j RaptorRes. 37(3):228-235 ¸ 2003 The Raptor ResearchFoundation, Inc.

NEST AND ROOST HABITAT CHARACTERISTICS OF THE GREY-FACED IN NORTHEASTERN CHINA

WEN-HONG DENG 1 Ministry of EducationKey Laboratory for BiodiversityScience and EcologicalEngineering, College of Lij•bSciences, BeijingNormal University,Beijing, 100875, China

GAO WEI Schoolof Li•[bSciences, Northeast Normal University,Changchun, 130024, China

ZHENG GUANG-MEI Ministry of EducationKey Laboratory for BiodiversityScience and EcologicalEngineering, College of Lij• Sciences, BeijingNormal University,Beijing, 100875, China

ABSTRACT.--Thehabitat of the Grey-facedBuzzard (Butasturindicus) has diminished substantiallybe- causeof forest managementfor timber production and farmland reclamationin recent years.An un- derstandingof the characteristicsof nest and roost sitesof this is important for its conservation. We studiedGrey-faced during their breeding seasonin Zuojia Nature Reserve,Jilin Province, China, from March 1996-August 1998. This speciesselected both nesting and roosting sitesin mixed- deciduousforests that containedKorean larch (Larix olgens)more frequently than availablein the study area.Most nests were located in Koreanlarches and in Chinesepines (Pinus tabulaeformis; 70%), whereas nestsin broadleaftrees were relativelyinfrequent (30%). Eight nests(75%) were locatedon the upper third of a slope,three nests(19%) were located on the middle third, and one nest (6%) on the lower third. Buzzardsused 11 tree speciesfor roosting;however, (58%) of all roostswere locatedin three tree (Korean larch, Chinese pine, and River birch [Betulanigra] ). Higher canopyclosure and taller trees best separatedroost sitesfrom random plots with a discriminantanalysis. Mean roost height was 9.6 _+0.5 m. Roosttrees averaged11.2 -+ 0.6 m in height with mean DBH (diameter at breastheight) of 16.9 _+0.3 cm. Buzzardsselected nest and roost sitesin forestscharacterized by tall (>12 m) conifers, hardwoodunderstory, and high canopyclosure (>70%). I•Y WORDS: Butasturindicus; habitat use;, Grey-faced Buzzard; nest site;, nest tree;, roost site.

CARACTERISTICAS DEL HABITAT DE LOS DORMIDEROS Y DE ANIDACION DEL BUITRE DE CARA GRIS DEL NORESTE DE CHINA RESUMEN.--E1habitat del buitre de cara gris (Butasturindicus) ha disminuidosubstancialmente en los filtimos aftos,a causadel manejo del bosquepara la producci6n de madera y la recuperaci6nde tierras para agricultura. Una comprensi6nde las caracteristicasde los skios de nido y de los dormiderosde esta ave, es importante para su conservaci6n.Estudiamos el buitre de cara gris durante su temporada de cria en la ReservaNatural de Zuojia, Provinciade Jilin, China, desde marzo del 1996 hastaagosto del 1998. Esta especieescogi6 sitios de anidaci6n y dormiderosen los bosquesdeciduos mixtos que contenian una mayor frecuencia de Larix olgensen el •irea del estudio. I,a mayorla de los nidos se localizaronen Larix olgensy Pinustabulaeformis; 70%, mientrasque los nidosque estabanen •irbolesde hojasanchas fueron relativamentepoco frecuentes(30%). Ocho nidos (75%) fueron localizadosen el tercio superiorde una pendiente,tres nidos (19%) fueron localizadosen el tercio medio, y un nido (6%) en el tercio inferior. Los buitresutilizaron 11 especiesde •irbolescomo dormideros;sin embargo el 58% fueron localizadosen tres especiesde •irbolesLarix olgens,Pinus tabulaeformisy Betula nigra. La cobertura del dosel y los arbolesmas altos fueron separadosde los dormideros mediante la utilizaci6n del analisisdiscriminante. La media de la altura de los dormiderosfue de 9.6 _+0.5 m. Los •irbolesque sirvieron como dormideros tuvieron una altura promedio de 11.2 -+ 0.6 m, con media DAP (di•imetro a la altura del pecho) de 16.9 _+0.3 cm. Los buitres escogieronnidos y dormideros en bosquescarac-

1 E-mail address:[email protected]

228 SEPTEMBER 2003 GREY-FACEDBUZZARD HABITAT 229

terizadospor altas coniferas (>12 m), firbolesdel sub-dosel,de madera dura con copascerradas y altas (>70% alto de dosel). [Traduccitn de Casar Marquez]

The large number of published reports that de- STUDY AREA scribe habitat selection and habitat characteristics The studyarea, ca. 84 km2 in size,was located in Zuojia among atteststo the enormousvariation ob- Natural Protection Area and included the Tumengling Mountains and Zhujia Mountains ranging from the east- servedand to the biologicalimportance of this top- ern ChangBai Mountains to the western plain (126ø1'- ic. In addition to its importance, an understanding 127ø2'N,44ø6'-45ø5'E). Elevation at the siteranged from of the responseby birds to environmentalhabitat 200-500 masl. The climate is east monsoon, character- ized by hot, dry summers and cold, snowywinters. The change is necessarybefore conservationstrategies vegetation within the study area was quite diverse, al- can be developed and implemented (Schmutz though the existing forest is secondary.The most com- 1989). A theoretical framework for habitat selec- mon trees present in the study area were Mongolian oak tion has been provided by Fretwell and Lucas ( Quercusmongolica), dahur birch (Betula davurica), Man- churian linden ( Tilia mandschmica),Japanese elm ( UImus (1970), and functionaland theoreticalaspects of japonica), Scotch pine (Pinus sylvestris),Korean larch (P•- habitat selection have been summarized by Gody nus koraiensis) and Masson pine (Pinus massoniana) (1985). Within these frameworks, nest and roost (Deng et al. 1997). In the studyarea, hawthorn raspberry characteristicsare very important factorsrelated to ( Rubus crataegifolius),dahurian rose (Rosa dahurica), Ko- rean rose (Rosadoreana), willowleaf spiraea ( Spiraeasah- arian habitat selection. cifolia),ural falsespiraea(Sorbaria sorbifolia), prickly rose Among subtropicalbirds, raptors are one of the (Rosaacicularis) , amur barberry ( Berberisamurensis) , amur least-studiedgroups and relatively little is known honeysuckle (Loniceramaacki), manchur honeysuckle about their nest and roost characteristics.Grey- (Loniceraruprechtiana), and sakhalin honeysuckle(LonG- ceramaximowiczii) dominated the shrub layer. The study faced Buzzard ( indicus) is a summer mi- area consistsof ca. 30% open habitat and 70% forest grant in northeast China (Cheng 1987, Anony- habitat. mous 1988). It seemsthat most of the Grey-faced METHODS Buzzardsthat breed in northeastern China migrate Survey Methods. We surveyedthe study area at least to Okinawa, Taiwan, the Philippines, Indonesia, four times each year, 20 March-20 August, 1996-98, us- Malaysiaor nearbyregions to winter (Chang 1980, ing conventions suggestedby Newton and Marquiss Gheng 1987, Ehimekensibuet al. 1989, Deng et al. (1982) and Steenhof (1987) to describeoccupancy and 1997). This buzzard has been listed as a threatened activitiesat nesting sites.We attempted to locate all buz- zard nestswithin the studyarea. We used behavioral cues speciesin National SecondGlass Protected Species and systematicsearches of potential nest substratesto lo- in China Data Book of Endangered cate nests.We determined the sex of buzzards by body (Zheng and Wang 1998). The habitats of this spe- size and plumage characteristics(Deng et al. 1997). We establishedsix transectsin the study area (2 = 3 km, cies have been substantiallyreduced because of range = 2-5 km). Each transectwas 50 m wide and par- forest management for timber production and allel to the forest edge. Each transect was divided into farmland reclamation in recent years (Zheng and 100 m segments from one end to the other. Sites were Wang 1998). An understanding of the characteris- surveyedwithin a 4-hr period beginning 30 rain after sun- rise by walking along each side, with 5-rain stopsat each ticsof nest and roostsites of this bird is particularly 100 m interval. We walked along one side and came back •mportant for its conservation.However, very little along another one. A nesting area was consideredoccu- is known about nestingecology of Grey-facedBuz- pied if a territorial pair or evidence of a territorial pair (such as observationsof an incubating bird, nest con- zards either in China or in other areas.In this pa- struction, or nest maintenance) was observed, otherwise per we describe the nest and roost characteristics the area was classifiedas unoccupied. We located roosts of this raptor in northeastern China. Our null hy- (the perch location where a bird spendsthe night) by potheses were: (1) no difference exists between observingbuzzards at roost sites.Only roost siteswhere buzzardsremained stationaryupon initial detection were nesting sitesand randomly-placedplots within the used in analyses.We recorded eagle nest and roost lo- study area, (2) no difference exists between the cationswith GPS receiversand plotted these on geologic immediate nest site and general habitat within nest surveymaps to the nearest 10 m using Universal Trans- stands, and (3) no difference exists between roost- verse Mercator (UTM) coordinates. Most of our habitat-sampling protocol was adopted ing sites and randomly-placedplots within the from Searoans and Gutitrrez (1995). We measured site study area. characteristicsin sample plots centered on buzzard nest 230 DENG ET AL. VOL. 37, NO. 3 trees and roost trees and at random locations in forest tween groups. We used cross validation (Capen et al. habitat. Detailed vegetationswere collected in 12 nest 1986) to evaluatethe stabilityof the DA model. stands and 12 random plots. Nesting habitat data were We took detailed measurementsof all nest trees (age, collected at four sample points 25 m from each nest in height, DBH, canopy and nest height) using the same each cardinal direction. Also, four sample points were techniques used for measuring trees in sample plots. We located in each cardinal direction and at a random dis- estimatednest tree age by extracting a core samplewith tance between 100 m and 1000 m from the nest tree to an increment borer and counting the rings. We used cir- represent availablehabitat. This samplingwas designed cular statistics to estimate mean orientation of the nest to addressnest-site level habitat selection occurring with- relative to the tree trunk. We used a chi-squareanalysis in a hypothetical Grey-facedBuzzard home range. In to test for differences in tree species between nest and each nest stand, four sample points were measured with- random tree distributions.We used paired-samplet tests in both nesting and availablehabitat to increasethe num- (Zar 1984) to compare nest tree height and DBH to ran- ber of points availablefor use in the assessmentof model dom trees within nest stands. stability(Mueller-Dombois and Ellenberg1974). We cat- egorized the forest type (conifer, if the proportion of co- RESULTS refers was >70%; mixed conifer/broadleaf, if the pro- Roost-site Characteristics. Of 86 roost sites we portion of conifer and broadleaf was near equal; broadleaf forest, if the proportion of broadleaf was measured, we used 58 (one each for 30 males and >70%) and slope position (lower, middle, upper third) 28 females) as independent samples for analysis. at each site, and measured 12 habitat characteristics. At The distribution of forest typesat Grey-facedBuz- each plot center, we estimated slope aspectwith a com- zard roostsdiffered from random sitesthroughout pass,slope angle (%) with a clinometer,and relative can- the studyarea (X2 = 9.06, df = 2, P < 0.05), with opy closure (%) with a sphericaldensiometer. We mea- sured tree DBH (diameter at breast height in cm) with most roosts in the mixed conifer/river birch forest calipers, and tree height with a clinometer. We used a type. Positionof roostssites on the slope differed varible radius-plot method (Mueller-Dombois and Ellen- from a random distribution (X2 = 9.06, df = 2, P berg 1974) to estimatebasal area (m9/ha) of conifers < 0.05). Forty-sevenroosts (81%) were locatedon and broad-leaf trees. In addition, we recorded nest and the middle third, eight (14%) roostswere on the roost tree species, percent height (roost height/tree height) relative to the height of the tree, and distance upper third, and three roostswere (5%) on the from nest trees to roost trees. We used the variance of lower third of the slopes.Mean aspectat roost sites tree height and variance of tree diametersof all trees in differed fYom aspect at random sites (F = 8.94, df each sample plot as an index of forest structuralhetero- = 1, 112, P < 0.05; mean aspect= 6.8ø, mean vec- geneity. tor length = 0.60, angulardeviation = 68.5ø).Buz- Data Analysis.We pooled data among yearsafter find- ing no difference using a series of Kruskall Wallis tests zards used 11 tree speciesfor roosting, however (Zar 1984), with sequentialBonferroni adjustments(Rice over half (58%) of all roosts were located in three 1990). We compared forest types and slope position of tree species(Korean larch, Chinese pine, and river eagle sitesand random sitesusing chi-square analysis. We birch). Mean roost height was 9.6 -+ 0.5 m in the estimatedthe mean slopeaspect of eaglenests and roosts roost tree. Roost trees averaged 11.2 --+0.6 m tall using circular statistics(Batschelet 1981), and compared aspect of eagle nestsand roostswith random sitesusing in height with mean DBH of 16.9 + 0.3 cm. a Watson-Williams test (Zar 1984). We assessedunivariate Roost plots differed from random plots (MA- normality of the variablesusing skewhess,kurtosis, and NOVA; Wilk's Lambda = 0.51, F = 16.8, df = 10, probability plots. We assessedthe equality of variance of 109, P < 0.05). The t-testsindicated most variables variables between groups using an F-max test. We used differed between roost and random plots (Table logarithmic and square-roottransformations to normal- lze variables and equalize variances for variables that de- 1). Higher canopyclosure and taller treesbest sep- viated t?om normal distribution. For analyses,we only arated roostsfrom random plots in the DA (Table uscd those variables which approximated a normal dis- 2). The pooled DA correctlyclassified 78.6% of the trlbution and had comparablevariances between groups, roost and random plots (Table 3). e•ther before or after transformation. We tested the null Nest-site Characteristics.We found 12 Grey- hypothesesof no difference in variable means between faced Buzzard nests and used all nest sites as in- eagle nest and roost plots and random plots using mul- tivariate analysisof variance (MANOVA, Stevens1996). dependent samplesfor analysis(Fig. 1). The dis- For the MANOVA, we used Wilk's Lambda to compare tribution of forest types at nests differed from hnear combinations of variables between used and ran- random sites (X2 = 15.62, df = 2, P < 0.05), with dom sites.If the MANOVA was significant,we tested in- most nests in the mixed-conifer/river birch forest dividual variablesusing a seriesof t testswith sequential Bonferroni adjustments.We used discriminant analysis type. Position of nest sites on the slope differed (DA, Stevens 1996) to model data, to estimate which from an independent distribution (X2 = 11.26, df characteristics contributed the most to differences be- = 2, P < 0.05). Eight nests(75%) were locatedon SEPTEMBER 2003 GREY-FACED BUZZA•t) EAGLE HABITAT 231

Table 1. Habitat characteristicsat Grey-facedBuzzard roost sites(N = 58) and random plots (N = 58) in the Zuojia and Tumengling mountains,northeastern China, 1996-98.

MEAN + SD

VARIABLE ROOST RANDOM a tb P

Tree height (m) 13.8 _+ 1.5 9.1 _+ 1.2 5.63 <0.01 Tree DBH (cm) 28.7 -+ 11.3 21.2 +- 8.6 2.84 0.01 Tree basal area (m'•ha-•) 10.6 _+4.9 9.3 -+ 5.8 0.98 0.46c Canopy closure (%) 81.9 _+ 18.5 62.4 -+ 13.1 8.56 <0.01 Tree height variance 1.9 _+ 1.6 0.8 + 0.5 4.29 <0.01 Tree DBH variance 3.9 + 1.6 2.1 -+ 0.9 2.91 0.01 Distance from water (m) 348.6 _+57.9 401.2 -+ 345.5 0.42 0.61 Tree crown volume (m 3) 12.2 --+5.5 7.9 _+ 3.8 3.21 <0.01 Slope angle (%) 22.5 -+ 11.4 13.6 _+6.8 8.27 <0.01 Random siteswere located throughout the studyarea by using GIS. Degreesof freedom = 98. No significantdifference. the upper third of the slope, three nests (19%) Nest-tree Characteristics. Grey-faced Buzzard were located on the middle third, and one nest nestswere locatedin five tree species.Fifty percent (6%) on the lower third. Mean slopeaspect at nest (N = 6) of nests were located in Korean larches, siteswas northerly (mean aspect= 295ø, mean vec- 25% (N = 3) were in Chinesepines, and 8% each tor length = 0.37, angular deviation = 56.5ø),and were in river birch, Mongolian oak, Manchurian differ significantlyfrom random sites (F = 15.13, linden (T ilia mandschurica),respectively. Mean as- df = 1, 22, P < 0.05). pect deviation of the nests in the trees was north- Nest plots differed from random plots (MANO- westerly (mean aspect = 342.5ø, mean vector VA; Wilk's Lambda = 0.68, F = 7.96, df = 8, 23, P length = 0.51, angular deviation = 56.5ø), but did < 0.05). The t-testsindicated that four of the eight not differ from a random distribution (z = 1.2, P variables differed between nest and random plots > 0.05). The distribution of random tree species (Table 3). Larger DBH, taller trees,greater canopy differed from nest tree species(X 2 = 8.9, df = 2, closure,and greater basalarea of mature treesbest P < 0.05). Nest trees were larger, denser,and taller separated nest sites from random sites in the DA than trees randomly located within the nest stand (Table 3). (Table 4).

Table 2. Mean habitat characteristicsat Grey-facedBuzzard nest (N = 12) and random plots (N = 12) in the Zuojia and Tumengling mountains,northeastern China, 1996-98.

MEAN _+ SD

VARIABLE NEST RANDOM a tb P

Tree height (m) 15.1 +_2.2 10.6 _+ 1.7 8.24 <0.01 Tree DBH (cm) 31.9 + 9.2 25.2 + 6.7 3.13 0.01 Tree basal area (m2ha-•) 17.6 _+5.1 9.6 _+ 3.7 4.94 <0.01 Canopy closure (%) 84.5 -+ 15.6 62.7 -+ 22.0 8.76 <0.01 Tree height variance 1.5 _+ 1.1 1.3 -+ 0.6 0.87 0.42c Tree DBH variance 3.2 -+ 1.8 3.1 -+ 0.8 0.69 0.49 • Distance from water (m) 229.5 -+ 112.6 314.2 -+ 258.6 0.28 0.87 ½ Tree crown volume (m 3) 13.8 -+ 6.9 8.1 _+ 3.8 3.47 <0.01 Slope angle (%) 25.5 -+ 8.5 21.3 + 7.1 1.22 0.06• Random sites located in each cardinal direction from nest at a random distance between 100 and 1000 m. Degreesof freedom = 94. No significantdifference. 232 DENG ET AL. VOL. 37, No. 3

Table 3. Discriminant analysisresults of habitat characteristicsat Grey-facedBuzzard roost and nest plots in the Zuojia and Tumengling mountains,northeastern China, 1996-98.

ROOSTPLOTS (N = 58) NEST PI,OTS (N = 12)

MEAN POOLED DATA MEAN POOLED DATA MF•N STRUCTURE STRUCTURE MFAN STRUCTURE STRUCTURE VARIABLE RANK COEFFICIENT a COEFFICIENT a RANK COEFFICIENT a COEFFICIENT a

Tree height (m) 1.9 0.57 0.64 2.5 0.54 0.62 Tree DBH (cm) 6.3 0.21 0.33 6.6 0.29 0.28 Tree BA (m2ha-•) 8.7 0.16 0.13 9.2 0.11 0.09 Canopyclosure (%) 1.2 0.76 0.81 1.0 0.80 0.83 Tree height variance 2.7 0.53 0.58 3.6 0.35 0.59 Tree DBH variance 4.9 0.29 0.45 5.7 0.39 0.36 D•stance from water (m) 9.9 0.04 0.03 9.9 0.07 0.01 Tree crown volume (m '•) 3.1 0.42 0.53 3.9 0.55 0.44 Slope angle (%) 1.3 0.68 0.77 1.8 0.68 0.72 Structure coefficient is correlation between a singlevariable and discriminant function.

DISCUSSION tributed to the forest structure. Raptors have Most Grey-faced Buzzard nest and roost sites shown some selection for slopes,but slope orien- were found on the upper third of north-facing tation patternswere not alwaysconsistent (Klopfer slopes. This corresponded to the distribution of 1965, Delannoy and Cruz 1988, Mcintyre and Ad- mature mixed conifer/river birch forests on the ams 1999, Nijman et al. 2000). In selecting nest study area. In addition, most nest and roost sites sites,Grey-faced Buzzards avoid slopeswith south- had an understoryof Mongolian oak, which con- ern aspects.We suggestthat the birds are selecting exposuresto insulatethe nest againsthot weather conditions during incubation and to place their nests close to hunting habitat. Data from accipiter studies (Shuster 1980, Moore and Henny 1983, Speiserand Bosakowski1987) alsoshow an obvious avoidanceof southern slope aspectsfor nesting. The majorityof nests(75%) were built in coni- fers rather than in deciduous hardwood trees. Grey-facedBuzzards preferred to build their nests in dense canopy closure of Korean larches and •} 1: 33 000 000 Chinese pines (Table 4). Broad-leaf trees were rarely used as nest trees despite the preference for broad-leaf trees in nest stands. We propose that broad-leaftrees are partly avoidedbecause they sel- dom have a larger triple and quadruple primary crotches.All nest trees were generally greater in DBH and height when comparedto thoserandom trees (Table 4). Large raptors require large tree- forks to place the nest in (Newton 1979, Mader

N 1982). Grey-faced Buzzard nests were alwayssitu- ated below or in the bottom quarter of the tree 15 km canopy,which had an open branch structure.Rap- tors nest in the lower quarter of the canopy with Figure 1. The studyarea and locationsof the nestsites open branch structureprobably to allow the buz- (solid squaresindicate nest location) of the Grey-faced zards access to the nest both above and within the Buzzard in Zuojia Nature Reserve, northeastern China. canopy; this positioning may also hide the nest SEPTEMBER 2003 GREY-FACEDBUZZARD EAGLE HABITAT 233

Table 4. Characteristicsof Grey-facedBuzzard nest (N = 12) and random trees (N = 12) in the Zuojia and Tu- mengling mountains,northeastern China, 1996-98.

MEAN -4- SD

VARIABLE NEST TREES RANDOM TREESa tb P-VALUE

Age (yr)c 56.2 _+5.1 54.8 _+3.4 0.16 0.91 Tree height (m) 18.8 _+3.1 14.3 _+2.6 4.23 <0.01 Tree DBH (cm) 32.5 -+ 9.7 24.9 -+ 9.2 3.44 <0.01 Tree crown volume (m 3) 3.6 + 1.1 2.4 - 0.7 3.16 <0.01 Nest height (m) 12.5 -+ 3.4 ------Random trees were located a random distance (10-100 m) in a random direction from the nest tree. t valuesfrom matched pair test, significantat ot = 0.05. Estimatedby extractingat core samplewith an increment borer and counting the rings. from potential predators (Selas 1996, Malan and each nest tree to measure the characteristics of nes- Robinson 2001). Cerasoli and Penteriani (1996) tlings in every five daysduring nestling period, the suggestedthat tree-nesting raptors might select buzzardsdid not change their roost sites.Belthoff trees for their size and structural features, such as and Ritchison (1990) suggested that Eastern a tall and open canopy, that allow unobstructed Screech-Owls (Otus asio) did not use the same roost accessto nests.In this study,however, Grey-faced site on successivedays, and suggestedreuse of sites Buzzard seem to prefer dense to open canopy for could attract potential predators. However, repeat- nesting. ed use of roost sitesby Grey-facedBuzzards is very In our study,Grey-faced Buzzards selected nest common, probably because they have few preda- and roost sites primarily in mixed conifer/river tots in the studyarea (Deng et al. 1997). Also,suit- birch forestsin the largest and tallest trees, with able roosts with dense foliage may be extremely relativelyhigh canopyclosure. Forests composed of limited in the study area. larger treeswith high variation in tree heightsmay The forest tracts inhabited by buzzards were provide an accessibleprey base for buzzards and large in extent as exemplifiedby the comparatively provide protection from potential disturbance long distances (>15 km) to human habitation from human activities. This kind of vegetation (Deng et al. 1997). Nest siteswere alsomore often structure allows more small animals to occur and found in dense forests, where little disturbance of also preventshumans from entering. In addition, any kind occurredand that may have had a greater Grey-facedBuzzards are heat intolerant and may density and diversityof prey speciesthan edge ar- require mature, multi-storiedforests to thermoreg- eas (Kojima 1982, Ricketts and Ritchison 2000). ulate effectively(Feng 1991,Deng et al. 1997).The However,we found most nest sitescloser to logging middle partition of north-facing slopes, forested roads (or discernable trails) than random sites: six with multi-storiedmixed conifer habitat, may pro- nests (50%) were very close (<30 m) and four vide suitable cooler microclimates for buzzards. (33%) were within 60 m. In our study,forests roads However, habitats such as hardwood forests seem often represented the break in deep contiguous to lack the complete vertical structure of most typ- forests.In dense contiguousforests, logging roads ical nest and roost sites. may aid the buzzardsby providing open flyways. Trees frequently used for roosting were often On several different occasions, we observed buz- thosewith densefoliage or high canopyclosure. In zards flying, perching, and plucking prey along contrast, available trees rarely or never used for logging roads in the studyarea. Speiserand Bosa- roosting appeared to provide little cover.In addi- kowski (1987) found Northern Goshawks(Accipzter tion to concealingbirds from potential human dis- gentilis)often nested near logging roads in north- turbance, the dense cover of most roost sitesprob- ern New Jersey and southern New York. Raptors ably provided a favorable microclimate.We found often nest near logging roads or in an exposedpo- that individuals often used the same roost site on sition that allows easyaccess to and from the nest successivedays, although our presence may have to deliver sticksand prey. disturbed the buzzards.Although we climbed up Moore and Henny (1984) pointed out the ira- 234 DENG ET AL. VOL. 37, NO. 3 portance of past experience (successor failure) in CAPEN,D.E., J.W. FENWICK,D.B. INKLEY,AND A.C. BOYN- nest site selectionby raptors, but at least for first-• TON. 1986. Multivariate models of songbird habitat in time nesters,the role of early experience and im- New Englandforests, Pages 171-176 inJ. Verner,M.L printing to the natal habitat may be of primary im- Morrison, and CJ. Ralph [EDS.],Wildlife 2000: mod- portance (Newton 1979). For a variety of eling habitat relationshipsof terrestrial vertebrates. Univ. of Wisconsin Press, Madison, WI U.S.A. nonpasserines,studies of marked individualshave CERASOLI, M. AND V. PENTERIANI. 1996. Nest-site and ae- shown that nesting sites are more likely to be re- rial point selectionby Common Buzzards( buteo) occupied in years following successfulnesting at- in Central Italy.J. RaptorRes. 30:130-135. tempts and abandonedafter nestingfailures (e.g., CHANG, W.F. 1980. Travellers who do not need visas, the Newton and Marquiss 1982, Marks 1986, Thors- Grey-facedBuzzard Eagle. Environment1:58-64. trom and Quixch/m 2000). We did not quantifyfor- CHENG,TSO-HIN. 1987. A synopsisof the avifaunaof Chi- aging habitat of the buzzardsin our study area. na. Science Press,Beijing, China. Available information indicated that Grey-faced CHING, H.C., S.T. Yu, AND C.H. CHIANG.1989. A survey Buzzardscommonly forage in open areas(e.g., pas- of Grey-faced Buzzard Eagle hunting in ManChou tures,marshes, paddy fields) where they find most Area. KenTing National Park, Taipei, China. of their prey (Kojima 1982, Ehimekensibuet al. CODY, M.L. 1985. Habitat selection in birds. Academic 1989). Their foods mainly include frogs, reptiles, Press, New York, NY U.S.A. rodents,and somebirds (Cheng 1987, Ching et al. DELANNOY,C.A. ANDA. CP,uz. 1988. Breeding biology of the Puerto Rican Sharp-shinnedHawk (Accipiterstria- 1989, Severinghaus1991). Accordingto our obser- tus venator). Auk ]05:649-662. vations, Grey-faced Buzzards often hunt from DENt3,W.H., X. WANg,AND W. GAO.]997. Breedinghab- perches,typically at a top of dead tree; once de- itat selectionof the Grey-facedBuzzard Eagle. J. North- tected the buzzards then dive down to capture east Normal Univ. 9:57-6]. prey. Becauseour study did not involve marked EHIMEKENSIBU,KOCHIKENSIBU, OITAKENSIBU, AND MIYAZA- Grey-faced Buzzards among years, whether the KIKENSIBU.]989. The migration route of Grey-faced same individual birds reoccupied nestsis open to Buzzard Eagle Butasturindicus from Shikoku to Kyu- question. Future studies should examine Grey- shu. Strix 6:]25-]3]. faced Buzzard Eaglesin more detail and should FENG,H.L., G.Q. XIANG, AND K.Q. HANG. 199]. Obser- address the nature and extent of individual varia- vation on breeding Grey-facedBuzzard Eagle. Chi- tion in habitat use.Additional factorssuch as prey, nese Avian Studies, No. 3., Chinese Science Press, Bei- density, prey accessibility,and competition with jing, China. FRETW•LL, S.D. AND H.L. LUCAS. ]970. On territorial be- other raptors need to be addressedin future stud- havior and other factorsinfluencing habitat distribu- ies of the nestling ecologyof Grey-facedBuzzards. tion in birds. I. Theoreticaldevelopment. Acta Biotheor ACKNOWLEDGMENTS 9:]6-36. KOJIMA,Y. ]982. Territory and territorial behavior of the We thank Liu Yang, Yan-hui Li, Ren-kai Song, Lu-ye Tang, Ya-meiYu for help in the field. Wei-dongLuo and Grey-facedBuzzard Eagle Butasturindicus. Tori 30: Bao-xue Zhang were very patient in helping with data ]]7-]47. collection. Earlier drafts of this manuscript were im- KLOPFEV,,P. ]965. Behavioralaspects of habitat selection. proved by commentsfroln Zheng-wangZhang and Yan- a preliminary report on foliage preferencesof birds. yun Zhang. The staff of Zuojia Natural Protection Area Wilson Bull. 77:376-38]. providedfacilities during the studyperiod. We especially MAD•V,,WJ. ]982. Ecologyand breeding habitats of the thank the editorsand the anonymousreferees of theJ0ur- Savanna Hawk in the Llanos of Venezuela. Condor84. nal of RaptorResearch for their excellent and helpful re- 261-27]. marks on the manuscript. •, G. AND E.R. ROBINSON.2001. Nest-site selection LITERATURE CITED by Black SparrowhawksAccipiter melanoleucus: imph- ANONYMOUS.1988. A study of the northward migration cationsfor managingexotic pulpwoodand sawlogfor- of Grey-facedBuzzard passingPaKua Moun- estsin South Africa. Environ.Manage. 28:195-205. tain in the spring.Ann. TaipeiWild. Bird Soc.12:12- MAm4S,J.S. 1986. Nest-sitecharacteristics and reproduc- 18. tive successof Long-earedOwls in southwesternIda- BATSCHELET,E. 1981. Circular statisticsin biology.Aca- ho. Wilson Bull. 98:547-560. demic Press, San Francisco, CA U.S.A. MCINTYRE,C.L. AND L.G. ADAMS. ]999. Reproductive BELTHOFF,J.R. ANDG. R•TCHISON.1990. Roostingbehav- characteristicof migratory Golden Eaglesin Denah ior of postfledgingEastern Screech-Owls. Auk 107: National Park, Alaska. Condor ]0]:]]5-]23. 567-579. MooRE, K.R. AND CJ. H•NNY. 1983. Nest site character- SEPTEMBER 2003 GREY-FACED BUZZARD EAGLE HABITAT 235

isticsof three coexistingAccipiter hawksin northeast manipulatedvariation in tree density.J. Avian Biol.27: Oregon. RaptorRes'. 17:65-76. 56-62. -- ANDJ. HENNY.1984. Age specificproductivity and SEVERINGHAUS,L.L. 1991. The status and conservation of nest site characteristicsof Gooper'sHawks. Northwest Grey-facedBuzzard-Eagles and Brown Shrikesmigrat- Sci. 58:290-299. ing through Taiwan. Pages 203-223 in T. Salath6, MUELLER-DOMBOIS,D. AND H. ELLENBERG.1974. Aims and [ED.]. ICBP Tech. Publ. 12. Taiwan, China. methods of vegetation ecology.John Wiley and Sons, SHUSTER,W.C. 1980. Northern Goshawknesting require- New York, NY U.S.A. ments in the Colorado Rockies. West. Birds' 11:89-96 NEWTON,I. 1979. Populationecology of raptors.Buteo SPEISER,R. AND T. BOSAKOWSKI.1987. Nest-site selection Books, Vermillion, SD U.S.A. by Northern Goshawksin northern New Jersey and --AND M. MARQUISS.1982. Fidelity to breeding area southeastern New York. Condor 89:387-394. and mate in sparrowhawksAccipiter nisus. J. Anim. Ecol. STEENHOF,K. 1987. Assessingraptor reproductivesuccess 5:327-341. and productivity.Pages 157-170 in B.A. Giron Pen- NIJMAN,V., S. VAN BALEN,AND R. SOZER.2000. Breeding dleton, B.A. Millsap, K.W. Cline, and D.M. Bird biology of Javan Hawk-Eagle 5pizaetusbartelsi in west [EDS.], Raptor management techniquesmanual. Sci- Java, Indonesia. Emu 100:125-132. entific and Tech. Series No. 10, Natl. Wildl. Fed., PriCE,W.R. 1990. A consensuscombined p-value test and Washington,DC U.S.A. the family-widesignificance of component tests.Bio- STEVENS,J. 1996. Applied multivariate statisticsfor the metrics 46:303-308. social sciences, 3rd Ed. Lawrence Erlbaum Publishers, RICKETTS,M.S. AND G. RITCHISON.2000. Nesting success Mahwah, NJ U.S.A. of yellow-breastedChats: effect of nest site and terri- THORSTROM,R. AND A. QU•XCHAN.2000. Breeding biol- tory vegetation strucure. l/VilsonBull. 112:510-516. ogy and nest site characteristics of the Bicolored SCH•tUTZ,J.K. 1989. Hawk occupancyof disturbedgrass- Hawk in Guatemala. Wilson Bull. 112:195-202. lands in relation to models of habitat selection. Condor ZAR, J.H. 1984. Biostatistical analysis. Prentice-Hall, En- 91:362-371. glewood Cliffs, NJ U.S.A. S•s, M.E. AND RJ. GUTIERREZ.1995. Breeding hab- ZHENG,G.M. AND Q.S. WANG. 1998. China red data book itat of the Mexican Spotted Owl in Tularosa Moun- of endangered animals. Chinese Science Press,Bm- tains, New Mexico. Condor97:944-952. jing, China. SELAS,V. 1996. Selection and reuse of nest standsby spar- rowhawk (Accipiternisus) in relation to natural and Received 2 October 2002; accepted 18 May 2003