sign in sign up
<<
Home , Common buzzard, Northern goshawk

J. RaptorRes. 31(1):16-25 ¸ 1997 The Raptor ResearchFoundation, Inc.

NEST-SITE SELECTION BY FOUR SYMPATRIC FOREST RAPTORS IN SOUTHERN

VmAa SELAS Departmentof Biologyand Nature Conservation,Agricultural University of Norway, P.O.Box 5014, N-1432 Ks, Norway

ABsTR•CT.--Differencesbetween 0.1 ha nest-siteplots of Honey ( apivorus), European Sparrowhawks(Acdpiter nisus), Northern Goshawks(A. gentilis)and Common Buzzards (Buteobuteo) were comparedto randomlysampled 0.1 ha control plotswithin a 400 km2 area with 80% forestand <2% agriculturalland in southernNorway. At Honey Buzzardnest sites,forests were more productive than in control plots and there wasa higher proportion of ,older and a higher density at Northern Goshawknest sitesthan in control plots. Nests of European Sparrowhawkswere also in siteswith higher tree densitythan expected.Common Buzzardnest siteswere situatedin steeperterrain than control plots and more often had a southern aspect.For sparrowhawks,nesting in forestswith high tree densitymay be an adaptation to avoid goshawksand (Martesmartes) which are their main nest predators. For the larger ,nest-site selection may be a responseboth to nest predation risk, microclimate,foraging habitat and food supply. KEYWORDS: HoneyBuzzard; European Sparrowhawk; Common ; ; nisus; Accipiter gentilis; buteo; Pernis apivorus;forest;, nest-site selection; Norway.

Seleccitn del nido de cuatro rapacesde bosquesin que no aparean en el sur de Norway. RESUMEN.---Diferenciasentre 0.1 ha parcela de sitio de nido de Pernisapivorus, Acdpter nisus, A. gentilis y Buteobuteo fueron comparadoscon muestrasalazar 0.1 ha parcelasmanejadas dentro de una area de 400 km2 con 80% bosquey <2% tierra agricolaen el sur de Norway.En nidos de Pernisapivorus, los bosquesfueron mas productivo en las parcelasmanejadas y habia una proporcitn alta de Picea,firboles maduros y densidad alto de firboles en nidos de A. nisustambi•n estaban en sitios con densidad alta de firbolesmas de 1oque esperfibamos.Nidos de B. buteoestaban situados mas en terreno abrupto que en parcelasmanejadas y con frecuencia tenla aspectodel sur. Para A. nisus, nidos en el bosque con densidad alta de firboles puede ser un adaptacitn para evitar A. gentilisy Martes martesque son su principal depredador de nido. Para la especiemas grande, la seleccitn del nido puede ser reaccitn a riesgode depredador al nido, microclima,hfibitat de forraje y suministrode cornida. [Traduccitn de Rafil De La Garza, Jr.]

Breeding pairs of raptors use relatively large ar- hand, dense foliage may decreasethe possibilityfor eas, and thus have a good opportunity to select breeding to detect and escape from preda- nesting placesthat maximize the probability of suc- tots (G6tmark et al. 1995). Thus, selection of nest cessfulbreeding and lifetime reproduction (New- site may be a trade-off between concealment and ton 1979). Interspecificdifferences in nest-sitese- opportunities to escapeor attack predators, which lection may be due to differences in body size and also depend on flight ability, body size or other flight performance of different species,but it can characteristicsof the species.Selection may alsobe also be due to interspecificdifferences in nest pre- affected by a trade-off between current and future dation risk, climatic conditions during breeding reproduction, since short-lived specieswith large and feeding habits (Newton 1979,Janes 1985), or brood sizes have more to lose when nesting at- to interspecificcompetition for nest sitesand ter- tempts fail than long-lived species with smaller ritories (Newton 1979). brood sizes. For several species,dense foliage close to Cover may also be an important factor since it the nest both reduces the rate of detection, and can shield nests from wind or rain and limit ex- impedes the ability of predators to hunt in the vi- cessive nocturnal radiation loss or excessive diur- cinity of the nest (Martin 1993). On the other nal heat-gain from solar radiation (Walsberg1985).

16 M_•CH 1997 NEST-SITE SELECTION BY FOUR RA_PTORS 17

Protection from thermal extremesmay be the most 400 km •and is situated 50-300 m a.s.l. and 10-30 km important factor in nest-siteselection by medium- inland from the , in the boreo-nemoral zone (Abra- hamsen et al. 1977). The climate is suboceanic,and snow and large-sizedraptors where nest predation is low usually covers the ground from December-April. The (Newton 1979,Janes 1985). At higher latitudes,the studyarea is hilly and sharplyundulating. It is dominated timing of breeding in these birds should be im- by forests(80%), with scatteredlakes (10%), bogs (5%) portant, since early breeders are faced with more and lessthan 2% agriculturalland. Forestsare character- ized by a fine-grained mosaicof young-, medium- and severeclimatic conditionsthan thosespecies which old-aged coniferous, mixed and stands,with begin nesting later in spring. Scotspine (Pinussylvestris), Norway spruce (Piceaatnes), If prey are not evenlydispersed throughout the (Quercusspp.), (Populustremula) and landscape,raptors should selectnest sitesclosest to (Betulaspp.) the dominant tree species. the best hunting areasin order to reduce time and Forestrybased on clear cutting, replanting and thin- ning of the regrowth was introduced to the area in the energy connectedwith foraging. Thus, local varia- 1950s.At the time of my study,approximately 30% of the tion in the availabilityof food may influence the area had been clear-cut,with most regeneration<20 yr. nest-siteselection (Janes1985), and explain inter- The area is dividedinto numeroussmall ownerships with specific differences in nesting habitat. management of forests controlled by each of the land owners.Most of the properties are managed to provide In ,four raptor specieshunt and a mosaicof forest types.Thus, there is a heterogeneous nest in forest-dominatedlandscapes. The - environmenton a small scale,but a homogeneous,frag- an Sparrowhawk (Accipiternisus; mean body mass mented environmenton a large scale. male 150 g, female 260 g) is the main predator on The study area was searchedfor nest sites each year small birds (Sulkava 1964a, Selfis 1993), while the (cf. Forsman and Solonen 1984), and habitat variables were describedat one nest site in each nestingterritory Northern Goshawk (A. gentilis;mean body mass located. If possible,the nest site used in 1988 was select- male 870 g, female 1330 g) primarilyfeeds upon ed. In territories where the 1988 nest was not found, I larger bird speciesand (H6glund 1964, usuallydescribed the nest site usedin 1989. Alternatively, Sulkava 1964b, Widan 1987, Selfis 1989). The Com- the nest site used closest in time to 1988 or 1989 was described.The breeding densityof goshawksincreased mon Buzzard (Buteobuteo; mean body massmale during the time of the study.To get a larger number of 740 g, female 1100 g) is a generalistpredator that nest sitesof this species,I first selectedone nest site from respondsfunctionally to changesin populationsof each of the nesting territories used since 1985. Then, I its (Microtusspp.) prey (Suomus1952, Spidso selectedone nest site in each of the 11 new nesting ter- ritories establishedduring 1986-88, even though these and Sel•s 1988), while the Honey Buzzard (Pernis territories substitutedfive of the existingones. Since the apivorus;mean body massmale 750 g, female 910 goshawksin the five old territories had all been illegally g) mainly feeds on the larvae and pupae of social shot by game keepers, I regarded the data to be inde- hymenoptera species(Holstein 1944, Hagen and pendent. Thus, I describeda total of 48 nest site plots of Bakke 1958, It/imies and Mikkola 1972). the sparrowhawk,30 of the goshawk,50 of the ,and 21 of the Honey Buzzard. Severalauthors have describednest sitesused by Control plotswere describedduring 1989. Aerial pho- sparrowhawks (Tinbergen 1946, Holstein 1950, tographsof the studyarea taken in spring 1989 (scale1: Hald-Mortensen1974), goshawks(Holstein 1942, 15000) were covered by a grid with 100 numbered inter- Dietzen 1978, Link 1986), Common Buzzards sectionsof which two were randomly selectedas control plots. Out of 122 selectedpoints, 80 (65.6%) were locat- (Holstein 1956, Kntwer and Loske 1980, Solonen ed in forests>20 yr old and 25 (20.5%) were in forests 1982, Jedrzejewskiet al. 1988, Hubert 1993) and <20 yr old (clear-cutsand regrowth), while 9 (7.4%) Honey Buzzards (Holstein 1944, Arecoif et al. were on lakes,4 (3.3%) on bogs,and 4 (3.3%) on agri- 1994) in Europe. However, no one has compared cultural land or developed areas. Measurements were made only in control plotsin habitatsapparently suitable nest-siteselection of sympatricpopulations of these for raptors (i.e., forests >20 yr old, N = 80). speciesin a continuousforest habitat. My aim was Each of the nest site plots and the control plots cov- to study the importance of different habitat vari- ered 0.1 ha within a circle with a radius of 17.8 m. In ableson nest-siteselection of thesespecies by com- nest site plots, the nest was in the center of the circle. paring habitat variablesfrom plots at nest siteswith The followinghabitat variableswere used: those from plots placed randomly in the study 1) Site type, determined from the plant community area. (Kielland-Lund 1981, 1994). Plots dominated by Bar- bilophozio-Pinetumor Vaccinio-Pinetumwere classifiedas STUDY AREA AND METHODS siteswith poor productivity,plots dominated by Leu- cobryo-Pinetum,Eu-Piceetum myrtilletosum, or Populo-Quer- The studywas conductedfrom 1985-93 in southern cetumwere classified as sites with intermediate pro- Norway (58ø 43'N, 8ø44'E). The studyarea coversabout ductivity, and plots dominated by Melico-Piceetumtypz- 18 SE•S VOL. 31, No. 1

Table 1. Test results (upper, right) and P-values(lower, left) of correlation analysesof habitat variablesfrom ran- domly-sampledcontrol plots (N = 80). Categoricalvariables were tested againsteach other by use of contingency table analysis(X e value given) and againstcontinuous variables by useof Mann-WhimeyU-test (two categories, U-value given) or Kruskal-Wallistest (more than two categories,H-value given), while continuousvariables were testedagainst each other by use of Spearmanrank correlation (correlation coefficientgiven).

SITE FOREST FOREST TREE ALTITUDE TYPE TYPE AGE DENSITY SLOPE ASPECT CATEGORY

Site type (3 categories) 40.60 5.78 28.51 9.70 4.97 5.52 Foresttype (5 categories) <0.01' 1.79 18.12 22.17 11.25 17.30 Forestage (continuous) 0.06 0.77 -0.26 0.01 770.5 2.79 Tree density(continuous) <0.01' <0.01' 0.02* 0.13 775.5 7.67 Slope (continuous) <0.01' <0.01' 0.91 0.25 778.0 4.00 Aspect(2 categories) 0.08 0.02* 0.81 0.85 0.87 1.30 Altitude category (3 categories) 0.24 0.03* 0.25 0.02* 0.14 0.52 * Statisticallysignificant.

cum, Melico-Quercetum,Alno incanae-Prunetumpadi or 5) Slope,measured from 0-100ø. Ulmoglabrae-Tilietum cordatae were classifiedas sites 6) Aspect, defined as one of two categories:north (1- with the highest productivity. 100ø , 301-400ø ) or south (101-300ø). Nest-siteplots 2) Foresttype, defined according to % pine and spruce and control plotswith slopes<5 ø were omitted. trees with diameters>7 cm at breastheight (DBH, 7) Altitude, definedas three altitudepossible categories 1.3 m aboveground). Pine forestwas >50% pine and in relation to the altitude variation within a radius of sprucewith pine >67%. Mixed coniferousforest was I km from the plot. Plotswere assignedto the lower >50% pine and sprucewith pine and spruce-<67%. altitude zone if situated in the lower third of the al- Spruceforest was >50% pine and sprucewith spruce titude difference between the lowest and highest >67%. Mixed forestwas 25-50% pine and spruce.De- point within this area. Middle and upper altitude ciduousforest was <25% pine and spruce. zones were assignedcorrespondingly. 3) Forest age, defined as the mean age of four trees judged to representthe age of all treeswith DBH >7 When consideringthe randomly-sampledcontrol plots, cm. Ages were measuredusing an increment borer at several of the habitat variableswere highly correlated (Table 1). To control for the effect of these correlations breast height. 4) Number of trees,regardless of specieswith DBH >7 when comparingnest site plots and control plots, I used cm. likelihood-ratiotests (SAS 1994) in a logisticregression model, with nest-siteplots and control plots asresponses and all habitatvariables as explanatory variables (cf. Man- Table 2. Results (P-values) from Likelihood-Ratio tests ly et al. 1993).

•n a logisticregression model with nest-siteplots and ran- RESULTS domly-sampledcontrol plots (N = 80) as responses,and all habitatvariables as explanatory variables. R e is the pro- Site Type and ForestType. Of sparrowhawknest- portion of variation that is explained by the logisticre- site plots, none were on siteswith poor productiv- gressionmodel. ity, 66.7% were on intermediatesites and 33.3% in the highest productivity sites. Corresponding val- SPARROW- COMMON HONEY ues for goshawknest-site plots were 16.7%, 73.3% HABITAT GOSHAWK BUZZARD BUZZARD and 10.0%; for Common Buzzard 16.0%, 60.0% VARIABLES (N = 48) (N = 30) (N = 50) (N = 21) and 24.0%, and for Honey Buzzard0.0%, 57.1% Site type 0.58 0.80 0.54 0.004* and 42.9% comparedto 42.5%, 47.5% and 10%for Forest type 0.38 0.023* 0.23 0.35 control plots. When controlling for effects of cor- Forest age 0.94 0.027* 0.39 0.50 relations between all habitat variables, there was a Tree density <0.001' 0.013' 0.15 0.10 significant difference between Honey Buzzard Slope 0.82 0.98 <0.001' 0.31 nest-siteplots and control plots, while the other Aspect 0.92 0.06 0.037* 0.05 speciesdid not differ from the control plots (Like- Altitude category 0.53 0.47 0.20 0.59 lihood-ratio tests, Table 2). 0.90 0.27 0.34 0.47 The distribution of goshawknest-site plots in dif- * Statisticallysignificant. ferent forest types differed significantly from that MARCH 1997 NEST-SITE SELECTION BY FOUR RAPTORS 19

lOO

•,, Sparrowhawkj. o •. Goshawk

• 50

25,

o o 25 50 75 100 0 25 50 75 1 O0

% pine % pine lOO- 1oo,,, I • Common 80-J, •' Honey 75t_ •. Buzzard [ •. Buzzard o5, ' /

i 0 25 50 75 1 O0 0 25 50 75 100

% pine % pine 100-• 100- I •, ControlPlots • Forest • Type

50 • 50-

2 25- •.mixedX pine •

v ,i- ---i i -t- 0 O 25 50 75 1 O0 0 25 50 75 1 O0

% pine % pine Figure 1. Distributionof nest-siteplots of European Sparrowhawks,Northern Goshawks,Common Buzzardsand Honey Buzzards,and randomly-selectedcontrol plots on different forest types.Forest types were defined according to the frequencyof pine and spruceamong all trees>7 cm in breastheight (1.3 m aboveground). The sizeof each plot is 0.1 ha.

of control plots, with a higher proportion of nest tween habitat variables were adjusted for (Likeli- sites in spruce forests (Table 2, Fig. 1). Sparrow- hood-ratio tests,Table 2, Fig. 1). hawk, Common Buzzard and Honey Buzzard nest- Forest Age and Tree Density. The mean forest site plots did not differ from control plotswith re- age was 36.8 - 18.5 (SD) yr in nest-siteplots of spectto forest type when effectsof correlationsbe- sparrowhawks,99.3 + 19.1 yr in thoseof goshawks, 20 SE•.s VoI•. 31, No. 1

. Sparrowhawk Goshawk

•,? ß o :" =!:: :'::' '

o i • • i i i i i i i • i

3oo- CommonBuzzard _ Honey Buzzard

200 -

o ; i i i i i i • ] i [ i 25 50 75 lOO 125 15o 25 50 75 100 125 150

Forest age

Figure 2. Forest age and number of trees (>7 cm in breastheight) in nest-siteplots of European Sparrowhawks, Northern Goshawks,Common Buzzardsand Honey Buzzards(solid squares),and in randomly-selectedcontrol plots (open squares).The size of each plot is 0.1 ha.

98.5 -----20.4 yr in those of Common Buzzardsand Sparrowhawk nest sites were on south-facing 86.7 + 28.3 yr in thoseof Honey Buzzards.When slopes 34.1% of the time while 63.3%, 76.0%, using likelihood-ratio tests,only nest sites of gos- 33.3% and 53.8% of goshawk,Common Buzzard, differed significantly from control plots, Honey Buzzard, and control plots were on south- where the mean forestage was90.7 __-29.0 yr (Ta- facing slopes,respectively. Only nest sitesof Com- ble 2, Fig. 2). mon Buzzards were on south-facing significantly The mean number of trees was 190.4 --- 47.7 in more than control plots (Likelihood-ratio tests,Ta- nest-siteplots of sparrowhawks,84.9 + 27.7 in ble 2, Fig. 3). those of goshawks,73.3 --- 19.9 in those of Com- None of the nest sitesof the four raptor species mon Buzzardsand 86.4 + 33.4 in those of Honey differed significantlyfrom control plots in termsof Buzzards.The number of trees in nest-siteplots of their altitude (Likelihood-ratio tests,Table 2, Fig. sparrowhawksand goshawkswas significantly high- 4). er than in control plots, where the mean number DISCUSSION of trees was 62.3 --- 26.3 (Likelihood-ratio tests, Ta- ble 2, Fig. 2). Site Type and ForestType. Only the Honey Buz- Topographical Variables. The mean slope was zard showeda significantpreference for nestingin 8.6 + 5.6ø in nest-site plots of sparrowhawks,15.1 sites with the highest productivity. This finding + 9.1ø in those of goshawks,28.6 ñ 13.3ø in those agreedwith that ofAmcoffet al. (1994). Unlike oth- of Common Buzzards, 16.3 ___7.2 ø in those of Hon- er raptor species,Honey Buzzardmales do not pro- ey Buzzardand 15.0 ___10.5 ø in controlplots. Only vision femaleswith food during the egg-layingand nest sites of Common Buzzards differed signifi- incubationperiod (Holstein 1944), possiblybecause cantly from control plots (Likelihood-ratio tests, their prey are too small to be profitablytransported Table 2, Fig. 3). to the nest. Becauseof small size of its prey, short MARCH 1997 NEST-SITE SELECTION BY FOUR RAPTORS 21

Sparrowhawk Goshawk N N

W E W E

N

S S

Control Plots w E

N N

S

W E W E

$ $ Common Buzzard Honey Buzzard

Figure 3. Slope (0-100ø) and aspect (NE, SE, SW, NW) of nest-siteplots of European Sparrowhawks,Northern Goshawks,Common Buzzardsand Honey Buzzards.The distancefrom the origin reflectsthe slopewhile the direction reflectsthe aspectof the plot. The size of each plot is 0.1 ha. distancesbetween nesting and foraging areasduring urogallus,Swenson and Angelstam 1993) in spruce incubation may be especiallyvaluable for this spe- forest. However, preference for spruce may simply cies.Highly productiveforests may be importantfor be related to the fact that it gives the best cover Honey Buzzardsbecause they support high densities and thus the bestprotection against the main pred- of juvenile (Tiainen 1981, Helle 1985, ator of the goshawk,the (Bubobubo, Ut- Stokland1994) which appear to be important prey tendtrfer 1952, Mikkola 1983). in early stagesof the breeding season(Amcoff et al. Forest Age and Tree Density. Goshawknest sites 1994). These forestsalso support high biomassof were situated in older forests than control plots. (Birkemoe 1993, Stokland 1994) on Old forest is an important hunting habitat for the which Honey Buzzardsmay also rely. goshawk (Widtn 1989) and it provideslarge trees Preference for forest type wassignificant only for for nest building (Dietzen 1978, Anonymous1989, the goshawk,which selectedspruce forest for nest- Siders and Kennedy 1996, Squires and Ruggiero ing. This preference may be related to the larger 1996). Goshawk nests were also found in forests number of important winter and spring prey spe- with a higher tree densitythan control plots. Gos- cies such as (Sciurusvulgaris, Andrtn and hawksmay reduce the risk of predation by nesting Delin 1994), Hazel (Bonasa bonasia,Swen- in dense forests,since Eagle prefer to hunt son and Angelstam1993), and Capercaillie(Tetrao in open or semi-openlandscapes (Mikkola 1983). 22 Svs•s VoL. 31, No. 1

• Lowerzone 'øø1 '[] Middlezone • 751 [] Upperzone

r- 50

Sparrowhawk Goshawk Common Honey ControlPlots n = 48 n = 30 Buzzard Buzzard n = 80 n=50 n=21

Figure 4. The distributionof nest-siteplots of European Sparrowhawks,Northern Goshawks,Common Buzzards and Honey Buzzards,and of randomly-selectedcontrol plotsby altitudecategories. A plot wasassigned to the lower altatudezone if situated in the lower third of the altitude difference between the lowest and highest point within 1 km from the plot, to the middle zone if situatedin the middle third of this altitude difference, and to the upper zone if situated in the higher third.

For the smallest speciesinvestigated, the spar- Goshawksand pine martens are also important rowhawk, the only variable that discriminatedbe- predators of Honey Buzzard nestlings (Kostrzewa tweennest-site plots and control plotswas tree den- 1991). The Honey Buzzardseems to prefer spruce, sity.I obtained similar resultsafter a thinning ex- which givesbest cover, as nest trees (Amcoff et al. periment, where the reuse of nest stands in 1994). In contrast to the other species studied, thinned young forestswas lower than of nest stands Honey Buzzardsare usually silent when disturbed in young forestsnot thinned (Selfis1996). Place- by humans at the nest site (Holstein 1944, Hagen ment of nestsin dense forest could hardly be prof- 1952, Kostrzewa1985). Rather than selectingnest- itable with respect to food of sparrowhawks,be- ing habitats to avoid nest predation, Honey Buz- causethe densityof passerinesis usuallylow here zards appear to behave as crypticallyas possibleat (Haapanen 1965, odegaard1982, Glowacinskiand the nest site, possiblybecause they are lessefficient Weiner 1983, Helle 1985). Probably,predation is than other raptors in defending their nestsagainst the most important aspectin the nest-siteselection predators. In addition, low annual mortality and of sparrowhawks(Selfis 1996), sinceits main pred- low clutch size of the Honey Buzzard (Holstein ators, goshawksand pine martens (Mattes mattes), 1944, Kostrzewa 1985, Tjernberg and Ryttman both prefer mature forestrather than young,dense 1994) may make nest defense lessprofitable than forestwhen hunting (Pulliainen 1984, Widtn 1989, for Common Buzzards and goshawks. Storch et al. 1990). Actually, dense forest seemsto TopographicalVariables. The only specieswhich be less important as nesting habitat for the spar- showedany preference for slope wasthe Common rowhawk when the goshawk is absent (Bornholt Buzzard,which usuallynested in steep terrain. Sim- 1983, Newton 1986, Tommeraas 1994). Pine mar- ilar results have been found for the Red-tailed tens will probably find raptor nests easy,because Hawk (Buteojamaicensis, Titus and Mosher 1981, of the smell from pellets and prey remains. Since Speiser and Bosakowski1988). Flight energetics the pine is also known to remember dif- may be more favorableon steeperslopes for larger ferent sites of food resources (Sonerud 1985), it is soaringraptors like eaglesand large (Speis- likely to be familiar with most of the old raptor er and Bosakowski1988). It may however also be nestswithin its home range. This may be one rea- important that thesebroad-winged species can best son for why sparrowhawksrarely use nests for two escape, or attack, predators in this habitat. Even successiveyears, unlike goshawksand Common though the Common Buzzard is able to rob prey Buzzards which are probably less vulnerable to from the goshawk(Fischer 1980, Jtrgensen 1983), pine marten predation due to their large size. its breeding successhas been found to be nega- MARCH 1997 NEST-SITE SELECTION BY FOUR RAPTORS 23 tively correlated with the distanceto goshawknests The observed interspecific differences in nest- (Kostrzewa 1991). One reason for the difference site selectionbetween the raptor speciesinvestigat- in nest-site selection between the Common Buz- ed may be explainedby interspecificdifferences in zard and goshawkmay be that the goshawk,which body size and flight performance, nest-predation is better adapted for flight and foraging in dense risk, time of breeding and feeding habits.The risk forest, is more dangerousto Common Buzzard in of predation probably affectsnest-site selection or dense forest. breeding habits of all these species, but mostly Common Buzzardsalso preferred nest siteswith sparrowhawksand Honey Buzzards which were southern aspects.There was also a tendencyfor a most vulnerable to nest predation. Common Buz- higher percentageof nest-siteplots of goshawksto zards,goshawks, and Honey Buzzardsalso showed have southern aspectsthan expected, while those nest-sitepreferences which could be explained as of Honey Buzzards tended to have northern as- an adaptationto microclimate.For thesethree spe- pects.Common Buzzardsand goshawksstart their cies, nest-site selection could also be connected to breeding nearly one month earlier than sparro- the availability of food in the early stage of the whawks and more than one month earlier than breedingseason. These species may have a broader Honey Buzzards(Forsman and Solonen 1984), at habitat choice and it is possiblethat factors other a time of the year when the temperaturesmay still than the habitat variables I selected for study may be far below freezing in southern Norway.Nests of have been of importance.This may have been es- both specieswere most often found at siteswith a pecially true for the goshawk,which builds larger southeasternaspect, which are the first heated by neststhan the other speciesand may be influenced the morning sun when nest building occurs (Hol- by charactersdirectly connectedto the nest tree. stein 1942, 1956). Also in , goshawkshave ACKNOWLEDGMENTS been found to favor southern slopes (McGowan 1975), while in more temperateareas, southern ex- I wish to thank I. Selfis and K_O. Selfis for assistance posuresare avoided(Dietzen 1978, Reynolds et al. with the field work, G.A. Sonerud for valuable advice dur- ing the analysesof the data, and T. Bosakowski,S.M. Brai- 1982, Moore and Henny 1983, Link 1986, Speiser nerd, S. Dale, G.A. Sonerud and two anonymous referees and Bosakowski1987). A similarpattern has been for constructivecomments on the manuscript.The study observed for nest sites of Golden (Aquila wassupported by the Nansen Endowment. chrysaetos,Mosher and White 1976, Pfaff 1993). LITERATURE CITED In , Common Buzzards place their nest near forest edges (Knfiwer and Loske A•RAHAMSEN,J., N. I• JAKOBSEN,E. DAHL, R. KALLIOLA, 1980, Spitzer 1980, Hubert 1993), probably be- L. WILBORGAND L. Pf•ILSON. 1977. Naturgeografisk cause they hunt from perches in open areas or regioninndelingav Norden. Nordiske utredninger B from forest edges (Wid6n 1994). In my studyarea, 1977, 34, Stockholm, . Amcoff, M., M. Tjern- open areas were usually covered by snow when berg and A. Berg. 1994. Nest site choice of Honey BuzzardPernis apivorus. Ornis Svecica 4:145-158. Common Buzzards arrived to breed, making the ANONYMOUS.1989. Goshawkbreeding habitat in lowland (Microtusagrestis), which is the most im- Britain. British Birds 82:56-67. portantprey species in thishabitat (Hansson 1978, ANDRI•N, H. AND t. DELIN. 1994. Habitat selection in the Spidsoand Selfis1988), nearly unavailable(Hans- Eurasian red , Sciurusvulgaris, in relation to son 1982, Sonerud 1986). Early snow-free areas forestfragmentation. Oikos 70:43-48. availablefor vole hunting in the spring are found BIRKEMOE,T. 1993. Distribution of ground dwellingbee- on southfacingslopes and in steep terrain, where des along the main vegetation gradient in an old Common Buzzard nestsare usuallyfound. growth boreal forest. Cand. Scient. thesis, Univ. of Goshawks,Common Buzzards and Honey Buz- Oslo, Norway. zards rarely used nest siteswith southwesternas- BOMHOLT,P. 1983. Population trends in Danish raptors since 1970. Proc.Third Nordic Congz Ornithol. 1981:39- pect, possiblybecause too much sun may be harm- 44. ful to newly-hatchednestlings (c.f. Holstein 1942, DIETZEN,W. 1978. Der brutbiotop des habichtsAcci]nter Hald-Mortensen1974, Reynoldset al. 1982, Link gentilisin drei Gebieten Bayerns.Anz. orn. Ges.Bayern 1986). Unless there is good shelter,as in the dense 17:141-159. youngforests used by sparrowhawks,nest siteswith FISCHER,W. 1980. Die Habicht. Neue Brehm-Bficherei, a southwesternaspect are probably unprofitable re- Wittenberg, Lutherstadt. gardlessof when egg-layingbegins. FORSM&•,D. ANDT. SOLON•N.1984. Censusingbreeding 24 Svsd•s VOL. 31, No. 1

raptors in southern : methods and results. JORGENSEN,J. 1983. Buzzard, Buteo buteo,forcing gos- Ann. Zool. Fenn. 21:317-320. hawk, Acdpitergentilis,to drop prey. Calidris12:35. GLOWACINSKI,Z. AND J. WEINER. 1983. Successional KIELLAND-LUND,J. 1981. Die WaldgesellschaftenSO-N- trends in energeticsof forest bird communities.Ho- orwegens.Phytocoenologia 9:53-250. larct. Ecol. 6: 305-314. 1994. Syntaxonomy of Norwegian forest vege- GOTMARK,F., D. BLOMQVIST,O. C. JOHANSSONAND J. tation 1993. Phytocoenologia24:299-310. BERGKVIST. 1995. Nest site selection: a trade-off be- KNOWER,H. ANDK.H. LOSKE.1980. Zur frage der habl- tween concealment and view of the surroundings?J. tat-ansprfichedes MSusebussards(Buteo buteo) bei der Avian Biol. 26:305-312. horstplatzwahl. Vogelwelt101:18-30. HAAeANEN, A. 1965. Birdfauna of the Finnish forest in KOSTRZEWA,A. 1985. Zur Biologie des wespenbussards relation to forest succession. Ann. Zool. Fenn. 2:153- (Pernis apivorus) in Teilen der Niederrheinishen 196. Bucht mit besonderen Anmerkungen zur Methodik HAGEN,Y. 1952. Rovfuglene og viltpleien. Gyldendal, bei Greifvogeluntersuchungen.Okol. Vb•el 7:113-134. Oslo, Norway. 1991. Interspecific interference competition in --AND t. BAKKE.1958. The food of some Honey three Europeanraptor species.Ethology, Ecology &Evo- Buzzards (Pernisapivorus (L)) in Norway. Papers of lution 3:127-143. the Norwegian State Game Research,Ser. No. 2. LINK, H. 1986. Untersuchungen am Habicht (Acdpiter HALD-MORTENSEN,P. 1974. Nest and nest site of the spar- gentilis). Deutscher Falkenorden--Schriftenreihe, rowhawk Acdpiternisus•and some comparisonswith Heft 2. the Goshawk A. gentilis.Dansk Orn. Foren. Tidsskr.68: MANLY,B.FJ., L.L. MCDONALDAND D.L. THOMAS.1993. 91-115. Resource selection by . Statisticaldesign and HANSSON,L. 1978. Small abundance in relation analysisfor field studies. Chapman & Hall, London, to environmental variables in three Swedish forest UK. phases.Stud. Forest. Suec., No 147. MARTIN,T.E. 1993. Nest predation and nest sites.New 1982. Dfiggdjurensvinterutnyttjande av kanter perspectiveson old patterns. BioSdence43:523-532. mellan hyggenoch barrskog.Fauna och Flora 77:301- McGowAN,J.D. 1975. Nesting habits and reproductive 308. successof goshawksin interior Alaska.Raptor Res. Rep. HELI•E, P. 1985. Habitat selection of breeding birds in No. 3:147-152. relation to forest succession in Northeastern Finland. MIKKOLA,H. 1983. Owls of Europe. T. & A.D. Poyser, Ornis Fenn. 62:113-123. London, UK. HOGLUND, N.H. 1964. Der Habicht Accipitergentilis MOORE,K.R. AND CJ. HENNY. 1983. Nest-sitecharacter- Linn in Fennoskandia.Viltrevy 2:195-270. isticsof three coexistingAcdpiter Hawks in northeast- HOLSTEIN,V. 1942. Duehogen Acdpiter gentilis dubius ern .J. RaptorRes. 17:65-76. (Sparrman). Biologiskestudier over danske rovfugle MOSHER,J.A. ANDC.M. WHITE. 1976. Directional expo- I. Hirschsprungs,Copenhagen, . sure of nests.Can. Field-Nat. 90:356-359. 1944. HvepsevaagenPernis apivorus apivorus (L.). NEWTON,I. 1979. Population Ecology of Raptors.T. & Biologiske studier over danske rovfugle II. Hirschs- A.D. Poyser,London, UK. prungs, Copenhagen,Denmark. 1986. The sparrowhawk.T. & A.D. Poyser,Lon- 1950. Spurvehogen Accipiternisus nisus (L.). don, UK. Biologiske studier over danske rovfugle III. Hirschs- ODEGAARD,R. 1982. Spurvefuglsamfunneti suksesjonss- prungs, Copenhagen, Denmark. tadier blfiba•r-bregnegranskog(Eu-Piceetum) i Sor- 1956. MusvaagenButeo buteo buteo (L.). Biologis- Norge. Cand. Scient. thesis,Univ. of Oslo, Norway. ke studier over danske rovfugle IV. Hirschsprungs, PfAff, A. 1993. Bestandsstorrelse, reproduksjon og Copenhagen, Denmark. na•ringsvalghos kongeorn Aquila chrysaetosi Aust-Ag- HUBERT,C. 1993. Nest-sitehabitat selectedby Common der. Cand. Scient. thesis,Univ. of Oslo, Norway. Buzzard (Buteobuteo) in southwesternFrance. J. Raptor PULI•IAINEN,E. 1984. Use of the home range by pine Res. 27:102-105. martens (Martes martes). Acta Zool. Fennica 171:271- ITAMIES,J. ANDH. MIKKOLA.1972. The diet of Honey 274. BuzzardsPernis apivorus in Finland. OrnisP)nn. 49:7- REYNOLDS, R.T., E.C. MESLOW AND H.M. WIGHT. 1982. 10. Nesting habitat of coexistingAccipiter in Oregon. J. JANEs,S. W. 1985. Habitat selectionin raptorial birds. wildl. Manage.46:124-138. Pages159-184 in M.L. Cody [Ed.], Habitat selection SAS INSTITUTE,INC. 1994. JMP statisticsand graphics in birds. Academic Press, London, UK. guide, Version 3 ofJMP. Cary, NC U.S.A. JEDRZEJEWSKI,W., B. JEDRZEJEWSKIAND M. KELLER.1988. SEL•S,V. 1989. Prey selectionin the goshawkduring the Nest site selectionby the buzzard Buteobuteo L. in the breeding season.Fauna 42:104-110. extensive forest of eastern . Biol. Cons. 43:145- 1993. Selection of avian prey by breeding spar- 158. rowhawksAccipiter nisus in southern Norway:the im- MARCH 1997 NEST-SITE SELECTION BY FOUR RAPTORS 25

portance of size and foraging behaviour of prey. Ornis and habitat selection of the pine marten in relation Fenn. 70:144-154. to competition with the . Acta The•:35:311-320. ß 1996. Selectionand reuseof nest standsby spar- SULI•V^, P. 1964a. On the behaviour and food habits of rowhawksAccipiter nisus in relation to natural and ma- the sparrowhawk(Accipiter nisus) during the nesting nipulated variation in tree density.J. Avian Biol. 27: season. Suomen Riista 17:93-105. 56-62. 1964b. Zur nahrungsbiologiedes habichtsAcc•p- SIDERS,M.S. AND P.L. KENNEDY. 1996. Forest structural iter gentilis.Aquilo, So: Zoologica3:1-103. SUOMUS,H. 1952. Studieson the feeding habits of the characteristicsof Accipiternesting habitat: is there an Common Buzzard.Riistat. Julk. 8:121-127. allometric relationship?Condor 98:123-132. SW•NSON,J.E. ANDP. ANGELST)•M.1993. Habitat separa- SOLONEN, T. 1982. Nest sites of the Common Buzzard tion by sympatric forest grouse in Fennoscandia •n Buteo buteo in Finland. Ornis Fenn. 59:191-192. relation to boreal forest succession.Can. J. Zool.71. SONERUD,G.A. 1985. Nest hole shift in Tengmalm's Owl 1303-1310. Aegoliusfunereus as defenseagainst nest predation in- TI•dN•N, J. 1981. Habitat preferencesby breeding forest volving long-term memory in the predator.J. Anim. passerinesin an area in Finland. Proc. SecondNordic Ecol. 54:179-192. Cong•Ornithol. 1979:38-42. 1986. Effect of snow cover on seasonalchanges TINBERGEN,L. 1946. De sperver als roofvijand van za- in diet, habitat, and regional distribution of raptors ngvogels.Ardea 34:1-213. that prey on small mammalsin boreal zonesof Fen- TITUS, K. ANDJ.A. MOSH•R. 1981. Nest-site habitat se- noscandia. Holarct. Ecol. 9:33-47. lected by hawksin the central Appalachi- SPEISER,R. AND T. BOSAKOWSKI.1987. Nest-site selection ans. 98:270-281. by Northern Goshawksin northern and TJERNBERG,M. ANDH. RYYTMAN.1994. Survivaland pop- southeastern . Condor 89:387-394. ulation developmentof the Honey BuzzardPernis ap•- vorus in Sweden. Ornis Svecica 4:133-139. -- and •. 1988. Nest-sitepreferences of Red- TOMMERAAS,P.J. 1994. The GoshawkAccipiter gentilis •n tailed Hawks in the highlands of southeasternNew Leksvik--a sacrifice to modern forestry. Fauna 46. York and northern New Jersey.J. Field Ornithol.59: 180-195. 361-368ß UTTENDORFER,O. 1952. Neue Ergebnissefiber die Er- SPIDSC},T.K. AND V. SELAS.1988. Prey selection and nShrung der Greifv6gel und Eulen. Eugen Ulmer, breeding successin the Common Buzzard Buteobuteo Stuttgart, . in relation to smallrodent cyclesin southernNorway. WALSBERG,G.E. 1985. Physiologicalconsequences ofmi- Fauna Norv. So: C, Cinclus 11:61-66. crohabitat selection. Pages 389-413 in M.L. Cody SPITZER,g. 1980. Zum Dispersionsmusterund dessen [Ed.], Habitat selection in birds. Academic Press, Bedeutung in der biologie des mausebussards(Buteo London, UK. buteo).Line basisstudiezu einem roodell seiner po- WIDI•N, P. 1987. Goshawk predation during winter, pulationsdynamik.Z. Jagdwiss.26:11-23. spring and summer in a boreal forest area of central Swedenß Holarct. Ecol. 10:104-109. SQUIRES,J.R. AND L.F. RUGGIERO.1996. Nest-siteprefer- ß 1989. The hunting habitatsof GoshawksAccip•ter ence of northern goshawksin southcentralWyoming. gentilisin boreal forests of central SwedenßIbis 131: J. Wildl.Manage. 60:170-177. 205-213. STOKt•tND,J.N. 1994. Biological diversityand conserva- 1994. Habitat quality for raptors:a field exper- tion strategies in Scandinavian boreal forests. Dr. iment. J. Avian Biol. 25:219-223. Scient. thesis,Univ. of Oslo, Norway. STORCH,I., E. LINDSTROMAND J. DEJOUNG.1990. Diet Received18 June 1996; accepted 1 December 1996