The Condor 91:362-371 0 The CooperOrnithological Society 1989

HAWK OCCUPANCY OF DISTURBED GRASSLANDS IN RELATION TO MODELS OF HABITAT SELECTION ’

JOSEF K. SCHMUTZ Department of Biology, Universityof Saskatchewan,Saskatoon, SK S7N 0 WO, Canada

Abstract. I counted nesting pairs of FerruginousHawks (Buteo regalis)and Swainson’s Hawks (B. swainsoni)on 76 41-km2 study plots in 1982 and 1987 to study the effect of agricultural cultivation of grasslandon hawk densities. The study plots were selected at random within a 74,686-km2 mixed-grassprairie region in southeasternAlberta. The results confirm a curvilinear relationship between hawk density and extent of cultivation. Nesting densitiesof both specieswere higher in areasof moderate cultivation than in grassland.On plots with extensive cultivation (>30%), Ferruginous Hawks declined and so did ground squirrels,their main prey. Swainson’s Hawks did not decline with extensive cultivation but probably shifted to other prey where ground squirrelswere scarce.The observedcurvilinear relationship between hawk density and habitat is similar in shape to changesin the pro- ductivity of ecosystemssubjected to varying degreesof disturbance. The pattern of hawk abundance in relation to habitat quality did not change between years despite a more than 50% increase in hawk densities. The hawks therefore did not conform to the general assumptionin models of habitat selectionthat optimal habitats are saturatedbefore suboptimal areas are occupied. Instead, new breeding pairs continued to settle in optimal habitat but not suboptimal habitat. To account for these results,I propose a graphical model which incorporatesterritoriality as a factor causingregular dispersion at medium population densitiesacross the rangeof habitat typesincluded in this study.Optimal habitat may be the last to reach saturation becauseonly there can territory size be reduced furthest under increasingintruder pressure. Key words: FerruginousHawk; Buteo regalis; Swainsons’ Hawk; Buteo swainsoni;Al- berta; habitat-selectionmodels: grassland cultivation; population dynamics; territoriality.

INTRODUCTION and individuals disperse (“free distribution”), The large number of published reports that de- such that in the end they experience relatively scribe habitat selection among birds attests to equal feeding opportunity in sites varying in re- the enormous variation observed and to the bi- source availability. Given differences in suita- ological importance of this topic. In addition to bility between habitats, the prediction is that areas its basic biological importance, an understanding of lower suitability are occupied only after the of the responses by birds, and other animals, to suitability of the most optimal areas has been changes in habitat characteristics is important for reduced by the need to share resources with oth- conservation. Habitat degradation emerges as a ers. The model predicts that members of a species major threat to conservation in many countries. for which this model applies experience rela- Functional and theoretical aspects of habitat tively equal habitat suitability and hence equal selection have been summarized by Cody (1985). fitness for that component of fitness that is de- Fretwell and Lucas (1970) and Fretwell (1972) rived from habitat suitability. provide a theoretical framework for habitat se- Species that exhibit territoriality such that their lection. Their model illustrates the choices in- density can be greatly dependent on this behavior dividuals are forced to make when attempting to (e.g., Brown 1969) violate the “free” assumption maximize fitness in habitats varying in suitabil- of the ideal free distribution model. To include ity for optimal survival and reproduction. They the effect of dominance and territoriality, Fret- postulate first that organisms are able to judge well (1972) incorporated a risk accruing to un- an area’s basic suitability and seek those areas settled individuals in the “ideal despotic distri- best suited (“ideal distribution”). Second, under bution model.” Because territories are generally increasing intruder pressure suitability decreases smaller and hence more easily defended in the most suitable areas, territoriality can create dif- 1Received 27 September 1988. Final acceptance30 ferences in fitness between individuals in areas January 1989. differing in their basic suitability.

[36-V HABITAT SELECTION BY PRAIRIE HAWKS 363

In this analysis, I report on a “natural exper- lected at random within a 74,686-km* survey iment” in which the habitat selection of Ferru- region in southeasternAlberta. This region was ginous and Swainson’s hawks was studied twice, bounded by Saskatchewanand Montana. Within 5 years apart, on randomly selected study plots Alberta, the study area included semi-arid, in southeastern Alberta. During the 5-year in- mixed-grass prairie habitat located south of terim, the abundance of Richardson’s ground 5 l”42’N latitude and east of 114”OO’W longitude. squirrels (Spermophilus richardsonii) has in- I searchedplots for nests in 2 years. In 1982, I creasedmarkedly (J. K. Schmutz and D. J. Hun- searched80 plots (Schmutz 1984) and in 1987, gle, unpubl. data), and so has the population den- 78 plots, 76 of which were searchedduring both sity ofboth hawks (Schmutz, unpubl.). The study years. plots used were the same in both years. Changes I searchedstudy plots completely for all nests in the degree of cultivation between years were of buteonine hawks. The primary mode of trans- small. In a within-year comparison, I examine portation on plots was by motorcycle but some differences in the densities of breeding hawks sites were searchedusing a truck or on foot. In between plots varying in degree of cultivation. I most areas, nests were found easily in the gently also examine habitat selection between years undulating landscapewith few trees. In addition when the overall population size of hawks dif- to nestscontaining eggsor young, some deserted fered. The observed patterns of habitat selection nests (4.7% of all nests) that showed evidence of are compared with predictions based on models use during the present year (Postupalsky 1974) of habitat selection. I make the fundamental as- were counted. In these cases,species was deter- sumption in this analysis that individuals on dif- mined by the type of nest material (Schmutz et ferent plots are drawn at random from the pop- al. 1980) and color offeatherspresent. The 6-week ulation pool and that they do not represent searching period, from late May to early July, different ecotypes. included the hatching and nestling periods of The Ferruginous and Swainson’s hawks stud- Ferruginous Hawks, and the latter part of laying, ied exhibit intraspecific territoriality (Schmutz et all of incubation, and the early nestling period al. 1980, unpubl.). Their habitat selection has of Swainson’s Hawks. been described (Olendorff 1973; Schmutz et al. Steeply eroded banks existed on the study area 1980; Bechard 1982; Gilmer and Stewart 1983, along rivers and creeks. Some grasslands had 1984; Schmutz 1984, 1987; Janes 1985). Unlike never been tilled, others had undergone second- Swainson’s Hawks, Ferruginous Hawks are high- ary successionfrom fields abandoned during the ly selective in their choice of grassland habitat 1930s (Dormaar and Smoliak 1985). The pri- for nesting. This habitat dependence is so pro- mary land useswere grazing of cattle and growing nounced that an availability of grasslandhas been of cereal grain, depending on soil quality for ag- considered important in limiting this species’ riculture and nature of terrain. In 1982 the av- breeding distribution. erage percent cultivation was 50% and in 1987 The present study differs from most others (cf. it was 53%. Cody 1985) in some potentially important ways. For each plot, the locations of nests, fields, First, the speciesstudied are large raptors which, pastures, trees, and water bodies were recorded as adults at least, are threatened by few if any on a data sheet at the time of searching.I deter- heterospecificpredators. Second, density is that mined the percentageof land area that was cul- of breeders only. Third, comparisons of hawk tivated from thesemaps for each year separately. density are made within a grasslandhabitat that Fields that had been reseededwith “tame” grass varies in degreeof disturbance due to agricultural varieties and used for grazing were recorded as cultivation. Thus, this study examines responses grassland. Fields of clover, that were irrigated by the hawks to alterations (cultivation) of one and used for forage production, were recorded type of habitat (mixed-grass paririe) and not dif- as cultivated becauseof their similarity to cereal ferencesbetween two natural habitats. crops in vegetation height and density. I recorded the number and activity of hawks presentwhen their nest was found. Nest contents STUDY AREA AND METHODS were inspectedwhen accessiblewithout climbing Nesting pairs of Ferruginous and Swainson’s aids. In late May and early June, clutches of hawks were counted on 4 1-km2 study plots se- Swainson’s Hawks were considered complete if 364 JOSEF K. SCHMUTZ down feathers were present in the nest cup and ously (Schmutz et al. 1980, 1984). To minimize the incubating hawk flushed at a distance of 20 the confounding influence of tree availability on m or less. Because the nests were visited only hawk density, I excluded plots which bore no once between the latter period of laying (Swain- trees. This approach was slightly conservative son’s) and prefledging(Ferruginous), the number because some pairs could nest on the ground. of eggsor young, found on different days from Fifteen percent of Ferruginous Hawks and < 1% the mean hatching date, was expressedas a de- of Swainson’s Hawks nested on the ground. In viation from a local population average. This an attempt to account for intraspecific territo- allowed a comparison between pairs on different riality, some treed plots were also excluded if plots. An average clutch and a change in brood every 0.64 km* (quarter section) of treed land size throughout the nestling period was derived supported a conspecific, nesting pair of hawks. from a total of 29 nests of Ferruginous Hawks The number of plots that remained for analysis and 25 nestsof Swainson’s Hawks monitored on was 66 for Ferruginous Hawks and 64 for Swain- the Hanna study area in 2 years (Schmutz et al. son’s Hawks. 1980). Those nests were visited for a study of food habits every l-3 days beginning prior to CULTIVATION AND HAWK DENSITY hatching. Most loss of nestlings occurred early Cultivation on plots ranged from 0 to 99%, ex- with little change in the second half of the nest- cepting farmyards and roadsides.In a curvilinear ling period (J. K. Schmutz, unpubl. data). I con- relationship, the hawksincreased in density where siderthe Hanna study area representativeof many cultivation increased up to 30% and then either plots becausethis area was located in the north- declined (Ferruginous Fig. 1) or showed no fur- central portion of the survey region (Schmutz, ther change (Swainson’s Fig. 2). unpubl.) and because 14% of the area was cul- To determine whether the observed curvilin- tivated. ear relationship between hawk density and cul- To monitor the abundance of the hawks’ main tivation differed significantly from a straight line, prey, I recorded the number of burrows showing I first compared hawk density on plots of grass- signs of use by ground squirrels (Spermophilus land (O-10% cultivation) with plots of moderate spp.) on transectswithin study plots in 1987. One cultivation (1 l-30%) in an a posterori manner. to four transects were done on each of 37 plots For each speciesand during each year, the mean which contained. at least 3 km2 of contiguous number of nests was greater in areas with 1 l- grassland. Burrows within 1 m of an observer 30% cultivation than O-IO% (Table 1). When the were counted while driving a motorcycle at slow apparently positive effect of moderate cultiva- speed in a straight line. Transect length varied tion on hawk density was excluded by consid- from 0.5 to 1.5 km depending on accessibility. ering only plots with 1 l-100% cultivation, Fer- Assumptions of normality were not met in the ruginous Hawks declined with increasing data. Hence, tests of significance were frought cultivation (1982: Spearman’s rr = -0.477, P = with difficulty when using parametric statistical 0.003; 1987: r, = -0.684, P < 0.001). Swain- approachesfor hypothesis testing. Nonparamet- son’s Hawks, in contrast, showed no significant ric approacheswere also not entirely valid in all declining trend with increasingcultivation (1982: casesowing to up to 27% tied values. I therefore Y, = -0.149, P = 0.134; 1987: r, = -0.042, P = used a combination of nonparametric, and para- 0.377). metric analyses (Feldman et al. 1986). The null The number of suitable nest trees on a plot hypothesis was rejected at the 0.05 level. was dependenton land use. Frequently, treeswere planted as “shelterbelts” near fields, trees were RESULTS protected from damage by livestock when grow- Hawk density and distribution on study plots ing near fields, and trees benefitted from addi- was potentially affected by nest-site availability, tional moisture present in the irrigated/cultivat- degree of cultivation, inter- and intraspecific ed regions. In an additional attempt to separate competition, and hawk population size. While the effectof tree availability (number of 0.64 km2 degree of cultivation was the primary factor ex- per plot bearing one or more trees/shrubs)from amined in this study, other factors may have the effectsof cultivation on hawk density, I used interacted with it. An influence of nest-site avail- a multiple regressionanalysis after hawk density ability on hawk density has been shown previ- was subjectedto squareroot transformation. This HABITAT SELECTION BY PRAIRIE HAWK8 365

0 0 20 40 50 80 100 Percent Cultivation on Plots FIGURE 2. The number ofnestsof Swainson’sHawks FIGURE 1. The number of nests of Ferruginous on plots in 1982 and 1987 are shown in relation to Hawks on plots in 1982 and 1987 are shown in relation percentcultivation on plots. A fourth order polynomial to the percentcultivation on plots. A fourth order poly- expressionwas fitted to the data (see Fig. 1). The mean nomial expression,represented by a line, was fitted to number of nests in each 10% cultivation categoryand the data by increasingthe order of the expressionuntil the range in the number of nests are given in Table 2. no further resolution in the shape of the curve was obtained. The mean number of nests in each 10% cul- tivation categoryand the range in the number of nests abundance with percent cultivation on plots. are given in Table 2. Squirrel abundance was most variable on plots with little or no cultivation (Fig. 3). Because of was done to ensure that the hawks’ responses their loosely colonial dispersion, squirrels were to habitat change (cultivation) did not merely sometimes abundant on small patches of grass- reflect tree availability. For this analysis I used land. The decline in squirrel abundance on a plot all 76 plots searchedin 1982 and 1987. In the basis was actually much greater than suggested caseof Ferruginous Hawk density, there was no by these data because squirrels occupied crop- improvement in the amount of variation ex- land at low density if at all and no burrow counts plained when the additional variable of tree were carried out there. Squirrel abundance rel- availability was added to cultivation (1982: cul- ative to cultivation resembled the Ferruginous tivation r2 = 0.210, cultivation and trees r2 = Hawks’ density pattern. Squirrels appeared to 0.190; 1987: cultivation r* = 0.417, cultivation increase in abundance with an increase in cul- and trees r2 = 0.344). By nesting on the ground, tivation up to 30% (Fig. 3). However, this in- on rocks or on man-made structures (Schmutz creasewas not statistically significant (r, = 0.256, et al. 1984), Ferruginous Hawks were able to n = 27, P = 0.096) possibly because of a wide occupyareas where treeswere absent. Swainson’s variance in burrows counted and a small sample Hawk density was more strongly affectedby tree of transects.Squirrels declined where the degree availability. The square of the regressioncoeffi- of cultivation exceeded 30% (r, = -0.649, n = cient for cultivation was substantially greater 26, P -c 0.001). when trees were added as a variable (1982: cul- tivation r2 = 0.070, cultivation and trees r2 = HAWK REPRODUCTION IN 0.288; 1987: cultivation r2 = 0.148, cultivation RELATION TO HABITAT and trees r2 = 0.255). In 1987, even those pairs that nested on the most To examine a possiblelink between hawk den- intensively cultivated plots raised a normal num- sity and cultivation, I plotted ground squirrel ber of young. Four pairs of Ferruginous Hawks, 366 JOSEF K. SCHMUTZ

TABLE 1. The number of nesting pairs of Ferruginousand Swainson’s hawks on plots in grassland(O-l 0% cultivation) and on plots with moderate cultivation (1 l-30%) in southeasternAlberta. Differences in nesting density betweenthe two cultivation categoriesare compared usinga modified signtest (Conover 1971). I recorded the number of plots whose densitieswere above or below the combined mean density (x) for all 35 or 3 1 plots combined. This method assumesthat if hawk density were not affected by the degree of cultivation, a similar proportion of plots in each cultivation categorywould fall above or below the combined mean.

Grassland Moderatecultivation Plots Plots R Rallge n abovex; R RW+? n above R

Ferruginous 1982 0.88 o-2 8 2 (25%) 1.54 o-7 11 4 (36%) 1987 1.14 o-3 7 1 (14%) 2.44 04 9 8 (89%) Total 1.oo o-3 15 3 (20%) 1.95 o-7 20 12 (60%)’ Swainson’s 1982 1.43 o-3 7 2 (29%) 3.00 O-7 10 6 (60%) 1987 1.33 o-2 6 0 4.38 O-8 8 5 (63%) Total 1.38 o-3 13 2 (15%) 3.61 O-8 18 11 (61%)”

*x2 = 5.60, P = 0.018. bx ’ = 6.48, P = 0.011. for which the brood size was known and which were avoided. The pattern of occupancy was nested on plots with 70% or more cultivation, highly similar between years. raised 2.8 young, compared to 3.0 young raised The additional Swainson’s Hawks that nested by 59 pairs on the Hanna study area in 1987. in 1987 also exhibited a habitat preference. In Similarly, 10 pairs of Swainson’s Hawks on plots contrast to Ferruginous Hawks, this preference with 90% or more cultivation raised 2.3 young, was for moderately and extensively cultivated the same number that was raised by 8 1 pairs on areas. Only plots with O-10% cultivation were the Hanna study area. Furthermore, using the avoided (Fig. 1). Again, the pattern of occupancy coarse measure of reproduction obtained in this was similar between years. study, there was no evidence to suggestthat re- COMPETITION productive performance declined as degree of cultivation increased (Ferruginous rs = -0.070, Although most new pairs of Ferruginous Hawks n = 30, P = 0.353; Swainson’s rs = 0.086, n = in 1987 colonized plots in the 1 l-40% cultiva- 77, P = 0.228). tion range, their settlement on individual plots depended on previous densities. Ferruginous POPULATION SIZE AND HABITAT Hawk numbers tended not to change or to de- OCCUPANCY BETWEEN YEARS creaseon thoseplots on which pairs already nest- Populations of nesting Ferruginous and Swain- son’s hawks have increased by 60% and 57% respectively between 1982 and 1987 following a regional increase in the abundance of ground squirrel prey (J. K. Schmutz and D. J. Hungle, unpubl. data), The additional hawks that nested in 1987 could have settled in one of three ways: (1) in each cultivation categoryin equal numbers, (2) in the same proportions as they nested in 1982, or (3) preferentially in areas with either a low or high degree of cultivation. An increase in density of Ferruginous Hawks 0 30 60 Percent Cultivation on Plots between 1982 and 1987 occurred primarily in the 1 l-40% cultivation categories(Table 2), pref- FIGURE 3. The number of ground squirrel burrows counted on transectson the grasslandportion of study erentially near the low end of the cultivation con- plots in relation to the degree of cultivation on plots. tinuum. The increase in nests in the O-10% cat- A fourth order polynomial expressionwas fitted to the egory was small. Areas of extensive cultivation data (see Fig. 1). HABITAT SELECTION BY PRAIRIE HAWKS 361

TABLE 2. The mean number of Ferruginous and Swainson’s hawk nests in each cultivation category in southeasternAlberta is shown including sample range and the number of plots in each category. Only those plots which bore unused nest trees are included here.

Percent cultivation on plots O-10 II-20 21-30 31-40 41-50 51-60 6 l-70 71-80 81-90 91-100 Total - FerruginousHawk 1982 Mean 0.88 1.67 1.40 0.67 1.00 0.67 0.25 0.43 0.25 0.00 0.62 Range o-2 o-3 o-7 o-2 o-3 o-2 o-1 o-3 o-1 0 O-l Plots 8 6 5 6 5 3 4 7 8 14 66 1987 Mean 1.14 3.00 2.17 1.80 0.86 2.50 0.40 0.17 0.29 0.13 0.94 Range o-3 2-4 l-3 l-3 o-3 O-6 o-2 o-1 o-1 o-2 O-6 Plots I 3 6 5 I 4 5 6 7 16 66 Mean increment 0.26 1.33 0.77 1.33 _ 0.14 1.83= 0.15 -0.26 0.04 0.13 Swainson’s Hawk 1982 Mean 1.43 2.20 3.80 2.67 1.60 2.33 1.75 4.57 3.38 1.36 2.44 Range o-3 O-5 2-7 O-6 l-2 l-4 l-3 2-8 l-7 04 O-8 Plots I 5 5 6 5 3 4 7 8 14 64 1987 Mean 1.33 3.00 4.83 3.60 4.86 6.00 3.20 5.00 5.71 4.13 4.23 Range o-2 1-5 2-8 l-6 2-10 2-12 O-9 3-9 l-8 O-14 O-14 Plots 6 2 6 5 7 4 5 6 7 16 64 Mean increment -0.10 0.80 1.03 0.93 3.26 3.67 1.45 0.43 2.33 2.11

s This increase was due to one plot with 7 nests and 29% cultivation in 1982 and 6 nests and 52% cultivation in 1987 ed in 1982. Conversely, on plots where no hawks It is conceivablethat one speciesof hawk gained were found in 1982 nesting densities tended to a competitive advantage in some habitat types increase(Table 3). Swainson’s Hawks showedno and displaced the other accordingly. This was evidence to suggestthat 1982 density affected unlikely becauseno significant negative relation- settlement in 1987 (Table 3). ship between the densities of the two species within yearscould be found on the 76 plots (1982: TABLE 3. The number of plots on which density r. = 0.230, P = 0.025; 1987: rs = -0.005, P = changed in 1987 as compared to 1982 in relation to 0.484). Furthermore, in 1987, the year with the nesting density in 1982. Rows and columns are greatestdensities, correlations between densities collapsedso that the sample size in each cell is max- of the two specieshad a positive sign whether imized (Conover 1971). cultivation ranged from O-50% or 5 l-l 00%.

- Number of nests in 1982 Change in 1987 0 l-7 DISCUSSION

FerruginousHawk HABITAT AND HAWK DENSITY Increase 13 8 The results confirm that cultivation was impor- No change 31 4 tant in determining Ferruginous and Swainson’s Decrease 0 10 hawk distribution. Hawk response to habitat

Number of nests in 1982 change was curvilinear. Despite a negative effect Change in 1987 O-l 2-3 4-8 on density when cultivation was extensive (>30%), Ferruginous Hawks occupied areas of Swainson’s Hawkb moderate cultivation (1 O-30%) at greater density Increase 20 17 10 Decrease/no change 5 6 5 than they did grassland. Ferruginous Hawks were habitat specialists (e.g., Rosenzweig 198 1). Their ax ’ = 28.06, P < 0.001. b xi = 3.56, P = 0.470. density reached a distinct peak along the habitat 368 JOSEFIL SCHMUTZ continuum. Swainson’s Hawks were largely op- Hungle, unpubl. data). The similarity between portunistic becausetheir densities changed little Ferruginous Hawk and ground squirrel abun- on plots with 1 l-100% cultivation. Although dance in relation to cultivation on plots supports Ferruginous and Swainson’s hawks used similar the view that food availability is a prominent resourceson the Hanna study area (Schmutz et factor determining bird density and distribution al. 1980) they responded differently to habitat (e.g., Newton 1980). Ferruginous Hawks relied alteration. Swainson’s Hawks preferred cultivat- primarily on ground squirrels for food (Schmutz ed areas over grassland regardlessof the extent et al. 1980) and hence it is not surprising that of cultivation. this hawks’ pattern of abundance is similar to The shape of the curvilinear response to cul- the abundance of ground squirrels. This hawk’s tivation, particularly by Ferruginous Hawks, re- responseto habitat modification may have been sembled a well-documented general pattern ex- mediated by the response of ground squirrels. hibited by ecosystemsresponding to disturbance. Swainson’s Hawks, in contrast,raised more young Reimer (1983) found that as the frequency of when they nested near fields on the Hanna study disturbance increased, net primary productivity area (Schmutz 1987). in grasslandor forest ecosystemsalso increased, In this study, hawk density was more strongly but only at low disturbance levels. Net primary influenced by cultivation than was hawk repro- productivity declined as disturbance increased duction. This may indicate that the hawks ac- further. During this study disturbance (cultiva- curately assessedthe suitability of a given nesting tion) was similar between plots in the number of area before they attempted to breed there. This times with which it occurredbut the aereal extent would therefore support Fretwell’s (1972) notion of this disturbance on a plot varied. The outcome of an “ideal distribution.” of cultivation was a checkered distribution of COMPETITION grasslandand fields with varying degreesof hab- itat edge.A lack of a decline in Swainson’s Hawk The avoidance by newly settling Ferruginous density with increasing cultivation was consis- Hawks of plots that, although basically suitable, tent with this species’ generalized food habits. were already occupiedby conspecificsagrees with Where ground squirrels became scarce, Swain- resultsfrom the Hanna study area. Evidence there son’s Hawks probably fed on cricetid rodents and suggestedthat a surplus of breeders, especially grasshoppers. among Ferruginous Hawks, was prevented from The results confirm Rosenzweig’s (198 1) as- breeding by the territorial competition for space sumption that habitat selection is independent by nearby residents (Schmutz, unpubl.). In view of bird density. Despite a more than 50% in- of the extensive loss of grasslandin western Can- crease in breeding density, the pattern of hawk ada and given this species’ dependence on grass- abundance remained the same. This result was land, the existence of a surplus of breedersis not surprising becauseit may be expectedthat intra- surprising. The Canadian breeding range of Fer- and interspecific competition under increased ruginous Hawks is smaller now than it was in nesting densities may alter the hawk distribution the past (Schmutz 1984). Swainson’s Hawks used pattern along the habitat continuum. The spe- smaller territories and may not yet have reached cializations in resourceutilization within a species densitiesat which territoriality limited their nest- on the one hand and the limitations faced by ing density. This may be so even though the individuals on the other (e.g., adaptations to cope densities encountered in this study were among with aridity), may be far more important in de- the highest reported for these species. termining habitat distribution than the actions Despite exceptionally high densities of both of competitors. speciesof hawks on many plots, there was no evidence for one speciesdisplacing another from HABITAT AND FOOD plots. The two speciesmaintained intraspecific Changesin prey densities on plots between years but not interspecificterritories on the Hanna study were not recorded during this study. However, area (Schmutz et al. 1980). When a shortage of data on the amount of poison used by landown- nests forced the later arriving Swainson’s Hawk ers in the northern 12% of the survey region to nest near Ferruginous Hawks, aggressionbe- suggestthat ground squirrels have increased in tween pairs has resulted in reduced reproductive abundance in 1986 (J. K. Schmutz and D. J. success.Interspecific aggression may have oc- HABITAT SELECTION BY PRAIRIE HAWKS 369

curred in this study also, but interspecific com- petition apparently played a minor role if any in affecting densities.

MODELS OF HABITAT SELECTION The preferential settlement of optimal habitat by new hawks in 1987 was contrary to predictions arising from models of habitat selection. Newly settling hawks did not “spill over” into subop-

timal plots within their breeding range as ex- HlQh LOW pected (Fretwell and Lucas 1970, Fretwell 1972, Habitat Quality Klomp 1972, Grant 1975). The expectation that FIGURE 4. Presumedpattern by which habitat vary- birds settle optimal habitat first and spill over ing in quality is occupiedby the hawks (see text). The areaunder eachline represents25% ofthe total carrying into suboptimal spacelater is rooted in optimal capacity K.The negative slope of the lines increasesas foraging and competition theory (Rosenzweig population density approachesK. 1985) and has considerable empirical support from other studies (Cody 1985). that all potential settlers are equally well adapted Before the applicability of models of habitat to exploit the habitat in question and are thus selectionis evaluated in the context of this study, equally free to enter those habitats. it may be instructive to examine first whether In the context of the hawks studied, the model special conditions (e.g., philopatry) violated fun- may be interpreted as follows. At low densities damental assumptions of the model and not the in the past (e.g., 25% of carrying capacity), the model itself. Philopatric tendencies among ani- hawks may have occupied habitats of different mals may affect habitat selection by restricting quality at relatively equal density. Any negative the freedom to disperse (Fretwell 1972, Cody slope in density acrossa habitat continuum may 1985). Such restrictions to dispersal among have been statistically indistinguishable from breeders may be related to selection manifested zero slope (Fig. 4). In the absence of the stabi- in local adaptation (e.g., Mayr 1963) or optimal lizing effectof territoriality, dispersion may have inbreeding (e.g., Shields 1982). Fretwell (1972) been close to random and density in a given suggeststhat in suchcases habitat selectionshould area variable over time. Only as density in- be consideredonly among members of local pop- creased(e.g., up to 75% ofcarrying capacity) may ulations. Given the checkereddistribution of plots a habitat preference have become evident, as it with varying degreesof cultivation and the com- did in the 1982 data. When hawk density ap- paratively high natal dispersal exhibited by the proachedcarrying capacity in 1987, the less suit- hawks (J. K. Schmutz, unpubl. data), philopatry able habitat could apparently no longer accom- is unlikely to be a complicating factor in this modate additional pairs. On the Hanna study study. area at least, breeding pairs deviated from a ran- Wiens et al. (1986) concluded that philopatry dom toward a regular pattern of dispersion was responsible for a time lag in the responseby (Schmutz et al. 1980). The rapid replacement of breeding birds to experimental manipulations of six dead adults on the Hanna study area in recent habitat quality. Given the long history of culti- years suggestedthat a surplus of potential breed- vation in the study region and the 5 years be- ersexisted and that most or all available breeding tween counts, any time lag in responseprobably opportunities were saturated(Schmutz, unpubl.). would have been of minor importance. Further- The proposedmodel is consistentwith Fretwell’s more, it is presumably the newly settling indi- (1972) despotic distribution in that those indi- viduals that behaved contrary to expectation and viduals occupying the most suitable areas may not established breeders. be able to defend concentrated resources with The data suggesta new model for the occu- comparatively less expenditure of energy and pancy of habitat by the speciesstudied (Fig. 4). time, and may survive and reproduce more suc- This model may only apply to speciesin which cessfullythan individuals occupying lesssuitable individuals defend all-purpose territories and in areas. which territoriality can limit density (Brown One possiblefunctional explanation for the de- 1969, Fretwell 1972). The model presupposes scribed pattern (Fig. 4) includes a tendency for 370 JOSEF K. SCHMUTZ individuals to claim more spacein optimal areas lation regulation in birds. Wilson Bull. 81:293- atnrler lnxw intn~rlor ~TPECIW-P toquired to ULLUU. I” I. . ..c. UUUl &JIUc.LILLIL. thanL11c&11 ieI.3 Ic-’ 329. CODY, M. L. 1985. Habitat selection in birds. Aca- obtain resourcesnecessary for success1Ful repro- 1 ,. PT.,.. . . . * “ demic Press, New York. auction. I nis is nor to empty a --superterritory” COLGAN,P. 1979. Is a superterritorialstrategy stable? (see Verner 1977, Colgan 1979, Robertson and Am. Nat. 114:604-605. Gibbs 1982). The defense of a larger than re- CONOVER,W. J. 1971. Practical nonparametric sta- quired territory is contrary to a substantial body tistics. John Wilev and Sons. New York. DAVIES,N. B. 1980. The economicsof territorial be- of evidence that suggeststhat many birds min- havior in birds. Ardea 68:63-74. imize territory size and maximize benefit (Da- DORMAAR,J. F., ANDS. SMOLIAK. 1985. Recovery of vies 1980). However, most studies addressing vegetative cover and soil organic matter during this question are carried out in high density areas revegetationof abandonedfarmland in a semiarid where territorial optimization may be expected. climate. J. Range Manage. 38:487-49 1. FELDMAN,D. S., JR., J. GAGNON,R. HOFMAN,AND J. An avoidance of distant neighbors at low density SIMPSON. 1986. Stat-View 512+. Brain Power, may be subtle (Tinbergen 1967) and serve to Calabasas,CA. reduce interference during pair formation and FRETWELL,S. D. 1972. Populationsin a seasonalen- mating (Hinde 1965, Wells 1977). vironment. Princeton Univ. Press, Princeton, NJ. FRETWELL.S. D.. AND H. L. LUCAS. 1970. On terri- Reasons for the unexpected settlement in the torial’behavior and other factors influencing hab- optimal and not the suboptimal areas may in- itat distribution in birds. Acta Biotheor. 19:16- clude unequal changesin suitability. For exam- 36. ple, a given patch may be suitable for occupancy GILMER,D. S., AND R. E. STEWART.1983. Ferrugi- by an individual if conditions along each of 10 nous Hawk populations and habitat use in North Dakota. J. Wildl. Manage. 47:146-157. different resourcegradients reached or exceeded GILMER,D. S., AND R. E. STEWART.1984. Swainson’s a threshold level. Patchesin optimal habitat may Hawk nesting ecology in North Dakota. Condor on average contain more resource dimensions 86:12-18. above threshold than patchesin suboptimal hab- GRANT,P. R. 1975. Population performance of Mi- crotuspennsylvanicus confined to woodland hab- itat. Even if a beneficial change occurs acrossall itat, and a model of habitat occupancy. Can. J. habitats and even if this change (e.g., climatic Zool. 53:1447-1465. changes) causesimprovements in two or more HINDE, R. A. 1956. The biological significance of resourcedimensions, there will be a greater like- territories of birds. Ibis 98:340-369. JANES,S. W. 1985. Habitat selectionin raptorial birds, lihood that patchesin the optimal as opposedto p. 159-l 88. In M. L. Cody [ed.], Habitat selection suboptimal habitat will become suitable for oc- in birds. Academic Press. New York. cupancy. This could explain why new individuals KLOMP,H. 1972. Regulation of the size of bird pop- could settle in the already densely occupied areas ulations by means of territorial behavior. Neth. J. and not in the sparselyoccupied suboptimal hab- Zool. 22:456-488. MAYR, E. 1963. Animal speciesand evolution. Bel- itat. knap Press, Cambridge, MA. NEWTON, I. 1980. The role of food in limiting bird ACKNOWLEDGMENTS numbers. Ardea 68: 1 l-30. I gratefullyacknowledge the field assistanceprovided OLENDORF~,R. R. 1973. The ecologyof nestingbirds by R. G. Johnson, R.J.P. Meschishnick, and H. P. of prey of northeastern Colorado. U.S. Int. Biol. Samoil. S. M. Schmutz contributed to many aspectsof Program, Grassland Biome, Tech. Report 2 Il. this study. K. Martin, F. Messier, and anonymousref- POSTUPALSKY,S. 1974. Raptor reproductive success: erees provided helpful discussionand comments on Some methods, criteria and terminology, p. 21- this manuscript. M. G. Bickis kindly provided statis- 31. In F. N. Hamerstrom, Jr., B. E. Harrell, and tical advice. Financial support was provided by the R. R. Olendorff [eds.], Management of raptors. Alberta Recreation, Parks and Wildlife Foundation, Raptor ResearchFoundation, Vermillion, SD. the World Wildlife Fund Canada, the Alberta Division REIMER,W. A. 1983. Disturbance and basic prop- of Fish and Wildlife, the Canadian Wildlife Service, erties of ecosystemenergetics, p. 83-98. In H. A. the Special Areas Board of Hanna, and the University Mooney and M. Godron [eds.], Disturbance and of Saskatchewan. ecosystems.Springer-Verlag, New York. ROBERTSON,R. J., ANDH. L. GIBBS. 1982. Superter- LITERATURE CITED ritoriality in tree swallows:A reexamination. Con- dor 84:313-316. BECHARD,M. J. 1982. Effect of vegetative cover on ROSENZWEIG,M. L. 1981. A theoryof habitat selec- foraging site selectionby Swainson’s Hawk. Con- tion. Ecology 62~327-335. dor 84:153-159. ROSENZWEIG,M. L. 1985. Some theoretical aspects BROWN,J. L. 1969. Territorial behavior and popu- of habitat selection, p. 517-540. In M. L. Cody HABITAT SELECTION BY PRAIRIE HAWKS 371

[ed.], Habitat selection in birds. Academic Press, SHIELDS,W. M. 1982. Philopatry, inbreedingand the New York. evolution of sex. State Univ. of New York Press, SCHMUTZ,J. K. 1984. Ferruginous and Swainson’s Albany, NY. hawk abundance and distribution in relation to TINBERGEN,N. 1967. Adaptive featuresof the Black- land use in southeasternAlberta. J. Wildl. Man- headed Gull Larus ridibundus.Proc. XIV Int. Or- age. 48:1180-1187. nithol. Congr. (1966):43-59. SCHMUTZ,J. K. 1987. The effect of agriculture on VERNER,J. 1977. On the adaptive significanceof ter- Ferruginous and Swainson’s hawks. J. Range ritoriality. Am. Nat. 111:769-775. Manage. 40:438-440. WELLS,K. D. 1977. Territoriality and male mating SCHMUTZ,J. K., S. M. SCHMUTZ,AND D. A. BOAG. successin the greenfrog (Rann clamitans). Ecology 1980. Coexistenceof three speciesof hawks (BU- 58:750-762. tea spp.) in the prairie-parkland ecotone. Can. J. WIENS,J. A., J. T. ROTENBERRY,AND B. VAN HORNE. Zool. 58:1075-1089. 1986. A lesson in the limitations of field exper- SCHMUTZ,J. K., R. W. FYFE,D. A. MOORE,AND A. R. iments: Shrub-steppebirds and habitat alteration. SMITH. 1984. Artificial nestsfor Ferruginousand Ecology 671365-376. Swainson’s hawks. J. Wildl. Manage. 48:1009- 1013.