RESPONSES OF COYOTES AND LYNX TO THE SNOWSHOE HARE CYCLE by MARK O'DONOGHUE B.Sc, The University of Maine, 1981 M.Sc, The University of British Columbia, 1991 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF ZOOLOGY We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA January 1997 © Mark O'Donoghue, 1997 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial" gain shall not be allowed without my written permission. Department of Z o o lo 0 into the analysis. We used logistic regression (Trexler & Travis 1993) to estimate the probability of encountering any tracks in a transect segment, and multiplied the least squares means by these probabilities to give our yearly means. Live-trapping and Radio-tracking of Predators We live-trapped coyotes and lynx, mostly during the fall and winter months, using padded foot-hold traps (No. 3 Soft-catch ® traps, Woodstream Corp., Lititz, Pennsylvania), cable snares (Fremont leg snares ®, Fremont Humane Traps, Candle Lake, Saskatchewan), and box traps (techniques similar to Mowat, Slough & Rivard 1994). We checked traps at least once per day, and locked traps open when the temperature was less than -20 C. During the first 4 years of the study, we immobilized captured animals using a 7:1 mixture of ketamine hydrochloride and xylazine hydrochloride, but we switched to using a mixture of tiletamine hydrochloride and zolazepam hydrochloride (Telazol ®, A. H. Robbins Co., Richmond, Virginia) since then (Poole, Mowat & Slough 1993). We subjectively classified all captured animals as adults, yearlings, or kits, based on body size and tooth wear, and after 1990, we pulled lower incisors from most animals, and determined their ages from cementum annuli (Crowe 1972). 20 We fitted animals with radio-collars weighing 250-350 g (Telonics, Inc., Mesa, Arizona); most of these had internal mercury switches sensitive to movement, which allowed us to monitor the activity of collared animals. Using ground-based telemetry, we attempted to get at least one radio-location per week of each radio-collared coyote and lynx. We periodically monitored the activity of most animals much more intensively. We calculated locations and error ellipses using the program LOCATE II (Nams 1990), and plotted home ranges using 95% minimum convex polygons (Mohr 1947) with the program CALHOME (Kie, Baldwin & Evans 1994). Estimates of Predator Populations We estimated the numbers of coyotes and lynx in the study area each December, from 1987 to 1995, using data from several sources. We kept detailed records of the home ranges and movements of radio-collared animals from telemetry locations and snow-tracking. Each winter, we also had extensive snow-tracking programs, in which we followed the tracks of coyotes and lynx throughout the study area (Chapters 3 and 4). During 8 winters, from 1987 to 1995, we tracked coyotes an average of 237 km, and lynx an average of 279 km per winter. These data, supplemented by track data from our transect counts, allowed us to estimate the number of uncollared animals in our study area. We also kept records of the locations and group sizes of all howling by coyotes throughout the year, which gave us a further check on the size of family groups of coyotes. An example of our method is illustrated in Fig. 2.1 for lynx during the winter of 1994-95. 21 Results Prey Populations Populations of snowshoe hares peaked between 1988 and 1990 at approximately 2/ha (Fig. 2.2; Boutin et al. 1995). There were locally abundant pockets of hares through the fall of 1991 (Boutin et al. 1995), and populations declined to very low numbers by the end of winter 1992-93. Densities of hares started to increase again in 1994. Depending on whether fall or late winter estimates of hare numbers are used, the cyclic amplitude was 26-44-fold (Boutin et al. 1995). Population trends of other potential prey species varied. Numbers of red squirrels stayed relatively stable, with a slight increase during the last two winters of the study (Fig. 2.3). Populations of small mammals fluctuated with a 10-50-fold amplitude, with the highest numbers occurring from 1991 to 1993 (Fig. 2.3). Densities of ground squirrels increased to a peak in 1991 (a year after the cyclic peak of hares), declined to low numbers by 1993, and increased again in 1994 (Fig. 2.3). During the winter, ground squirrels hibernate, and they are therefore not available as prey. There were few other alternative prey species. Spruce grouse (Dendragapus canadensis), ruffed grouse (Bonasa umbellus), and ptarmigan (mostly Lagopus lagopus) were present (Boutin et al. 1995), but were seldom utilized by predators (2% of all kills in winter from 1987 to 1995; Chapter 3). Snowshoe hares were by far the most abundant potential food source during winter for predators from 1987 to 1992, comprising 63-81% of the total biomass of the main prey species (Fig. 2.4). Red squirrels were the second largest food source during these years. During the last three winters of the study (1992-93 to 1994-95), however, once hare numbers had crashed, red squirrels comprised 58-72% of the total biomass (Fig. 2.4). Small mammals represented less than 10% of the total biomass of prey in all winters except 1992-93 (14%; Fig. 22 Fig. 2.2. Estimated densities of snowshoe hares (means of autumn and late winter estimates from live-trapping on 1-3 60-ha grids), coyotes, and lynx from 1986 to 1995 in the southwest Yukon. Counts of predator tracks along a 25-km transect are presented as least squares means; raw track counts were adjusted for covariates track location, date, days since last snowfall, and temperature. Mean 95% confidence intervals of track counts were ± 13%. 23 Hares per 100 ha 10 Coyotes per 100 sq km ( • ) i40 -20 0 0 Tracks per Night per 100 km ( • ) Winter 24 Fig. 2.3. Estimated densities of red squirrels (means of late summer and spring estimates from live-trapping on 2-3 8-ha grids), small mammals (means of late summer and spring estimates from live-trapping on two 2.8-ha grids for Clethrionomys rutilus, and on two 1.5-ha grids for Microtus oeconomus and M. pennsylvanicus), and Arctic ground squirrels (means of spring and late summer estimates from live-trapping on 1-3 8-ha grids) from 1986 to 1995 in the southwest Yukon. Arctic ground squirrels hibernate during the winter, and are therefore not available to predators. 25 400 Red Squirrels per 100 ha 200 0 Ground Squirrels per 100 ha 86- 88- 90- 92- 94-87 89 91 93 95 Winter 26 Fig. 2.4. Percent prey biomass of hares, red squirrels and small mammals available to predators in winter from 1987 to 1994. These species made up 97% (666/689) of all kills by coyotes and lynx during the winters of these years. 27 Percent Biomass 100 80 -60 -40 -20 0 I IP i i l l Winter 28 2.4). From the winter of 1989-90 to winter 1992-93, there was an approximately 4-fold decrease in the total prey biomass available. Snowshoe hares are the only one of these prey species with no arboreal or subnivean refuge from predation, so the decrease in effective availability of prey during this time was likely considerably greater. Predator Populations Coyotes increased in numbers from 1987 to 1990, and then decreased to low numbers again by 1992-93 (Fig. 2.2). The density of coyotes was highest, at about 9/100 km2, in 1990-91, the first winter of declining hare numbers (Fig. 2.5). The low density of coyotes was 1.4/100 km2, and the cyclic amplitude was approximately 6-fold. Over the whole cycle, coyote numbers were strongly correlated with hare numbers the previous year (Fig. 2.5; Boutin et al. 1995). The trend in track counts, based on 11,140 km of transect sampled, closely mirrored the trend suggested by our population estimates (Spearman rank correlation coefficient = 0.88, P = 0.01). The trend in lynx numbers was similar to that of coyotes (Fig. 2.2). Populations of lynx increased from 1987 to 1990 to a peak density of about 17/100 km2, and then declined by 1992-93 to a low of about 2.3/100 km2. The cyclic amplitude was 7.5-fold. Lynx numbers were also highly correlated with densities of hares during the previous year (Fig. 2.6; Boutin et al. 1995). As with coyotes, the trend in track counts closely followed our population estimates of lynx (Spearman rank correlation coefficient = 0.95, P < 0.005). Survival and Dispersal of Predators We live-trapped and radio-collared 21 different coyotes and 56 different lynx between 1986 and 1995 (Tables 2.1 and 2.2). In general, we were able to capture a greater percentage 29 Fig. 2.5. Numerical response of coyotes to changes in the density of snowshoe hares from 1987 to 1995. Coyote density was highly correlated with density of hares during the previous winter. Numbers next to the data points indicate years. 30 Coyotes per 100 sq km 0 1 1 1 1 1 1 0 50 100 150 200 250 Hares per 100 ha Coyotes per 100 sq km 0 50 100 150 200 250 Hares per 100 ha in Previous Winter Fig. 2.6. Numerical response of lynx to changes in the density of snowshoe hares from 1987 to 1995. Lynx density was highly correlated with density of hares during the previous winter. Numbers next to the data points indicate years. 32 20 Lynx per 100 sq km 0 1 1 1 1 1 1 0 50 100 150 200 250 Hares per 100 ha Lynx per 100 sq km 0 50 100 150 200 250 Hares per 100 ha in Previous Winter Table 2.1. Numbers of radio-collared coyotes, and estimates of untagged animals, during winter from 1986 to 1995. Numbers of untagged animals were estimated from intensive snow-tracking and recording howling year-round. Winter Estimated Population Number of Collared Coyotes Number of Adult Males Number of Adult Female Number of Immature Males Number of Immature Females Estimated Number of Uncollared Coyotes 1986-87 0 ? 1987-88 8 3 1 1 1 0 5 1988-89 12 2 0 2 0 0 10 1989-90 20 0 0 0 0 0 20 1990-91 30 6 4 1 0 1 24 1991-92 17 8 2 2 1 3 9 1992-93 9 4 1 2 1 0 5 1993-94 5 3 3 0 0 0 2 1994-95 6 0 0 0 0 0 6 34 Table 2.2. Numbers of radio-collared lynx, and estimates of untagged animals, during winter from 1986 to 1995. Numbers of untagged animals were estimated from intensive snow-tracking. Winter Estimated Population Number of Collared Lynx Number of Adult Males Number of Adult Female s Number of Immature Males Number of Immature Females Estimated Number of Uncollared Lynx 1986-87 4 2 2 0 0 ? 1987-88 10 7 2 2 0 3 3 1988-89 16 6 2 1 2 1 10 1989-90 50 11 4 3 2 2 39 1990-91 60 22 13 5 1 3 38 1991-92 28 16* 8 5 2 0 12 1992-93 15 11 8 3 0 0 4 1993-94 9 6 4 2 0 0 3 1994-95 8 6 3 3 0 0 2 The sex of one adult was not noted in 1991-92. 35 of the coyotes and lynx present in the study area during the second half of the study. The mean weights of trapped animals were: adult male coyotes = 12.2 kg (n = 11), adult female coyotes = 10.7 kg (n = 6), adult male lynx = 10.8 kg (n = 26), and adult female lynx 9.6 kg (n = 18). Survival of radio-collared lynx varied widely during the study (Fig. 2.7). Survival was high (>70% per winter) during the increase phase of the hare cycle, slightly lower (45-63% per winter) during the early decline phase, and virtually all animals died in 1992-93, the first winter of very low hare numbers. Those animals that did survive into the next winter (1993-94) survived quite well. Causes of mortality of radio-collared lynx in the study area were mostly human-caused (fur-trapping) in all years except during the decline phase of the cycle when predation/scavenging (n = 5) and starvation (n = 2) were of greater or equal importance (Fig. 2.7). We confirmed three cases of predation on lynx (by a wolf, a wolverine, and another lynx), and suspected predation by wolverines in two other cases. We recorded only 5 mortalities of coyotes, and 4. were human-caused. Yearly dispersal rates of lynx varied from 0-32%, with the highest rates occurring at the peak and in the early decline phase of the hare cycle (Fig. 2.7; dispersal here is defined as leaving the study area). In contrast, 62% of all coyotes radio-collared dispersed or were lost (and presumed to have dispersed) from our study area, and there was no clear relationship between the timing of dispersal and the abundance of prey. Dispersal rates of both adults (coyotes = 56%, lynx = 44%) and juveniles (coyotes = 80%, lynx = 58%) were high for both species. We determined the minimum dispersal distances for 2 coyotes and 10 lynx, most of which were captured by fur-trappers (Table 2.3). Distances between capture and mortality sites varied from 23-40 km for coyotes, and 24-830 km for lynx. Six lynx dispersed further than 200 km from their capture sites-one of these was trapped in the Northwest Territories, and the other 5 were trapped in Alaska. 36 Fig. 2.7. Survival and causes of mortality of radio-tagged lynx during winter, from 1986 to 1995. Sample sizes are given above each bar. Of the 5 mortalities classified as due to "Predation" in 1992-93, 3 were confirmed, and two suspected, based on signs at the sites of mortality. Sample sizes were too small for a comparable figure for coyotes, but 62% of all radio-tagged coyotes dispersed from the study area, and only 5 mortalities, 4 of them human-caused, were recorded. 37 4 7 6 1 1 2 2 16 11 6 6 = n Percent of Collared Lynx Winter 38 Table 2.3. Known dispersal distances, from capture sites to sites of mortality, of radio-collared coyotes and lynx from 1986 to 1995. Species Age When Last in Study Area-Sex Date of Last Location in Study Area Days from Last Location to Mortality Total Dispersal Distance (km) Fate Coyote A - F 18 Nov 91 1117 40 Trapped Coyote A - M 10 Mar 94 0 23 Shot Lynx Y - M 26 Mar 90 660 140 Trapped Lynx Y - M 01 Jul 90 858 355 Shot Lynx A - F 15 Jul 90 167 320 Trapped Lynx A - M 16 Feb 91 296 830 Trapped Lynx A - M 02 Apr 91 680 405 Trapped Lynx A - M 30 Sep 91 483 300 Trapped Lynx Y - F 18 Oct 91 58 130 Trapped Lynx A - M 02 Nov 92 202 245 Trapped Lynx A - F 10 Feb 93 0 24 Unk. Cause Lynx A - M 04 Mar 93 11 40 Trapped A = adult, Y = yearling. 39 Recruitment of Predators While we did not directly measure reproduction of coyotes or lynx, we recorded the mean group sizes for all observations of tracks along our track transect (Figs. 2.8 and 2.9). We defined "family groups" of coyotes as all groups with more than 2 animals, and of lynx as all groups with kittens present (determined by track size). Groups of coyotes were noted only during the 4 winters from 1988-89 to 1991-92, and groups of more than 3 only during the first two winters of the hare decline (Fig. 2.8). Family groups of lynx were seen in all winters except the last two years of the study, but there were very few sightings after 1990-91 (Fig. 2.9). Groups of larger than 4 animals were seen only in the three winters from 1988-89 to 1990-91. Discussion Coyotes and lynx showed similar responses to changes in prey abundance over the cycle in numbers of snowshoe hares. The 6-fold change in coyote numbers was slightly less than the 7.5-fold fluctuation of the lynx population, but measures of amplitude are strongly affected by estimates at the low of the cycle, so small changes in numbers at the low can translate into large differences in amplitude. Populations of both carnivores peaked in 1990-91, the year following the highest hare densities, and both declined at a similar rate; coyote abundance was 30% of peak density two years afterwards, compared to 25% for lynx. Populations of the main species of alternative prey did not fluctuate in synchrony with snowshoe hares (Boutin et al. 1995). Therefore, our null hypothesis, that the densities of coyotes and lynx in the boreal forest are largely determined by the abundance of snowshoe hares, is supported by these data. 40 Fig. 2.8. Mean group sizes of family groups (± S. D.) of coyotes (defined as groups with more than 2 individuals), and percent of total number of track observations comprised of family groups, along the 25-km track transect, from 1987 to 1995. 41 Mean Size of Family Groups ( • ) i 15 10 0 Percent Family Groups ( - ) Winter 42 Fig. 2.9. Mean group sizes of family groups (± S. D.) of lynx (defined as groups with kittens present), and percent of total number of track observations comprised of family groups, along the 25-km track transect, from 1987 to 1995. 43 Winter 44 Densities of Predators The peak density of coyotes in our study area (9/100 km2) was much lower than that observed in Alberta (44/100 km2), and more comparable to their low density (8/100 km2; Keith et al. 1977; Todd, Keith & Fischer 1981). Populations of coyotes in Alberta also declined at a much slower rate than those in the Yukon after the cyclic peak (22% decline after 2 years relative to 70% decline in the Yukon), despite similar trends in hare abundance. Coyotes made heavy use of livestock carcasses during periods of low prey abundance in Alberta (Todd, Keith & Fischer 1981), which were not available to coyotes in our study. Further south in their range, densities of coyotes can range as high as 50-100/100 km2 (Camenzind 1978; Andelt 1985), so the densities we observed in the boreal forest are quite low. Densities of lynx at the cyclic peak in our study area (17/100 km2) were higher than those observed in Alberta (10/100 km2; Brand, Keith & Fischer 1972; Keith et al. 1977), but considerably lower than in south-central Yukon (50/100 km2; Slough & Mowat 1996) or in the Northwest Territories (30/100 km2; Poole 1994). The peak abundances of hares were greater in all 3 other study sites (Alberta: 17/ha, Brand, Keith & Fischer 1976; south-central Yukon: 8/ha, Slough & Mowat 1996; Northwest Territories: 8/ha, Poole 1994; vs. 2/ha at Kluane). Despite apparently higher prey density further south in their range then, lynx attain higher densities in the north. In our study area, densities of lynx were approximately double those of coyotes at all phases of the cycle. The reverse was true in the Alberta study area (Brand, Keith & Fischer 1976; Keith et al. 1977; Todd, Keith & Fischer 1981). Numerical responses of specialist predators such as lynx are usually delayed, and they therefore tend to destabilize predator-prey interactions, while the functional responses of generalists are stabilizing (Murdoch & Oaten 1975; Hassell & May 1986; Crawley 1992). However, strong numerical responses of generalist predators to cyclic prey have also been reported for coyotes (approximately 10-fold change) responding to changes in the abundance 45 of black-tailed jackrabbits {Lepus californicus) in Utah (Clark 1972; Wagner & Stoddart 1972; Knowlton & Stoddart 1992), bobcats (Lynx rufus; 9-fold change) in relation to jackrabbit densities in Idaho (Knick 1990), and red foxes (Vulpes vulpes) to European vole populations (Goszczynski 1977; Angelstam, Lindstrom & Widen 1985; Small, Marcstrbm & Willebrand 1993). Survival of Predators The decline in lynx numbers in our study area was in part due to lower survival of adults during the decline phase of the cycle. Our results are comparable to the two other concurrent studies of lynx in northern Canada (Poole 1994; Slough & Mowat 1996). Survival of radio-collared lynx was high (75-90% per year) in both of these studies during all years except for the decline phase of the cycle. In our study, only 1 of 11 lynx (9%) survived the second winter after the cyclic peak, while 40% (n = 30) survived in the south-central Yukon (Slough & Mowat 1996), and 27% (n = 16) survived in the Northwest Territories (Poole 1994) during the same time period. In Idaho, the annual survival of bobcats declined from 78% to 16% during a cyclic decline in numbers of jackrabbits (Knick 1990). These are the first studies to document causes of mortality in untrapped or lightly-trapped populations of lynx. Both other studies in northern Canada noted that starvation, predation, and cannibalism were the main causes of mortality that they could identify (Poole 1994; Slough & Mowat 1996). Predation among predators (intraguild predation) has been recorded among many species of carnivores (e.g., Elsey 1954; Eaton 1979; Stephenson, Grangaard & Burch 1991; O'Donoghue, Hofer & Doyle 1995). Intraguild predation may have important consequences in some predator-prey systems (review in Polis, Myers & Holt 1989), but it is not known whether the lynx killed in our studies would have survived or starved had they not been killed by predators. 46 In our study, and in the study of Poole (1994, 1995), there was a complete turn-over of resident lynx in the study area during the population decline. We know of only 1 animal which survived on our study area from the previous cyclic peak (1981) until the most recent decline (1992). We therefore have little evidence supporting the "core population hypothesis", which proposes that a cohort of permanent residents persist through cyclic lows in relatively stable home ranges (Breitenmoser, Slough & Breitenmoser-Wiirsten 1993). In south-central Yukon, the majority of animals surviving into the low phase of the cycle were born in the study area (Slough & Mowat 1996), but it remains to be seen whether they survive to breed during the next cyclic increase. We have insufficient data on the survival of radio-collared coyotes during the hare cycle, since most animals dispersed out of range before they died. The few mortalities that we did identify were all human-caused. Recruitment of Predators The decline in populations of both predators was associated with lower recruitment. We did not directly measure the reproductive output of either species, but since we failed to hear howling by any family groups of coyotes after 1991, and lynx kits stay with their mother through their first winter (Mowat, Slough & Boutin 1996), we propose that the changes in group sizes that we observed were due to lower recruitment rather than early dispersal of young. There is ample evidence that cyclic declines in lynx numbers are associated with drastic declines in their reproductive output (Brand, Keith & Fischer 1976; Brand & Keith 1979; Parker et al. 1983; O'Connor 1986; Poole 1994; Mowat, Slough & Boutin 1996; Slough & Mowat 1996). During the low phase of the hare cycle, recruitment is generally zero. Declines in litter sizes, pregnancy rates, age at first reproduction, and kitten survival all contribute to 47 the lower recruitment. Bailey (1974) and Knick (1990) also documented drops in recruitment by bobcats during crashes in jackrabbit numbers in Idaho. Likewise, although coyotes select a broader range of prey species across their range, cyclic declines of snowshoe hares (Todd, Keith & Fischer 1981; Todd & Keith 1983) and jackrabbits (Clark 1972), have also been associated with lower body condition and decreased reproductive output. Lower recruitment has also been noted for red foxes during cyclic declines in vole numbers (Lindstrom 1989), and wolves (Canis lupus) during a decline in deer populations (Mech 1977). Canids in general have higher reproductive potential than felids, and can thus take greater advantage of short-term fluctuations in food abundance (Eisenberg 1986), but lynx have one of the largest potential litter sizes of felids and do not differ from coyotes in their potential rate of population increase. Dispersal of Predators The cyclic decline of hares on our study area was accompanied by high rates of dispersal of lynx. This was also the case in studies in the south-central Yukon (Slough & Mowat 1996) and Northwest Territories (Poole 1994). Mech (1980) reported influxes of lynx into Minnesota that were associated with declining populations of hares to the north. Likewise, bobcats in Idaho went on long forays and dispersed at higher rates during periods of declining and low jackrabbit populations in Idaho (Knick 1990). Lynx are quite vulnerable to trapping mortality (Ward & Krebs 1985), and because of this, we were able to learn the eventual mortality sites of some very long-distance dispersers. Slough & Mowat (1996) also reported dispersal distances of up to 1100 km for radio-tagged lynx. Although dispersal was biased towards male kits in the south-central Yukon, we did not observe a sex-bias in our study. 48 Most of the loss of coyotes from our study area was due to dispersal. Few trappers concentrate their effort on coyotes in the north, so we received collar returns from only two dispersers, both less than 50 km from their capture sites. Long-distance natal dispersal by coyotes may be typical, however (Harrison 1992). High mobility of predators has been cited as a potential factor in causing regional synchrony of cyclic populations (Finerty 1980; Ims & Steen 1990), and in allowing nomadic avian predators to track and dampen local outbreaks of vole numbers in Fennoscandia (Korpimaki & Norrdahl 1991a). Long-range movements of mammalian predators during cyclic declines may also act to keep the fluctuations of local hare populations in synchrony. How Accurate are our Estimates of Predator Density? Our estimates of the numbers of coyotes and lynx in our study area were based on a comparison of the locations of radio-tagged animals with the locations and numbers of tracks found by extensive snow-tracking, as well as howling for coyotes. Since we did not estimate densities of predators using a statistical model, we can put no confidence limits on our estimates. During years of low hare numbers, the task of estimating the number of animals in the study area was relatively easy, since there were very few animals present (5 coyotes and 8 lynx at the cyclic low). At peak predator densities though, there were a great many tracks, and we were not as certain of our estimates. We based our estimates of lynx numbers on the assumption that the core areas of lynx home ranges were relatively intrasexually exclusive. Based on our radio-telemetry locations, this assumption was valid for all years except for the decline winters of 1991-92 and 1992-93 (U. Breitenmoser, unpublished data). Data from the Northwest Territories also support this assumption at cyclic peaks, and the social system of lynx seemed to break down at low hare densities there as well (Poole 1995). We had an additional check on monitoring numbers of 49 resident females at the peak of the cycle, since each traveled with their kits, which often differed in number. Our estimates of coyote numbers were based on the assumption that coyotes lived in relatively exclusive family group territories. This assumption has been supported by numerous studies (Bowen 1982; Messier & Barrette 1982; Bekoff & Wells 1986), although most studies have also noted that a portion of coyotes persist as transients or solitary residents, with home ranges that overlap those of resident pairs and packs. Since we recorded howling throughout the year, it is not likely that we misjudged the numbers of residents, but we may have missed an unknown number of transients. In summary, our estimates of predator numbers were likely very accurate before and after the peak and early decline phases of the cycled During the peak and early decline though, greater movement of animals (Ward & Krebs 1985; Slough & Mowat 1996), a possible increase in the number of transients and "floaters", and a break-down of the social system (Brand, Keith & Fischer 1976; Poole 1995) may have caused us to miss animals. Therefore, our estimates of the peak densities of predators may be conservative, although we do not believe we underestimated by more than 5 animals in the study area. Our indices of predator numbers, from track counts, were closely correlated with our estimates of population sizes. Although we used tracks to make our assessment of predator numbers, we only used the locations and group sizes of these tracks, rather than their abundance, in estimating populations of predators. Therefore, the two methods were essentially independent. Track counts suggest that we may have overestimated the number of lynx on our study area during the winter of 1989-90 (Fig. 2.2). Track counts have been often used as an indices of animal numbers (e.g., Stephenson & Karczmarczyk 1989; Thompson et al. 1989). In an evaluation of the method for lynx, Stephenson & Karczmarczyk (1989) found a poor correlation between the index and an estimate of lynx numbers based on radio-telemetry, especially at low densities of lynx. They 50 postulated that increased movements of lynx during cyclic lows (Ward & Krebs 1985) would bias track counts as indicators of population trends. However, these authors counted tracks for only 2-5 weeks each year on an average of 542 km, and factors such as weather, location of the transect, and time of year could have introduced a great amount of variability among years. We ran our transect throughout each winter, for an average of 1238 km per year. By controlling for factors such as date, track location, and weather in our analysis, we were also able to factor out some of the year-to-year variance which would have affected our counts. Conclusions Coyotes and lynx responded numerically in much the same way to the fluctuation in numbers of hares in our study area. The limited availability of alternative prey in the boreal forest gave both predators restricted options when densities of hares were low. Lynx were apparently more skilled at capturing red squirrels than coyotes (Chapter 3), and although coyotes killed more small mammals than lynx, the total biomass and availability of voles was quite low. "Generalist" predators may behave as specialists in such low-diversity systems. Based on work in Idaho, Johnson & Hansen (1979a) concluded that coyotes were highly selective predators of rabbits rather than opportunistic predators. The relatively low density of coyotes in our study area suggests that the boreal forest is not optimal habitat for them. We found evidence that changes in survival, recruitment, and dispersal rates all contributed to changes in the populations of predators. Numerical responses which are due to changing rates of reproduction are likely to be more de-stabilizing to predator-prey interactions than those based on aggregation or changes in survival (Crawley 1975). The time-lag of the decline of predator numbers also tends to lead to cyclic dynamics (May 1973). Although, as a rule, numerical responses are more pronounced in specialist than generalist 51 predators (Crawley 1992), and generalists tend to stabilize predator-prey interactions (Hanski, Hansson, & Henttonen 1991), both coyotes and lynx responded numerically to the cycle in hare numbers more in the manner of hare specialists, which would contribute to the cyclic behaviour of the system. 52 CHAPTER 3 FUNCTIONAL RESPONSES OF COYOTES AND LYNX TO THE SNOWSHOE HARE CYCLE Introduction Populations of snowshoe hares undergo regular cycles in abundance throughout the northern boreal forest in North America, with amplitudes over two orders of magnitude, and 8-11 years between cyclic peaks (Elton & Nicholson 1942; Keith 1990). These cycles have major effects on predators of hares, as well as on other herbivores, in the relatively simple boreal ecosystem (Finerty 1980). Recent empirical and theoretical studies have suggested that predation may be a necessary factor in causing cyclic dynamics of populations of hares (Akcakaya 1992; Royama 1992; Stenseth 1995; Krebs et al. 1995). Likewise, Scandinavian researchers have presented theoretical (Hanski, Hansson & Henttonen 1991; Hanski et al. 1993; Hanski & Korpimaki 1995) and empirical (Henttonen et al. 1987; Korpimaki & Norrdahl 1991b; Korpimaki 1993) evidence that predation is an essential factor in generating 3-4-year cycles of microtine rodents in northern Fennoscandia. Strong functional responses by a large suite of generalist predators may regulate vole numbers in southern regions (Erlinge et al. 1983, 1984, 1988), while delayed numerical responses of specialist predators—mostly weasels—cause a lag and cyclic dynamics in the north (Korpimaki, Norrdahl & Rinta-Jaskari 1991). The effect of any predators on populations of their prey are determined by their numerical (changes in reproduction, survival, or aggregation) and functional (changes in kill rates) responses to prey density (Solomon 1949). In order for predation to have a regulatory effect, the proportion of a prey population killed must be both increasing with prey density (i.e., density-dependent), and greater than the net production of prey at prey densities 53 exceeding an equilibrium (Sinclair & Pech 1996). Depensatory predation, in which the percent of prey killed increases with declining prey abundance, can result from time lags in the responses of predators to fluctuations in prey abundance. The shapes and timing of the numerical and functional responses of predators to prey are thus critical to determining the effects of predation. Holling (1959b) described three basic forms of functional responses of predators to changes in density of prey. Type-1 responses describe linearly increasing kill rates with prey density up to a threshold, above which the rate is constant. Type-2 responses describe predation increasing at a monotonically decreasing rate with prey density, up to an asymptote, and type-3 responses are sigmoidal in shape. Type-2 responses may be typical of specialist predators, predators with few alternative prey, or they may arise due to adaptive adjustment of search rates and time to the benefits and costs of foraging (Abrams 1990, 1992a). Type-3 responses are more typical of generalist predators, and they may be the result from a number of mechanisms: 1) predators learning to recognize, capture, or handle prey better with increasing prey density (Holling 1959b), 2) predators switching among prey types, habitats, or foraging tactics (Royama 1970; Murdoch & Oaten 1975; Akre & Johnson 1979), 3) adaptive variation in foraging rates (Holling 1966; Hassell, Lawton & Beddington 1977; Dunbrack & Giguere 1987), and 4) changes in the behaviour of prey, or prey having a refuge below a fixed density (Murdoch & Oaten 1975; Taylor 1984). Only type-3 responses describe density-dependent predation, and there is evidence from field (e.g., Pech et al. 1992) and theoretical (e.g., Hassell & Comins 1978; Nunney 1980) studies that sigmoidal functional responses may stabilize predator-prey interactions. The main predators of snowshoe hares are the same through much of their range-lynx, coyotes, great horned owls, and goshawks. Keith and colleagues (Keith et al. 1977) estimated the population sizes and kill rates of these predators during a complete hare cycle in order to analyze the effects of predation on hare dynamics. They found that both coyotes and lynx 54 responded to the increase in hare numbers with increasing kill rates. Coyotes killed virtually no hares at cyclic lows, while they killed an estimated 0.66/day per coyote when numbers of hares were high. The kill rate by lynx increased about 3-fold over the same time period. The functional responses of coyotes were judged to be type-3, while those of lynx were type-2 (Keith et al. 1977). Lynx are usually considered prototypical specialists on snowshoe hares, and virtually all studies of their food habits have shown hares to be their predominant prey (Saunders 1963a; Brand & Keith 1979; Parker et al. 1983; Ward & Krebs 1985). They are morphologically well-adapted to hunting in the deep snows of the north (Murray & Boutin 1991) , and their geographic range overlaps almost exactly with that of hares (Banfield 1974). Coyotes, in contrast, are often considered prototypical generalists. Over their large and expanding range in North America, they are adapted to a wide variety of habitats, climates, and foods (see papers in Bekoff 1978). Morphologically, coyotes are not well-suited for hunting in deep, soft snow, having a relatively high foot-load (Murray & Boutin 1991). Coyotes are fairly recent immigrants into the far north, appearing in the Yukon between 1910 and 1920 (G. Lotenberg, 1994, unpublished report for Parks Canada). The effects of generalist and specialist predators on prey populations are likely to be very different, based on theoretical (Murdoch & Oaten 1975; Hassell & May 1986; Crawley 1992) and empirical (e.g., Erlinge et al. 1983; Henttonen et al. 1987; Korpimaki 1993) evidence. Rapid functional responses of generalist predators may increase the stability of predator-prey interactions, while the delay in the numerical responses of specialists may introduce lags and dynamic instability. While there is ample evidence that the hare cycle has persisted since long before coyotes immigrated into the north (Elton & Nicholson 1942), there is little quantitative evidence of their effect on hare populations, populations of alternative prey, and other predators. There have been very few studies conducted of coyotes in contiguous boreal forest 55 (Theberge & Wedeles 1989; Murray & Boutin 1991; Murray, Boutin & O'Donoghue 1994; Murray et al. 1995)-Keith's study area was approximately one-third agricultural land, which was used by coyotes extensively during periods of low hare numbers (Todd, Keith & Fischer 1981). The functional responses in Keith's study area were calculated using an unverified energetic model for coyotes (i.e., assuming no surplus killing or wastage), and an assumption that lynx only rested once per day (Keith et al. 1977). There is therefore a need for direct measurement of the functional responses of coyotes and lynx in the boreal forest to better understand their effects on the snowshoe hare cycle. Objectives, Hypotheses and Predictions The goals of this chapter are to present and contrast the functional responses of coyotes and lynx to the snowshoe hare cycle, and to examine the associated changes in the "components" of these responses (Holling 1959b, 1966)-rate of successful search, foraging time, and handling time. In Chapter 4,1 examine the behavioural responses—switching between prey types, habitat patches, and foraging tactics-behind the functional responses. This research is aimed at meeting part of a larger objective of evaluating the total impact of mammalian predators on the dynamics of snowshoe hares. Both predators are approximately the same size (9-12 kg adult body size in the southwest Yukon), and they are therefore reliant on essentially the same prey base. Specifically, I present the average kill rates of coyotes and lynx each winter during a cycle in abundance of hares in the southwest Yukon. I then examine the predators' travel rates, reactive distances, capture success, foraging time, and handling time to determine mechanisms contributing to the functional responses. Adaptive changes in these components with prey density can result in either type-2 or type-3 functional responses, and it is valuable 56 to look at the separate components to understand the predators' reactions to changing density of hares (Abrams 1990). From the data of Keith and co-workers, I assume that snowshoe hares are the preferred (i.e., most profitable) food for both coyotes and lynx. Data presented in Chapter 4 supports this assumption. I postulate two null hypotheses: 1) The shapes of the functional responses of coyotes and lynx are determined only by the relative abundances of hares and alternative prey. If this were true, then the functional responses and dietary composition of both species should be similar as hare numbers fluctuate. Alternatively, lynx may respond very quickly to increasing hare density and have type-2 functional responses, while coyotes may switch to and from alternative prey as the relative abundances of prey change, resulting in type-3 responses. 2) The asymptotes of the functional responses to hares of both predators are determined by their respective daily metabolic requirements (estimated at about 0.9 hares per day for coyotes (Litvaitis & Mautz 1980), and 0.4 hares per day for lynx (Nellis, Wetmore & Keith 1972)). Alternatively, one or both species may exhibit surplus killing or wastage (Kruuk 1972), and kill more animals than expected at high densities of hares. Study Area This study was conducted in the southwest Yukon Territory, Canada (60° 57' N, 138° 12' W), in a broad glacial valley approximately 350 km2 in area (Fig. 2.1). The study area is described in Chapter 2. 57 Methods This study was conducted as a part of the Kluane Boreal Forest Ecosystem Project, an experimental study of the vertebrate food web in the northern forests (Krebs et al. 1995). These results summarize the efforts of a great many biologists, technicians, and students who have worked with the Kluane Project. Field work was started in 1986, and continued through the winter of 1994-95. With the exception of some scat collection, live-trapping and radio telemetry, field work was conducted during the winter months, from October to April. Estimation of Hare Densities We estimated densities of snowshoe hares by live-trapping on 1-3 60-ha grids each March and October-November, from 1986 to 1995 (Boutin et al. 1995). Population estimates were made using the jackknife estimator in program CAPTURE (White, Burnham & Otis 1982). I used the mean of density estimates of hares in fall and spring for calculation of functional responses. A sample of hares was radio-collared during this period (radio collars from Lotek, Newmarket, Ontario), and they were monitored each 1-2 days for estimation of survival rates (Krebs et al. 1995). Snow-tracking of Coyotes and Lynx We followed the tracks of coyotes and lynx each winter from 1987-88 to 1994-95, beginning as soon as there was enough snow, usually in mid- to late October, and finishing at the end of March. Nineteen different observers snow-tracked predators during the 8 winters of study. This was the main method that we used for determining species and frequencies of kills and attempted kills, hunting success rates, lengths of chases, proportions of prey eaten, 58 frequencies of scavenging and caching, and frequencies of beds. We also collected scats along these tracks for later analysis. Fresh tracks were selected along snowmobile trails and roads in our study area on any days that weather and snow conditions permitted. We attempted to spread our tracking effort as evenly as possible over the study area, between coyotes and lynx, and among different group sizes of predators. Once a track was selected, it was usually followed backwards (relative to the animal's direction of travel) until it was lost due to poor snow conditions or confusion with the tracks of other animals. We then tried to follow it forwards as far as possible, to have continuous segments of maximum length—this sometimes required several days of following the same tracks. We counted the distance tracked on hand-held tally counters (numbers of paces were later converted to meters, using observer-specific conversion factors), and recorded events along the tracks on micro-cassette recorders. At each site of a kill, we recorded the prey species, number of bounds by the predator, proportion of the carcass consumed (estimated subjectively), which parts were left uneaten, whether or not any of the carcass was cached, and a detailed description of any signs left. We recorded the prey species and number of bounds for all attempted kills. At sites where predators scavenged old kills or retrieved caches, we estimated the amount of food eaten when possible (e.g., impressions from cached prey were sometimes clear enough to see what parts of the prey were cached), and noted any evidence of the initial cause of death of the prey. All beds were classified as crouches (or hunting beds), "short" beds (where the predator had laid down but did not stay long enough to melt the snow), or resting beds. Additional data gathered about use of habitat, travel on trails, and hunting tactics will be summarized in Chapter 4. We analyzed tracking data using the MGLH (Multivariate General Linear Hypothesis) procedure in SYSTAT (Wilkinson 1990), and considered P < 0.05 the criterion for rejecting 59 null hypotheses. Proportions of diet by biomass were calculated by multiplying numbers of prey killed by 1500 g for hares, 250 g for red squirrels, and 20 g for small mammals. Scat Analysis We analyzed scats of coyotes and lynx collected in 1987-88 to 1993-94, to supplement our estimates of the predators' diets from the tracking data (scats collected in 1994-95 have not yet been analyzed). Scats were frozen and stored until ready for analysis, at which time they were autoclaved for 20-30 minutes. Scats were analyzed by several different observers and labs, each using slightly different techniques for handling scats and quantifying the undigested remains of prey species. In all protocols, scats were manually broken apart, 10-30 random samples of hairs were selected, and hairs from these sub-samples were identified based on colour-banding, and patterns of cuticular scales and medullary pigments (e.g., Moore et al. 1974). We report the results of scat analyses here only as the relative frequency of occurrence of different prey items in scats, since this measure does not depend on the protocol used for quantifying the contents of scats. Although smaller-sized prey are frequently over-represented in scat samples (Floyd, Mech & Jordan 1978; O'Gara 1986), Johnson and Hansen (1979b) reported that relative frequency of occurrence of prey in scats of coyotes was closely related to the percent intake by biomass. Monitoring of Radio-collared Predators We live-trapped coyotes and lynx, mostly during the fall and winter months, using padded foot-hold traps (No. 3 Soft-catch ® traps, Woodstream Corp., Lititz, Pennsylvania), cable snares (Fremont leg snares ®, Fremont Humane Traps, Candle Lake, Saskatchewan), 60 and box traps (techniques similar to Mowat, Slough & Rivard 1994). We checked traps at least once per day, and locked traps open when the temperature was less than -20 C. During the first 4 years of the study, we immobilized captured animals using a 7:1 mixture of ketamine hydrochloride and xylazine hydrochloride, but we switched to using a mixture of tiletamine hydrochloride and zolazepam hydrochloride (Telazol ®, A. H. Robbins Co., Richmond, Virginia) since then (Poole, Mowat & Slough 1993). We fitted animals with radio-collars weighing 250-350 g (Telonics, Inc., Mesa, Arizona); most of these had internal mercury switches sensitive to movement, which allowed us to monitor the activity of collared animals. The pulse rate of the radio signal switched between 60 and 90 beeps per minute, depending on the inclination of the mercury switch. We monitored the radio-collared animals to gather information on their travel rates and activity patterns, in order to interpret the snow-tracking data. From 1989 to 1995, we opportunistically measured normal travel rates of coyotes and lynx when they were in accessible locations. To do this, we first walked in on a collared animal until we gained a very accurate fix on its location (either by actually seeing the animal or waiting until it left a bed, which we subsequently located). We then monitored the animal's activity, by listening to its radio signal, until it either settled down into a bed or was in a location that we could once again pinpoint. On occasions that we successfully determined two exact locations without disturbing the animal, we then followed the animal's tracks between the two sites and measured its exact travel distance, from which we calculated a travel rate. We also monitored the signals of radio-collared coyotes and lynx during continuous blocks of time from 1990 to 1995, to determine their activity patterns. We set up monitoring stations at several high-altitude locations in the study area, where we could regularly receive the signals of collared animals wherever they moved in their home ranges. We use a Lotek SRX-400 programmable receiver with a data logger (Lotek, Newmarket, Ontario) to monitor the activity of predators. We set this to record the pulse rate and signal strength of a target 61 animal's collar each 30 seconds, based on 3 signal pulses each reading. At these settings, we could record about 18 hours of continuous activity before the memory of the data logger was full. The resulting data was a series of short (<: 3 seconds) measures of pulse rate and signal strength, taken each 30 seconds, from which we interpreted the animal's activity. In general, strings of consecutive readings with constant pulse rates and signal strengths meant that the animal was inactive-if the pulse rate was fast and constant, the animal's head was down (resting); if the pulse rate was slow and constant, the animal's head was up, and it could be lying down, sitting, or standing. Radio signals from moving animals typically were variable in strength, and the pulse rate changed between slow and fast as the orientation of the mercury switch in collar changed. We conducted a series of trials with activity monitoring to develop a statistical technique to interpret these data. We simultaneously monitored the signals of target animals using the automatic receiver and an observer listening constantly to the signal. Observers recorded continuous records of the activity of 1 coyote for 20 hours (in 4 different sessions), and 5 lynx for 45.75 hours (in 15 different sessions). We used the activity recorded by the observers as the standards for interpreting the data records from the automatic receiver. We used discriminant function analysis to analyze the activity data. We first calculated separate discriminant functions for coyotes and lynx based on our trial data. We calculated the mean, median, variance, and range of the pulse rate and signal strength in a 5-minute sliding window from the data record of the automatic receiver. These were standardized within each collar to a mean = 0 and standard deviation = 1, and each period was classified as "active" or "inactive" based on our trial monitoring. The discriminant functions calculated with these variables correctly predicted active periods for coyotes 92% of the time and for lynx 79% of the time, and inactive periods for coyotes 77% of the time, and for lynx 78% of the 62 time. We used these discriminant functions to classify all of the activity records from the automatic receiver. We calculated the overall percent time active each winter for each animal monitored, in six 4-hour blocks, beginning with 24:00-03:59. The mean of these was taken as the overall percent time active per animal per year, and then the means of all coyotes and all lynx within each year were calculated to give us estimates of the average percent time that coyotes and lynx were active each winter. Confidence limits were calculated using arcsine-transformed data. We carried out a further check on the accuracy of this method in 1991-92, by listening to the signals of 3 coyotes for 122 hours and of 9 lynx for 295 hours, and subjectively classifying them as active or inactive each 5 minutes based on variations in signal strength and pulse rate. We calculated the overall percent time active of these animals in the same manner as above. Calculation of Functional Responses We used travel rates and activity patterns estimated from radio-collared animals to transform kills per km trail from our snow-tracking data to kills per day. Kill rates were adjusted for group size for all tracks where more than one animal was followed. We calculated the amount of "active" time represented by our tracks in each winter, by predator species, by dividing the total track distances by measured travel rates. We then divided the active times for coyotes and lynx by the percent times they were active each year to estimate the total amount of coyote and lynx time (including resting periods) represented by our snow-tracking each winter. Mean kill rates of hares per day were calculated for each predator each winter. We calculated the functional responses of coyotes and lynx by plotting mean kill rates against the means of fall and spring densities of hares, based on live-trapping. For 63 descriptive purposes, we fitted Holling's disc equation (1959a) to these data, using the NONLIN least-squares curve-fitting procedure in SYSTAT (Wilkinson 1990). It would have been desirable to statistically test whether the data were best described by type-1, type-2, or type-3 curves. However, given that we only have 8 data points (from 8 winters) per predator, and no replication at each density of hares, this would have been little more than an exercise in curve-fitting, with little power to distinguish among the true shapes of the curves (Trexler, McCulloch & Travis 1988). Monitoring of Caches We observed caching of prey and use of caches by both coyotes and lynx during the first years of this study. We had no way of knowing, however, how often and when caches were returned to, or whether the caches that we located while snow-tracking had been killed by the same species that ate it later in the winter, killed by a different species, or had originally died of causes besides predation. Distinguishing among these alternatives is important in understanding the kill rates and foraging tactics of the predators. Beginning in the winter of 1992-93, we began monitoring caches made by coyotes and lynx, which we located while snow-tracking or from kills of radio-collared hares between October and April. When we located caches with any substantial meat left on them, we disturbed the sites as little as possible, and set up monitoring radios. These each consisted of a beacon radio, hung in a tree overhead so the site could be located again, and a second "cache radio", which was buried in the snow near the cache. The cache radio had a steel wire wrapped around its battery, and the batteries were left uncemented to the radio transmitters. The wire ran under the snow, and was twisted around an accessible part of the cache (often a leg). When caches were disturbed, the battery pulled loose from the radio, which then stopped transmitting. The cache radios were monitored each 2 days; when we could not hear 64 signals from them, we followed the signals of the beacons to the sites, and recorded data about the cause of the disturbance, how much of the cache was eaten, and details of the cache retrieval. Results Density of Hares Densities of hares increased from 1986-87 to a peak of about 2/ha in 1989-90, and then declined to very low numbers by the winter of 1992-93 (Fig. 3.1). The cyclic amplitude was 26-44-fold, depending on whether fall or spring densities are considered (Boutin et al. 1995). Numbers of hares began to climb again in 1994. Diets of Coyotes and Lynx Coyotes. We followed the tracks of coyotes for 1897 km during the 8 winters from 1987-88 to 1994-95 (mean per winter, 237 ± 100 (S.D.) km). We found 189 kills by coyotes, 47.1% of which were hares, 13.2% red squirrels, and 37.6% small mammals (Fig. 3.2). Coyotes killed mostly hares from 1987-88 to 1991-92, plus some red squirrels and small mammals from 1989-90 to 1991-92. Small mammals comprised most kills in 1992-93 and 1993-94, while hares and red squirrels were killed more in 1994-95. In terms of biomass, the diets of coyotes were comprised largely of hares in all winters (Fig. 3.3). Small mammals made up a maximum of 9.5% in 1993-94, a winter of high vole density, and red squirrels made up maximum of 19.9% in 1994-95. 65 Fig. 3.1. Estimated densities of snowshoe hares (means of autumn and late winter estimates from live-trapping on 1-3 60-ha grids) from 1986 to 1995 in the southwest Yukon. 66 Winter 67 Fig. 3.2. Percent of kills by prey species located along tracks of coyotes and lynx during winter, 1987-88 to 1994-95. Sample sizes are given above bars. "Other" kills for coyotes were 2 grouse, 1 flying squirrel, and 1 unknown prey, and for lynx were 12 grouse, 2 muskrats, 2 flying squirrels, 1 short-tailed weasel, 1 red fox, and 1 unknown prey. 68 n = 5 27 20 24 12 13 69 19 a Other Winter 69 Fig. 3.3. Percent biomass of kills by prey species located along tracks of coyotes and lynx during winter, 1987-88 to 1994-95. Sample sizes (number of kills) are given above bars. 70 87- 89- 91- 93-88 90 92 94 Winter 71 Scat analyses suggest that coyotes used a greater diversity of prey during the cyclic increase (1987-88 and 1988-89) and low (1992-93 and 1993-94) phases in numbers of hares than indicated by our snow-tracking (Fig. 3.4). The relative frequency of occurrence of small mammals in scats ranged from 18.0 to 42.2% during these years~we likely missed many of these kills while snow-tracking. Hares comprised most of the diets of coyotes from 1989-90 to 1991-92 (relative frequency of occurrence 82.9-90.7%), but only 31.1-56.4% (based on scats) in other years. Red squirrels were a minor component of the diet in all winters (relative frequency of occurrence 0.0-6.7%). Lynx. We followed the tracks of lynx for 2232 km during the 8 winters from 1987-88 to 1994-95 (mean per winter, 279 ± 85 (S.D.) km). We found 502 kills by lynx, 50.2% of which were hares, 34.7% red squirrels, and 11.0% small mammals (Fig. 3.2). Lynx killed mostly hares from 1987-88 to 1991-92, plus some red squirrels in 1987-88. Red squirrels comprised most kills from 1992-93 to 1994-95, with some small mammals in 1993-94. In terms of biomass, the diets of lynx were comprised largely of hares from 1987-88 to 1991-92, but red squirrels became increasingly important from 1992-93 to 1994-95 (20.4-43.9%; Fig. 3.3). Small mammals made up a negligible proportion of the diet in all winters. Scat analyses suggest that lynx used a greater diversity of prey during the cyclic increase (1987-88 and 1988-89) and low (1993-94) in numbers of hares than indicated by our snow-tracking (Fig. 3.4). The relative frequency of occurrence of small mammals in scats ranged from 23.1 to 25.5% during these years. Hares comprised most of the diets of lynx from 1989-90 to 1991-92 (relative frequency of occurrence 84.6-94.4%), and 38.5-60.0% of diets (based on scats) in other years. Red squirrels were an important component of the diets of lynx in 1992-93 and 1993-94 (relative frequency of occurrence 25.0-36.5%). 72 Fig. 3.4. Relative frequency of occurrence of prey species in the scats of coyotes and lynx during winter, 1987-88 to 1993-94. Sample sizes (number of scats analyzed) are given above bars. Relative frequency of occurrence is roughly equivalent to biomass ingested, although smaller-sized prey species may be overestimated. "Other" prey species for both predators include moose, Arctic ground squirrel and unidentified prey remains in scats. 73 Winter 74 s J Travel Speed of Coyotes and Lynx Between 1989 and 1995, we successfully completed measures of the travel rates of coyotes 7 times (average duration 46 minutes, range 15-150 minutes), and of lynx 16 times (average duration 99 minutes, range 11-274 minutes). Coyotes traveled at 2.49 ± 0.39 (S.D.) km/hr (range 1.77-3.06 km/hr), while lynx traveled at 1.09 ± 0.21 (S.D.) km/hr (range 0.75-1.46 km/hr). Activity of Coyotes and Lynx Coyotes. We monitored the activity of 3 coyotes in 1990-91 (522 hours), 3 in 1991-92 (279 hours), 1 in 1992-93 (73 hours), and 2 in 1993-94 (468 hours). The percent time active averaged 46.7%, and varied little among years (44.3-49.7%; Fig. 3.5). Likewise, the number of beds of coyotes that we found while snow-tracking was approximately the same in all years (range 2.5-4.3 beds/10 km trail; Fig. 3.6). Lynx. We monitored the activity of 5 lynx in 1990-91 (1595 hours), 4 in 1991-92 (181 hours), 6 in 1992-93 (195 hours), 5 in 1993-94 (845 hours), and 6 in 1994-95 (1441 hours). The percent time active averaged 41.5%, and varied little among years (39.2-43.5%; Fig. 3.5). Unlike coyotes though, the number of beds of lynx that we found while snow-tracking increased over 3-fold from 1989-90 (5.1 beds/10 km trail) to 1993-94 (18.4 beds/10 km trail; Fig. 3.6). The frequency of resting beds was more stable (range 1.9-5.0/10 km trail) than that of "short" beds (range 3.2-13.4/10 km trail) during this period. Accuracy_of_Methad. We listened to the signals of 3 coyotes (122 hours) and 9 lynx (295 hours) during the winter of 1991-92 to assess the accuracy of inferring activity patterns from the output of the automatic receiver. Coyotes were judged to be active 47.4% of the time based on this monitoring, compared to 48.1% of the time from the automatic receiver, and 75 Fig. 3.5. Percent time active for coyotes and lynx from 1990-91 to 1994-95, based on monitoring the signals of radio-collared animals. Error bars represent 95% confidence limits with arcsine-transformed data from all animals which were monitored during all time periods of the day. 76 80 Percent of Time Active 60 40 -20 [ 0 90- 91- 92- 93- 94-91 92 93 94 95 Winter 77 Fig. 3.6. Frequency of resting beds and "short" beds (where the animal had laid down, but not stayed long enough to melt the snow) along trails of coyotes and lynx from 1987-88 to 1994-95. 78 Winter 79 lynx were judged to be active 41.9% of the time, compared to 40.0% of the time from the automatic receiver. Analyzing the output of the automatic receiver therefore seems to be an accurate method of assessing activity of predators. Functional Responses of Coyotes and Lynx Coyotes. Coyotes showed a strong functional response to changes in density of hares (Fig. 3.7). (Activity measurements from 1990-91 were used for the 3 previous winters, and those from 1993-94 were used for 1994-95, to calculate the kill rates in those winters in which no measurements of activity were made). The disc equation was fitted to these data for the purpose of description (fitted parameters a = 0.021, h = 0.418), but its fit (r2 = 0.83) is little better than the fit of a linear functional response (r2 = 0.80). The maximum kill rate per coyote suggested by the disc equation is 2.4 hares per day, which is close to what we observed in 1988-89 (2.3 hares/day). Kill rates peaked in 1988-89, a year before hare numbers peaked, and they were generally higher during periods of increase in density of hares (1987-88 to 1988-89, and 1994-95) than during the peak, decline, and low phases of the cycle (1989-90 to 1993-94). Lynx. Lynx also showed a functional response to changes in density of hares (Fig. 3.7). (Activity measurements from 1990-91 were used for the 3 previous winters to calculate the kill rates in those winters in which no measurements of activity were made). The disc equation was fitted to these data for the purpose of description (fitted parameters a = 0.031, h = 0.756), and it fits the observed kill rates quite closely (r2 = 0.97). The maximum kill rate per lynx suggested by the disc equation is 1.3 hares per day, which is close to what we observed in 1990-91 (1.2 hares/day). Kill rates peaked in 1990-91, the year after hare numbers peaked, and unlike coyotes, they were generally lower during periods of increase in density of hares 80 Fig. 3.7. Functional responses of coyotes and lynx (kills per day per individual predator) to the density of snowshoe hares from 1987-88 to 1994-95. Numbers next to each point indicate the year starting each winter (e.g., 87 = 1987-88). 81 2.5 r 2.0 -Kills of 1.5 -Hares per 1.0 -Day 0.5 -0 /-2.5 -2.0 -Kills of Hares 1.5 -per 1.0 -Day 0.5 -0 L 88 0 50 100 150 200 250 Hares per 100 ha 82 (1987-88 to 1988-89, and 1994-95) than during the peak, decline, and low phases of the cycle (1989-90 to 1993-94). Components of the Functional Responses of Coyotes and Lynx Holling (1959a, 1966) considered that there were three basic components of functional responses of predators to prey: the rate of successful search (determined by the reactive distance of the predator to its prey, the rates of movements of both the predator and prey, and the predator's capture success), the time the predator was exposed to the prey, and the handling time (the sum of time spent pursuing, subduing, eating, and digesting prey). These parameters were constants in Holling's disc equation (Holling 1959a), but it is likely they vary adaptively with changing densities of prey (Abrams 1990). We briefly examine our data here for evidence of changes in these components with hare density; other behavioural changes of predators are considered in Chapter 4. EatejJLSaiccessfuLSearch. The only measures that we made of the reactive distances of coyotes and lynx to hares were the lengths of their chases, and we use these here as indices of the actual reactive distances. Chases of hares made by both coyotes and lynx were considerably longer during the low and early increase phases of the hare cycle (1992-93 to 1994-95) than during the late increase, peak, and decline winters (1987-88 to 1991-92) (Fig. 3.8; ANOVA on log-transformed data, Effect of Year, F = 32.26, d.f. = 7,990, P = 0.000). Unsuccessful chases were significantly longer (F = 98.77, d.f. =1,990, P = 0.000) than chases at successful kills. Since 6 of 7 of our measures of rates of travel by coyotes were made during years of high hare density, we can only evaluate changes in travel rates for lynx. The mean travel rate of lynx did not differ between periods of higher numbers of hares (1989-90 to 1991-92; Mean travel rate = 1.02 ± 0.23 (S.D.) km/hr) and periods of lower numbers (1992-93 to 1994-83 Fig. 3.8. Lengths of chases of hares in successful and unsuccessful attempts by coyotes and lynx from 1987-88 to 1994-95. Mean coefficient of variation of chase lengths = 103% (error bars are omitted for the sake of clarity). 84 Winter 85 95; Mean travel rate = 1.13 ± 0.16 (S.D.) km/hr; T-test, t = 1.63, d.f. = 14; P = 0.13). We gathered no data on the travel rates of hares. Hunting success of coyotes preying on hares was relatively constant from 1989-90 to 1993-94 (range 26.8-37.5%; Fig. 3.9), but considerably higher in the cyclic increase years of 1988-89 (57.4%) and 1994-95 (68.8%, although the sample size was quite small this year). Success rates of lynx preying on hares were less variable than those of coyotes (range 20.0-38.8%); the lowest success rates were in the last 3 winters of the study (20.0-21.9%; Fig. 3.9) and lynx was fairly constant from 1990 to 1995 (Fig. 3.5). For lynx, however, the percent of time active may not be a good measure of time exposed to prey. Lynx often use ambush beds, and use of ambush and "short" beds increased during the cyclic decline of hares (Fig. 3.6; Chapter 4). Handling Time. We have no measures of the amount of time necessary to digest hares, but we did record the amount of prey eaten per kill for both predators. The percent of the carcasses of hares eaten by coyotes and lynx remained fairly constant from 1987-88 to 1994-95 (Fig. 3.10; ANOVA on arcsine-transformed data, Effect of Year, F = 0.43, d.f. = 7,285, P = 0.886). Coyotes ate 82.2-95.6% of carcasses in all years except 1993-94, when an average of only 62.1% per hare carcass was eaten (but only 5 hares were killed in this winter). Lynx ate an average of 78.2-95.2 % per carcass. The total amount of time devoted to chasing, subduing, and eating hares are likely very minor components of the time budgets of coyotes and lynx though. Scavenging and Caching by Coyotes and Lynx Signs of scavenging were noted frequently along the trails of coyotes, and more scavenging was noted during winters of high hare abundance (Fig. 3.11). We found less . As previously shown, the percent of time active by both coyotes 86 Fig. 3.9. Hunting success rates (% of all chases that were successful) of coyotes and lynx preying on hares and red squirrels from 1987-88 to 1994-95. Sample sizes (number of chases) are given above the bars. 87 Success Rate n = 15 47 56 55 27 8 17 16 n = 4 2 2 5 2 1 1 21 15 100 80 60 40 20 0 Prey = hare Prey = red squirrel • * _i i i Success Rate n = 63 85 184 160 138 114 78 55 n = 1 0 2 0 1 3 151 191 107 100 80 . 60 . 87- 89-88 90 91- 93-92 94 Prey = hare Prey = red squirrel X Winter 88 Fig. 3.10. Percentages of hare carcasses eaten per kill by coyotes and lynx from 1987-88 to 1994-95. 89 90 Fig. 3.11. Frequency of signs of scavenging (visiting old kills or caches) along trails of coyotes and lynx from 1987-88 to 1994-95. 91 92 evidence of scavenging by lynx, and the increase in frequency of scavenging with hare abundance was less pronounced. We used our monitoring of caches to infer whether or not the coyotes and lynx were scavenging animals which they had likely killed previously, as opposed to finding caches opportunistically. From 1992-93 to 1994-95, we monitored 37 caches--30 of these were made by coyotes (27 hares, 1 red squirrel, 1 flying squirrel, and one piece of a moose carcass), and 7 by lynx (all hares). Coyotes cached mostly entire hare carcasses (24 of 27 carcasses were of whole hares; mean percent of prey body weight cached = 99.3% (96.7-100.0%, 95% C.I.) per kill). Of the 27 caches of hares, coyotes returned to 14 of them (51.9%), and ate an average of 74.0% (37.0-97.7%, 95% C.I.) of each carcass retrieved. No caches made by coyotes were eaten by lynx, 2 were scavenged by red squirrels, and 11 were unused. Coyotes returned to caches an average of 56.1 ± 34.6 (S.D.) days (range 9-140 days) after they made the kill. Lynx cached mostly only portions of hare carcasses (1 of 7 carcasses were of whole hares; mean percent of prey body weight cached = 61.1% (32.9-85.7%, 95% C.I.) per kill). Of the 7 caches of hares, lynx returned to 6 of them (85.7%), and ate an average of 99.5% (96.1-99.7%, 95% C.I.) of each carcass retrieved. No caches made by lynx were eaten by coyotes, none were scavenged, and 1 was unused. Lynx returned to caches an average of 0.9 ± 0.5 (S.D.) days (range 0-2 days) after they made the kill. Most caches of hares made by coyotes were made early in the winter (85.2% of monitored caches were made in October and November), while those made by lynx were more evenly distributed over the winter (42.8% in October and November). Evidence from monitoring the survival of radio-collared hares supports this pattern. Most kills of radio-collared hares by coyotes in winter were made in October and November (77.3% of 141 kills; Fig. 3.12), while lynx kills occurred more evenly across the winter. Coyotes cached the entire carcasses of 37.0% of all hares they killed; they cached the carcasses of 41.4% of all kills made 93 Fig. 3.12. Percentages of depredated radio-collared hares killed by coyotes and lynx by month during winter from 1986 to 1995. Sample sizes (total number of mortalities) are given above bars. "Other" includes avian predators, red foxes, wolverines, wolves, and cases where the identity of the predator could not be determined. 94 110 92 65 153 48 62 60 = n Percent of All Radio-Tagged Hares Killed by Predators Oct. Dec. Feb. Apr. Month 95 in October and November, and none after January. Lynx cached the entire carcasses of only 1.6% of kills. Discussion Coyotes and lynx showed clear functional responses to changing densities of snowshoe hares. Kill rates of hares by coyotes increased at a faster rate than those by lynx during the cyclic increase in hare abundance. The functional response of coyotes is described equally well by linear and type-2 curves, while that of lynx is well-described by a type-2 curve. Both predators killed more hares than energetically required when hares were abundant. At lower densities of hares, coyotes preyed more on small mammals, and lynx on red squirrels. Coyotes frequently scavenged carcasses of hares, and many of these may be hares previously killed and cached by coyotes. In the following discussion, we will discuss these main points in more detail, and finish with an evaluation of potential biases in our estimates of functional responses. Functional Responses of Coyotes and Lynx While we had too few data points to accurately determine the shapes of the functional responses of coyotes and lynx, it is clear that the two predators responded differently to changing densities of hares in our study. Coyotes responded to increasing hare numbers, both in 1987-88 to 1988-89 and in 1994-95, with kill rates that were higher than those at comparable densities of hares during the cyclic decline (Fig. 3.7). Similarly, in a study in Utah, predation rates by coyotes on black-tailed jackrabbits were higher during a cyclic increase in jackrabbits than during the decline in prey numbers (Knowlton & Stoddart 1992). 96 Lynx showed the opposite pattern. This was not what we had expected based on the simple contrast of coyotes as generalists and lynx as hare specialists-typically, specialists respond with kill rates increasing at a faster rate at low densities of their preferred prey than do generalists. In Alberta, the functional response of lynx was stronger than that of coyotes at low numbers of hares (type-2 versus type-3 responses; Keith et al. 1977). The availability of alternative prey in the Yukon may be one reason for the difference in the relative responses of coyotes and lynx. The main prey of coyotes besides hares in our study area were small mammals (Fig. 3.2 to Fig. 3.4). Populations of small mammals were high in 1987-88 and from 1991-92 to 1993-94, and low in other years (Chapter 2). Based on scat analyses (Fig. 3.4), they were the most important alternative prey of coyotes during years that they were abundant. The availability of small mammals during winter, however, is limited by snow cover (Wells & Bekoff 1982; Halpin & Bissonette 1988) and affected by species-specific vulnerabilities-Microns, which are mostly confined to grassy meadows (which comprised only about 7% of our study area), appear to be much more vulnerable to predation than the more widespread Clethrionomys (Henttonen et al. 1987). Coyotes may have opportunistically taken advantage of increasing numbers of hares because they had few alternatives. Coyotes in Alberta fed mostly on the carrion of livestock during periods of low hare numbers (Todd, Keith & Fischer 1981; Todd & Keith 1983), an alternative food source not available to coyotes in our study area. Alternatively, lower kill rates by coyotes during cyclic declines, and, in our case, at the cyclic peak of hare abundance (1989-90; Fig. 3.7), may reflect an increased proportion of inexperienced young animals in the predator population, derived from the pulse of reproduction from 1988-89 to 1991-92 (Chapter 2). The main alternative prey of lynx during our study was red squirrels (Fig. 3.2 to Fig. 3.4). Populations of red squirrels were relatively stable during the 8 winters of this study (Chapter 2; Boutin et al. 1995).) Behavioural data suggest that lynx increased active foraging 97 for squirrels during the cyclic low in hare abundance (1992-93 and 1993-94; Chapter 4). Despite the increase in hare numbers in 1994, lynx which had mostly hunted squirrels during the previous 2 winters, continued to do so during the winter of 1994-95. While hunting success of lynx preying on hares declined from 1990-91 to 1994-95 (38.8% to 20.0%; Fig. 3.9), it increased from 1992-93 to 1994-95 (30.5% to 50.5%; Fig. 3.9) for lynx hunting red squirrels. These data suggest that lynx surviving into the low of the hare cycle became skilled at hunting squirrels, and may have been less plastic than coyotes in modifying their foraging behaviour when hare numbers started to increase again. There is ample evidence from the literature that foraging decisions are more strongly influenced by recent feeding choices than those made over a longer time-frame (Shettleworth, Reid & Plowright 1993), and this effect may be stronger in some predators than others. The maximum kill rates of hares by coyotes (2.3 hares/day/coyote) and lynx (1.2 hares/day/lynx) calculated in this study are higher than those reported in the literature, and those based on estimates of energetic requirements. The energetic needs of coyotes have been estimated to be from 0.7 to 0.9 hares per day (Keith et al. 1977; Litvaitis & Mautz 1980), and those of lynx to be from 0.4 to 0.5 hares per day (Nellis, Wetmore & Keith 1972; Aldama, Beltran & Delibes 1991). The maximum kill rate by coyotes in Alberta was estimated to be 0.7 hares/day, but this was based on the assumption that coyotes ate only what they needed energetically (Keith et al. 1977). Estimated kill rates by lynx have ranged from 0.5 hares/day in Newfoundland (Saunders 1963a) to 0.8 hares/day in Alberta (Brand, Keith & Fischer 1976; Keith et al. 1977) to about 1 hare/day in Nova Scotia (Parker 1981). The studies in both Alberta and Newfoundland based their estimates of kill rates on the assumption that the distance between 2 resting beds represented 1 day of travel by lynx. Several studies have noted though, that lynx may rest more than once per day (Haglund 1966; Parker 1981; this study), and estimates made with this assumption are likely underestimates. 98 Our estimates of kill rates of hares by lynx, and particularly coyotes, are likely conservative, due to potential biases in our field sampling (discussed later in this chapter). Apparently then, coyotes, and to a lesser extent, lynx, killed more than they energetically needed during years of high abundance of hares. Some of this excess food was cached by coyotes, and some of these caches were lost to scavengers or not retrieved. Storing excess food may guard against future periods of scarcity or losses of caches (review in Vander Wall 1990). While lynx seldom cached food, our data suggest some wastage (incomplete consumption) of prey by lynx during peak years of hare abundance (Fig. 3.10). Components of the Functional Responses of Coyotes and Lynx Our data suggest that the reactive distances of coyotes and lynx to hares increased once hares had declined to low abundance (Fig. 3.8), and they provide weaker evidence that the average hunting success of coyotes was higher during years of increasing densities of hares (Fig. 3.9). Increases in either of these parameters would lead to higher rates of successful search. Reactive distances of predators may be a function of hunger or environmental factors (Holling 1965; Abrams 1990; Bell 1991), and there is evidence from other field studies that satiated predators react to prey within smaller perceptual fields (e.g., Wood & Hand 1985). The average hunting success of coyotes preying on hares was 36.9% and of lynx 28.7% in our study-this was higher than success rates of 6-10% for coyotes reported in the literature (Ozoga & Harger 1966; Berg & Chesness 1978), but within the range of 19-57% reported for lynx (Saunders 1963a; Haglund 1966; Brand, Keith & Fischer 1976; Parker 1981; Major 1989). As in this study, hunting success of lynx was not related to hare density in Alberta (Brand, Keith & Fischer 1976). The apparently higher hunting success by coyotes in 1988-89 and 1994-95 may have been partly due to the absence of young, inexperienced coyotes in the population during these years. 99 We have no evidence that coyotes or lynx changed their rate of travel as densities of hares changed (we did not, however, have an adequate sample size of coyote travel rates at low hare densities). Several studies have noted that travel rates vary with density of prey (e.g., Smith 1974; Bell 1991), but Holling (1966) considered hunting speed to be adaptive for specific prey, and unlikely to vary. There have been no comparable direct measures of the travel speed of coyotes and lynx of which we are aware, but Witmer & DeCalesta (1986) noted that mean straight-line travel distances of coyotes were 1.7-times longer than those of sympatric bobcats in Oregon. Our measures of the activity patterns of coyotes and lynx showed that neither species increased their amount of active time in response to declining numbers of hares (Fig. 3.5). During times of food shortage, many arthropod predators decrease the amount of time they spend foraging (review in Bell 1991), but this is not true for all predators—lions do not change foraging time even during periods when cubs are starving (Schaller 1972). While Ward & Krebs (1985) found that lynx increased their daily movements during periods of low hare density in the Yukon, this was not the case in studies in Alberta (Brand, Keith & Fischer 1976). As discussed previously, "active" time is not likely equivalent to foraging time for lynx. Lynx increased their use of hunting beds during the decline and low phases of the hare cycle (Chapter 4), which likely represented an increase in the amount of time they were exposed to prey. We have no evidence that the handling time of coyotes and lynx preying on hares changed significantly during our study. We did note, however, that coyotes scavenged (and possibly cached) more often, and lynx consumed slightly less of hare carcasses, during periods of high hare abundance. Partial consumption of prey may decrease handling time (Abrams 1990). 100 Diets of Coyotes and Lynx Coyotes depended heavily on snowshoe hares during most years of our study, and preyed on small mammals during winters when voles were abundant (Fig. 3.2 to Fig. 3.4), which coincided with years of low hare densities. Although not quantified here, most small mammals killed were Microtus (unpublished data). We likely underestimated the number of kills of small mammals from our snow-tracking, because while it was easy to distinguish attempted captures of voles by coyotes "mousing", it was often difficult to judge whether they were successful. The results of our scat analyses suggest that we did miss many kills of voles. In Alberta (in the only other comparable study of diets of coyotes during a cycle in abundance of hares), the proportion of hares in the diets of coyotes ranged from 0% to 77% during cyclic lows and highs, respectively (Nellis & Keith 1976; Todd, Keith & Fischer 1981; Todd & Keith 1983). Coyotes on the Kenai Peninsula in Alaska fed mostly on carrion of wolf-killed or starved moose in winter, during a period of low hare numbers (Staples 1995). The main alternative prey of coyotes in both these other studies was either not available (livestock carcasses) or available in low numbers (moose carcasses) for coyotes in our study area. Coyotes were evidently not as skilled at catching red squirrels as lynx, since they made few kills of this abundant potential prey. The preponderance of hares in the diets of lynx in our study was consistent with virtually all studies of North American lynx (e.g., Saunders 1963a; Van Zyll de Jong 1966; Brand, Keith & Fischer 1976; Brand & Keith 1979; Parker et al. 1983; Staples 1995). Extensive use of red squirrels as the main alternative prey has not been noted previously. Only two other studies, one in Washington (Koehler 1990) and one in Alaska (Staples 1995), have recorded squirrels as more than minor components of the diets of lynx; in both these areas, populations of hares were low. Our behavioural data show that lynx actively hunted for squirrels from 1992-93 to 1994-95, and may have switched (sensu Murdoch 1969) from 101 preying on hares to preying on red squirrels during these years. Our scat analyses suggest that lynx also preyed on small mammals in significant numbers during winters when voles were abundant (Fig. 3.4). Neither coyotes nor lynx consumed a substantial number of grouse, despite the fact that they were relatively abundant in our study area in 1989 and 1990 (Boutin et al. 1995). Ruffed grouse comprised up to an estimated 12% of the diets of lynx during the cyclic low in the Alberta study (Brand, Keith & Fischer 1976). The contrast between coyotes and lynx as generalists versus hare specialists is obviously not appropriate at the local scale in our study area. Where few alternative prey are available, coyotes may be as or more dependent on hares than are lynx. Scavenging and Caching by Coyotes and Lynx The evidence from our monitoring of radio-collared hares and caches suggests that many of the instances of scavenging of hares by coyotes, which we noted while snow-tracking, may have been of their own previous kills. Lynx seldom cached prey, and when they did, they usually retrieved it the next day. Coyotes, in contrast, returned to about half of their caches, as long as 4-1/2 months after they were made. The wires and radios which we attached to the caches may have discouraged some animals from retrieving their caches, so this return rate is a minimum estimate. This nevertheless suggests that coyotes have a long-term spatial memory that allows them to remember the location of caches even after they have been covered by a half-meter of snow (review of spatial memory by other species in Vander Wall 1990). This corresponds with our field observations of coyotes deviating well away from their travel routes (sometimes over 0.5 km) to retrieve caches. Coyotes killed and cached most hares early in the winter (Fig. 3.12). This is also the time when the snow is shallow, and travel is likely easier for them. Hares may also be more 102 vulnerable at this time of year due to a higher proportion of young animals in their population. How Accurate are Our Estimates of Kill Rates? There are a number of potential biases that could have affected our estimates of kill rates of hares by coyotes and lynx: 1) In some winters, snow conditions were poor for snow-tracking in October, and so we could have missed some tracking during the time of year when kill rates by coyotes were highest. 2) During the winters of peak abundance of coyotes and hares, the abundance of tracks often made snow-tracking in prime habitats difficult, due to confusion caused by criss-crossing and obliterated tracks. Tracks were easier to follow when they were away from these habitats. We therefore could have sampled less intensively in the best habitats where more kills would have been located. We attempted to avoid this by tracking right after fresh snows when tracks were least confusing. 3) Coyotes and lynx often circled many times in the areas of kills, and made mazes of tracks that were confusing to follow. This was particularly true of coyotes. We may have missed some kills because we were unable to follow them to kills. We attempted to avoid this by searching very carefully in any areas with frequent circling. 4) Coyotes were very skilled at hiding caches of hares with minimal deviation from their trail and little noticeable disturbance to the snow. We may have missed some of these, although we searched very carefully around any areas where coyotes broke stride. 5) Coyotes and lynx which were the most successful hunters may have moved less than unsuccessful predators, and therefore crossed our trails, where we started snow-tracking, less frequently. 103 6) We may have overestimated the amount of "active" time of radio-collared coyotes and lynx. Activities such as eating or grooming, in which the animal frequently moved its head up and down, were probably classified as "active" in our analyses, even though the animal was not traveling. This bias would cause us to underestimate the total amount of time represented by our tracks, and therefore overestimate kill rates. In Alberta, Bowen (1982) estimated that coyotes were active 50% of the time during winter, and Bekoff and Wells (1981, 1986) estimated that coyotes spent 46% of their time traveling and hunting during winter—these are quite close to our estimates of 44-50% time active for coyotes. In summary, most of the potential biases in our method of estimating kill rates would cause us to underestimate the true kill rates. We therefore consider the calculated kill rates from this study to be conservative, particularly for coyotes. Conclusions Coyotes and lynx both showed clear functional responses to changes in the densities of snowshoe hares during their cyclic fluctuation in abundance. Coyotes responded with higher kill rates during the increase in hare abundance than did lynx, and their functional response is described equally well by linear and type-2 curves. Kill rates of hares by coyotes were lower during the cyclic decline in hare densities than during the increase. Coyotes killed the maximum number of hares per coyote one year before the peak in hare abundance~the maximum observed kill rate was 2.3 hares/day, over double their estimated energetic needs. Small mammals, mostly Microtus, were the most important alternative prey of coyotes. Coyotes killed more hares early in the winter, and cached many of these for later retrieval. Kill rates of hares by lynx changed almost 4-fold during the cyclic fluctuation. Their functional response is well-described by a type-2 curve. Kill rates by lynx were lower during 104 the increase in density of hares than during the cyclic decline. Lynx killed the maximum number of hares per lynx one year after the peak in hare abundance~the maximum observed kill rate was 1.2 hares/day, which is also more than double their estimated energetic requirements. The main alternative prey of lynx during periods of low abundance of hares was red squirrels. Lynx that survived into the low of the hare cycle apparently became more skilled at catching squirrels, and continued to actively hunt them even when numbers of hares began to climb again. 105 CHAPTER 4 BEHAVIOURAL RESPONSES OF COYOTES AND LYNX TO THE SNOWSHOE HARE CYCLE Introduction Populations of snowshoe hares undergo regular cycles in abundance throughout the northern boreal forest in North America, with amplitudes over two orders of magnitude, and 8-11 years between cyclic peaks (Elton & Nicholson 1942; Keith 1990). These cycles have major effects on predators of hares, as well as on other herbivores, in the relatively simple boreal ecosystem (Finerty 1980). This chapter considers the behavioural responses of coyotes and lynx, the two most important mammalian predators of hares throughout much of their range, to the population cycle. Predators may respond to fluctuating abundance of prey with demographic (e.g., changes in reproduction or survival) and behavioural (e.g., switching prey or habitats) adjustments to prey density. Demographic responses and migratory movements relative to prey density are termed "numerical responses", while responses leading to changes in kill rates with density of prey are "functional responses" (Solomon 1949). The total impact of predators on their prey are determined by their combined numerical and functional responses (review in Murdoch and Oaten 1975). The shapes and timing of the numerical and functional responses of predators to their prey are therefore critical to determining the effects of predation. Holling (1959b) described three basic forms of functional responses of predators to changes in density of prey. Type-1 responses describe linearly increasing kill rates with prey density up to a threshold, above which the rate is constant. Type-2 responses describe predation increasing at a monotonically decreasing rate with prey density, up to an 106 asymptote, and type-3 responses are sigmoidal in shape. Type-2 responses may be typical of specialist predators, predators with few alternative prey, or they may arise due to adaptive adjustment of search rates and time to the benefits and costs of foraging (Abrams 1990, 1992a). Type-3 responses are more typical of generalist predators, and they may be the result from a number of mechanisms: 1) predators learning to recognize, capture, or handle prey better with increasing prey density (Holling 1959b), 2) predators switching among prey types, habitats, or foraging tactics (Royama 1970; Murdoch & Oaten 1975; Akre & Johnson 1979), 3) adaptive variation in foraging rates (Holling 1966; Hassell, Lawton & Beddington 1977; Dunbrack & Giguere 1987), and 4) changes in the behaviour of prey, or prey having a refuge below a fixed density (Murdoch & Oaten 1975; Taylor 1984). Functional responses of predators thus can result from increasing encounter rates with prey at higher densities of prey, density-dependent changes in the behaviour of prey, or changes in the behaviour of predators. Any of these can lead to changes in the shape of the functional response, and on predator-prey dynamics. "Type-3", or sigmoidal, functional responses have received a great deal of attention in theoretical models, because they describe density-dependent, and potentially regulatory predation over some range of prey density (e.g., Hassell & Comins 1978; Nunney 1980). Abrams (1992b) described how adaptive changes in foraging by predators, in response to balancing costs and benefits, could have major effects on predator-prey interactions. In Chapter 3, we examined the functional responses of coyotes and lynx, and changes in the "components" of these responses (rate of successful search, time exposed to prey, and handling time), to fluctuating density of hares. In this chapter, we will examine behavioural responses-in particular, switching among prey, habitat types, and hunting tactics-that accompanied and affected the functional responses of these predators. As the density of a prey species changes, so often does its relative density to alternative prey. Predators may "switch" between prey species based on relative prey abundances. 107 "Switching" is defined as feeding on a prey species disproportionately less when its relative abundance to other prey is low, and disproportionately more when it is high (Murdoch 1969). A number of mechanisms for switching have been proposed: 1) predators may develop "search images", or learn to "see better" the prey as it becomes relatively more abundant (Tinbergen 1960; Lawrence & Allen 1983), 2) predators may switch habitat types, which leads to switches in prey (e.g., Royama 1970), or 3) predators may change foraging tactics as relative densities of prey change (e.g., Lawton, Beddington & Bonser 1974). There is ample evidence that switching among prey species does occur (e.g., Murdoch & Oaten 1975; Akre & Johnson 1979). Laboratory and theoretical investigations suggest that switching usually, but not always, leads to type-3 functional responses by predators (Murdoch & Oaten 1975), which can stabilize predator-prey interactions (Oaten & Murdoch 1975; Hassell 1979). Switching among habitat types by predators may also occur in response to changes in the relative frequency of prey, and the resulting changes in relative profitabilities of foraging in different habitats (Royama 1970; Murdoch & Oaten 1975). Concentration of predators in habitat patches with high density of prey can lead to type-3 functional responses (Murdoch & Oaten 1975). Such aggregation can be considered as a functional response or a numerical response, depending on the scale of movement of predators, but the effect on prey dynamics may be similar (Solomon 1949; Murdoch & Oaten 1975; Hanski, Hansson & Henttonen 1991). Switching foraging tactics in response to changes in the relative frequency of prey has also been documented in a number of studies (Lawton, Beddington & Bonser 1974; Davies 1977; Akre & Johnson 1979; Formanowicz & Bradley 1987). Changing tactics may involve switches between active searching and ambush, differences in travel patterns, or changes in the size of foraging groups. Keith et al. (1977) suggested that coyotes had a type-3 functional response to changes in the density of snowshoe hares due to a switch to frequent travel along trails of hares as they increased in number. 108 Animals vary in their foraging strategies, and some are more versatile than others in the range of prey types, habitats, and foraging tactics that they will use (review in MacNally 1995). "Generalists" are more plastic in their behaviour than "specialists", and may more readily switch among foraging strategies. Coyotes are often considered prototypical generalists. Over their large and expanding range in North America, they are adapted to a wide variety of habitats, climates, and foods (see papers in Bekoff 1978). Lynx, in contrast, are usually considered prototypical specialists on snowshoe hares, and virtually all studies of their food habits have shown hares to be their predominant prey (Saunders 1963a; Brand & Keith 1979; Parker et al. 1983; Ward & Krebs 1985). Coyotes usually stalk and rush prey larger than small mammals, while lynx may either stalk or ambush their prey (Saunders 1963a, 1963b; Wells & Bekoff 1982; Murray et al. 1995) In Chapter 3, we showed that both coyotes and lynx showed clear functional responses to changing densities of snowshoe hares in the Yukon. This was accompanied by changes in the reactive distance of both predators to hares. In this chapter, we examine changes in prey choice, use of habitat patches, and foraging tactics. Objectives, Hypotheses and Predictions The main objective of this chapter is to investigate and contrast the behavioural responses of coyotes and lynx to changing densities of snowshoe hares. Specifically, we will look at evidence for prey-switching, changes in habitat use, and changes in foraging tactics, including hunting technique, use of trails, and group hunting. As the null hypothesis, we postulate that coyotes and lynx feed on prey in proportion to their relative abundances (i.e., no preferences, no switching; Murdoch 1969). Alternatively, 109 coyotes may be more flexible in their use of prey and exhibit switching as the relative availabilities of prey change, while lynx may always concentrate their hunting on hares. Study Area This study was conducted in the southwest Yukon Territory, Canada (60° 57' N, 138° 12' W), in a broad glacial valley approximately 350 km2 in area (Fig. *2.1). The study area is described in Chapter 2. Methods This study was conducted as a part of the Kluane Boreal Forest Ecosystem Project, an experimental study of the vertebrate food web in the northern forests (Krebs et al. 1995). These results summarize the efforts of a great many biologists, technicians, and students who have worked with the Kluane Project. Field work was started in 1986, and continued through the winter of 1994-95. With the exception of some scat collection, live-trapping and radio telemetry, field work was conducted during the winter months, from October to April. Estimation of Prey Densities We estimated densities of snowshoe hares, red squirrels, and small mammals by live-trapping, as described in Chapter 2. 110 Snow-tracking of Coyotes and Lynx We followed the tracks of coyotes and lynx each winter from 1987-88 to 1994-95, beginning as soon as there was enough snow, usually in mid- to late October, and finishing at the end of March. Nineteen different observers snow-tracked predators during the 8 winters of study. This was the main method that we used for determining species and frequencies of kills and attempted kills, frequencies of beds, use of habitats, and use of trails, as well as other data discussed in Chapter 3. We chose and followed tracks of coyotes and lynx as described in Chapter 3. We recorded the prey species and characteristics of chases and feeding for all kills and attempted kills. All beds were classified as crouches (or hunting beds), "short" beds (where the predator had laid down but did not stay long enough to melt the snow), or resting beds. We kept a continuous record of the habitat through which the animal was traveling. We characterized the overstory by cover (subjectively classed as <5%, 5-25%, 25-50%, 50-75%, or 75-100%), dominant species (white spruce, trembling aspen, or balsam poplar), and age (immature or mature). We also recorded when the animals we were following were traveling on trails made by other predators, hares, snowmobiles, or snowshoes. At high densities of hares, it became very difficult to keep track of every time a predator followed or left trails of hares, so we changed our protocol to recording whether the coyote or lynx was on or off a hare trail at each 100th step. For subsequent analyses, we converted all trail-use data to "on" or "off at 100-m intervals. We analyzed tracking data using the MGLH (Multivariate General Linear Hypothesis) procedure in SYSTAT (Wilkinson 1990), and considered P < 0.05 the criterion for rejecting null hypotheses. Availabilities and proportions of diet by biomass were calculated by multiplying numbers of prey present or killed by 1500 g for hares, 250 g for red squirrels, and 20 g for small mammals. Ill Analysis of Prey Switching by Coyotes and Lynx We used a graphical analysis similar to that suggested by Murdoch (1969) to investigate evidence for prey switching by coyotes and lynx. Analysis of switching is typically carried out by comparing the percent of a given prey species in the diet of a predator, relative to its availability in the environment (Murdoch 1969). The null hypothesis of no preference and no switching predicts that use and availability should be equal (i.e., when plotted, the points should fall on a straight line) at all relative availabilities of the prey. Predators may have "innate preferences" for certain prey species though, in which case the null hypothesis for no switching predicts that, when percent use is plotted against percent availability, the points should all lie on a curve (convex for preferred prey, concave for those not preferred) determined by the degree of preference (Murdoch 1969). Alternatively, if points lie below the null-model curve at low relative availabilities (indicating disproportionately low use), and above the curve at high availability (indicating disproportionately high use), this is taken as evidence of prey switching. Most analyses of switching have been carried out in the laboratory or experimentally-controlled situations, using similar-sized prey (review in Murdoch and Oaten 1975). Relative availability and use have been determined, in these cases, by relative frequencies of the prey. In our field study though, the main available prey-hares, red squirrels, and small mammals-were of very different body sizes. Considering their relative frequencies as equivalent to the relative amounts of food available to predators was clearly inappropriate. A further difficulty is presented by the fact that abundance and availability were almost certainly not equivalent in the field-red squirrels have refuges from predation in trees and under frozen ground, and small mammals have subnivean refuges. 112 Our analysis of switching was designed to take these potential biases into account. We calculated preference indices for hares, squirrels and small mammals for each predator in each winter using Manly's alpha (Manly, Miller & Cook 1972), with the proportions of prey available and in diets expressed as biomass rather than frequencies. Manly's alpha measures the probability that an individual prey item is selected from a prey class when all prey species are equally available (Krebs 1989). We took the means of preference indices for the 8 winters of this study as the overall preference index for each prey group. The ratio of the preference index for a given species to those of the other prey was then used as the "proportionality constant" to calculate null-model curves to test for switching (equation 2 in Murdoch 1969). While this analysis is not strictly equivalent to that proposed by Murdoch (1969), it nonetheless provides a graphical method of comparing relative use of prey to relative availabilities, to test for prey switching. Analysis of Habitat Use by Coyotes and Lynx We collected data on habitat use by coyotes and lynx continuously along their trails. However, it is obvious that the habitat we measured at each step cannot be considered an independent sample, since it is likely to be the same as that at an adjacent step simply by virtue of being in the same habitat patch. Before analyzing these data, therefore, we had to first determine the appropriate interval along trails at which our observations of habitat were independent. We used data from parallel linear transects run through the study area at 2-km intervals, along which habitat was classified in the same manner as when snow-tracking. From these data, we calculated the set of "patch lengths" for each overstory type. We then used a bootstrapping procedure in which each set of patch lengths was randomly sampled 5000 times, and for each random sample, we calculated the 95% quantile (patch length longer 113 than all but 5% of measured patch lengths). We took the median of these for each overstory type-these ranged from 79-188 m for different habitat types. We therefore chose to consider our measurements of habitat each 200 m along our trail to be independent samples in subsequent analyses. A second potential bias with our habitat data became apparent in early analyses. Over the course of our study, 22 different observers snow-tracked predators. Preliminary analyses of the data indicated that there were differences among observers in their subjective classification of overstory cover. We therefore conducted separate analyses of habitat use, using t-tests to compare between years and between species, for each observer. We then combined the statistical results of the tests from different observers to evaluate overall statistical significance using "meta-analyses" (Arnqvist & Wooster 1995); specifically, we used the method of adding Z's (Rosenthal 1978). Track Counts of Coyotes and Lynx around Experimental Grids We conducted counts of the tracks of coyotes and lynx around the perimeters of control and food-addition grids, established by the Kluane Project, in order to determine if predators were concentrating their activities in the pockets of high hare density created by the food addition. Beginning in the winter of 1988-89, we counted all tracks of predators crossing a transect around each of 2 food addition grids and 1-2 control grids, 1-5 days after fresh snowfalls. We compared track counts between control and food-addition grids using ANOVAs with log-transformed data, controlling for number of days after snowfall. 114 Results Prey Populations Populations of snowshoe hares peaked between 1988 and 1990 at approximately 2/ha (Fig. 4.1; Boutin et al. 1995). There were locally abundant pockets of hares through the fall of 1991 (Boutin et al. 1995), and populations declined to very low numbers by the end of winter 1992-93. Densities of hares started to increase again in 1994. Depending on whether fall or late winter estimates of hare numbers are used, the cyclic amplitude was 26-44-fold (Boutin et al. 1995). Population trends of other potential prey species varied. Numbers of red squirrels stayed relatively stable, with a slight increase during the last two winters of the study (Fig. 4.1). Populations of small mammals fluctuated with a 10-50-fold amplitude, with the highest numbers occurring from 1991 to 1993 (Fig. 4.1). There were few other alternative prey species. Arctic ground squirrels hibernate, and were therefore not available to predators during the winter. Spruce grouse, ruffed grouse, and ptarmigan were present (Boutin et al. 1995), but were seldom utilized by predators (2% of all kills from 1987 to 1995; Chapter 3). Snowshoe hares were by far the most abundant potential food source during winter for predators from 1987-88 to 1992-93, comprising 63-81% of the total biomass of the main prey species. Red squirrels were the second largest food source during these years. During the last three winters of the study (1992-93 to 1994-95), however, once hare numbers had crashed, red squirrels comprised 58-72% of the total biomass. Small mammals represented less than 10% of the total biomass of prey in all winters except 1992-93 (14%). 115 Fig. 4.1. Estimated densities of snowshoe hares (means of autumn and late winter estimates from live-trapping on 1-3 60-ha grids), red squirrels (means of late summer and spring estimates from live-trapping on 2-3 8-ha grids), and small mammals (means of late summer and spring estimates from live-trapping on two 2.8-ha grids for Clethrionomys rutilus, and on two 1.5-ha grids for Microtus oeconomus and M. pennsylvanicus) from 1986 to 1995 in the southwest Yukon. 116 200 Hares per 100 100 ha 0 400 Red Squirrels per 100 ha 200 -0 _j i i i i i i i_ Winter 117 Prey Switching by Coyotes and Lynx Coyotes. We followed the tracks of coyotes for 1897 km during the 8 winters from 1987-88 to 1994-95 (mean per winter, 237 ± 100 (S.D.) km). We found 189 kills by coyotes, 47.1% of which were hares, 13.2% red squirrels, and 37.6% small mammals (see Chapter 3). Coyotes killed mostly hares from 1987-88 to 1991-92, plus some red squirrels and small mammals from 1989-90 to 1991-92. Small mammals comprised most kills in 1992-93 and 1993-94, while hares and red squirrels were killed more in 1994-95. In terms of biomass, the diets of coyotes were comprised largely of hares in all winters (Fig. 4.2). Small mammals made up a maximum of 9.5% in 1993-94, a winter of high vole density, and red squirrels made up maximum of 19.9% in 1994-95. Coyotes showed a strong preference for hares throughout the 8 winters of this study (a = 0.88 ± 0.11 (S.D.), range 0.73-1.00). The null model for switching is therefore a convex curve for hares, and concave curves for red squirrels and small mammals (Fig. 4.2). Based on these data, there is no evidence that coyotes switched among prey species (i.e., no consistent pattern of data points below the null-model curves at low relative availabilities, and above at high availabilities), as their relative abundances changed. Lynx. We followed the tracks of lynx for 2232 km during the 8 winters from 1987-88 to 1994-95 (mean per winter, 279 ± 85 (S.D.) km). We found 502 kills by lynx, 50.2% of which were hares, 34.7% red squirrels, and 11.0% small mammals (see Chapter 3). Lynx killed mostly hares from 1987-88 to 1991-92, plus some red squirrels in 1987-88. Red squirrels comprised most kills from 1992-93 to 1994-95, with some small mammals in 1993-94. In terms of biomass, the diets of lynx were comprised largely of hares from 1987-88 to 1991-92, but red squirrels became increasingly important from 1992-93 to 1994-95 (20.4-43.9%; Fig. 4.3). Small mammals made up a negligible proportion of the diet in all winters. 118 Fig. 4.2. Graphical tests for prey switching by coyotes preying on snowshoe hares, red squirrels, and small mammals from 1987-88 to 1994-95. The curves are the null hypotheses of no switching, taking into account the "innate" dietary preferences of coyotes. Proportions of prey in the diets of coyotes are based on the observed kills from snow-tracking. Evidence of switching is inferred when data points lie below the null-model curve at relatively low prey availability, and above it at high availability. 119 Percent Biomass in Diet Percent Biomass in Diet 20 r Percent 1 5 Biomass 10- • in Diet c 5 -° 0 5 10 15 20 Percent Biomass Available 120 Fig. 4.3. Graphical tests for prey switching by lynx preying on snowshoe hares, red squirrels, and small mammals from 1987-88 to 1994-95. The curves are the null hypotheses of no switching, taking into account the "innate" dietary preferences of lynx. Proportions of prey in the diets of lynx are based on the observed kills from snow-tracking. Evidence of switching is inferred when data points lie below the null-model curve at relatively low prey availability, and above it at high availability. 121 Percent Biomass in Diet Percent Biomass in Diet 20 r 15-Percent Biomass 10 -in Diet c o - • 0 5 10 15 20 Percent Biomass Available 122 Lynx also showed a strong preference for hares throughout the 8 winters of this study (a = 0.90 ± 0.12 (S.D.), range 0.65-1.00). Preference was very high from 1987-88 to 1992-93 (range 0.92-1.00), but declined to 0.79 in 1993-94 and 0.65 in 1994-95. The null model for switching is therefore a convex curve for hares, and concave curves for red squirrels and small mammals (Fig. 4.3). Based on these data, there is some evidence that lynx switched from hares to red squirrels during the last 3 winters of this study (1992-93 to 1994-95). Preference for small mammals was very low in all years. Use of Habitat by Coyotes, Lynx and Hares Coyotes showed a general pattern of using progressively more dense cover from the peak year of hare abundance (1989-90) into the early cyclic decline (Fig. 4.4). They used the most dense cover in 1991-92, the second year of decline in hare abundance, and then more open habitat in each of the next two winters (1992-93 and 1993-94). There was an increase the density of cover used again in 1994-95. Lynx showed a pattern of habitat use similar to that of coyotes—they used progressively more dense cover from 1988-89 (late increase phase of the cycle) to the early cyclic decline (Fig. 4.4). They used the most dense cover in 1991-92, the second year of decline in hare abundance, and then showed a general trend towards using more open habitat during the last 3 winters of the study (1992-93 to 1994-95). The pattern of habitat use by coyotes and lynx described above roughly paralleled that of hares from 1988-89 (late increase of cycle) to 1992-93 (late decline; Fig. 4.4). Hares used progressively more dense cover to a maximum in 1991-92, and then more open cover the next winter. Hares were consistently in more dense habitat than either coyotes or lynx (Fig. 4.5). In summary, these data show that coyotes and lynx did change their patterns of habitat use over the course of the cycle in numbers of hares, and that these changes roughly 123 Fig. 4.4. Use of habitat (by overstory density) by coyotes, lynx, and snowshoe hares in the southwest Yukon, from snow-tracking data, 1987-88 to 1994-95. Each symbol represents a different observer who tracked in two consecutive winters. Differences in use of habitat between consecutive years were tested for each observer individually, and the results of these tests were combined in meta-analysis to analyze overall trends between the two years. "+" signs indicate overall statistically significant increases in the density of cover used between years, and "-" signs indicate significant decreases. 124 80 r + + Percent Overstory 40 Cover 0 Winter 125 Fig. 4.5. Differences in use of habitat (by overstory density) by coyotes, lynx, and snowshoe hares in the southwest Yukon, from snow-tracking data, 1987-88 to 1994-95. Each data point represents a different observer—data points above the line at 0% indicate the species on top in each graph used denser cover; those below the line show use of denser cover by the species on bottom. Differences in use of habitat between species were tested for each observer individually, and the results of these tests were combined in meta-analysis to analyze overall trends between the two species. "-" signs indicate overall statistically significant use of less dense cover by the species on top of each graph. 126 Percent Difference in Overstory Cover 50 0 50 -i i i_ Percent Difference in Overstory Cover 50 r 0 50 ! I I I _ i _ 50 Percent Difference in Overstory Cover 0 50 i i ! t 87- 89- 91- 93-88 90 92 94 Winter 127 paralleled changes in habitat use by hares. Coyotes generally used more open habitat than lynx, particularly during the cyclic decline (Fig. 4.5). Additional evidence that predators concentrated their activity in habitats with high numbers of hares comes from our track counts around experimental hare grids. Food addition grids had roughly 3-times the .density of hares as controls during the peak and decline in the hare cycle (Krebs et al. 1995). Coyotes consistently used food addition grids more intensively than controls from 1990-91 to 1993-94 (ANOVA, Effect of Treatment, F = 19.46, d.f. = 1,150, P = 0.000; Fig. 4.6; too few data were gathered in 1988-89 and 1989-90 for meaningful analyses), and lynx used them more from 1990-91 to 1992-93 (ANOVA, F = 4.52, d.f. = 1,150, P = 0.035; Fig. 4.6). Foraging Behaviour by Coyotes and Lynx Hunting Tactics. While lynx used both ambush and stalking to hunt prey, coyotes seldom made hunting beds (Fig. 4.7). There was a large increase in the frequency of hunting beds by lynx during the decline in hare abundance (1990-91 to 1992-93), and they continued to use them frequently in 1993-94 and 1994-95. Coyotes made few crouches during any winters. Lynx initiated progressively more chases of hares from hunting beds during the decline and low years of the hare cycle (1990-91 to 1993-94; Fig. 4.8). (This is based on the criterion of chases actually starting from beds, and does not include cases where lynx may have seen hares from beds, but stalked them before initiating a chase). From 1992-93 to 1994-95, 20-30% of all chases of red squirrels by lynx were made from beds (Fig. 4.8); sample sizes are too low before this time period for meaningful analyses. Hunting success of lynx chasing hares was approximately the same whether they were chased directly from hunting beds or not (Fig. 4.9), while success was generally lower from beds for lynx chasing squirrels (Fig. 4.9). 128 Fig. 4.6. Use of food addition (high density of hares) and control grids by coyotes and lynx at Kluane in winter from 1990-91 to 1994-95. Error bars are not presented for clarity; mean C.V. = 179%. 129 5 Number 4 of 3 Tracks per Grid 2 per Day 1 0 -•— Control Grids -*- - Food Grids 5 Number 4 of 3 Tracks per Grid 2 per Day 1 0 90-91 92-93 X -•— Control Grids -•- - Food Grids 94-95 Winter 130 Fig. 4.7. Frequency of crouches (hunting beds) along trails of coyotes and lynx in the southwest Yukon, from 1987-88 to 1994-95. 131 Winter 132 Fig. 4.8. Percentages of chases of hares and red squirrels by lynx that initiated from hunting beds in the southwest Yukon, during winter, 1987-88 to 1994-95. Successful and unsuccessful chases are plotted separately; sample sizes (number of chases) are given above graphs. Note very low sample sizes for chases of squirrels before 1992-93. 133 n = 18 30 27 36 34 22 15 11 —•— Kills n = 44 53 137 97 97 89 66 44 - # - Attempted kills 40 r n = 5 0 0 0 0 45 61 54 m Kills n = 6 2 o i 2 105 121 51 Attempted kills 40 r 87- 89- 91- 93-88 90 92 94 Winter 134 Fig. 4.9. Hunting success rates (% of all chases that were successful) of lynx chasing hares and red squirrels from hunting beds, compared to success rates not initiated from beds, from 1987-88 to 1994-95. Sample sizes (number of chases) are given above the bars. Note very low sample sizes for chases of squirrels before 1992-93. 135 n = 56 79 156 no i n 87 56 38 - - * - Not from bed n = 6 4 8 23 20 24 20 17 —•— From bed ioo r Success Rate n = 9 2 o i 2 107 140 86 - # - Not from bed n = l o o o o 43 42 19 —•— From bed ioo r Success Rate 87- 89- 91- 93-88 90 92 94 Winter 136 During winters when voles were abundant, coyotes hunted them with a "mousing" foraging tactic, mostly in open habitat (Wells & Bekoff 1982). Coyotes moved through grassy areas slowly, listening for subnivean prey; when located, coyotes pounced and tried to pin then with their forepaws. Frequency of mousing was very low (0.00-0.05 attempts per 10 km trail) during winters of high and declining abundance of hares (1988-89 to 1991-92; except for 1991-92, these also corresponded with winters of low densities of voles (Fig. 4.1)). During winters of higher vole and lower hare abundance, coyotes hunted voles more actively (0.70-1.91 attempts per 10 km trail in 1987-88, 1992-93, and 1994-95), and in 1993-94, they spent a great deal of effort "mousing" (13.98 attempts per 10 km trail). IIsejsfTjcails. Coyotes spent more time than did lynx on the trails of other predators (ANOVA with arcsine-transformed data, F = 5.82, d.f. = 1,1617, P = 0.016) and human-made trails (ANOVA, F = 8.29, d.f. = 1,1617, P = 0.004; Fig. 4.10). Both lynx and coyotes spent more time on hare trails during the peak and early decline phases of the cycle (Fig. 4.10). Lynx used hare trails more than did coyotes, particularly during periods of declining and low numbers of hares (ANOVA, F = 6.63, d.f. = 1,2150, P = 0.010). GjmipJHunting. Both coyotes and lynx sometimes traveled in and hunted in groups. To investigate whether group hunting conferred advantages of higher food intake to predators, we calculated kill rates of hares by coyotes and lynx in different group sizes (we limited our analyses to group sizes which we tracked for more than 10 km in a given winter), as in Chapter 3. We adjusted these by group size to calculate per-individual kill rates. Coyotes hunted in groups of 1-3; groups larger than 2 were tracked only in 1990-91 and 1991-92. We were unable to determine the ages of animals in these groups. The relationship between group size and per-individual kill rates was variable (Fig. 4.11). Larger groups generally killed more hares, but this usually translated in to fewer hares per individual, although this was not statistically significant (ANOVA, F = 11.31, d.f. = 1,2, P = 0.078). 137 Fig. 4.10. Percentages of trails of lynx and coyotes on trails made by other predators, snowshoe hares, and humans, in the southwest Yukon during winter, 1987-88 to 1994-95. 138 Percent of Track on Trails of Other Predators Percent of Track on Trails of Hares 40 30 20 10 0 _i i i i i_ Percent of Track on Human-made Trails 25 20 15 10 5 0 _i i i i i i_ 87- 89- 91- 93-88 90 92 94 Winter 139 Fig. 4.11. Kill rates per individual relative to group size of coyotes and lynx in the southwest Yukon during winter, 1987-88 to 1994-95. Numbers next to the data points indicate group sizes; data points within the same winter are connected by lines. 140 Kills of Hares per Day per Individual 2.5 Kills of 2.0 Hares 1.5 per Day per 1.0 Individual 0.5 - ft 0 i 87 2 f 88 89- 91- 93-90 92 94 Winter 141 Lynx hunted in groups of 1-5. All of these were family groups (females with kittens) up to the winter of 1991-92. In the winters from 1991-92 to 1993-94, however, we observed few family groups (none after 1991-92), but rather several groups of adults hunting together for the first time. As with coyotes, larger groups of lynx generally killed hares more frequently, but per individual kill rates were lower from 1987-88 to 1991-92 (Fig. 4.11). In 1992-93 and 1993-94, there was a suggestion that larger adult groups may have benefited individuals in them with equal or higher kill rates, but sample sizes were small. The overall relationship between group size and individual kill rates was statistically insignificant (ANOVA, F = 0.82, d.f. = 1,6, P = 0.401). Discussion Coyotes and lynx showed strong preferences for snowshoe hares throughout the cycle. Our data suggest that lynx switched from preying on hares to red squirrels during the winters of low and early increasing densities of hares. Habitat use by both predators roughly paralleled that of hares, and both concentrated their hunting efforts in areas of high hare density. As numbers of hares declined, lynx increasingly used hunting beds for ambushing prey. Coyotes and lynx frequently traveled along the trails of hares while hunting, and coyotes, in particular, often used trails made by other predators and humans. We observed lynx hunting in adult groups during the late decline phase of the cycle, and this may have led to higher kill rates. In the following discussion, we will examine these main points above in more detail. 142 Prey Switching by Coyotes and Lynx "Switching", as defined by Murdoch (1969), strictly refers to disproportionately low and high use of prey at low and high relative availabilities, respectively, where availabilities are determined by the relative abundances of prey. In this chapter, we have used relative biomasses rather than abundances because of the large differences in body sizes of available prey. Our data show a clear increase in the relative use of red squirrels by lynx when squirrels represented more than about 55% of the available biomass of prey; virtually no squirrels were killed when they represented 30% or less of the biomass (Fig. 4.2). We would argue that this represents a definite "switch" in prey by lynx to red squirrels during periods of low densities of hares. Hares remained the preferred prey throughout the cycle, but more hunting effort was devoted to pursuing squirrels during these winters. Prey switching by the "specialist" lynx, rather than the "generalist" coyote was the opposite of what we expected. Coyotes fed heavily on hares during all winters of this study. As discussed in Chapters 2 and 3 though, coyotes likely had few other options in our study area. While numbers of small mammals were high during some winters, their availability was limited by snow cover. Coyotes spent a large amount of time hunting voles during only one winter, 1993-94. Red squirrels can escape predation by coyotes by climbing trees, and they often spend long periods in their arboreal nests during cold periods (Stuart-Smith & Boutin 1995). The slower hunting speed and ambush beds of lynx seem better-suited to hunting squirrels, and lynx were observed to successfully pursue squirrels into trees on several occasions, an option not open to coyotes. "Facultative specialists" (Glasser 1982) may better describe both coyotes and lynx in the boreal forest. Prey switching is not expected when predators have strong preferences for specific prey, and when there is little variation among individuals in their choice of prey (Murdoch 1969; Murdoch & Oaten 1975). Our data suggest that both predators preferred hares over other 143 prey. We do not know though, whether the shift in diets of lynx towards red squirrels from 1992-93 to 1994-95 was due to all individuals hunting more squirrels, or certain individuals specializing on them. We will need to conduct intensive analyses of our data by area to try to distinguish between these alternatives. Individual preferences for prey, probably learned, have been noted in many felids (Kruuk 1986). Prey switching may, in some cases, lead to type-3 functional responses (Murdoch & Oaten 1975). Hares were at very low densities before lynx actively pursued squirrels in large numbers though. Our data on kill rates of hares by lynx suggest that the switch back from squirrels to hares may occur at higher densities of hares during the increase phase of the cycle than during the decrease phase (Chapter 3). We suggest then, that lynx may not be as plastic as coyotes in modifying their foraging behaviour in response to changing relative availabilities of prey. Use of Habitat by Coyotes and Lynx Patterns of habitat use by coyotes and lynx changed over the course of the hare cycle, and they were similar to those of hares (Fig. 4.4). Increasing use of more dense cover by hares during population declines has been noted in a number of studies (e.g., Wolff 1980; Hik 1995). Very dense cover may act as refuges for hares during population lows (Wolff 1980; Akcakaya 1992). Coyotes and lynx used less dense cover than hares at all phases of the cycle (Fig. 4.5), and several researchers have concluded that the predators were ineffective at hunting hares in very dense habitats (e.g., Major 1989). During the increase and peak phases of the hare cycle in this study, the hunting success of lynx was about the same in all habitat types, while coyotes were more successful in closed cover (Murray, Boutin & O'Donoghue 1994; Murray et al. 1995). 144 Coyotes and lynx both concentrated their activities in more closed forest types, and in areas with higher densities of hares (Fig. 4.4, Fig. 4.6). This is consistent with other studies of habitat use by these species in the boreal forest (Brand, Keith & Fischer 1976; Ward & Krebs 1985; Staples 1995; Poole, Wakelyn & Nicklen 1996). Coyotes used more open cover during periods of low abundance of hares, which, in 1992-93 and 1993-94, corresponded with winters in which they were often "mousing" in meadows. Lynx used denser habitats during these years, while they hunted red squirrels and hares. We have no evidence that habitat selection by coyotes and lynx was affected by intraspecific competition. Both species used the most dense habitats, where higher numbers of hares were found, during periods of peak abundance of predators, which is the opposite pattern expected if interference among predators was pushing them to suboptimal habitats. Hunting Tactics of Coyotes and Lynx Greater use of ambush, as opposed to active searching, is characteristic of felids, whereas most canids typically run down their prey (Eisenberg 1986; Kruuk 1986). The use of hunting beds by lynx has been noted in many studies (Saunders 1963a, 1963b; Haglund 1966; Nellis & Keith 1968; Parker 1981; Murray et al. 1995), but the percent of kills from beds have ranged from 12% (Nellis & Keith 1968) to 61% (Saunders 1963a, 1963b). The density of hares and range of alternative prey available undoubtedly affects the frequency of use of hunting beds. The large increase in use of hunting beds by lynx, during the cyclic decline and low in our study (Fig. 4.7), suggests an adaptive change in hunting tactics. In Alberta, Brand, Keith & Fischer (1976) suggested that relatively short daily travel distances by lynx at cyclic lows may have been due to increased use of hunting beds. Hunting success of lynx was not greater from beds when they were preying on hares or squirrels (Fig. 4.9). As discussed previously though, we used a conservative estimate of the 145 number of kills from beds. Using the criterion that all kills within 30 m of hunting beds were considered ambushes, Murray et al. (1995) found that chases of hares from beds were more successful (46%) than those not initiated from beds (27%) during the cyclic increase and peak at Kluane. Even with comparable or lower hunting success (once chases are initiated) though, hunting from beds may be more energy-efficient during periods of low prey abundance, and may be better suited for hunting alternative prey such as squirrels. Coyotes made a major shift in hunting tactics during the second winter of the cyclic low in abundance of hares (1993-94), when the frequency of "mousing" greatly increased along their trails. They shifted back to hunting hares in more closed cover the following winter though. Use of trails by coyotes and lynx can help them conserve energy while traveling, and, when following trails of hares, can increase encounter rates with prey. Coyotes have a relatively high foot-load, and deep snow inhibits their movements and hunting success (Wells / & Bekoff 1982; Murray & Boutin 1991). Both predators may follow hare trails as a hunting technique (Brand, Keith & Fischer 1976; Keith et al. 1977). Our data showing higher use of hare trails by both predators during periods of increasing hare abundance are consistent with the suggestion of Keith et al. (1977) that this may contribute to increasing kill rates by coyotes, but they may simply be a reflection of increased abundance of trails to follow as well. While certainly not conclusive, our data suggest the possibility that foraging success of adult groups of lynx may have been higher than single animals, at least in 1993-94 (Fig. 4.11). Family groups of lynx are typical of the species during winter, and, since kittens are apparently less skilled at hunting than adults (Saunders 1963a), per-individual kill rates are lower. Families do hunt as a unit though, with young animals usually flanking the female through good hunting habitat (Saunders 1963a; Haglund 1966; Parker 1981). We have found only one other reference to groups of adult lynx hunting together (Barash 1971). Increased foraging success has been suggested as one of the benefits of group-living by carnivores 146 (review in Gittleman 1989), and while it is certainly not typical of lynx, it may be temporarily beneficial to them during periods of low prey abundance. Coyotes often hunt in pairs or family groups, particularly when hunting larger prey (Bowen 1981; Messier & Barrette 1982). Possible cooperative hunting of hares by coyotes, in which one animal runs through good patches of cover while the other circles the perimeter, has been reported by Ozoga & Harger (1966), and observed in this study as well. Increased foraging efficiency with group size has not been generally noted for coyotes though (Messier & Barrette 1982; Bekoff & Wells 1986). Conclusions We found clear behavioural responses by coyotes and lynx to the snowshoe hare cycle. Coyotes preferred hares to other prey at all densities, and changes in their use of habitat followed those of hares. They concentrated their hunting activities in areas of high hare numbers. During years of low abundance of hares, and high vole numbers, coyotes used more open cover while "mousing". Coyotes frequently used trails of other predators, humans, and, particularly during cyclic highs, hares for travel and hunting. Lynx also preferred hares to other prey at all densities, but switched to preying on red squirrels during the cyclic low and subsequent early increase in hare abundance. Habitat use by lynx followed the same general pattern as that of hares, and lynx concentrated their activities in areas of high hare numbers. During the cyclic decline and low, lynx increasingly used ambush beds, from which they hunted both hares and squirrels. Hunting success was not higher from beds, but hunting by ambush may have been more energetically efficient during periods of low prey abundance. We observed groups of 2-3 adult lynx hunting together 147 during the decline and low phases of the cycle, and limited data suggest that this may have increased their foraging success, at least in 1 year. 148 CHAPTER 5 GENERAL CONCLUSIONS: RESPONSES OF COYOTES AND LYNX TO THE SNOWSHOE HARE CYCLE Summary of Conclusions The main objectives of this study were to determine and contrast the numerical and functional responses of coyotes and lynx to the large changes in prey abundance associated with the 10-year cycle of snowshoe hares. Secondly, we wanted to investigate the behavioural mechanisms leading to changes in kill rates (i.e., the functional response) of the two predators. Coyotes and lynx responded to cyclic changes in the abundance of their main prey species, snowshoe hares, with major demographic and behavioural changes. I will summarize the main conclusions of this study, and then follow with a discussion of the total impact of predation by coyotes and lynx on hares and alternative prey. Finally, I will briefly discuss evidence for competition between the two predators, and the usefulness of the specialist-generalist contrast in characterizing them. The most important findings of this study are: 1) Coyotes and lynx responded numerically in much the same way to the 26-44-fold fluctuation in numbers of hares in our study area. Numbers of coyotes varied 6-fold and those of lynx 7.5-fold, and the abundances of both predators were maximal a year later than the peak in numbers of snowshoe hares. 2) Cyclic declines in numbers of coyotes were associated with lower reproductive output and high dispersal rates. Likewise, few to no kits were produced by lynx after the second winter of declining numbers of hares. High dispersal rates were characteristic of lynx during the early cyclic decline, but low in situ survival was observed later in the decline. 149 3) The total biomass of available prey fluctuated approximately 4-fold as densities of hares cycled. Snowshoe hares represented more than 60% of this biomass in all but the three lowest years of the cycle. Red squirrels were the most important alternative prey, representing 17-31% of total biomass during winters of higher numbers of hares, and 58-73% in winters of low hare abundance. Small mammals comprised less than 10% of total biomass of prey in all but one year, when they comprised 14%. 4) Coyotes and lynx both fed mostly on hares during all winters except during cyclic lows. Coyotes killed more voles than hares during two winters when abundance of hares was lowest, and numbers of small mammals were high. The main alternative prey of lynx was red squirrels, which they killed more than hares during the two winters of low hare numbers, and the first winter of the subsequent cyclic increase. 5) Coyotes and lynx both showed clear functional responses to changes in the densities of snowshoe hares. Coyotes responded with higher kill rates during the increase in hare abundance than did lynx, and their functional response is described equally well by linear and type-2 curves. Kill rates of hares by coyotes were lower during the cyclic decline in hare densities than during the increase. Kill rates of hares by lynx changed almost 4-fold during the cyclic fluctuation. Their functional response is well-described by a type-2 curve. Kill rates by lynx were lower during the increase in density of hares than during the cyclic decline. 6) Coyotes killed the maximum number of hares per coyote (2.3 hares/day) one year before the peak in hare abundance. Lynx killed the most hares (1.2 hares/day) one year after the peak in hare abundance. Both of these kill rates are more than the estimated energetic requirements of the predators. 7) Coyotes killed more hares early in the winter, and cached many of these for later retrieval. 150 8) Coyotes and lynx preferred hares to other prey species during all phases of the cycle. At low densities of hares, lynx switched to hunting more actively for, and preying on red squirrels. 9) Habitat use by coyotes and lynx changed over the cycle, and roughly paralleled trends in habitats used by hares. All three species used the densest cover during the second winter of decline in hare abundance. Coyotes and lynx concentrated their activities in areas of high densities of hares during most winters. Coyotes hunted for voles in more open cover during the cyclic low in hare abundance, when voles were numerous. 10) Lynx increasingly used hunting beds for ambushing both hares and red squirrels during the cyclic decline and low. Hunting success was not higher from these beds, but hunting by ambush may have been more energetically efficient. Coyotes switched their hunting tactics to active foraging for voles during only one winter. Both predators frequently traveled on the trails of hares, and coyotes also often used trails made by other predators and humans. 11) Lynx hunted in adult groups for the first time during the cyclic decline and low in hare numbers. Our data suggest the per-individual foraging success may have been higher for groups during one winter. Discussion Total Impact of Predation by Coyotes and Lynx on Hares We estimated the total impact of predation by coyotes and lynx on populations of snowshoe hares by calculating the total number of hares killed by these predators each winter, from October to April, based on our measured functional and numerical responses. 151 The percentage of the fall population of hares killed by coyotes and lynx over the winter falls in a counter-clockwise temporal pattern typical of delayed-density dependent predation (Sinclair & Pech 1996; Fig. 5.1). The low predation rate calculated in 1991-92~largely the result of rapidly declining populations of coyotes and lynx (Fig. 2.2)~departs from the pattern expected with a simple time lag of the effects of predation though. We used data collected on the survival of radio-collared hares to make a second, independent estimate of the effects of predation by coyotes and lynx on hares. This analysis suggests that predation by these predators was 1.5-4-times higher than our estimates from this study (Fig. 5.2), particularly during the decline and low phases of the cycle. Rates of predation calculated from the hare survival data are almost certainly overestimates, as evidenced by the fact that the impacts of predation of coyotes and lynx alone (i.e., not even accounting for mortalities from predation by avian predators), estimated from telemetry data from 1990-91 to 1992-93, exceeded the total measured declines in densities of hares from live-trapping during these winters (Table 5.1). Higher vulnerability of radio-collared hares to predation, or attraction of predators to trapping grids could both account for this bias. Nonetheless, the data from the hare telemetry suggest that the effect of predation by coyotes and lynx was highest in 1991-92, the winter we measured quite low rates from our predator data. Many radio-collared hares were killed by coyotes in October-November this winter, and we could have missed many kills, since snow conditions were poor for snow-tracking until November in 1991. Our estimates of the total impact of predation by coyotes and lynx ranged from 9-13% of the fall population of hares during the cyclic increase, 21% at the peak, 21-40% during the decline, and 15-47% during the cyclic low in hare abundance (Fig. 5.1). These are approximately double those calculated in Alberta, of 3-6% during the increase, and 9-20% during the decline of hare populations (Keith et al. 1977). We suggest that our estimates of kill rates by predators, particularly coyotes, are conservative, because of potential biases 152 Fig. 5.1. Total impact of predation by coyotes and lynx during winter (October-April) on populations of snowshoe hares in the southwest Yukon, based on measured numerical and functional responses from 1987 to 1995. Densities of hares are fall estimates, based on live-trapping in October-November. Numbers next to data points indicate years (e.g., 87 = winter of 1987-88). 153 50 92 Percentage of Hare Population Killed 40 30 L 20 h 10 0 0 50 100 150 200 250 Hares per 100 ha 154 Fig. 5.2. Total impact of predation by coyotes and lynx during winter (October-April) on populations of snowshoe hares in the southwest Yukon, based on survival of radio-collared hares from 1987 to 1995. Densities of hares are fall estimates, based on live-trapping in October-November. Numbers next to data points indicate years (e.g., 87 = winter of 1987-88). 155 100 91 Percentage of Hare Population Killed 80 60 40 20 0 0 50 100 150 200 250 Hares per 100 ha 156 Table 5.1. Estimated proportions of hares surviving the winters from 1987 to 1995 in the southwest Yukon, based on survival rates of radio-collared hares and estimates of population size. Kaplan-Meier survival rates were calculated considering only predation by coyotes and lynx as sources of mortality (i.e., mortalities from all other sources were censored for this analysis). Mark-recapture estimates of population size were made by live-trapping hares at the beginning and end of each winter. Winter Proportion of Hares Surviving Predation by Coyotes and Lynx, based on Radio-Telemetry Proportion of Hares Surviving, based on Population Estimates from Live-Trapping 1987-88 0.784 1.000* 1988-89 0.755 0.325 1989-90 0.711 0.630 1990-91 0.440 0.537 1991-92 0.159 0.259 1992-93 0.291 0.615 1993-94 0.427 0.250 1994-95 0.562 0.415 * Based on live-trapping on 3 control grids, 2 of which had higher estimated populations of hares at the end of winter than at the beginning. 157 discussed in Chapter 3. Furthermore, we have estimated the effects of predation over only 7 months of the year (October-April). The diets of coyotes (Todd, Keith & Fischer 1981; Andelt et al. 1987; Theberge & Wedeles 1989) and lynx (Van Zyll de Jong 1966; Parker et al. 1981) are generally more varied during summer, so we expect kill rates of hares to be lower. Both predators do prey on hares during summer though, and so the total impact on hare numbers is greater than we present here. The impact of all predators, in all seasons, needs to be assessed for a full evaluation of the effects of predation on the hare cycle. Our data suggest that the delayed numerical (coyotes and lynx) and functional (lynx) responses of mammalian predators contribute to the cyclic dynamics of the interaction, and that the magnitude of the effect of predation by coyotes and lynx is greater than previously measured. Recent models of the hare cycle are consistent with the assertion that predation is an essential component in generating the cycle (Akcakaya 1992; Royama 1992), but these models also incorporate intrinsic density-dependent regulation of hare numbers. A 3-factor explanation, incorporating hare-vegetation and hare-predator dynamics is supported by both empirical (Krebs et al. 1995) and theoretical (Akcakaya 1992; Royama 1992; Stenseth 1995) evidence. Impact of Predation by Coyotes and Lynx on Alternative Prey There is ample evidence from field studies that cyclic fluctuations of a preferred prey species may also occur in populations of alternative prey of predators (Marcstrom, Kenward & Engren 1988; Marcstrom et al. 1989; Henttonen 1985; Lindstrom et al. 1994). In our study area, trends in populations of Arctic ground squirrels, grouse, and ptarmigan were correlated with those of hares (Boutin et al. 1995), suggesting there may be a causal connection. We evaluated the effects of predation by coyotes and lynx during winter on populations of red squirrels, using our measured numerical responses, and kill rates calculated in the 158 same manner as those for hares in Chapter 3. We were unable to evaluate the effects of predation on numbers of small mammals, because of the difficulty of judging whether attempted kills were successful (see Chapter 4). The estimated total effect of predation by coyotes and lynx on red squirrels was very small (<4% of the fall populations) in all winters, and showed no pattern relative to densities of squirrels (Fig. 5.3). This agrees with the experimental results of Stuart-Smith & Boutin (1995), showing predation by mammals had little effect on population sizes of red squirrels. Coexistence of Coyotes and Lynx Competition between sympatric predators, and how it affects partitioning of resources and breadth of resource use, has been much explored and debated in the literature (Rosenzweig 1966; Schoener 1982; Wiens 1993). In general, overlap in resource use between sympatric species is less during "lean" times (high competition; Schoener 1982) than during periods of resource abundance, but this prediction may not hold due to complex relationships between resource abundance and species-specific thresholds of resource limitation (Wiens 1993), or different strategies of resource acquisition (Glasser & Price 1982; MacNally 1995). Accordingly, studies of patterns of overlap among predators have shown no change (Jaksic, Feinsinger & Jimenez 1993; Meserve et al. 1996), lower overlap (Korpimaki 1987), or higher overlap (Wiens 1993) with lower resource abundance. We conducted no experiments enabling us to add empirical evidence on mechanisms permitting coexistence of coyotes and lynx. The question of how these two predators, of very similar size, survive using a very limited resource base is especially interesting though, given that coyotes are relatively new immigrants in the north. We summarize our observations here to suggest hypotheses rather than provide answers. 159 Fig. 5.3. Total impact of predation by coyotes and lynx during winter (October-April) on populations of red squirrels in the southwest Yukon, based on measured numerical and functional responses from 1987 to 1995. Densities of squirrels are fall estimates, based on live-trapping in August. Numbers next to data points indicate years (e.g., 87 = winter of 1987-88). 160 Percentage of Red Squirrel Population Killed , 200 300 400 500 Squirrels per 100 ha 161 We calculated Horn's index of overlap in diet (based on number of kills and biomass of kills) and habitat use between coyotes and lynx for each winter of our study. Overlap in diet was high (>0.75 for number of kills, >0.86 for biomass) in all winters (Fig. 5.4). The lowest dietary overlap was in 1992-93, when coyotes killed mostly small mammals and lynx killed more red squirrels (Fig. 3.2), at a time when numbers of hares were low. Overlap in habitat use was high (range 0.82-0.95) in all winters (Fig. 5.4). These observations are consistent with those made of high overlap in diets between bobcats and coyotes, in areas where coyotes were relatively recent colonizers (Major & Sherburne 1987; Witmer & DeCalesta 1986; Litvaitis & Harrison 1989). There is correlative evidence that coyotes may depress numbers of bobcats in the western United States (Nunley 1978), and both interspecific competition and intraguild predation have been suggested as mechanisms. We observed only one case of a coyote killing a young lynx (O'Donoghue, Hofer & Doyle 1995), so the influence of predation by coyotes on lynx is not likely large in our area. Coyotes and lynx had very similar diets and used similar habitats during our study, but lynx concentrated their hunting on red squirrels during periods of low abundance of hares, and coyotes concentrated on small mammals. The degree of overlap in their diets during cyclic lows is likely highly dependent on the availability of alternative prey—in our study area, population fluctuations of red squirrels and voles, which are largely unrelated to cyclic hare dynamics (Boutin et al. 1995), would have a great influence on this. Lynx appear better able than coyotes to take advantage of the most consistently available alternative prey species, red squirrels, and during periods when few voles or hares are available, lynx may persist in greater numbers than coyotes. In areas where other alternative food sources, such as the carrion of moose (e.g., Staples 1995) or livestock (e.g., Todd, Keith & Fischer 1981), are consistently available, coyotes may persist in higher numbers. Even with very high or even complete overlap in resource use though, recent models of competition have suggested that species may coexist for long periods of time (Hubbell & Foster 1986; MacNally 1995). 162 Fig. 5.4. Overlap in diet (number of kills and biomass of kills) and habitat use by coyotes and lynx in the southwest Yukon from 1987-88 to 1994-95. 163 Horn's Index of Overlap 1.00 0.8 0.6 0.4 0.2 0 • Number of kills • • - Biomass of kills J L Horn's Index of Overlap 1.00 0.8 0.6 0.4 0.2 0 87-88 J L 89-90 91-92 93-94 Habitat use Winter 164 The Specialist-Generalist Contrast The degree of versatility in resource use by any consumer is a function of species-specific constraints, and external factors, such as the relative availabilities of resources. Predators that are generalists, when considered over their whole geographic range, may be local specialists (MacNally 1995). Clearly, in a low-diversity environment such as the boreal forest in our study area, dietary options are limited for terrestrial predators. Given the limited availability of voles and other alternative prey, coyotes in our study area appeared to "specialize" on hares even during periods (1994-95) when lynx were preying more heavily on red squirrels. In fluctuating environments, facultative foraging strategies, in which predators concentrate their foraging efforts on the most profitable prey, which may vary in time, should be favored over obligate strategies (Glasser 1982). Over their geographic range, North American lynx can certainly be said to be more dependent on snowshoe hares than are coyotes—on this basis, they can be considered hare specialists. Locally though, at least some individual lynx can "specialize" on preying on red squirrels during periods of low hare abundance, and so they may perhaps better be called "facultative specialists". Our data from 1994-95 suggest that lynx that became skilled at hunting squirrels continued to do so despite increasing numbers of hares. Several studies have demonstrated that there is a cost to switching among prey species, which if too high, may result in consumers maintaining dietary preferences rather than accurately tracking resource abundance (Lewis 1986; Dukas & Clark 1995). Studies comparing the plasticities of individual lynx and coyotes in adjusting their patterns of resource use to changing availabilities of prey would be valuable for clarifying mechanisms behind shifts at the population level. 165 LITERATURE CITED Abrams, P. A. 1989. Decreasing functional responses as a result of adaptive consumer behavior. Evolutionary Ecology 3:95-114. Abrams, P. A. 1990. The effects of adaptive behavior on the type-2 functional response. Ecology 71:877-885. Abrams, P. A. 1992a. Adaptive foraging by predators as a cause of predator-prey cycles. Evolutionary Ecology 6:56-72. Abrams, P. A. 1992b. Predators that benefit prey and prey that harm predators: unusual effects of interacting foraging adaptations. American Naturalist 140:573-600. Adamcik, R. S., A. W. Todd, & L. B. Keith. 1978. Demographic and dietary responses of great horned owls during a snowshoe hare decline. Canadian Field-Naturalist 92:156-166. Akcakaya, H. R. 1992. Population cycles of mammals: evidence for a ratio-dependent predation hypothesis. Ecological Monographs 62:119-142. Akre, B. G., & D. M. Johnson. 1979. Switching and sigmoidal functional response curves by damselfly naiads with alternative prey available. Journal of Animal Ecology 48:703-720. Aldama, J. J., J. F. Beltran, and M. Delibes. 1991. Energy expenditure and prey requirements of free-ranging Iberian lynx in southwestern Spain. Journal of Wildlife Management 55:635-641. Andelt, W. F. 1985. Behavioral ecology of coyotes in south Texas. Wildlife Monographs 94:1-45. Andelt, W. F., J. G. Kie, F. F. Knowlton, & K. Cardwell. 1987. Variation in coyote diets associated with season and successional changes in vegetation. Journal of Wildlife Management 51:273-277. Andrewartha, H. G., & L. C. Birch. 1954. The distribution and abundance of animals. University of Chicago Press, Chicago. Angelstam, P., E. Lindstrom, & P. Widen. 1984. Role of predation in short-term population fluctuations of some birds and mammals in Fennoscandia. Oecologia 62:199-208. Angelstam, P., E. Lindstrom, & P. Widen. 1985. Synchronous short-term fluctuations of some birds and mammals in Fennoscandia-occurrence and distribution. Holarctic Ecology 8:285-298. Arditi, R., & H. R. Akcakaya. 1990. Underestimation of mutual interference of predators. Oecologia 83:358-361. Arnqvist, G., & D. Wooster. 1995. Meta-analysis: synthesizing research findings in ecology and evolution. Trends in Ecology and Evolution 10:236-240. 166 Bailey, T. N. 1974. Social organization in a bobcat population. Journal of Wildlife Management 38:435-446. Banfield, A. W. F. 1974. The mammals of Canada. University of Toronto Press, Toronto. Barash, D. P. 1971. Cooperative hunting in the lynx. Journal of Mammalogy 52:480. Bekoff, M. (ed.). 1978. Coyotes: biology, behavior, and management. Academic Press, New York. Bekoff, M., J. Diamond, & J. B. Mitton. 1981. Life-history patterns and sociality in canids: body size, reproduction, and behavior. Oecologia 50:386-390. Bekoff, M., & M. C. Wells. 1981. Behavioural budgeting by wild coyotes: the influence of food resources and social organization. Animal Behaviour 29:794-801. Bekoff, M., & M. C. Wells. 1986. Social ecology and behavior of coyotes. Advances in the Study of Behavior 16:251-338. Bell, W. J. 1991. Searching behaviour. Chapman and Hall, London. Berg, W. E., & R. A. Chesness. 1978. Ecology of coyotes in northern Minnesota. Pages 229-247 in M. Bekoff (ed.). Coyotes: biology, behavior, and management. Academic Press, New York. Bond, R. M. 1939. Coyote food habits on the Lava Beds National Monument. Journal of Wildlife Management 3:180-198. Boutin, S. 1995. Testing predator-prey theory by studying fluctuating populations of small mammals. Wildlife Research 22:89-100. Boutin, S., & H. D. Cluff. 1989. Coyote prey choice: optimal or opportunistic foraging? A comment. Journal of Wildlife Management 53:663-666. Boutin, S., C. J. Krebs, R. Boonstra, M. R. T. Dale, S. J. Hannon, K. Martin, A. R. E. Sinclair, J. N. M. Smith, R. Turkington, M. Blower, A. Byrom, F. I. Doyle, C. Doyle, D. Hik, L. Hofer, A. Hubbs, T. Karels, D. L. Murray, V. Nams, M. O'Donoghue, C. Rohner, & S. Schweiger. 1995. Population changes of the vertebrate community during a snowshoe hare cycle in Canada's boreal forest. Oikos 74:69-80. Boutin, S., C. J. Krebs, A. R. E. Sinclair, & J. N. M. Smith. 1986. Proximate causes of losses in a snowshoe hare population. Canadian Journal of Zoology 64:606-610. Bowen, W. D. 1981. Variation in coyote social organization: the influence of prey size. Canadian Journal of Zoology 59:639-652. Bowen, W. D. 1982. Home range and spatial organization of coyotes in Jasper National Park, Alberta. Journal of Wildlife Management 46:201-216. Brand, C. J., & L. B. Keith. 1979. Lynx demography during a snowshoe hare decline in Alberta. Journal of Wildlife Management 43:827-849. 167 Brand, C. J., L. B. Keith, & C. A. Fischer. 1976. Lynx responses to changing snowshoe hare densities in Alberta. Journal of Wildlife Management 40:416-428. Breitenmoser, U., B. G. Slough, & C. Breitenmoser-Wursten. 1993. Predators of cyclic prey: is the Canada lynx victim or profiteer of the snowshoe hare cycle? Oikos 66:551-554. Camenzind, F. J. 1978. Behavioral ecology of coyotes on the National Elk Refuge, Jackson, Wyoming. Pages 267-294 in M. Bekoff(ed.). Coyotes: biology, behavior, and management. Academic Press, New York. Clark, F. W. 1972. Influence of jackrabbit density on coyote population change. Journal of Wildlife Management 36:343-356. Crawley, M. J. 1975. The numerical responses of insect predators to changes in prey density. Journal of Animal Ecology 44:877-892. Crawley, M. J. 1992. Overview. Pages 476-489 in M. J. Crawley (ed.). Natural enemies. Blackwell Scientific Publications, Oxford. Crowe, D. M. 1972. The presence of annuli in bobcat tooth cementum layers. Journal of Wildlife Management 36:1330-1332. Davies, N. B. 1977. Prey selection and search strategy of the spotted flycatcher (Muscicapa striata), a field study on optimal foraging. Animal Behaviour 25:1016-1033. Douglas, G. W. 1974. Montane zone vegetation of the Alsek River region, south-western Yukon. Canadian Journal of Botany 52:2505-2532. Dukas, R., & C. W. Clark. 1995. Searching for cryptic prey: a dynamic model. Ecology 76:1320-1326. Dunbrack, R. L., & L. A. Giguere. 1987. Adaptive responses to accelerating costs of movement: a bioenergetic basis for the type-Ill functional response. American Naturalist 130:147-160. Eaton, R. L. 1979. Interference competition among carnivores: a model for the evolution of social behavior. Carnivore 2:9-16. Eisenberg, J. F. 1986. Life history strategies of the Felidae: variations on a common theme. Pages 293-303 in S. D. Miller, & D. D. Everett (eds.). Cats of the world: biology, conservation, and management. National Wildlife Federation, Washington, D. C. Elsey, C. A. 1954. A case of cannibalism in Canada lynx (Lynx canadensis). Journal of Mammalogy 35:129. Elton, C. 1927. Animal ecology. Sidgwick & Jackson Ltd., London. Elton, C , & M. Nicholson. 1942. The ten year cycle in numbers of lynx in Canada. Journal of Animal Ecology 11:215-244. Erlinge, S. 1987. Predation and non-cyclicity in a microtine population in southern Sweden. Oikos 50:347-352. 168 Erlinge, S, G. Gbransson, L. Hansson, G. Hogstedt, O. Liberg, I. N. Nilsson, T. Nilsson, T. von Schantz, & M. Sylven. 1983. Predation as a regulating factor on small mammal populations in southern Sweden. Oikos 40:36-52. Erlinge, S, G. Gbransson, G. Hogstedt, G. Jansson, O. Liberg, J. Loman, I. N. Nilsson, T. von Schantz, & M. Sylven. 1984. Can vertebrate predators regulate their prey? American Naturalist 123:125-133. Erlinge, S, G. Gbransson, G. Hogstedt, G. Jansson, O. Liberg, J. Loman, I. N. Nilsson, T. von Schantz, & M. Sylven. 1988. More thoughts on vertebrate regulation of prey. American Naturalist 132:148-154. Errington, P. L. 1946. Predation and vertebrate populations. Quarterly Review of Biology 21:144-177, 221-245. Fichter, E., G. Schildman, & J. H. Sather. 1955. Some feeding patterns of coyotes in Nebraska. Ecological Monographs 25:1-37. Finerty, J. P. 1980. The population ecology of cycles in small mammals. Yale University Press, New Haven, Connecticut. Fitzgerald, B. M. 1977. Weasel predation on a cyclic population of the montane vole (Microtus montanus) in California. Journal of Animal Ecology 46:367-397. Floyd, T. J., L. D. Mech, & P. A. Jordan. 1978. Relating wolf scat content to prey consumed. Journal of Wildlife Management 42:528-532. Formanowicz, D. R., & P. J. Bradley. 1987. Fluctuations in prey density: effects on the foraging tactics of scolopendrid centipedes. Animal Behaviour 35:453-461. Gittleman, J. L. 1989. Carnivore group living: comparative trends. Pages 183-207 in J. L. Gittleman (ed.). Carnivore behavior, ecology and evolution. Cornell University Press, Ithaca, New York. Glasser, J. W. 1982. A theory of trophic strategies: the evolution of facultative specialists. American Naturalist 119:250-262. Glasser, J. W., & H. J. Price. 1982. Niche theory: new insights from an old paradigm. Journal of Theoretical Biology 99:437-460. Goss-Custard, J. D. 1980. Competition for food and interference among waders. Ardea 68:31-52. Goszczynski, J. 1977. Connections between predatory birds and mammals and their prey. Acta Theriologica 22:399-430. Haglund, B. 1966. Winter habits of the lynx (Lynx lynx L.) and wolverine (Gulo gulo L.) as revealed by tracking in the snow. (In Swedish; English summary.) Viltrevy 4:81-310. Halpin, M. A., & J. A. Bissonette. 1988. Influence of snow depth on prey availability and habitat use by red fox. Canadian Journal of Zoology 66:587-592. 169 Hamlin, K. L., S. J. Riley, D. Pyrah, A. R. Dood, & R. J. Mackie. 1984. Relationships among mule deer fawn mortality, coyotes, and alternative prey species during summer. Journal of Wildlife Management 48:489-499. Hanski, I., L. Hansson, & H. Henttonen. 1991. Specialist predators, generalist predators, and the microtine rodent cycle. Journal of Animal Ecology 60:353-367. Hanski, I., & E. Korpimaki. 1995. Microtine rodent dynamics in northern Europe: parameterized models for the predator-prey interaction. Ecology 76:840-850. Hanski, I., P. Turchin, E. Korpimaki, & H. Henttonen. 1993. Population oscillations of boreal rodents: regulation by mustelid predators leads to chaos. Nature 364:232-235. Hansson, L. 1979. Food as a limiting factor for small rodent numbers: tests of two hypotheses. Oecologia 37:297-314. Hansson, L. 1987. An interpretation of rodent dynamics as due to trophic interactions. Oikos 50:308-318. Hansson. L., & H. Henttonen 1988. Rodent dynamics as community processes. Trends in Ecology and Evolution 3:195-200. Harrison, D. J. 1992. Dispersal characteristics of juvenile coyotes in Maine. Journal of Wildlife Management 56:128-138. Hassell, M. P. 1971. Mutual interference between searching insect parasites. Journal of Animal Ecology 40:473-486. Hassell, M. P. 1979. The dynamics of predator-prey interactions: polyphagous predators, competing predators and hyperparasitoids. Pages 283-306 in R. M. Anderson, B. D. Turner, & L. R. Taylor (eds.). Population dynamics. Blackwell Scientific Publications, Oxford. Hassell, M. P., & H. N. Comins. 1978. Sigmoid functional responses and population stability. Theoretical Population Biology 14:62-67. Hassell, M. P., J. H. Lawton, & J. R. Beddington. 1977. Sigmoid functional responses by invertebrate predators and parasitoids. Journal of Animal Ecology 46:249-262. Hassell, M. P., & R. M. May. 1986. Generalist and specialist natural enemies in insect predator-prey interactions. Journal of Animal Ecology 55:923-940. Henttonen, H. 1985. Predation causing extended low densities in microtine cycles: further evidence from shrew dynamics. Oikos 45:156-157. Henttonen, H., T. Oksanen, A. Jortikka., & V. Haukisalmi. 1987. How much do weasels shape microtine cycles in the northern Fennoscandian taiga? Oikos 50:353-365. Hik, D. S. 1995. Does risk of predation influence population dynamics? Evidence from the cyclic decline of snowshoe hares. Wildlife Research 22:115-129. 170 Holling, C. S. 1959a. Some characteristics of simple types of predation and parasitism. Canadian Entomologist 91:385-398. Holling, C. S. 1959b. The components of predation as revealed by a study of small-mammal predation of the European pine sawfly. Canadian Entomologist 91:293-320. Holling, C. S. 1965. The functional response of predators to prey density and its role in mimicry and population regulation. Memoirs of the Entomological Society of Canada 45:1-60. Holling, C. S. 1966. The functional response of vertebrate predators to prey density. Memoirs of the Entomological Society of Canada 48:1-86. Hornfeldt, B. 1978. Synchronous population fluctuations in voles, small game, owls, and tularemia in northern Sweden. Oecologia 32:141-152. Hubbell, S. P., & R. B. Foster. 1986. Biology, chance, and history and the structure of tropical rain forest tree communities. Pages 314-329 in J. Diamond & T. J. Case (eds.). Community ecology. Harper & Row, New York. Ims, R. A., & H. Steen. 1990. Geographic synchrony in microtine population cycles: a theoretical evaluation of the role of nomadic avian predators. Oikos 57:381-387. Inoue, T., & T. Matsura. 1983. Foraging strategy of a mantid, Paratenodera augustipennis S.: mechanisms of switching tactics between ambush and active search. Oecologia 56:564-271. Jaksic, F. M., P. Feinsinger, & J. E. Jimenez. 1993. A long-term study on the dynamics of guild structure among predatory vertebrates at a semi-arid Neotropical site. Oikos 67:87-96. Johnson, M. K., & R. M. Hansen. 1979a. Coyote food habits on the Idaho National Engineering Laboratory. Journal of Wildlife Management 43:951-956. Johnson, M. K., & R. M. Hansen. 1979b. Estimating coyote food intake from undigested residues in scats. American Midland Naturalist 102:363-367. Keith, L. B. 1990. Dynamics of snowshoe hare populations. Pages 119-195 in H. H. Genoways (ed.). Current mammalogy. Plenum Press, New York. Keith, L. B., J. R. Cary, O. J. Rongstad, & M. C. Brittingham. 1984. Demography and ecology of a declining snowshoe hare population. Wildlife Monographs 90:1-43. Keith, L. B., & D. H. Rusch. 1988. Predation's role in the cyclic fluctuations of ruffed grouse. International Ornithological Congress 19:700-732. Keith, L. B., A. W. Todd, C. J. Brand, R. S. Adamcik, & D. H. Rusch. 1977. An analysis of predation during a cyclic fluctuation of snowshoe hares. Proceedings of the International Congress of Game Biologists 13:151-175. Keith, L. B., & L. A. Windberg. 1978. A demographic analysis of the snowshoe hare cycle. Wildlife Monographs 58:1-70. 171 Kie, J. G., J. A. Baldwin, & C. J. Evans. 1994. CALHOME home range analysis program. Electronic user's manual. California Department of Fish and Game. Kleiman, D. G., & J. F. Eisenberg. 1973. Comparisons of canid and felid social systems from an evolutionary perspective. Animal Behaviour 21:637-659. Knick, S. T. 1990. Ecology of bobcats relative to exploitation and a prey decline in southeastern Idaho. Wildlife Monographs 108:1-42. Knowlton, F. F., & L. C. Stoddart. 1992. Some observations from two coyote-prey studies. Pages 101-121 in A. H. Boer (ed.). Ecology and management of the eastern coyote. Wildlife Research Institute, Fredericton, New Brunswick. Koehler, G. M. 1990. Population and habitat characteristics of lynx and snowshoe hares in north central Washington. Canadian Journal of Zoology 68:845-851. Korpimaki, E. 1986. Predation causing synchronous decline phases in microtine and shrew populations in western Finland. Oikos 46:124-127. Korpimaki, E. 1987. Dietary shifts, niche relationships and reproductive output of coexisting kestrels and long-eared owls. Oecologia 74:277-285. Korpimaki, E. 1993. Regulation of multiannual vole cycles by density-dependent avian and mammalian predation? Oikos 66:359-363. Korpimaki, E. 1994. Rapid or delayed tracking of multi-annual vole cycles by avian predators? Journal of Animal Ecology 63:619-628. Korpimaki, E., & K. Norrdahl. 1989a. Predation of Tengmalm's owls: numerical responses, functional responses and dampening impact on population fluctuations of microtines. Oikos 54:154-164. Korpimaki, E., & K. Norrdahl. 1989b. Avian predation on mustelids in Europe 2: impact on small mustelid and microtine dynamics~a hypothesis. Oikos 55:273-276. Korpimaki, E., & K. Norrdahl. 1991a. Do breeding nomadic avian predators dampen population fluctuations of small mammals? Oikos 62:195-208. Korpimaki, E., & K. Norrdahl. 1991b. Numerical and functional responses of kestrels, short-eared owls, and long-eared owls to vole densities. Ecology 72:814-826. Korpimaki, E., K. Norrdahl, & T. Rinta-Jaskari. 1991. Responses of stoats and least weasels to fluctuating food abundances: is the low phase of the vole cycle due to mustelid predation? Oecologia 88:552-561. Krebs, C. J. 1966. Demographic changes in fluctuating populations of Microtus californicus. Ecological Monographs 36:239-273. Krebs, C. J. 1989. Ecological methodology. Harper & Row, New York. 172 Krebs, C. J., S. Boutin, R. Boonstra, A. R. E. Sinclair, J. N. M. Smith, M. R. T. Dale, K. Martin, & R. Turkington. 1995. Impact of food and predation on the snowshoe hare cycle. Science 269:1112-1115. Krebs, C. J., B. S. Gilbert, S. Boutin, A. R. E. Sinclair, & J. N. M. Smith. 1986. Population biology of snowshoe hares. I. Demography of food-supplemented populations in the southern Yukon, 1976-84. Journal of Animal Ecology 55:963-982. Krebs, C. J., & J. H. Myers. 1974. Population cycles in small mammals. Advances in Ecological Research 8:267-399. Kruuk, H. 1972. Surplus killing by carnivores. Journal of Zoology 166:233-244. Kruuk, H. 1986. Interactions between Felidae and their prey species: a review. Pages 353-374 in S. D. Miller & D. D. Everett (eds.). Cats of the world: biology, conservation, and management. National Wildlife Federation, Washington, D. C. Lack, D. 1954. The natural regulation of animal numbers. Clarendon Press, Oxford. Lawrence, E. S., & J. A. Allen. 1983. On the term 'search image'. Oikos 40:313-314. Lawton, J. H., J. R. Beddington, & R. Bonser. 1974. Switching in invertebrate predators. Pages 141-158 in M. H. Williamson (ed.). Ecological stability. Chapman and Hall, London. Lewis, A. C. 1986. Memory constraints and flower choice in Pieris rapae. Science 232:863-865. Lindstrom, E. R. 1989. Food limitation and social regulation in a red fox population. Holarctic Ecology 12:70-79. Lindstrom, E. R., H. Andren, P. Angelstam, G. Cederlund, B. Hornfeldt, L. Jaderberg, P.-A. Lemnell, B. Martinsson, K. Skold, & J. E. Swenson. 1994. Disease reveals the predator: sarcoptic mange, red fox predation, and prey populations. Ecology 75:1042-1049. Lindstrom, E. R., P. Angelstam, P. Widen, & H. Andren. 1987. Do predators synchronize vole and grouse fluctuations?~An experiment. Oikos 48:121-124. Lindstrom, E. R., & B. Hornfeldt. 1994. Vole cycles, snow depth and fox predation. Oikos 70:156-160. Litvaitis, J. A. & D. J. Harrison. 1989. Bobcat-coyote niche relationships during a period of coyote population increase. Canadian Journal of Zoology 67:1180-1188. Litvaitis, J. A., & W. W. Mautz. 1980. Food and energy use by captive coyotes. Journal of Wildlife Management 44:56-61. MacCracken, J. G., and R. M. Hansen. 1987. Coyote feeding strategies in southeastern Idaho: optimal foraging by an opportunistic predator? Journal of Wildlife Management 51:278-285. 173 MacNally, R. C. 1995. Ecological versatility and community ecology. Cambridge University Press, Cambridge. Major, A. R. 1989. Lynx, Lynx canadensis canadensis (Kerr) predation patterns and habitat use in the Yukon Territory, Canada. M. Sc. Thesis, State University of New York, Syracuse. Major, J. T., & J. A. Sherburne. 1987. Interspecific relationships of coyotes, bobcats, and red foxes in western Maine. Journal of Wildlife Management 51:606-616. Manly, B. F. J., P. Miller, & L. M. Cook. 1972. Analysis of a selective predation experiment. American Naturalist 106:719-736. Marcstrom, V., L. B. Keith, E. Engren, & J. R. Cary. 1989. Demographic responses of arctic hares (Lepus timidus) to experimental reductions of red foxes (Vulpes vulpes) and martens (Martes martes). Canadian Journal of Zoology 67:658-668. Marcstrom, V., R. E. Kenward, & E. Engren. 1988. The impact of predation on boreal tetraonids during vole cycles: an experimental study. Journal of Animal Ecology 57:859-872. May, R. M. 1973. Stability and complexity in model ecosystems. Princeton University Press, Princeton, New Jersey. Mech, L. D. 1977. Productivity, mortality, and population trends of wolves in northeastern Minnesota. Journal of Mammalogy 58:559-574. Mech, L. D. 1980. Age, sex, reproduction, and spatial organization of lynxes colonizing northeastern Minnesota. Journal of Mammalogy 61:262-267. Meserve, P. L., J. R. Gutierrez, & F. M. Jaksic. 1993. Effects of vertebrate predation on a caviomorph rodent, the degu (Octodon degus), in a semiarid thorn community in Chile. Oecologia 94:153-158. Meserve, P. L., J. R. Gutierrez, J. A. Yunger, L. C. Contreras, & F. M. Jaksic. 1996. Role of biotic interactions in a small mammal assemblage in semiarid Chile. Ecology 77:133-148. Messier, F., & C. Barrette. 1982. The social system of the coyote (Canis latrans) in a forested habitat. Canadian Journal of Zoology 60:1743-1753. Messier, F., & M. Crete. 1985. Moose-wolf dynamics and the natural regulation of moose dynamics. Oecologia 65:503-512. Mohr, C. O. 1947. Table of equivalent populations of North American mammals. American Midland Naturalist 37:223-249. Moore, T. D., L. E. Spence, C. E. Dugnolle, & W. P. Hepworth. 1974. Identification of the dorsal guard hairs of some mammals of Wyoming. Wyoming Game and Fish Department Bulletin 14. 174 Mowat, G., B. G. Slough, & S. Boutin. 1996. Lynx recruitment during a snowshoe hare population peak and decline in southwest Yukon. Journal of Wildlife Management 60:441-452. Mowat, G., B. G. Slough, & R. Rivard. 1994. A comparison of three live capturing devices for lynx: capture efficiency and injuries. Wildlife Society Bulletin 22:644-650. Murdoch, W. W. 1969. Switching in generalist predators: experiments on predator specificity and stability of prey populations. Ecological Monographs 39:335-354. Murdoch, W. W., & A. Oaten. 1975. Predation and population stability. Advances in Ecological Research 9:1-131. Murray, D. L., & S. Boutin. 1991. The influence of snow on lynx and coyote movements: does morphology affect behavior? Oecologia 88:463-469. Murray, D. L., S. Boutin, & M. O'Donoghue. 1994. Winter habitat selection by lynx and coyotes in relation to snowshoe hare abundance. Canadian Journal of Zoology 72:1444-1451. Murray, D. L., S. Boutin, M. O'Donoghue, & V. O. Nams. 1995. Hunting behaviour of a sympatric felid and canid in relation to vegetative cover. Animal Behaviour 50:1203-1210. Nams, V. O. 1990. Locate II User's Guide. Pacer Computer Software, Truro, Nova Scotia. Nellis, C. H., & L. B. Keith. 1968. Hunting activities and success of lynxes in Alberta. Journal of Wildlife Management 32:718-722. Nellis, C. H., & L. B. Keith. 1976. Population dynamics of coyotes in central Alberta, 1964-68. Journal of Wildlife Management 40:389-399. Nellis, C. H., S. P. Wetmore, & L. B. Keith. 1972. Lynx-prey interactions in central Alberta. Journal of Wildlife Management 36:320-329. Newsome, A. 1990. The control of vertebrate pests by vertebrate predators. Trends in Ecology and Evolution 5: 187-191. Newsome, A., I. Parer, & P. C. Catling. 1989. Prolonged prey suppression by carnivores-predator removal experiments. Oecologia 78:458-467. Nicholson, A. J. 1933. The balance of animal populations. Journal of Animal Ecology 2:132-178. Nunley, G. L. 1978. Present and historical bobcat population trends in New Mexico and the west. Proceedings of the Vertebrate Pest Conference 8:177-184. Nunney, L. 1980. The influence of the Type 3 (sigmoid) functional response upon the stability of predator-prey difference.models. Theoretical Population Biology 18:257-278. Oaten, A., & W. W. Murdoch. 1975. Switching, functional response, and stability in predator-prey systems. American Naturalist 109:299-318. 175 O'Connor, R. M. 1986. Reproduction and age distribution of female lynx in Alaska, 1961-1971-preliminary results. Pages 311-325 in S. D. Miller & D. D. Everett (eds.). Cats of the world: biology, conservation, and management. National Wildlife Federation, Washington D. C. O'Donoghue, M., E. Hofer, & F. I. Doyle. 1995. Predator versus predator. Natural History 104(3):6-9. O'Gara, B. W. 1986. Reliability of scat analysis for determining coyote feeding on large mammals. Murrelet 67:79-81. Ozoga, J. J., & E. M. Harger. 1966. Winter activities and feeding habits of northern Michigan coyotes. Journal of Wildlife Management 30:809-818. Parker, G. R. 1981. Winter habitat use and hunting activities of lynx (Lynx canadensis) on Cape Breton Island, Nova Scotia. Pages 221-248 in J. A. Chapman & D. Pursley (eds.). Worldwide Furbearer Conference proceedings, Aug. 3-11, 1980. Frostburg, Maryland. Parker, G. R., J. W. Maxwell, L. D. Morton, & G. E. J. Smith. 1983. The ecology of the lynx (Lynx canadensis) on Cape Breton Island. Canadian Journal of Zoology 61:770-786. Pearson, O. P. 1964. Carnivore-mouse predation: an example of its intensity and bioenergetics. Journal of Mammalogy 45:177-188. Pearson, O. P. 1966. The prey of carnivores during one cycle of mouse abundance. Journal of Animal Ecology 35:217-233. Pearson, O. P. 1971. Additional measurements of the impact of carnivores on California voles. Journal of Mammalogy 52:41-49. Pech, R. P., A. R. E. Sinclair, A. E. Newsome, & P. C. Catling. 1992. Limits to predator regulation of rabbits in Australia: evidence from predator-removal experiments. Oecologia 89:102-112. Polis, G. A., C. A. Myers, & R. D. Holt. 1989. The ecology and evolution of intraguild predation: potential competitors that eat each other. Annual Review of Ecology and Systematics 20:297-330. Poole, K. G. 1994. Characteristics of an unharvested lynx population during a snowshoe hare decline. Journal of Wildlife Management 58:608-618. Poole, K. G. 1995. Spatial organization of a lynx population. Canadian Journal of Zoology 73:632-641. Poole, K. G., G. Mowat, & B. G. Slough. 1993. Chemical immobilization of lynx. Wildlife Society Bulletin. 21:136-140. Poole, K. G., L. A. Wakelyn, & P. N. Nicklen. 1996. Habitat selection by lynx in the Northwest Territories. Canadian Journal of Zoology 74:845-850. 176 Price, P. W. 1987. The role of natural enemies in insect populations. Pages 287-312 in P. Barbosa & J. C. Schultz (ed.). Insect outbreaks. Academic Press, New York. Readshaw, J. L. 1973. The numerical response of predators to prey density. Journal of Applied Ecology 10:342-351. Rosenthal, R. 1978. Combining the results of independent studies. Psychological Bulletin 85:185-193. Rosenzweig, M. L. 1966. Community structure in sympatric Carnivora. Journal of Mammalogy 47:602-612. Royama, T. 1970. Factors governing the hunting behaviour and selection of food by the great tit (Parus major L.). Journal of Animal Ecology 39:619-668. Royama, T. 1992. Analytical population dynamics. Chapman and Hall, London. Saunders, J. K. 1963a. Food habits of lynx in Newfoundland. Journal of Wildlife Management 27:384-390. Saunders. J. K. 1963b. Movements and activities of lynx in Newfoundland. Journal of Wildlife Management 27:390-400. Schaller, G. B. 1972. The Serengeti lion. University of Chicago Press, Chicago. Schoener, T. W. 1982. The controversy over interspecific competition. American Scientist 70:586-595. Searle, S. R., F. M. Speed, & G. A. Milliken. 1980. Population marginal means in the linear model: an alternative to least squares means. The American Statistician 34:216-221. Shettleworth, S. J., P. J. Reid, & C. M. S. Plowright. 1993. The psychology of diet selection. Pages 56-77 in R. N. Hughes (ed.). Diet selection. Blackwell Scientific Publications, London. Sinclair, A. R. E. 1989. Population regulation of animals. Pages 197-241 in J. M. Cherritt (ed.). Ecological concepts. Blackwell Scientific Publications, Oxford. Sinclair, A. R. E., C. J. Krebs, J. N. M. Smith, & S. Boutin. 1988. Population biology of snowshoe hares. III. Nutrition, plant secondary compounds and food limitation. Journal of Animal Ecology 57:787-806. Sinclair, A. R. E., P. D. Olsen, & T. D. Redhead. 1990. Can predators regulate small mammal populations? Evidence from house mouse outbreaks in Australia. Oikos 59:382-392. Sinclair, A. R. E., & R. P. Pech. 1996. Density dependence, stochasticity, compensation and predator regulation. Oikos 75:164-173. Skogland, T. 1991. What are the effects of predators on large ungulate populations? Oikos 61:401-411. 177 Slough, B. G., & G. Mowat. 1996. Population dynamics of lynx in a refuge and interactions between harvested and unharvested populations. Journal of Wildlife Management 60: in press. Small, R. J., V. Marcstrom, & T. Willebrand. 1993. Synchronous and nonsynchronous fluctuations of some predators and their prey in central Sweden. Ecography 16:360-364. Smith, J. N. M. 1974. The food searching behaviour of two European thrushes. II. The adaptiveness of search patterns. Behaviour 59:1-59. Smith, J. N. M., C. J. Krebs, A. R. E. Sinclair & R. Boonstra. 1988. Population biology of snowshoe hares. II. Interactions with winter food plants. Journal of Animal Ecology 57:269-286. Solomon, M. E. 1949. The natural control of animal populations. Journal of Animal Ecology 18:1-35. Staples, W. R. 1995. Lynx and coyote diet and habitat relationships during a low hare population on the Kenai Peninsula, Alaska. M. Sc. Thesis, University of Alaska, Fairbanks. Stenseth, N. C. 1995. Snowshoe hare populations: squeezed from below and above. Science 269:1061-1062. Stephenson, R. O., D. V. Grangaard, & J. Burch. 1991. Lynx, Felis lynx, predation on red foxes, Vulpes vulpes, caribou, Rangifer tarandus, and Dall sheep, Ovis dalli, in Alaska. Canadian Field-Naturalist 105:255-262. Stephenson, R. O., & P. Karczmarczyk. 1989. Development of techniques for evaluating lynx population status in Alaska. Alaska Department of Fish and Game, Division of Wildlife Conservation, Federal Aid in Wildlife Restoration, Research Final Report, Project W-23-1, Juneau, Alaska. Stuart-Smith, A. K., & S. Boutin. 1995. Predation on red squirrels during a snowshoe hare decline. Canadian Journal of Zoology 73:713-722. Taylor, R. J. 1984. Predation. Chapman and Hall, London. Theberge, J. B., & C. H. R. Wedeles. 1989. Prey selection and habitat partitioning in sympatric coyote and red fox populations, southwest Yukon. Canadian Journal of Zoology 67:1285-1290. Thompson, I. D., I. J. Davidson, S. O'Donnell, & F. Brazeau. 1989. Use of track transects to measure the relative occurrence of some boreal mammals in uncut forest and regeneration stands. Canadian Journal of Zoology 67:1816-1823. Tinbergen, L. 1960. The natural control of insects in pine woods. I. Factors influencing the intensity of predation by songbirds. Archives Neerlandaises de Zoologie 13:266-336. Todd, A. W., L. B. Keith, & C. A. Fischer. 1981. Population ecology of coyotes during a fluctuation of snowshoe hares. Journal of Wildlife Management 45:629-640. 178 Todd, A. W., & L. B. Keith. 1983. Coyote demography during a snowshoe hare decline in Alberta. Journal of Wildlife Management 47:394-404. Trexler, J. C , C. E. McCulloch, & J. Travis. 1988. How can the functional response best be determined? Oecologia 76:206-214. Trexler, J. O, & J. Travis. 1993. Nontraditional regression analyses. Ecology 74:1629-1637. Trostel, K., A. R. E. Sinclair, C. J. Walters, & C. J. Krebs. 1987. Can predation cause the ten-year hare cycle? Oecologia 74:185-193. Vander Wall, S. B. 1990. Food hoarding in animals. University of Chicago Press, Chicago. Van Vuren, D., & S. E. Thompson. 1982. Opportunistic feeding by coyotes. Northwest Science 56:131-135. Van Zyll de Jong, C. G. 1966. Food habits of the lynx in Alberta and the Mackenzie District, N. W. T. Canadian Field-Naturalist 80:18-23. Wagner, F. H., & L. C. Stoddart. 1972. Influence of coyote predation on black-tailed jackrabbit populations in Utah. Journal of Wildlife Management 36:329-342. Ward, R. M. P., & C. J. Krebs. 1985. Behavioural responses of lynx to declining snowshoe hare abundance. Canadian Journal of Zoology 63:2817-2824. Wells, M. O, & M. Bekoff. 1982. Predation by wild coyotes: behavioral and ecological analyses. Journal of Mammalogy 63:118-127. Werner, E. E., & D. J. Hall. 1977. Competition and habitat shift in two sunfishes (Centrarchidae). Ecology 58:869-876. Werner, E. E., & D. J. Hall. 1979. Foraging efficiency and habitat switching in competing sunfishes. Ecology 60:256-264. White, G. C , K. P. Burnham, & D. L. Otis. 1982. Capture-recapture and removal methods for sampling closed populations. Los Alomos National Laboratory, Los Alomos, New Mexico. Wiens, J. A. 1993. Fat times, lean times and competition among predators. Trends in Ecology and Evolution 8:348-349. Wilkinson, L. 1990. SYSTAT: the system for statistics. SYSTAT, Inc., Evanston, Illinois. Witmer, G. W., & D. S. DeCalesta. 1986. Resource use by unexploited sympatric bobcats and coyotes in Oregon. Canadian Journal of Zoology 64:2333-2338. Wolff, J. O. 1980. The role of habitat patchiness in the population dynamics of snowshoe hares. Ecological Monographs 50:111-130. 179 Wood, C. C , and C. M. Hand. 1985. Food-searching behaviour of the common merganser (Mergus merganser) I: Functional responses to prey and predator density. Canadian Journal of Zoology 63:1260-1270. 180