coyote predation and habitat segregation of white-tailed deer and mule deer

2037

Ecology, 83(7), 2002, pp. 2037–2048q 2002 by the Ecological Society of America

COYOTE PREDATION AND HABITAT SEGREGATION OFWHITE-TAILED DEER AND MULE DEER

SUSAN LINGLE1

Large Animal Research Group, Department of Zoology, Downing Street, University of Cambridge, Cambridge CB2 3EJ UK

Abstract. Predation has been proposed as a major factor maintaining segregation amongspecies of ungulates, but predator–prey interactions have not been observed to test thisidea directly. Here, observations of coyote (Canis latrans) packs hunting deer are used toshow that mule deer (Odocoileus hemionus), which typically stand high on slopes and onrugged terrain, increased both their risk of being encountered and attacked by coyotes bystanding low rather than high on slopes. The risk incurred at certain heights was not fixed:a mule deer’s risk of being approached by coyotes was also affected by the height of othermule deer groups present during the same hunt. White-tailed deer (O. virginianus), whichtypically use gentle terrain, were not similarly disadvantaged by remaining low on slopesor on flat terrain. When confronted by coyotes, mule deer moved to and up slopes, whereaswhite-tails moved down and away from slopes. Species differences in behavior were in-dependent of starting position and were observed for animals in mixed-species groups.Unlike their response to coyotes, feeding preferences or competition did not lead mule deerto use rugged habitats: mule deer moved down and left slopes to feed, bringing them closerto white-tails. These results suggest that coyote predation contributes to the habitat seg-regation of white-tails and mule deer (1) by selective predation against mule deer but notwhite-tails in gentle habitats and (2) by eliciting differing antipredator behavior that in-creases their segregation. Unlike prey involved in other examples of predator-mediatedresource partitioning, white-tails and mule deer are similar in size and morphology. Con-trasting antipredator strategies, specifically, their ability to avoid predation using flight orconfrontation, are likely to explain why the species differ in their behavior and in theirrisk on gentle terrain.

Key words: antipredator behavior; Canis latrans; conservation; coyotes; deer; habitat segre-gation; Odocoileus; predator-mediated resource partitioning; reproductive isolation; rugged terrain.

INTRODUCTION

The risk of predation can affect habitat use by preyin two ways. First, predators may capture more preyin certain habitats than in others, thereby selectivelyremoving animals that use unsafe habitats (Taylor1984). Second, the threat of predation may influenceprey to remain in or move to a certain habitat to avoidpredators (Jeffries and Lawton 1984, Sih 1987, Limaand Dill 1990, Brown et al. 1999). When groups ofprey differ in their risk in a particular habitat or in theirresponse by moving to different habitats, predator-me-diated resource partitioning can result. Predation hasbeen shown to play a significant role in maintainingsegregation among aquatic organisms (Mercurio et al.1985, Sih et al. 1985, Werner and McPeek 1994, Kruukand Gilchrist 1997), birds (Martin 1996), and desertrodents (reviewed by Kotler and Brown 1988). Mech-anisms of this kind have been proposed as segregatingspecies of ungulates (Jarman 1974, Sinclair 1985), but

Manuscript received 6 November 2000; revised 15 September2001; accepted 6 October 2001; final version received 29 October2001.

1 Present address: Department of Psychology and Neuro-science, University of Lethbridge, Lethbridge, Alberta, Can-ada T1J 3M4. E-mail: [email protected]

there has been no direct test of this idea, primarilybecause predators are rarely observed hunting theselarge prey and their size restricts experimentation.

Rather than leading to segregation, there is evidencethat the risk of predation can lead to the convergenceof ungulate species. Antipredator benefits of mixed-species groupings have been identified for ungulates(FitzGibbon 1990), as well as for birds (Burger 1984)and primates (Noe and Bshary 1997). The segregationof ungulates has been attributed to body size, feedingpreferences, and competition, with the underlying pre-mise that interspecific competition originally led tothese forms of niche separation (Gwynne and Bell1968, Sinclair 1979, Wilmshurst et al. 2000). However,studies taking advantage of natural perturbations in theenvironment that should influence competition amongungulate species have not supported this hypothesis(Sinclair 1985, Singer and Norland 1994).

White-tailed deer (Odocoileus virginianus) and muledeer (O. hemionus) are closely related species of sim-ilar size (Mackie 1964, Wishart 1986) that occupy dif-ferent habitats in areas of allopatry and sympatry. Muledeer use more rugged and open terrain, whereas white-tails are found on more gentle terrain with greater cov-er. However, both species tolerate a wide range of hab-itats and frequently coexist (Swenson et al. 1983, Wig-

2038 SUSAN LINGLE Ecology, Vol. 83, No. 7

gers and Beasom 1986, Wood et al. 1989). Little in-formation has been obtained to explain the speciesdifference in habitat. Mule deer seem tolerant of coldertemperatures, which may be an adaptation to their moreopen habitats (Mautz et al. 1985). Studies of other fac-tors have mostly yielded negative results. Neither spe-cies directly excludes the other (Kramer 1973, Anthonyand Smith 1977). White-tails and mule deer use similarfoods when available (Martinka 1968, Anthony andSmith 1977, Krausman 1978) with no evidence of com-petition (Martinka 1968, Krausman 1978, Swenson etal. 1983, Wiggers and Beasom 1986, Wood et al. 1989).Biologists who measured the structure of topographyused by white-tails and mule deer concluded that large-scale topographical segregation precludes most oppor-tunities for feeding competition (Martinka 1968, Hud-son et al. 1976, Krausman 1978, Swenson et al. 1983,Wood et al. 1989). The only ones suggesting that feed-ing competition was likely to occur due to similar feed-ing habits failed to measure the structure of topographyused by each species (Anthony and Smith 1977, Smith1987). While the white-tail’s frequent association withagricultural land has led to the view that it is betterable to utilize succulent, higher quality foods (Mackieet al. 1998), to my knowledge no one has tested thishypothesis. The white-tail’s association with croplandmay simply reflect their use of gentle terrain, as wasfound to explain their greater use of native grassland(Wood et al. 1989).

Despite their similarities, white-tails and mule deershow pronounced differences in response to predators.White-tails exhibit a variety of antipredator signals(Caro et al. 1995) and, when attacked, typically flee toescape (Lingle and Pellis 2002). Mule deer are slowerthan white-tails and seldom able to outdistance pred-ators (Lingle 1992, 1998). Instead, they form tightlybunched groups that confront predators (Lingle 2001,Lingle and Pellis 2002). Given the difference in anti-predator behavior, it seems likely that their ability towithstand predation in specific habitats differs. White-tails and mule deer could select habitats with differingtopographical structure because of differences in theirrisk of predation or antipredator behavior associatedwith these habitats (Kramer 1972, Geist 1981). In thispaper, I present data collected while observing coyotes(Canis latrans) hunt sympatric white-tails and muledeer to test the two parts of the predation hypothesis.

Hypothesis 1.—Selective predation contributes tothe habitat (topographical) segregation of white-tailsand mule deer. Predictions. White-tails and mule deerfind the same habitat safest, but one species is able toaccept higher levels of risk in the unsafe habitat (Kotler1984, Longland and Price 1991, Christensen and Pers-son 1993, Werner and McPeek 1994, Kruuk and Gilchr-ist 1997). Alternatively, they find different habitats saf-est (Mercurio et al. 1985).

Hypothesis 2.—Differing antipredator behavior con-tributes to the habitat segregation of white-tails and

mule deer. Predictions. As a response to predators, onespecies moves further into its exclusive habitat, whilethe second species makes no change or a small-enoughchange in the same direction that segregation increases(reviewed by Sih 1987). Alternatively, each speciesmoves further into its exclusive habitat (Mercurio etal. 1985).

Either mechanism, selective predation or differingantipredator behavior, is sufficient to show that coyotepredation contributes to the habitat segregation of thesespecies. A final prediction was tested for an initial testof an alternative hypothesis.

Hypothesis 3.—Distinct feeding preferences or com-petition for similar food contribute to the habitat seg-regation of white-tails and mule deer. Predictions.White-tails or mule deer should move further into theirexclusive habitats or at least make no change to feed,compared with times when they rest.

Observations reported in this paper are restricted towinter, the season in which white-tails and mule deerare most segregated (Wood et al. 1989, Lingle 2000).Coyote predation was a significant source of mortalityfor deer at this site with most of it occurring duringwinter. Coyotes primarily captured fawns, which were5–9-mo-old during winter, but also attacked and killedsome adults (Lingle 2000, Lingle and Pellis 2002).

STUDY SITE AND SUBJECTS

The research was conducted in a 20-km2 portion ofa 225-km2 cattle ranch in southern Alberta, Canada (498N, 1128 W, elevation 1080–1380 m). The study sitewas selected to include exclusive white-tailed deer(Odocoileus virginianus) habitats, exclusive mule deer(O. hemionus) habitats, and shared habitats, each form-ing approximately one-third of the site (see map inLingle 2000). Exclusive areas were contiguous withmuch larger exclusive areas outside the study site forboth species. The landscape was open prairie, with fes-cue-dominated and mixed grassland in 83% of the studyarea. The remainder had been cultivated until the mid-1980s and was since reseeded with introduced grasses.Most habitat variation was topographical. There weretwo slope systems, an escarpment with slopes rising150 m from their base and a river valley with slopesrising 10–60 m, with rolling or flat terrain elsewhere.Census data showed that ;135 mule deer and 250white-tailed deer lived in the study area in the winterfrom 1994 to 1996 (Lingle 2000). Seven coyote packshaving dens in or near the study area in summer wereobserved hunting deer in winter. More recent obser-vations of radio-collared individuals indicate that ad-ditional coyotes may also have hunted deer in the area.

METHODS

Observation and definition of hunts

Coyote packs went on regular excursions duringwhich they hunted deer, typically around midday in

July 2002 2039PREDATOR-MEDIATED SEGREGATION IN DEER

FIG. 1. Illustration of coyote hunt of deer. Research was conducted in a 20-km2 portion of a 225-km2 cattle ranch insouthern Alberta, Canada. The coyote’s search route is shown, along with groups of mule deer that were present. The stageto which each interaction progressed is indicated by size and darkness of symbols. For definitions of hunt stages, see Methods:Hunt stages. The gap in the coyotes’ route is due to travel during an attack. Coyotes commenced their search at the pointshown after the gap.

winter (Lingle 2000). The entire excursion was definedas a hunt, during which coyotes could encounter severalgroups of deer or none at all. In a hunt, an average of4.4 coyotes traveled at a fast walk with group memberswalking in line, following a leader, who usually re-mained the same throughout the hunt. They appearedto search for deer by looking around as they traveled,pausing to scan from high points and looking directlyat visible deer. Hunts of small prey, typically voles orground squirrels, were distinguishable from hunts ofdeer by smaller pack sizes, the absence of a fixed leaderor formation, and by different forms of searching andattack behavior (Wells and Bekoff 1982, Lingle 2000).Hunts were classified as ‘‘single species’’ if coyotesrestricted their travel to a habitat used exclusively byone deer species or as ‘‘both species’’ if they traveledthrough a shared habitat or began in a habitat used byone deer species and then moved to a habitat used bythe other.

Hunts of deer were observed by sitting at a vantagepoint located relatively far from the coyotes (500–1000m), which enabled monitoring of coyotes as theymoved amongst different groups of deer, often for theentire hunt. Four main vantage points were used inwinter, each offering two or three distinct views. Allprovided views of exclusive white-tail habitats, threeprovided views of shared habitats, and two providedviews of exclusive mule deer habitats. After spottinga pack starting to hunt, groups of deer that were presentwere plotted on a topographical map. The coyotes’route was drawn on the same map (Fig. 1). Mapped

locations of deer and coyotes were used to obtain large-scale habitat information while smaller scale features,habitat changes, group composition, and behavioral re-sponses were recorded on audiotape and transcribedfollowing the hunt. This method of viewing hunts wasused to collect data on 39 hunts in the winter of 1995–1996. The same behavioral and habitat characteristicswere recorded for deer encountered by coyotes whileobserving smaller portions of hunts the two previouswinters.

Hunt stages

Predators must complete several steps before a huntis successful (Wells and Bekoff 1982, Endler 1986). Inthis paper, I identify groups that were ‘‘passed,’’ ‘‘en-countered,’’ ‘‘approached,’’ ‘‘attacked,’’ and ‘‘cap-tured.’’ The stage of being ‘‘passed’’ only applied todeer on slopes. Coyotes generally followed a routeroughly parallel to the base of a slope. Two lines weredrawn from the base to the top of the slope, with thefirst intersecting the coyotes’ starting point and the sec-ond intersecting the spot where they finished hunting.The portion of slopes lying between the two lines wasconsidered the region in which coyotes passed deer. Agroup of deer was defined as being passed when thecoyotes’ main travel route (excluding sharp turns toapproach a group) was at its shortest distance to thegroup. Deer could be passed regardless of the distancebetween them and the coyotes (Fig. 1). If coyotes trav-eled directly up-slope, which was only observed for


one short section of two hunts, they were defined aspassing deer within 400 m.

Coyotes ‘‘encountered’’ deer when they were within200 m and either looked at the deer or there appearedto be no obstructions preventing detection. An ‘‘ap-proach’’ occurred when coyotes started to stalk, walk,or run toward a group or toward an individual deer.‘‘Attacks’’ were intense attempts to capture individualdeer, either by chasing or lunging, that appeared inimmediate danger of being captured. Deer that wereattacked could escape without serious injury or be cap-tured. The last category, ‘‘captured,’’ was used to de-scribe deer that were killed immediately and also thosewounded seriously enough to die later as a result ofthe attack.

Social and habitat characteristics

Group size, the number of juveniles, and group type(female, male, mixed sex, yearling male or juvenile,following Hirth 1977) were recorded for each group ofdeer present during a hunt, a group being defined asan aggregation of animals in which each individual hadanother deer within 50 m (Clutton-Brock et al. 1982).A group size category was recorded when there wasinsufficient time to make an exact count (‘‘x-small,’’one adult, with or without fawns, or less than threefawns with no adult; ‘‘small,’’ 2–5 deer excluding x-small groups; ‘‘medium,’’ 6–10 deer; ‘‘large,’’ 11–20deer; ‘‘x-large,’’ .20 deer). The median value in thesize category was used when exact size was not avail-able to estimate per capita attack and kill rates.

Habitat characteristics other than snow depth wereidentified for each group of deer present during a hunt.Variation in topography included ‘‘slopes’’ (covering28% of the study area), ‘‘steep rolling terrain’’ (14%),and ‘‘gently rolling or flat terrain’’ (58%), outlined ona 1:50 000 topographical map having 7.6-m contourintervals at the start of the study (Lingle 1998). Rollingterrain was undulating terrain that was not embeddedin either slope system; it was considered steep if itformed hills .58 and 7.6 m in height, gentle if ,58.Geographic Information System data and PAMAP soft-ware (PAMAP GIS [1995] version 4.2, Essential Plan-ning Systems, Victoria, British Columbia, Canada)were used to produce a map showing steepness ofslopes within 25-m2 cells throughout the study site.Slope was distinguished into 58 increments, poolingthose .158. Height of animals above the base of a slopewas identified using the topographical map. Height wasrecorded for coyotes every 500 m along their route andused to calculate a median value for each hunt. Thetopography and, for those found on slopes, height offresh carcasses (estimated ,24 h old) attributed to coy-ote predation were also identified. Signs of predationincluded bite marks, blood and fur where contact wasmade prior to death, and tracks showing the animals’paths merging and leading to the kill site. To minimizebias in their location I only used carcasses found while

walking or driving the route established for censuses.Carcasses were usually spotted after seeing coyoteseating.

Because of its homogeneity on the prairie landscapein winter, the structure of vegetation was lumped intotwo categories for analysis, ‘‘short and thin’’ (,40 cmhigh, obscuring ,50% of the portions of the animal inthe vegetation) and ‘‘taller or denser.’’ The former con-sisted of desiccated forbs and grass while the lattertypically consisted of shrubs ranging from 0.4 to 2.0m in height, which rarely concealed deer in winter dueto a lack of foliage. Animals were considered as beingassociated with taller or denser vegetation if they werewithin 2 m. Snow depth was measured in a gently roll-ing (#58) area each day a hunt was observed. If thedepth of snow differed at a location where coyotesencountered deer, these values were adjusted using theheight of snow relative to the animals’ bodies.

Data analysis: predation hypotheses

The main data set used to analyze the relationshipbetween habitat traits and deer vulnerability was from1995–1996, when deer locations were mapped, becausethis data set depicted habitat traits for all groups pre-sent, not only those encountered or attacked. Habitattraits of groups as they were passed, before makingchanges, were used in this analysis. Groups were ex-cluded if they joined or were joined by another groupbeing pursued.

Data from the winter of 1995–1996 were also usedto examine habitat features of deer before and afterhunts, while data from all three winters were used toanalyze habitat changes and the responses of deer inmixed-species groups. Only groups that moved wereincluded in the analysis of habitat changes. Data forgroups that joined other groups were used to describefeatures used before hunts, but excluded from the anal-yses of habitat changes or features used after hunts incase a group moved simply to join another group andso that a newly formed group would not be representedtwice.

Habitat characteristics were based on the majority ofindividuals in a group for all analyses. If a group splitand made different changes, the portion involved in themost advanced hunt stage (e.g., attacked rather thanencountered) was used. If this distinction could not bemade (two white-tail groups, no mule deer), charac-teristics were used for the largest portion of the group.Characteristics of deer groups, including size, com-position, and several habitat traits, changed frequentlyduring the day so were unlikely to be the same duringmore than one hunt, thereby avoiding potential prob-lems with pseudoreplication. It was more important toavoid repeated observation of individuals at the attackstage, since this is when coyotes focused on one in-dividual. Based on locations at which individuals wereattacked, physical differences among individuals (age,sex, natural markings, and the presence of ear tags on


TABLE 1. Height on slopes (in meters) of deer before and after coyote hunts at the study sitein southern Alberta, Canada (1995–1996 data) and change in height for groups that movedwhen they had the option of moving in any direction (data from all three winters).

Time

Mule deer

n median IQR

White-tails

n median IQR

Mann-Whitney

z P

Before†After†Change

817131

616916

38–8446–91

5–21

483819

158

28

8–233–15

223–0

27.9928.1024.82

,0.0001,0.0001,0.0001

Note: Abbreviations: n 5 number of groups; IQR 5 interquartile range.† There was a significant difference between height of groups before and after hunts for both

species (Wilcoxon paired signed ranks test: mule deer, z 5 23.86, P , 0.0001; white-tails, z5 23.06, P 5 0.002).

20% of fawns), the outcome of attacks (some werekilled so could not be observed a second time), it wasunlikely more than one attack was observed on anyindividual.

Precautions were taken to identify whether speciesdifferences in behavior were due to species-typical re-sponses or due to differences in their starting habitatsand options. The type of habitat change made by eachgroup was used to compare changes made by the twospecies instead of simply comparing data for a group’shabitat before and after coyote hunts. The former ap-proach restricts data to groups for which both startingand end points were known and does not conceal hab-itat changes made in different directions if these dependon different starting points. Only groups near a topo-graphical boundary were used to analyze changes intopography: those within 100 m of slopes if on rollingterrain or those under 30 m if on slopes. For directionmoved and changes in height, deer were only includedwhen they were above the base (over 3.5 m) and belowthe top of a slope and so had the option of moving inany direction.

Habitat use while feeding

Fawns were marked with ear tags within a few daysof birth and resighted regularly thereafter (Lingle2000). The type of topography and height on slope wererecorded during these sightings and used to comparehabitat traits of fawns at times they were active, soprimarily feeding, with times when they rested. Theaverage proportion of sightings in each topographicaltype and average height on slope were used as datavalues for individual fawns.

Statistical analyses

Standard parametric and nonparametric tests wereused for analyses (Siegel and Castellan 1988, Sokaland Rohlf 1995). Williams’s correction was applied toG values and a was adjusted using the sequential Bon-ferroni correction in subsequent pairwise tests to main-tain the experimentwise error rate at 0.05. Z values arereported for Mann-Whitney tests when either n1 or n2

. 20; U is reported for smaller samples. Reported Pvalues are two-tailed.

RESULTS

Species differences in habitat before hunts

Mule deer were more likely to stand on slopes,whereas white-tails were more likely to occupy gentlyrolling or flat terrain or steep rolling terrain (98% of83 mule deer groups on slopes; 30% of 143 white-tailgroups on slopes, 19% on steep rolling, and 51% ongently rolling or flat terrain; G test, G 5 120.84, df 52, P , 0.0001). When on slopes, mule deer assumedhigher steeper positions than white-tails (Table 1,height; median slope, 11–158 for mule deer vs. 0–58for white-tails, Mann-Whitney z 5 26.48, P , 0.0001).Based on these results and census data (Lingle 1998),the higher steeper slopes, $61 m, were defined as theexclusive habitat of mule deer in winter and gentlyrolling terrain as the exclusive habitat of white-tails.Lower gentle slopes, ,61 m high and typically ,108,were physically intermediate between the exclusivehabitats of the two species and therefore defined as anintermediate habitat. Much of this area was shared bythe two species.

Hypothesis 1: selective predation by coyotescontributes to habitat segregation

Where do coyotes hunt deer, and how is this relatedto deer distributions?—Coyotes hunted deer both onrolling terrain and on slopes. When on slopes, theygenerally followed gentle routes having a median slopeof 6–108. Coyotes travelled significantly lower than themedian height at which mule deer occurred (Fig. 2,Wilcoxon paired sign rank test, z 5 23.7, n 5 20 hunts,P 5 0.0002), and this was also lower than the medianheight of deer when both species were included (z 522.80, n 5 20, P 5 0.005). There was no differencebetween the height of coyotes and white-tails in huntsin which coyotes passed both deer species on slopes(Fig. 2, z 5 20.63, P 5 0.53), and no indication of adifference when coyotes hunted white-tails on slopeswith no mule deer present (median, interquartile range5 5 m, 4–9 m for coyotes; 10 m, 6–15 m, n 5 5 forwhite-tails, too few hunts to compare statistically).

Does height on slope affect risk for mule deer?—Mule deer standing high on slopes were at far less risk


FIG. 2. Height on slope of coyotes and deer during coyotehunts of deer. Boxes represent interquartile ranges, horizontallines within boxes show medians, and capped bars represent10th and 90th percentiles. Data shown are for hunts in whichcoyotes traveled in habitats used strictly by mule deer (5hunts) or in habitats used by both species (15 hunts). Theypassed and encountered white-tails in 13 of these.

FIG. 3. Effect of height on slope on vulnerability of muledeer. (a) Height of mule deer groups involved at each huntstage (mean 6 1 SD). Arrows indicate that groups involvedat each hunt stage resulted in two subsets of data by thesubsequent hunt stage: groups for which predation attemptsended (end) and groups for which predation attempts contin-ued (cont) to the more advanced stage. The number of groupsreaching each stage is shown below each hunt stage. Datashow differences between the height on slope of groups whencoyotes first passed them, not changes made while coyoteswere present. (b) Effect of relative position of mule deergroups on their vulnerability. A hunt section is a portion ofa hunt that includes groups passed from the time that coyotesstarted searching for deer to the time they made an approach.If coyotes made more than one approach during a hunt, itwould have more than one hunt section. Vertical lines connectthe height on slope of groups that were approached with theaverage height on slope of groups that were encountered orsimply passed during the same hunt section. Numbers belowthe x-axis indicate different hunt sections.

than mule deer low on slopes. They were less likely tobe encountered or attacked (Fig. 3a, unpaired t test,encounter of pass, t 5 5.4, df 5 67, P , 0.0001; attackof approach, t 5 3.7, df 5 17, P 5 0.002; parametrictests are used here but not for interspecific comparisonsbecause height is only distributed normally for muledeer). Only 2 kills were made from 10 attacks, but itis important to note that data for kills were consistentwith the relationship between height and vulnerabilityfound for earlier hunt stages (Fig. 3a).

The height of a group relative to other mule deerpresent during the same hunt also affected their risk.When data from all hunts were considered, there wasa wide range in heights at which coyotes approachedmule deer, from 0 to 84 m, and an unpaired comparisonrevealed only a tendency for height to affect the prob-ability of being approached once encountered (Fig. 3a,unpaired t test, t 5 1.9, df 5 35, P 5 0.06). However,a pattern emerges when looking at individual sectionsof a hunt, which includes groups passed from the timethat coyotes started searching for deer to the time theymade an approach. In each hunt section, the approachedgroup was significantly lower than the average heightof groups that were simply encountered (Fig. 3b, pairedt test, t 5 5.8, n 5 9, P 5 0.0004). In fact, coyotesapproached the lowest mule deer group with fawns pre-sent in 13 cases for which a comparison could be made.Nonetheless, coyotes did not indefinitely travel higherto find mule deer to attack. They were less likely toapproach or attack mule deer at any time during a huntif all deer were above 61 m than if at least one groupstood below this height (an approach in all 13 huntsin which mule deer were below 61 m and an attack in7 of these vs. an approach in 3 of 7 hunts in which all

mule deer were high and no attacks; Fisher’s exact test,approach, P 5 0.007; attack, P 5 0.04).

Even though the position of other mule deer appearedto affect an animal’s risk, there was no indication theywere affected by the presence of white-tails. Coyoteswere as likely to approach and attack mule deer en-countered during hunts in which they traveled in hab-itats also used by white-tails (at least one approach andattack in 80% and 57%, respectively, of 15 hunts) asduring hunts in exclusive mule deer habitats (80% and


FIG. 4. Effect of height on slope on vulnerability of white-tailed deer (WT) in comparison with mule deer (MD). Theproportion of groups encountered by coyotes that were in-volved at more advanced hunt stages is shown. Numbers ofencounters appear in parentheses. Groups in low habitats in-clude those encountered below 30.5 m on slopes, includingthose on gently rolling terrain. Whether or not coyotes huntedin a habitat used by one or both species affected their interestin white-tails but not in mule deer (see Results). Consequent-ly, only white-tail groups are distinguished by this trait.

40% of 5 hunts; Fisher’s exact test, approach, P . 0.99;attack, P 5 0.66).

Attacks observed in 1994–1995 and attacks in pro-gress when first seen similarly occurred at low heights(eight attacks on fawns with only one escaping, height5 12 6 20 m [mean 6 1 SD]; three attacks on adultsseen from the start resulting in no kills, 20 6 19 m).Five of the attacks on fawns involved animals that werealready wounded so were heavily weighted towardkills. Mule deer carcasses attributed to coyote predationwere also on gentle terrain. Over one-third were foundon flat or gently rolling terrain away from slopes (5 of13, excluding those resulting from attacks mentionedabove). Those found on slopes tended to be low (11 610 m, n 5 8).

The relationship between height and predation riskfor mule deer cannot be explained by group member-ship. Group size was not related to the height of groups(ANOVA, F 5 1.7, df 5 3, 64, P 5 0.18, x-small andsmall groups, large and x-large groups were pooled).Furthermore, there was no difference between theheight of groups with or without juveniles (unpaired ttest, t 5 21.2, df 5 67, P 5 0.23). In fact, the meanheight of groups with fawns tended to be higher ratherthan lower as would be expected if fawns attractedcoyotes to lower heights (mean 5 64 m for groups withfawns vs. 45 m for groups having only adults).

Height of a group was not correlated with its veg-etation structure (rs 5 0.06, P 5 0.64, n 5 58 groups).There was no relationship between snow depth and theoutcome of hunts (rs 5 0.20, z 5 0.85, P 5 0.39, n 520 hunts), indicating the effect of height cannot be dueto variation in that variable. The amount of standingsnow was greater that year than usual due to a highersnowfall, but still small compared with most northernlocations (snow depth during hunts, median, inter-quartile range 5 10 cm, 8.75–15 cm).

Does height affect predation risk for white-tails?—White-tails did not appear disadvantaged by using gen-tle terrain or by standing lower than other deer. Coyotespaid little attention to white-tails when hunting in hab-itats having both species, even though they encounteredwhite-tails lower than mule deer in 13 of these hunts.Coyotes approached white-tails in 7% of 15 hunts inhabitats used by both species and made no attacks, butmade an approach in 79% of 19 hunts in exclusivewhite-tail habitats, attacking deer in 21% of these (ap-proach, G 5 19.24, df 5 1, P , 0.0001; attack, G 54.49, df 5 1, P 5 0.03). Subsequent analyses of white-tail vulnerability were therefore restricted to single spe-cies hunts. When coyotes hunted on slopes used ex-clusively by white-tails, they still showed no tendencyto approach the lowest white-tail group in three huntsfor which a comparison could be made, making noapproaches in two other hunts.

White-tails in the most gentle habitat of all, rollingor flat terrain, were at no more risk of being approachedthan white-tails on slopes (22% of 94 groups on rolling

terrain vs. 20% of 15 on slopes, G test, G 5 0.04, df5 1, P 5 0.84). These groups were pooled for com-parison with mule deer at low heights. Coyote en-counters with white-tails were less likely to result inattacks than encounters with mule deer at low heightsand, despite a small sample of kills, showed a nonsig-nificant tendency to result in fewer kills (Fig. 4, G testfor low mule deer, all hunts vs. low white-tails, singlespecies hunts, attack, G 5 14.20, df 5 1, P 5 0.0002;kill, G 5 2.84, df 5 1, P 5 0.09). There was no dif-ference in the rate at which white-tails and mule deerlow on slopes were encountered (94% of 16 white-tailgroups vs. 93% of 15 mule deer groups), but both ofthese groups were encountered significantly more oftenthan mule deer higher on slopes (28% of 36 groups; G5 32.50, df 5 2, P , 0.0001). When data are adjustedto show risk to individuals, 9.5% of mule deer indi-viduals encountered below 30.5 m were attacked and2.4% were killed (mean group size for deer encounteredat low heights was 6.5 for mule deer, 7.6 for white-tails). In contrast, only 0.51% of encountered white-tails were attacked and 0.26% were killed. No muledeer encountered above 61 m were attacked or killed(Fig. 4).

Hypothesis 2: differing antipredator behaviorcontributes to habitat segregation

The species difference in risk in gentle terrain wasreflected in their antipredator behavior. Mule deer weremore likely to move up-slope, either directly or at anangle. White-tails were more likely to move down(80% of 30 mule deer groups moved up-slope, 20%cross-slope; 29% of 17 white-tail groups moved cross-


slope and 71% downslope; G 5 38.39, df 5 2, P ,0.0001), increasing differences in height and topog-raphy seen before hunts (Table 1). If deer started onslopes, mule deer were more likely to be on slopes andwhite-tails on rolling terrain once coyotes passed(100% of 19 mule deer groups stayed on slopes vs.64% of 28 white-tail groups, G test, G 5 11.52, df 51, P 5 0.0007). If deer started on rolling terrain, muledeer were more likely to have moved to slopes whilewhite-tails remained on rolling terrain (0 of 3 muledeer groups stayed on rolling terrain vs. 100% of 11white-tail groups, Fisher’s exact test, P 5 0.003).

The difference between the behavior of white-tails(wt) and mule deer (md) was not explained by thedirection from which coyotes approached. Mule deerwere still more likely to have increased their heightwhen the analysis was restricted to groups approachedfrom cross-slope (Mann-Whitney test, z 5 24.47, nmd

5 22, nwt 5 16, P , 0.0001). The species differencein response was not explained by the intensity of threatgiven it was found when the analysis was restricted togroups involved in the same hunt stage: groups thatwere encountered but not approached (U 5 7.5, nmd 510, nwt 5 8, P 5 0.003). Finally, the difference inbehavior was not restricted to groups having fawns.The species difference in height at the start of huntsalso applied to groups consisting only of adults, as didthe change in height (height at start, z 5 22.94, nmd 511, nwt 5 17, P 5 0.003; change in height, U 5 0, nmd

5 6, nwt 5 6, P 5 0.004). Too few adult groups wereavailable to test for a sex-related difference. However,male groups that were present and moved, includingtwo white-tail and three mule deer groups, traveled ina similar direction to others of their species.

Species-typical habitat changes were observed whencoyotes encountered mixed-species groups or groupshaving a heterospecific neighbor within 100 m. Sixmixed associations disbanded when encountered within100 m by coyotes, after white-tails and mule deer ex-hibited different patterns of habitat use (six mixedgroups disbanded during two encounters and four at-tacks, binomial test, P 5 0.03). White-tails either re-mained on or fled to rolling terrain, while all mule deerfrom these groups remained on or moved to slopes,with the mixed-species association ending as a result.

Hypothesis 3: feeding preferences or competitioncontribute to segregation

In contrast to their response to coyotes, groups withmule deer fawns were more likely to leave slopes orto move downslope when active, so mostly feeding,compared with times when they rested (topography,median proportion of sightings on slopes when activeor resting, 85% vs. 100%; Wilcoxon signed-ranks test,six mule deer fawns less likely to occupy slopes whenactive, four ties, z 5 22.20, P 5 0.02; height on slope,median height when active or resting, 65 m vs. 76 m;paired t test 5 2.5, n 5 9, P 5 0.04). White-tails

showed no difference between the type of topographythey used when active or resting (Wilcoxon signed-ranks test, four of five fawns less likely to occupyslopes when active, three ties, z 5 1.48, P 5 0.14).Any tendency for white-tails to move, which couldhave been overlooked due to the small sample, was touse slopes less often when active than when resting(median 14 vs. 47% of sightings, respectively), similarto mule deer.

DISCUSSION

Coyote predation and habitat segregation

These results suggest coyote predation contributesto the habitat segregation of white-tails and mule deerin winter by at least two mechanisms, selective pre-dation and differing antipredator behavior. In supportof hypothesis 1 (selective predation contributes to seg-regation), observations of coyote–deer interactions andcarcass data indicated that mule deer were more vul-nerable in shared or in atypical habitats (i.e., low onslopes or gently rolling terrain) than in their exclusivehabitat (high on slopes), while white-tails were not sim-ilarly disadvantaged by standing low on slopes or ongently rolling terrain. This was true even though white-tails were 2.5 times as common as mule deer in thewinter of 1995–1996 (Lingle 2000) and nearly alwaysoccupied gentle terrain, while mule deer tended to oc-cur higher on slopes. In support of hypothesis 2 (an-tipredator behavior contributes to segregation), thepresence of coyotes led the two species to move indifferent directions even when starting in the same hab-itat or in mixed species groups.

Results indicated the relationship between height andvulnerability of mule deer was not due to confoundingfactors such as vegetation structure, snow depth, or thecomposition of groups. Although snow depth was notrelated to the outcome of hunts, it is important to realizethere was little standing snow compared with other siteswhere snow depth has been reported to be related tothe risk of predation for ungulates (Mech et al. 1987,Paquet 1992, Huggard 1993). Species differences invulnerability and behavior were also not due to packswith different skills hunting different species of deer.Five of the seven packs were observed hunting bothspecies, and packs hunting white-tails and mule deerwere similar in size (Lingle and Pellis 2002).

The different habitat changes made by white-tailsand mule deer in response to coyotes were not deter-mined by their starting habitats. Two outcomes couldhave been expected if that were the case. First, twospecies may become more similar if they start in dif-ferent habitats but prefer and so move to the samehabitat. Second, species may modify their behavior inresponse to the habitat they occupy; for instance, Ves-per Sparrows, Pooecetes gramineus, seek cover whennear shrubs, but make evasive maneuvers when farfrom cover (Pulliam and Mills 1977). If this occurs, a


species will show a bimodal response that depends ontheir starting position. Changes in topography andheight were compared for white-tails and mule deerthat occurred in shared habitats and had overlappingstarting positions. Despite these restrictions, the twospecies moved in different directions and each speciesmade unimodal changes in height and topography.

This paper examined whether predation risk con-tributed to broad interspecific differences in habitat use.Secondary analyses indicated that groups containingonly adults, including all-male groups, moved in thesame direction in response to coyotes as did groupshaving fawns. Finer-scale differences in behavior arestill likely to exist between the age and sex classes ofeach species that were overlooked due to the smallsample of adult groups (Berger 1991, Bleich 1999), butthey do not appear to conflict with the species-typicalpatterns identified in this study. It is important to em-phasize that these results apply to winter, the time whenthe species are most segregated and the sexes the least(Lingle 1998). From June to August, many white-tailmothers rear their newborn fawns on slopes alongsidemule deer females. The habitats that are safest are likelyto change if behavioral defenses and vulnerability ofprey change with maturity.

In contrast to their response to coyotes, mule deerleft slopes or moved downslope to feed, bringing themcloser to white-tails. This trend is in conflict with hy-pothesis 3, that distinct feeding preferences or com-petition for similar food items contribute to the seg-regation of these species. This result is consistent withstudies reporting similar feeding habits (Martinka1968, Anthony and Smith 1977, Krausman 1978) withno evidence of competition between white-tails andmule deer (Martinka 1968, Krausman 1978, Swensonet al. 1983, Wiggers and Beasom 1986, Wood et al.1989). This suggests mule deer abandoned safe posi-tions on slopes to obtain food, a finding consistent withreports for other ungulates associated with rugged ter-rain (Bergerud et al. 1984, Festa-Bianchet 1988, Berger1991) and, more generally, with reports for animalsliving in a variety of habitats that make a trade-offbetween foraging and predation risk (reviewed by Limaand Dill 1990). The benefit obtained by these deer byfeeding on gentle terrain has not been identified andcould involve quality, quantity, or access to a specificresource.

While altitude is known to correspond to variationin food quality (Festa-Bianchet 1988, Albon and La-ngvatn 1992), the base of two slope systems within thestudy site and one just outside it started at differentelevations. Height on a slope did not correspond to anelevational gradient, other than within a particularslope system. Mule deer used the full range of slopespresent. White-tails used gentle terrain above and be-low the different slope systems, as did deer of bothspecies that left slopes to feed.

This was not a case of certain individuals spending

their lives in high rugged habitats while others stayedin low gentle ones, in which case a difference amongindividuals might account for the risk faced in lowhabitats. Some mule deer lived in areas encompassingmore rugged terrain than areas used by other individ-uals, and they might have been at less overall risk.Nonetheless, all individuals had a home range includ-ing gentle terrain and the tendency for coyotes to en-counter and attack animals standing in gentle terrainwas found within each area. Most mule deer moved upand down during the course of a day or between days,affecting their risk on a short-term basis. If coyoteswere simply looking for vulnerable individuals at lowheights and were not deterred by rugged terrain, theyshould have persisted to move into rugged areas to findprey on days when all deer stood high. The same patternof leaving slopes or moving downslope to feed wasfound for all individuals, not only some as would haveoccurred if dominant animals forced others out of pre-ferred feeding locations. Although these data were ob-tained for tagged fawns, these animals occurred ingroups with their mothers, other females, and oftenmales so cannot be viewed as a subordinate class thatcould be pushed into an undesirable habitat (Cresswell1994). It is certainly possible individuals in poor con-dition were more likely to move to low habitats to feed.While condition could compound their risk, the lowhabitat is still the risky one. Otherwise, animals in goodcondition would also move there.

Detailed studies of feeding behavior and physiologyare needed to assess their role in the habitat selectionand segregation of these species. What matters here isthat no factor other than predation risk can explain thepatterns observed in this study, specifically, why thetwo species moved in different directions in responseto coyotes or why mule deer, but not white-tails, werevulnerable in gentle habitats. These findings have sig-nificant implications for wildlife management. Even ifwhite-tails and mule deer have similar needs for food,they require different habitats for safety. Mule deersurvival could be threatened by the elimination of rug-ged terrain or by construction of barriers that interferewith their ability to move into rugged areas.

Unlike most examples of predator-mediated resourcepartitioning (but see Kruuk and Gilchrist 1997), thesetwo species are able to interbreed and their female off-spring are fertile (Wishart et al. 1988). Anything lead-ing to their physical separation during the Novemberbreeding season will contribute to their reproductiveisolation. These mechanisms, selective predation anddivergent habitat changes, should reduce their oppor-tunities to meet potential mates of the other species.Nonetheless, deer can and do move long distanceswhen searching for mates (Geist 1981), so it is unlikelythese are the only or even the most important factorsrestricting interbreeding.

The type of topography and height on slope proveduseful to distinguish large-scale differences between


white-tail and mule deer habitats at this site. However,it is important to acknowledge that the apparent effectof height was undoubtedly due to the ruggedness ofterrain at different heights (Lingle 1998). Height haslimited application given there are times when animalshave to travel downslope to find rugged terrain. More-over, mule deer are known to inhabit rolling terrainwhen it is sufficiently steep or when topographical re-lief is near (Hudson et al. 1976, Wood et al. 1989). Acomposite variable, incorporating different compo-nents of rugged terrain such as height, slope, breaks interrain, and distance to topographical relief, shouldhave broader application. Indices of the ruggedness ordiversity of terrain have been developed (Nicholson etal. 1997) that may be useful for comparing the overallquality of mule deer habitats in avoiding predation.However, an index of terrain needs to incorporate ananimal’s position relative to specific habitat features toassess how individual animals use these features tothwart predators (Lingle 1998). For instance, one an-imal may be vulnerable standing in the base of a gullydue to late detection of predators and easy access bypredators. Another may be safe standing 5 m higher onthe gully’s edge, where it can both detect predatorsearly and make access more difficult.

Prey behavior and habitat selection

In contrast to prey involved in other examples ofpredator-mediated segregation (Kotler and Brown1988, Christensen and Persson 1993, Kruuk and Gilchr-ist 1997), white-tails and mule deer are similar in sizeand morphology. Antipredator strategies, rather than aphysical difference, are likely to explain why the spe-cies differed in their risk in gentle terrain and why theymoved to different habitats. White-tail juveniles havelittle difficulty galloping faster than coyotes on levelterrain by winter (Lingle 1998), which may enablethem, and their associates, to accept more overlap withpredators. Mule deer typically employ a jumping gaitcalled a ‘‘stot’’ rather than the gallop (Eslinger 1976,Lingle 1992). Regardless of their gait, mule deer ju-veniles are slower than white-tails and coyotes, makingthem vulnerable if they flee (Lingle 1998, Lingle andPellis 2002). Mule deer are successful in avoiding pre-dation if they bunch together and attack coyotes, butthey sometimes fail to employ these tactics (Lingle2001, Lingle and Pellis 2002). These results show thatmule deer are less able than white-tails to avoid attackand capture once encountered on gentle terrain. Thealternative for mule deer, and their first line of defence,is to remain in high and rugged habitats as much aspossible where they can minimize their exposure topredators.

Mule deer standing high on slopes reduced both theirrisk of being encountered and attacked, advantages thathave been proposed in other studies of ungulates oc-cupying rugged terrain (Murie 1944, Geist 1971, Ber-gerud et al. 1984, Festa-Bianchet 1988, Berger 1991,

Main and Coblentz 1996, Bleich et al. 1997, Bleich1999). A more detailed analysis of the way in whichmule deer use elements of rugged terrain revealedstrong tendencies for them to move directly up-slopewhen detecting coyotes early or when encountered(Lingle 1998). This movement increased the steepnessof slope beneath them, but not the steepness of groundwhere they stood. If mule deer fled when attacked, theywere more likely to move up-slope at a shallow angle.In fact, they lost ground rapidly if they fled directlyup-slope (Lingle 1998). These observations suggestthat mule deer use rugged terrain to impose physicalbarriers on coyotes to avoid encounters and dissuadeattack, rather than for escape by flight (Lingle 1998).Mule deer may also gain an offensive advantage simplyby standing above coyotes. Coyotes frequently aban-doned pursuits once mule deer moved to small knolls,even though there was no physical barrier blockingthem from the deer.

White-tails seldom occurred on steeper slopes inwinter, so it was not possible to determine whether theywere safest on gentle terrain or simply capable of tol-erating more risk than mule deer. That white-tails al-ways remained on rolling terrain or frequently movedfrom slopes to rolling terrain in response to coyotessuggests they perceived this type of topography as saf-est in winter. White-tails might be safer in rolling areasif advantages in escape, once attacked, outweigh theapparently disadvantageous encounter rates (Lima1992).

During this study, I observed species that adjustedtheir patterns of habitat use during their interactionswith predators. Such flexibility in habitat use shouldnot automatically be expected, even if prey select cer-tain habitats to lower their risk. Whether or not a be-havioral response is seen in response to predatorsshould depend on the hunt stage at which a habitatfeature is useful. If an animal inhabits a depression tohide from predators, there is no reason for it to moveinto another depression once detected by a predator,unless it has the possibility of escaping from the pred-ator’s view. A second reason that flexibility in habitatuse may not be seen is that habitat preferences may befixed or limited in their flexibility. Some prey speciesremain in refuges regardless of variation in numbersof predators (reviewed by Sih 1987). Moreover, ani-mals may be able to respond to the presence of pred-ators by moving to safer habitats, without having theability to respond equally strongly to the absence ofpredators (Sih 1987). This last point is particularly rel-evant to studies of habitat selection in ungulates. Un-gulates cannot be expected to abandon rugged terrainsimply because they live in locations where predatorshave been eliminated since historical times.

The effects of predation risk on prey behavior havereceived relatively little direct study for larger animals,primarily because predation events involving such preyare seldom witnessed. Instead, biologists often form


predictions based on their assumptions of how riskcould affect prey animals. In many cases, we do notunderstand how prey respond to predators sufficientlywell to form predictions about which habitat is safestand whether different prey groups (sexes, size classes,or species) will converge or segregate in response topredators. The approach used here of mapping preygroups that are passed, encountered, attacked, andkilled is one method that can be used to directly mea-sure the risk faced in different habitats.

ACKNOWLEDGMENTS

I am grateful to the Thrall family for the opportunity toconduct research on their land and to the employees and fam-ilies of the McIntyre Ranch for their hospitality and assis-tance. I thank Catherine Bellerive and Finbarr Wilson for helpin the field; Michele Gruninger for mapping and GIS analysis;Bill Samuel and Bill Wishart for general assistance; and JoelBrown, Tim Clutton-Brock, Larissa Conradt, Nick Davies,Hans Kruuk, Loeske Kruuk, Sue McCrae, Sergio Pellis, Fin-barr Wilson, and two anonymous reviewers for helpful com-ments on the work or drafts of the manuscript. This work wasprimarily funded by an Alberta Conservation AssociationChallenge Grant in Biodiversity, with additional equipmentand assistance provided by the University of Cambridge, theUniversity of Lethbridge, and Alberta Fish and Wildlife.

LITERATURE CITED

Albon, S. D., AND R. Langvatn. 1992. Plant phenology andthe benefits of migration in a temperate ungulate. Oikos65:502–513.

Anthony, R. G., and N. S. Smith. 1977. Ecological relation-ships between mule deer and white-tailed deer in south-eastern Arizona. Ecological Monographs 47:255–277.

Berger, J. 1991. Pregnancy incentives, predation constraintsand habitat shifts: experimental and field evidence for wildbighorn sheep. Animal Behaviour 41:61–77.

Bergerud, A. T., H. E. Butler, and D. R. Miller. 1984. An-tipredator tactics of calving caribou: dispersion in moun-tains. Canadian Journal of Zoology 62:1566–1575.

Bleich, V. C. 1999. Mountain sheep and coyotes: patterns ofpredator evasion in a mountain ungulate. Journal of Mam-malogy 80:283–289.

Bleich, V. C., R. T. Bowyer, and J. D. Wehausen. 1997. Sexualsegregation in mountain sheep: resources or predation?Wildlife Monographs 134:1–50.

Brown, J. S., J. W. Laundre, and M. Gurung. 1999. Theecology of fear: optimal foraging, game theory, and trophicinteractions. Journal of Mammalogy 80:385–399.

Burger, J. 1984. Grebes nesting in gull colonies: protectiveassociations and early warning. American Naturalist 123:327–337.

Caro, T. M., L. Lombardo, A. W. Goldizen, and M. Kelly.1995. Tail-flagging and other antipredator signals in white-tailed deer: new data and synthesis. Behavioral Ecology 6:442–450.

Christensen, B., and L. Persson. 1993. Species-specific an-tipredatory behaviours: effects on prey choice in differenthabitats. Behavioral Ecology and Sociobiology 32:1–9.

Clutton-Brock, T. H., F. E. Guinness, and S. D. Albon. 1982.Red deer: behavior and ecology of two sexes. Universityof Chicago Press, Chicago, Illinois, USA.

Cresswell, W. 1994. Age-dependent choice of redshank(Tringa totanus) feeding location: profitability or risk?Journal of Animal Ecology 63:589–600.

Endler, J. A. 1986. Defense against predators. Pages 109–134 in M. E. Feder and G. V. Lauder, editors. Predator–prey relationships: perspectives and approaches from the

study of lower vertebrates. University of Chicago Press,Chicago, Illinois, USA.

Eslinger, D. H. 1976. Form, function and biological role inthe locomotory apparatus of the genus Odocoileus in Al-berta (Mammalia: Artiodactyla). Thesis. University of Cal-gary, Calgary, Alberta, Canada.

Festa-Bianchet, M. 1988. Seasonal range selection in bighornsheep: conflicts between forage quality, forage quantity,and predator avoidance. Oecologia 75:580–586.

FitzGibbon, C. D. 1990. Mixed-species grouping in Thom-son’s and Grant’s gazelles: the antipredator benefits. Ani-mal Behaviour 39:1116–1126.

Geist, V. 1971. Mountain sheep: a study in behavior andevolution. University of Chicago Press, Chicago, Illinois,USA.

Geist, V. 1981. Behavior: adaptive strategies in mule deer.Pages 157–223 in O. C. Wallmo, editor. Mule and black-tailed deer of North America. University of Nebraska Press,Lincoln, Nebraska, USA.

Gwynne, M. D., and R. H. V. Bell. 1968. Selection of grazingcomponents by grazing ungulates in the Serengeti NationalPark. Nature 220:390–393.

Hirth, D. H. 1977. Social behavior of white-tailed deer inrelation to habitat. Wildlife Monographs 53:1–55.

Hudson, R. J. D., D. M. Herbert, and V. C. Brink. 1976.Occupational patterns of wildlife on a major East Kootenaywinter-spring range. Journal of Range Management 29:38–43.

Huggard, D. J. 1993. Effect of snow depth on predation andscavenging by gray wolves. Journal of Wildlife Manage-ment 57:382–388.

Jarman, P. J. 1974. The social organization of antelope inrelation to their ecology. Behaviour 48:215–216.

Jeffries, M. J., and J. H. Lawton. 1984. Enemy free spaceand the structure of ecological communities. BiologicalJournal of the Linnean Society 23:269–286.

Kotler, B. P. 1984. Risk of predation and the structure ofdesert rodent communities. Ecology 65:689–701.

Kotler, B. P., and J. S. Brown. 1988. Environmental hetero-geneity and the coexistence of desert rodents. Annual Re-view of Ecology and Systematics 19:281–307.

Kramer, A. 1972. A review of the ecological relationshipsbetween mule and white-tailed deer. Occasional Paper 3.Alberta Fish and Wildlife Division, Edmonton, Alberta,Canada.

Kramer, A. 1973. Interspecific behavior and dispersion oftwo sympatric deer species. Journal of Wildlife Manage-ment 37:288–300.

Krausman, P. R. 1978. Deer-forage relationships in Big BendNational Park, Texas. Journal of Wildlife Management 42:101–107.

Kruuk, L. E. B., and J. S. Gilchrist. 1997. Mechanisms main-taining species differentiation: predator-mediated selectionin a Bombina hybrid zone. Proceedings of the Royal Societyof London, B 264:105–110.

Lima, S. L. 1992. Strong preferences for apparently danger-ous habitats: a consequence of differential escape frompredators. Oikos 64:597–599.

Lima, S. L., and L. M. Dill. 1990. Behavioral decisions madeunder the risk of predation: a review and prospectus. Ca-nadian Journal of Zoology 68:619–640.

Lingle, S. 1992. Escape gaits of white-tailed deer, mule deerand their hybrids: gaits observed and patterns of limb co-ordination. Behaviour 122:153–181.

Lingle, S. 1998. Antipredator behavior, coyote predation andhabitat segregation of white-tailed and mule deer. Disser-tation. University of Cambridge, Cambridge, UK.

Lingle, S. 2000. Seasonal variation in coyote feeding be-haviour and mortality of white-tailed deer and mule deer.Canadian Journal of Zoology 78:85–99.


Lingle, S. 2001. Anti-predator strategies and grouping pat-terns in white-tailed deer and mule deer. Ethology 107:295–314.

Lingle, S., and S. M. Pellis. 2002. Fight or flight? Antipred-ator behavior and the escalation of coyote encounters withdeer. Oecologia, 131:154–164.

Longland, W. S., and M. V. Price. 1991. Direct observationsof owl and heteromyid rodents: can predation risk explainmicrohabitat use? Ecology 72:2261–2273.

Mackie, R. J. 1964. Montana deer weights. Montana Wildlife4:9–14.

Mackie, R. J., D. F. Pac, K. L. Hamlin, and G. L. Dusek.1998. Ecology and management of mule deer and white-tailed deer in Montana. Montana Fish, Wildlife and Parks,Helena, Montana, USA.

Main, M. B., and B. E. Coblentz. 1996. Sexual segregationin Rocky Mountain mule deer. Journal of Wildlife Man-agement 60:497–507.

Martin, T. E. 1996. Fitness costs of resource overlap amongcoexisting bird species. Nature 380:338–340.

Martinka, C. J. 1968. Habitat relationships of white-tailedand mule deer in northern Montana. Journal of WildlifeManagement 32:558–565.

Mautz, W. W., P. J. Pekins, and J. A. Warren. 1985. Coldtemperature effects on metabolic rate of white-tailed, mule,and black-tailed deer in winter coat. Pages 453–457 in P.K. Fennessy and K. R. Drew, editors. Biology of deer pro-duction. Bulletin 22. The Royal Society of New Zealand,Wellington, New Zealand.

Mech, L. D., R. E. McRoberts, R. O. Peterson, and R. E.Page. 1987. Relationship of deer and moose populationsto previous winters’ snow. Journal of Animal Ecology 56:615–627.

Mercurio, K. S., A. R. Palmer, and R. B. Lowell. 1985. Pred-ator-mediated microhabitat partitioning by two species ofvisually cryptic, intertidal limpets. Ecology 66:1417–1425.

Murie, A. 1944. The wolves of Mount McKinley. U.S. Na-tional Park Services, Fauna Series 5:1–238.

Nicholson, M. C., R. T. Bowyer, and J. G. Kie. 1997. Habitatselection and survival of mule deer: tradeoffs associatedwith migration. Journal of Mammalogy 78:483–504.

Noe, R., and R. Bshary. 1997. The formation of red colobus–diana monkey associations under predation pressure fromchimpanzees. Proceedings of the Royal Society of London,B 264:253–259.

Paquet, P. C. 1992. Prey use strategies of sympatric wolvesand coyotes in Riding Mountain National Park, Manitoba.Journal of Mammalogy 73:337–343.

Pulliam, H. R., and G. S. Mills. 1977. The use of space bywintering sparrows. Ecology 58:1393–1399.

Siegel, S., and N. J. Castellan, Jr. 1988. Nonparametric sta-tistics for the behavioral sciences. Second edition. Mc-Graw-Hill, New York, New York, USA.

Sih, A. 1987. Predators and prey lifestyles: an evolutionaryand ecological overview. Pages 203–224 in W. C. Kerfoot

and A. Sih, editors. Predation: direct and indirect effectson aquatic communities. University of New England Press,Hanover, New Hampshire, USA.

Sih, A., P. Crowley, M. McPeek, J. Petranka, and K. Stroh-meier. 1985. Predation, competition and prey communities:a review of field experiments. Annual Review of Ecologyand Systematics 16:269–311.

Sinclair, A. R. E. 1979. Dynamics of the Serengeti ecosystem:process and pattern. Pages 1–20 in A. R. E. Sinclair andM. Norton-Griffiths, editors. Serengeti: dynamics of anecosystem. University of Chicago Press, Chicago, Illinois,USA.

Sinclair, A. R. E. 1985. Does interspecific competition orpredation shape the African ungulate community? Journalof Animal Ecology 54:899–918.

Singer, F. J., and J. E. Norland. 1994. Niche relationshipswithin a guild of ungulate species in Yellowstone NationalPark, Wyoming, following release from artificial controls.Canadian Journal of Zoology 72:1383–1394.

Smith, W. P. 1987. Dispersion and habitat use by sympatricColumbian white-tailed deer and Columbian black-taileddeer. Journal of Mammology 68:337–347.

Sokal, R. R., and F. J. Rohlf. 1995. Biometry: the principlesand practice of statistics in biological research. Third edi-tion. W. H. Freeman, New York, New York, USA.

Swenson, J. E., S. J. Knapp, and H. J. Wentland. 1983. Winterdistribution and habitat use by mule deer and white-taileddeer in southeastern Montana. Prairie Naturalist 15:97–113.

Taylor, R. J. 1984. Predation. Chapman and Hall, New York,New York, USA.

Wells, M. C., and M. Bekoff. 1982. Predation by wild coy-otes: behavioral and ecological analyses. Journal of Mam-malogy 63:118–127.

Werner, E. E., and M. A. McPeek. 1994. Direct and indirecteffects of predators on two anuran species along an envi-ronmental gradient. Ecology 75:1368–1382.

Wiggers, E. P., and S. L. Beasom. 1986. Characterization ofsympatric or adjacent habitats of 2 deer species in westTexas. Journal of Wildlife Management 50:129–134.

Wilmshurst, J. F., J. M. Fryxell, and C. M. Bergman. 2000.The allometry of patch selection in ruminants. Proceedingsof the Royal Society of London, B 267:343–349.

Wishart, W. D. 1986. White-tailed deer and mule deer. Pages134–143 in Alberta Fish and Game Association, editor.Alberta wildlife trophies. Alberta Fish and Game Associ-ation, Edmonton, Alberta, Canada.

Wishart, W. D., F. Hrudka, S. M. Schmutz, and P. F. Flood.1988. Observations on spermatogenesis, sperm phenotypeand fertility in white-tailed 3 mule deer hybrids and a yak3 cow hybrid. Canadian Journal of Zoology 66:1664–1671.

Wood, A. K., R. J. Mackie, and K. L. Hamlin. 1989. Ecologyof sympatric populations of mule deer and white-tailed deerin a prairie environment. Montana Department of Fish,Wildlife, and Parks, Bozeman, Montana, USA.

coyote predation and habitat segregation of white-tailed deer and mule deer

Documents