Social and nutritional factors influencing the dispersal of resident coyotes
ERIC M. GESE*, ROBERT L. RUFF* & ROBERT L. CRABTREE†
(Received 3 August 1995; initial acceptance 3 November 1995; final acceptance 25 March 1996; MS. number: A7379R)
Dispersal plays a major role in the regulation, spatial distribution, size and genetic structure of animal populations (Hamilton 1972; Lidicker 1975; Taylor & Taylor 1977). Although dispersal has been documented in many coyote, Canis latrans, and wolf, C. lupus, populations (e.g. Andelt 1985; Mech 1987; Fuller 1989; Gese & Mech 1991), the mechanisms triggering an animal to leave its pack or social unit are not well understood. Christian (1970) proposed the social subordination hypothesis, in which a high level of aggression from dominant animals forces low-ranking individuals to disperse. In contrast, Bekoff (1977a) proposed the social cohesion hypothesis, that individuals that do not develop strong ties to their group early in life will be most likely to disperse.
Other proximate causes for dispersing may include lack of breeding opportunities, physiological changes (Holekamp 1984, 1986), reduced food intake or availability (Messier 1985; Harrison 1992), increased social pressures associated with increased density (Snyder 1961; Van Vleck 1968) and ectoparasite load.
Among canids, captive studies of coyotes (Knowlton & Stoddart 1983) and wolves (Zimen 1976, 1981) suggest that increased aggression and reduced access to carcasses may either force a subordinate animal to disperse or cause the animal to leave voluntarily (Packard & Mech 1980). Similarly, a study of free-ranging red foxes, Vulpes vulpes, in England showed an increase in sub-adult fighting injuries at the start of the dispersal period and a greater level of bite wounding on smaller males (White & Harris 1994). In contrast, Harris & White (1992) reported that red fox pups that received more grooming than other litter-mates were less likely to disperse. They concluded, however, that neither the social sub-ordination hypothesis nor the social cohesion hypothesis alone explained the dispersal behaviour of foxes (White & Harris 1994).
Owing to the secretive and elusive nature of canids (Mech 1974; Kleiman & Brady 1978), examination of the social and nutritional factors influencing dispersal in free-ranging canid populations is difficult. Detailed observation of identified individuals is prerequisite to increasing understanding of why some animals leave their natal pack but others stay (Bekoff 1989).
Coyotes were last studied in Yellowstone National Park in the 1940s and early 1950s, after the predator control programme in the park had ceased (Murie 1940; Robinson & Cummings 1951). Coyotes in Yellowstone have since been unexploited and are now tolerant of a stationary observer (e.g. Gese & Grothe 1995). We were able to collect information on each individual coyote in five resident packs during three winters. Coyote pups spent less time feeding on ungulate carcasses than alpha and beta coyotes, suggesting resource partitioning between pack members (Gese et al. 1996a). Moreover, pups were less experienced hunters of small mammals than older coyotes (Gese et al. 1996b).
Differences in social rank combined with reduced access to carcasses may cause pups or other older individuals to disperse. Therefore, the objective of this study was to examine the interaction of the social and nutritional factors influencing dispersal of individual coyotes from their resident pack. We predicted that, compared with philopatric individuals, dispersing coyotes would (1) be less dominant in interactions with other pack members and hence would be lower-ranking individuals in the pack, (2) spend less time with other pack members, (3) have less access to ungulate carcasses and (4) be less successful capturing small mammals.
STUDY AREA AND METHODS
This study was conducted in a 70-km 2 area located in the Lamar River Valley in Yellowstone National Park, Wyoming (44)52*N, 110)11*E); elevation is about 2000 m above sea level. The climate is characterized by long, cold winters and short, cool summers (Houston 1982). Mean annual temperature and precipitation is 1.8)C and 31.7 cm, respectively, with most of the annual precipitation falling as snow (Houston 1982).
Seven habitats were identified in the study area including forest, mesic meadow, mesic shrub-meadow, riparian, grassland, sage grassland and road (Gese et al. 1996a). Major ungulate species in the study area during winter included elk, Cervus elaphus, mule deer, Odocoileus hemionus, bison, Bison bison, and big-horn sheep, Ovis canadensis. A few moose, Alces alces, and white-tailed deer, O. virginianus, were in the valley, and pronghorn antelope, Antilocapra americana, were present during summer. A major food source for coyotes during the winter was elk carrion (Murie 1940; Gese ET al., 1996a). Small mammal species in the area included microtines, Microtus spp., mice, Peromyscus spp., pocket gophers, Thomomys talpoides, and ground squirrels, Spermophilus armatus. Lagomorphs were not present in the valley.
Coyotes were captured with padded leg-hold traps with attached tranquilizer tabs (Balser 1965). Coyotes were immobilized (Cornely 1979) for handling, then weighed, sexed and radio-collared. We removed the first vestigial pre-molar from the lower jaw for aging by counting the cementum annuli (Linhart & Knowlton 1967).
Pups were captured at the den when 10–12 weeks old, ear-tagged, and surgically implanted with an intra-peritoneal transmitter. Coyotes were classified into age classes of pup ( <12 months), yearling (12–24 months), or adult (>24 months).
Coyotes were classified either as members of a resident pack or as transients. Resident packs used and actively defended one unique area or territory, and transients displayed nomadic movements over a large area (Bowen 1981; Gese et al. 1988).
Direct open-field observations were made of both marked (radio collared, implanted or ear-tagged) and unmarked but identifiable (pelage coloration, pelage pattern and physical characteristics) coyotes during daylight hours, usually between 0700 and 2000 hours, with a #10–45 spotting scope. Nocturnal observations were collected using a night-vision scope during clear, moonlight nights.
To maintain reliable and consistent interpretation of behaviours (Lehner 1979; Martin & Bateson 1993), E.M.G. trained each observer for a minimum of 5–7 days. To avoid repeated sampling of the same pack or individual (Morrison et al. 1992), coyote packs were chosen using a random numbers table prior to going into the field. Stratification of individuals within the pack then allowed for selection of the animal to be observed (Gese et al., 1996a,b). The animal chosen was observed continually, recording all social and predatory behaviour (Gese et al. 1996a,b) as well as interactions with other coyotes.
Coyotes were observed from mid-October to July; tall grass in the study area precluded observation in August and September.
Visual locations of predation events (Gese et al. 1996b), bed sites, scent marks, carcass sites, territorial defense and other activities were recorded to the nearest 10 m on 1:24 000 U.S. Geological Survey topographic maps using the Universal Transverse Mercator (UTM) grid system.
Territorial boundaries were determined by locations of pathways used during scent marking and territorial defense. Visual locations were used because radiotelemetry locations in the valley were highly inaccurate (triangulation errors averaged 1.4 km from reference transmitters; range of error was 0.25–6 km).
To examine the influence of social factors on dispersal, coyotes within a resident pack were classified into different social classes based upon the sex-specific dominance hierarchies observed within each resident pack. Coyotes were ranked within the linear dominance hierarchy of each pack based upon their display of expressive behaviours for dominance or submission with each pack member (Schenkel 1947, 1967; Mech 1970; Zimen 1975) and the direction of submission and agonistic behaviour (Lockwood 1979). Annual dominance matrices were constructed for each sex class in each pack to identify the social order of the pack members.
Social classes included alphas (the dominant, breeding adult male and female), betas (adults and yearlings subordinate to alphas, but dominant over pups) and pups (young-of-year which were subordinate to both alphas and betas).
We recorded the number of times each individual was located with another pack member at the beginning of each observation bout as a measure of the level of disassociation from the pack. We examined the influence of food resources on dispersal by recording the number of carcasses (elk, mule deer and bison) each coyote was observed to visit and the length of time spent feeding on the carcass. We calculated a carcass access index by multiplying the mean length of time spent feeding on a carcass (in hours) by the number of carcasses visited by that individual.
Because coyotes may hunt small mammals in response to reduced access to carcasses (Gese ET al., 1996a), we also measured each coyote’s capture rate of small mammals. We recorded the number of prey taken by each coyote per hour it was active. Because many dispersers left in early winter, we used the small mammal capture rates recorded in October–December for coyotes observed during the winters of 1991–1992 and 1992–1993. For the winter of 1990–1991, we used the capture rate during April–May, because observations were not initiated until after January.
Litter sizes presented in the text were the numbers of pups emerging from the natal den in early May. A coyote was defined as a ‘known disperser’ if the animal left its territory and was subsequently located or killed outside the territory. ‘Probable dispersers’ were animals that left the territory and were not located again.
Coyotes with transmitters that were found dead were ‘known deaths’. Pups classed as ‘probable deaths’ were animals that disappeared from the pack at an age too young for them to be independent and survive on their own. Coyotes classed as ‘unknown fate’ were pups that did not have transmitters and disappeared from the pack prior to initiation of behavioural observations. These animals either died or dispersed (i.e. they were old enough to disperse and disappeared during the time of parvovirus susceptability, but they did not have transmitters). Daily total snow depth was recorded by the National Park Service at a permanent weather station located at the Lamar Valley Ranger Station in the study area. The amount of ungulate carcass biomass available to the coyotes was measured by recording the size and number of carcasses fed upon by the coyotes (Gese et al. 1996a), then converting each carcass into carcass biomass following the procedure by Houston (1978). This conversion accounted for the weight loss of the animal at the time of death, then subtracted the weight of parts of the carcass not eaten by coyotes (i.e. carcass weight minus the rumen and skeleton). We compared social and nutritional characteristics of philopatric individuals to dispersing coyotes using a Student’s t-test (Steel & Torrie 1980).
We compared the proportion of observations that each coyote was located with another pack member, the proportion of interactions in which each coyote was dominant over another pack member, and each coyote’s small mammal capture rate. We used only pups and betas in these analyses because alpha coyotes were not observed to disperse. All values presented for dispersing coyotes are the social and nutritional characteristics of those individuals prior to dispersal. Many coyotes dispersed or died in early fall, prior to initiation of behavioural observations in October. Therefore, our observations, analyses and conclusions are based solely upon those coyotes still present in the pack in mid-October.
From January 1991 to June 1993, we observed 49 resident coyotes from five packs for 2456 h. Of the 49 coyotes observed, 27 were males, 20 were females and two unmarked coyotes were of unknown sex. We collared or implanted 28 coyotes with radiotransmitters, and 21 were recognizable from physical characteristics. We observed 2366 interactions between pack members, allowing us to construct dominance matrices in each resident pack. We recorded 9349 visual locations of the coyotes for determination of territory size and boundaries.
The first winter of observation was mild, with little carcass biomass available to the coyotes in the Lamar River Valley (Fig. 3a). Maximum snow depth was 30 cm and the amount of known carcass biomass was less than 170 kg/week. Coyotes were dependent upon small mammals, mostly voles (Microtus spp.), as their major food item during this winter. The second winter of observation was characterized by deeper snow cover and higher carcass biomass (Fig. 3b).
This winter had an early snowfall followed by a thaw, which re-froze into an ice layer on the ground and led to an early initiation of winter die-off of ungulates. Maximum snow depth was 46 cm and known carcass biomass exceeded 200 kg/week for 10 weeks. The third winter of observation was similar to the second winter, with deep snow cover and high amounts of carcass biomass. Maximum snow depth was 63 cm and there were 6 weeks in which known carcass biomass was greater than 200 kg/week.
The coyotes in Lamar Valley were organized into relatively large packs with distinct territories. Territorial boundaries were scent-marked and actively defended (E. M. Gese, unpublished data). A pack consisted of an alpha pair and associated pack members, usually related offspring.
Construction of dominance matrices for each pack demonstrated the presence of a social order or hierarchy among females and males. The social organization and presence of a dominance hierarchy in each pack was similar to that described in a wolf pack (Schenkel 1947, 1967; Mech 1970; Zimen 1975, 1981). The large packs we observed were probably a consequence of the combination of abundant prey biomass and the lack of exploitation in the study area.
The alpha female had four pups in 1990, with only one pup remaining through winter. She had six pups in 1991. Only three pups remained by October when observations began.
Of the three remaining pups, the low-ranking male pup dispersed into an adjacent territory in January 1992 and was found dead in March. The lone female pup probably dispersed in June 1992. The alpha male of the previous winter was killed by a car in December 1991, and the highest-ranking beta male assumed the alpha position.
The pack produced two litters of pups in 1992. The alpha female had seven pups, and the high-ranking beta female had five pups in a separate den. Only three pups remained by winter; none of these dispersed.
The alpha male of the previous winter was displaced by the highest-ranking beta male. The former alpha male was relegated to the position of the second-ranking beta male. This change in the alpha position occurred between the time of the last observations in July and our first observations in October.
No pups were observed in 1990. The alpha female had six pups in 1991, but only four remained by June. The low-ranking pup dispersed more than 60 km in January, and two other pups remained through the winter. The alpha female had five pups in 1992, and only one pup remained through the winter. The alpha male of the two previous winters was killed by a car in December, and the highest-ranking beta male assumed the alpha position.
He was paired with the alpha female within 4 days of the death of her mate. A beta male that had dispersed as a pup the previous winter returned in May as the low-ranking beta male.
Fossil Forest pack
The alpha female produced four pups in 1990, but only two pups remained by January. The alpha female had six pups in 1991.
At the initiation of observations in October, only three pups remained. The high-ranking female pup was killed by a car in December. The two remaining pups stayed through the winter. The alpha female had a minimum of five pups in 1992.
We also captured and radio-collared two other pups in the autumn, but it is not known whether they were from the original litter or dispersers from other packs. Both pups left the area immediately after capture.
A third pup captured in the fall was the low-ranking male in the pack and dispersed in December. A non-marked female pup, which was the low-ranking female, probably dispersed in February. The low-ranking beta male apparently dispersed in January. Another beta male (M379) dispersed into an adjacent pack (Norris) in an attempt to become alpha male in that pack. He was observed paired and scent-marking with the alpha female of Norris, but was displaced by another male and returned to his natal pack, where he remained the second-ranking beta male below his dominant brother.
No pups were whelped in 1990. We observed only three pups emerge from the natal den in May 1991, with only one pup remaining by winter. The low-ranking beta female, who was dominated by the female pup, dispersed in January.
The low-ranking beta male, who was integrated into the pack in November, probably dispersed in May 1992. The alpha female had eight pups in 1992 with two pups dispersing in the autumn; the female pup dispersed more than 40 km away. Only one pup, which was the low-ranking animal in the female hierarchy, remained during winter, then apparently dispersed in April 1993. The alpha male of the previous two winters was found dead in January 1993 and the social structure of the pack deteriorated.
The alpha female left the territory for a month and was observed mating with three different males in the valley. During her absence, the Soda Butte pack usurped half of the Norris territory (Fig. 2).
When she returned with a male from the Fossil Forest pack (M379), they were continually chased from the territory by the Soda Butte alpha pair and finally settled in a smaller remnant of the original Norris territory. Another male displaced M379 from the alpha position, then that male was also displaced from the alpha position (Fig. 2). No pup production was observed in 1993. The high-ranking beta male of the pack, which did not assume the alpha position, dispersed and became a transient in the valley.
Soda Butte pack
The alpha female had five pups in 1990, with four pups remaining through the winter. The alpha female had four pups in May 1991, but only one pup remained by June (Fig. 2).
The lone pup then dispersed 17 km in January 1992. The low-ranking beta female made many pre-dispersal forays into neighbouring territories during February and March, but did not disperse until the next winter. The alpha female had nine pups in 1992. Two pups dispersed by November.
Of the nine pups born, only one pup remained through the winter. The low-ranking beta female dispersed in March 1993 and apparently became the alpha female in the adjacent pack north of Soda Butte.
Influence of Social Rank and Dominance
The social rank and level of dominance in the pack hierarchy influenced whether a coyote dispersed or stayed. In the first winter of study, observations of the five resident packs began in January 1991 after many pups had died or dispersed. Thirteen pups were known to have been born in three packs in 1990, of which six had either died or dispersed prior to the beginning of observations in January 1991. None of the seven remaining pups, nor any of the alphas and betas, dispersed throughout the winter. These philopatric individuals were dominant in an average of 30% of their interactions with other pack members (Table V). Comparisons between philopatric coyotes and dispersing individuals were not possible for this winter, because no dispersal occurred from January to July 1991.
During the second winter of study, all six coyotes that dispersed or probably dispersed during winter were the low-ranking individuals in their respective packs; i.e. they were subordinate to all the pack members above them in the dominance hierarchy. Four of the dispersers were pups and two were older beta coyotes; no alphas or high-ranking betas dispersed.
Dispersers were dominant in an average of 8% of their interactions with other pack members, and philopatric animals were dominant in an average of 37% of their interactions (t=3.26 DF=21, P=0.0019). When we controlled for age (i.e. used only pups), we found that philopatric pups were the dominant individual in an average of 30% of their interactions with other pack members, but dispersing pups were dominant in an average of 6% of their interactions (t="2.14, df=8, P=0.065). In packs with two surviving pups of the same sex (i.e. Bison and Druid packs), the dominant pup stayed and the subordinate pup dispersed.
During the third winter of study, eight coyotes dispersed or apparently dispersed from their respective packs. Three dispersers were pups and five others were betas. All three pups were the lowest-ranking individuals in their packs, and three of the five betas that dispersed were the lowest or next-to-lowest ranking betas. None of the alphas and only one high-ranking beta dispersed. Philopatric animals were dominant in an average of 35% of their inter-actions with other pack members, and dispersers were dominant in an average of 11% of their interactions (t=2.54, DF=28, P=0.008).
Percentage of Observations with Another Pack Member The percentage of observations with another pack member differed between philopatric coyotes and dispersers. Behavioural observations collected on 10 philopatric pups and betas in the first winter showed that they were observed with another pack member an average of 38% of the time. During the second winter, dispersing coyotes (N=6) and philopatric individuals (N=17, pups and betas only) were observed an average of 15% and 37% of the time, respectively, with other pack members (t=4.18, DF=21, P=0.0002). When we controlled for age (i.e. used only pups), we found that philopatric pups (N=6) were located with other pack members a mean of 36% of the time, and pups that later dispersed (N=4) were located with other pack members a mean of 18% of the time (t="2.33, df=8, P=0.048).
During the third winter, dispersers (N=8) were observed with other pack members an average of 19% of the time, and philopatric animals (N=22) were with other pack members during an average of 27% of the observations (t=1.38, DF=28, P=0.08).
Access to Ungulate Carcasses
We were unable to compare the carcass access index of dispersers to philopatric animals directly, owing to differences in the number of carcasses available to each pack. A consistent trend existed, however. We had no observations on dispersers in the first winter to compare with philopatric coyotes. During the second winter, however, all the dispersing coyotes typically had low access to carcasses within their respective packs. Alphas and higher-ranking betas typically had the highest carcass access index in their pack. Again, in the third winter the dispersing coyotes were typically low-ranking coyotes that were subordinate to the other coyotes in the pack, and typically had little or no access to carcasses within their respective pack.
Capture Rate of Small Mammals
A coyote’s ability to capture small mammals appeared to be important in determining whether an animal dispersed or stayed. During the first winter, the philopatric coyotes captured an aver-age of 1.4 small mammals/h. During the second winter, capture rates by dispersers (N=6) and philopatric pups and betas (N=17) averaged 2.4 and 2.3 prey/h, respectively (t="0.227, DF=21, P=0.411). In the third winter, dispersing coyotes captured small mammals at about half the rate (X=1.2 prey/h) attained by philopatric pups and betas (X=2.2 prey/h; t=3.42, DF=28, P=0.001).
Proximate mechanisms influencing mammalian dispersal patterns are varied and involve both intrinsic and extrinsic factors. Captive studies of both wolves and coyotes indicate that subordinate animals are harassed by animals of higher social rank and denied equal access to food resources. This behaviour pattern either prompts low-ranking individuals to dissociate from the pack or requires the investigator to remove them to prevent injury (Zimen 1976, 1981; Knowlton & Stoddart 1983). The researchers concluded that, if given the opportunity, these captive subordinates would have dispersed (Zimen 1976, 1981; Knowlton & Stoddart 1983). A common characteristic among all dispersing coyotes on our study area was that they were low-ranking pups or beta animals in their packs. They spent little time with other pack members, were almost always subordinate when interacting with other coyotes and had little access to ungulate carcasses. These findings all support the hypothesis that young animals may be less successful at competing for resources against older individuals in the pack, and thus are predisposed to dispersal (Macdonald 1980; Fritts & Mech 1981).
Low-ranking coyotes may be predisposed to disperse from their natal pack (Bekoff 1977b), because an animal’s social rank within the dominance hierarchy may be established early in life (Fox 1969; Mech 1970; Bekoff 1974, 1978; Knight 1978). Our findings that dispersing coyotes (prior to their dispersal move) were submissive, spent little time with other pack members and had little access to ungulate carcasses, suggest that these interrelated factors were a direct result of their social rank. The factors that influence an animal’s social rank early in life are speculative. Body size may influence the social ranking of a pup early in life (Fuller & DuBuis 1962; Knight 1978). White & Harris (1994) found a higher incidence of wounding among smaller male red foxes that dispersed.
Food resources appeared to influence the timing of coyote dispersal and the number of individuals that remained in the pack over winter.
During the first winter with low carcass biomass in the valley, dispersal had already occurred by January and pack size was 4.6 coyotes. During the second winter, with more ungulate carcasses in the valley, some animals did not disperse until mid- or late-winter, and pack size increased to 5.8 coyotes as more coyotes remained in their pack. During the final winter, with similar high ungulate carcass biomass, many coyotes did not disperse until late winter and pack size increased to 6.6 coyotes. Food resources influenced how long and how many coyotes could remain in the pack. In Maine, low densities of deer and alternative prey were believed to prevent delayed dispersal and pack formation in coyotes (Harrison 1992). In Canada, Messier (1985) found a higher incidence of extra-territorial movements by wolves in an area with low prey abundance compared with an area with high prey density. The finding that small mammal capture rates by dispersers and philopatric animals were not different during the second winter, but were different the third winter, may indicate the influence of both pack size and food resources on dispersal.
When pack size was 5.8 coyotes, the level of competition around carcasses may have been low enough that the ability to capture small mammals was unimportant. When pack size was higher during the third winter, however, dispersers were less successful than philopatric animals at capturing small mammals.
Perhaps with increased pack size, competition at carcasses intensified. Animals that compensated for this reduction in carcass access by hunting small mammals could fulfill their energy requirements and remain in the pack. In contrast, individuals with low access to carcasses and that were also unable to capture small mammals at a high rate may have elected to disperse from their territory and seek resources elsewhere. Thus, when competition for carcasses increases to certain levels with increasing pack size, an individual’s skill and ability to capture small mammals may become very important in determining whether it remains or disperses. Nine of the 14 dispersers left immediately before (December) or during the breeding season (January and February). All interactions between pack members become more intense and aggression increases during the breeding season (Schenkel 1947; Rabb et al. 1967), which may force the subordinate animals to leave the pack (Zimen 1976).
The breeding season for coyotes in Lamar Valley also coincides with the time of year when deep snow accumulates, making capture of small mammals difficult (Gese ET al., 1996b), and forcing greater reliance upon ungulate carcasses.
The young, subordinate individuals that we observed had little chance of breeding within their natal territory in the near future (Figs 1, 2) and had little access to carcasses. The combination of increased aggression during the breeding season and competition around carcasses could culminate in their dispersal. When more than one individual left the pack (Fossil Forest and Norris packs in 1992–1993), the lowest-ranking coyote left first. Although we observed dominant–submissive interactions between all the coyotes when they interacted with one another, and some subordinate individuals were harassed by older coyotes causing these subordinates to dissociate from the pack, we never observed overt aggression in which pack members forcibly drove the subordinate individual out of the territory. Rather, we believe that a culmination of low social rank, reduced access to carcasses and little opportunity for breeding causes an animal to leave its territory voluntarily. The fact that two dispersers were later re-integrated into their natal pack suggests that they had left voluntarily the first time. Possibly, they were able to remain in the pack the second time because the social and/or nutritional pressures within the natal territory had lessened.
One of the primary objectives of dispersal is to find a mate and reproduce (Howard 1960; Lidicker 1975). Two of the dispersing coyotes were successful in integrating into another pack. One pup (M412) joined a pack outside Lamar Valley, but we do not know whether it reproduced in that pack. A beta female (F600) dispersed into an adjacent pack and successfully acquired the alpha female position. Two dispersers (M379, M240) returned to their natal pack after unsuccessfully attempting to join another pack. The success or failure of the other long-range dispersers was unknown because they dispersed to areas outside the park. Wolf pups in Minnesota had low success in pairing with another wolf after dispersal, but adults that dispersed had relatively high success (Gese & Mech 1991). The size, experience and sexual maturity of dispersing adults may allow them to successfully compete against other animals, but pups may be easily displaced from a new pack or area (Gese & Mech 1991).
In contrast to the dispersing individuals, some high-ranking philopatric coyotes were able to eventually advance to the breeding position within their pack. When an alpha member of the pack was killed (N=2) or displaced (N=1), the highest-ranking beta assumed the alpha position or was responsible for displacing the alpha animal. In another case when the alpha male died (Norris pack), the high-ranking male was apparently not accepted by the alpha female, and she left the territory in search of a new mate. During her 1-month absence, the adjacent pack usurped half of her territory.
Inbreeding avoidance, mate competition and resource competition have been proposed as ultimate reasons for dispersal (Greenwood 1980; Moore & Ali 1984; Waser 1985). Our findings indicate that all three hypotheses may be involved in the dispersal patterns of coyotes in our study area. Resource competition (i.e. access to ungulate carcasses) was related to social rank and influenced the likelihood of dispersing. The ability to capture small mammals also influenced dispersal when pack sizes were highest. Increased pack size may have caused increased competition at the primary winter food source (carcasses). Mate competition could also be involved, because many of the coyotes dispersed before or during the breeding season.
Increased aggression during the breeding season (Zimen 1976), suppression of breeding behaviour (Rabb et al. 1967) and lack of breeding opportunities could all influence dispersal patterns. A balance between out breeding and inbreeding may also exist within the coyote social system. The observation that many pups and betas dispersed, and that the Norris alpha female did not pair with the older beta male in the pack (possibly her father), suggests some level of inbreeding avoidance. The three observations of a beta male within the pack becoming the alpha male suggests that some inbreeding could occur, if those beta males were offspring or closely related to the alpha female.
Unfortunately, the genetic relatedness of those beta males and alpha females was unknown. In conclusion, our findings offer support for the social subordination hypothesis (Christian 1970).
Coyotes that dispersed were low-ranking individuals that were subordinate to other animals in the dominance hierarchy, spent little time with other pack members, had little access to carcasses and were less skilled at hunting small mammals during the year when pack size was greatest. We emphasize, however, that we never observed dominant coyotes chasing a pack member and forcing the subordinate coyote to disperse.
Instead, we believe that the culmination of different social and nutritional pressures reaches a certain level, and the individual voluntarily leaves the territory to seek resources (food and/or breeding opportunities)elsewhere. Whether affiliative behaviour (i.e. the social cohesion hypothesis) played a role in the early stages of life for these dispersing coyotes is unknown. We were unable to collect information on individual pups at the den, and our observations began in the autumn when many pups had already dispersed. The social subordination and social cohesion hypotheses are not necessarily mutually exclusive, and both may play a role at different life stages in influencing the dispersal of coyotes in Yellowstone National Park.
We thank Patricia Terletzky, Ed Schauster, AldenWhittaker, Alexa Calio, Melissa Pangraze, LaraSox, Levon Yengoyan, Danny Rozen, Jeanne Johnson, John Roach and Valeria Vergara for assistance with behavioural observations; Scott Grothe, Kezha Hatier and numerous technicians for field assistance; John Varley of the National Park Service for logistical support; John Cary and John Coleman for computer programming; and Peter Arcese, Jeffery Baylis, Robert Garrott, James Malcolm, Warren Porter and Peter Waserfor review of the manuscript. Funding and support was provided by the Department of Wildlife Ecology and the College of Agricultural and Life Sciences at the University of Wisconsin-Madison, Max McGraw Wildlife Foundation, U.S. Fish and Wildlife Service, the National Park Service (cooperative agreement 1268-1-9001 to R. L. Crabtree), National Geographic Society, the Biology Department at Montana State University, Earthwatch and the Hornocker Wildlife Research
*Department of Wildlife Ecology, University of Wisconsin †Biology Department, Montana State University