco requires 28-30 d of incubation. The incubation period for Seychelles Kestrels was 28-31 d (Watson 1992), for Mauritius Kestrels about 30 d (Cade 1982) , and we found a similar period for the Madagascar Kestrel ranging from 27-29 d. Like Seychelles Kestrels (Watson 1993) and Mauritius Kestrels (Cade 1982), both female and male Madagascar Kestrels contributed to prey pro- visioning during the nestling period. Madagascar Kestrels dropped prey into the nest cavity during the latter stage of the nestling period, and this has also been observed in American Kestrels (Balgooy- en 1976). Madagascar Kestrels were most aggres- sive against other cavity nesters during the nestling period, and this has been observed with other kes- trel species (Balgooyen 1976). Madagascar Kestrels fledged at 23-24 d of age, much earlier than Seychelles Kestrels at 35-42 d (Watson 1993) and Mauritius Kestrels at 38—39 d (Cade 1982). The shorter nestling period for Mad- agascar Kestrels might be attributed to their ad- aptation to nesting in open habitat, whereas the protracted breeding season of the Mauritius and Seychelles kestrels are adapted to tropical forested situations (Cade 1982, Watson 1992). Adult Madagascar Kestrels continued to use the nest for prey deliveries to the fledglings. Young Madagascar Kestrels caught prey by 14 d after fledging and young American Kestrels also were successful in catching prey at 12-14 d after fledg- ing (Balgooyen 1976). The prey delivered by adult Madagascar Kestrels decreased progressively dur- ing the post-fledgling period until young dispersed and were independent at 44—45 d old. In comparison to an adjacent inland kestrel spe- cies, Seychelles and Mauritius kestrels, the nest suc- cess of 50% of Madagascar Kestrels was lower due to the predation on eggs and nestlings, inclement weather, and possibly the smaller sample size. The Madagascar Kestrel’s diet was reported to comprise mainly of insects and some vertebrates (Rand 1936, Cade 1982, Langrand 1990, Ferguson- Lees and Christie 2001). On Masoala Peninsula, the kestrel’s diet during the breeding season was vertebrates, and predominantly terrestrial plated lizards (Zonosaurus spp.; 93.8%, N= 317). This was similar to the Seychelles Kestrel, which fed mainly on skinks (Mabuya seychellensis) and some day geck- os, {Phehuma spp.) and the Mauritius Kestrel, which consumed mostly day geckos (93%; Cade 1982, Jones 1984, Watson 1992). The sympatric Banded Kestrel studied in the same area also preyed on lizards, but more on arboreal species, such as chameleons {Furcifer and Calumna spp.) and day geckos (Rene de Roland et al. 2005) . In this study, almost half of the identified prey, 158 of 338 (46.7%) , were captured during the nest- 154 Rene de Roland et al. VoL. 39, No. 2 ling period. The majority of prey was delivered by male Madagascar Kestrels (82.3%, N = 130) rela- tive to females (17.7%, N = 28). Watson (1993) reported male Seychelles Kestrels increased their hunting effort during nestling period, delivering 92% of the food to the young. In comparing the Madagascar Kestrel with the two other closely-related insular kestrels, the Sey- chelles and Mauritius kestrels, and the larger sym- patric Banded Kestrel, this species seems to be adapted to open and disturbed habitats. Thus, the Madagascar Kestrel may have benehted from de- forestation, human activities, and villages. Also, in comparison to the other two kestrels, the Mada- gascar Kestrel frequently hovers for hunting, has two color morphs, and has a relatively short breed- ing season. The Mauritius Kestrel is morphologi- cally similar to an accipiter and its habits and be- havior reflect this based on its occupancy of forested habitat, and the Seychelles Kestrel is found both in forested and open habitat (Cade 1982, Watson 1992). Acknowledgments We would like to thank the Masoala technicians of The Peregrine Fund for their assistance in data collection in the field. We are also indebted to the Association Nation- al pour la Gestion des Aires Protegees, Direction de la Gestion Durable des Resources Forestieres, and Antana- narivo University, The Peregrine Fund’s collaborators, for their invaluable administrative help, and especially, for permitting us to undertake this current study at two dif- ferent sites. We kindly thank L. Kiff, J. Bednarz, and two anonymous reviewers for comments on the manuscript. A special thanks the Liz Claiborne and Art Ortenberg Foundation, MacArthur Foundation, and Environment Now for their financial support. Literature Cited Balgooyen, T.G. 1976. Behavior and ecology of the American Kestrel {Falco sparverius) in the Sierra Ne- vada of California. Univ. Calif. Publ. Zool. 103:1-83. Berger, D.D. and H.C. Mueller. 1959. The bal-chatri; a trap for birds of prey. Bird-Banding Brown, L.H. and D. Amadon. 1968. Eagles, hawks, and falcons of the world. Country Life Books, London, U.K. Cade, TJ. 1960. Ecology of the peregrine and Gyrfalcon populations in Alaska. Univ. Calif. Publ. Zool. 63:151- 290. . 1982. The falcons of the world. Cornell Univ. Press, Ithaca, NY U.S.A. AND C.G. Jones. 1993. Progress in the restoration of the Mauritius Kestrel. Conserv. Biol. 7:169-175. Ferguson-Lees, J. and D.A. Christie. 2001. Raptors of the world. Houghton Mifflin Co., New York, NY U.S.A. Guillaumet, J.-L. 1984. The vegetation: an extraordinary diversity. Pages 27-54 in A. Jolly, P. Oberle, and R. Albignac [Eds.], Key environments: Madagascar. Per- gamon Press, New York, NY U.S.A. Jones, C.G. 1984. Eeeding ecology of the Mauritius Kes- trel. Page 209 in J. Mendelsohn and C.W. Sapsford [Eds.], Proceedings of the second symposium on Af- rican predatory birds. Natal Bird Club, Durban, South Africa. Kenward, R.E. 1990. RANGES IV: software for analyzing animal location data. Institute of Terrestrial Ecology, Warham, U.K. Langrand, O. 1990. Guide to the birds of Madagascar Yale Univ. Press, New Haven, CT U.S.A. AND B.-U. Mevburg. 1984. Birds of prey and owls in Madagascar: their distribution, status and conser- vation. Pages 3-13 inj. Mendelsohn and C.W. Saps- ford [Eds.], Proceedings of the second symposium on African predatory birds. Natal Bird Club, Durban, South Africa. Newton, I. 1979. Population ecology of raptors. Buteo Books, Vermillion, SD U.S.A. NicOll, M.E. and O. Langrand. 1989. Madagascar: revue de la conservation et des aires Protegees. World Wide Fund for Nature International, Gland, Switzerland. Rand, A.L. 1936. The distribution and habits of Mada- gascar birds. Bull. Amer. Mus. Nat. Hist. 72:143-499. Rene de Roland, L.-A. 1994. Methode d’inventaire de rapaces dans la presqu’ile de Masoala. M.S. thesis, Univ. d’Antananarivo, Antananarivo, Madagascar. . 2000. Contribution a Fetude biologique, ecolo- gique et ethologique de trois especes d’Acapzfcr dans la Presqu’Ile Masoala. Ph.D. dissertation, Univ. d’Antananarivo, Antananarivo, Madagascar. , J. Rabearivony, G. Razafimanjato, H. Roben- arimangason, and R. Thorstrom. 2005. Breeding biology and diet of Banded Kestrels Falco zoniventris on Masoala Peninsula, Madagascar. Ostrich. 76:32- 36. Robenarimangason, H. 1999. Contribution a Fetude de la biologique, ecologique, et ethologique de la famille de falconidae, faucon de newton, Falco newtoni, et fau- con a ventre raye, F. zoniventris, dans la partie occi- dentale de la presqu’ile Masoala. M.S. thesis, Univ. d’Antananarivo, Antananarivo, Madagascar. Siegfried, W.R. and P.G.H. Frost. 1970. Notes on the Madagascar Kestrel Falco newtoni. Ibis 112:400—402. Temple, S. 1977. The status and conservation of endemic kestrels in the Indian Ocean islands. Pages 78-82 in R.D. Chancellor [Ed.], Proceedings world conference of birds of prey. International Council for Bird Pres- ervation, Cambridge, U.K. Thorstrom, R. 1996. Methods for capturing tropical for- est birds of prey. Wildl. Soc. Bull. 24:516-520. June 2005 Breeding Biology of the Madagascar Kestrel 155 . 1999. A description of nests, diet, and behaviour of the Banded Kestrel. Ostrich 70:149-151. , A. Andrianarimisa, and L.-A. Rene de Roland. 1997. Weather at Andranobe Field Station, Western Masoala Peninsula. Pages 127—130 mR. Watson [Ed.], Progress Report IV. The Peregrine Fund, Boise, ID U.S.A. , AND L.-A. Rene de Roiand. 2000. Status and con- servation of raptors on the Masoala Peninsula, Mad- agascar. Pages 35-41 in R.D. Chancellor and B.-U. Meyburg [Eds.], Raptors at risk. World Working Group on Birds of Prey and Owls, Hancock House, Berlin, Germany. Vesey-Fitzgerai.d., D.F. 1940. The birds of Seychelles. Ibis, 14th Sex 4:480-489. Watson, J. 1992. Nesting ecology of the Seychelles Kes- trel Falco araea on Mahe, Seychelles. Ibis 134:259-267 . 1993. Breeding cycle of the Seychelles Kestrel Pages 73-79 in M.K. Nichols and R. Clarke [Eds], Proceedings of the 1991 hawk and owl trust confer- ence, September 1991. The Hawk and Owl Trust, London, U.K. Willoughby, E.J. and T.J. Cade. 1964. Breeding behavior of the American Kestrel. Living Bird 3:75-96. Received 7 October 2003; accepted 14 February 2005 Short Communications J Raptor Res. 39 (2): 156-1 59 © 2005 The Raptor Research Foundation, Inc. Interspecific Aggression and Nest-site Competition in a European Owl Community Inigo Zuberogoitia^ Estudios Medioambientales Icarus s.L, Pintor Sorolla 6, 1-C, 26007 Logrono, La Rioja, Spain Jose Antonio Martinez Juan de la Cierva 43, E-03560 El Campello, Alicante, Spain Jabi Zabala Sebero Ochoa, 45, 5~ B, 48480 Arrigorriaga, Bizkaia, Spain Jose Enrique Martinez Dpto de Ecologia e Hidrologia, Universidad de Murcia, Campus de Espinardo, E-301 00 Espinardo, Murcia, Spain Key Words: Bam Owl\ Tyto alba; Little Owl; Athene noc- tua; Scops Owl; Otus scops; Tawny Owl; Strix aluco; com- munity; competition; predation. Interspecific killing among predators of the same guild has been extensively reported, but is still relatively un- studied (Mikkola 1983, Kostrzewa 1991, Palomares and Caro 1999). Although Mikkola (1983) summarized 1363 cases of owls killed by other owls, it was not always clear whether birds were taken as prey or killed for other rea- sons. Indeed, some of these owls may have been killed during defense of nest sites, as food competitors, or a few may have been found dead and scavenged. Others may have been killed, but not actually eaten. Palomares and Caro (1999) pointed out that interspecific killing may remove potential predators or their offspring, free up re- sources that would otherwise be consumed by competi- tors or provide energetic benefits as prey, although atyp- ical in the diet. On the other hand, Jaksic and Braker (1983) and Mar- ti et al. (1993) showed that predator assemblages can be organized in feeding guilds (i.e., clusters of species within which interspecific dietary overlap is more extensive), al- though they did not take into account the habitat di- mension of these respective niches. Herrera and Hiraldo (1976) showed a weak clustering effect due to interspe- cific dietary overlap in owl communities in the Iberian Peninsula. In this case, we would expect that spatial seg- regation would be the most common dimension of re- source partitioning in the owl community (Schoener 1974, Nilsson 1984, Danielson 1991, Venier and Fahrig 1996). 1 Email address: inigo.zuberogoitia@wanadoo.es Competition among species is difficult to assess, and in spite of great interest in such interactions, the actual influence of direct and indirect effects of this process is still far from clear (Palomares and Caro 1999). Mikkola (1983) explained that existing data are too circumstantial to allow an evaluation of the important benefits related to the competition. As several factors may be influencing population dynamics, the importance and degree of in- terspecific killing among raptors needs to be assessed by long-term, intensive studies exploring owl interactions Bizkaia offers a unique opportunity to examine this issue in Europe, as the owl population has been studied for over a decade (e.g., Zuberogoitia and Campos 1998, Zu- berogoitia and Martinez 2000, Zuberogoitia 2002). Here, we report rates of interspecific aggression and nest-site competition among seven species of owls that we cen- sused during the above-mentioned research. Methods Study Area. This work was conducted in Bizkaia, in northern Spain (43°22'N, 2°4TW) between 1992-2002. This is a 2300-km^ area covered primarily by forest (70%), mainly conifers, especially Monterey pine {Pinus radiata), which occupies 53% of the forested area (De- partamento de Ordenacion del Territorio y Medio Am- biente 2001). In Bizkaia, Tawny Owls (Strix aluco) reach one of the highest densities found in Europe, with 1700 known territories (Zuberogoitia and Campos 1998, Zu- berogoitia 2002). The lowlands and rural areas are sur- rounded by old fields and agriculture, where owls more characteristic of open space live (e.g.. Barn Owl [Tyto alba] , Little Owl [Athene noctua] , Scops Owl [ Otus scops ] ) , with 407, 272, and 26 known territories, respectively (Zu- berogoitia 2002). The rest of the owl guild is comprised of Long-eared Owls (Asio otus) , with nine known territo- ries; Short-eared Owls (Asio flammeus) , present only dur- 156 June 2005 Short Communications 157 Table 1. Number and proportion of total playbacks and territories where interspecific attacks occurred. For calcu- lating the percentages we considered all the cases when we used the broadcast method (2056) and used the number of known territories for each attacked species. Prey Attack on Tape Recorder Attack on Another Owl Attacker N Percent N Percent Little Owl 4 0.2 1 0.4 Barn Owl 2 0.1 1 0.4 Tawny Owl Long-eared Owl 2 0.1 1 0.1 Tawny Owl mg the migratory fluxes and winter; and Eagle Owls {Bubo bubo) with three territories. Survey Methods. We used three different techniques to assess competition and aggression among owls. The first was the playback method, conducted between De- cember 1992 and December 1996, that we used to elicit territorial vocalizations from the seven owl species at 2056 point count stations and recorded the number of interspecific attacks on the broadcast speaker. Scops Owls were surveyed between April and September because they do not winter in the study area; otherwise, we sur- veyed all species twice a week throughout the year (x = 7.26 count stations, SD = 2.13). We broadcasted taped vocalizations (male, female, and owlet vocalizations re- corded in a continuous format) according to the size of the owl (smallest to largest) for 5 min, and then assessed reactions of owls during the subsequent 10 min. The broadcast speaker was placed 1.5 m above the ground, while two to four observers surrounded it separated by 10-20 m. Surveys began at dusk, continued for an aver- age of 5 hr, and were performed in all kinds of weather except during very windy (>35 km/hr) and stormy nights. Further details about survey methods are de- scribed by Zuberogoitia and Campos (1998), Zuberogo- itia and Martinez (2000, 2001), Martinez and Zubero- goitia (2002), and Martinez et al. (2002). Second, we reviewed our notes made during 3084 hr observing behavior of owls in our study area, from which instances of interspecific interactions involving physical contact among owl species were tallied. Finally, following Zuberogoitia and Campos (1998), we located breeding sites of 181 Barn Owls, 83 Little Owls, 77 Tawny Owls, nine Scops Owls, and three Long-eared Owls, and ex- amined nest-site competition among these species. We considered that competition for nest sites existed when one species displaced another from its nest before the end of the nesting season. Data Analysis. We calculated relative frequency and proportion of interspecific attacks during the 2056 point counts and recorded species of both the attacker and target species. Independent of the call broadcasts, we cal- culated number of interspecific attacks observed inciden- tally during the study period. We used number of terri- tories of each species as the total sample when calculating proportion of territories at which attacks occurred. For example, one attack involving a Little Owl in its territory is 1/272, as there were 272 known territories of this spe- cies. Third, we calculated frequency and proportion of species that were expelled from their nests by another species. Results We registered eight cases of owls attacking the playback station while broadcasting the call of a different species (Table 1), and three cases of interspecific aggression ob- served incidentally. Most attacks were aimed at Little Owls, and the main aggressors were Barn and Tawny owls. We also documented interspecific competition for breeding sites. Tawny Owls displaced Barn Owls six times (3.3% of the recorded nests), and all such cases occurred during the egg-hatching period. However, we also found evidence that many owl species within the guild did not interact with each other aggressively, even though they nested in close proximity. In five cases, two or three dif- ferent species shared the same building for breeding (Ta- ble 2). All bred successfully, and we did not record pre- dation or aggressive behaviour among them. Discussion The frequency of direct attacks by an owl species on another and the frequency of interspecific attacks to play- back stations were very low. Tawny Owls appeared to be the most aggressive species of the guild, attacking Little Table 2. Number of cases in which two or more owl species nested in the same building at the same time. The percentage data were obtained considering all nests monitored for each species. N Percent Percent Percent 2 Little Owl 2.4 Barn Owl 1.1 Tawny Owl 2.6 2 Little Owl 2.4 Barn Owl 1.1 2 Barn Owl 1.1 Tawny Owl 2.6 158 Short Communications VoL. 39, No. 2 and Long-eared owls and expelling Barn Owls from their nests. Similarly, Tawny Owls can show a high degree of mtraspecific competition, as territoriality is often the cause of fights that can lead to the killing of an intruder (Zuberogoitia and Martinez 2000) . Hence, it may not be surprising that such an aggressive species would defend Its resources vigorously against other species. Barn Owls were also aggressive against other species. All observed cases were aimed at Little Owls, although the frequencies of such interactions were almost negli- gible. Our results are similar to those of Mikkola (1983), who found that the only owls killed by Barn Owls were Little Owls, but very infrequently. In our study areas in Valencia (eastern Spain), we also have witnessed two cas- es of resident male Barn Owls expelling Long-eared Owls from their territories after brief aerial fights (J.A. Marti- nez and I. Zuberogita, unpubl. data). According to Mik- kola (1983), shortage of suitable breeding places for owls may lead to interspecific conflicts. Natural cavities are in short supply, and therefore, presumably a limited re- source for owls in Bizkaia, which helps explain why owls tend to breed in alternative sites. Such sites include vaults of churches, attics of houses, and piles of hay or branches (Zuberogoitia 2002). Thus, both interspecific and intra- specific competition for such limited resources would be expected (Newton 1979), especially if food availability is high, and the structural characteristics of the habitat suit the hunting mode of several species. Tawny Owls are ex- tremely abundant in our study area despite that avail- ability of suitable nest holes is low because of timber har- vesting (Zuberogoitia 2002). Accordingly, these owls recently have increased use of anthropogenic structures (mainly buildings) for nesting. Barn Owls also select buildings for nesting (Zuberogoitia 2002), but they seem to be at a disadvantage when confronted by the more aggressive Tawny Owls in competition for nest sites. Nev- ertheless, Bunn et al. (1982) described a single case of a pair of Barn Owls chasing away a Tawny Owl that had entered a barn where they were nesting. Therefore, even if we were not monitoring all the nests m the owl guild, our results suggested that competition between Tawny Owls and Barn Owls occurred at least at the nest-site level, although sharing of structures sup- porting breeding sites occurred occasionally. Current land management practices favoring timber plantations over deciduous woods (which provide natural cavities for forest owls) have created Tawny Owl hunting habitat ar- tificially by increasing the availability of edges within woods (Zuberogoitia 2002). In these habitats, Tawny Owls have adapted to breeding in diurnal-raptor nests and even in buildings, which may support a high density of Tawny Owls (Zuberogoitia 2002) competing for a lim- ited number of nest sites with less aggressive, open-space dwellers such as Barn Owls, Little Owls, and Scops Owls (Taylor 1994, Zuberogoitia 2002). Jaksic (1988) wondered about effects of removing dominant owls on the abundance and diversity of local predator assemblages. For example, Eurasian Eagle-Owls can kill smaller owls and raptors (Mikkola 1983, Saurola 1995) or influence the composition of predator guilds (Sergio et al. 2003). Whether the wide range of habitats occupied by Tawny Owls and their high density in Bizkaia are also a consequence of the lack of competition by a larger owl is still an open question. AgRESION iNTERESPEClFICA Y COMPETENCIA POR SiTIOS DE NiDIFICACION EN UNA COMUNIDAD EUROPEA DE BtJHOS Resumen. — La depredacion entre depredadores de una misma comunidad no ha sido bien estudiada. Con objeto de comprender la frecuencia y la magnitud de las agre- siones interespecificas en una comunidad europea de ra- paces nocturnas, analizamos la frecuencia de contactos agresivos (ataques) y apropiaciones de nidos entre las sie- te especies de buhos presentes en un area de 2300 km^ ubicada en Bizkaia (Espana) entre 1992-2002. Repro- dujimos reclamos previamente grabados de las siete es- pecies en 2056 puntos de censo, comenzando con los de la especie mas pequena y finalizando con los de la mas grande. Durante los reclamos registramos (1) la frecuen- cia con la que se producian ataques interespecificos, y (2) las especies implicadas. Solo registramos ocho ataques, los cuales fueron dirigidos a especies de menor tamano que la especie atacante. Ademas, durante mas de 3000 horas de observaciones de rapaces nocturnas en el area de estudio, registramos tres casos de ataque directo de una especie contra otra. For ultimo, constatamos siete casos de competencia directa por los lugares de nidifi- cacion, en los que una especie fue desplazada del nido por otra especie antes de finalizar el periodo reproduc- tivo. Sugerimos que el nivel de agresion esta relacionado con el tamano de la especie, de forma que las especies de mayor tamano atacan a las mas pequenas. Sin embar- go, las agresiones son muy poco frecuentes, por lo que nuestros datos sugieren que estas especies rara vez com- piten directamente entre si de forma directa o apropian- dose de los nidos. En cambio, las especies podrian estar compitiendo de foma menos evidente. [Traduccion de los autores] Acknowledgments Agurtzane Iraeta, Ainara Azkona, Sonia Hidalgo, Luisa Fernanda Campos, Lander Astorkia, Julen Zuberogoitia, Inaki Castillo, Fernando Ruiz-Moneo, Javier Elorriaga, and Raul Alonso helped in the fieldwork. The manu- script was greatly improved by comments from Geir A Sonerud, Kent Livezey, Tania Tripp, Jim Belthoff, and an anonymous referee. Literature Cited Bunn D.S., A.B. Warburton, and R.D.S. Wilson. 1982. The Barn Owl. T. & A.D. Poyser, Cal ton, U.K. Danielson, BJ 1991. Communities and landscape: the influence of habitat heterogeneity on the interactions between species. Am. Nat. 138:1105-1120. June 2005 Short Communications 159 Departamento de Ordenacion del Territorio y Medio Ambiente. 2001. El medio ambiente en la comunidad autonoma del Pais Vasco. Servicio editorial del Go- bierno Vasco, Vitoria-Gasteiz, Spain. Herrera, C. and F. Hiraldo. 1976. Food-niche relation- ships among European owls. Ornis Scand. 7:29-41. Jaksic, F. 1988. Trophic structure of some nearctic, neo- tropical and palearctic owl assemblages: potential roles of diet opportunism, interspecific interference and resource depression./. Raptor Res. 22:44—52. AND H.E. Braker. 1983. Food-niche relationships and guild structure of diurnal birds of prey: compe- tition versus opportunism. Can.]. Zool. 61:2230-2241. Kostrzewa, a. 1991. Interspecific interference competi- tion in three European raptor species. Ethol. Ecol. Evol. 5:127-145. Marti C.D., K. Steenhof, M. Kochert, and J. Marks. 1993. Community trophic structure: the roles of diet, body size and activity in vertebrate predators. Oikos 67: fi- 18 . Martinez, J. A. and I. Zuberogoitia. 2002. Factors affect- ing the vocal behaviour of Eagle Owl Bubo bubo: effects of sex and territorial status. Ardeola 49:1-9. , , J. Colas, and J. Macia. 2002. Use of re- corder calls for detecting Long-eared Owls Asio otus. Ardeola 4Q:^7— 101. Mikkoia, H. 1983. Owls of Europe. T. & A.D. Poyser, Gal- ton, U.K. Newton, I. 1979. Population ecology of raptors. T. & A.D. Poyser, Hertfordshire, U.K. Nilsson, I. 1984. Prey weight, food overlap, and repro- ductive output of potentially competitive Long-eared Owls and Tawny Owls. Ornis Scand. 15:176-182. Palomares, F. and T.M. Caro. 1999. Interspecific killing among mammalian carnivores. Am. Nat. 153:492-508. Saurola, P. 1995. Owls of Finland. Hiijayhtyma Oy, Hel- sinki, Finland. Sergio, F, L. Marches:, and P. Pedrini. 2003. Spatial re- fugia and the coexistence of a diurnal raptor with in- traguild owl predator./. Appl. Ecol. 72:232-245. ScHOENER, TW. 1974. Resource partitioning in ecological communities. Science 185:27-39. Taylor, I. 1994: Barn Owls: predation-prey relationships and conservation. Cambridge University Press, Cam- bridge, U.K. Venier, L.A. and L. Fahiug. 1996. Habitat availability causes the species abundance-distribution relation- ship. Oikos 76:564—570. Zuberogoitia, I. 2002. Ecoethology of the Bizkaia’s owls Ph.D. dissertation, Univ. Pais Vasco, Leioa, Spain. AND L.F Campos. 1998. Censusing owls in large areas: a case study. Ardeola 45:47-53. AND J.A. MartInez. 2000. Methods for surveying Tawny Owl Strix aluco populations in large areas. Biota 1:137-146. AND . 2001. The Little Owl in the “Proy- ecto Noctua.” Pages 103-108 mJ.C. Genot, J.M. La- pios, P. Lecomte, and R.S. Leigh, [Eds.], Chouette cheveche et territorie. Actes du colloque de Champ- sur-Marne, Paris, France. Received: 31 December 2003; accepted 22 February 2005 Associate Editor: James R. Belthoff /. Raptor Res. 39(2):159-163 © 2005 The Raptor Research Foundation, Inc. Prey Partitioning between Mates in Breeding Booted Eagles (Hieraaetus pennatus) Jose E. Martinez and Jose F. Calvo^ Departamento de Ecologia e Hidrologia, Universidad de Murcia, Campus de Espinardo, 30100 Murcia, Spain Keywords: Booted Eagle, Hieraaetus pennatus; food par- titioning, forest; prey provisioning, reversed size dimorphism (RSD). Reversed sexual-size dimorphism (RSD) is widespread m raptors and owls, with females being larger than males (Newton 1979). Several researchers have proposed that this trait is driven by different selective forces acting on breeding adults (Mueller and Meyer 1985, Massemin et al. 2000, Simmons 2000). However, no explanation has gained universal acceptance (Bildstein 1992). One of the most popular explanations is the prey-partitioning hy- pothesis or female supplementary feeding hypothesis (Reynolds 1972, Korpimaki 1985), which suggests that RSD is advantageous because it allows females to hunt larger prey, widening the prey base available for the pair and reducing intersexual competition for food (Snyder and Wiley 1976, Andersson and Norberg 1981, Massemin et al. 2000). Several authors (e.g., Snyder and Wiley 1976, ^ Email address: jfcalvo@um.es 160 Short Communications VoL. 39, No. 2 Newton 1979, Simmons 2000) have noted that the degree of RSD among raptor species shows a strong relationship with the proportion of birds in the diet. Nonetheless, studies addressing differential prey-size choice between sexes have been equivocal (Opdam 1975, Collopy 1984, Kennedy and Johnson 1986, Boal and Mannan 1996). Between 1998 and 2000, we conducted a study on a breeding population of Booted Eagles {Hieraaetus penna- tus) in southeastern Spain. This species is a medium-sized raptor showing a moderate degree of reversed size di- morphism {x male body mass = 709 g, female = 975 g; del Hoyo et al. 1994, Balbontin et al. 2001). Although the Booted Eagle is a common bird of prey of the forests and woodland areas of the Iberian Peninsula (Veiga and Vihuela 1994), little is known about its diet. Earlier stud- ies in Europe and South Africa describe it as a small- and medium-sized bird hunter, also preying on lizards and mammals (Steyn and Grobler 1985, Veiga 1986, Martinez et al. 2004). Here, our objective was to analyze the prey items de- livered to the nest by male and female Booted Eagles, examining differences in the kind and body mass of prey between the genders. Methods We carried out the study between 1998—2000 in central Murcia (southeastern Spain; 38°00'N, 1°45'W). The study area included about 10 000 ha and ranges from 550—1521 m above sea level, with a topography characterized by rugged slopes dominated by pine forests {Pinus halepen- sts) interspersed with traditional agroecosystems (cereal plots, vineyards, and olive and almond groves). Dietary differences between males and females were assessed from observations of prey deliveries at five dif- ferent nests. Prey items were identified from blinds lo- cated 30—50 m away from the nests using a spotting scope and binoculars. One nest was observed in 1998, two in 1999, and other two in 2000. Observations were con- ducted every 4-8 d from the early nestling stages (late May) until 5-7 d after fledging (early August) . Nest ob- servations started at 1100 H, lasted until 1900 H, and continued the following day from 0700-1100 H. In total, 509 hr of direct observation were made, during which we recorded visually all prey deliveries. A prey item was as- sumed to have been captured by the male when: (1) we observed the male delivered prey to the nest or (2) the prey captured by the male was delivered to the nest by the female after a food transfer involving characteristic vocalizations. A prey item was assumed to have been cap- tured by the female when she delivered it and no food transfer was observed. Biomass of prey was estimated based on data reported by Van den Brick and Barruel (1972) and Manosa (1994). Mass data was log transformed for analysis. Prey Items were also assigned to the following categories: mammals, large adult birds (>200 g), small adult birds (<200 g), nestlings (<100 g) and reptiles. We also con- sidered two nesting periods (early, until the chicks were ca. 30 d old, and late, until the end of observations), to assess temporal differences in the type and mass of the prey captured. We used Morisita’s index (Krebs 1999) to assess over- lap in the prey species caught by male and female eagles. A linear mixed-effects model was employed to evaluate differences in the mean mass of prey captured by each mate. Nest was considered as a random factor to avoid pseudoreplication and the temporal factor was included as a fixed effect. The proportions of each prey type and size were also assessed by the analysis of four-way contin- gency tables using Poisson log-linear models and likeli- hood ratio test (Venables and Ripley 2002) to examine the effects of sex, nest, and nesting period as explanatory factors. Statistical analyses were performed with the R sta- tistical package (Maindonald and Braun 2002). Results We identified to the species level 117 of 127 prey items delivered to nests (Table 1). Birds made the bulk of the Booted Eagle diet (65.35%), followed by ocellated lizards (Lacerta lepida\ 26.77%), and mammals (7.87%). Females brought 40 prey items (31.50%) and males 87 (68.70%). Of these, 54 were delivered directly by the male, and 33 were previously transferred and then delivered by the fe- male. The prey-provisioning rate was 0.24 prey items/hr (0.08 for females and 0.16 for females). A moderate degree of dietary similarity between the sexes was found (Morisita’s index = 0.67). However, the mean mass of prey captured by males and females dif- fered significantly (Tj iiy = 11.50, P< 0.001). The tem- poral factor and its interaction with the gender were not significant (Ti ng ~ 0.20, P = 0.66 and Tj j^g = 0.19, P = 0.59, respectively). The analysis of the prey-type contingency table re- vealed that sex, nesting period, and nest factors signifi- cantly influenced prey type (sex: = 25.42, P < 0.001; nesting period: x^ = 23.79, P < 0.001; nest: x^ = 72.40, P< 0.001;). Based on the Poisson model, males delivered greater proportions of small birds, nesdings, and reptiles than females (Table 2) . Both sexes captured similar pro- portions of large adult birds and mammals. Discussion Our results show that differences in provisioning rates between sexes was moderate and similar to those report- ed by Simmons (1986), Manosa (1994), and Gronnesby and Nygard (2000) for a variety of small and medium- sized forest raptors. The Morisita’s index points to a mod- erate dietary similarity between sexes, a finding which differs from those of Kennedy and Johnson (1986) and Boal and Mannan (1996) for Cooper's Hawk {Acdpiter cooperit) and Northern Goshawk {Accipiter gentilis) , respec- tively. These workers found extensive niche overlap be- tween the sexes in these species that also exhibit substan- tial sexual dimorphism. Our study suggests prey-size partitioning between pair members of Booted Eagles during the chick-rearing stage. Previous studies have reported a similar tendency June 2005 Short Communications 161 o o 0 CM I 00 01 oi a, c V c/3 C3 OJ O c/3 u « rH Ij t 3 (U (-1 Ph h 2 ; u o Pi Pi Ph to 0 d 2 D Q Q H 2 ; to to Ph to O d 2 D Q Q 2: p Q o 00 irj O CM o> OJ O 1— i CO O CM o o ir: q lo in o CM iri CM CM o CM m q q q q CM hH i-h O in CD q CM GO O J> CM o in CM r-H CM O in o o o o o m m m O CM rH CM J 8 -I s I 8 S |l O >3 .6 c?'qqqqq odooooooooo q iri t- qqqqqqqqqqqqqqqqoqmooq CMt^dtPcMOinindddcMddcMCMcBdcMddd GO rto 00 i-HOOOmi— lOCMCMOTfO^OO^rHOO^OO CM 00 O O d d GO q q q CM CM in m o d qqqqqqqqqcMqq O O rto O rH O O O rH O GO m O O Tfi d d GO 00 J> (D O in d GO CM O d ooi— ii— (oo-^TfieooooooootD CM^^’cMCOGOCMCMdCMCMod m OOOOCD^m'sfOCMOCMrH,— iCMOOGOGMCMOCMCMi-H in ooooomoomooooooomooooo ooi>oini>oair-iooJ>j>'^-Gficr>in-^incDCM <0 GO CM Tto t-H rto rto CM Xi _to s 0 D 0 cn OOO fzi V q q GO 00 q q tP i> GO GO GO m +1 nr rto GO o o d d GO GO q q in in GO GO 1 1 1 1 I 'r -}- I GO GO 00 00 o m • 04 53 t3 to N JJ d X! OJ ■to d Cp QJ to bO O O Ji OJ o to CIh Q X3 OJ 'to' to X3 § s >- o; u O. T3 o; u > o u o; u 0 =+-( 0 bJD I Sd cG C PCX P D « o A ^ w 2 -C u 162 Short Communications VoL. 39, No. 2 Table 2. Estimated probabilities from the main effects log-linear model applied to the prey-type frequency table. Probabilities from the hve nests are summed. Prey Type Earia Nestling Period Late Nestling Period Males Females Males Females Mammals 0.43 0.27 0.07 0.22 Large adult birds (s200 g) 0.20 0.10 0.28 0.44 Small adult birds (<200 g) 0.53 0.22 0.16 0.11 Nestlings (<100 g) 0.67 0.07 0.22 0.05 Reptiles 0.69 0.04 0.22 0.04 in raptors and owls (Simmons 1986, Mahosa and Cordero 1992, Overskaug et al. 1995), although several authors found only weak evidence of prey partitioning between sexes during the nestling period (Widen 1984, Toyne 1998, Delannoy and Cruz 1999). Males in our study, on average, delivered smaller prey items than females. The taking of smaller prey by males could be related to their smaller body size and greater agility, which would favor the search, pursuit, and hunting of such prey (Temeles 1985) and reduce the costs of handling prey (Villaran 2000). Females may reduce competition for food with males by taking larger prey; Booted Eagle females weigh 27% more than males on average (del Hoyo et al. 1994), which is consistent with theories concerning the selective advantage of RSD (Andersson and Norberg 1981, Te- meles 1985). In summary, our findings suggest that in- tersexual prey-size partitioning may be related to the sex- ual dimorphism of this species. Captura Diferencial de Presas for Machos y Hembras DE Hieraaetus pennatus Resumen. — Entre 1998 y 2000 estudiamos la dieta de ma- chos y hembras de Hieraaetus pennatus, mediante el con- trol visual de cinco nidos en una zona forestal del sureste de Espaha. Machos y hembras capturaron respectiva- mente el 69% y el 31% del total de las presas aportadas a los nidos. La tasa de aporte fue de 0.24 presas/hora, siendo las aves la dieta predominante de los polios (65%). Encontramos un moderado nivel de similitud en- tre las capturas de machos y hembras, pero la biomasa media y las frecuencias de los diferentes tipos de presas capturados fueron significativamente distintos entre am- bos sexos. Nuestros resultados sugieren la existencia de diferencias entre la dieta de ambos sexos, probablemente relacionadas con el dimorhsmo de tamaho. [Traduccion de los autores] Acknowledgments We thank S. Ovidio, C. Garcia, M. Clemente, R. Ruiz, and M. Carrete for help in the held work. We are also indebted to M.A. Sanchez, J. Rodriguez, and C. Sobrado for their assistance in the identihcation of several prey remains. J. Balbontin, J.C. Senar, and E. Cerezo made valuable comments on the manuscript. This work was funded by Spanish Ministry of Science and Technology (project REN2002-01884/GLO, partially hnanced by FEDER funds). Literature Cited Andersson, M. and A. Norberg. 1981. Evolution of re- versed sexual size dimorphism and role partitioning among predatory birds, with a size scaling of flight performance. Biol. J. Linn. Soc. 15:105-130. BALBONTiN, J., M. Ferrer, and E. Casado. 2001. Sex de- termination in Booted Eagles {Hieraaetus pennatus) us- ing molecular procedures and discriminant function analysis. J. Raptor Res. 35:20—23. Bildstein, K.L. 1992. Causes and consequences of re- versed sexual-size dimorphism in raptors: the head start hypothesis. /. Raptor Res. 26:115-123. Boat, C.W. and R.W. Mannan. 1996. Prey sizes of male and female Northern Goshawks. Southwest. Nat. 41. 355-358. COLLOPY, M.W. 1984. Parental care and feeding ecology of Golden Eagle nestlings. Auk 101:753-760. DEL Hoyo, J., A. Elliott, and J. Sargatal. 1994. Hand- book of the birds of the world, Vol. 2. Lynx Edicions, Barcelona, Spain. Delannoy, C.A. and A. Cruz. 1999. Patterns of prey abundance and use by male and female Puerto Rican Sharp-shinned Hawks. Caribb. J. Sci. 35:38-45. Gronnesby, S. and T. Nygard. 2000. Using time-lapse vid- eo monitoring to study prey selection by breeding gos- hawks Accipiter gentilis in central Norway. Ornis Fenn 77:117-129. Kennedy, P.L. and D.R. Johnson. 1986. Prey-size selec- tion in nesting male and female Cooper's Hawks. Wil- son Bull. 98:110-115. Korpimaki, E. 1985. Diet of the kestrel Falco tinnunculus in the breeding season. Ornis Fenn. 62:130-137. Krebs, C.J. 1999. Ecological methodology, 2nd Ed. Ben- jamin/Cummings, MenloPark, CA U.S.A, Maindonald, j. and j. Braun. 2002. Data analysis and graphics using R. Cambridge Univ. Press, Cambridge, U.K. June 2005 Short Communications 163 Manosa, S. 1994. Goshawk diet in a Mediterranean area of northeastern Spain. J. Raptor Res. 28:84-92. AND P. Cordero. 1992. Seasonal and sexual vari- ation in the diet of the Common Buzzard in north- eastern Spain. J. Raptor Res. 26:235-238. Martinez, J.E., M. Cremades, I. Pagan, and J.F. Calvo. 2004. Diet of Booted Eagles in southeastern Spain. Pages 593-599 in R.D. Chancellor and B.-U. Meyburg [Eds.], Raptors worldwide. World Working Group on Birds of Prey & Owls. MMD/BirdLife, Hungary, Bu- dapest. Massemin, S., E. KorpimAki, and J, Wiehn. 2000. Reversed sexual size dimorphism in raptors: evaluation of the hypotheses in kestrels breeding in a temporally changing environment. Oecologia 124:26-32. Mueller, H.C. and K. Mever. 1985. The evolution of re- versed sexual dimorphism in size: a comparative anal- ysis of the Falconiformes of the western Palearctic. Pages 65-101 in R.F. Johnston [Ed.], Current orni- thology. Plenum, New York, NY U.S.A. Newton, I. 1979. Population ecology of raptors. T. Sc A.D. Poyser, Berkhamsted, U.K. Opdam, P. 1975. Inter- and intraspecific differentiation with respect to feeding ecology in two sympatric spe- cies of the genus Accipiter. Ardea 63:30-54. OvERSKAUG, K., E. Kristiansen, and P. Sunde. 1995. Sex- specific diet analysis of the Tawny Owl {Strix aluco) in Norway./. Raptor Res. 29:137-140. Reynolds, R.T. 1972. Sexual dimorphism in Accipiter hawks; a new hypothesis. Condor 74:191-197. Simmons, R.E. 1986. Ecological segregation of the Red- breasted Sparrowhawk Accipiter rufiventris and six co- existing Accipitrine raptors in southern Africa. Ardea 74:137-149. . 2000. Harriers of the world: their behavior and ecology. Oxford Univ. Press, Oxford, U.K. Snyder, N.F.R. and J.W. Wiley. 1976. Sexual size dimor- phism in hawks and owls of North America. Ornithol Monogr. 20:1-96. Steyn, P. and J.H. Grobler. 1985. Supplementary obser- vations on the breeding biology of the Booted Eagle in southern Africa. Ostrich 56:151-156. Temeles, E. 1985. Sexual size dimorphism of bird-eating hawks: the effect of prey vulnerability. Am. Nat. 125 485-499. Toyne, E.P. 1998. Breeding season diet of the goshawk Accipiter gentilis in Wales. Ibis 140:569-579. Van den Brick, FH. and P. Barruei,. 1972. Guia de cam- po de los mamiferos salvajes de Europa occidental. Omega, Barcelona, Spain. Veiga, J.P. 1986, Food of the Booted Eagle (Hieraaetus pennatus) in central Spain. Raptor Res. 20:120-123. AND J. ViNUELA. 1994. Booted Eagle Hieraaetus pen- natus. Pages 182-183 in G.M. Tucker and M.F Heath [Eds.], Birds in Europe: their conservation status. BirdLife International, Cambridge, U.K. Venables, B.D. and W.N. Ripley. 2002. Modern applied statistics with S, 4th Ed. Springer, New York, NY U.S.A. VillarAn, a. 2000. Analisis comparativo de la dieta de ambos sexos en el carabo comun Strix aluco en la Pen- insula Iberica. Ardeola 47:203-213. Widen, P. 1984. Activity patterns and time budgets in the goshawk Accipiter gentilis in a boreal forest area of cen- tral Sweden. Ornis Fenn. 61:109-112. Received 6 May 2004; accepted 22 March 2005 Associate Editor: Juan Jose Negro J Raptor Res. 39 (2); 163-1 66 © 2005 The Raptor Research Foundation, Inc. Predation of Small Mammals by Rufous-legged Owl, Barn Owl, and Magellanic Horned Owl in Argentinean Patagonia Forests Daniel E. Udrizar Sauthier,^ Analia Andrade, and Ulyses F.J. Pardinas Centro National Patagonico, Casilla de correo Key Words: Magellanic Horned Owl; Bubo magellanicus; Rufous-legged Owl; Strix rufipes; Bam Owl; Tyto alba; diet, stgmodontine rodents. Despite the large number of forest owls in the Neo- tropics, there are few data available on their diets that ^ Email address: dsauthier@cenpat.edu. ar 128, 9120 Puerto Madryn, Chubut, Argentina reflect foraging inside forested habitats. For the southern cone of South America, including Argentina and Chile, only a few contributions have addressed this topic (Mar- tinez and Jaksic 1996, 1997, Ramirez-Llorens 2003, Trejo and Ojeda 2004). The Rufous-legged Owl {Strix rufipes) inhabits dense and old-growth temperate forests in southern Argentina and Chile (Straneck and Vidoz 1995, Martinez and Jaksic 164 Short Communications VoL. 39, No. 2 Table 1. Number and percent frequency of prey items and proportion of scansorial (SC), arboreal (AR), cursorial (C), and fossorial (S) small mammals in the three analyzed owl pellet samples from Argentinean Patagonia forest Percent frequency is in parentheses. Rodents MAGEI.LANIC Horned Owi. Rufous-legged Owl Barn Owt Abrothrix longipilis (C) — 7 (6.9) 6 (7.2) Abrothrix olivaceus (C) — 36 (35.6) 11 (13.3) Chelemys macronyx (S) 9 (27.3) 1 (1.0) — Eligmodontia morgani (C) 4 (12.1) — — Geoxus valdivianus (S) — 2 (2.0) 1 (1-2) Irenomys tarsalis (AR) — 7 (6.9) 6 (7.2) Loxodontomys micropus (C) 20 (60.6) 19 (18.8) 21 (25.3) Oligoryzomys longicaudatus (SC) — 29 (28.7) 38 (45.8) Total 33 101 83 SC and AR (%) — 35.6 48.2 C and S (%) 100 64.4 51.8 1996, 1997) . Diet information about this owl is only known from few localities in Chile (Martinez and Jaksic 1996, 1997, Diaz 1999). The Barn Owl {Tyto alba) and the Magellanic Horned Owl {Bubo magellanicus; del Hoyo et al. 1994) are widespread in South America and nu- merous contributions have described their diets (Pardi- has and Cirignoli 2002 and references therein) . However, these works focused on pellet samples from open areas; consequently, the knowledge of owl predation in forested habitats is extremely limited. The only available data for the Barn Owl in a forested habitat were reported by Trejo and Ojeda (2004) on the basis of an owl pellet sample recovered in Nahuel Huapi Forest (Argentina). Here, we report the first data on the diet of the Rufous- legged Owl in Argentina and the Magellanic Horned Owl in forested landscapes. In addition, we present informa- tion about the small mammals preyed upon by the Barn Owl in Patagonian forests. Study Area and Methods Owl pellet samples were collected in three localities of Neuquen and Rio Negro provinces (Argentina): Magel- lanic Horned Owl (18 pellets), Las Brenas Ranch (39°30'S, 7l°02'W, 1437 m elevation, Neuquen); Rufous- legged Owl (43 pellets), Lago Steffen (41°32'S, 7l°35'W, 730 m, Rio Negro); and Barn Owl (36 pellets), Cacique Foyel Ranch, Alto Rio Villegas (41°35'S, 7l°31'W, 700 m, Rio Negro). Las Brenas Ranch is located in the Pehuen district, Su- bantartic phytogeographic province (Cabrera 1971). This environment is found between 900-1800 m and is dom- inated by forests of pehuen {Araucaria araucana) associ- ated with lenga {Nothofagus pumilio) and coligue cane ( Chusquea culeou) . Lago Steffen and Cacique Foyel Ranch are located in the Caducifolious Forest district (sub- antarctic phytogeographic province; Cabrera 1971) and are characterized by the presence of hire and coihue {No- thofagus antarctica, N. dombeyi)y lenga, and cipres {Austro- cedrus chilensis). Osteological remains were picked apart by hand and prey were identified by comparison with the reference mammal collection of the Centro Nacional Patagonico, Puerto Madryn, Argentina, and identification keys (Pear- son 1995). The taxonomic criteria follow Galliari et al (1996). The minimum number of mammal prey items was estimated by counting mandibles. The Rufous-legged Owl sample was previously analyzed by O. Pearson (field notes 1982) and kindly provided to U. Pardihas for in- clusion here. Results and Discussion We recovered 217 prey items, exclusively sigmodontine rodents, from the examined pellets (Table 1). Only three species were eaten by the Magellanic Horned Owl; most of which was the austral greater mouse {Loxodontomys mi- cropus) . The Rufous-legged Owl and Barn Owl consumed seven and six species, respectively. Both owls preyed upon the same small mammal assemblage, primarily consum- ing the olivaceous field mouse {Abrothrix olivaceus), the long-tailed mouse {Oligoryzomys longicaudatus) , and the austral greater mouse. Diet of Magellanic Horned Owl differed from that of Rufous-legged Owl and Barn Owl in the proportion and kind of small mammal species found in pellets (Table 1). These differences may be at- tributed to the specific forest areas where these respec- tive owls foraged (see Study Area and Methods). Despite the small pellet samples examined, we regis- tered almost exclusively small mammals related to forest- ed or bushy-forested cover types (Pearson 1995). The aus- tral greater mouse, heavily preyed upon by the three owls studied, was a common sigmodontine in dense nire- bushy and coligue cane formations. The same is true for the long-tailed mouse and the olivaceous field mouse, main prey items in the diets of the Rufous-legged Owl and the Barn Owl. In addition, these owls consumed the poorly known (in Argentina) Chilean tree rat {Irenomys tarsalis), a specialist of Nothofagus iore^Xs, (Kelt 1993). The dominance of forest species in the diet of the Rufous- June 2005 Short Communications 165 legged Owl and Barn Owl suggests that these owls hunted within the forest. This fact is reinforced by the absence of those species characteristic of open areas, like the Pa- tagonian leaf-eared mouse (Phyllotis xanthopygus) , the rab- bit rat {Reithrodon auritus), the silky rat {Euneomys chin- chilloides), and the southern cavy {Microcavia australis). The same is true for the Magellanic Horned Owl, al- though we cannot discard the possibility of forest-steppe ecotone hunting by these raptors. The olivaceous field mouse constituted a high propor- tion of the Rufous-legged and Barn Owl diets, being the dominant prey in the diet of the former. These results contrast with those of Trejo and Ojeda (2004) and Mar- tinez and Jacksic (1997), who reported a low consump- tion and an apparent avoidance of this species by the owls, despite its high abundance in the field. Martinez and Jaksic (1997), working with the Rufous-legged Owl in Chile, mentioned the preference of this owl for ar- boreal and scansorial small mammals (e.g., Dromiciops gli- roides, Irenomys tarsalis, Oligoryzomys longicaudatus) over cursorial ones (e.g., Abrothrix olivaceus, A. longipilis). These authors proposed that different escape tactics or detectability of cursorial species in addition to the sit-and- wait predation strategy of the Rufous-legged Owl as an explanation for the high representation of arboreal and scansorial species in the owl’s diet. In contrast, our data showed that, at least in the study area, this owl preyed upon cursorial and fossorial rodents (64.4%) over arbo- real and scansorial species (35.6%; Table 1). This fact could be related to the presence of small open areas (gaps) in the forest, where the cursorial rodents could be more detectable. These results were in agreement with Diaz (1999) who observed a preference of Rufous- legged Owl for terrestrial prey in the Mediterranean sclerophyllous forest of central Chile. In spite of the low number of pellets analyzed for Ma- gellanic Horned Owl, the results suggested that this spe- cies took exclusively cursorial and fossorial rodents asso- ciated with dense vegetative cover. This finding demonstrated the capability of this owl hunting inside the forest, although it may be foraging mostly in open patches. This agrees with Teta et al. (2001) who proposed that horned owls preferentially consumed open-area spe- cies rather than arboreal and scansorial ones, as a result of its sit-and-wait hunting strategy (Marti 1974). The scarce number of forested sites studied in south- ern South America may be attributed mainly to the dif- ficulty of identifying owl roosts coupled with a low pres- ervation rate of pellets in humid forest environments. Depredacion de MamIferos Pequenos por Srmx rvfipes, Tyto alba y Bubo magellanicus en Bosques Patagonicos DE Argentina Resumen. — Se estudiaron los mamiferos depredados por Bubo magellanicus, Strix rufipes y Tyto alba en tres locali- dades boscosas de las provincias de Neuquen y Rio Ne- gro. Sobre un total de 217 presas (exclusivamente roe- dores sigmodontinos) , Oligoryzomys longicaudatus, Loxodontomys micropus y Abrothrix olivaceous fueron los mas consumidos. Estos resultados, sumados a la baja o nula frecuencia de especies de micromamiferos de areas abiertas, sugieren que las tres rapaces estudiadas cazaron en el interior del bosque. Este es el primer estudio de dieta para S. rufipes en Argentina y para B. magellanicus en el bosque patagonico. [Traduccion de los autores] Acknowledgments We want to thank to H. Povedano, D. Glaz, D. Podesta, S. Cirignoli, F. Cremonte, and O. Pearson who collected samples or kindly gave information or other help. This work was supported by Consejo Nacional de Investiga- ciones Cientificas y Tecnicas (CONICET, Argentina) . We appreciate the improvements in English usage made by Stacy Small through the Association of Field Ornitholo- gists’ program of editorial assistance. Literature Cited Cabrera, A. L. 1971. Fitogeografia de la Republica Ar- gentina. Bol. Soc. Argent. Bot. 14:1—43. DEL Hoyo, J., A. Elliott, and J. Sargatal. 1994. Hand- book of the birds of the world, Vol. 2. Lynx Edicions, Barcelona, Spain. Diaz, I. 1999. Food habits of the Rufous-legged Owl {Stnx rufipes) in the Mediterranean sclerophyllous forest of central Chile. J. Raptor Res. 33:260—264. Galliari, C., U.F.J. Pardinas, and F. Coin. 1996. Lista comentada de los mamiferos Argentinos. MastozooL Neotrop. 3:39—61. Kelt, D.A. 1993. Irenomys tarsalis. Mamm. Species 44:7:1— S Marti, C.A. 1974. Feeding ecology of four sympatric owls. Condor 76:45-61. Martinez, D.R. and F.M. Jaksic. 1996. Habitat, relative abundance, and diet of Rufous-legged Owls {Strix ru- fipes King) in temperate forest remnants of southern Chile. Ecoscience 3:259-263. AND . 1997. Selective predation on scanso- rial and arboreal mammals by Rufous-legged Owls {Strix rufipes) in southern Chilean rainforest./. Raptor Res. 31:370-375. Pardinas, U.F.J. and S. Cirignoli. 2002. Bibliografia co- mentada sobre los analisis de egagropilas de aves ra- paces en Argentina. Ornitol. Neotrop. 13:31—59. Pearson, O.P. 1995. Annotated keys for identifying small mammals living in or near Nahuel Huapi National Park or Lanin National Park, southern Argentina. MastozooL Neotrop. 2:99—148. RamIrez Llorens, P.M. 2003. Ecologia trofica de Strigi- formes en Argentina: Pulsatrix perspicillata (Lechuzon Mocho Grande). M.S. thesis, Univ. Nacional Buenos Aires, Buenos Aires, Argentina. Straneck, R.J and F. Vidoz. 1995. Sobre el estado taxono- mico de Strix rufipes (King) y de Strix chacoensis (Gher- rie and Reichenberger) . Not. Faunist. 74:1—5. 166 Short Communications VoL. 39, No. 2 Teta, R, C. Panti, a. Andrade, and A. Perez. 2001. Am- plitud y composicion de la dieta de Bubo virginianus (Aves, Strigiformes, Strigidae) en la Patagonia noroc- cidental Argentina. Bol. Soc. Biol. Concep. 72:125-132. Trejo, A. and V. Ojeda. 2004. Diet of Barn Owl {Tyto alba) in forested habitats of northwestern Argentine Patagonia. Ornitol. Neotrop. 15:1-5. Received 30 July 2004; accepted 6 December 2004 Associate Editor: Michael I. Goldstein J. Raptor Res. 39(2): 166-1 68 © 2005 The Raptor Research Foundation, Inc. Changes in Site Occupancy and Nesting Performance of Peregrine Falcons in Colorado, 1963-2004 James H. Enderson^ Department of Biology, Colorado College, 14 East Cache la Poudre Street, Colorado Springs, CO 80903 U.S.A. Keywords: Peregrine Falcon-, Falco peregrinus; occupancy rate, population change, productivity, reproductive success; Col- orado. In 1965, Professor Joe Hickey held a conference of more than 50 people at the University of Wisconsin to review marked declines of several species of bird-eating and fish-eating raptors, particularly the Peregrine Falcon {Falco peregrinus; Hickey 1969). By that time, peregrines were extirpated in parts of Europe and the eastern Unit- ed States and were greatly reduced in several other re- gions. The purpose of this paper is to compare territory occupancy and nesting performance in Colorado in the 1960s and 1970s, when the population was in decline, with falcon activities and occupancy at most of the same cliffs in 2004. In 1964, a broad survey in the Rocky Mountain region was accomplished (Enderson 1965) and adult pairs in Colorado were present at five (33%) of 15 cliffs visited. Single peregrines were seen at two other sites, and eight were apparently vacant; only four young were seen at these sites. Another survey was done in Colorado in 1973 and the results were included in a broader regional re- port (Enderson and Craig 1974); eight (44%) pairs were found at 18 sites and only two young were seen. These published reports included records from only a single year. Because additional unpublished data are available for Colorado in other years during those early periods, it would be useful to include them in a wider analysis and to contrast those historical results with a recent survey that I completed in 2004. Methods In the period from 1963—65, 1 visited cliffs in Colorado where peregrines had been reported in the literature or ^ Email address: jenderson@coloradocollege.edu gleaned from correspondence with interested people, of- ten falconers. I also visited several sites in the 1973-75 period that included most of the territories from the ear- lier survey plus two additional peregrine cliffs found m the interim. The sites visited in 2004 included 15 sites selected arbitrarily from those included in the earlier sur- vey periods. In all, 21 different sites were surveyed at least once in the three periods. In 2004, 1 also visited an additional 32 sites found since 1975. I did not include the latter group in the activity and occupancy comparisons between periods. The addi- tional 32 cliffs were checked because they were logisti- cally accessible while I visited historical sites, or were re- ported by other observers. They were selected without regard for past occupancy or reproductive performance Because of small sample size in any one year, I pooled data within the first two periods. As a result, data for as many as three seasons within a period were tallied for some locations. For this analysis, reproductive perfor- mance by pairs between years at the same site was con- sidered independent, but the validity of that notion was not tested. A spotting scope or binoculars on a tripod or window mount were used to discover falcons on cliffs. In the two earlier periods, I usually walked the tops of the cliffs in addition to the distant viewing. Unless weather inter- fered, I usually searched the cliff for at least 4 hr or until peregrines were seen. Incubation, brooding, or food ex- changes usually occurred in that span of time. When fresh excrement was seen, or poor weather interrupted, searching was extended or often repeated on another day. Sites were excluded from the analysis if I assessed that sampling was inadequate to determine occupancy. I believe the effectiveness of observations was generally similar between periods, except that in 1963-65 when I was less familiar with the cliffs and peregrine activities at the cliffs. Sites were usually visited once or twice to verify occupancy, and usually once to count young. This pattern did not change substantially between periods, except that in the first two periods I often checked, based on incu- bation behavior, to see if females had laid eggs as well. June 2005 Short Communications 167 Table 1. Occupancy and productivity of Peregrine Falcons at 21 cliffs during three survey periods from 1963-2004 in Colorado. \Tars 1963-65 1973-75 2004 Number of cliffs checked in period 18 18 15 Number of site visits in period 34 47 15 Number of pairs 16 19 13 Occupancy (%) 47 40 87 Number of pairs with known reproductive outcome 9 19 10 Number of young recorded 11 12 21 Number of young per pair 1.2 0.7 2.1 Productivity^ was defined as the number of large young (ca. 30 d or older) counted for all pairs on territory. The estimate is a minimum because some young on the ledge or recently fledged may have been overlooked. When a site was not checked at the time large young should have been present, it was not included in productivity calculations. Results In the 1963-65 period, 1 determined occupancy by per- egrines during 34 visits at 18 nesting locations (Table 1). In that period pairs were present in 16 of 34 instances (47% occupancy). In all, productivity was determined for nine pairs and they produced 11 young (1.2 young/pair). In 1973-75, J. Craig and others assisted me in visits to 18 sites. In all, occupancy was ascertained 47 times over the 3 yr, 19 pairs were found holding territories (40% occu- pancy) and 12 large young were seen (0.63 young per pair). For comparison, in 2004, 1 visited 15 of these same sites (including 14 from the 1963—65 sample); 13 had pairs (87% occupancy). The outcome of nesting was checked for 10 pairs; 21 large nestlings or flying young were found (2.1 young per pair). Additionally, the expanded 2004 survey included other sites discovered in the last three decades. In all, 47 sites (among about 160 places where peregrines have been reported to nest in Colorado in the last decade) were checked in 2004. Pairs were found at 45 sites (96% oc- cupancy) and 30 pairs of which I checked for production, reared 54 large young (1.80 young/pair). Discussion How severe was the DDT/DDE induced decline in Col- orado in the 1960s and 1970s? To what extent have per- egrines become reestablished at sites vacated in those earlier periods? The biology and management of pere- grines in Colorado, 1973-2001, have recently been sum- marized (Craig and Enderson 2004). From 1994-2001, annual occupancy rate was in the range of 82-89%, prob- ably typical of the species. However, in 1963-65 and 1973-75 occupancy rates were 47% and 40% (Table 1), respectively. If these low rates represented the general condition, the population was depressed to about one- half in the early periods compared to the last decade. Indeed, the low occupancy rate persisted for some time. In 1977 we visited 29 sites in Colorado and 11 pairs were found and in 1985, only six (21%) of those 29 sites were occupied by pairs (Enderson et al. 1988). In 2004, 96% of the sites surveyed (N = 47) had pairs of falcons. Documentation of further population increase and ex- pansion of nesting pairs will depend on colonization (and our discovery) of sites not known to have held per- egrines in the past. It is tempting to speculate on the size of the pre-DDT peregrine population in Colorado and on the potential for further increase. A coarse estimate of total pairs at recently-used cliffs can be obtained by multiplying the approximate number of cliffs used in the last decade by the mean recently observed for occupancy rate (160 cliffs X 0.85 occupancy rate = 136 pairs). Craig and Enderson (2004) also estimated that there were another 250-300 potentially useable cliffs in the state where the falcons could nest without undue crowding. Because there is much suitable habitat in Colorado, I suggest that the total population might eventually double or even triple in size The current population plus this potential projected in- crease could also reflect the pre-DDT population size (250-400 pairs). Productivity in 1963-65 was 1.2 young per pair on ter- ritory, and 0.7 in 1973-75. After that time, augmentation of broods from captive stocks with 229 peregrines through 1990 obscured natural reproduction. However, in the late 1970s and early 1980s, 22 non-augmented pairs averaged 1.2 young per pair (Craig and Enderson 2004). In 1995-2001, productivity averaged 1.7 young per pair (range 1.3-2.1). The current reproductive rate of 1.8/pair seen in 2004 is a marked improvement com- pared to the mid-1970s. Cambios en la Ocupacion de Sitios y Desempeno en la Nidificacion de Falco peregrinus en Colorado entre 1963-2004 Resumen. — El halcon Falco peregrinus desaparecio de mu- chos de los acantilados que utilizaban para nidificar en 168 Short Communications VoL. 39, No. 2 las Montanas Rocosas en las decadas de 1960 y 1970. En el ano 2004 revise 15 de los 21 acantilados utilizados por Falco peregrinus en el pasado en Colorado para determi- nar los cambios en la ocupacion y en las tasas de produc- tividad. En el ano 2004, la tasa de ocupacion por pareja fue de un 87% en comparacion con un 47% y un 40% entre 1963-65 y 1973-75, respectivamente. La tasa re- productiva basada en todas las parejas con territorios fue de 2.1 juveniles/ pareja, en comparacion con una tasa de 1.2 y 0.7 para los periodos anteriores, respectivamente. Se estima que 136 parejas nidifican en unos 160 acanti- lados donde F. peregrinus estuvo presente en la ultima de- cada, pero el numero real es seguramente mayor y pod- ria aumentar a 250-400 parejas dada la estimacion de la disponibilidad de habitat apropiado. [Traduccion del equipo editorial] Acknowledgments Colorado College and the Colorado Division of Wild- life funded travel in 2004. Myron Chase and Jason Ferrell of the National Park Service, and Brent Bibles, graciously provided occupancy information from eight sites. Tom Cade, Brian Millsap, and an anonymous reviewer helped importantly with revision of the original manuscript. Literature Cited Craig, G.R. and J.H. Enderson. 2004. Peregrine Falcon biology and management in Colorado 1973-2001 Tech. Publ. No. 43. Colorado Division of Wildlife, Fort Collins, CO U.S.A. Enderson, J.H. 1965. A breeding and migration survey of the Peregrine Falcon. Wilson Bull. 77:327-339. ANDJ. Craig. 1974. Status of the Peregrine Falcon in the Rocky Mountains in 1973. Auk 91:727-736. , , AND W.A. Burnham. 1988. Status of per- egrines in the Rocky Mountains and Colorado Pla- teau. Pages 83-86 in T.J. Cade, J.H. Enderson, C.G Thelander, and C.M. White [Eds.], Peregrine Falcon populations: their management and recovery. The Peregrine Fund Inc., Boise, ID U.S.A. Hickey, J.J. (Ed.). 1969. Peregrine Falcon populations: their biology and decline. Univ. Wisconsin Press, Mad- ison, WI U.S.A. Received 10 September 2004; accepted 28 March 2005 /. Raptor Res. 39(2): 168-1 73 © 2005 The Raptor Research Foundation, Inc. Analysis of Reservoir Selection by Wintering Ospre\s {Pandion haliaetus haliaetus) in Andalusia, Spain: A Potential Tool for Reintroduction Eva Casado^ and Miguel Ferrer Department of Biodiversity Conservation, Estacion Biologica de Donana, CSIC, Pabellon del Peru, Avda. Maria Luisa s/n, 41013 Sevilla, Spain Key Words: Osprey, Pandion haliaetus; reintroduction', pre- dictive model', reservoirs:, southern Spain', fish production. The extant Mediterranean Osprey {Pandion haliaetus haliaetus) breeding population is largely fragmented in Morocco, Corsica, and on a few islands from the Balearic and Canary archipelagos, which support isolated-rem- nant populations (Gonzalez et al. 1992, Thibault et al. 1996). The disappearance of the Osprey as breeding bird in the coastal region of mainland Spain was due to the loss of suitable nesting sites resulting from the develop- ment of a tourist infrastructure (Gonzalez et al. 1992) and human persecution (especially by theft of eggs or chicks) . Ospreys have been extinct as a breeding species in continental Spain since the 1980s (Gonzalez et al. 1 Email address: casado@ebd.csic.es 1992). However, Ospreys still winter in some parts of Spain. Historically, Andalusia was an important breeding area for this species in Spain, and currently is an impor- tant wintering and stopover area for migrant birds (Os- terloff 1977, Saurola 1994). Reservoirs are occupied extensively by breeding Os- preys in most of their range, but they are a relatively new ecosystem in Spain. Reservoirs of 150 ha or more covered 25 500 ha (0.3% of Andalusia) during the 1960s, but now cover twice this area (MOPU 1991). Construction of ar- tificial impoundments may have enhanced the spread of Osprey populations in other areas due to habitat creation (Van Daele and Van Daele 1982, Houghton and Rymon 1997). Reservoirs often provide foraging advantages over rivers and lakes because they are shallow and open-water areas, vfith reduced turbidity that improve the detectibil- ity of prey (Vana-Miller 1987 and references therein). An June 2005 Short Communications 169 Osprey reintroduction project involving recently-created reservoirs was initiated in Andalusia in 2003 to re-estab- lish a breeding population on the Iberian Peninsula (Ca- sado and Ferrer 2004). Requirements of wintering Osprey involve primarily abundance of food supplies (Newton 1979, Prevost 1982), but physical structure of habitat may also affect accessibility to prey (Moore et, al. 1993 and references therein). Fish production or standing crop depends on reservoir features such as mean depth, surface area, shoreline development, water level fluctuation, age, and storage ratio, factors that also affect availability of prey to Osprey (Jenkins 1970, 1976, Sancho Royo and Granados 1988). Also, the quality of a reservoir for a wintering Os- prey may be affected by human activity or by the avail- ability of hunting and resting perches. We surveyed struc- tural characteristics, human disturbance, and availability of perches on reservoirs to determine factors affecting occupation by Osprey. Study Area Andalusia is the southernmost Spanish autonomous re- gion, with 64 man-made freshwater impoundments >150 ha in size, covering a total of 50 183 ha and have a storing capacity of 9403 hm^ of water. Reservoirs studied were situated among mountains near the seacoast (Betic Chain) and in valleys of the Guadalquivir River and its tributaries. Climate of Andalusia is mild, with maximum and minimum temperatures in winter 20.7°C and — 2°C, respectively (Instituto Nacionalde Meterologia 2004) . Barbel {Barbus spp.), Iberian Nose {Chondrostoma spp.), and Carp {Cyprinus carpio) are the primary prey species of Osprey in Spanish reservoirs (Sancho Royo and Gra- nados 1988, Gil Sanchez 1995, Lekuona 1998). Predom- inant vegetation was crop fields in valleys, and pines {Pi- nus spp.) and cork oak {Quercus suber) scrub in the mountains. Eucalyptus {Eucalyptus sp.) have been planted around reservoirs with recreational facilities. Methods Selection of Reservoirs and Survey of Wintering Os- prey. Historical data about presence of wintering Osprey from published sources and from official state reports, from banding centers, wildlife-recuperation centers, and unpublished observations of naturalists and researchers were used to identify occupied reservoirs in our study area. Records of Osprey sightings in the study area be- tween November and mid-March over the winters of 1984-85 to winter 1996-97 were collected. Systematic surveys on reservoirs were conducted in the winter of 1997-98 (mid-November-mid-March) to inves- tigate the occupancy of each reservoir by Ospreys. Two 45-min Osprey searches were conducted with binoculars (8 X 30) and spotting scope (20 X 60) from different observation points at each reservoir. Searches were con- ducted by the same observer between 0800—2000 H (so- lar time) . Reservoir searches were conducted with similar weather conditions, avoiding those days with precipita- tion or high winds, both in the morning and in the even- ing. One observation point was always at the head of the dam, and the other at the opposite end of the reservoir. Surveying all of Andalusian reservoirs was not possible due to the limited time of the study period (4 mo m winter). Thus, we had to reduce the sample size to 20 reservoirs which were selected on the basis of previously reported sight records of Osprey. First, 10 reservoirs with previous historical records of wintering Ospreys were chosen. Then, for comparison, we randomly selected 10 reservoirs without previous records of wintering Ospreys. Habitat Measurements. We collected data on 17 vari- ables possibly affecting both abundance and availability of fish to Ospreys (Table 1 ) . We assumed that shorelines of reservoirs were more likely to be affected by human activity than reservoir centers. Thus, distances from res- ervoir shoreline to the nearest distribution and transmis- sion power line, nearest paved road, and nearest urban center were measured to provide estimators of human activity. Other habitat features included in analyses were dis- tance from a reservoir to nearest reservoir >150 ha, dis- tance to nearest coast, and number of arms (thin prolon- gation of water) of over 100 m length X reservoir surface^^ (“number of arms”). Reservoirs on flat terrain usually have few arms, but more large tails than those situated in mountain valleys. Number of reservoir arms was an index of surrounding topographical relief (i.e., higher index values indicated more relief) . Distances, number of arras, and shoreline length were measured on 1:50 000 topographic maps prepared by the Spanish Army Cartographic Service, using a ruler and a digital curvimeter, respectively. Percent of tree cover within a 20-m wide band around the reservoir edge was considered to be an index of perch availability and was obtained from land-use maps of the Spanish Ministry of Agriculture, using SYGMAS- CAN pro 4.0 image analysis software (Fox and Ulrich 1995). Water-exchange rate (the percentage of the difference between the water entry and the water exit, in relation to the mean volume of the reservoir; Table 1) between March 1997 and March 1998 was calculated from data coming from Guadalquivir and south hydrographic con- federacies (Spanish Environment Ministry) . Data Analyses. Statistical analyses were conducted us- ing STATISTICA (1986). Variables were transformed (square-root and log [1 + square root]) when necessary to achieve a normal distribution. When variables could not be normalized, nonparametric statistics were used for comparisons. Correlation among variables was evaluated by a factor analysis and covariates were removed. The remaining variables were included in a discriminant func- tion analysis, which was employed to determine the res- ervoir features that were associated with Osprey pres- ence. Variables for the discriminant analysis were introduced three at a time due to the small sample size. One variable from each of the three factors obtained was included in every entry group in order to avoid co-rela- tions; thus several combinations of variables were ana- lyzed with the discriminant function approach. The for- ward stepwise method was used and significance was set at P < 0.05. As a measure of the model’s ability to predict Osprey occupation of reservoirs, validation was carried out using 170 Short Communications VoL. 39, No. 2 Table 1 . Comparison of habitat variables between occupied and unoccupied surveyed reservoirs. Occupied Unoccupied N Mean N Mean pa Meters above sea level 7 10.53 13 11.79 0.362 Years since reservoir construction to winter 1997-98 7 30.29 13 33.85 0.662 Mean surface area (ha) 7 1476.53 13 471.86 0.166 Mean water depth (m) Distance from reservoir shoreline to nearest >150 ha 7 18.14 12 13.31 0.397 reservoir (km) Distance from reservoir shoreline to nearest paved road 7 88.31 13 98.94 0.403 (m) Distance from reservoir shoreline to nearest urban center 7 35.71 13 42.31 0.591 (km) 7 35.30 13 56.39 0.048 Distance from reservoir shoreline to nearest distribution power line (m) 7 24.71 13 5.27 0.172 Distance from reservoir shoreline to nearest coast (km) Number of arms longer than 100 m in length/reservoir 7 169.30 13 228.92 0.143 surface (ha) 7 0.16 13 0.23 0.030 Shoreline (km) Water exchanged, which was calculated by: (entry hm^ — 7 55.41 13 33.86 0.285 exit hm^) X 100/ (mean volume hm^) Shoreline development. Ratio of the shoreline length to 7 0.084 13 0.024 0.063 the circumference of a circle with an area equal to that of the reservoir. Calculated through the function L/2 X V(n X A) , being L = shoreline length in km, A = area of reservoir (ha) 7 0.40 13 0.43 0.781 Depth where a Secchi disk of 20 X 20 cm was not visible Distance from reservoir shoreline to nearest pole of trans- 7 68.44 13 78.35 0.968 mission power line (m) 7 27.01 13 34.27 0.838 Percent of shoreline length occupied for dense canopy 7 3.06 8 1.92 0.270 Trophic state = 10 (6-logD); D = Secchi disk depth 7 -1.96 13 2.35 0.405 Probability that occupied and unoccupied distributions were different was obtained by the Mann-Whitney Ctest. Probability values in bold indicate statistical significance. 42 more Andalusian reservoirs with previous records of occupation by Osprey. Results Twenty reservoirs covering 16 482 ha were included in the analyses (Table 2); these represented 31.25% of res- ervoirs by number and 33% by surface in Andalusia. Ac- cording to major operational function, one reservoir was classified as a hydropower reservoir, six as irrigation, nine as municipal water supply, three as recreational use, and one for mining use (MOPU 1991). Observations of wintering Ospreys in Spain increased from 15 individuals in winter 1984—85 to 47 in winter 1996-97. Of 522 sightings of wintering Osprey in Spain, 400 were from Andalusia. Every Andalusian estuary and marsh was occupied by wintering Ospreys, and 19 out of 64 Andalusian reservoirs were occupied. Seven of the 20 sampled reservoirs were occupied by Osprey. Altitude (square-root transformed) of reservoirs, age, water-exchange rates (log transformed), shoreline devel- opment, water transparency, distances from reservoir shoreline to the nearest distribution and transmission power line, to the nearest paved road, to the nearest ur- ban center, to the nearest reservoir, to the nearest coastal point, square root of number of arms, and percent of tree cover were the non-related variables that were con- sidered for further analysis. Shoreline length (km) , mean surface area (ha), concentration of dissolved organic ma- terial in water (trophic state), and water depth (m) were redundant and thus, not included in further analysis. Fac- tor analysis provided three principal factors. First, one contributed to 23.77% of total variance and showed the highest factor loading with distance to nearest distribu- tion power line, to nearest reservoir, forest cover, and the transformed index of number of arms and altitude. The second factor explained 18.82% of total variance and was related to water transparency, distance to nearest urban center, and log transformed water-exchange rates. The third factor, which explains 15.81% of total variance, and loaded with age of reservoir, shoreline development. June 2005 Short Communications 171 Table 2. Probability of occupation and presence of Os- prey on surveyed reservoirs in Andalusia, Spain. Reservoirs Osprey Presence Occupation Probability Agrio No 0.26 Almoddvar No 0.33 Arcos Yes 0.70 Barbate Yes 0.55 Bornos Yes 0.70 Gala No 0.10 Celemin No 0.03 Charco Redondo No 0.33 Cordobilla No 0.55 El limonero No 0.43 Gergal No 0.33 Guadalcacin Yes 0.55 Guadarranque Yes 0.43 La Concepcion Yes 0.12 La Minilla No 0.26 Los Hurones No 0.33 Pintado No 0.20 Retortillo No 0.12 Torre del Aguila No 0.43 Zahara Yes 0.33 nearest distances to transmission power line, to nearest coastal point, and to nearest paved road. After analyzing several combinations of variables, the simplest discriminant function with the highest correct classification was: Ln (P/1 - P) = -6.59259 + 54.30499 (number of arms), which correctly classified 80% of cases: 92.3% of unoccupied reservoirs and 57.1% of occupied (Wilks’s Lambda = 0.8, T) jg = 4.49, P < 0.04). The model clearly discriminated occupied versus unoccupied reservoirs on the basis of number of arms of the reservoir. The reser- voirs with the highest probability of occupancy were Ar- cos (P = 0.70), Bornos (P = 0.70), Barbate (P = 0.55) and Guadalcacin (P = 0.55). To compare univariate differences between occupied and unoccupied reservoirs Mann-Whitney G-test was used (Table 1) and these indicated a significant differences in number of arms and distance to nearest urban center. Occupied reservoirs had a fewer number of arms and a shorter distance to nearest urban center. Validation. Validation with the test set of reservoirs in- dicated that the discriminate function classified Osprey occupancy well. Only nine (21.4%) of 42 non-surveyed reservoirs were misclassified. Discussion The variable number of arms was negatively correlated with Osprey presence both in univariate test and in dis- criminant function analysis. High topographic relief, re- flected by this variable, seemed to be avoided by Osprey Reservoirs with a circular shape are shallower and have higher exchange between the entry and exit of water These factors are associated with higher productivity (Jenkins 1976, Ryder 1982), providing a better food sup- ply to fish-eating birds than those reservoirs located with- in the mountains. Volume fluctuations enhance nutrient movement, and a greater area of shallow water allows a high production of macrophytes, and consequently, of other organisms including fish. High fish abundance may enhance Osprey foraging efficiency (Flook and Forbes 1983). Deeper water inhibits photosynthesis, depressing primary production (Ryder 1982). Rawson (1952) dem- onstrated a negative correlation between mean depth and long-term fish production in reservoirs. The same is true for Andalusian reservoirs, where productivity is largely determined by mean depth and the level of eu- trophy (Sancho Royo and Granados 1988). Sancho Royo and Granados (1988) investigated the re- lationship between fish standing crop and characteristics of seven Andalusian reservoirs, six of which were includ- ed in the present study. These authors found that the largest and most shallow reservoirs supported the highest fish density (fish/m^). Tolerance of human activity depends on timing, inten- sity, and frequency of activity and degree of habituation to such activities (Odsjo and Sondell 1976, Swenson 1979, Van Daele and Van Daele 1982, Levenson and Kop- lin 1984). Occupied reservoirs were closer to an urban center than the unoccupied reservoirs. However, this may be due to topographical constraints when towns were es- tablished. No other human-activity indicator was related to occupation by Ospreys, but this result should be taken with caution because we were using data on wintering Ospreys that may respond to human activities in a differ- ent way than breeding Ospreys. Wintering Ospreys in Spanish reservoirs do not appear to choose reservoirs according to foraging perch avail- ability, as also was observed in Senegambia, Africa (Pre- vost 1982). Areas with high tree coverage will not neces- sarily be occupied by Ospreys if habitat does not provide an adequate food supply. In Spain, Ospreys seem to choose reservoirs based on fish productivity over foraging perch availability or hu- man activity. In our analysis, the number of reservoir arms relative to total surface area was identified as a key variable both because it is easy to derive and its relation to reservoir productivity. Management Implications. Using the discriminant function analysis, we derived an estimate of the proba- bility that one individual would be observed on a reser- voir under the conditions that existed during this study This model developed for wintering Ospreys in Andalusia was focused on specific habitat characteristics (e.g., prey and perch availability, human disturbance). Combining this knowledge with an analysis of the status and devel- opment plans for the Andalusian reservoirs, environmen- 172 Short Communications VoL. 39, No. 2 tal management agencies would have an important tool to reduce the threats to migratory and wintering Ospreys. Results of the present study should be helpful for the Osprey reintroduction project in Spain. One advantage of our model is that the data needed are easily measured from maps, aerial photographs or Geographic Informa- tion Systems. The Osprey reintroduction project in Spain has two interesting aspects. First, the program will involve use of reservoirs as release areas that are relatively new man-made habitats. To consider man-induced changes in the environment as new habitat opportunities for endan- gered species represents a novel approach in conserva- tion. Second, this program can use information about wintering areas as an additional indicator of good-quality habitat for selection of the release areas. Analisis de ia Seleccion de Reservorios de Agua for PaNDION HAUAETUS HAIJAETUS DURANTE EL InVIERNO EN AN- DALUClA, ESPANA: UNA HeRRAMIENTA POTENCIAL PARA LA Reintroduccion de Poblaciones Resumen. — Se considera que Pandion haliaetus es un ave amenazada en el Mediterraneo, donde se encuentra solo en poblaciones pequehas y fragmentadas. Las pobla- ciones reproductivas de esta especie en el area continen- tal de Espaha se extinguieron desde los aiios ochenta. Sin embargo, Espaha arm representa un area importante de invernada y de escala para las poblaciones migratorias europeas de P. haliaetus. En este estudio desarrollamos un modelo de seleccion de habitat, en parte para evaluar la factibilidad de reintroducir a esta especie en Espaha. Especificamente, estudiamos la ocupacion de reservorios de agua por parte de P. haliaetus durante el invierno en Andalucia (sur de Espaha). Comparando caracteristicas del habitat de reservorios ocupados y no ocupados, em- pleamos un analisis discriminante para desarrollar un modelo para predecir la seleccion de reservorios por esta especie. La funcion discriminante clasifico correcta- mente el 80% de los reservorios como ocupados o no ocupados. Los reservorios con mayor probabilidad de es- tar ocupados por P. haliaetus presentaron una forma mas circular, menor profundidad general y alta abundancia de pcces, y se cncontraban a nivel del terreno. Este mo- delo de prediccion podria ser util para identificar los re- servorios optimos para la reintroduccion de individuos mdificantes. [Traduccion del equipo editorial] Acknowledgments This study was supported by Consejeria de Medio Am- biente of Andalucia and Consejo Superior de Investiga- ciones Cientificas for a predoctoral grant. We thank Ja- vier Balbontin for as.sistance in the field, and also Jordi Figuerola and Luis Palma for reviewing the manuscript critically. To Lourdes Encina and Mikael Hake for their comments and to Guadalquivir and Southern Hydro- graphic Confederacies for providing data and informa- tion. We are indebted greatly to the naturalists who gave us insights about looking for Ospreys. Comments of A. Harmata, P. Martin, and M. Pandolfi improved the man- uscript greatly. Literature Cited Casado, E. and M. Ferrer. 2004. Osprey {Pandion haliae- tus) reintroduction project in Andalusia: annual re- port. Dohana Biological Station, Seville, Spain. Fi.OOK, D.R. and L.S. Forbes. 1983. Ospreys and water management at Creston, British Columbia. Pages 281-286 mD.M. Bird [Ed.], Biology and management of Bald Eagles and Ospreys. Harpell Press, Ste. Anne de Bellevue, Quebec, Canada. Fox, E. and C.G. Ulrich. 1995. SygmaScan and SygmaScan Pro user’s manual. Jandel Corporation, San Rafael, CA U.S.A. Gil Sanchez, J.M. 1995. Alimentacion y seleccion de pre- sa por el aguila pescadora {Pandion haliaetus) en el embalse del Cubillas. 42:133-138. Gonzalez, G., J.M. Santiago, and L. Fernandez. 1992. El aguila pescadora {Pandion haliaetus) en Espaha: censo, reproduccion y conservacion. Institute para la Conservacion de la Naturaleza, Madrid, Spain. Houghton, L.M. and L.M. Rymon. 1997. Nesting distri- bution and population status of U.S. Ospreys 1994./. Raptor Res. 31:44—53. Instituto Nacionalde Meterologia. 2004. http.// www.inm.es Jenkins, R. 1970. The influence of engineering design and operation and other environmental factors on reservoir fishery resources. /. Am. Water Res. Assoc. 16: 110-119. .1976. Prediction of fish production in Oklahoma reservoirs on the basis of environmental variables. Ann. Okla. Acad. Sci. 5:11-20. Lekuona, J.M. 1998. Distribucion, fenologia y ecologia trofica del aguila pescadora {Pandion haliaetus) en Na- varra durante el periodo no reproductor. Anu. Ormt. Navarra 3:29-34. Levenson, H. and J.R Koplin. 1984. Effects of human activity on productivity of nesting Ospreys. /. Wildl. Manag. 48:1374-1377. Ministry OF the Road Works (Mopu). 1991. Inventario de presas espaholas. Publicaciones de la Direccion General de Obras Hidraulicas, Madrid, Spain. Moore, R.E, S.A. Gauthreaux, Jr., P. Kerlinger, and T.R. Simons. 1993. Stopover habitat: management im- plications and guidelines. Pages 58-69 in D. Finch and P. Stangel [Eds.] , Status and management of neo- tropical migratory birds. USDA Forest Service, Rocky Mountain Forest Experimental Station, Ft. Collins, CO U.S.A. Newton, I. 1979. Population ecology of raptors. T. & A.D. Poyser, London, U.K. Odsjo, J. and J. Sondell. 1976. Reproductive success in Ospreys Pandion haliaetus in southern and central Sweden, 1971-73. Ornis Scand. 7:71-84. June 2005 Short Communications 173 OSTERLOFF, S. 1977. Migration, wintering areas, and site tenacity of the European Osprey, Pandion haliaetus ha- liaetus (L.). Ornis Scand. 8:60-78. Prevost, Y. 1982. The wintering ecology of Osprey in Se- negambia. Ph.D. dissertation, Univ. Edinburgh, Ed- inburgh, U.K. Rawson, D.S. 1952. Mean depth and the fish production of large lakes. Ecology 33:513-521. Ryder, R.A. 1982. The morphoedaphic index-use, abuse, and fundamental concepts. Trans. Am. Fish. Soc. Ill: 154-164. Sancho Royo, F. and C. Granados. 1988. La pesca en los embalses andaluces. Instituto de Desarrollo Re- gional de la Univ. Sevilla, Sevilla, Spain. Saurola, P. 1994. African non-breeding areas of Fenno- scandian Ospreys {Pandion haliaetus). Ostrich Qb\\'2!7- 136. STATISTICA. 1986. Version 5. StatSoft, Inc., Tulsa, OK U.S.A. Swenson, J.E. 1979. The relationship between prey spe- cies ecology and dive success in Ospreys. Auk 96:408— 412. Thibault, J.-C., R. Triay, P.L. Beaubrun, D. Poukhalfa, J.-M. Dominici, and a. Torre. 1996. Osprey {Pandion haliaetus) in the Mediterranean: characteristics of a resident population with a patchy distribution. Pages 135-144 in]. Muntaner andj. Mayol [Eds.], Ecologia y conservacion de las rapaces Mediterraneas. SEO, Madrid, Spain. Van Daele, LJ and H.A. Van Daele. 1982. Factors af- fecting the productivity of Ospreys nesting in west- central Idaho. 84:292-299. Vana-Miller, S.L. 1987. Habitat suitability index models. Osprey. U.S. Fish and Wildlife Service, Washington, DC U.S.A. Received 1 1 February 2004; accepted 5 March 2005 J. Raptor Res. 39(2):173-179 © 2005 The Raptor Research Foundation, Inc. Introduced Animals in the Diets of the Ogasawara Buzzard, an Endemic Insular Raptor in the Pacific Ocean Yuka Kato^ and Tadashi Suzuki Department of Biological Sciences, Tokyo Metropolitan University, Minami-ohsawa Tl, Hachi-ohji, Tokyo 192-0397 Japan Key Words: Common Buzzard; Buteo buteo toyoshimai; insular subspecies, diet, introduced animals; Bonin Islands; Ogasawara Islands. The Ogasawara buzzard {Buteo buteo toyoshimai) is an insular subspecies of the Common Buzzard {B. buteo ) , en- demic to the Ogasawara (Bonin) Islands, in the Pacific Ocean (Momiyama 1927, Ornithological Society of Japan 2002). This hawk may be distinguished from a closely- related subspecies, the Japanese Common Buzzard {B. buteo japonicus), by its less brown or lighter plumage, a longer bill, and shorter wings and tarsi (Momiyama 1927). The distribution of the Ogasawara buzzard is very restricted, and this hawk is classified as endangered in Japan (Ministry of Environment 2002) . Recently, Suzuki and Kato (2000) reported on the abundance of the Oga- sawara buzzard on Chichijima, and estimated that less than 85 pairs of this subspecies inhabited the Ogasawara Islands. Insular raptors are likely to be sensitive to environmen- ^ Email address: yukak@bh.mdn.or.jp tal changes as are many other insular predators (e.g.. Cade and Jones 1993). Therefore, ecological information including dietary data are needed to develop conserva- tion strategies for the population. However, little infor- mation on the food habits of the Ogasawara buzzard are currently available. Many researchers have investigated the diet of the con- tinental subspecies of the Common Buzzard, especially in Europe. As a result, the Common Buzzard is well known to capture and consume various kinds of inver- tebrates and small- to medium-sized vertebrates. Com- mon prey include reptiles, birds, and rodents depending on environmental conditions (e.g.. Cramp and Simmons 1980, del Hoyo et al. 1994, Jedrzejewski et al. 1994, Swann and Etheridge 1995, Reif et al. 2001, Sergio et al. 2002 ). The native fauna of the Ogasawara (Bonin) Islands was originally characterized by low species richness because of the island’s volcanic origin, and its small size and rel- atively great distance from the other islands and main- land of Japan (Tsuyama and Asami 1970). Human colo- nization of the islands began in the 1830s. After that 174 Short Communications VoL. 39, No. 2 time, it was likely that a number of exotic animals were introduced intentionally or unintentionally. Some of these introduced species have become very common. Conversely, many native animals have become extinct. For example, about a third of breeding land birds have gone extinct (Ornithological Society of Japan 2002). As a result, the present fauna of the Ogasawara Islands is composed of a mix of introduced exotic species and some surviving native species. Several authors have incidentally reported that the Ogasawara buzzard consumed rodents, some birds, green anoles {Anolis carolinmsis) , and marine toads {Bufo mari- nus; Takano et al. 1970, Wild Bird Society of Japan 1975, Nakane et al. 1980, Ueda and deForest 1988, Suzuki and Chiba 1995, Kawakami 2000). Villagers also reported that this buzzard preyed upon domestic fowls (Suzuki 1982) and migrant egrets {Egretta sp. and Bulbulcus ibis; Y. lida pers. comm.). Despite these anecdotal reports, the food habits of the Ogasawara huzzard have not received any quantitative study. Here, we provide quantitative data on the diet of the Ogasawara buzzard, derived from a broader ecological study of this endemic insular raptor. We also describe some of the basic ecology of this species, information that IS necessary to provide for its conservation. Methods Study Area. The Ogasawara (Bonin) Islands are com- posed of three island groups and situated in the Pacific Ocean, ca. 1000 km south of the mainland of Japan (Fig. 1). It has a volcanic origin and subtropical climates, with an annual mean temperature and rainfall of 23.0°C and 1333 mm, respectively (means from 1986-99 at Chichi- jima Observatory). Chichijima (ca. 27°N, 142°E), where we studied the di- ets of the Ogasawara buzzard, is the largest island of the Ogasawara Islands and 24 km^ in area with a high ele- vation of 326 m. Chichijima was extensively deforested from the late 1880s to the mid 1900s (Katahira 1981). Today, Chichijima is generally covered with relatively low vegetation. About 73% of Chichijima is regenerated na- tive forests and scrubs, and the rest includes coastal for- ests, exotic low shrubs {Leucaena leucocephald) , grasses {Stachytarpheta jamaicensis) , and residential areas; canopy trees in all areas consist of native and introduced species that do not exceed 15 m in height (Shimizu and Tabata 1991). Land Animals on Chichijima. In addition to the extant native migratory and introduced land vertebrates in Chichijima (Table 1) and accidental visitors of birds, some breeding seabirds are also found on the islands. Nonbreeding visitors are infrequent and breeding sea- birds occur in a relatively restricted season of the year depending on the species (e.g., Momiyama 1930, Chiba and Funatsu 1991, Ornithological Society of Japan 2002, Kato pers. observ.). The native land invertebrates include about 236 spe- cies of insects and 63 species of land snails, although many of these have exhibited declines in numbers or al- ready have gone extinct (Kurozumi 1988, Kato 1991). Figure 1. The location of the study area, Ogasawara (Bonin) Islands. Introduced animals (Table 1; see Tomiyama 1998) are very common now in the islands as well as introduced invertebrates, such as European honeybees {Apis melh- fera) and African giant snails {Achatina fulica), Norway rats {Rattus norvegicus) have also been recorded as intro- duced rodents on the Ogasawara Islands, but recently have not been found in Chichijima. Data Collection. Data were collected by two methods: (1) observation of hunting or carrying behavior by buz- zards from fixed points and (2) observation at nest sites We conducted general behavioral observations, prey-car- rying and hunting behavior for 491 hr in January-June in 2000, 543 hr throughout 2001, and 668 hr in 2002. We also included opportunistic observations in 1998 and 1999. The buzzards were observed with binoculars (8X) and/or a telescope (60X). We pooled all data, because sample size in each year was small. We observed four nests on cliffs during 148.9 hr in May-June in 2000 and one nest for a total of 28.8 hr in May-June in 2001. We watched those nests from hides, which were more than 300 m away. Prey items brought to the nests were iden- tified with a telescope (60X). We took care not to disturb the nesting activity; we generally stopped the observation when a parent on the nest noticed us, or a parent outside the nest displayed an alarm for more than 5 min. Data June 2005 Short Communications 175 Table 1. Present fauna of resident land vertebrates on Chichijima, the Ogasawara (Bonin) Islands. Species Status^ Mammals Ogasawara fruit bat {Pteropus peslaphon) Native Feral goat {Capra hircus) Introduced Feral cat (Felis catus) Introduced Black rat {Rattus rattu^) Introduced House mouse {Mus musculus) Introduced Birds Brown-cheeked Bulbul {Hypsipetes amaurotis) Native Bull-headed Shrike {Lanius bucephalus) Recent immigrant Bush Warbler {Cettia diphone) Native Japanese White-eye {Zosterops japonicus) Introduced Rock Thrush {Monticola solitarius) Native White’s Ground Thrush {Zoothera dauma) Recent immigrant Japanese Wood Pigeon {Columba janthina) Native Common Buzzard {Buteo buteo) Native Reptiles Ogasawara snake-eyed skink ( Cryptoblepharus boutonii) Native Green anole {Anolis carolinensis) Introduced House gecko {Hemidactylus frenatus) Introduced Morning gecko {Lepidodactylus lugubris) Introduced Brahminy blind snake {Typholops braminus) ? Amphibians Marine toad {Bufo marinus) Introduced ^ Categorizations are referred from Tsuyama and Asami (1970), and the Ogasawara Natural Environment Study Group (1992). ^ Another introduced rat, K norvegicus, formerly resident, has not heen found recently. from nest observations were also combined because of small sample sizes for each nest. The two methods were not employed on the same day in the same territory, which should have prevented the double counts of the same prey items. For both methods, we were usually able to identify prey animals, because all prey observed were likely species on the limited list of terrestrial vertebrate fauna present in the Ogasawara Is- lands (Table 1). As black rats {R. rattus) and Norway rats are similar in size and shape, it was possible that we mis- took one species for the other. However, several sampling surveys have been unable to document Norway rats on Chichijima (e.g., Yabe and Matsumoto 1982, K. Watanabe pers. comm., pers. observ.). Therefore, we assumed that the black rat were the only large rodent prey taken by buzzards and that Norway rats was inconsequential in the diet of this hawk. In order to evaluate the relative importance of prey species as in the diet on a biomass basis, we weighed dominant prey species using specimens collected by us in Chichijima in 2001. Twenty-three black rats were caught with live traps in August-October, along with 29 green anoles, and 32 marine toads caught by hand in April-May and on 16 September, respectively. As marine toads were caught around a flume where many toads con- gregated for mating in the night, they were considered all adults. Adult biomass was appropriate for this prey species because all but one partial food item identified as a marine toad was large and determined to be an adult. Results We recorded 156 prey items using the two methods of observations. Among them, we omitted from the analyses 46 prey items that we could not identify. The Ogasawara buzzard hunted or carried 49 prey items, and 40 of them were identified (Table 2). Black rats {N = 23, 57.5%) and green anoles {N = 9, 22.5%) predominated in this sample. Marine toads {N = 2, 5.0%) and petrels {Bulweria bulwerii or Oceanodroma sp.; N = 2, 5.0%) were next most frequent. A house mouse (Mus musculus) and a White’s Ground Thrush {Zoothera dauma) were also observed {N = 1, 2.5%, respectively). One small- or medium-sized bird (2.5%) and one lizard (2.5%) were also recorded as prey, though we could not identify these to the species level. We recorded 35 (87.5%) of the 40 identified prey items during January-June when the Ogasawara buzzard engages in breeding activities. Six (17.1%; all were black rats) of the 35 prey items were used for flight display by territorial males. Two birds (one thrush and one petrel, 176 Short Communications VoL. 39, No. 2 Table 2. Diets of the Ogasawara buzzards observed in Chichijima (1998-2002). Prey items identified at the class level were shown.^ Prey No. Observed Hunted or Observed Carried at Nest Total Percent Mammals Black rat {Rattus rattus) 23 32 55 50.0% House mouse {Mus musculus) 1 0 1 0.9% Birds White’s Ground Thrush {Zoothera dauma) 1 1 2 1.8% Petrel*’ 2 0 2 1.8% Small- to medium-sized birds'^ 1 3 4 3.6% Reptiles Green anole {Anolis carolinensis) 9 27 36 32.7% Unknown 1 0 1 0.9% Amphibians Marine toad {Bufo marinus) 2 7 9 8.2% Total 40 70 no 100.0% “ Forty-six animals (9 observed hunting and 37 prey deliveries) were excluded from the table because their identity was uncertain. •’ Bulweria bulwerii or Oceanodroma sp. Small- to medium-sized birds include Zosterops japonicus, Cettia diphone, Hypsipetes amaurotis, Ijinius bucepharus, Monticola solitarius, and Zoothera dauma. 5.0%) were fed to fledglings. Though we could not follow the fate of the remaining prey, five (14.3%) of the 35 were likely brought to the nests based on the flight di- rection of the breeding Ogasawara buzzard. Other prey seemed to he also used for breeding activities (fed to mates and/or offspring) rather than for self consump- tion. We recorded 107 prey items during the nest site ob- servations. Sixteen of the prey items were already on the nests when we began the observations. We observed that 101 (94.4%) of the 107 prey items were consumed by either nestlings or the female. Six other items were car- ried off from the nests by the female parent after the nestling stopped feeding. We identified 70 (65.4%) of the 107 prey items ob- served at nest sites (Table 2). Among them, black rats {N — 32, 45.7%) and green anoles {N = 27, 30.6%) were dominant, followed by marine toads {N = 7, 10.0%). The remaining prey items were small- to medium-sized birds. We combined the data collected by the two different methods, because the prey composition between the two methods was not significantly different (x^ = 4.13, df = S, P = 0.24). Based on the pooled sample, we identified 110 prey items. There were three predominant species in the diet; black rats (N = 55, 50.0%), green anoles (N = 36, 32.7%), and marine toads {N = 9, 8.2%; Table 2). In addition, we also collected eight pellets in fields inci- dentally during the study. All of these contained only the remnants of black rats. To evaluate the relative importance of each prey spe- cies as foods, we took body mass into consideration. Many black rats (x = 114.5 ± 44.2 g [SD], range = 50-180 g, N= 23) and marine toads (x = 91.6 ± 36.8 g, range = 35-205, N = S2) were more than 100 g, but all green anoles {x = 6.0 ± 1.6 g, range = 3. 0-9. 5, N = 29) were less than 10 g. Total biomass consumed during observa- tion was 6298 g for black rats, 216 g for green anoles, and 824 g for marine toads. Discussion During this study, we documented the current food habits of the Ogasawara buzzard, an insular endemic sub- species. The number of species taken was very small (Ta- ble 2) compared with other populations of the Common Buzzard (e.g.. Cramp and Simmons 1980). The main rea- son for this difference was undoubtedly the low diversity fauna inhabiting the Ogasawara Islands (Table 1). The Ogasawara buzzard primarily caught black rats, green anoles, and marine toads, all of which were introduced animals in the Ogasawara Islands (Tomiyama 1998). These three species accounted for ca. 90% of prey items in terms of frequency (Table 2). Black rats are known to be common on many islands of the Ogasawara archipel- ago. However, green anoles and marine toads are only present and common on two major islands (Chichijima and Hahajima) of the Ogasawara chain. Based on the predominance of these three prey species in the diet of June 2005 Short Communications 177 Table 3. Possible native prey available for the Ogasawara buzzard in the Ogasawara (Bonin) Islands before coloni- zation by human beings and exotic animals.® Species Status Vertebrates Mammals Ogasawara fruit bat (Pteropus peslaphon) Endangered Birds Brown-cheeked Bulbul {Hypsipetes amaurotis) Bush Warbler (Cettia diphone) Bonin White-eye {Apalopteron familiare) Rock Thrush (Monticola solitarius) Bonin Islands Thrush ( Chichlopasser terrestris) Oriental Greenfinch ( Carduelis sinica) Bonin Islands Grosbeak ( Chaunoproetus ferreorostris) Jungle Crow {Corvus macrorhynchos) Bonin Wood Pigeon {Columba versicolor) Japanese Wood Pigeon {Columba janthina) Rufous Night Heron {Nycticorax caledonicus) Small- to medium-sized sea birds (e.g., Bulweria, Oceanodroma) Occasional migrant visitors (e.g., Ardeidae, Charadriidae, Scolo- pacidae, and passerines) Reptiles Ogasawara snake-eyed skink ( Cryptoblepharus boutonii) Green turtle {Chelonia mydas) Invertebrates Extinct in Chichijima Extinct Extinct in Chichijima Extinct Extinct Extinct Endangered Extinct Large insects (e.g., Cicadidae, Buprostidae) Land crabs {Geograpsus, Sesarma) and hermit crabs (e.g., Coenobi- ta) Large land snails (e.g., Mandarina) extinct or endangered de- pending on species) Extinct or Endangered ® Main sources are Tsuyama and Asami (1970) and Ogasawara Natural Environment Study Group (1992). the buzzard, the density of these three species appears to be very high on Chichijima Island. We observed the Ogasawara buzzard to prey upon only few birds (Table 2). The birds appeared to be auxiliary food for buzzard on Chichijima Island during this study. But other researchers have reported some birds as prey of the Ogasawara buzzard. Ueda and deForest (1998) found that the Japanese White-eye {Zosterops japonicus) was prey for the buzzard in Chichijima and Kawakami (2000) recorded the White’s Ground Thrush, Pacific Golden Plover {Pluvialis dominica) , and Providence Petrel {Pterodroma hypoleuca) in the diet on Hahajima. Further- more, the Ogasawara buzzard was reported by villagers to hunt migrant egrets (Y. lida pers. comm.). The Common Buzzard is known to prey frequently (e.g.. Cramp and Simmons 1980) and selectively (Jedrze- jewski et al. 1994) on medium-sized birds. The Ogasawara buzzard also hunted White’s Ground Thrushes. We sug- gest that medium-sized birds, especially fledglings and large nesdings because of their relatively large body mass. could be important prey at times for the Ogasawara buz- zard. For example, the mean mass of six adult White’s Ground Thrushes was 128 g (Y. Kato, unpubl. data). We also observed the Ogasawara buzzard preying upon seabirds (e.g., petrels; Table 2), which are generally large compared with passerines. The Ogasawara Islands are known as breeding grounds of seabirds (e.g., Bulwers Pe- trel [Bulweria bulwerii] , Wedge-tailed Shearwater [Pufpnus padficus]. Brown Noddy [Anous stolidusi, and Brown Boo- by [Sula leucogaster]; Ornithological Society of Japan 2002). We observed that three pairs of the Ogasawara buzzard had breeding territories that included seabird breeding colonies on the islets near Chichijima (Y Kato, unpubl. data) . These three territories were not included in our nest-site studies. As breeding Ogasawara buzzards seem to hunt only within their territories, it was not sur- prising our list of the prey species was limited to land fauna. Thus, our data may have underestimated buzzard predation on seabirds by excluding territories without seabird colonies. 178 Short Communications VoL. 39, No. 2 The number of marine toads (x = 91.6 g) taken was less than that of green anoles recorded (Table 2). How- ever, the biomass of marine toads was more than 10 times that of green anoles (x = 6.0 g). Thus, we suggest that marine toads may be the more important prey for Oga- sawara buzzards. Of course, green anoles might also be an important food if frequent feeding provides the buz- zards with stable nutrition. During the study period, the Ogasawara buzzard large- ly fed on introduced vertebrates. This dietary pattern is similar as reported in the Hawaiian Hawk (B. solitarius) in the Hawaiian Islands (Griffin et al. 1998). These re- sults indicated that the endemic buzzards surviving on selected islands, with drastically modified environments, m the Pacific Ocean now feed on exotic animals intro- duced by human beings. Then, what animals did the Ogasawara buzzard feed on before the arrival of introduced animals? Based on the knowledge of the current food habits of the Ogasawara buzzard and of the continental Common Buzzard (e.g.. Cramp and Simons 1980) , we can speculate about the his- torical diet of this species. We suggest that the former diet of this buzzard consisted of several medium-sized birds, a couple reptiles, some large invertebrates, and perhaps, the Ogasawara fruit bat {Pteropus peslaphon; Table 3) . The Common Buzzard is known to switch from one species of primary prey to another depending on their abundance (Truszkowski 1976, Reif et al. 2001, Selas 2001). In the Ogasawara Islands, about a third of native land bird species have expired since the human coloni- zation (Ornithological Society of Japan 2002), and the abundance of native animals have generally decreased and many of them are threatened with extinction (Tsuy- ama and Asami 1970). On the contrary, the number of species and the density of introduced animals increased. The predominance of introduced animals in the present diets of the Ogasawara buzzard has probably been caused by the shift in prey availability. Animates Introducidos en ia Dieta de Buteo buteo to- YOSHIMAI, UN AvE RaPAZ EnDEMICA A LAS ISLAS DEL OC EA NO PacIfico Resumen. — Buteo buteo toyoshimai es una subespecie insu- lar endemica distribuida exclusivamente en las islas Oga- sawara (Bonin), en el Oceano Pacifico subtropical. Esta subespecie se considera amenazada en Japon. La ecolo- gia de esta rapaz, incluyendo sus habitos alimenticios, ha sido poco investigada. En este estudio examinamos cuan- titativamente su dieta mediante observaciones directas realizadas entre 1998-2002 en Chichijima, la isla de may- or area (ca. 24 km^) entre las islas Ogasawara. Las presas fueron registradas mientras las aves estaban cazando, mientras aves adultas las cargaban o cuando eran llevadas a los nidos. De los 110 items identificados, 55 (50.0%) correspondieron a Rattus rattus, 36 (32.7%) a Anolis car- olinensis y 9 (8.2%) a Bufo marinus. Con base en la con- tribucion estimada en terminos de biomasa, R. rattus fue predominante, seguida por B. marinus. [Traduccion del equipo editorial] Acknowledgments We thank K. Morimoto, N. Kachi, H. Chiba, and the staff of the Institute of Boninology for their helpful in- formation and field support. A. Suzuki kindly informed us of the mass of green anoles before we took our own measurements. We also thank three anonymous referees for their helpful advice that improved this manuscript. This study was partially supported by the Research Pro- ject on Conservation Methods of Subtropical Island Eco- systems coordinated by S. Nohara and funded from Japan Environmental Agency and a public trust Zoshinkai Na- ture and Environmental Conservation Research Grant. Literature Cited Cade, TJ. AND C.G. Jones. 1993. Progress in restoration of the Mauritius Kestrel. Conserv. Biol. 7:169-175. Chiba, H. and T. Funatsu. 1991. Birds in Chichijima-ret- to and Hahajima-retto. Pages 135-147 in M. Ono, M Kimura, K. Miyashita, and M. Nogami [Eds.], Reports of the second general survey of natural environment of the Ogasawara (Bonin) Islands. Tokyo Metropoli- tan Univ., Tokyo, Japan. Cramp, S. and K.E.L. Simmons. (Eds.). 1980. Handbook of the birds of Europe, the Middle East and North Africa. Vol. 2. Oxford Univ. Press, Oxford, U.K. DEL Hoyo, J., A. Elliott, and J. Sargatal. (Eds.). 1994. Handbook of the birds of the world. Vol. 2. Lynx Edi- tions, Barcelona, Spain. Grifein, C.R., P.W.C. Paton, and T.S. Baskett. 1998. Breeding ecology and behavior of the Hawaiian Hawk. Condor 100:654-662. Jedrzejewski, W., a. Szymura, and B. Jedrzejewska. 1994. Reproduction and food of the buzzard Buteo buteo in relation to the abundance of rodents and birds in Bi- alowieza National Park, Poland. Ethol. Ecol. Evol. 6: 179-190. Katahira, H. 1981. The characteristics of land use in Chichijima island, the Ogasawara islands. Pages 43- 63 in The general survey of natural environment of the Ogasawara (Bonin) Islands research group of To- kyo Metropolitan University [Eds.], Reports of the general survey of natural environment of the Ogasa- wara (Bonin) Islands. Tokyo Metropolitan Univ., To- kyo, Japan. Kato, M. 1991. A list of insects from the Ogasawara Is- lands. Ogasawara Res. 17/18:32-59. Kawakami, K. 2000. Bird deaths in the Bonin Islands. An- imals and Zoos 52:12-16. Kuroda, N. 1930. The geographical distribution of mam- mals in the Bonin Islands. Bull. Biogeogr. Soc. Japan 1: 81-88. KurOZUMI, T. 1988. Species composition and abundance of land mollusks and factors affecting their extinction in the Ogasawara (Bonin) Islands. Ogasawara Res. 14/ 15:59-109. Ministry oe Environment. (Eds.). 2002. Threatened wildlife of Japan -red data book. Japan Wild Research Center, Tokyo, Japan. June 2005 Short Communications 179 Momiyama, T. 1927. Discrip tion of twenty-five new birds and three additions from Japanese territories. Annot. Orn. Orient. 1:81-101. . 1930. On the birds of Bonin and Iwo-islands. Bull. Biogeogr. Soc. Japan. 1:89—186. Nakane, M., T. Matsumoto, and K. Miyashita. 1980. Sta- tus of the birds on Chichijima and Haha-jima in the Ogasawara Islands. Pages 43—63 in The general survey of natural environment of the Ogasawara (Bonin) Is- lands research group of Tokyo Metropolitan Univer- sity [Eds.], Reports of the general survey of natural environment of the Ogasawara (Bonin) Islands. To- kyo Metropolitan University, Tokyo, Japan. Ogasawara Natural Environmental Group. 1992. The nature of the Ogasawara Islands. Kokin-Shoin, Tokyo, Japan. (In Japanese.) Ornithological Society of Japan. 2002. Check-list of Japanese birds. Ornithological Society of Japan, Hok- kaido, Japan. Reif, V., R. Tornberg, S. Jungell, and E. Korpimaki. 2001. Diet variation of Common Buzzards in Finland supports the alternative prey hypothesis. Ecography 24: 267-274. Selas, V. 2001. Predation on reptiles and birds by the Common Buzzard, Buteo buteo, in relation to changes in its main prey, voles. Can. J. Zool. 79:2086-2093. Sergio, E, A. Boto, C. Scandolara, and G. Boglianti. 2002. Density, nest sites, diet, and productivity of Common Buzzards (Buteo buteo) in the Italian pre- Alps. J. Raptor Res. 36:24—32. Shimizu, Y. and H. Tabata. 1991. Forest structures, com- position, and distribution on a Pacific island, with ref- erence to ecological release and speciation. Pac. Sci. 45:28-49. Suzuki, T. 1982. Status of the Ogasawara buzzard on Chichijima: estimation of distribution and abun- dance. Ann. Rep. Ogasawara Res. 6:23-34. AND H. Chiba. 1995. Distribution and ecology of the Ogasawara Buzzard. Pages 61-93 in Tokyo Re- gional Forest Office [Eds.], The Ogasawara buzzard- report for conservation act of rare plants and animals Tokyo Regional Forest Office, Tokyo, Japan. AND Y Kato. 2000. Abundance of the Ogasawara buzzard on Chichijima, the Pacific Ocean. J. Raptor Res. 34:241-243. Swann, R.L. and B. Etheridge. 1995. A comparison of breeding success and prey of the Common Buzzard Buteo buteo in two areas of northern Scotland. Bird Study 42:37—43. Takano, S., Y. Uchida, N. Yanagisawa, and N. Sugiyama 1970. The birds of the Bonin Islands and the Volcano Islands. Pages 61-87 in Agency for Cultural Affairs [Eds.], Nature of the Ogasawara Islands. Agency for Cultural Affairs, Tokyo, Japan. Tomiyama, K. 1998. Disturbance of island ecosystem by introduced species in Ogasawara Islands. Jpn. J. Ecol , 48:63-72. Truszkowski, J. 1976. Role of the Common Buzzard (Bu- teo buteo L.) in agrocenoses of the middle Wielkopol- ska. Pol. Ecol. Stud. 2:103-111. Tsuyama, T. and S. Asami. 1970. Nature of the Ogasawara. Hirokawa Publishing Co., Tokyo, Japan. Ueda, K. and L. N. deforest. 1988. Food and behavior of male Ogasawara buzzards during the courtship feeding period. J. Ornithol. 37:34—36. Wild Bird Society of Japan. 1975. Hahajima Islands. Pages 93-136 in Wild Bird Society of Japan [Eds.], Survey on special birds requiring protection. Japan Environment Agency, Tokyo, Japan. Yabe, T. and T. Matsumoto. 1982. A survey on the ma- rine rodents on Chichijima and Hah^ima, the Oga- sawara Islands./. Mammal Soc. Jpn. 9:14-19. Received 2 November 2003; accepted 4 October 2004 J. Raptor Res. 39(2):179-183 © 2005 The Raptor Research Foundation, Inc. The Diet of Eurasian Griffons {Gypsfulvus) in Crete Stavros M. Xirouchakjs^ Natural History Museum oj Crete, University oj Crete, P.O. Box 2208, Heraklion 71409, Crete, Greece ICey Words: Eurasian GriJJon-, Gyps fulvus; pellet analysis', diet, Crete. Between a quarter and half of the global range of the Eurasian Griffon (Gyps Julvus) population occurs within Europe (Arroyo 1994). The species is distributed mainly in countries bordering the Mediterranean basin and feeds primarily on livestock carrion (Cramp and Sim- mons 1980, Donazar 1993). In Crete, the griffon’s feed- ing ecology may be more closely related to pastoralism than anywhere else in Europe. Paleontological findings indicate that all autochthonous-mammal species (apart ^ Email address: sxirouch@nhmc.uoc.gr 180 Short Communications VoL. 39, No. 2 from some rodents) disappeared from the island by the end of the Pleistocene (Sondaar et al. 1996, Willemsen 1996), while the presence of griffons dates back to the upper Pleistocene and Holocene (Rich 1983). Moreover, the introduction of sheep and goats by farmers occurred 7000 years ago (Cavalli-Sforza 1996, Jarman 1996). Since the last century the main potential food sources for this species on Crete consisted of domestic livestock and one wild species (i.e., the Cretan wild goat \Capra aegagrus cretica]). Given that (1) studies of the Eurasian Griffon’s diet are completely lacking in Greece, (2) that the most important griffon population in the country is found on Crete (Handrinos 1985, Handrinos and Akri- otis 1997), and (3) that the number of species that are available as carrion on the island is low, this study on the griffons’ diet should lead to more effective management of the species. Pellet analysis is the most common tech- nique for the study of the diet of carrion-eating species (Hiraldo 1976, Coleman and Fraser 1987, Ceballos and Donazar 1990, Thibault et al. 1993, Larraz, 1999). How- ever, an analytical problem arises due to the low occur- rence or even absence of bones and a high concentration of hair or feathers in pellets (Houston 1976). The latter provide information concerning the qualitative compo- sition of the diet, but they do not reveal the number of animals on which the bird has fed upon. Thus, no quan- titative data or biomass quantity can be estimated from such pellets (Marti 1987) . Here, 1 report qualitative data on the diet of the Eurasian Griffin on the island of Crete, based on the analysis of regurgitated pellets. Methods Fieldwork was undertaken in 1 1 griffon colonies at the end of the 2000 and 2001 breeding seasons. This period was considered to be most suitable, because it offers the opportunity to collect pellets while causing minimum dis- turbance to the birds. As nests are progressively destroyed by the chicks, a relatively large sample of pellets may be found among the nest material at the bottom of the breeding cliffs. Unfortunately, collection of pellets below roost sites that would account for the diet of the species throughout the year was not feasible. The bases of most cliffs where roosting sites were located were searched in- tensively but did not provide any pellet material. Pellets were found below 14 nests in four colonies (two in western and two in eastern Crete, which hosted 90 and 79 individuals in 2000 and 2001, respectively). They were sun-dried in order to remove water (Coleman and Fraser 1987), their mass measured and separated from bones and feathers. They were dissected into small pieces for subsequent identification. The hair was identified through a photonic stereoscope Leica MZ8 (Leica Micro- systems Ltd., Heerbrugg, Switzerland), with the aid of an identification-key (Papageorgiou and Sfougaris 1989) and by comparison to reference samples from specimens of local mammal species, which constitute potential prey for griffons (i.e., domestic sheep {Ovis aries^ and domes- tic goat [Capra hircus], beech marten [Martes foina], bad- ger [Meles meles], brown hare [Lepus europaeus], and the Cretan wild goat). Table 1. Items found in Eurasian Griffon pellets in Crete (Greece, 2000-01). Item N (Percent) Domestic sheep hair {Ovis aries) 127 (89.4) Domestic goat hair ( Capra hircus) 22 (15.5) Rabbit hair {Oryctolagus cuniculus) 4 (2.8) Beech marten hair {Martes foina) 2 (1.4) Bones (Caprines) 9 (6.3) Gramineae spp. 61 (42.9) Holm-oak leaves Quercus coccifera 24 (16.9) Dirt 14 (9.86) Coleoptera (Family Dermestidae) 8 (5.6) Olive core 1 (0.7) Eurasian Griffons are exclusively carrion eaters, feed- ing mainly on soft tissues, thus only the frequency of the presence of prey items was estimated (namely the num- ber of pellets containing each prey species X 100/total number of pellets; Ceballos and Donazar 1990). The null hypothesis that sheep and goat remains were found in the pellets in exact ratio to their proportion in livestock numbers was tested by a chi-square test with a Yates’ cor- rection (Zar 1996). When the remains of one or more animals were found in the same pellet it was assumed that they provided information on the frequency that griffons fed on them (Hiraldo 1976). Results My field assistant and I collected 811 different items among nest ruins. However, only 142 items were consid- ered as regurgitated material and only 93 (65.5%) had the typical oval shape of the pellets produced by raptors (Fitz- ner et al. 1977). The rest of the items were tufts of sheep and goat hair, which probably originated from dissolved pellets, and were not included in this analysis. Pellet size was rather small, with a mean diameter of 19.3 mm (SD = 8 mm, range = 5-50 mm), mean length of 57.7 mm (SD = 24.5 mm, range = 18-138 mm), and mean width of 33.5 mm (SD = 14.6 mm, range = 10-105 mm). Their mass ranged from 0.6-15.3 g (x = 4.05 ± 2.86 g) and contained 10 different constituents (Table 1). The dominant species in the griffons’ diet was domes- tic sheep, which was present in almost 90% of the pellets. Domestic goat was the second most important prey item (Table 1). Sheep were represented in the pellets more than goats (x^ = 16.5, P< 0.001). The rabbit {Oryctolagus cuniculus) and the heech marten were the only additional species detected in the analysis, but in a substantially smaller number of pellets. Pieces of long bones or whole vertebra were found in 6.3% (N = 9) of the pellets, all originating from adult caprines. Plant material occurred in 42.9% (N = 61) of the pel- lets and was mainly grass (i.e., Gramineae), while holm oak leaves (Quercus coccifera) were found in 16.9% {N = 24) of them. Fourteen pellets (9.86%) consisted exclu- sively of dirt with a few stones, one contained an olive June 2005 Short Communications 181 core (0.7%) and eight (5.6%) contained cocoons of co- leoptera of the family Dermestidae. Discussion In Europe, the griffon usually feeds on cattle (Bos tau- rus) , equines (Equus caballus, Equus asinus) , caprines (Ovis aries, Capra hircus, Rupicapra rupicapra), red deer (Cervus elaphus), pigs (Sus scrofa var. domesticus) , canids (Canis familiaris, Vulpes vulpes), lagomorphs (Lepus capen- sis, Oryctolagus cuniculus), birds and insects (Fernandez 1975, Beven 1979, Cramp and Simmons 1980, Gonzalez et al. 1984), but it prefers large ungulates when these are available (Don^ar 1993). Fernandez (1975) reports that the species’ diet in southern Spain consists mainly of do- mestic animals such as goats (34.9%), sheep (20.9%), donkeys (11.6%), and cows (7%); while in northern Spain, sheep were found in more than 75% of pellets analyzed (Marco and Garcia 1981). The results of the present study do not differ from relevant ones in other Mediterranean regions in the sense that griffons feed op- portunistically on the carrion of the most common spe- cies in their foraging range. The findings of my pellet analysis probably reflect the availability of the different prey species and the way that griffons feed on different types of carrion. Medium sized domestic animals, such as sheep and goats, constitute the majority of livestock (i.e., 95.3%; National Statistical Ser- vice of Greece 1991) and the totality of the free grazing livestock of Crete. They are present in the uplands from late April until late October, and suffer greater losses than those that are homebred due to accidents, malnu- trition, and sickness. Moreover, they are well represented in the pellets because griffons start feeding on their car- casses from any part of the body, thus consuming lots of hair (pers. obs.). By contrast, large ungulates were com- pletely absent from the pellets. In Crete the number of such animals has decreased by 86% (National Statistical Service of Greece, pers. comm.) in recent decades (i.e., since 1961) and their corpses are buried currendy. Typ- ically during the 1960s and early 1970s old equines were left to die in remote upland areas or their carcasses were thrown in a few waste dumps near the villages (primarily in gorges) . However, griffons are incapable of tearing the skin of these animals. Instead, griffons access nutrition from carcasses mainly from the natural orifices (e.g., mouth) , which produces minimal intake of hair in their food (Houston 1976). I attributed the high frequency of sheep hair in the pel- lets to the spatial distribution of domestic animals up to the time that pellet collection took place. Pellets were col- lected from nest material and account for the period that nests were occupied by incubating individuals or by parent birds attending their young, specifically from early January to late May/early June. The majority of livestock in the vicinity of the colonies during this period was sheep. Small waste dumps that operate illegally near these stockyards also provided dead biomass to the griffons. The rabbit remains recorded were suspected to have been taken from a Bearded Vulture (Gypaetus barbatus) nest. A breeding pair had been supplied with dead rabbits during the chick-rearing period (Xirouchakis et al. 2003) 3 km from the colony where the pellets containing rabbit hair were collected. After the Bearded Vultures’ breeding attempt, griffons were observed to visit the nest several times and consume bones and prey remains. In other cas- es, griffons were observed to take bone fragments from the base of a cliff 4 m below another nest while a Bearded Vulture was incubating. This habit has also been reported in Spain, where griffons occasionally explore the ossuaries of Bearded Vultures, where the latter drop the bones to break them (Bertran and Margalida 1997). Based on the present results, bone consumption seemed frequent in Crete, suggesting that griffons do not suffer from calcium deficiency as some African vultures do in areas where large carnivores are extinct. This substance is not available in large herbivores’ carrion if no broken bones are found resulting from the action of large mammalian predators (Houston 1978, Richardson et al. 1986). In contrast, grif- fons in Crete have been seen to break the bones of small ungulates easily, and thus are not dependent on mam- malian carnivores or other vulture species with powerful beaks, which are absent from the island (e.g.. Cinereous Vulture [Aegypius monachus^). Vegetation in the griffon’s diet was probably consumed accidentally as well as the dirt, or was contained in the ungulates’ digestive tract. However, griffons have been observed to eat green plants after feeding bouts, a pro- cess that probably facilitates pellet formation (Hiraldo 1976). Finally, the coleoptera prey that was recorded was likely consumed incidentally along with hair as dormes- tids (Dermestidae) feed on the wool of dead animals. Overall, irrespective of the source of food used by grif- fons (i.e., carrion from waste dumps or from transhu- mance herds) , pellet analysis showed that their diet was dominated by small domestic ungulates. Considering the low use of poisoned baits for vermin (due to the absence of large predators) and the dramatic increase in livestock numbers (>70%; National Statistical Service of Greece 1991) promoted by subsidy policies when Greece joined the European Union in 1981, the future of the species seems rather favorable. However, increased food abun- dance does not necessarily imply higher food availability. The latter is also determined by other factors such as stock-rearing methods, sanitary conditions, closing down of rubbish dumps, or outdoor abattoirs. In the mean- time, the strong dependence of griffons on pastoralism means that socio-economic factors may place them at risk of a human-induced food shortage. La Dieta de Gyps fulvus en Greta Resumen. — Recolectamos egagropilas de 14 nidos de Gyps fulvus en Creta al final de las estaciones reproduc- tivas de 2000 y 2001. El analisis mostro que G. fulvus se 182 Short Communications VoL. 39, No. 2 alimenta casi exclusivamente de carrona proveniente de ovejas y cabras domesticas. Los equinos y bovinos estuvi- eron ausentes de las egagropilas, probablemente debido al modo en que los buitres consumer! los cadaveres y a la escasez de estos animales como carrona. La deficiencia de calcio en los volantones no fue considerada como un problema porque se encontraron huesos de pequenos ungulados en el 6.3% de las egagropilas. Se considera que el incremento en la abundancia de ganado de las ultimas dos decadas ha sido beneficioso para la especie; sin embargo, los metodos de cria del ganado tambien pueden afectar su estado de conservacion al largo plazo. [Traduccion del equipo editorial] Acknowledgments I thank C. Grivas for assistance in the fieldwork. Dr. I. Robert (Natural History Museum of Paris) for assistance with bone identification, A. Copland for improving the English text; A. Margalida, V. Penteriani, and F. Sergio for their comments on an earlier draft of this manuscript. This study was supported by the University of Crete and the Ministry of Education through the PYTHAGORAS II research grants. Literature Cited Arroyo, B. 1994. Griffon Vulture Gyps fulvus. Pages 156— 157 in G.M. Tucker and M.E Heath [Eds.], Birds in Europe; their conservation status. BirdLife Conserva- tion Series No. 3, BirdLife International, Cambridge, U.K. Bertran, J. and a. Margalida. 1997. Griffon Vultures {Gyps fulvus) ingesting bones at the ossuaries of Bearded Vultures {Gypaetus barbatus) . J. Raptor Res. 31: 287-288. Beven, G. 1979. Griffon Vultures apparently feeding on beetles. Br. Birds 72:336. Cavalli-Sforza, L.L. 1996. The spread of agriculture and nomadic pastoralism: insights from genetics, linguis- tics and archaeology. Pages 51-69 in D.J. Harris [Ed.], The origins and spread of agriculture and pastoralism in Eurasia. Univ. College London Press, London, U.K. Ceballos, O. and J.A. Donazar. 1990. Roost-tree char- acteristics, food habits and seasonal abundance of roosting Egyptian Vultures in northern Spain. J. Rap- tor Res. 24:19-25. Coleman, J.S. and J.D. Fraser. 1987. Food habits of Black and Turkey vultures in Pennsylvania and Maryland. J. Wildl. Manag. 51:733-739. Cramp, S. and K.E.L. Simmons. (Eds.). 1980. The birds of the western Palearctic. Oxford Univ. Press, Oxford, U.K. DonAzar, J.A. 1993. Los buitres ibericos: biologia y con- servacion. J.M. Reyero, Madrid, Spain. Fernandez, J.A. 1975. Consideraciones sobre el regimen alimenticio de Gyps fulvus. Ardeola 21:209-217. Fitzner, R.E., L.E. Rogers, and D.W. Uresk. 1977. Tech- niques useful for determining raptor prey-species abundance. Raptor Res. 11:67-71. Gonzalez, L.M., J.L. Gonzalez, and C. Llandres. 1984 Tree-nesting colony of griffon vultures in Spain. Vul- ture News 11:12-13. Handrinos, G. 1985. The status of vultures in Greece. Pages 103-115 in I. Newton and R.D. Chancellor [Eds.], Conservation studies in raptors. World Work- ing Group on Birds of Prey, Salonica, Greece. AND T. Akriotis. 1997. The birds of Greece. Helm-A. & C. Black Ltd., London, U.K. Hiraldo, F. 1976. Diet of Black Vulture {Aegypius mona- chus) in the Iberian Peninsula. Don. Acta Verteb. 3:19- 31. Houston, D.C. 1976. Breeding of the White-backed and RueppelFs Griffon vultures. Gyps africanvs and G. ruep- pellii. Ibis. 118: 14—40. . 1978. The effect of food quality on breeding strategy in griffon vultures {Gyps spp.)./. Zool. Lond 186: 175-184. Jarman, M.R. 1996. Human influence in the develop- ment of the Cretan mammalian fauna. Pages 211-229 in D.S. Reese [Ed.], Pleistocene and Holocene fauna of Crete and its first settlers. Prehistory Press, Madi- son, WI U.S.A. Larraz, D.S. 1999. Dumps for dead livestock and the con- servation of wintering Red Kites {Milvus milvus) . J. Raptor Res. 33:338-340. Marco, J. and D. Garcia. 1981. Situation actuelle des populations de necrophagus ( Gyps fulvus, Gypaetus bar- batus et Neophron percnopterus) en Catalogne. Rapaces Mediterraneens 1:119-129. Marti, C.D. 1987. Raptor food habit studies. Pages 67- 80 in B.G. Pendleton, B.A. Millsap, K.W. Cline, and D.M. Bird [Eds.], Raptor management techniques manual. Natl. Wildl. Fed., Washington, DC U.S.A. National Statistical Service of Greece. 1991. Statisti- cal year book of Greece, Athens. National Statistical Service of Greece, Athens, Greece. Papageorgiou, N. and a. Sfougaris. 1989. Identification of Greek mammals by the morphology of hair. Aris- totelian Univ. Salonica, Salonica, Greece. Rich, P.V. 1983. The fossil history of vultures: a world perspective. Pages 3-54 in S.R. Wilbur and J.A. Jack- son [Eds.], Vulture biology and management. Univ. California Press, Berkeley, CA U.S.A. Richardson, P.R.K., PJ. Mundy, and I. Plug. 1986. Bone crushing carnivores and their significance to osteo- dystrophy in griffon vulture chicks. /. Zool. Lond. 200. 23-43. Sondaar, P.Y., M.D. Dermitzakis, and J. De Vos 1996. The paleogeography and faunal evolution. Pages 61- 67 in D.S. Reese [Ed.], Pleistocene and Holocene fau- na of Crete and its first settlers. Prehistory Press, Mad- ison, WI U.S.A. Thibault, J.C., J.D. ViGNE, andJ. Torre. 1993. The diet of young Lammergeier Gypaetus barbatus in Corsica: its dependence on extensive grazing. Ibis 135:42-48. WiLLEMSEN, G.F. 1996. The Cretan otter Lutrogale cretensis. June 2005 Short Communications 183 Pages 153-157 in D.S. Reese [Ed.], Pleistocene and Holocene fauna of Crete and its first settlers. Prehis- tory Press, Madison, WI U.S.A. XiRoucHAKis, S., C. Grivas, M. Probonas, A. Sakoulis, AND G. Andreou. 2003. Evaluation des actions pour la conservation du Gypaete barbu {Gypaetus barbatus) en Crete. Pages 124-132 in J.-F. Sarrazin and J.-M. Thiollay [Eds.], Actes du colloque international: con- servation des populations de Gypaete barbu. LPO, Tende, France. Zar, J. 1996. Biostatistical analysis. Prentice Hall, Prince- ton, NJ U.S.A. Received 28 June 2004; accepted 7 March 2005 Letters J Raptor Res. 39(2) :184— 186 © 2005 The Raptor Research Foundation, Inc. Are Earlier Estimates of Accipitriformes Crossing the Channel of Sicily (Central Mediterranean) During Spring Migration Accurate? Each spring thousands of raptors cross the Channel of Sicily, between the Cap Bon Peninsula and western Sicily, during northward migration. The Cap Bon Peninsula in northeastern Tunisia is a 25-km-wide plain dominated in the north by a promontory that reaches a height of 392 m (Fig. 1). At this promontory, Thiollay (1975, Nos Oiseaux 33' 109-121; 1977, Alauda 43:115-121) made counts of migrants between 26 March-14 April 1974 and between 2- 18 May 1975 to estimate the migratory flow of raptors using this route (Table 1). Thiollay divided the period 22 March-20 May into six 10-d periods and extrapolated numbers of migrants in the periods for which observations were lacking. Three decades later these counts and estimates are still quoted in publications (Ferguson-Lees and Christie 2001, Raptors of the world, Helm Edition, London, U.K.; Brichetti and Fracasso 2003, Ornitologia Italiana, Perdisa Editore, Bologna, Italy; Sara 2003,/. Raptor Res. 37:167-172). However, subsequent studies carried out at the Cap Bon Promontory, at the Strait of Messina and over the islands of Marettimo and Ustica (Eig. 1) were not consistent with these earlier data (Giordano 1991, Birds Prey Bull. 4:239-249; Agostini and Duchi 1994, Bird Behav. 10’45-48; Agostini et al. 1994a, Avocetta 18:73-76; Agostini et al. 1994b, Atti VI Conv. Ital. Ornitol. 451-452; Agostini and Malara 1997, Riv. Ital. Ornitol. 66:174-176; Agostini and Logozzo 1998, Riv. Ital. Ornitol. 68:153-157; Agostini 2001, Buteo 12:99-102; Corso 2001, Br. Birds 94:196-202; Panuccio et al. 2004, Br. Birds 91:400-403). At the Cap Bon Promontory, several collaborators and I collected data focusing on the flight behavior of raptors on the coastline in the spring of 1990 and 1992 (Agostini and Duchi 1994, Agostini et al. 1994a, 1994b). During these observations, upon reaching the coast near the promontory, most raptors hesitated at the water barrier; they often stopped migrating and exhibited an unexpected series of movements, first disappearing across the sea, then returning to the coast, and then flying back inland. During our observations, my colleagues and I recognized indi- viduals or entire flocks circling over the promontory, sometimes repeatedly over several days. Because of this behavior, it was impossible to avoid recounting migrating raptors and we concluded that the Tunisian promontory was unsuit- able for assessing numbers of passing raptors. Factors probably influencing a hawk’s decision whether or not to cross the Mediterranean included the morphology and age of the hawk, weather, and in the case of Honey-buzzards {Perms apivorus) and Black Kites (Milvus migrans), flock size. Was the hesitation to cross this water barrier not conspicuous when Thiollay made his observations? Was it possible to avoid recounting migrants during spring 1974 and 1975^ The Channel of Sicily is ca. 150 km wide, implying that long, powered flight, with a considerable expenditure of energy, is necessary to make landfall. Because thermals are mostly absent over water, raptors cannot alternate soaring with gliding during the crossing, as they do over land (Kerlinger 1989, Flight strategies of migrating hawks, Univ. Chicago Press, Chicago, IL U.S.A.) and this could explain their hesitation at the Cap Bon Promontory when faced with a substantial water crossing. In particular, counts made at the Strait of Messina (Table 1 ) and over the islands of Marettimo and Ustica (Giordano 1991, Agostini 2001, Agostini and Logozzo 1998, Corso 2001, Panuccio et al. 2004, Br. Birds 97:400-403), showed that few broad-winged raptors, such as Common Buzzards {Buteo buteo). Booted Eagles {Hieraaetus pennatus) , Short-toed Eagles {Circaetus gallicus) , and Egyptian Vultures {Neophron percnopterus) cross the Channel of Sicily. However, Thiollay (1977) estimated the total passage of more than 2800, 450, 400, and 620 birds, respectively (Table 1). Moreover, at the Strait of Messina, the Long-legged Buzzard {Buteo rufinus) and the Lesser Spotted Eagle {Aquila pomarina) were irregular migrants (Table 1; see also Zalles and Bildstein 2000, Raptor watch: a global directory of raptor migration sites, BirdLife International, Cambridge, U.K. and Hawk Mountain Sanctuary, Kempton, PA U.S.A). Whereas, Thiollay estimated a regular passage of more than 200 and 150 individuals of these species. Undoubtedly, our counts missed some proportion of the birds that passed overhead; however, difference between data recorded in southern Italy since 1980s and the estimate made by Thiollay in the 1970s was substantial. Moreover, the few broad-winged raptors recorded in southern Italy was consistent with the notion that species with a lower-aspect ratio (shorter) wings are not suited to undertake crossings of large bodies of water. Such wings increase the induced drag, and thus, the energy needed for powered flight (Kerlinger 1985, Wilson Bull. 97: 109-113). I suggest that the earlier estimates offered by Thiollay in the 1970s were too high probably because, during his survey, the author recounted birds returning after they abandoned the water crossing attempt. Other explanations 184 June 2005 Letters 185 Tyrrhenian Sea Ustica Marettimo Cap Bon Sictiy Strait of Messina Tunisia Libya 200 km Figure 1. The Channel of Sicily in the central Mediterranean area. for differences between the 1970s and the more recent data include: (1) numbers of broad-winged raptors (eagles, buteos, vultures), and of Black Kites, may have measurably declined in recent years and (2) a climatic shift may have occurred possibly making water crossings more difficult. However, when comparing counts made at the Strait of Messina between 1986—90 and 1996-2000 (Table 1), I noted a substantial increase in the case of Honey Buzzards, Marsh Harriers (Circus a^^ginosus), and Black Kites. This comparison suggests, for at least these species, that water crossings have not become more difficult in recent years. Counts involving Short-toed Eagles made at the Cap Bon Promontory by Thiollay were recently cited by Ferguson- 186 Letters VoL. 39, No. 2 Table 1. Counts and estimates of raptors crossing the Channel of Sicily in the central Mediterranean area during three decades. Thiollay’ s Estimates 1970s“ Strait of Messina Counts 1986-90 (Min-Max)*’ Strait of Messina Counts 1996-2000 (Min-Max)'^ European Honey-buzzard {Pernis apivorus) >16.000 6.032-8.516 16.700-27.297 Black Kite {Milvus migrans) >15.000 155-397 546-1.008 Western Marsh Harrier {Circus aeruginosus) >700 125-978 1.621-3.074 Montagu’s Harrier (C. pygargus) >220 5-273 155-866 Pallid Harrier (C. macrourus) >50 4-15 25-83 Montagu’s/Pallid Harrier (C. pygargus X macrourus) — 0-29 33-159 Northern Harrier (C. cyaneus) >15 11-59 3-84 Common Buzzard {Buteo buteo) >2.800 15-42 30-103 Long-legged Buzzard {B. rufinus) >200 0-4 6-12 Short-toed Eagle ( Circaetus gallicus) >400 0-3 1-4 Egyptian Vulture {Neophron percnopterus) >620 4-8 3-12 Booted Eagle {Hieraaetus pennatus) >450 5-22 5-19 Lesser Spotted Eagle {Aquila pomarina) >150 0-5 0-4 Eurasian Sparrowhawk {Accipiter nisus) >70 0-7 2-14 Osprey {Pandion haliaetus) >20 2-20 10-25 ••Thiollay (1975, Nos 0iseaux3S: 109-121; 1977, Alauda 43: 115-121). ^ Giordano (1991, Birds Prey Bull. 4:239-249). Corso (2001, Br. Birds 94: 196-202). Lees and Christie (2001, Raptors of the world, Helm Edition, London): “Italian population (380-415 pairs) presum- ably crosses by Sicilian Channel to and from Tunisia’s Cap Bon, whence total of ca. 200 travelled northward on spring migration during 2-18 May 1975.” However, recent studies in Italy showed that nearly all the Italian population of this species crosses the Mediterranean Sea at the Strait of Gibraltar (14 km wide), with hundreds of birds breeding m central Italy using a circuitous migratory route both during spring and autumn migration (Agostini et al. 2002a, J. Raptor Res. 36:111-114; Agostini et al. 2002b, Ardeola 49:287—291; Agostini et al. 2004, Avocetta 28:37—40; Baghino 2003, Avocetta 27:67; Premuda 2004, Riv. Ital. Ornitol. 74:119-124). The number recorded at Cap Bon by Thiollay during the first half of May was relatively late for the spring migration of this species (Cramp and Simmons 1980, The birds of the western Palearctic, Vol. 2, Oxford Univ. Press, Oxford, U.K.). I suggest that perhaps the relatively large number of Short-toed Eagles reported by Thiollay was the result of recounting immature birds in northern Tunisia. In agreement with this conclusion, recent observations by some colleagues and I made over Marettimo showed a late autumn passage of juvenile Short-toed Eagles across the Sicilian Channel (Agostini et al. 2004, Avocetta 28:37-40). I wish to thank M. Gonzalez Forero, M. Sara, J. Bednarz, A. Green, F. Sergio, and one anonymous referee for their useful comments on this manuscript. — Nicolantonio Agostini (e-mail address: nicolantonioa@tiscalinet.it), Via Carlo Alberto n°4, 89046 Marina di Gioiosa Jonica (RC), Italy. Received 10 October 2003; accepted 10 March 2005 J. Raptor Res. 39 (2): 186— 187 © 2005 The Raptor Research Foundation, Inc. Ground Nesting by Egyptian Vultures (Neophron percnopterus) in the Canary Islands Ground nesting is a relatively rare occurrence in raptors, except for areas lacking any elevated nesting substrates (e.g., tundra habitats), or islands devoid of mammalian predators (Newton 1979, Population ecology of raptors, Buteo Books, Vermillion, SD U.S.A.). Moreover, this behavior has not been described for large diurnal raptors with long June 2005 Letters 187 breeding cycles that typically breed in protected cavities of cliffs, a trait presumably favored because it provided security against adverse weather. The Egyptian Vulture {Neophron percnopterus) is a medium-sized scavenger living mainly in open landscapes of arid and rugged regions of Eurasia and Africa. Although strongly migratory, this species also includes sedentary popula- tions on several archipelagos such as the Balearic Islands, Cape Verde, Canary Islands, and Socotra. Breeding takes place in cavities or caves of cliffs of variable height and nests are usually reused year after year. Occasionally, alternative sites are occupied within the same territory (Cramp and Simmons 1980, The birds of the western Palearctic, Vol. 2, Oxford Univ. Press, Oxford, U.K.). Egyptian Vultures have been extensively studied in Spain since the late 1970s. More than 1000 breeding attempts have been monitored. Most of them were in inaccessible nesting places, with only a few (<5%) in caves with easy access to large mammals, including humans. No nest was located directly on the ground (Donazar and Ceballos 1988, Ardeola 35:3-14). In this paper, we describe the first recorded case of ground nesting in Canarian Egyptian Vultures {Neophron percnopterus majorensis) . Fuerteventura (1662 km^) is the most eastern island of the Canary archipelago. It is relatively flat with a dry climate (<100 mm rain annually; Donazar et al. 2002,/. Raptor Res. 36:17-23). The island harbors the last population of an endangered endemic subspecies of the Egyptian Vulture, with no more than 130 individuals and 25 breeding pairs (Donazar et al. 2002, Biol. Conserv. 107:89-97). Twenty, 23, 21, 25, and 27 breeding territories have been monitored in 1998, 1999, 2000, 2001, and 2002, respec- tively. On 29 March 2002, we visited the breeding territory of one of these pairs, which had bred successfully in a cave on a hillside between 1998-2001. The old nest, easily accessible by foot, was unoccupied, but ca. 600 m away, we discovered an adult Egyptian Vulture incubating an egg on the ground. The new nest site was placed on a flat and exposed surface, with scattered shrubs {Launaea arborescens) . On 13 July 2002, we visited the nest to mark and measure the chick, which fledged successfully at the beginning of August. In 2003, the pair moved back to the cave it used in previous years and bred successfully there. Nesting in accessible caves is common for this species in Fuerteventura (in 2002, 41% of the nest sites were accessible by foot, N = 27), although inaccessible sites are not a limiting factor on the island (pers. observ.). Terrestrial predators were not existent on the island until the human colonization, 2500 yr ago. Currently, the only carnivores present are feral dogs and cats, in very low numbers. In addition, the dry climate may favor open nesting. Further- more, human density has been always extremely low (1000-3000 habitants before the European colonization; Cabrera 1996, La prehistoria de Fuerteventura: un modelo insular de adaptacion, Servicio de Publicaciones del Cabildo Insular de Fuerteventura, Puerto del Rosario, Islas Canarias). However, during the last several decades the human population in the island has increased sharply (11 668 in 1900, 69 260 in 2000; Anonymous 2001, Anuario estadistico de Fuer- teventura, Cabildo Insular de Fuerteventura, Puerto del Rosario, Islas Canarias), around a million tourists visit the island every year, and the number of pets has presumably increased too. These factors may lead to the loss of a number of nesting territories accessible to potential predators and, consequently, have a negative affect on this endangered population. We thank Cabildo de Fuerteventura and the Projects REN 2000-1556 GLO and CGL 2004-00270 for having funded this research. We also thank Jose A. Donazar, Jordi Figuerola, M. Di Vittorio, and an anonymous referee for reviewing early drafts of this letter. — Laura Gangoso (e-mail address: laurag@ebd.csic.es) and Cesar-Javier Palacios, Department of Applied Biology, Estacion Biologica de Donana, C.S.I.C., Avda. M“ Luisa s/n 41013 Sevilla, Spain. Received 30 July 2004; accepted 26 March 2005 Associate Editor: Fabrizio Sergio J. Raptor Res. 39(2):187— 188 © 2005 The Raptor Research Foundation, Inc. First Summer Records of Ospre\s {Pandion haliaetus) Along the Coast of Oaxaca, Mexico Between October 2000 and September 2001, we conducted 10 trips by sea, once during almost every month, to survey Ospreys {Pandion haliaetus) in Oaxaca, Mexico. Ospreys breed in temperate North America and along the coast of the Gulf of California (Henny and Anderson 1979, Bull. So. Calif. Acad. Sd., 78:89-106; Judge 1983, Wilson 188 Letters VoL. 39, No. 2 Bull. 95:243-255), but generally only winter in southern Mexico (Henny 1988, Pages 73-101 in R. Palmer [Ed.], Handbook of North American birds, Vol. 4. Yale Univ. Press, New Haven, CT U.S.A.), including Oaxaca. We surveyed islands and adjacent coasts between 0848-1614 H, from Zipolite Beach (96°29'W, 15°40'N), near Puerto Angel, to Huatulco Bay (96°05'W, 15°48'N). Total surveyed area was ca. 2.5 km^. The Pacific coast of Oaxaca is characterized by rocky beaches interspersed among extensive sandy portions of coastline. The adjacent mainland is almost exclusively deciduous forest. Small islands and rocky formations are concentrated throughout Huatulco Bay, where Ospreys are a moderately common wintering species. During surveys we observed 13 Ospreys in October, 12 in November, 11 in January, six in February, nine in March, two during both May and June, three in July, two in August, and finally, one individual in September. Because of the blackish patch in their crowns and above their necks, we assumed that our observations corresponded to Pandion haliaetus carolinensis, which winters in low densities in Oaxaca (Poole 1989, Ospreys, Cambridge Univ. Press, Cam- bridge, MA U.S.A.). During summer months there were two previous June records of Ospreys in southeastern Mexico: one record involved three birds in central Oaxaca, far from the coast (Binford 1989. A distributional survey of the birds of the Mexican state of Oaxaca. The American Ornithologists’ Union, Washington, DC U.S.A.), and a second record noted a few individuals on the Bay of Zihuatanejo, Guerrero (Erickson and Hamilton 1993, Euphonia 2:81—91). Our work now provides new summer records of nonbreeding Osprey for southern Mexico: two individuals from both May and August, and at least three individuals in July. It is possible that Ospreys we observed during these summer months were immature individuals that spend at least 16 continuous months in the tropics before returning to northern breeding areas (Henny 1988). We appreciate the improvements in English usage made by Kerri Vierling through the Association of Field Orni- thologists’ program of editorial assistance. — Juan Meraz (e-mail address: sula@angel.umar.mx), Instituto de Recursos, Universidad del Mar, Puerto Angel, Oaxaca, C.P. 70902 Mexico; Betzabeth Gonzalez-Bravo, Biologia Marina, Uni- versidad del Mar, Puerto Angel, Oaxaca, C.P. 70902, Mexico. Received 10 May 2004; accepted 8 March 2005 Associate Editor: James R. Belthoff A Telemetry Receiver Designed with The Researcher in Mind What you've been waiting for! Finally, a highly sensitive 999 channel synthesized telemetry receiver that v»eighs less than 13 ounces. Is completely user programmable ar>d offers variable scan rates over all frequencies. For each animal belr>g tracked, the large ICO display provides not only the frequency (to 100Hz) and channel number, but also a 7 character alphanumeric comment held and a digital signal strength meter. Stop carrying receivers that are the size of a lunch bo* or cost over $1500. The features and performance of the new R-1000 pocket sized telemetry receiver will impress you. and the price will convince you. Othf Ftatures Indudt: • Factory tuned to any 4MHz wide segment in the 148 174MHz Band • Very high sensitivity of -148dBm to -ISOdBm • Illuminated display and keypad for use in low light or darkness • User selectable scan rates from 1-30 seconds in 1 second steps • Rechargeable battenes operate the receiver for 12 hours and can be replaced with standard AA Alkaline batteries m tf>e held. \ Visit our website for complete specifications, operating manual and information on the R-IOOO or call our toll free number to order your receiver now. Try the New R IOOO and YoaV Impressedl 6.r 2 . 6 ‘ 1.5' (15.5cm) high, (6.6cm) wide. (3.8cm) deep. • 3 year warranty • 1 day delivery S69S.OO Please specify desired 4MHz wide segment in the 148- 174MHz band Bute Boo Sks 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, and hundreds of out-of-print titles available- ( K;i|)l