UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Predation risk and the 10-year snowshoe hare cycle Hik, David Sherwood 1994

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1994-894245.pdf [ 1.81MB ]
JSON: 831-1.0088088.json
JSON-LD: 831-1.0088088-ld.json
RDF/XML (Pretty): 831-1.0088088-rdf.xml
RDF/JSON: 831-1.0088088-rdf.json
Turtle: 831-1.0088088-turtle.txt
N-Triples: 831-1.0088088-rdf-ntriples.txt
Original Record: 831-1.0088088-source.json
Full Text

Full Text

PREDATION RISK AND THE 10-YEAR SNOWSHOE HARE CYCLE by DAVID SHERWOOD HIK B.Sc. (Hons), Queen’s University, 1986 M.Sc., University of Toronto, 1988  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 1994 © David Sherwood Hik, 1994  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.  (Signature)  Department of The University of British Columbia Vancouver, Canada Date  DE-6 (2188)  1/  II  ABSTRACT I examined the effects of predation risk on the behaviour and population dynamics of snowshoe hares (Lepus americanus) during a cyclic peak and decline (1989-1993) near Kluane Lake, Yukon. Like most heavily preyed upon animals, snowshoe hares have to balance conflicting demands of obtaining food at a high rate and avoiding predators. The consequences of adopting predator avoidance behaviours under high risk of predation in winter may influence population dynamics of hares. Changes in patterns of winter habitat use, survival, body mass, and female reproduction were compared on four experimental areas: (I) where predation risk was reduced by excluding-out terrestrial predators (FENCE), (ii) where food supply was supplemented with ad lib rabbit chow (FOOD), (iii) a combination of these two treatments (FENCE+FOOD), and (iv) an unmanipulated CONTROL. Three hypotheses were compared. The food hypothesis predicts that hares use habitats with the highest amounts of food: body mass remains high, but survival is reduced. The predator avoidance hypothesis predicts that hares use habitats with the lowest risk: survival is high, but body mass decreases. The predationsensitive foraging (PSF) hypothesis predicts that both survival and body mass decline because a trade-off exists between predation risk and food availability. At peak densities hares used open habitats where food was readily available. However, as predation risk increased during the population decline, hares increased their use of safer, closed habitat and shifted their diet to include a greater proportion of poorer quality spruce twigs. This change in behaviour resulted in lower female body mass and reduced fecundity on the CONTROL area, even though sufficient winter forage was available. A similar decrease in body mass was observed on the FOOD treatment during the third year of the  III  population decline. On FENCE+FOOD, female body mass and fecundity remained high during the decline. Similarly, body mass did not decline on the FENCE treatment. These results supported the PSF hypothesis where terrestrial predators were present (CONTROL and FOOD), and the food hypothesis where terrestrial predators were absent (FENCE and FENCE+FOOD). Hares appear to have a limited ability to reduce exposure to predators because they have no absolutely safe refuge from predators, and they have limited reserves of energy during winter. Preliminary evidence suggests that physiological stress associated with high risk and poor condition is elevated during the population decline. I suggest that deleterious maternal effects mediated by predation risk may introduce a lag of one generation into the 10year population cycle of snowshoe hares.  iv TABLE OF CONTENTS page  Abstract  Ii  Table of Contents  iv  List of Tables  vii  List of Figures  viii  Acknowledgments  x  Chapter 1: The Predation Risk Hypothesis of the Snowshoe Hare Cycle 1 Hypotheses to Explain the Snowshoe Hare Cycle The Predation Risk Hypothesis  Chapter 2:  Food, Risk of Predation, and Patterns of Habitat Use by Snowshoe Hares  Introduction Methods Study Area Habitat Characteristics: Food and Cover Hare Survival Patterns of Habitat Use based on Live Captures Results Hare Density Ratio of Predators to Hares Prediction 1: Forage Availability Other Habitat Characteristics Prediction 2: Hare Survival Predictions 3 and 4: Habitat Use Discussion Significance of the Wolff Model Prediction 1: Winter Forage Prediction 2: Predation and Hare Survival Predictions 3 and 4: Habitat Use Cover and Other Factors Influencing Habitat Use Conclusions  1 4  8 8 13 13 14 15 16 17 17 17 22 28 30 37 41 41 41 43 44 45 47  V  Chapter 3:  Behavioural Responses of Snowshoe Hares to Changes in Food and Predation Risk  Introduction Hypotheses and Predictions Methods Study Area Habitat Use Movement: Home Range and Distance Between Browse Points Diet Body Mass, Time-constraints, and Sex Results Prediction 1: Hares Increase Use of Closed Habitat Prediction 2: Movements Decrease Prediction 3: Diet Quality Declines Prediction 4: Hares Increase Use of Open Habitat in Late Winter Prediction 5: Hares in Poor Condition Use More Open Habitats Prediction 6: Females Adopt Safer Foraging Strategies Discussion Prediction 1: Measuring Patterns of Habitat Use Prediction 2: Home Range and Movements Prediction 3: Diet Composition Predictions 4, 5, and 6: Time of Season, Body Mass, and Sex How Can Snowshoe Hares Minimize Risk of Predation? Conclusions  Chapter 4:  Does Predation Risk Influence Population Dynamics? Evidence from the Cyclic Decline of Snowshoe Hares  Introduction Hypotheses and Predictions Methods Results Winter Survival Body Mass Reproductive Output of Female Hares Discussion Evidence Supports the PSF Hypothesis Why do Body Mass and Reproduction Decline? Reproductive Costs of Predation Risk Predation Risk, Stress, and Population Cycles Conclusions  48 48 49 54 54 54 56 57 58 58 58 64 67 67 72 72 75 77 78 79 80 82 84  86 86 88 91 92 92 92 97 97 97 101 103 105 107  vi Chapter 5:  Population Cycles, Predation Risk, and the Ghost of Predators Past  What Causes Population Cycles? Food Predation Spacing Behaviour Multi-factor Explanations The Stress Hypothesis Revisited Summary of the Main Results of this Thesis Predator Avoidance and Population Dynamics Conclusions Literature Cited  108 108 109 109 110 111 112 116 120 122  124  vi’ LIST OF TABLES page Table 2.1 Forage availability, distance to cover, and snow depth on CONTROL, FENCE, and FENCE+FOOD  23  Table 2.2 Winter browse available to hares in mid-March and May  29  Table 2.3 Summary of hare mortalities determined from radio-telemetry data between January and April 1989-1 993 on CONTROL, FENCE, and FENCE+FOOD  35  Table 2.4 Manly’s alpha habitat preference index for hare trapping data between January and April 1988-1 993 on CONTROL, FENCE, and FENCE+FOOD  40  Table 3.1 Habitat use in late winter based on tracking data on CONTROL and FENCE+FOOD, 1990-1992  59  Table 3.2 Home Range size of males and females on CONTROL and FENCE+FOOD, 1989-1992  65  Table 3.3 Distance between browse stations on CONTROL and FENCE+FOOD, 1990-1992  66  Table 3.4 Summary of predictions and results of the food, predator avoidance, and predation-sensitive foraging hypotheses  76  Table 4.1 Mean litter size of hares at Kluane, Yukon, 1989-1992 on CONTROL areas and FENCE+FOOD  98  Table 4.2 Changes in body mass and survival on CONTROL, FENCE, FOOD, and FENCE+FOOD grids 1989-1 993, compared to predictions of food, predator avoidance, and PSF hypotheses  99  VIII  LIST OF FIGURES page Figure 2.1 Conceptual model of changing distribution of hares in closed, open, and shrub habitat over the course of the hare cycle (after Wolff 1981)  11  Figure 2.2 Hare densities in April 1988-1993 on CONTROL, FENCE, and FENCE+FOOD trapping grids  19  Figure 2.3 Ratio of predator abundance to CONTROL hare density at Kluane, Yukon, 1988-1993  21  Figure 2.4 Browse available in mid-March, 1991-1 993, in shrub, open, and closed habitat on CONTROL, FENCE, and FENCE+FOOD  27  Figure 2.5 Survival proportions estimated from radio-collar data between January and May, 1989-1 993 on CONTROL, FENCE, and FENCE+FOOD trapping grids  32  Figure 2.6 Habitat use by hares based on live-trapping captures during the period January to April 1988-1993 on CONTROL, FENCE, and FENCE+FOOD  39  Figure 3.1 Predictions of the FOOD, PREDATION RISK, and BALANCE hypotheses to explain foraging behaviour of snowshoe hares during winters 1989-1 992  52  Figure 3.2 Distance to cover while foraging on CONTROL and FENCE+FOOD, 1990-1992  63  Figure 3.3 Diet composition of hares on CONTROL and FENCE+FOOD, 1990-1 992  69  Figure 3.4 Effect of time of season on use of safer habitat on CONTROL and FENCE+FOOD by males and females, 1990-1992  71  Figure 3.5 Effect of body mass on use of closed habitat on CONTROL and FENCE+FOOD, 1990-1992  74  ix Figure 4.1 Predicted changes in body mass and winter survival under two levels of predation risk and food availability  90  Figure 4.2 Estimates of 30-day survival in winter on CONTROL, FENCE, FOOD, and FENCE+FOOD, 1989-1 993  94  Figure 4.3 Changes in body mass of female hares during January to May, 1989-1993, on CONTROL, FENCE, and FENCE+FOOD  96  Figure 5.1 Summary of key results in support of predictions of the predation risk hypothesis of the snowshoe hare cycle  118  x ACKNOWLEDGMENTS I owe substantial intellectual debts to many people who have helped me over the past years. I would especially like to thank everyone who was part of the Kluane Project for their suggestions and criticism, companionship, and help in the field and lab. Tony Sinclair challenged me, encouraged my ideas, and helped me to understand this fragile green planet we live on. I haven’t figured it out yet, but will keep trying. Rudy Boonstra was a constant source of inspiration about stress and hormones, and many other things. Charley Krebs was open to all of my crazy ideas and his vision of the Kluane Project was always inspiring. Stan Boutin and Jamie Smith were always available to talk about hares and plants and predators. They, along with the rest of my supervisory committee, Ken Hall and Don Ludwig, offered many valuable suggestions, critically reviewed my early proposals, and read drafts of this thesis. I am particularly grateful to Dennis Chitty for his interest in my work. John Boulanger, Karen Hodges, Alex End, Mark O’Donoghue, Don Reid, Christoph Rohner, and Locke Rowe made many helpful suggestions or read parts of this thesis. A number of people assisted with the field work and in many other ways, including Cathy Doyle, Frank Doyle, Remie Dionne, Judy Graham, Kerry Grisely, Robin McQueen, Ryan Patterson, Alistair Sarre, and Sabine Schweiger. Thanks! Anders Angerbjorn, Vilis Nams, and Os Schmitz provided good advice early on. And I am not sure I can ever express my appreciation to Fritz (‘let’s build a sauna’) Mueller and Nic (‘the refrigerator goes’) Larter, who went out of their way to make my life here and in the North memorable. Scott Gilbert provided me with a sanctuary on the other side of the runway. Andy and Carole Williams, and the rest of the Kluane Base crew shared great food, good company, and many ridiculous situations. Funding for this project was provided by the National Scientific and Engineering Research Council (NSERC) of Canada through CSP grants to C.J. Krebs et al., and operating grants to A.R.E. Sinclair. I was supported by an NSERC post-graduate fellowship, a University of British Columbia Graduate Fellowship, a Canadian Northern Studies Trust Fellowship, and the Department of Indian Affairs and Northern Development. All research was conducted under permit from the Wildlife Branch, Yukon Territorial Government, and facilities were maintained at Kluane Lake by the Arctic Institute of North America, University of Calgary. Most importantly, my parents provided love and encouragement throughout my studies. Thank you.  1  CHAPTER 1 THE PREDATION RISK HYPOTHESIS OF THE SNOWSHOE HARE CYCLE  The ‘10-year’ population cycle of snowshoe hares (Lepus americanus) and their predators across boreal North America is a remarkable natural phenomenon that has attracted the attention of naturalists and scientists for over 300 years (Finerty 1980). Fur return records from the Hudson Bay Company provide a long-term chronology of these cyclic fluctuations (MacLulich 1937; Elton and Nicholson 1942; Keith 1963), which appear to persist across the continent, with some regional exceptions (Smith 1983; Sinclair eta!. 1993). Many studies have considered the role of winter food, predation, and social behaviour as causes of the snowshoe hare cycle. Four major hypotheses based on these primary factors have been proposed (Keith 1990; Krebs et a!. 1992; Royama 1992), and I outline these below.  Hypotheses to Explain the Snowshoe Hare Cycle  (1) The prevailing explanation is the winter-food and predation, or Keith hypothesis described by Keith (1974, 1983, 1990), and Keith et al. (1984). At peak hare populations the depletion of winter food resources is supposed to lead to poor nutrition, and subsequently reduced fecundity and increased susceptibility to predation. Then, according to this model, there is a delayed density-dependent increase in predators, which continue to drive the hares to low numbers before declining themselves. Pease et a!. (1979) documented an absolute shortage of winter food for hares at the population peak, but nevertheless, body weights are highest at this time (Keith and Windberg 1978; Chilly 1987; Smith et al. 1988; Keith 1990). In a comparison of three cyclic  2  declines Keith et a!. (1984) concluded that food shortage is probably restricted to a period of a few months during late winter of the hare peak. Some models of the hare cycle have suggested that food limitation is a crucial factor (Akcakaya 1992), but do not distinguish between absolute food limitation and relative food limitation, where hares cannot get access to abundant food resources (Smith et a!. 1988; Royama 1992). It may be important to make this distinction to understand the mechanisms of the hare cycle. Relative food limitation implies that poor nutrition of hares during the decline is mediated by some factor other than direct starvation.  (2) The plant chemistry hypothesis (Bryant eta!. 1991a,b) suggests that qualitative nutritional changes in winter food plants may explain why hares suffer poor nutrition during the decline in spite of having a surplus of food available. Forage plants are defended by an array of antifeedant secondary compounds (Sinclair and Smith 1984; Sinclair et a!. 1988; Bryant et a!. 1991 a,b; Schmitz et aL 1992) that are known to result in poor nutrition of hares. However, Sinclair et a!. (1988) reported that while some plant defenses increased in forage species following the hare decline, there was no evidence that antifeedant chemicals caused the decline. At best, an increase in chemical defenses of forage plants may slow the recovery of hares during the low phase of the cycle.  (3) The polymorphic behaviour hypothesis (Chitty 1967; Krebs 1978) suggests that the spacing behaviour of hares themselves is a necessary condition for the occurrence of cycles. This hypothesis was not supported in an experimental test of its predictions (de Poorter 1984). Boutin (1980, 1984) also claimed that spacing behaviour was not a potentially limiting factor that could increase mortality and cause a decline. Results of behavioural experiments  3  (Sinclair 1986; Ferron 1993) suggested that hare populations are controlled by social behaviour only when food is limiting. Resident hares may limit juvenile immigration and recruitment (Boutin 1 984a) and react aggressively to unfamiliar conspecifics (Graf and Sinclair 1987), but these interactions appear to be related to the local availability of food. Social behaviour does not appear to act as a mechanism of self-regulation of hare population size (sensu Caughley and Krebs 1983).  (4) The predation hypothesis involves a delayed density-dependent interaction between predators and hares, according to which specialist and generalist predators increase in abundance some years after the hares’ increase because of their reproductive time-lags. These predators then face starvation as hares become increasingly rare, thereby generating the cycle (Keith et al. 1984; Hanski eta!. 1991; Royama 1992). Support for this hypothesis comes from recent modeling (Trostel et a!. 1987) and experimental studies which show that predation alone may be sufficient to generate the hare population decline (Krebs et a!. 1 986a,b). Winter food is not absolutely limiting for hares at any time during the cycle (Sinclair et a!. 1988, Smith et a!. 1988), and most hares died of predation rather than starvation (Boutin eta!. 1986). This predation hypothesis appears to be a reasonable explanation for the hare cycle. However, as Krebs et al. (1992) point out, there are two problems. First, Keith et a!. (1977, 1984) concluded that predation alone could not regulate hare numbers because, according to their calculations, predators alone could not account for the numerical losses of hares during the peak and decline phase of the cycle. Direct starvation must therefore be a significant factor in the hare decline. Recently, Boutin (1994) and C.J. Walters etal. (unpublished model) have suggested that predation rates and predator densities estimated in  4  previous studies may be too low, and consequently predation may play a larger role than previously postulated. The second argument against the predator hypothesis is that a significant reduction in reproduction during the hare decline (Cary and Keith 1979) cannot be a direct result of predation. Natality and juvenile survival are the two parameters most highly correlated with the rate of population change (Green and Evans 1940, Keith and Windberg 1978, Krebs et a!. 1 986a, Keith 1990, Royama 1992). Reduced fecundity is correlated with low body mass during the previous winter (Keith and Windberg 1978; Vaughan and Keith 1981), and the major factor determining winter body mass is the availability and quality of winter food supply. It has been suggested that predation risk may limit access to food resources, resulting in relative food shortage during winter (Wolff 1980; Keith et a!. 1984; Sievert and Keith 1985).  The Predation Risk Hypothesis  The effects of predators on prey populations are usually considered to be lethal, involving the removal of individual prey. However, there is increasing evidence that nonlethal effects of predators on the behaviour of prey may also be important, since many animals use poorer habitat and reduce their food intake in the presence of predators (for a review see Lima and Dill 1990). Anti predator behaviour of prey in response to increased predation risk may result in decreased fecundity or increased mortality caused by factors other than predation. There is considerable interest in how these predator avoidance behaviours influence predator-prey population dynamics (Hassell and May 1985; Ives and Dobson 1987; McNamara and Houston 1987, 1990; Abrams 1989,  5  1990, 1991, 1992a,b,c, 1993; Williamson 1993; FitzGibbon and Lazarus 1994), but this effect has been difficult to demonstrate experimentally. In this thesis I formulate and test a cohesive predation risk hypothesis of the snowshoe hare cycle. The main elements of this hypothesis have been anticipated for more than 50 years (MacLu lich 1937; Green and Evans 1940; Adams 1958; Wolff 1980; Keith et al. 1984; Sievert and Keith 1985; Krebs et a!. 1986a; Smith eta!. 1988; and others), and recent theoretical advances in understanding effects of food and predation on behaviour of prey (Ludwig and Rowe 1990; Lima 1992; Clark 1993, 1994) are used to integrate these earlier observations. Predation is clearly the proximate cause of death of most hares (Keith et a!. 1984; Boutin et a!. 1986). It also seems that observed declines in body mass and reproduction following the cyclic peak cannot be explained by absolute food limitation (Pease et a!. 1979; Smith et a!. 1988). The distribution of food and cover offering safety from predators may play an important role in determining habitat preferences and foraging behaviour of hares (Wolff 1980). There may be a decrease in relative food availability (e.g. a reduction in food intake or quality) if hare foraging activity is reduced as risk of predation is increased (Gilbert and Boutin 1990). The essence of the predation risk hypothesis is that those hares which survive the initial population decline or are born and recruited during this time, live in an environment of high predation risk. If hares adopt anti-predator behaviours that reduce foraging effort, this change may lead to decreases in body mass and fecundity even though food is abundant. The adverse consequences of poor condition may persist for more than one generation. Anti-predator behaviours may be reinforced throughout the life of a hare because predation rates are high and predator capture success is low. O’Donoghue and Krebs (1992) showed that mortality during the first 40 days of  6  life was 60-90%, and estimates of 30-day adult mortality rates averaged 10-20% (C.J. Krebs eta!., unpublished data). In addition, most adult hares have probably survived one or more encounters with a predator. For example, Murray (1990) reported that the precent of chases that result in a kill was 3381% for coyote, and 17-38% for lynx. The experiments described in this thesis were conducted as part of the Kluane Boreal Forest Ecosystem Project (Krebs et aL 1992), a series of community-scale, decade-long experiments focused on the effects of predators on the snowshoe hare cycle, and consequences of the hare cycle on the structure of northern boreal forest communities. I monitored the behaviour and demography of hares during the cyclic population decline when hare mortality is high (Keith 1990), with emphasis on the late winter (pre-reproductive) period, when a tradeoff between survival and reproduction in hares is expected for theoretical reasons (McNamara and Houston 1987; Ludwig and Rowe 1990; Clark 1993, 1994). In this thesis the following predictions of the predation risk hypothesis were examined: (1) Predation risk is not constant during the 10-year population cycle of snowshoe hares, and patterns of winter habitat use by hares reflect changes in predation risk in addition to the distribution of winter food resources (Chapter 2). (2) Hares, like other heavily preyed-upon animals, adopt foraging behaviours that minimize risk of predation (Chapter 3). (3) The adoption of anti-predator behaviours by hares in response to increased predation risk results in reduced body mass and fecundity during the decline  7  phase of the 10-year cycle, and contributes to the persistence of low hare numbers (Chapter 4). In Chapter 5 the evidence in support of the predation risk hypothesis and its role in generating population cycles of hares and other small mammals is reviewed. Particular emphasis is placed on the role of physiological stress associated with predation-sensitive foraging behaviour.  8  CHAPTER 2 FOOD, RISK OF PREDATION, AND PATTERNS OF HABITAT USE BYSNOWSHOE HARES INTRODUCTION The availability of food and protective cover often have a strong influence on the behaviour of vertebrate herbivores. Classical foraging theory suggests that animals select habitats that provide the highest rate of energy return (Krebs and Kacelnik 1991). Recent studies of animal foraging behaviour have emphasized the importance of predation risk in determining patterns of habitat use (Gilliam and Fraser 1987; Ludwig and Rowe 1990; McNamara and Houston 1987, 1990; Clark 1993). Many animals select environments offering protection from predators (e.g. dense vegetation) even if foraging efficiency there (e.g. intake rate) is lower (Lima and Dill 1990; Cassini and Galante 1992; Dickman 1992; Hughes et a!. 1993). Patterns of habitat use may reflect behavioural processes that result in reduced predation risk. In this chapter I examine differences in food, cover, and predation that may determine the patterns of habitat use by snowshoe hares in winter. Populations of snowshoe hares (Lepus americanus) fluctuate in number, with population peaks occurring at periods of 8 to 11 years throughout most of their range in the northern boreal forest (Keith 1963, 1990; Krebs eta!. 1986). Keith et a!. (1984) proposed that successive food-hare and hare-predator interactions could explain the ‘10-year’ cycle (see Chapter 1). Recent experimental studies of the snowshoe hare cycle have shown that the major proximate cause of hare mortality is predation (Keith et as’. 1984; Boutin et a!. 1986; Krebs et a!. 1986; Trostel et a!. 1987), and that winter forage is not absolutely limiting for hares at Kluane (Smith et al. 1988).  9  In this thesis I propose a mechanism for the interaction between food and predation based on changes in predation risk and resource distribution among different habitats. The spatial heterogeneity of habitats utilized by hares may influence the cyclic dynamics of these populations, since the environment often consists of a mosaic of refuges containing little food, surrounded by relatively unsafe areas where animals feed (Pullianinen 1983; Holmes 1984; Anderson 1986; Sih 1987; Lima and Dill 1990; Cassini and Galante 1992). If safe patches are too small, the herbivore (hares) may go extinct. Buehler and Keith (1982) suggested that large-scale clearing of land in Wisconsin may have caused local extinction of hare populations because patches of dense refuge habitat were insufficient to maintain viable populations through the cyclic low. In a further study, Keith et a!. (1993) suggested that the probability of local extinction in fragmented habitat depends on patch size and the number of resident hares. Predation risk has been recognized as an important factor influencing habitat use by hares (Keith et a!. 1984; Sievert and Keith 1985). Wolff (1980, 1981) developed a conceptual model of changing patterns of winter habitat use in which three habitat types were available to hares. In order of decreasing suitability based on predation risk, these were closed spruce forest, open spruce forest, and open shrub habitat (Fig. 2.1). In this model, utilization of open spruce and shrub habitat increases with hare density, and at peak hare densities all habitat is inhabited (darker shading in Fig. 2.1). Predation is highest in open habitats and hares decline there first. During the decline and low phase of the cycle, hares survive in closed spruce habitat, which provides refuge from predators. There is some empirical support for the patterns of habitat use described by Wolff (1980, 1981). First, many studies have found that hares are more abundant in dense, closed habitat than in more open habitats (Buehler and  10  Fig. 2.1 Conceptual model of changing distribution of hares in three different  habitats of decreasing suitability (closed  >  open  >  shrub), over the course of the  hare cycle (after Wolff 1981). Hare density increases with darker shading. Arrows indicate direction of dispersal movements between habitats.  11  LOW  INCREASING  4  EARLY PEAK  DECLINING  LATE PEAK  c1osed  12  Keith 1982; Orr and Dodds 1982; Wolfe et a!. 1982; Pietz and Tester 1983; Keith  et a!. 1984; Litvaitis et a!. 1985a,b; Scott and Yahner 1989; Litvaitis 1991; and others). Snowshoe hares are found in virtually all woody and brushy habitats during cyclic population peaks, but are restricted to densely vegetated habitat during population lows (MacLulich 1937; Keith and Windberg 1978; Wolff 1980, 1981; Litvaitis eta!. 1985; Keith 1990). Second, Sievert and Keith (1985) found that mortality due to predation was higher in open habitat. Finally, movements of marked snowshoe hares into dense cover were recorded during cyclic population declines (Keith 1966; Keith and Windberg 1978; Wolff 1980, 1981; Boutin 1984b). In this Chapter I describe changes in availability of forage and cover in different habitats, predation on hares in different habitats, and patterns of habitat use by hares during a cyclic decline in the southwest Yukon in order to evaluate three key predictions of the Wolff model. Prediction 1: Closed and open habitats differ in the amounts of forage available: more food is available in open habitats. Prediction 2: The survival rate of hares is higher in closed habitats than in open habitats. Prediction 3: Habitat use is determined by risk of predation rather than forage availability. Use of open shrub habitat increases at the population peak, but is reduced during the population decline in favour of more closed habitat.  Finally, terrestrial predators were excluded from two experimental study sites which allowed me to examine a fourth prediction of the effects of predation risk on patterns of habitat use by hares.  13  Prediction 4: When predation risk is reduced, habitat use by hares reflects the availability of food more than cover (predation risk). Therefore, hares use open habitat more often then they would in areas with higher predation risk.  METHODS  Study Area  This study was conducted between 1988 and 1993 as part of the Kluane Boreal Forest Ecosystem Project (Krebs et a!. 1992). This period spanned the peak (1989/90) and decline years of one hare cycle. Field sites were located in the Shakwak Trench along the Alaska Highway, east of Kluane Lake, Yukon, Canada (61° N, 138° W; c. 900 m a.s.l). The forest is dominated by white spruce (Picea glauca) with an understory of grey willow (Salix glauca) and bog birch (Betula glandulosa). These three plant species are the primary forage of snowshoe hares during winter (Smith et a!. 1988). Experimental feeding trials showed that hares prefer twigs of 5 mm basal diameter or less of mature Picea, followed by Betula, and then SaIix (Sinclair and Smith 1984). Hares were studied on three 34-ha experimental grids containing 400 grid points located in a 20 x 20 array: (i) an unmanipulated CONTROL (Sulphur), (ii) FENCE+FOOD, a 1-km 2 area surrounded by an electric fence to deter terrestrial predators (lynx, Lynx L canadensis, and coyote, Canis latrans), and provisioned weekly with pelleted rabbit chow (16% crude protein). Chow was distributed along four cut lines spaced evenly across the grid. Avian predators (mainly Great Horned Owl, Bubo virginianus, and Northern Goshawk, Accipitergentilis) 2 area surrounded had unrestricted access to this site, and (iii) FENCE, a 1-km  14  by an electric fence as above. Approximately 12-ha of this grid was covered by monofilament-line in an attempt to deter avian predators; however, much of this was buried by snow in late winter and therefore ineffective. Hares were live-trapped (Tomahawk Live Trap Co., Tomahawk, Wis.) on the three 34-ha trapping grids between January and May, 1988-1993. At least 86 traps were placed on four equally spaced rows across the grid. Traps were baited with alfalfa cubes, and hares were trapped over 1 to 6 days, at 2- to 4week intervals between January and May. We eartagged (No. 3 monel tags, National Band and Tag Co., Newport, Ky.), weighed, and determined the sex of all animals trapped. Hare densities were estimated from 5-6 day trapping sessions in April of each year using the mark-recapture estimators for a closed population (Otis et a!. 1978; Boulanger 1993). The effective trapping area was estimated to be 60 ha which includes a buffer of one home range (about 5 ha) around the edge of the trapping grid.  Habitat characteristics: food and cover  Between late February and mid-March 1991, the following habitat characteristics were recorded for an area of 15-rn radius centered on the 400 grid stations on each of the three grids: (i) habitat type (shrub:  <  cover; open spruce: 10 to 50% spruce cover; closed spruce:  50% spruce  >  10% spruce  cover); (ii) distance to nearest cover (places used by hares that provided cover on at least three sides and overhead; e.g. deadfall trees, large spruce trees, snow-burrows); (iii) the number of 5-mm stems of each species of browse , offset from each grid 2 (Picea, Salix, and Betula) available in an area of 3 m stake by 2-rn, and that were 80-cm or less above snow level; and (iv) snow depth. Measures of forage availability were repeated at approximately 100  15  randomly selected sites on each grid in mid-March 1992 and 1993. Differences between grid, habitat type, and year were compared using ANOVA (Wilkinson 1988). In order to convert stem density to twig biomass the fresh weight (fwt) and  dry weight (dwt) of fifty 5-mm stems of Salix and Betula were determined. The biomass of 5-mm stems of Picea was calculated from a relationship between stem diameter and stem mass of 86 twigs (r=0.80; C.J. Krebs, unpublished data). Biomass of 5-mm stems were 10.1±0.6 g fwt for Salix, 9.8±0.6 g fwt for  Betula, and 26.5 g fwt for Picea stems.  Hare su,vival  During trapping sessions, some hares were fitted with 40-g radio-collars equipped with mortality sensors (Lotech Inc., Newmarket, ON). Radio frequencies were monitored daily to determine survivorship of hares. On the death of a radio-collared hare, details of the cause of death and the location of the kill site were recorded (C. Doyle et a!., unpublished data). The proportion of hares surviving was calculated for each month (January to May) using the nonparametric Kaplan-Meier maximum likelihood estimator described by Pollock  et al. (1 989a,b), which allows for the staggered entry of animals during the study and censoring of data for lost radios. Usually 25-35 hares were radio-collared on each grid, but following the decline all hares (as few as 4) were radio-collared on CONTROL and FENCE. Survival distributions between treatments were compared using the Wilcoxon signed ranks test (Pollock et a!. 198gb).  16  Patterns of habitat use based on live-captures  At least 86 traps on each grid were placed systematically at grid stations and later assigned to one of the three habitat types (shrub, open spruce, closed spruce). The total number of hares captured during the first night of each trapping session during the January to April period was used to determine the proportion of hares utilizing each habitat. Only the first night of a multi-day trapping session was used in order to minimize problems associated with trapping hares, such as trap-saturation and increased stress. The number of hares captured during this period ranged from 7-217 on CONTROL, 10-131 on FENCE, and 47-230 on FENCE+FOOD, depending on hare density and the number of trap-nights (Table 2.4). An index of habitat use for each habitat (Pu) was calculated as 1 Pu  =  1 (P  -  )*Pai, where Pa 1 Pa 1 is the proportion of habitat i available, and Pc is  the proportion of hares captured in habitat i. A value of Pu 1  =  0 indicated that  there was no difference in the proportion of habitat available and hares captured there. Positive values indicated a preference and negative values indicated avoidance. For comparison, I also calculated Manly’s alpha as an index of habitat preference (Krebs 1989). This value was based on the capture frequency of hares in each habitat. Differences between the number of hares captured in each habitat in each year and the number of traps available in each habitat type were evaluated using the Kolmogorov-Smirnov test statistic (Wilkinson 1988).  17  RESULTS  Hare density  1 on Peak hare densities in April 1990 were approximately 1.2 hares ha1 on FENCE+FOOD 1 on FENCE, and 5.5 hares haCONTROL, 1.6 hares ha, respectively, 1 (Fig. 2.2). Hare densities declined to 0.2, 0.38, and 4.0 hares haby April 1992. Peak numbers on FENCE and FENCE+FOOD were reached one year after peak numbers on CONTROL.  Ratio of predators to prey  The ratio of an index of predator activity in the entire valley to CONTROL hare density increased each year during the increase, peak and decline phase of the 10-year cycle (Fig. 2.3), indicating that relative risk of predator encounters per individual hare increased over this time. Activity of lynx and coyote was estimated using the number of tracks observed per 100 km of winter transects (M. O’Donoghue eta!., unpublished data), goshawk activity by the number of birds seen in late winter per 1000 hrs of observation (Doyle and Smith 1994), and great horned owl abundance based on the number of resident individuals in 2 in spring (Rohner and Krebs 1994). It is possible that this ratio 100 km overestimates predation risk for hares during the late decline if predators search for alternative prey species (Stuart-Smith 1992). I have not tried to adjust the predation risk index to account for this possibility.  18  Fig. 2.2 Hare densities (number per ha  +  95% C.l.) in April 1988-1993 on  CONTROL, FENCE and FENCE+FOOD trapping grids.  19  7  6  5  = Ui C,) Ui  4  3  = 2  1  0 1988  1989  1990  YEAR  1991  1992  1993  20  Fig. 2.3 Ratio of predator abundance to CONTROL hare density in late winter at Kluane, Yukon (1988-1993). Predator index described in the text.  21  500  400  Ui  z 300 >Ui 0.  0  I  200  Ui •0.  100  0 1988  1989  1990  1991  YEAR  1992  1993  22  Prediction 1: Forage availability is higher in open habitat  Habitat was classified into three types based on spruce overstory: shrub (<10% spruce cover), open spruce (10%-50%) and closed spruce (>50%). ) in mid-March for each of 2 Forage availability (number of 5-mm stems in 3-rn these three habitats was generally highest for Salix, followed by Picea, on all three grids (Table 2.1). Betula twigs were rare on CONTROL and FENCE+FOOD, but more abundant on FENCE. Salix forage was generally. more abundant in shrub and open habitat than in closed habitat in all three years. Results of a three-way ANOVA for Picea twigs, with main effects of grid, habitat type, and year, indicated that only the effect of grid was significant 2 17421 0.13, P.<0.001). No other main effect and interaction terms were (F significant (P>0.10). For Salix, two of three main effects were significant (grid: 21742 F 21742 P=0.018; year: =8.81, F 21742 P=0.051; habitat: =4.04, F =2.98, P.<0.001). The interaction between grid*habitat and habitat*year were also significant (P<.009), owing to the lower twig biomass on FENCE+FOOD. Forage availability (stem density) of all species decreased in all habitats between 1991 and 1992. In 1993 there was an increase in the number of Picea and Salix stems on FENCE+FOOD. The biomass of forage available per individual hare was calculated based on the number of stems available to hares in each habitat in mid-March. These biomass values were calculated for each habitat and summed, then divided by April hare density (Fig. 2.2). In addition, I also assumed that 60 days of woody browse were required to maintain hares until the end of the season, and that hares require about 300 g (fwt) of forage per day to maintain body mass (Pease et al. 1979). On this basis, the forage available each year was at least ten times the amount required on CONTROL for Picea and Salix three times the amount  23 Table 2.1 Forage availability, distance to cover, and snow depth. Availability of 5-mm ) of Picea, Sallx and Betula, distance to cover (m), and snow depth (cm) 2 stems (in 3 m in shrub, open spruce, and closed spruce habitats in March 1991 to 1993 on CONTROL, FENCE+FOOD, and FENCE grids. Mean values ± 1 S.E. Sample size in brackets. Approximately 400 quadrats were sampled in 1991, and 100 quadrats in 1992/93.  2 # stems /3 m  CONTROL Habitat  N  Year  Picea  Salix  Betula  Distance to Cover 15.0± 1.7  Snowdepth  9.8± 2.7  0.4± 0.2  (12)  1991 5.8± 2.2 1992 1.2± 0.7  3.5±0.7  0  48±7  (8)  1993 1.5± 0.8  2.5±0.6  0  52±7  6.9± 1.0  0.1± 0.01  (55)  1991 4.4± 0.7 1992 3.2± 0.8  5.1± 1.2  0.02± 0.02  44±3  (60)  1993 2.1±0.4  3.6±0.8  0  49±2  CLOSED (160)  1991 3.9± 0.7  0.01± 0.01  (35)  1992 1.7±0.5  3.8± 0.7 2.7± 1.1  0  45±5  (35)  1993 2.9± 1.0  1.9±0.6  0  43±3  SHRUB (37)  OPEN (203)  10.2± 0.5  10.3±0.6  52±2  48±1  43±2  24 Table 2.1 continued  FENCE Habitat  2 #stems/3m N  Year  Picea  Salix  Betula  Distance to Cover  16.4± 1.5  Snowdepth  1991 0.8± 0.4  4.2± 1.0  2.5± 0.7  (37)  1992 0.5±0.5  2.1±0.7  0.1±0.1  71±4  (15)  1993 0  3.0± 1.4  2.5± 0.5  87±4  OPEN (195)  1991 0.9±0.2  3.3± 0.3  0.9± 0.2  (82)  1992 0.7±0.2  1.7±0.3  0  65±3  (57)  1993 1.0±0.5  5.4±0.9  1.3±0.6  72±2  2.2± 0.5  0.1± 0.1  (10)  1991 0.8± 0.4 1992 0.2±0.2  1.1±0.5  0  54±4  (28)  1993 0.8± 0.4  2.0± 0.6  2.8± 1.3  68±3  SHRUB (154)  CLOSED (34)  7.6± 0.5  6.2± 0.6  25 Table 2.1 continued  # stems /3 m  FENCE+FOOD Habitat  N  SHRUB (92)  Year  Picea  1991 0.2±0.1  Salix  14.1±2.4  Betula  Distance to Cover  0.6±0.3  15.5± 1.5  Snowdepth  66±2  1992 0.03±0.03 2.4± 0.5 1993 1.0±0.35 6.3± 1.5  0  74±3  0  80±4  OPEN (289)  1991 0.9±0.1  0.5± 0.1  (65)  1992 0.3±0.1  3.3± 0.5 1.5±0.3  0  --  (72)  1993 1.3± 0.4  3.3± 0.5  0  --  1991 0  2.9± 1.0  0  (5)  1992 0.3±0.3  0.7±0.7  0  --  (4)  1993 1.0±0.7  0  0  --  (35) (27)  CLOSED (16)  9.8± 0.5  8.7± 1.7  59±2 67±3 75±2 56±5 47±21 64± 19  26  1 day ) in mid-March 1991-1 993 based on 1 Fig. 2.4 Browse available (kg hare number of 5-mm stems of Picea, Salix and Betula in each habitat (shrub (S), open (0), closed (C)), on CONTROL, FENCE and FENCE+FOOD trapping grids. Dashed line indicates daily requirement of 300 g per hare, and x indicates no browse available. Bars, from left to right, indicate values for 1991, 1992, and 1993.  27 100cOF’ffROL 10—  H  a  F  i-i  a  I- —I  :  4.——  01  o.olJ  0.001  S  0 Picea  S  C  0 Salix  C  ELFRrL S  0 C Betula  100 FENCE 10-  F 2 -  1-  0.1-  0  0.01-  0.001  S  0 Picea  S  C  0 Salix  C  S  a  0 C Betula  100  FENCE+FOOD 10-  a  0  a  1 I  0.01-  0.001•  NH NH S  S  Picea  0 Sd&  C  ft H S  0 C Bettia  28  required on FENCE, but less than the amount required on FENCE+FOOD, particularly in 1992 (Table 2.2). Total forage available on each grid (Table 2.2) generally increased as hare numbers declined. Amounts of Picea available were lowest on FENCE÷FOOD, but large quantities were available on CONTROL and FENCE. Estimates of forage available on FENCE+FOOD are lower, but this grid was supplied with ad lib rabbit chow (16% crude protein). Forage available in mid-March in each habitat (Fig. 2.4) also appears to be sufficient for hares, except for closed habitat on FENCE+FOOD. Forage available in mid-March was sampled only from 1991-1993. In order to compare these amounts with forage available during the peak years (1989/90), I calculated amounts of Salix and Betula browse available at the end of each winter. These estimates are based on the dry weight of 5-mm stems estimated from clip-plots at the end of winter (early May; C.J. Krebs et a!., unpublished data). The results show that even after the winter of peak hare density amounts of woody browse available to hares ranged from 68 to 280 kg 1 on FENCE+FOOD and CONTROL (Table 2.2). Measures were not hareavailable for FENCE, but the same pattern was apparent (pers. obs). The biomass values estimated at the end of winter were higher than in mid-March because the May values include forage buried under snow and therefore not continuously available to hares. In summary, winter forage was not absolutely limiting for hares at Kluane, a result consistent with that of Smith et a!. (1988).  Other habitat characteristics: distance to cover and snow depth  There was no difference in the distance to cover between open and closed spruce habitats on any of the grids (P>0.83), but these distances were significantly less on FENCE, averaging 7 m compared with about 10 m on  29 Table 2.2 Winter browse (5-mm stems of Salix, Betula, Picea) available to hares (above snow) in mid-March (kg fwt hare1 day), and in May at end of winter (kg dwt 1 ). Water content of browse is approximately 50% (i.e. 1 kg dwt 2 kg fwt). 1 hareSalix MARCH (kg fwt hare 1 day) CONTROL FENCE+FOOD  FENCE  MAY (kg dwt hare) 1 CONTROL FENCE+FOOD  1989  184  1990  280  438  1991  3.6  0.6  1.3  280  68  1992  9.8  0.2  2.5  1304  40  1993  22.6  1.9  15.0  3929  845  Betula MARCH (kg fwt hare1 day ) 1 CONTROL FENCE+FOOD  FENCE  MAY (kg dwt hare) 1 CONTROL FENCE+FOOD  1989  0.8  1990  0.8  57.3  1991  0.05  0.06  0.4  0  20.3  1992  0  0  0.03  8.7  13.6  1993  0  0  5.6  Picea MARCH (kg fwt hare day’) CONTROL FENCE+FOOD  FENCE  1991  6.9  0.2  0.9  1992  15.5  0.1  2.3  1993  49.8  1.6  6.4  71.4  199.1  30  CONTROL and FENCE+FOOD (P <0.002). Distance to cover was greatest in shrub habitat, averaging 15-16 m on all grids (Table 2.1). Results of a two-way =53.44, P.<0.001), since 21116 ANOVA indicated a significant effect of habitat (F distance to cover was less in open spruce and closed spruce habitats. Snow depth was lower on CONTROL than on FENCE or FENCE+FOOD in all three years (Table 2.1). CONTROL grid is located 5 km east of the other grids and at slightly lower elevation, therefore local differences in snowfall could explain this difference. On CONTROL, snow depth was similar in all years, but was consistently higher in shrub and open spruce habitat than in closed spruce habitat (P.<0.001). The same pattern among habitats was observed on FENCE+FOOD and FENCE where snow depth varied by 10 tol5 cm between years.  Prediction 2: Hare suiviva! is higher in closed habitat  In 1989, survival between January and May on CONTROL was similar to that on FENCE (Wilcoxon signed ranks test: P=0.08), but was significantly lower than that on FENCE+FOOD (Wilcoxon test: P<0.05), particularly in February and March (Fig. 2.5). However, in 1990 survival on CONTROL was generally similar to FENCE and FENCE+FOOD. In 1991, the first year of the decline, survival of hares on the CONTROL grid dropped significantly lower than that on FENCE and FENCE+FOOD (c. 60% on CONTROL compared with 80-90% on the two fenced grids between January and May; Wilcoxon test, P.<.05). In 1992 survival was lower on CONTROL than on FENCE+FOOD, but was similar to that on FENCE. In 1992, survival on FENCE was also significantly lower than on FENCE+FOOD (P<0.05). In 1993 survival on the three grids was similar (P>0.5).  31  Fig. 2.5 Hare survival each month estimated from radio-collar data during the period January to May, 1989-1993 on CONTROL, FENCE, and FENCE+FOOD trapping grids.  SURVIVAL  30-DAY  .  0.3  1992  1991 1990  .  1989  .  J FMAM  .  J FMAM  .  J FMAM  •  J FMAM  .  FENcE+FcDo cXDNTFOL QFWDE  o •  0.4  0.5  0.6  II \  •  •  •  1993  J FMAM  .  CI)  33  On FENCE+FOOD survival showed a small decline in 1991 and 1992 compared with peak years. On FENCE, survival was also lower in 1991 than in 1992. In 1993 survival of hares on CONTROL fluctuated widely between months (1.0 to 0.39), but was higher overall than in 1992. Survival on FENCE in 1993 was similar to that in 1992, while survival on FENCE+F000 was lower than in 1992. Increased use of the predator reduction grids by avian predators may account for much of the observed decrease in survival on these areas in 1993 (Ch. Rohner, pers. comm.). Overall, survival of both sexes was similar. More dead hares located during the January to April period were in shrub or open habitat than in closed habitat (Table 2.3). On CONTROL, the proportion of kills in shrub and open habitat was 50-100%. About 25% of mortalities occurred in closed habitat, except in 1992 when 50% (5/10) occurred there. The increase in the proportion of hares killed in closed habitat in 1992 reflects increased use of this habitat by hares (see Fig. 2.6 below). On FENCE÷FOOD the sample of hares killed was small except in winter 1993 when 12 kills were observed; 67% of these occurred in shrub habitat. No kills were recorded in closed habitat. On FENCE the majority of kills also occurred in shrub or open habitat. No statistical tests were conducted because of small sample sizes. During this winter period of 1989 to 1993, 83% of all recorded radio-collar mortalities were due to predation, 9% were attributed to starvation, and the remaining 8% were due to unknown or accidental causes. Both terrestrial (41%) and avian (49%) predators contributed equally to mortality on CONTROL (Table 2.3). On FENCE, there were more kills by great horned owls (61%) than by goshawks (18%), but on FENCE+FOOD these avian predators accounted for an equal proportion of kills (about 30% each). Most goshawk predations took place during the population peak and first year of decline, but were also observed in 1992 and 1993 on FENCE+FOOD. The majority of hares suspected of starving  34  Table 2.3 Summary of hare mortalities determined from radio-telemetry data between January and April on CONTROL, FENCE, and FENCE÷FOOD, 1989-1993. Habitats are shrub (S), open (0), and closed (C). Cause of death was recorded as coyote (CY), lynx (LX), goshawk (GOS), great horned owl (GHO), other AVIAN, unknown predator (UNK), suspected starvation (STARVE). Sample size (N) is indicated for (a) the total number of hares predated, and (b) the total number of mortalities (includes number recorded as starved).  CONTROL HABITAT (%) Year  N  1989 1990 1991 1992 1993  9 6 25 10 2  1989-1993  S  0  C  44 22 33 50 17 33 52 28 20 40 50 10 01000  CAUSE (%) N 9 6 30 10 2  CY  LX  GOS GHO AVIAN UNK STARVE  22 0 56 0 50 17 7 27 13 10 50 10 01000 9  32  19  0 33 17 20  22 0 20 0 00  16  14  0 0 0 10  0 0 17  0  0 0  2  9  35  Table 2.3 continued  FENCE  HABITAT (%) Year  N  1989  3  0  100  1990 1991  5 18  20 44  1992  11  1993  4  1989-1993  S  0  C  CAUSE (%) N  CY  LX  0  4  0  0  25  0  50  25  60  20  5  0 0  0  20  0  0  50  6  20  0  0  35  80 45  5  5  0 10  27  55  18  11  0  0  0  100  0  0  0  75  25  0  4  0  0  0  50  0  25  25  0  0  18  61  2  9  9  G0S GHO AVIAN UNK STARVE  36 Table 2.3 continued  FENCEi- FOOD CAUSE (%)  HABITAT (%) Year  N  S  0  C  N  CY  LX  GOS GHO AVIAN UNK STARVE  1989 1990 1991 1992 1993  3 2  33 50 20 50 67  67 50 80 50 33  0 0 0 0 0  5 2  0 0 0 0 0  0 0 0 0 0  0 50 60 43 17  0 0 0 14 75  0 0 40 14 0  60 50 0 14 8  40 0 0 14  0  0  29  32  10  19  10  5 6 12  1989-1993  5 7 12  0  37  (5/11) occurred on CONTROL during 1991, the first year of the population decline. Hares recorded as starved were found dead with no external injuries. Since the cause of death could not be proven conclusively and may actually have been caused by other factors (i.e. disease, other stress), the number of hares reported starved should be considered a high estimate. It is also possible that some of these other factors may be starvation-induced.  Predictions 3 and 4: Use of closed habitat increases during the hare decline  There were three general trends in the patterns of habitat use by hares (Fig. 2.6).  First, shrub habitat was usually avoided (P 1  <  0). Second, as  predation risk increased hares decreased use of open spruce habitat on CONTROL and FENCE. Third, hares used closed spruce habitat increasingly between 1988 and 1993. These results are unlikely to be a consequence of trap saturation as no more than 62% of all available traps were filled on any one evening, even on FENCE+FOOD where hare densities were highest. Habitat preferences based on Manly’s alpha (Table 2.4) showed the same pattern as the index of habitat use (P). Shrub habitat was least preferred on all three grids, and overall there was a decrease in use of open habitat and a concomitant increase in use of closed habitat during the decline. The results of Kolmogorov-Smirnov tests comparing patterns of habitat use and availability based on winter trapping data showed significant differences (P<0.001) in 1988, 1989,1992, and 1993 on FENCE; 1992 and 1993 on CONTROL; and 1992 on FENCE+FOOD. These significant differences occurred mostly during the late decline phase of the cycle (1992/93) on all grids. Hares preferred closed habitat to more open habitat on CONTROL, where both terrestrial and avian predators were present. A similar pattern was observed  38  Fig. 2.6 Habitat use by hares based on live-trapping captures during the period  January to April, 1988-1993 on CONTROL, FENCE and FENCE+FOOD trapping grids. Shaded bars (1989/90) indicate years of peak hare density on CONTROL. Positive values indicate preference, negative values indicate avoidance.  0.5  1.0  SHRUB  CONTROL  OPEN  CLOSED  ri[fi  0.5  1.0  SHRUB  FENCE  OPEN  []  CLOSED  0.5  1.0  SHRUB  11  OPEN  fi 1 tL.  FENCE-i-FOOD  CLOSED  40 Table 2.4 Manly’s alpha habitat preference index based on proportion of hares captured in each habitat between January and April, 1988-1993, on CONTROL, FENCE, and FENCE+FOOD (N = total number of hares captured). Habitat types are shrub (S), open spruce (0), and closed spruce (C).  CONTROL Year  N  S  0  FENCE C  1988 (21) 0.17 0.44 0.39 1989  (40) 0.08 0.51  1990 (217) 0.13 0.41 1991  (81) 0.21  0.41  NS  0  FENCE+FOOD C  (10) 0  0.79 0.21  (38) 0.11  0.43 0.46  N  SO  ---  (51) 0.21  C  ---  0.79 0  0.46 (120) 0.19 0.47 0.34 (201) 0.33 0.40 0.27  0.34 0.45 (131) 0.18 0.42 0.40 (230) 0.37 0.05 0.57  1992 (22) 0.19 0.28 0.47  (42) 0.08 0.30 0.62  (68) 0.09 0.08 0.83  1993  (30) 0.17 0.23 0.60  (47) 0.06 0.75 0.19  (7) 0  0.33 0.66  41  inside the fences where only avian predators were present (Fig. 2.6)  DISCUSSION  Significance of the Wolff model  The results of this study provide evidence that predation risk, rather than changes in food availability, can account for observed shifts in habitat use over the hare cycle. Hares increased their use of habitat with lower food (i.e. stem density and biomass) and higher cover during the hare decline. Several authors have suggested that this sort of habitat heterogeneity may influence population dynamics (Rosenzweig and Abramsky 1980; Sih 1987; Morris 1988; Oksanen et al. 1992; Ostfeld 1992). The late winter period is perhaps the most critical time of the year for hares, because female condition at this time is positively correlated with reproductive output. Although the Wolff model makes no quantitative predictions, it allows testable predictions of the mechanisms (i.e. dispersal, predation, food shortage) responsible for this pattern to be examined.  Prediction 1: Winter forage is most abundant in open habitats  The results of this study support the first prediction of the Wolff hypothesis, that winter food is more available in open habitats. It also appears that winter forage is not absolutely limiting for snowshoe hares. If forage were absolutely limiting, hares would suffer mortality from direct starvation. Relative food limitation implies that poor nutrition is mediated by some factor other than outright food shortage causing starvation. Hares require about 300 g of mixed species forage (fresh wt) per day to  42  maintain body mass (Pease et a!. 1979), and results in Table 2.2 and Fig. 2.2 show that sufficient winter food is available for hares even during the cyclic peak. These results are consistent with those of Smith et a!. (1988). However, Pease et a!. (1979) and Wolff (1980) suggested that some hares actually ran out of food at the peak. One explanation that could account for their results is that peak hare densities were higher at Rochester, Alberta in the 1960’s and 1970’s, than at Kluane, Yukon during the 1980’s and 1990’s (Keith 1990; Fig. 2.2). In Fairbanks, Alaska, peak hare densities were similar to those at Kluane, but Wolff (1980) did not include Picea, which is an important component of winter diet (Wolff 1978), in his estimates of forage availability. Experimental studies at Kluane suggest that food shortage is not a necessary cause of the cycle (Krebs  eta!. 1986; Sinclair eta!. 1988). Although it is possible that some hares did run out of food (Sinclair et al. 1988), this was not because food was absolutely limiting. Even though forage appears to be available (Table 2.2), body mass of hares decreased during the population decline (Keith et a!. 1984; Smith et a!. 1988; Chapter 4). One possible explanation is that forage availability is overestimated based on the methods described above.  Forage quality may be  poor because plants are defended by an array of antifeedant compounds (Sinclair and Smith 1984; Bryant et a!. 1985; Sinclair et a!. 1988; Schmitz et a!. 1992). Sinclair et a!. (1988) reported that the amounts of some secondary defenses increased in forage plants at Kluane following the hare decline, and therefore may have contributed to poor nutrition at this time. The diet of hares on CONTROL shifted from predominately Sa!ix twigs in 1990, to one dominated by Picea in 1992 (Chapter 3). In short term feeding trials, captive hares were unable to maintain body mass on a diet of Picea twigs, and could just maintain it on a diet of Salix twigs (Rogers and Sinclair 1994); the observed shift in diet may  43  account in part for observed declines of body mass on CONTROL (Chapter 4). If quality and digestibility of twigs change over the hare cycle, estimates of forage biomass may not be the best measure of winter forage availability. Hares may also lose mass during the decline because foraging efficiency is lower if predation risk reduces time spent foraging. Hares may forage less efficiently if winter browse is less available during the decline as the spatial patchiness of browse increases (e.g. twig density declines as shown in Table 2.1). The observed decrease in survival on CONTROL during the first year of the hare decline (Fig. 2.5) may also, in part, reflect differences in social dominance of hares interacting with the relative availability of food (Boutin 1984b; Sinclair 1986; Ferron 1993). Subordinate individuals may be excluded from food and cover, and therefore suffer higher predation (Keith et a!. 1984).  Prediction 2: Predation and hare suivival  The second prediction, that predation is higher in open habitats, was also supported. Results showed that 70% of all mortalities occur in open habitats (Table 2.3), suggesting that differences in cover between habitats are important to survival of hares in winter. Many of the predators at Kluane also showed habitat use preferences. Rohner and Krebs (1994) found that great horned owls killed more hares in open habitat types at Kluane. They concluded that owls avoid, or have less hunting success, in closed forests and shrub with dense cover. Longland and Price (1991) also reported that capture success of great horned owls hunting heteromyid rodents was higher in open habitat than in brushy, closed habitat. In their study, the proportion of successful attacks ranged from 5% to 60%. Similarly, Murray (1990) found that during the hare population increase (1 987-1988), coyote and lynx used open habitat significantly  44  more than closed habitat, but the proportion of chases that resulted in a kill was similar in both habitat types. However, hares were also killed in closed habitat, particularly during the late decline. Therefore, closed spruce habitat does not provide absolute refuge from predators, but use of these areas may increase the probability of survival. The behavioural mechanisms by which hares might be able to reduce risk of predation are discussed in Chapter 3.  Predictions 3 and 4: Habitat use  The third prediction of the Wolff model, that shrub habitat is generally avoided and use of closed habitat increases when predation risk is higher, is supported by live-trapping data (Fig. 2.6). These results indicate increased use of more closed habitats during the hare decline. Based on the amounts of forage available in each habitat type, classical foraging models (Krebs and Kacelnik 1991) suggest that hares might prefer open spruce habitat, where foraging rates are higher than in shrub or closed spruce habitat. This pattern does not emerge from the results. Rather, when terrestrial predators were present, hares displayed a preference for closed, relatively safe habitat over riskier open habitat. Absolute amounts of available browse appear to be sufficient (Smith et a!. 1988, Table 2.2), but relative availability may be reduced if predation risk limits access to forage. The spatial distribution of twigs increased during the decline and this may result in lower foraging efficiency (e.g. increased travel time between browse patches; Chapter 3). Support for the fourth prediction, of increased use of open habitat on FENCE and FENCE+FOOD, was weak. This may be explained in part by the activity of avian predators on both fenced areas (Table 2.3). Even though terrestrial predators were removed from inside the fences, no pattern of habitat  45  preference or avoidance was detected at peak numbers. As CONTROL density dropped and the intensity of avian predation increased on the fenced grids (Table 2.3), hares shifted to more closed habitat, as they did in unfenced areas. These results provide additional evidence that hares are influenced by risk of predation. There are several problems in interpreting patterns of habitat use based on trapping data. Alfalfa baits may attract hares to habitats where they would not normally be found, but this problem was minimized by using a large number of traps and thereby avoiding trap saturation. Trapping may over-estimate use of open habitat by drawing animals into areas they would not normally travel (Kotler 1985; Hughes et a!. 1993); however, Boulanger (1993) found no evidence that snowshoe hares could be drawn off of their normal home ranges into traps. Despite these potential problems, there is still a strong trend in the data suggesting that hares increasingly preferred closed habitats as predation risk increased. In general, shrub habitat is under-represented in the trapped sample, and there is an increase in use of closed habitat. Other methods of determining habitat use confirm this pattern (Chapter 3), and Litvaitis et a!. (1985a) found that pellet counts, track counts, and live captures provided similar information about the use of habitat types by snowshoe hares.  Cover and other factors influencing habitat use  Density of understory cover has been shown to be a key component influencing the distribution of snowshoe hares in many studies (Bider 1961; Buehler and Keith 1986; O’Donoghue 1983; Wolfe eta!. 1982; Wolff 1980), and preference for closed habitat by snowshoe hares appears to be a general result. Litvaitis et a!. (1 985b) and Pietz and Tester (1983) found that patterns of habitat  46  use by hares were based primarily on the density of understory species rather than composition of forage species. Similarly, Buehler and Keith (1982) and Rogowitz (1988) reported that the availability of cover influenced habitat use more significantly than the availability of food. In western New York state, hares preferred sites with a well developed overstory of mature spruce even though forage availability was limited (Rogowitz 1988). Scott and Yahner (1989) reported that habitat use by hares was positively correlated with distance to cover and forage availability. Several studies (Litvaitis eta!. 1985; Wolfe etal. 1982; Pietz and Tester 1983; Sullivan and Moses 1986; Carreker 1985) have reported that hare population densities were highly correlated with understory cover> 2-m tall providing at least 40% visual obstruction. At Kluane the shrub understory rarely provides this amount of cover and therefore overstory vegetation is a better index of habitat suitability. Understory shrubs provide forage for hares, but overstory spruce provides most cover. Although overall cover was higher in closed habitat (by definition), distance to point cover was similar in both closed and open habitat. Several other factors might influence foraging decisions of hares, including snow depth, temperature, and moonlight. Snow accumulation may decrease forage availability and cover, but snow also facilitates access to forage that normally would be out of reach (Keith 1990). Snow burrows are also used extensively by hares for shelter. Kluane is in the rain shadow of the St. Elias Mountains and mean snow depth is shallow, averaging 60 cm and rarely exceeding 1 m. Hares also excavate craters up to 36 cm deep to obtain buried forage (Gilbert 1990). Deeper snow may also reduce the hunting success of some predators (Murray and Boutin 1991; Huggard 1993). Pease et al. (1979) and Keith (1990) reported that mortality of malnourished captive and wild hares was significantly related to ambient  47  temperature. Low temperatures limit foraging and lead to a decrease in condition. Since snowshoe hares cannot maintain reserves of energy for more than 3-4 days (Whittaker and Thomas 1983), periods of lost foraging due to low temperature may require hares to increase risk during subsequent foraging bouts. Gilbert and Boutin (1991) found that snowshoe hares reduced activity in open areas, away from cover, during bright moonlit nights. This well known response of small mammals to moonlight may modify predator-prey interactions by increasing prey vulnerability. The combined effects of low temperature and bright moonlight may greatly reduce time spent foraging by hares. This aspect of foraging ecology requires further investigation.  CONCLUSIONS  The results presented in this chapter support the conceptual model of patterns of habitat use by hares in winter (Wolff 1980, 1981). Habitat use by hares appears to be determined by risk of predation from both terrestrial and avian predators, rather than by availability of winter forage. Boutin et al. (1985) reported that less than 10% of hares dispersed off their home range, therefore differential predation pressure between habitat types may generate the observed pattern. More information is needed about the foraging success of hares foraging in different habitats, and the capture success of predators hunting in different habitats. The mechanisms by which hares may be able to avoid predators and the population-level consequences of predation risk are discussed in the following chapters.  48 CHAPTER 3 BEHAVIOURAL RESPONSES OF SNOWSHOE HARES TO INCREASING RISK OF PREDATION INTRODUCTION Predation risk is defined as the probability of being killed during some time period (Lima and Dill 1990). Predation risk may influence the time of day animals feed, the habitat patches utilized for feeding, the composition of the diet, and other aspects of behaviour. When predation risk is high, many prey species have been observed to reduce foraging effort and increase use of protected habitats (Lima and Dill 1990). The ability of animals to reduce predation risk may depend upon both their ability to detect predators, and the feasibility of alternative foraging strategies. Prey species may evaluate predation risk using cues based on illumination (Gilbert and Boutin 1991), the presence of predator odours such as faeces or urine (Sullivan et al. 1985; Dickman 1992; Jedrzejewski et aL 1993), observation of a predator, or predator vocalizations (Curio 1993; Kotler 1992), predation on neighbours (Charnov etal. 1976), or warning by conspecifics (Underwood 1982; Lazarus 1990). Yet, no matter how successful prey species are at assessing predation risk, their options for mitigating this risk may be limited. Recent theoretical work has suggested that the range of anti-predator behaviours available to prey species will often depend on their own condition, and the time remaining until reproduction (Ludwig and Rowe 1990; Clark 1993, 1994). Individuals in danger of starvation, or of not attaining a certain condition (e.g. fat level, body weight) by the end of a fixed period (e.g. end of winter), may incur increased predation risk in order to obtain sufficient food. Models which incorporate the current condition of an animal and the time remaining to achieve a certain minimum  49  condition (time-constraints) have been used to predict changes in patterns of habitat use (McNamara and Houston 1987, 1 990a; Ludwig and Rowe 1990; Clark 1993, 1994; and references therein). Winter food availability and predation may both be important components of the ‘10-year’ population cycle of snowshoe hares (Lepus americanus) in the northern boreal forest (Keith 1990; Krebs et a!. 1992; Chapter 1). In Chapter 2, I showed that there is spatial heterogeneity in winter food availability and predation risk among habitats used by hares: safer closed habitats contain less food than riskier open habitats. This heterogeneity may lead to a trade-off between food and predation risk since hares must balance their energy budget on a short-term (c. 72 hour) basis (Whittaker and Thomas 1983) while simultaneously avoiding predation, the proximate cause of most mortality (Keith  eta!. 1984; Boutin etal. 1986). In this chapter I describe patterns of foraging behaviour of snowshoe hares under different levels of predation risk during the late peak and early decline of the 10-year population cycle (1990-1992). During the cyclic population decline, predation risk (i) increased naturally, and (ii) was experimentally reduced in one area by excluding terrestrial predators. Supplemental food (rabbit chow) was also provided there.  Hypotheses and Predictions  The predictions of three hypotheses can be used to contrast the foraging behaviour of hares. The food hypothesis predicts that animals use habitats with the highest food availability (where energy gain is maximized; Krebs and Kacelnik 1991). The predator avoidance (PA) hypothesis predicts that animals use habitats with the lowest predation risk (minimize time spent in riskier  50  habitats). The predation-sensitive foraging (PSF) hypothesis predicts that animals balance predation risk and food availability when choosing habitats (McNamara and Houston 1987; Ludwig and Rowe 1990); as predation risk increases, so does use of safer habitats. Predictions of these three hypotheses are contrasted in Fig. 3.1. If closed habitat is relatively safe, but has less food or food of lower quality than more open habitats (see Chapter 2), then the food hypothesis predicts that closed habitat is avoided, independent of predation risk. The predator avoidance hypothesis predicts that closed habitat is preferred, independent of predation risk (i.e. minimize predation at all times). The PSF hypothesis predicts that there is a trade-off between food and predation, such that hares increase use of closed habitat as predation risk increases during the decline. Similar predictions can be made for distance to cover, movement (based on proxy measures such as home range and distance between browse points), and diet composition. These components of winter foraging behaviour (habitat use, movements, and diet) were used to examine the predictions of the PSF hypothesis.  Prediction 1: Patterns of habitat use by hares and movement of hares through the environment are not random. As risk of predation increases hares reduce exposure to predation by (i) increasing use of closed habitat, and (ii) reducing distance to protective cover.  Prediction 2: Hare movement (based on two proxy measures) decreases as predation risk increases: (i) home range decreases in size, and (ii) the distance between browse points decreases.  51  Fig. 3.1 Predictions for use of safer, closed habitat during the snowshoe hare  population decline according to the food, predation risk, and predation-sensitive foraging (PSF) hypotheses.  52  HIGH  PREDATOR AVOIDANCE HYPOTHESIS  —  % CLOSED HABITAT  PREDA11ONSENSfl1VE FORAGING HYPOTHESIS  FOOD HYPOTHESIS  LOW  PEAK  DECUNE HARE DENSITY  53  Prediction 3: As predation risk increases, the proportion of poorer quality items (Picea twigs) in the diet increases.  Additionally, theoretical investigations (McNamara 1987, 1990; Ludwig and Rowe 1990; Clark 1993, 1994) have suggested that an individual will counteract a decline in body condition (mass) by spending proportionally more time in open habitat where food is more available, but risk is higher. For hares, use of the riskier habitat may increase at the end of the winter. The assumptions of time-constraints (reproduction occurs at the end of winter), and condition-dependence (changes in body mass influence habitat use) which are implicit in these models may apply to snowshoe hares. In addition, Clark (1994) suggests that animals with a higher reproductive value (i.e. females) will adopt “safer” behaviours by increasing use of closed habitat than animals with a lower reproductive value (i.e. males). Although I do not have detailed information about changes in behaviour of individuals under different conditions, a number of predictions of these models can be examined by comparing a sample of hares from the population at different times.  Prediction 4: Use of riskier open habitat will be higher in late winter then in early winter.  Prediction 5: Use of riskier open habitat will increase as body mass decreases  within a season.  Prediction 6: In late winter, female hares will adopt safer foraging behaviours  and use a higher proportion of closed habitat then male hares in order to protect their reproductive investment, since the young are born in May.  54  METHODS  Study Area  This study was conducted near Kluane Lake, southwest Yukon (see Krebs et al. 1992; Chapter 2). The vegetation is a mosaic of open and closed white spruce (Picea glauca) forest, with the understory and open clearings dominated by grey willow (Salixglauca) and bog birch (Betula glandulosa). Two experimental hare live-trapping grids (34-ha) were used: (i) an unmanipulated 2 area surrounded by an electric CONTROL; and (ii) FENCE+FOOD, a 1-km fence to deter terrestrial predators (lynx, Lynx L canadensis, and coyote, Canis latrans), and provisioned weekly with pelleted rabbit chow (16% crude protein). Avian predators (mainly Great Horned Owl, Bubo virginianus, and Northern Goshawk, Accipitergentilis) had unrestricted access to this site. Predation risk, calculated as the index of predators to prey, increased significantly over the hare decline (Fig. 2.3). Habitat structure (percent cover of spruce trees), winter forage availability (number of 5-mm stems of Picea, Sallx, and Betula), distance to cover offering protection from predators (deadfall trees, dense thickets, snow burrows), and predation, in open shrub, open spruce, and closed spruce habitats, are described in Chapter 2.  Habitat Use  Habitat use and movements of foraging snowshoe hares were determined using a spool-and-line technique during winters of 1990-92 (Nams 1993). This method involved attaching spools of red thread (180-rn quilting  55  cocoon bobbins; Culver Textile Corp., 525-52nd St., West New York, NJ, U.S.A., 07093) to hares. Spools were tied in series (up to 3 spools) and secured in an envelope of clear shrink-wrap plastic to prevent thread from catching on vegetation. Spools were attached to a loop of wire on the radio-collar or glued to fur on the animals back using a non-irritating skin bond adhesive (Skin Bond Cement, Pfizer Hospital Products, 11775 Starkey Rd., Largo, FL, U.S.A., 33543). Hares were live-trapped in the evening, weighed, and fitted with spools were attached prior to release. The thread trail was followed the next day and information about patterns of habitat use and distance to cover was recorded every 10 m and at browse points. The initial 80-1 00 m when hares were fleeing the trap were disregarded. The mean length of a trail was 390±64 m. Each track was considered to be an independent sample. Hares were tracked in the same 30-ha area during each winter; therefore availability of habitats (shrub, open spruce, closed spruce) was similar between years. The results of this study are based on the foraging track of hares observed on part of a single night. This sample was probably representative of foraging behaviour of hares while active during the night, but I do not know what proportion of a 24-hr period was spent active. Most hares rest under cover when not foraging. The proportion of each habitat utilized by hares in each year was compared to the distribution of available habitat on the entire trapping grid using the log-likelihood ratio test statistic. The effect of year and sex on habitat use were also compared using G-tests. Manly’s alpha preference index (Krebs 1989) was also calculated for each year. Significant differences in distance to cover between habitat and year were compared using ANOVA (Wilkinson 1988).  56  Movement: Home Range and Distance Between Browse Sites  Home range area is weakly but positively correlated with hare movement rate in winter (Boulanger 1993; r=0.46, p=0.07); therefore home range area may be a useful index of hare movements. Home range was estimated using radiotelemetry data during the period February April, 1989-1992. Three permanent, -  3-rn high, null-peak telemetry towers were placed in a triangular array (White 1990) on CONTROL and FENCE+FOOD trapping grids. We oriented the towers using line-of-site triangulation from known grid locations, and by placing 20 survey beacon transmitters at known locations on the grid. Tower orientation was monitored by placing permanent beacons in known locations. During a telemetry session, bearing direction, signal strength, and an estimate of bearing reliability were made. Unreliable bearings were discarded. Bearings were usually taken by one observer moving between the three towers in rapid succession. Consequently all bearings were not taken simultaneously, but usually within a period of <45 minutes. Some hares may have moved between the recording of first and last bearings. This movement likely increased the error area of the location estimate, but did not appear to adversely affect location estimates. In a simulation study, Schmutz and White (1990) showed that when bearings were not obtained simultaneously, short-distance movements by animals increased location error estimates by 10-fold, but decreased location precision by <10%. Location estimates were made using Program LOCATE (Narns 1990) to estimate the location of the radio-transmitter and 95% confidence ellipse using the MLE procedure of Lenth (1981). Location estimates with error estimates of greater than 2 ha were eliminated from subsequent analysis. The mean error estimate of location estimates was 2.8 1±0.83 ha.  57  Location estimates were made once every 1-2 days during a 60-day period each winter (February-April). This ensured that successive telemetry locations were independent of each other (Swikhart and Slade 1985; Boulanger 1993). Home ranges were estimated using the harmonic mean estimator (Dixon and Chapman 1980), since it is the most accurate and least sensitive to differences in sample size (Boulanger and White 1990). Animals for which there were at least 15 independent locations were used to estimate home range area using Program LOCATE lVm (Kenward 1990). Asymptotic home range area was obtained after 15-20 fixes, and was calculated for both 95% and 80% (core) areas. Differences in home ranges were compared using ANOVA. The distance between browsing sites along the path of a foraging hare was also recorded and used as an index of movement. Hares were tracked using spools of thread described above. Longer distances between browse points may indicate higher movement rate. Alternatively, a shorter distance may indicate that hares are moving less.  Diet  Diet composition was determined by counting the number of times hares browsed on Salix or Picea along each threaded track. These points could be clearly identified along many trails, particularly after a snowfall. Sometimes, and particularly in 1992, there were problems with this method of identifying browse points because hares dig for food under the snow, climb spruce trees, eat bark of shrubs and trees, and consume bits of twigs discarded by red squirrels and other herbivores. On FENCE+FOOD the amounts of chow consumed could not be estimated. In addition, prolonged periods without fresh snow made detailed tracking impossible. In practice, therefore, estimation of diet composition is only  58  possible when conditions are ideal. For this reason, I consider the number of times hares were observed to stop and browse each species, rather than the number of stems or biomass of each species consumed.  Body Mass, Time-Constraints, and Sex  The influence of time of season, body mass, and sex on patterns of habitat use were examined using the methods described above. Early wirter was considered to be the period 15 January to 15 March, and late winter was the period from 15 March to the end of winter. Different individuals were tracked during each period. Habitat use was calculated as a proportion, and was arcsine transformed before analysis by ANOVA (Wilkinson 1988).  RESULTS  Prediction 1: Hares Increase Use of Closed Habitat During the Decline  Tracking data showed that use of exposed, open habitat in late winter decreased as predation risk increased between 1990 and 1992 on both grids (Table 3.1). In all years, the distribution of habitats used by hares on CONTROL was significantly different from the habitat available (G=7.9  -  16.2, df=2, P<0.02).  In 1990, hares on CONTROL utilized shrub and open habitat more than closed habitat, and this pattern was similar for both males and females (G=1 .06, df=2, P=0.588). In 1991 (early decline) females shifted use to include more closed habitat, whereas males did not (G=54.18, df=2, P.<0.001). This is consistent with the PSF hypothesis for females, and the food hypothesis for males. In 1992 (late decline) both males and females used closed habitat extensively (69-80%).  59 Table 3.1 Habitat use based on tracking data on CONTROL and FENCE+FOOD grids during late season. Percent of habitat available and mean percent (± S.E.) of total distance traveled in each habitat (N = number of hares tracked).  CONTROL % Habitat Year  N  Available  Shrub  Open  Closed  9  51  40  Males 1990  13  32±6  49±6  19±8  1991  9  51±13  36±11  13±8  1992  6  1± 0.3  19±3  80±3  Females  1990  8  29±8  46±9  25±9  1991  7  15±7  26±8  59± 13  1992  6  2±1  30±9  69± 10  60 Table 3.1 continued  FENCE+FOOD % Habitat  Year  N  Available  Shrub  Open  Closed  23  73  4  Males 1990  5  63± 10  37± 10  0  1991  5  55±5  42±5  3±2  1992  4  25±9  70± 12  5±4  1990  5  58±5  42±5  0  1991  4  62±6  38±6  0  1992  5  28±9  69±8  3±2  Females  61  In 1990 and 1991 hares on FENCE+FOOD used more shrub and less open spruce habitat compared with the available habitat (G-tests, df=2, P<0.01). In 1992 all habitats were used in approximately the proportion that they were available (G=0.75, df=2, P=0.69). During the winter of 1992 hare densities outside of the fence were low and avian predators frequented the fenced grid. The trend in patterns of habitat use suggest hares have increased sensitivity to higher predation risk. On FENCE+FOOD (Table 2), I compared use of shrub and open spruce habitat, since closed spruce habitat was uncommon. In 1991 and 1992 shrub habitat (higher food) was preferred to open spruce, but in 1992 both males and females changed patterns of use to include predominately open spruce habitat (lower stem density). Both sexes behaved similarly in all three years on FENCE+FOOD (G-tests, df=1, P>0.440). Habitat selection based on Manly’s alpha showed the same pattern: preference for closed spruce habitat increased during the population decline on both CONTROL and FENCE+FOOD. Distance to cover while traveling and browsing may also indicate sensitivity of hares to risk of predation (Fig. 3.2). Average distance to cover on each grid was discussed in Chapter 2 (Table 2.1). On CONTROL, hares traveled closer to cover than expected by chance (P<0.05) in shrub habitat, but not in open or closed habitat. However, distance to cover at browse sites was significantly less than at random for all three habitats, and was occasionally less than at random points along the track. This suggests that hares avoided browse far from cover. However, there was no difference in the distance to cover along a track or at browse points between years (P>0.80). Therefore, the prediction that distance to cover decreases as predation risk increases was not supported. Instead, it appears that hares consistently tried to minimize distance to cover in all years.  62  Fig. 3.2 Distance to cover (± S.E.) at random (squares), along foraging paths (circles), and at browse points (triangles) for male and female snowshoe hares on CONTROL and FENCE+FOOD, 1990-1992.  63  CONTROL Oicco •1991 01992  15RANDOM  DISTANCE TO COVER(M) 10 TRACK  ++  5-  0  BROWSE  I  I  SHRUB  OPEN  —  CLOSED  HABITAT  20 FENCE+FOOD Ol9co 15-  DISTANCE TO  COVER CM)  10  -  •1991 01992  RANDOM  +  TRACK  5-  BROWSE  AI  I  SHRUB  OPEN  HABITAT  CLOSED  64  On FENCE+FOOD distance to cover was not different from random when hares were traveling in shrub, open, and closed spruce habitat (Fig. 3.2). Distance to cover was only slightly less than along a track, except in 1992 in open habitat. Generally, there was no indication that hares traveled or browsed closer to cover on FENCE+FOOD than expected in 1990 or 1991. In 1992 browse sites were closer to cover than expected.  Prediction 2: Movements Decrease  Home range area during the period February to April was lowest for both male and female hares at peak density (1989/90), and then gradually increased on CONTROL (Table 3.2). Male home ranges were consistently larger than female ranges, but the difference between the sexes was not significant (2-way ANOVA of 80% core areas comparing sex and year: Fsex  =  0.014, df  =  1,35,  P=0.90). Similarly, the difference between years was not significant (Fyear  =  1.633, df=3,35, P=0.21). On FENCE+FOOD there was a significant difference between male and female home ranges (Fsex=0.45, df=1 ,29, P= 0.045), but the difference between years was not significant (P=0.09). Female home ranges were significantly larger in 1990 than in 1991 or 1992, but male home ranges were similar in all three years (Table 3.2). These results were not consistent with the prediction that home range would decrease as predation risk increased. The distances between browsing sites were also considered as an index of movement while foraging. These distances are minimum estimates because only paths with at least two browse sites were included. The distance between browsing sites was not significantly different between years on CONTROL (Table 3.3); mean.distances ranged from 32 to 53 m (P>0.40). On  65  Table 3.2 Home range area (ha ± S.E.) of male and female hares during the period February to April on CONTROL and FENCE+FOOD, 1989-1992. Home range areas shown for 95% and 80% areas (harmonic mean method).  CONTROL  Year N  FENCE+FOOD  95%  80%  N  95%  80%  Male 1989  6  2.05±0.68  1.23±0.47  1990  7  4.99±1.10  3.07±0.79  10  3.86±0.34  1.86±0.25  1991  8  6.13±3.69  2.95±1.17  4  3.93±0.34  2.17±0.33  1992  3  5.88±1.90  2.50±1.01  5  3.66±0.52  1.98±0.30  1989  3  0.98±0.48  0.50±0.15  1990  4  0.69±0.01  0.49±0.04  6  9.27±1.87  4.80±0.91  1991  7  4.65±1.86  2.22±0.38  6  2.33±0.64  1.33±0.29  1992  3  4.52±2.11  2.08±0.41  4  2.61±0.80  1.45±0.39  --  Female --  ---  66 Table 3.3 Distance between browse sites (± S.E.) on CONTROL and FENCE÷FOOD, winter 1990-1992. Values for male and female hares pooled. N is the number of browse sites recorded in each year.  CONTROL  Year  N  FENCE+FOOD  distance (m)  N  distance (m)  1990 49  41.0±7.7  31  32.8±9.1  1991 37  52.7±9.7  29  36.1±1 0.0  1992 32  31.6±7.9  16  23.7±5.8  67  FENCE+FOOD these distances were lower than on CONTROL, but were also similar between years. There is an indication on both grids that this distance decreased in 1992, but the difference was not significant (P>0.40).  Prediction 3: Diet Quality Declines  Diet composition was based on the number of times each browse species was consumed by hares along foraging tracks (Fig. 3.3). On CONTROL the proportion of Picea increased from 58% (29/50 sites) in 1990 to 91% (30/33) in 1992. The proportion of Salix browse sites decreased from 42% to 8% during the same period. On FENCE+FOOD SaIix dominated the diet in all three years (55-80%), but consumption of bark increased between 1990 and 1992 (5-25%; Fig. 3.3). The proportion of rabbit chow consumed could not be accurately calculated, but higher body mass of FENCE+FOOD hares compared to CONTROL hares (Chapter 4) suggests that all animals had access to the supplemental chow.  Prediction 4: Hares Increase Use of Open Habitat in Late Winter  The PSF model predicted that use of closed habitat would decrease late in the winter season. On CONTROL a decrease in use of closed habitat was observed for males in 1991 (Fig. 3.4), but the difference from early season was not significant  (Fi,i5  =  0.451, P=0.114). A small decrease was also observed for  females in 1992 but this was not significant. In 1991 females increased use of closed habitat between early and late season but this difference was also not significant (Ft,io  =  0.756, P  =  0.41). On FENCE+FOOD, there were no  68  Fig. 3.3 Estimated diet composition (mean % ± S.E.) of hares based on number of browse stations for each species. Species eaten in winter are twigs and bark of Picea (spruce), and Salix (willow).  69  100  80  60 % BROWSE SITES 40  20  0•  YEAR 100  FENCEi-FOOD  80  60’  % BROWSE SITES 40  20  0 YEAR  C  PICEAIWGS  •  SALJX1WGS SAUX BARK  70  Fig. 3.4 Proportion (± S.E.) of closed spruce habitat on CONTROL and open spruce habitat on FENCE+FOOD, used by female and male hares early and late in the winter, 1990-1992.  71  100-  CONTROL 92d  75929  % CLOSED SPRUCE HABITAT  919  .  25-  91 d  0-  I  I  EARLY  LATE  SEASON 100  % OPEN SPRUCE HABITAT  EARLY  LATE  SEASON  72  differences observed between early and late season use of the safer open spruce habitat (which was the only closed habitat available on this grid).  Prediction 5: Hares in Poor Condition Use More Open Habitats  Body mass was not correlated with use of closed habitat within a given year on CONTROL or FENCE+FOOD (Fig. 3.5). Over all three winter seasons there was a strong negative relationship between use of closed spruce habitat and body mass on CONTROL, but there was no indication that individuals increased use of open habitat as body mass declined within a season. On FENCE+FOOD there was no relationship between body mass and use of open spruce habitat, even over all seasons (Fig. 3.5). These results do not appear to support the prediction that use of open habitat increases as body mass declines.  Prediction 6: Females Adopt Safer Foraging Strategies than Males  Figure 3.4 indicates differences in the use of closed habitat between males and females in 1991 on CONTROL grid: females used closed habitat more than males late in the season. This shift is consistent with the hypothesis that females may reduce predation risk in order to protect their reproductive value (Prediction 6; Clark 1994), but this result was observed in one year only and the overall change in patterns of habitat use between males and females was not significant in 1991 (two-way ANOVA; sex: F1,25 season: F1,25  =  =  2.686, P  =  0.114;  0.013, P= 0.91). However, in the late winter season females  used closed habitat significantly more than did males (Ft,14  =  10.981, P< 0.005).  There were no differences between males and females on FENCE+FOOD. In summary, 1991 on CONTROL is the only year that a difference in use of closed  73  Fig. 3.5 Relationship between body mass of male and female hares and the proportion of closed habitat used on CONTROL or open spruce habitat used on FENCE+FOOD, 1990-1992.  74 100  --  CONTROL d9 DOl9coj  •  -  E J..  80  •1991I Q1992  0 60  0  I  %CLOSED SPRUCE HABITAT  40 D  o C  ii 20  C C  ii  1100  0.  •  0• D•Do• •..O 1600 ‘1500 1400 1700 1800 C  -  I  1200  —  I  1300  I  .  I  1900  2000  BODY MASS (g) 100  FENCE÷FOOD I9 ID 0 iccol ••i99i 09c2  0  80  0• 0 I  60 % OPEN SPRUCE HABITAT  1  0  40  00•0  .  0.  20  C  I  0  1100  1200  1300  I  1400  1500  I  1600  BODY MASS (g)  I  I  I  1700  1800  1900  2000  75  habitat was observed between males and females. No other differences in patterns of habitat use between males and females were observed.  DISCUSSION  The results suggest that the foraging behaviour of hares changed from being primarily influenced by the availability of food at the peak, to a strategy of predator avoidance during the decline (Table 3.4). On CONTROL, two of the results are consistent with the predation-sensitive foraging (PSF) hypothesis (% closed habitat used and diet composition). However, distance to cover was low in all years, suggesting that hares are always sensitive to predation risk. Conversely, home range size generally increased during the decline and possibly hares were moving more. Since I have no detailed information about how hares were using their home range, this may not be a good index of movement. Distance between browse points was also similar in all years, suggesting no differences in behaviour at this scale. The main differences between years were increased use of closed habitat and increased use of Picea browse. On FENCE+FOOD the changes in diet composition between 1990 and 1992 were consistent with the food hypothesis, but the other results support the predator avoidance and PSF hypotheses (Table 3.4). It is possible that hares have more alternative foraging strategies available when predation risk is lower or when supplemental rabbit chow is provided. Similar to results on CONTROL there was no change in distance to cover between years suggesting that hares minimize this distance most of the time. The main difference between years is an increase in use of safer habitat in 1992.  76  Table 3.4 Summary of predictions of the food, predator avoidance (PA), and predation sensitive foraging (PSF) hypotheses, for (a) use of closed habitat, (b) distance to cover relative to the peak, (c) home range size, and (d) % spruce in the diet. Changes in patterns of use relative to the peak are indicated as low or decrease (ii.), high or increase (U), or no change (=). Observed trends on CONTROL and FENCE+FOOD are shown and the hypothesis predicted (FOOD, PA, PSF) is indicated in bold type.  (a)  (b)  % use closed habitat Peak  PSF  Decline  Peak  U U  U  U  U  U  U  U  Male  CONTROL  (d)  distance to home range % spruce cover relative relative to in diet to peak peak  FOOD Predictions PA  (c)  U  U PSF  [  U  U  U  U  U  U  U  Female  UU PA  Decline  Food  PSF  Observed  FENCE+ FOOD  U  U PSF  U PA  PA  -  PSF  U  U  Food  77  The observed changes in patterns of habitat use within a season generally do not support predictions 4, 5, and 6, relating to the effects of body mass, time constraints, and sex. Only the change in patterns of habitat use on CONTROL during 1991 support the PSF hypothesis. Further testing of these predictions will require observation of individual changes over time. This is difficult with hares when winter survival is so low.  Prediction 1: Measuring Patterns of Habitat Use  As predation risk increased during the population decline, the use of closed and safer habitat increased (Table 3.1). This result was also obtained from hare live-trapping data (Fig. 2.6). In addition, hares on CONTROL often traveled and browsed closer to cover in shrub habitat than expected if hares moved randomly (Fig. 3.2). This pattern was similar in all years suggesting that hares have a limited ability to reduce this distance as predation risk increases. Several studies have found that the distance from cover that animals will travel to forage depends on predation risk (e.g. Anderson 1986; Hughes and Ward 1993). It seems that hares are more likely to switch to safer, closed habitat as risk increases, rather than reduce distance to cover. This may indicate the importance of dense closed forest as refuge areas for hares (Wolff 1980). One assumption of the spool-and-line technique is that behavior of hares was not influenced by trapping or application of spools. Mikesic and Drickamer (1992) have reported that both radio-collaring and application of fluorescent powders result in a short-term reduction in the activity of wild house mice (Mus musculus). In my study I attempted to minimize disturbance associated with trapping by only tracking hares that had been previously captured. These hares were often more calm in traps than individuals captured for the first time. In  78  addition, traps were checked at hourly intervals during the evening so that hares would not have time to consume alfalfa bait, which could modify their subsequent behavior. Boonstra and Singleton (1993) suggested that handling and trap confinement of hares lead to increased blood glucose levels and stress. It is possible, therefore, that trapping could have affected the results reported here.  Prediction 2: Home Range and Movements  Home range area of hares tended to increase as hare density declined (Table 3.2). These results are consistent with those of Boutin (1984c), who found that summer home range area decreased with increasing population density. However, it seems likely that home range is related to food availability and predation risk, as well as hare density. Increases in home range size on CONTROL and FENCE+FOOD during the population decline do not support the PSF hypothesis, but without knowing more about how hares use the space within their home range I cannot conclude that larger home ranges are correlated with higher movement rates. It is possible that hares increase home range size during the population decline in order to have access to additional cover, or to dilute their own scent, to reduce predator attraction. Larger home ranges may also be required to locate mates at low population density. Also, on FENCE+FOOD hares traveled between the four rows where rabbit chow was distributed. Therefore home range size could be increased by the experimental design. The distance between browse sites was similar in all years an CONTROL. There was a small decline in 1992, but this distance was not significantly different from previous years. On FENCE+FOOD distance between foraging  79  sites was slightly less than on CONTROL, but overall the pattern was similar between years. The results are inconclusive, but suggest that hares may reduce foraging activity during the decline. The development of more sensitive activity radio-collars may permit activity of hares to be measured in the future.  Prediction 3: Diet Composition  Increased use of Picea browse is known to lead to a loss of body mass in snowshoe hares (Rogers and Sinclair 1994). There was a major shift in diet composition (based on the number of browse sites for each species) between 1990 and 1992 on CONTROL: Salix was replaced by Picea as the dominant winter forage (Fig. 3.3). Increased use of Picea twigs during the hare decline was also noted by Smith et a!. (1988; Table 4). Wolff (1978) reported that Picea stems and needles were found in 90-100% of all stomachs, and comprised the largest proportion of the winter diet of hares in central Alaska; however, he provided no information about changes in use over the cycle. On FENCE+FOOD, Picea comprised a smaller proportion of the diet than on CONTROL, but there was also higher consumption of Salix bark (Fig. 3.3). This may reflect a dietary requirement for increased fibre when supplemental rabbit chow is provided (Cheeke 1983; J. P. Bryant, pers. comm.). The shift from Salixto Picea forage may also be related to the density of stems of each species (Table 2.1). Based on stem density in mid-March, the amount of Salix browse in closed spruce habitat is less than that of Picea (Table 2.1). As a result hares may be selecting the more available forage in this habitat, because the biomass of 5-mm stems of Picea are 2-3 times higher than for Salix (Chapter 2). If the intake rate of Picea stems is higher than that of Salix, then the amount of time spent foraging and exposed to predation would  80  decrease as consumption of Picea twigs increased. This would be particularly true if handling time is similar in each habitat and travel time between browse sites increased during the decline. Shipley and Spalinger (1992) tested the hypothesis that food intake rate of herbivores is limited by bite size rather than plant density. They found that bite size explained 40-90% of the variation in intake rate of moose, caribou, and white-tailed deer. However, intake rate was not a function of increasing bite size in snowshoe hares. In a similar experiment with captive lemmings (Dicrostonyx groenlandicus), Gross et a!. (1993) concluded that bite size had a larger effect on intake rate did than plant biomass or density. The time needed to process a bite in the mouth exceeded that necessary to travel between plants. However, they assumed that herbivores are able to process food while moving between browse sites. If hares are unable to do this easily, then stem density may have a significant effect on forage intake rate of hares, particularly during the decline phase of the cycle. Foraging on stems of Picea in closed habitat may reduce overall feeding time for hares and thereby reduce risk of predation. This hypothesis needs further testing.  Predictions 4, 5, and 6: Time of Season, Body Mass, and Sex  Body mass, time of season, and sex generally did not appear to influence foraging behaviour within a given year. One explanation for this is that differences in foraging strategies may not be detectable over the short period that hares were tracked on a single evening. It is also possible that changes in these behaviours are not apparent at the population level. Nevertheless, differences in habitat use between each year were significantly correlated with body mass in mid-April (Chapter 4). Hares did not show changes in habitat use  81  between early and late winter (Fig. 3.4), although there were differences between years. In 1991 on CONTROL there were also differences between males and females. Sex, body condition, and season are factors which, to varying degrees, may influence anti-predator behaviours of snowshoe hares, despite equivocal evidence (Litvaitis 1990, 1991). Other experimental studies may provide some clues about where to look for these behaviours in snowshoe hares. Recently, Pettersson and Brön mark (1993) reported that crucian carp (Carassius carassius) spent less time in open habitat when there was a predator present, and when not hungry. There was a significant interaction between predation and hunger level. Abrahams and Dill (1989) have reported similar results with guppies (Poecilia reticulata). Saarikko (1992) observed that the response of shrews (Sorex araneus) to the presence of a potential predator (weasel) was positively correlated with body mass. Under risk of predation, large shrews could decease foraging activity (at the cost of mass loss), but smaller individuals with lower energy reserves could not. Theoretically there are reasons to expect differences in the behaviour of animals as their condition changes, or the time to reproduction approaches (Ludwig and Rowe 1990; Houston et al. 1993; Clark 1994). There is clearly a need to find better indices of condition for hares (e.g. Boonstra and Singleton 1994), and for monitoring detailed individual behaviour in winter over a longer period. Sinclair and Arcese (1994) have suggested that predators affect wildebeest (Connochaetes taurinus) populations through food supply by influencing behaviour. They reported that the condition of predator-killed wildebeest was poorer than that of the general population, but better than in non-predator mortality, results which support the PSF hypothesis. Keith et a!. (1984) observed a similar relationship for snowshoe hares. Overall, there is considerable evidence that condition of animals influences foraging decisions  82  under risk of predation, but a great deal probably depends on the feasibility of alternative foraging strategies. Hares appear to have few options.  How Can Snowshoe Hares Minimize Risk of Predation?  Previous studies have suggested that patterns of habitat use by snowshoe hares in winter may be influenced by risk of predation (Wolff 1980, 1981; Keith et al. 1984; Sievert and Keith 1985). Results of my study support these observations, and also suggest that hares have a limited ability to reduce exposure to predators. There are two main ways in which hares may be able to minimize predation risk: (1) increase use of safer closed habitat, and (2) decrease time spent foraging. Opportunities to shift to safer habitats are limited, in part because winter forage in closed habitat is less available than in open habitats. Opportunities to decrease foraging effort or time are equally limited: hares are physiologically constrained by their poor ability to maintain reserves of fat or protein (Whittaker and Thomas 1983). Hares also have no absolute refuge from predators and are at risk in all habitat types (Table 2.3). The evidence suggests that the alternative anti-predator behaviours available to hares result in reduced foraging effort or a shift to a poorer diet. In each case this may lead to increased loss of body mass during winter. Indeed, if predators show adaptive behaviour (Abrams 1989) and follow hares into closed habitat (see Chapter 2), the most reasonable option for hares may be to reduce foraging activity or move. At a broader evolutionary level, snowshoe hares appear to be well adapted to minimize risk of predation. Hares undergo a seasonal pelage change from white in winter to brown in summer. In winter, white pelage has been associated with both improved thermal insulation (Hart and Pohl 1965) and  83  predator avoidance (Grange 1932; Litvaitis 1991). Other factors which might influence predator avoidance behaviours include bright moonlight (Gilbert and Boutin 1991), low temperature (Pease and Keith 1979; Keith 1990), and high wind (Bider 1961). I did not consider these factors explicitly, although they are probably important at certain times, and hares do change their foraging behaviour in response to environmental conditions (e.g. Gilbert and Boutin 1991). The mechanisms by which hares detect predators are not well understood, but I suspect that it is probably much easier for hares to detect the various faecal, urine, scent gland, and body odours of terrestrial predators, than of avian predators. Jedrzejewski et aL (1993) have recently shown this to be true for bank voles (Clethrionomys glareolus). Studies by Sullivan and his colleagues (Sullivan and Crump 1984, 1986; Sullivan et al. 1985) demonstrated that some volatile methyl sulfide constituents of canid, feline, and mustelid urine suppress feeding activity of snowshoe hares. They suggested that these odours induce a fear or avoidance response in hares. Perception of predation risk by individual prey is behaviourally complex, and predator avoidance behaviours may exist even in the absence of any overt predator cues (Curio 1993). Dickman (1992) found that the survival of predatorexperienced house mice (Mus domesticus) was 2.5 times higher than predatornaive mice, because predator-experienced mice used sites with greater vegetation cover. In a recent theoretical study, Bouskila and Blumstein (1992) showed that animals may overestimate predation risk if cues used to indicate risk are inaccurate or diffuse. The use of conservative “rules of thumb” by snowshoe hares to assess predation risk would tend to reduce foraging effort in a manner consistent with the observation that hares lose body mass as predation risk increases, even though food is apparently available.  84  It is also possible that hares over-estimate risk of predation when information about absolute predation risk is uncertain, or if hares are searching for both avian and terrestrial predators simultaneously. The presence of multiple predators can theoretically increase vigilance of animals, and therefore decrease time spent foraging (Lima 1992). Kotler et a!. (1992) have shown that gerbils (Gerbillus spp.) are unable to forage and remain safe from snakes and owls simultaneously. Like hares, gerbils appear to have few options available to reduce risk of predation when several types of predators are present.  CONCLUSIONS  Snowshoe hares act to minimize risk of predation during the snowshoe hare decline by reducing use of open habitats in favour of more closed habitats as predation risk increases, but finer scale changes were not apparent in this study. Distance to cover while foraging was not different from random in open and closed spruce habitat, and did not decrease as predation risk increased. In open shrub habitat hares traveled closer to cover than expected at random, but again, this did not change between years. The evidence that hares reduced foraging effort (movements) as predation risk increased was inconclusive. There was a marked shift in diet to include more lower quality Picea twigs even though hares seem to lose mass on this diet. There also seemed to be no difference in patterns of habitat use associated with body mass. Differences between males and females, and early and late winter were apparent only on the CONTROL grid in 1991. Females reduced use of open habitats, while males increased use of open habitats in late winter. Overall the results suggest that hares have limited options available to reduce risk of predation in winter. Since hares have no absolute refuge from predators, increased use of more closed  85  habitat, decreased foraging effort, and selection of poorer quality browse leads to loss of body mass. The consequences of this winter mass loss are discussed in the following chapter.  86  CHAPTER 4 DOES RISK OF PREDATION INFLUENCE POPULATION DYNAMICS? EVIDENCE FROM THE CYCLIC DECLINE OF SNOWSHOE HARES  INTRODUCTION Anti-predator behaviour of prey in response to increased predation risk may result in decreased fecundity or increased mortality caused by factors other than predation. The influence of short-term behavioural decisions by individual animals (e.g. tradeoffs between maximizing foraging rate and minimizing risk of predation), on their longer-term survival and reproduction have been considered in a number of models (McNamara and Houston 1987, 1990a; Mangel and Clark 1988; Ludwig and Rowe 1990; McNamara 1990; Clark 1993, 1994), and empirical studies (Lima and Dill 1990; Peckarsky eta!. 1993). Individuals in danger of starvation may accept increased predation risk in order to obtain sufficient food. However, if predation risk is high and individuals have only a finite amount of energy available, investment in present reproduction may be reduced in order to increase future growth, reproductive capacity, or survival (Partridge and Harvey 1985; Sibly and Calow 1986). There is considerable interest in determining how predator avoidance behaviours influence predator-prey population dynamics (Hassell and May 1985; Ives and Dobson 1987; McNamara and Houston 1987, 1990; Abrams 1989, 1990, 1991, 1992a,b,c; Williamson 1993; FitzGibbon and Lazarus 1994). Several authors (Wolff 1980; Keith et a!. 1984; Smith et a!. 1988) have suggested that predation risk may influence habitat use of snowshoe hares (Lepus americanus), but whether predation risk is a necessary component of the 10-year population cycle has not been considered. There are several alternative hypotheses involving food, predation, and behaviour of hares to explain the 10-  87  year cycle in North America (Sinclair et al. 1988; Keith 1990; Krebs et a!. 1992; Royama 1992; Chapter 1). Recent experimental studies have suggested that predation alone may be sufficient to generate the hare population decline (Krebs  et a!. 1 986a,b), and that winter food is not absolutely limiting for hares at any time during the cycle (Sinclair et a!. 1988, Smith et a!. 1988). Most hares die of predation (Boutin et al. 1986; Trostel et a!. 1987), although relative winter food shortage may contribute to the hare decline (Pease et a!. 1979; Smith et a!. 1988; Royama 1992). Summer food appears to be plentiful (Keith 1990; D. Hik  et a!., unpublished data). Previous studies have observed a significant reduction in female fecundity during the population decline and this demographic characteristic is closely linked to the cyclic dynamics (Cary and Keith 1979). Reproduction is correlated with female body mass at the end of the previous winter (Keith and Windberg 1978; Keith 1990; Royama 1992). Body mass at the end of winter is lowest during the population decline, and this reduction may result in smaller litters. However, poor female condition does not appear to affect all litter groups equally: the size of the first litter remains fairly constant, but the size of subsequent litters declines significantly. In the previous chapters I showed that predation risk is not constant during the 10-year population cycle of snowshoe hares, and that changing patterns of habitat use by hares are a behavioural response to minimize predation risk. In this Chapter the effects of increased predation risk on survival, body mass, and fecundity during the decline phase of the 10-year cycle are considered. Results reported here are compared with those from earlier studies of Lloyd Keith and his associates (see Keith 1990) and interpreted in light of the effects of poor nutrition on the physiology of snowshoe hares (Whittaker and Thomas 1983; Boonstra and Singleton 1993).  88  Hypotheses and Predictions  The interaction between food and predation risk has several possible outcomes. In Figure 4.1 changes in condition (body mass) and survival during winter predicted under each of three hypotheses are outlined. The predictions of these hypotheses were examined by manipulating food and predation risk on four experimental hare-trapping grids at Kluane, Yukon, during a hare population decline (1 991-1 993). When food is readily available and predators are scarce (conditions characteristic of the hare increase and peak, 1989/90), body mass (condition) and survival of hares should be high (initial state (+,+) in Fig. 4.1). As predation risk increases the food hypothesis (+,-) predicts that hares maintain body mass (maximized intake rate), but survive less well, relative to the initial state. The predator avoidance (PA) hypothesis (-,+) predicts that hares lose body mass, but still survive well (minimize predation). The predation-sensitive foraging (PSF) hypothesis (-,-) predicts that there is a tradeoff between foraging and predation, such that predation risk restricts access to food. Both survival and condition will decrease. Specific predictions of PSF (McNamara and Houston 1987; Ludwig and Rowe 1990) are (I) increased predation risk leads to decreased body mass (hence fecundity), and (ii) decreased food levels lead to increased mortality. Two further predictions that can be tested indirectly are, (iii) decreased food will have the greatest effect on condition when predation risk is highest, and (iv) increased predation risk will have the greatest effect on mortality when food level is lowest.  89  Fig 4.1.  low, +  =  Predicted changes in condition (body mass) and winter survival  (- =  high), under two levels of predation risk and food availability. The  arrows indicate the direction of change which supports each hypothesis: food, predator avoidance (PA), or predation sensitive foraging (PSF).  90  FOOD LOW  HIGH condition HIGH  +  FOOD  PSF  survival -  RISK condition LOW  survival  I  + +  INAL STATE  -  PA  91  METHODS  This study area is described in Chapter 2. Experiments were conducted on four 34-ha trapping grids between January and May of 1989-1993. Sulphur grid was an unmanipulated CONTROL; Beaver Pond (FENCE) was surrounded by an electric fence to deter terrestrial predators (lynx, Lynx L canadénsis, and coyote, Canis latrans); Hungry Lake (FENCE+FOOD) was surrounded by a fence, and provisioned weekly with pelleted rabbit chow (16% crude protein). Avian predators (mainly Great Horned Owl, Bubo virginianus, and Northern Goshawk, Accipiter gentilis) had unrestricted access to both fenced grids. Gravel Pit (FOOD) was provisioned weekly with pelleted rabbit chow, but predators had free access to this site. Peak hare densities on CONTROL were reached in 1989 and 1990 (Chapter 2), while peak densities were observed one year later on the other three trapping grids. Snowshoe hares were live-trapped at monthly or bimonthly intervals between January and May as described in Chapter 2. Body mass was determined at each capture, but only mass recorded at the first capture during a multi-day trapping session is used in this analysis. The deaths of radio-collared hares were used to calculate over-winter survival from the Kaplan-Meier estimator (Pollock 1 989a,b), as described in Chapter 2. Female mass in early April and 30-day survival (January to April) in 1990 were compared to corresponding values in each year of the decline (1991-1993), and significant differences are reported in Table 4.2. Mean size of each litter was determined from pregnant females placed in maternity cages 2-3 days prior to parturition (see O’Donoghue and Krebs 1992 for description of methods), on CONTROL grid and adjacent areas (1 989-92), FOOD (1989/90) and FENCE+FOOD (1 991/92) grids (O’Donoghue and Krebs  92  1992; Sovell 1993; C.J. Krebs eta!., unpublished data). I assumed that reproduction on FENCE+FOOD was similar to that on FOOD in 1989 and 1990. Reproductive output was not measured on FENCE or FOOD during the decline.  RESULTS  Winter Su,vival  Estimates of 30-day survival in winter (January to April) were highest on FENCE, FOOD, and FENCE+FOOD grids in 1989 and 1990, and were somewhat lower on CONTROL, particularly in 1989 (Fig. 4.2). In 1991, the first year of the decline, survival was highest on FENCE+FOOD (0.92), followed by FOOD (0.86), FENCE (0.76), and then CONTROL (0.64). In 1992 survival on FOOD and CONTROL was lower than on FENCE and FENCE+FOOD. In 1993, survival on FENCE+FOOD decreased significantly, but was similar to the other treatments (about 0.8).  Body Mass  Female hares generally maintained body mass throughout the winter and then increased rapidly in mass following conception in mid-April (Fig. 4.3), except on CONTROL after 1990, and FOOD after 1991. On CONTROL, female hares were significantly lighter in each winter during the decline than at the peak, and weighed less than 1400 g at the end of winter in 1991, 1992, and 1993. On FENCE grid, body mass did not decline, and values were consistently  93  Fig. 4.2 Winter (January to April) 30-day survival (± 95% C.L.) of hares on  CONTROL, FENCE, FOOD, and FENCE+FOOD, 1989-1993. The number of radio-collared hares monitored during each sampling period is indicated for each sample.  -  (JAN APRIL)  0.4  0.5  0.6  MEAN 30-DAY SURVIVAL 0.7  0.9  1.0  1989  1990  YEAR  1991  1992  1993  CD  95  Fig. 4.3 Mean body mass of female hares during the period January to May  1989-1993 on CONTROL, FENCE, FOOD, and FENCE+FOOD. Solid line joins estimates of mean mass in early April. Sample size for each mean value varied from 2 to 130 individuals depending on the year, and ±1 S.E. is less than the size of the symbol unless otherwise indicated. Dotted line at 1400 g is a reference for comparison between treatments.  C661 66[  66[66[  +  33N31  [661.  + [661  0661.  .  •  696[  .  +  6961.  COOd-’-33N3J  0661.  c66[ 66[ 166[  6961  OOLL  0OL  066[  OOL  009L  OOLL  OOgL  OOLL  00&  006L  009L  0011  0091  OOPL  4  ‘OOPL  C00i  OOQL  009L COIL ‘0091 •006L .0O  6961.  00L  0661.  009L  .  [661. •OOL  009L  661  .0091  COPI  66[  COPL  OOLL  0091  009L  ••  009L 0091  :  0091  COIL  0091  ,0OO  0061  1OI1NOD  0061 tVV7  97  between 1500-1 600 g. On FENCE+FOOD body mass was about 1600 g in all years. On FOOD, body mass was similar to that on CONTROL during the peak. During the decline body mass remained high, but decreased to about 1350 g in 1993.  Reproductive Output of Female Hares  Mean litter sizes (Table 4.1) showed that hares on CONTROL and adjacent unmanipulated areas were much less fecund during the decline than on FENCE+FOOD. These values are not adjusted for stillborn rates which ranged from 0 to 45% and were highest for third litters (mean  15%;  O’Donoghue and Krebs 1992). On CONTROL, the size of the first litter was similar in all four years in 1989 and 1990, but no third litter was produced in 1991. In 1992 neither second nor third litters were produced. This reduction in reproductive effort on the CONTROL areas suggests that body mass of less than 1400 g at the end of winter may contribute to reduced female fecundity.  DISCUSSION  Evidence Supports the PSF Hypothesis  The observed changes in winter survival (Fig. 4.2) and April body mass (Fig. 4.3) during the hare decline, relative to the population peak, are compared with predictions of the food, predator avoidance (PA), and predation sensitive foraging (PSF) hypotheses in Table 4.2. On CONTROL, the reduction in condition and survival during the hare decline supports the PSF hypothesis. In 1993, survival improved while body mass continued to decline, a result  98 Table 4.1 Mean litter size (± S.D. (sample size)) of hares at Kluane, Yukon on CONTROL and FENCE+FOOD (or FOOD in 1989/90) grids, during the population peak and decline. Data for 1989 and 1990 from O’Donoghue and Krebs (1992). Data from 1991 and 1992 from C.J. Krebs et a!. (unpublished data) and Sovell (1993).  CONTROL  Litter  Second  Third  Year  First  1989  3.6± 0.7 (8) 5.9± 1.6 (8)  4.2± 0.4 (5)  1990  3.9± 0.9 (7) 5.7± 1.4 (14)  4.1± 1.6 (12)  1991  3.9± 0.7 (8) 3.9± 1.4 (10)  0  1992  3.3± 0.5 (4) 0  0  FENCE+FOOD  Litter  Second  Third  Year  First  1989  3.8±1.2(9) 5.6±1.7(9)  4.6± 1.1 (9)  1990  3.8± 0.7 (15) 5.5± 2.2 (11)  5.8± 1.2 (10)  1991 1992  7.8± 1.3 (9) 4.2± 1.1 (13) 7.0± 0.9 (18)  4.7± 2.3 (6) 5.9± 1.7 (19)  99  Table 4.2 Observed changes in body mass and survival (+ = high; conditions during the peak; = low; significant decline from peak conditions, P <0.05) on CONTROL, FENCE, FOOD, and FENCE+FOOD grids during the snowshoe hare decline (1991-1993), relative to the peak years (1989/90). The hypothesis supported on each grid is indicated as FOOD (food hypothesis), PA (predator avoidance hypothesis), and PSF (predation-sensitive foraging hypothesis). -  Condition (body mass)  Survival Hypothesis Supported  Treatment  Year  CONTROL  1989 1990 1991 1992 1993  +  +  +  +  1989 1990 1991 1992 1993  +  +  +  +  +  -  +  -  +  -  1989 1990 1991 1992 1993  +  +  +  +  +  -  -  -  1989 1990 1991 1992 1993  +  +  +  +  +  +  +  +  +  -  FENCE  FOOD  FENCE+FOOD  -  -  +  -  PSF PSF PA  FOOD FOOD FOOD  FOOD PSF PSF  FOOD  100  consistent with the PA hypothesis. On FOOD, the results support the PSF hypothesis, suggesting that predation risk restricted access to food in the second and third year of the population decline. Survival did increase in 1993, even though mass declined, suggesting that hares were minimizing risk even though ad lib supplement food was available. On FENCE and FENCE+FOOD, body mass remained high even though survival decreased during the decline. These results support the food hypothesis. Predation may have increased inside the fences because avian predators were more active there following the hare decline outside of the fences (Chapter 2). Overall, the results indicate that in the absence of terrestrial predators hares are able to maintain body mass. If avian predators are more difficult to detect than mammalian predators (see Jedrzejewski et a!. 1993), hares may still use open habitat inside the fences. In this case, body mass will remain high, but survival will decrease. A comparison of all four treatments supports the third and fourth predictions of the PSF hypothesis. Where predators were not controlled, lower food level led to decreased condition on CONTROL compared with that on FOOD. Similarly, condition was lower on FENCE compared with that on FENCE+FOOD. Higher predation risk on CONTROL also resulted in lower survival compared to FOOD (higher food), at least during the first two years of the decline. Differences in hare densities among the treatments may have had some influence on the results, but I think that these effects were small. For example, I would predict that higher density would result in lower survival and poorer condition because of competition among hares for food or space (Sinclair 1986; Ferron 1993), and the observed results are in the opposite direction. The results also indicate a strong correlation between body mass and female reproduction, a result consistent with those of Cary and Keith (1979). In  101  1991 and 1992, body mass and female reproduction declined on CONTROL even though food was apparently not limiting (Chapter 2). The relationship between total mean reproductive output and mean female body mass (1989=0.71, N=4, 2 =0.53, N=4, P=0.17), and early May (r 2 1992) in mid-March (r P0.10), was not significant, but the trend was stronger late in the season. In summary, body mass at the end of winter appears to have the largest effect on reproduction (Keith 1990; Royama 1992).  Why do Body Mass and Reproduction Decline?  The body mass of hares on CONTROL decreased during the population decline, even though sufficient forage appeared to be available. It is possible that forage availability and quality was overestimated and therefore food may have been limited, but this is unlikely (Chapter 2). The observed increase in the proportion of Picea browse in the diet (Chapter 3) may account in part for observed declines in body mass (Fig. 4.1); captive hares were unable to maintain mass on a diet of Picea twigs, and could just maintain mass on a diet of Salix twigs (Rogers and Sinclair 1994). Hares may also lose mass during the decline if foraging efficiency is lower as the spatial patchiness of browse increases (Table 2.1). Alternatively, increased risk of predation may have contributed to the decline in body mass on CONTROL (and FOOD) by restricting access to food resources. If animals accept a certain probability of predation in order to obtain food, outright starvation may be uncommon, but effects of food availability on survival, body condition, and reproduction are potentially large (McNamara and Houston 1987). At peak densities, snowshoe hares utilized a significantly greater proportion of open habitat than expected by chance, a result consistent  102  with the food hypothesis (Chapter 3). However, as risk of predation increased during the cyclic population decline, hares increased their use of more closed habitats where forage was less available, and foraging rates may be lower. This switch in habitat is consistent with the predictions of the PSF hypothesis. On CONTROL there was a negative linear relationship between the proportion of closed habitat used by female hares and total reproductive output between 1990 and 1992 (r =0.98, N=3, P=0.066). The results from the FENCE+FOOD 2 treatment show that hares did not increase use of safer habitat until 1992, and that provisioning with supplemental rabbit chow prevented a decline of body mass, survival, and fecundity. Higher body mass was also observed on FOOD during the first year of the population decline. This result suggests that provisioning with supplemental food may allow hares to spend less time spent foraging away from cover. On FENCE body mass did not decline, suggesting that lower risk of predation may permit hares to forage more in open habitats. Reduction of predation risk seems to be a necessary factor for hares to maintain high reproductive output during the cyclic decline. Krebs et aL (1 986b) found that extra natural food supplied on a control area during the population decline prevented loss of body mass, but did not increase survival. It seems that even if hares have access to abundant forage, stress associated with higher predation risk may lead to reduced reproduction. Desy et a!. (1990) found that the presence of predators limited access of prairie voles (Microtus ochrogaste,) to food supplies and resulted in decreased body condition and slower rates of maturation. Keith and Windberg (1978) and Cary and Keith (1979) showed that female body weight changes during the winter are correlated with litter size and pregnancy rate in summer. When over-winter loss of body mass is high,  103  fecundity is low. Reproduction remained low for at least three years following the hare peak. Their interpretation of these results is that poor winter nutrition during the decline caused the reproductive decline. However, this does not overcome the problem of why this reproductive decline continues even when food increases in late summer. While I agree in general that poor nutrition contributes to reproductive loss, I suggest that predation risk restricts access to food resources and so causes physiological stress associated with poor condition. This results in suppression of hare reproduction over a long period. The results of the experiments reported here are consistent with this hypothesis. Behavioural interactions between hares may also play a role in determining condition in summer, but given that poor condition is evident when population density is declining, I think this is unlikely. Even when hares were in good condition at the population peak, O’Donoghue and Krebs (1992) observed the highest rate of stillborn young in third litters. They suggested that the costs associated with nursing two previous litters may contribute to the high loss.  Reproductive Costs of Predation Risk  Many studies have shown that mating activity increases predation risk (Ryan 1985; Magnhagen 1991). For example, Cushing (1985) found that estrous female prairie deer mice (Peromyscus maniculatus bairdi) were more vulnerable to predation by weasels (Mustela nivalis) than were diestrous mice. These results indicate increased risk of predation associated with a female’s attempt to reproduce, and are of considerable evolutionary significance (Magnhagen 1991). Increased predation risk has also been shown to reduce reproductive effort. Sih eta!. (1990) showed that food deprivation had no effect on the mating  104  behaviour of water striders (Gerris remigis); however, predation risk decreased the number of matings by about 50%. Ylönen et a!. (1992) found that the odour of small mustelids (M. nivalis and M. erminea) delayed sexual maturation of young voles (Clethrionomys spp.) and suppressed female reproduction under laboratory conditions. They also reported that female red-backed voles exposed to predator odour at the beginning of pregnancy had foetuses which were about 25% lighter than those of control females, suggesting that stress associated with high predation risk may influence the survival of offspring. These results suggest that the consequences of poor condition in one breeding season may carry over to the next generation. Korpimâki et a!. (1994) recently conducted a field experiment where reproduction of voles was measured at different weasel densities. Their results suggest that the presence or scent of small mustelids decreases the reproductive rate of voles. The mechanisms by which increased predation risk may lead to decreased reproduction have not been studied in detail. Bronson (1984) found that if access to food was limited (a common cause of elevated stress), reproduction in female house mice was adversely affected. These results are not unexpected because, in general, ovulation in mammals is regulated indirectly by female energy reserves (Bronson and Manning 1991). Bronson (1984) also reported that if the amount of food given to a weanling female was restricted, her subsequent reproductive development was inhibited. Mech et aL (1991) showed that the mass and survival of white-tailed deer (Odocoileus virginianus) fawns was directly related to maternal nutrition during gestation. Similarly, studies by Albon et a!. (1987) suggested that poor maternal condition in red deer (Genius elaphus) may have permanent repercussions for their offspring.  105  Predation Risk, Stress, and Population Cycles  Christian (1980) reviewed the evidence that the environment of cyclic species at peak densities may have long-term negative consequences on demography through impaired reproduction mediated by endocrine responses to elevated stress. Boonstra and Boag (1992) supported one of Christian’s predictions in a field population of meadow voles (Microtus pennsylvanicus) by showing that stress responses were positively correlated with population density. Although their high density vole population showed a low rate of population decline, they suggested that there were long-term consequences for young from being exposed to high free-corticosteroid levels at the peak. Under laboratory conditions, Mihok and Boonstra (1992) showed that the prior experience of decline-phase female meadow voles had long-term detrimental consequences for the performance of the next two generations. While the mechanism leading to increased stress in wild populations may vary (Chitty 1987), laboratory studies have shown that a variety of pre- and post-natal stresses can have long-lasting effects, and impair reproductive performance for one or two generations (Pollard 1986; Boonstra 1994). Reduced fecundity of snowshoe hares during the population decline is apparently not due to food being absolutely limiting, but to poor condition and elevated stress associated with foraging under high risk of predation. Recent experiments to examine the ability of snowshoe hares to recover from short-term stress in winter (Boonstra and Singleton 1993), suggested that the pituitary adrenocortical feedback system in hares from a declining population was operating normally, but that higher levels of free-cortisol were present in hares in poorer condition. In 1992 when CONTROL hares were in even poorer condition, the deleterious effects of short-term stress were even more pronounced  106  (Boonstra and Hik, unpublished data). The cumulative effects of this higher stress (chronic exposure to free-cortisol) may lead to reduced survival or reproduction. The conditions to which hares are exposed during the population decline may have long-term population consequences. Establishing the importance of predation risk on the population dynamics of hares should be a major focus of future research efforts. The predictions of the PSF hypothesis, some of which are supported by this study, can be directly tested in four ways. First, female reproduction should decrease on FOOD but not on FENCE. Second, other indicators of stress (i.e. free-cortisol levels; Boonstra and Singleton 1993), should be lower on FENCE than on FOOD. Third, female reproduction should remain low on CONTROL and FOOD for at least one year following the decline in predator numbers (and hence predation risk). This could be tested by monitoring reproduction of FOOD or CONTROL females inside one of the fenced treatment areas. Finally, it is not clear why females are unable to use abundant, high quality summer food to improve their condition during the summer to increase reproduction. The absence of second and third litters suggests that chronic stress associated with poor winter condition mediated by predation risk may limit reproductive output. If this is true, then providing high quality food to hares in summer will not increase reproduction in that season. Whittaker and Thomas (1983) showed that the total number of days of potential metabolic support derived from neutral lipid and protein reserves of hares was significantly higher in summer than winter (6.0 days compared with 3.8 days, respectively), but much of this energy may be used for lactation (O’Donoghue and Krebs 1992). In addition, the hares Whittaker and Thomas studied were collected during a population peak. During a population decline summer energy reserves may be  107  less. The allocation of variable resources to reproduction and other activities (i.e. Reznick and Yang 1993), has not been investigated in hares.  CONCLUSIONS  These experiments show that snowshoe hares trade-off body mass and survival during the population decline. Anti-predator behaviours in response to increased predation risk may result in reduced body mass and reproduction of hares. In the absence of terrestrial predators, hares behaved in a manner consistent with the predictions of the food hypothesis: body mass was maintained at the expense of lower survival. There appears to be a direct link between female body mass at the end of winter and subsequent reproduction. There is a suggestion that conditions of high predation risk during the decline phase have permanent detrimental effects on ability of hares to recover from ‘stress’. Hares may not be able to maintain physiological homeostasis in the face of environmental stress (higher predation risk), because of a deterioration in the endocrine feedback system. This may lead to a delay in the recovery of populations from low numbers if reproduction and offspring fitness are affected. Several ways to test this hypothesis using the ongoing experimental manipulations at Kluane are suggested.  108  CHAPTER 5 POPULATION CYCLES, PREDATION RISK, AND THE GHOST OF PREDATORS PAST In this concluding chapter, I describe the major factors thought to influence population cycles of small mammals. In particular, I examine the potential for sublethal effects of predation to influence snowshoe hare population dynamics, and I reconsider evidence supporting a variation of the stress hypothesis (Christian 1950, 1980). Finally, I summarize the main results of this thesis and the potential role of predation risk in shaping snowshoe hare behaviour and population dynamics.  What Causes Population Cycles?  Periodic, multi-annual fluctuations (cycles) in abundance are characteristic of a number of species, including some voles, lemmings, snowshoe hares, and their predators (Elton 1924; Krebs and Myers 1974; Finerty 1980; Keith 1990; Batzli 1992; Krebs 1993;). Demographic changes associated with population declines have been explained by food limitation (Keith 1974), predation (Hanski etal. 1991; Krebs etal. 1992; Korpimâki 1993), changes in spacing behaviour (Chitty 1967; Krebs 1978), or a variety of ‘multifactor’ interactions (Lidicker 1988; Batzli 1992). No common cause for these cycles has been discovered so far. A major enigma of these cycles is the failure of low density populations to increase more quickly. There is a characteristic extended low-phase which may persist for one generation or more after food has increased and predators have declined. In all populations, the decline and low phase are characterized by *  see Connell (1980) for more ghosts.  109  poor juvenile survival and growth, and the length of the breeding season may also be shortened.  Food  Food was considered to be the cause of density-dependent population regulation in 5 of 21 small mammal populations reviewed by Sinclair (1989). However, food shortage does not appear to be necessary to explain the decline of cyclic populations. For example, attempts to stop the snowshoe hare decline by adding supplemental food have been unsuccessful (Krebs et a!. 1 986a,b; Krebs et a!., unpublished data). Indeed, of nine studies that have attempted to prevent population declines of fluctuating populations by adding supplemental food (Boutin 1990), only one succeeded, and it was conducted in a predator-free environment (Ford and Pitelka 1984).  Predation  Some authors have suggested that predation may be sufficient to explain cycles of microtine rodents (Erlinge et a!. 1984; Hansson and Hettonen 1988; Korpimaki 1993), and snowshoe hares (Trostel et al. 1987). Some of these interpretations have been criticized on the basis that delayed density-dependent predation may be destabilizing (Kidd and Lewis 1987), but there is considerable evidence that predation alone is sufficient to generate cycles (Boutin 1994). Nevertheless, some microtine populations have declined in the absence of predators (Chitty 1967; Taitt and Krebs 1985; Krebs 1993; Lambin and Krebs 1991a,b; Boonstra 1994).  110  Spacing Behaviour  Chitty (1967, 1987), and others, have argued that predators cannot account for decreases in body mass and reproductive output characteristic of low density populations. Rather, changes in the spacing behaviour and dispersal of individuals are necessary to cause cyclic fluctuations of small mammals (Krebs 1985; Taitt and Krebs 1985; Lambin and Krebs 1991a). Spacing may affect aggression, the number of individuals gaining breeding status, and dispersal. In all of these cases, a decrease in resource (food) availability to resident females may result, leading to reduced juvenile survival and recruitment. In 14 of 21 studies of small mammal populations reviewed by Sinclair (1989), spacing behaviour was thought to be the cause of densitydependent population regulation. There are a number of explanations for the role of spacing behaviour in generating cyclic fluctuations, including interactions with food supplies and changes in the behaviour or genetic structure of these populations. There is evidence for changes in social mortality correlated with population density (Krebs 1985) that may explain some cycles. However, attempts to isolate a genetic-polymorphism responsible for cyclic fluctuations have generally been unsuccessful (Chitty 1987; Boonstra and Boag 1987). In lemmings, social interactions leading to mortality may be responsible for the population decline, and may also explain observed decreases in reproduction and juvenile survival (Krebs 1993). Crowding may reduce the relative availability of cover in a territorial system that is saturated at high density. However, predation may also play an important role in regulating lemming populations (Krebs 1993). Reid et a!. (1993) reported that when populations of collared lemmings (Dicrostonyx torquatus) were protected from  111  predation, survival of litters to weaning and adult survival were significantly enhanced. However, juveniles were not recruited into the population because they dispersed outside of the protected areas. The relative roles of predation, territoriality, and social spacing in regulating these populations is unknown.  Multi-factor Explanations  Since no single factor appears to be sufficient to explain population cycles, a number of authors have proposed multi-factor explanations (Hestbeck 1987; Lidicker 1988; Batzli 1992). Hestbeck (1987) argued that population densities can be regulated by predation, emigration, resource depletion, and behavioral or physiological collapse. He suggested that each of these mechanisms dominates the regulation process over a range of densities. Batzli (1992) has emphasized the importance of predation in a multi-factor hypothesis for vole and lemming cycles. According to this view, populations expand exponentially when food is available and predators are at low density. As populations increase, food becomes scarcer and predators increase. These factors lead to poor juvenile survival. Recovery from low densities is delayed by inversely density-dependent predation, slow recovery of the food supply, and possibly maternal effects. Desy and Batzli (1989) and Desy et al. (1990) have used such an approach to study population regulation of prairie voles. Keith and his colleagues (Keith 1974; Keith and Windberg 1978; Vaughan and Keith 1981) originally argued that the demography of snowshoe hare populations was primarily a delayed density-dependent nutritional problem. They suggested that food shortage at high or low density resulted in increased over-winter weight loss, decreased over-winter survival, decreased reproduction, and decreased growth rates. However, evidence that hares actually run short of  112  food at any time during the cycle is equivocal (Keith et a!. 1984; Smith et a!. 1988; Chapter 2). Keith et al. (1984) and Keith (1990) later emphasized the importance of predation in hare declines: “over-winter food shortage was still believed to trigger the decline, but with starvation deaths predominating over a shorter period and malnutrition markedly increasing hare vulnerability to predators. Such predation was further amplified by severe cold.” (Keith 1990, p. 177). Recent studies have also recognized the importance of predation risk on habitat use by hares (Wolff 1980; Keith et a!. 1984; Sievert and Keith 1985; Smith eta!. 1988). Relative food shortage at high hare numbers may facilitate the deaths of hares from predation. Furthermore, the consequences of greater predation risk on the behaviour and endocrine physiology of hares may provide a mechanism to explain the lag in recovery of populations.  The Stress Hypothesis Revisited  As mentioned previously, spacing behaviour or intraspecific competition for space has long been recognized as a possible regulatory factor of small mammal population cycles (Krebs 1985). Chitty (1952) suggested that the cause of the population decline of Microtus agrestis at Lake Vyrnwy (1936-1939) was increased intraspecific aggression, possibly leading to higher physiological stress. Juveniles born at the population peak had lower survival and lower fertility, and were in poor condition. Mechanisms that could account for maternal effects were poorly understood at the time, but physiological endocrine responses (stress) were recognized as being potentially important in regulating animal populations (Calhoun 1949; Christian 1950). This early stress hypothesis proposed that intraspecific interactions at high density lead to phenotypic physiological changes that reduce births and increase deaths.  113  The endocrine stress hypothesis of Christian was abandoned by Chitty (Chitty and Phipps 1966) because it did not explain the sudden decline of seemingly healthy individuals. Subsequent studies tested the polymorphic behaviour hypothesis (Chitty 1958, 1967), which postulates that individual differences in spacing behaviour have a genetic basis and respond to natural selection. Yet, despite many attempts to demonstrate a genetic basis for population fluctuations (see Krebs and Myers 1974; Chitty 1987; Spears and Clarke 1988), this hypothesis has not been widely supported (Boonstra and Boag 1987). Recent evidence supports a version of the original stress hypothesis (Christian 1980, Boonstra and Boag 1992, Mihok and Boonstra 1992, Boonstra and Singleton 1993). Chilly (1993) also suggests that inter generational maternal effects, of the type described below, have the potential to provide a common explanation for vole and snowshoe hare cycles. Nevertheless, some of the evidence supporting the genetic polymorphic behaviour hypothesis cannot be explained in any other way at present. The possibility that there is selection for different behavioural or physiological characters during population declines should not be abandoned (Pollard 1986). Interactions between individuals may cause population changes by altering the ability of animals to cope with increased environmental stress associated with high density. One of the early explanations for the cyclic decline of snowshoe hares was a physiological condition known as ‘shock disease’ (Green and Larson 1938), characterized by low levels of blood glucose and liver glycogen. It was suggested that overcrowding of hares at high density may result in strong maternal effects (referred to as “physiological derangements derived in utero”) leading to reduced viability of generations born during the population peak (Chitty 1952). However, this sort of stress was usually  114  observed only when hares were being trapped, handled, or crowded artificially (Chilly 1959), suggesting that physiological stress was rare in wild populations. The hypothalamo-pituitary-adrenocortical feedback system plays a central role in allowing animals to adapt to environmental challenges, or stress (Munck et al. 1984). Increased stress (i.e. poor nutrition, high risk of predation) may reduce the ability of this feedback system to respond. Adrenal glucocorticoids play a crucial role by shunting energy to muscles and away from other tissues, and in allowing animals to return to homeostasis following elevated stress. However, chronic exposure to glucocorticoids has deleterious consequences such as steroid diabetes, fatigue, infertility, inhibition of growth, and impaired resistance to disease (Munck et a!. 1984; Boonstra 1994). In the case of the brown antechinus (Antechinus stuartil), stress induced by aggressive interactions at the time of mating resulted in a tripling of plasma free corticosteriod concentration, which led to total post-mating male mortality (Bradley et a!. 1980). This is an extreme example, but demonstrates the possible role of increased adrenocortical activity for population regulation. The weight of adrenal glands provides a rough index of adrenocortical hypertrophy and cortical hormone production. Studies of snowshoe hares at Rochester, Alberta, found that mean adrenal weights were not significantly related to population density or levels of nutrition (Keith 1990). Feist (1980) also reported no difference in adrenal stress response in decline-phase populations. These studies suggested that hares showed no significant stress response as a result of high density. Recently, Boonstra and Singleton (1993) and Boonstra and Hik (unpublished data), have used radio-immunoassay techniques to measure hormone levels in field populations of hares during a population decline, from a control area and an area protected from terrestrial predators where rabbit chow was provided. Their results indicate that although the  115  pituitary-adrenocortical feedback system seems to be operating normally, hares in poorer condition (control areas) have higher levels of free cortisol. Chronic exposure to high levels of free cortisol may have long-term effects on the behaviour, growth, and reproductive performance of hares born to stressed mothers (see Pollard 1986). Maternal effects refer to the influence of environmental conditions experienced by mothers on the growth, survival, and fitness of offspring. The importance of maternal effects in regulating populations has recently been considered for insect (Rossiter 1991; Peckarsky et al. 1993) and mammal (Mech eta!. 1991; Mihok and Boonstra 1992) populations. Albon eta!. (1987) and Mech etal. (1991) have suggested that poor maternal condition has negative consequences for offspring fitness in wild deer populations. Is a one-generation lag in reproduction generated by maternal condition sufficient to account for population cycles? May (1974, 1976) used a single species model to predict that a 2-3 year lag-time for population recovery of hare populations could generate the observed 8 to 11 year cycle. Possibly, maternal effects lasting one generation may be sufficient to explain snowshoe hare population cycles. This potential reproductive lag complements the delayed density-dependent predation response suggested by Trostel eta!. (1987). Similarly, a lag of 9-12 months could generate microtine cycles of 3 to 4 years. Boonstra (1994) has suggested that such a lag could be generated by density dependent inhibition of maturation in peak years of microtine population density to produce time lags of the correct magnitude. One significant advance in our understanding of hare population dynamics over the last forty years, and one of the most important behavioural differences between hares and microtine rodents, is that spacing behaviour by itself appears to be relatively unimportant in regulating hare populations (Boutin  116  1980, 1984a; Krebs 1986). Although adult hares show dominance hierarchies (Graf 1985), and adult residents may exclude immigrant juveniles (Boutin 1 984a), they generally do not generally compete for space, unless food is limiting (Sinclair 1986; Ferron 1993). Thus, social interactions between individual hares may be less important than other types of environmental stress, such as risk of predation.  Summary of the Main Results of this Thesis  Predation seems to be the proximate cause of mortality of snowshoe hares and may be sufficient to generate the population decline. In this thesis I argue that sublethal effects associated with predation risk are the first step in a cascade of behavioural and physiological responses leading to a further decline during the low-phase of the cycle. The sensitivity of snowshoe hares to increased risk of predation during the cyclic population decline leads to changes in foraging behaviour which result in poor condition and lower reproductive output. Poor female condition during the cyclic population decline may also affect the fitness of juveniles, resulting in a one- or two-generation lag in recovery of the population. This may explain why the low period of the 10-year cycle persists when predators are scarce and food is abundant. The main elements of the predation risk hypothesis are shown in Fig. 5.1. During the hare population decline (a) there is an increase in the predator-to hare ratio, and therefore risk of predation (b). Survival of hares declines in the early decline (c). However, available forage increases during this period (d), and there is no evidence that hares are absolutely food limited. During this decline period female body mass (e) and reproduction (f) are reduced, even though per-capita food availability is increasing. However, there is a notable  117  Fig. 5.1 Summary of main results for CONTROL which support the predation risk hypothesis. (a) hare density (#/ha); (b) predator-prey ratio of terrestrial and avian predators; (c) survival of hares during January-April period based on radio-telemetry; (d) SaIix forage availability per hare at the end of winter (kg/hare), and 5-mm stems of Salix, Picea and Betula browse available in mid March; (e) body mass of female hares in mid-April; (f) total reproductive output of female hares; (g) habitat use by female hares during late winter.  118  (a) HARE DENSflY (NO./HA)  (b) PREDATOR HARE RAT1O  Cc) MEAN 30-DAY SURVIVAL (JAN MAY) -  50  (d) 40  SALIX FORAGE AVAILABLE AT END OF WINTER (I(G/HARE) ()  30  10 0  Ce) FEMALE BODY MASS IN MID-APRIL (g)  (f) TOTAL REPRODUCTIVE OUTPUT (YOUNG/YEAR)  (g)  LATE-WINTER HABITAT USE (%)  BROWSE AVAILABLE IN MID-MARCH (KG/HARE/DAY)  119  shift in patterns of winter habitat use by hares (g): hares increase the proportion of closed habitats used. In general hares appear to have few behavioural options available to reduce predation risk. Observations indicate that the quality of the diet and proportion of time spent foraging may also decline (Chapter 3). These changes in foraging behaviour contribute to poor condition. These observations support the predation risk hypothesis, which is a mixture of the four hypotheses of the snowshoe cycle summarized in Chapter 1. There is little evidence that winter food resources are depleted near the population peak. However, relative food shortage appears to lead to poor hare nutrition, lower body mass, reduced fecundity, and increased susceptibility to predation. The evidence from Kluane suggests that predators alone are sufficient to initiate the hare decline (Boutin 1994) and drive hare numbers down. However, hares are in poorest condition 2 or 3 years after the population peak, when food is abundant (Chapter 4). Therefore, there appears to be a synergistic interaction between food and predation, but I suspect it operates in the opposite order to that originally predicted by Keith (1974): predation causes the initial decline, and then predation-sensitive foraging leads to poor condition and higher rates of loss. Changes in plant quality (nutrients or defences) do not appear to cause the decline. However, differences in the chemistry of forage twigs may help to explain why hares lose body mass during this time. The increase in Picea forage used by hares during the decline may be an important cause of poor condition at this time. The polymorphic behaviour hypothesis is also relevant. There may not be a genetic polymorphism responsible for differences in body mass and reproductive output of low-phase hare populations, but there is a suggestion of physiological differences between high-phase and low-phase animals. During  120  the decline, females are exposed to high predation risk, which may result in reduced access to food. This leads to poor condition at the end of winter and may leave an indelible physiological imprint leading to lower fitness (Boonstra and Singleton 1993). These effects are phenotypic in origin, but may result in strong maternal effects that are passed to the next generation. Predation alone cannot explain why hares suffer a reproductive decline, and why the low-phase of the cycle persists for as long as it does. Further modelling and field studies are needed to test the idea that predation is the factor which initiates a cascade of sublethal effects, mediated by predation-sensitive foraging, which result in a physiological collapse of the population and lag in the cycle. This result has been anticipated by, and builds on the pioneering studies of Green and Evans (1940), and the extensive experimental work of Lloyd Keith and his colleagues at Rochester, Alberta (see Keith 1990). Further work by Charles Krebs and his colleagues at Kluane, Yukon (Krebs et a!. 1992) has refined these hypotheses of population regulation. Other factors, such as weather, may modify the amplitude and period of the cycle (Finerty 1980; Sinclair et al. 1993). The behavioural and physiological consequences of elevated predation risk at the cyclic peak provide a potential mechanism for the observed lag in the recovery of these populations.  Predator Avoidance Behaviour and Population Dynamics  Where predation is implicated in the regulation of prey populations, anti predator behaviour will probably play a role in the process. The cumulative effects of predation and starvation determine the state of animals, and theoretical work suggests that predation and starvation interact to cause population declines (McNamara and Houston 1987, 1990; Houston eta!. 1993).  121  The framework for exploring this problem of a trade-off between gaining energy and avoiding predation has developed rapidly in recent years (Ludwig and Rowe 1990; Houston etal. 1993; Clark 1993,1994). The foraging behaviour of individual animals will often depend on their energetic state and the timing of reproduction. There have been numerous studies on the effects of increased predation risk on foraging behaviour of small mammals (Holmes 1984; Kotler 1984; Anderson 1986; Desy eta!. 1990; Kieffer 1991; Cassini and Galante 1992; Dickman 1992; Kotler et a!. 1992; Hughes et a!. 1993), and in most of these studies increased risk of predation has resulted in shifts in patterns of habitat use, diet, and time spent active or foraging. The influence of predation risk on population dynamics has been more difficult to demonstrate (Ives and Dobson 1987; FitzGibbon and Lazarus 1994). Recent theoretical work by Abrams (1986, 1989, 1990, 1991, 1 992a,b,c, 1993), Abrams and Matsuda (1993), and Matsuda and Abrams (1993) has suggested that adaptive behaviour of predators and anti-predator behaviour of prey can have a large impact on predator-prey dynamics. Matsuda and Abrams (1993) emphasize that seemingly adaptive traits, such as anti-predator behaviour, do not always increase the mean fitness of a population. Under some circumstances it is possible that the evolution of anti-predator traits will lead to the extinction of the prey species, particularly if predator populations are maintained by alternative prey species. Few field tests to examine these possibilities have been conducted (Pech eta!. 1994). Our understanding of predator-prey dynamics has been greatly improved by accurate measurements of mortality in natural populations (e.g. Boutin et at 1986; Boutin 1994). The numerical and functional responses of predators have been used to interpret the effects of predators on the survival and recruitment of prey species, and vice versa. Predation plays a large role in generating hare  122  cycles (Trostel et a!. 1987), but the direct effects of predation seem to be insufficient to explain the slow recovery of low-phase populations. Direct sublethal effects of increased predation risk during the population decline also appear to be necessary. A complete understanding of snowshoe hare cycles requires an understanding not only of the causes of mortality, but also of the condition and behaviour of animals surviving or living in the decline. The “ghost of predators past” may have a profound influence on these cycles. Future research efforts should endeavour to see if maternal condition has an influence on the behaviour, survival, and reproductive success of hares. The latency of these effects may affect the population cycle. It will also be important to understand more about spacing behaviour and the interactions between individual hares at different times of the cycle. Although food does not appear to be limiting, hares may compete for sites of protective cover which may be limiting in the environment. If these sites of high quality cover are necessary for hares to reduce their risk of predation, then the availability of cover may influence the lower limit of the population size during the decline.  CONCLUSIONS  The results of the experiments described in this thesis show that antipredator behaviours in response to increased predation risk may result in reduced body mass and reproduction of snowshoe hares. Closed spruce forest appears to provide hares with protective cover from most predators during the population decline, but increased survival is achieved at the cost of lower fecundity. Thus, even though winter food is sufficient, predation risk reinforces the decline of hares during the low phase of the 10-year cycle. All individuals in a population will not be affected equally. The relative importance of predation  123  risk in regulating predator-prey dynamics will depend on the age, sex, reproductive status, and condition of individual prey that survive predator attack (FitzGibbon and Lazarus 1994). Habitat patchiness may also play an important role (Wolff 1981; Fig. 2.1). Further experimental work which explores the balance between conflicting behaviours by expressing them in terms of survival and future reproductive success will lead to greater understanding of the role of predation risk on life history strategies, population dynamics, and community structure (Kotler 1984; Abrams 1990, 1992a; Rowe and Ludwig 1991; Suhonen 1993). The non-lethal effects of predators are pervasive.  124  LITERATURE CITED Abrahams, MN. and Dill, L.M. 1989. A determination of the energetic equivalence of the risk of predation. Ecology 70, 999-1007. Abrams, P.A. 1986. Adaptive responses of predators to prey and prey to predators: the failure of the arms race analogy. Evolution 40, 1229-1247. Abrams, P.A. 1989. Decreasing functional responses as a result of adaptive consumer behavior. Evolutionary Ecology 3, 95-114. Abrams, P.A. 1990. The evolution of anti-predator traits in response to evolutionary change in predators. Qikos 59, 147-156. Abrams, P.A. 1991. Strengths of indirect effects generated by optimal foraging. Oikos62, 167-176. Abrams, P.A. 1 992a. Adaptive foraging by predators as a cause of predatorprey cycles. Evolutionary Ecology 6, 56-72. Abrams, P.A. 1 992b. Why don’t predators have positive effects on prey populations? Evolutionary Ecology 6, 449-457. Abrams, P.A. 1 992c. Predators that benefit prey and prey that harm predators: unusual effects of interacting foraging adaptations. American Naturalist 140, 573-600. Abrams, P.A. 1993. Why predation rate should not be proportional to predator density. Ecology 74, 726-733. Adams, P.A. and Matsuda, H. 1993. Effects of adaptive predatory and antipredator behaviour in a two-prey-one-predator system. Evolutionary Ecology 7, 312-326. Adams, L. 1959. An analysis of a population of snowshoe hares in northwestern Montana. Ecological Monographs 29, 141-1 70. Akçakaya, H.R. 1992. Population cycles of mammals: evidence for a ratio dependent predation hypothesis. Ecological Monographs 62, 119-142. Albon, S.D., Clutton-Brock, T.H., and Guinness, F.E. 1987. Early development and population dynamics in red deer. II. Density-independent effects and cohort variation. Journal of Animal Ecology 56, 69-81. Anderson, P.K. 1986. Foraging range in mice and voles: the role of risk. Canadian Journal of Zoology 64, 2645-2653.  125  Batzli, G.O. 1992. Dynamics of small mammal populations: a review. In Wildlife 2001: Populations. (D.R. McCullough and R.E. Barrett, eds.). pp. 831-850. Elsevier, London. Bider, J.R. 1961. An ecological study of the hare Lepus americanus. Canadian Journal of Zoology 39, 81-1 03. Boonstra, R. 1994. Population cycles in microtines: the senescence hypothesis. Evolutionary Ecology 8, in press. Boonstra, R. and Boag, P.T. 1987. A test of the Chilly hypothesis: Inheritance of life-history traits in meadow voles Microtus pennsylvanicus. Evolution 41, 929-947. Boonstra, R. and Boag, P.T. 1992. Spring declines in Microtus pennsylvanicus and the role of steroid hormones. Journal of Animal Ecology 61, 339-352. Boonstra, R. and Singleton, G.R. 1993. Population declines in the snowshoe hare and the role of stress. General and Comparative End rocrinology 91, 126-143. Boulanger, J. 1993. Evaluation of capture-recapture estimators using a cyclic snowshoe hare population. Unpublished M.Sc. Thesis, University of British Columbia. Boulanger, J. and White, G. 1990. A comparison of home-range estimators using Monte Carlo simulations. Journal of Wildlife Management 54, 310-315. Bouskila, A. and Blumstein, D.T. 1992. Rules of thumb for predation hazard assessment: predictions from a dynamic model. American Naturalist 139, 161-176. Boutin, S. 1980. Effect of spring removal experiments on the spacing behaviour of female snowshoe hares. Canadian Journal of Zoology 58, 2167-2174. Boutin, S. 1 984a. The effect of conspecifics on juvenile survival and recruitment of snowshoe hares. Journal of Animal Ecology 53, 623-637. Boutin, 5. 1 984b. Effect of late winter food addition on numbers and movements of snowshoe hares. Qecologia 62, 393-400. Boutin, S. 1984c. Home range size and methods of estimating snowshoe hare densities. Acta Zoologica Fennica 171, 275-278.  126 Boutin, S. 1990. Food supplementation experiments with terrestrial vertebrates: patterns, problems, and the future. Canadian Journal of Zoology 68, 203220. Boutin , S. 1994. Testing predator-prey theory by studying fluctuating populations of small mammals. Wildlife Research, in press. Boutin, S.A., Gilbert, B.S., Krebs, C.J., Sinclair, A.R.E., and Smith, J.N.M. 1985. The role of dispersal in the population dynamics of snowshoe hares. Canadian Journal of Zoology 64, 106- 6115. Boutin, S.A., Krebs, C.J., Sinclair, A.R.E., and Smith, J.N.M. 1986. Proximate causes of losses in a snowshoe hare population. Canadian Journal of Zoology 64, 606- 610. Bradley, A.J., McDonald, l.R., and Lee, A.K. 1980. Stress and mortality in a small marsupial (Antechinus stuartli, Macleay). General and Comparative Endrocrinology 40, 188-200. Bronson, F.H. 1984. The adaptability of the house mouse. Scientific American 250, 116-125. Bryant, J.P., Wieland, G.D., Clausen, T., and Kuropat, P. 1985. Interactions of snowshoe hares and feitleaf willow in Alaska. Ecology 66, 1564-1573. Bryant, J.P., Kuropat, P.J., Reichart, P.B., and Clausen, T.P. 1991a. Controls over the allocation of resources by woody plants to chemical antiherbivore defense. In Plant Defenses Against Mammalian Herbivory (Eds. R.T. Palo and C.T. Robbins) pp. 84-102. CRC Press, Boca Raton. Bryant, J.P., Reichart, P.B., Clausen, T.P., Provenza, F.D., and Kuropat, P.J. 1991b. Woody plant-mammal interactions. In Herbivores: Their Interaction with Plant Metabolites. Vol. 2 (Ed. G.A. Rosenthal and M.R. Berenbaum). pp. 344-370. Academic Press, New York. Buehler, D.A. and Keith, L.B. 1982. Snowshoe hare distribution and habitat use in Wisconsin. Canadian Field-Naturalist 96, 19-29. Calhoun, J.B. 1949. A method for self-control of population growth among mammals living in the wild. Science 109, 333-335. Carreker, R.G. 1985. Habitat suitability index models: snowshoe hare. U.S. Fish and Wildlife Service Biological Report 82(10.101). 21 pp. Cary, J.R. and Keith, L.B. 1979. Reproductive change in the 10-year cycle of snowshoe hares. Canadian Journal of Zoology 57, 375-390.  127 Cassini, M.H. and Galante, M.L. 1992. Foraging under predation risk in the wild guinea pig: the effect of vegetation height on habitat utilization. Annales Zoologi Fennici 29, 285-290. Caughley, G. and Krebs, C.J. 1983. Are big mammals simply little mammals writ large? Qecologia 59, 7-17. Charnov, E., Orians, G.H. and Hyatt, K. 1976. Ecological implications of resource depression. American Naturalist 110, 247-259. Cheeke, P.R. 1983. The significance of fibre in rabbit nutrition. Journal of Applied Rabbit Research 6, 103-1 06. Chitty, D. 1952. Mortality among voles (Microtus agrestis) at Lake Vyrnwy, Montgomeryshire in 1936-9. Philosophical Transactions of the Royal Society London, Series B, 236, 505-552. Chitty, D. 1958. Self-regulation of numbers though changes in viability. Cold Spring Harbour Symposium on Quantitative Biology 22, 277-280. Chitty, D. 1959. A note on shock disease. Ecology 40, 728-731. Chitty, D. 1967. The natural selection of self-regulatory behaviour in animal populations. Proceedings of the Ecological Society of Australia 2, 51-78. Chitty, D. 1987. Social and local environments of the vole Microtus townsend!!. Canadian Journal of Zoology 65, 2555-2566. Chilly, D. 1993. Do Lemmings Commit Suicide? Beautiful Hypotheses and Ugly Facts. Unpublished Manuscript. Chitty, D., and Phipps, E. 1966. Seasonal changes in survival in mixed populations of two species of vole. Journal of Animal Ecology 35, 313-331. Christian, J.J. 1950. The adreno-pituitary system and population cycles in mammals. Journal of Mammalogy 31, 247-259. Christian, J.J. 1980. Endocrine factors in population regulation. In Biosocial mechanisms of population regulation (Eds. M.N. Cohen, R.S. Malpass and H.G. Klein) pp. 55-115. Yale University Press, New Haven. Clark, C.W. 1993. Dynamic models of behaviour: an extenstion of life history theory. Trends in Ecology and Evolution 8, 205-209. Clark, C.W. 1994. Anti-predator behaviour and the asset protection principle. Behavioural Ecology, in press.  128  Connell, J.H. 1980. Diversity and the coevolution of competitors, or the ghost of competition past. Qikos 35, 131 -1 38. Curio, E. 1993. Proximate and developmental aspects of antipredator behaviour. Advances in the Study of Behaviour 22, 135-238. Cushing, B.S. 1985. Estrous mice and vulnerability to weasel predation. Ecology 66, 1976-1 978. de Poorter, M. 1984. An experimental test of predictions from different hypotheses of self-regulation in the snowshoe hare (Lepus americanus Erxleben, 1777). Unpublished Ph.D. Thesis, Vrije Universiteit Brussel. Desy, E.A. and Batzli, G.O. 1989. Effects of food and predation on behaviour of prairie voles: a field experiment. Ecology 70, 411-421. Desy, E.A., Batzli, G.O., and Jike, L. 1989. Comparison of vole movements assessed by live trapping and radiotracking. Journal of Mammalogy 70, 652656. Dickman, C.R. 1992. Predation and habitat shifts in the house mouse, Mus domesticus. Ecology 73, 313-322. Dixon, K. and Chapman, K. 1980. Harmonic mean measure of animal activity areas. Ecology 61, 1040-1044. Doyle, F.l. and J. N. M. Smith. 1994. Population responses of northern goshawks to the 10-year cycle in numbers of snowshoe hares. Condor, submitted. Elton, C.S. 1924. Periodic fluctuations in the numbers of animals: their causes and effects. Journal of Experimental Biology 2, 119-163. Elton, C. and Nicholson, M. 1942. The ten-year cycle of numbers of lynx in Canada. Journal of Animal Ecology 11, 215-244. Erlinge, S., Gäransson, G., Hogstedt, 3., Jansson, G., Liberg, 0., Loman,J., Nilsson, l.N., von Schantz, T., and Sylven, M. 1984. Can vertebrate predators regulate their prey? American Naturalist 123, 125-133. Feinsinger, P., Spears, E.E., and Poole, R.W. 1981. A simple measure of niche breadth. Ecology 62, 27-32. Feist, D.D. 1980. Corticosteroid release by adrenal tissue of Alaska snowshoe hares in a year of population decline. Journal of Mammalogy 61, 134-136.  129  Ferron, J. 1993. How do density and food supply influence social behaviour in the snowshoe hare (Lepus americanus)? Canadian Journal of Zoology 71, 1084-1089. Finerty, J.P. 1980. The population ecology of cycles in small mammals. Yale University Press, New Haven. FitzGibbon, C.D. and Lazarus, J. 1994. Anti-predator behaviour of Serengeti ungulates: individual differences and population consequences. In Serengeti Il: Research, Management and Conservation of an Ecosystem (Eds. A.R.E. Sinclair and P. Arcese), University of Chicago Press, Chicago. in press. Ford, R.G. and Pitelka, F.A. 1984. Resource limitation in populations of the California vole. Ecology 65, 122-136. Gilbert, B.S. 1990. Use of winter feeding craters by snowshoe hares. Canadian Journal of Zoology 68, 1600-1602. Gilbert, B.S. and Boutin, S. 1991. Effect of moonlight on winter activity of snowshoe hares. Arctic and Alpine Research 23, 61-65. Gilliam, J.F. and Fraser, D.F. 1987. Habitat selection under predation hazard: test of a model with foraging minnows. Ecology 68, 1856-1862. Graf, R.P. 1985. Social organization of snowshoe hares. Canadian Journal of Zoology 63, 468-474. Graf, R.P. and Sinclair, A.R.E. 1987. Parental care and adult aggression towards juvenile snowshoe hares. Arctic 40, 175-178. Grange, W.B. 1932. The pelage and colour changes of the snowshoe hare, Lepus americanus phaenotus Allen. Journal of Mammalogy 13, 99-116. Green, R.G. and Larson, C.L. 1938. A description of shock disease in the snowshoe hare. American Journal of Hygiene 28, 190-212. Green, R.G. and Evans, C.A. 1940. Studies on a population cycle of snowshoe hares on the Lake Alexander area. II. Mortality according to age groups and seasons. Journal of Wildlife Management 4, 267-278. Gross, J.E., Hobbs, N.T., and Wunder, B.A. 1993. Independent variables for predicting intake rate of mammalian herbivores: biomass density, plant density, or bite size? Qikos 68, 75-81.  130 Hanski, I., Hansson, L. and Henttonen, H. 1991. Specialist predators, generalist predators, and the microtine rodent cycle. Journal of Animal Ecology 60, 353-367. Hart, J.S. and Pohi, J. 1965. Seasonal acclimatization in varying hares. Canadian Journal of Zoology 43, 731-744. Hassell, M.P. and R.M. May. 1985. From individual behaviour to population dynamics. In Behavioural Ecology: Ecological Consequences of Adaptive Behaviour (Eds. R.M. Sibly and R.H. Smith) pp. 3-32. Blackwells, Oxford. Hestbeck, J.B. 1987. Multiple regulation states in populations of small mammals: a state-transition model. American Naturalist 129, 520-532. Holmes, W.G. 1984. Predation risk and foraging behavior of the hoary marmot in Alaska. Behavioural Ecology and Sociobiology 15, 293-301. Houston, A.l., McNamara, J.M., and Hutchinson, J.M.C. 1993. General results concerning the trade-off between gaining energy and avoiding predation. Philosophical Transactions of the Royal Society London, Series B, 341, 375397. Huggard, D.J. 1993. The effect of snow depth on predation and scavenging by wolves. Journal of Wildlife Managment 57, 382-388. Hughes, J.J. and Ward, D. 1993. Predation risk and distance to cover affect foraging behaviour in Namib Desert gerbils. Animal Behaviour, in press. Hughes, J.J., Ward, D. and Perrin, M.R. 1993. The effects of predation risk and competition on habitat selection and activity of gerbils in the Namib Desert. Ecology, in press. lves, A.R. and Dobson, A.P. 1987. Antipredator behaviour and the population dynamics of simple predator-prey systems. American Naturalist 130, 431447. Jedrzejewski, W., Rychlik, L., and Jedrzejewski, B. 1993. Responses of bank voles in odours of seven species of predators: experimental data and their relevance to natural predator-vole relationships. Oikos 68, 251-257. Keith, L.B. 1963. Wildlife’s ten-year cycle. University of Wisconsin Press, Madison. Keith, L.B. 1966. Habitat vacancy during a snowshoe hare decline. Journal of Wildlife Management 30, 828-832.  131 Keith, L.B. 1974. Some features of population dynamics in mammals. Proceedings International Congress of Game Biologists 11, 17-58. Keith, L.B. 1983. Role of food in hare population cycles. Qikos 40, 385-395. Keith, L.B. 1990. Dynamics of snowshoe hare populations. In Current Mammalogy (Ed. H.H. Genoways). pp. 119-195. Plenum Press, New York. Keith, L.B., Todd, A.W., Brand, C.J., Adamcik, R.S. and Rusch, D.H. 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. and Windberg, L.A. 1978. A demographic analysis of the snowshoe hare cycle. Wildlife Monographs 58, 1-70. Keith, L.B., Cary, J.R., Rongstad, O.J., and Brittingham, M.C. 1984. Demography and ecology of a declining snowshoe hare population. Wildlife Monographs 90, 1-43. Keith, L.B., Bloomer, S.E.M., and Willebrand, T. 1993. Dynamics of a snowshoe hare population in fragmented habitat. Canadian Journal of Zoology 71, 1385-1392. Kenward, R. 1990. Ranges IV. Institute of Terrestrial Ecology, Furzebrook, UK. Kidd, N.A.C. and Lewis, G.B. 1987. Can vertebrate predators regulate their prey? A reply. American Naturalist 130, 448-453. Kotler, B.P. 1984. Risk of predation and the structure of desert rodent communities. Ecology 65, 689-701. Kolter, B.P. 1985. Microhabitat utilization in desert rodents: a comparison of two methods of measurement. Journal of Mammalogy 66, 374-378. Kotler, B.P. 1992. Behavioural resource depression and decaying perceived risk of predation in two species of coexisting gerbils. Behavioural Ecology and Sociobiology 30, 239-244. Kotler, B.P., Blaustein, L., and Brown, J.S. 1992. Predation facilitation: the combined effects of snakes and owls on the foraging behavior of gerbils. Annales Zoologici Fennici 29, 199-206. Korpimaki, E. 1993. Regulation of multiannual vole cycles by density dependent avian and mammalian predation? Oikos 66,359-363.  132 Korpimaki, E., Norrdahl, K., and Valkama, J. 1994. Reproductive investment under fluctuating predation risk: microtine rodents and small mustelids. Evolutionary Ecology, in press. Krebs, C.J. 1978. A review of the Chitty Hypothesis of population regulation. Canadian Journal of Zoology 56, 2463-2480. Krebs, C.J. 1985. Do changes in spacing behaviour drive population cycles in small mammals. In Behavioural Ecology (Eds. R.M. Sibley and R.H. Smith). pp. 295-312. Blackwell Scientific Publications, London. Krebs, CJ. 1986. Are lagomorphs similar to other small mammals in their population ecology? Mammal Review 16, 187-194. Krebs, C.J. 1989. Ecological Methodology. Harper and Row, New York. Krebs, C.J. 1993. Are lemmings large Microtus or small reindeer? A review of lemming cycles after 25 years and recommendations for future work. In Biology of Lemmings (Eds. N.C. Stenseth and R.A. lms). pp. 247-260. Linnean Society Symposium Series 15. Academic Press, London. Krebs, C.J. and Myers, J.H. 1974. Population cycles in small mammals. Advances in Ecological Research 8, 267-399. Krebs, C.J., Gilbert, B.S., Boutin, S., Sinclair, A.R.E. and Smith, J.N.M. 1986a. Population biology of snowshoe hares. I. Demography of food-supplemented populations in the southern Yukon, 1976-1984. Journal of Animal Ecology 55, 963-982. Krebs, C.J., Boutin, S., and Gilbert, B.S. 1986b. A natural feeding experiment on a declining snowshoe hare population. Qecologia 70, 194-1 97. Krebs, C. J., Boonstra, R., Boutin, S., Dale, M. R. T., Hannon, S., Martin, K., Sinclair, A. R. E., Smith, J. N. M. and Turkington, R. 1992. What drives the snowshoe hare cycle in Canada’s Yukon? In Wildlife 2001: Populations. (D.R. McCullough and R.E. Barrett, eds.). pp. 886-896. Elsevier, London. Krebs, J.R. and Kacelnik, A. 1991. Decision-making. In Behavioural Ecology Third Edition (Eds. J.R. Krebs and N.B. Davies) pp. 105-136. Blackwells, London. Lambin, X. and Krebs, C.J. 1991 a. Spatial organization and mating system of Microtus townsendii. Behavioural Ecology and Sociobiology 28 353-363. Lambin, X. and Krebs, C.J. 1991b. Can changes in female relatedness influence microtine population dynamics. Oikos 61, 126-132.  133  Lazarus, J. 1990. Looking for trouble. New Scientist 27, 62-65. Lenth, R.V. 1981. On finding the source of a signal. Technometrics 23, 149154. Lidicker, W. Z., Jr. 1988. Solving the enigma of microtine “cycles”. Journal of Mammalogy 69, 225-235. Lima, S.L. 1992. Life in a multi-predator environment; some considerations for anti-predator vigilance. Annales Zoologici Fennici 29, 217-226. Lima, S.L. and Dill, L.M. 1990. Behavioural decisions made under the risk of predation; a review and prospectus. Canadian Journal of Zoology 68, 619640. Litvaitis, J.A. 1990. Differential habitat use by sexes of snowshoe hares (Lepus americanus). Journal of Mammalogy 71, 520-523. Litvaitis, J.A. 1991. Habitat use by snowshoe hares, Lepus americanus, in relation to pelage color. Canadian Field Naturalist 105, 275-277. Litvaitis, J.A., Sherburne, J.A. and Bissonette, J.A. 1 985a. A comparison of methods used to examine snowshoe hare habitat use. Journal of Wildlife Management 49, 693-695. Litvaitis, J.A., Sherburne, J.A. and Bissonette, J.A. 1 985b. Influence of understory characteristics on snowshoe hare habitat use and density. Journal of Wildlife Management 49, 866-873. Longland, W.S. and Price, M.V. 1991. Direct observations of owls and heteromyid rodents: can predation risk explain microhabitat use? Ecology 72, 2261-2273. Ludwig, D. and Rowe, L. 1990. Life-history strategies for energy gain and predator avoidance under time constraints. American Naturalist 135, 686707. Lundberg, P. 1988. Functional responses of a small mammalian herbivore: the disc equation revisited. Journal of Animal Ecology 57, 999-1006. MacLulich, D.A. 1937. Fluctuations in the Numbers of the Varying Hare (Lepus americanus). University of Toronto Studies, Biology Series 43. Magnhagen, C. 1991. Predation risk as a cost of reproduction. Trends in Ecology and Evolution 6, 183-186.  134  Manville, C.J., Barnum, S.A. and Tester, J.A. 1992. Influence of bait on arboreal behavior of Peromyscus leucopus. Journal of Mammalogy 73, 335336. Mangel, M. and Clark, C.W. 1988. Dynamic Modelling in Behavioural Ecology. Academic Press, Princeton, N.J. Matsuda, H. and Abrams, P.A. 1993. Timid consumers: self-extinction due to adaptive change in foraging and anti-predator effort. Theoretical Population Biology, in press. May, R.M. 1974. Stability and Complexity in Model Ecosystems. Princeton University Press, Princeton, N.J. May, R.M. 1976. Models for single populations. In Theoretical Ecology: Principles and Applications (Ed. R.M. May). pp. 4-25. Saunders, Philadelphia, PA. McNamara, J.M. 1990. The starvation predation tradeoff and some behavioural and ecological consequences. In Behavioural Mechanisms of Food Selection. (Ed. R.N. Hughes). pp. 39-59. Springer Verlag, Berlin. McNamara, J.M. and Houston, A.l. 1987. Starvation and predation as factors limiting population size. Ecology 68, 1515-1519. McNamara, J.M. and Houston, A.l. 1990. Starvation and predation in a patchy environment. In Living in a Patchy Environment. (Eds. B. Shorrocks and I. Swingland). pp. 23-43. Oxford University Press, Oxford. McNamara, J.M. and Houston, A.l. 1990. State-dependent ideal free distributions. Evolutionary Ecology 4, 298-311. Mech, L.D., Nelson, M.E. and McRoberts, R.E. 1991. Effects of maternal and grandmaternal nutrition on deer mass and vulnerability to wolf predation. Journal of Mammalogy 72, 146-151. Mihok, S. and Boonstra, R. 1992. Breeding performance in captivity of meadow voles (Microtus pennsylvanicas) from decline and increase-phase populations. Canadian Journal of Zoology 70, 1561-1566. Morris, D.W. 1988. Habitat-dependent population regulation and community structure. Evolutionary Ecology 2, 253-269.  135 Munck, A., Guyre, P., and Holbrook, N. 1984. Physiological functions of glucocorticoids during stress and their relation to pharmacological actions. Endocrinology Review 5, 25-44. Murray, D. 1990. Aspects of winter foraging in lynx and coyotes from southwestern Yukon during an increase in snowshoe hare abundance. M.Sc. Thesis, University of Alberta, Canada. Murray, D.L. and Boutin, S. 1991. The influence of snow on lynx and coyote movements in southwestern Yukon: Does morphology affect behaviour? Qecologia 88, 463-469. Nams, M. 1993. Bar coding bees and other wild ways to watch wildlife. Ranger Rick 27, 14-16. Nams, V.0. 1990. Locate II. Pacer, Truro, N.S., Canada. O’Donoghue, M. 1983. Seasonal habitat selection by snowshoe hare in eastern Maine. Transactions Northeast Fish and Wildlife Conference 40, 100-107. O’Donoghue, M. and Krebs, C. J. 1992. Effects of supplemental food on snowshoe hare reproduction and juvenile growth at a cyclic population peak. J. Animal. Ecology 61, 631-641. Oksanen, T, Oksanen, L., and Gyllenberg, M. 1992. Exploitation ecosystems in heterogenous habitat complexes II: impact of small-scale heterogeneity on predator-prey dynamics. Evolutionary Ecology 6, 383-398. Orr, C.D. and Dodds, D.G. 1982. Snowshoe hare habitat preferences in Nova Scotia spruce-fir forests. Wildlife Society Bulletin 10,147-150. Ostfeld, R.S. 1992. Small-mammal herbivores in a patchy environment: individual strategies and population responses. In Effects of Resource Distribution on Animal-Plant Interactions (Eds. M.D. Hunter, T. Ohgushi and P.W. Price). pp. 43-74. Academic Press, San Diego. Otis, D.L., Burnham, K.P., White, G.C., and Anderson, D.R. 1978. Statistical inference from capture data on closed animal populations. Wildlife Monographs 62, 1-135. Partridge, L. and Harvey, RH. 1985. Costs of reproduction. Nature 316, 20. Pease, J.L., Vowles, R.H. and Keith, LB. 1979. Interaction of snowshoe hares and woody vegetation. Journal of Wildlife Management 43, 43-60.  136 Pech, R.P., Sinclair, A.R.E., and Newsome, A.E. 1994. Predation models for primary and secondary prey species. Wildlife Research, in press. Peckarsky, B.L., Cowan, C.A., Penton, M.A., and Anderson, C. 1993. Sublethal consequences of stream-dwelling predatory stoneflies on mayfly growth and fecundity. Ecology 74, 1836-1846. Persson, L. 1993. Predator-mediated competition in prey refuges: the importance of habitat dependent prey resources. Oikos 68, 12-22. Pettersson, L.B. and Brönmark, C. 1993. Trading off safety against food: state dependent habitat choice and foraging in crucian carp. Qecologia 95, 353357. Pietz, P.J. and Tester, J.R. 1983. Habitat selection by snowshoe hares in north central Minnesota. Journal of Wildlife Management 47, 686-696. Pollock, K.H., Winterstein, S.R., Bunick, C.M., and Curtis, P.D. 1989a. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53, 7-15. Pollock, K.H., Winterstein, S.R., and Conroy, M.J. 1989b. Estimation and analysis of survival distributions for radio-tagged animals. Biometrics 45, 99109. Pollard, I. 1986. Prenatal stress effects over two generations in rats. Journal of Endocrinology 109, 239-244. Pullianinen, E. 1983. The refuge theory and habitat selection in the mountain hare on a subarctic fell in Finnish Forest Lapland. Finnish Game Research 41, 39-44. Radvanyi, A. 1987. Snowshoe hares and forest plantations: a literature review and problem analysis. Canadian Forestry Service, Northern Forestry Centre, Edmonton, AB, Information Report NOR-X-290. Reid, D.G., Krebs, C.J., and Kenney, A. 1993. Predation and collared lemming population dynamics. Abstracts of Sixth International Theriological Congress, Sydney (Ed. M.L. Augee). p. 255. Reznick, D. and Yang, A.P. 1993. The influence of fluctuating resources on life history: patterns of allocation and plasticity in female guppies. Ecology 74, 2011-2019.  137 Rodgers, A. R. and Sinclair, A. R. E. 1994. Diet choice and nutrition of snowshoe hares: interactions of energy, protein and plant secondary compounds. Canadian Journal of Zoology, submitted. Rogowitz, G.L. 1988. Forage quality and use of reforested habitats by snowshoe hares. Canadian Journal of Zoology 66, 2080-2083. Rohner, Ch. and Krebs, C.J. 1994. Snowshoe hares and great horned owl predation: which individuals are at greatest risk? Manuscript in preparation. Rosenzweig, M.L. and Abramsky, Z. 1980. Microtine cycles: the role of habitat heterogeneity. Qikos 34, 141 -146. Rossiter, M.C. 1991. Environmentally-based maternal effects: a hidden force in insect population dynamics? Qecologia 87, 288-294. Rowe, L. and Ludwig, D. 1991. Size and timing of metamorphosis in complex life cycles: time constraints and variation. Ecology 72, 41 3-427. Royama, T. 1992. Analytical Population Dynamics. Chapman and Hall, London. Ryan, M.J. 1985. The Tungara frog: a study in sexual selection and communication. University of Chicago Press, Chicago. Saarikko, J. 1992. Risk of predation and foraging activity in shrews. Annales Zoologici Fennici 29, 291-299. Schmitz, O.J., Hik, D.S., and Sinclair, A.R.E. 1992. Plant chemical defence and twig selection by snowshoe hares: an optimal foraging perspective. Qikos 65: 295-300. Schmutz, J.A. and White, G.C. 1990. Error in telemetry studies: effects of animal movement on triangulation. Journal of Wildlife Management 54, 506510. Scott, D.P. and Yahner, R.H. 1989. Winter habitat and browse use by snowshoe hares, Lepus americanus, in a marginal habitat in Pennsylvania (USA). Canadian Field Naturalist 103:560-563. Shipley, L.A. and Spalinger, D.E. 1992. Mechanics of browsing in dense food patches: effects of plant and animal morphology on intake rate. Canadian Journal of Zoology 70, 1743-1752. Sibly, R.M. and Calow, P. 1986. Physiological Ecology of Animals. Blackwell Scientific Publications, London.  138  Sievert, P.R. and Keith, L.B. 1985. Survival of snowshoe hares at a geographic range boundary. Journal of Wildlife Management 49, 854-866. Sih, A. 1987. Prey refuges and predator-prey stability. Theoretical Population Biology 31, 1-12. Sih, A., Krupa, J., and Travers, S. 1990. An experimental study on the effects of predation risk and feeding regime on the mating behaviour of the water strider. American Naturalist 135, 284-290. Sinclair, A.R.E. 1986. Testing multifactor causes of population limitation: an illustration using snowshoe hares. Qikos 47, 360-364. Sinclair, A.R.E. 1989. Population regulation in animals. In Ecological Concepts (Ed. J.M. Cherrett). pp. 197-241. Symposium of the British Ecological Society 29. Blackwell Scientific Publications, London. Sinclair, A.R.E. and Smith, J.N.M. 1984. Do plant secondary compounds determine feeding preferences of snowshoe hares? Qecologia 61, 403-410. Sinclair, A.R.E., Krebs, C.J., Smith, J.N.M., and Boutin, S. 1988a. Population biology of snowshoe hares. III. Nutrition, plant secondary compounds and food limitation. Journal of Animal Ecology 57, 787-806. Sinclair, A. R. E., M. K. Jogia, and R. J. Andersen. 1988b. Camphor from juvenile white spruce as an antifeedant for snowshoe hares. Journal of Chemical Ecology 14, 1505-1 51 4. Sinclair, A.R.E, Gosline, J.M., Holdsworth, G., Krebs, C.J., Boutin, S., Smith, J.N.M., Boonstra, R., and Dale, M. 1993. Can the solar cycle and climate synchronize the snowshoe hare cycle in Canada? Evidence from tree rings and ice cores. American Naturalist 141, 173-1 98. Sinclair, A.R.E. and Arcese, P. 1994. Population consequences of predation sensitive foraging: the Serengeti wildebeest. Ecology, submitted. Smith, C.H. 1983. Spatial trends in the Canadian snowshoe hare, Lepus americanus, population cycles. Canadian Field-Naturalist 97, 151-1 60. Smith, J.N.M., Krebs, C.J., Sinclair, A.R.E., and Boonstra, R. 1988. Population biology of snowshoe hares. II. Interactions with winter food plants. Journal of Animal Ecology 57, 269-286. -  139 Stuart-Smith, K. 1992. Do lemming, vole, and snowshoe hare cycles affect other small birds and mammals in northern ecosystems? Musk-Ox 39, 181188. Suhonen, J. 1993. Predation risk influences the use of foraging sites by Tits. Ecology 74, 1197-1203. Sullivan, T.P. and Crump, D.R. 1984. Influence of mustelid scent-gland compounds on suppression of feeding by snowshoe hares (Lepus americanus). Journal of Chemical Ecology 10, 1809-1821. Sullivan, T.P. and Crump, D.R. 1986. Feeding responses of snowshoe hares (Lepus americanus) to volatile constituents of red fox ( Vulpes vulpes) urine. Journal of Chemical Ecology 12, 729-739. Sullivan, T.P. and Moses, R.A. 1986. Demographic and feeding responses of a snowshoe population to habitat alteration. Journal of Applied Ecology 23, 5363. Sullivan, T.P. Nordstrom, L.O. and Sullivan, D.S. 1985. Use of predator odors as repellents to reduce feeding damage by herbivores. 1. Snowshoe hares (Lepus americanus). Journal of Chemical Ecology 11, 903-919. Swikhart, R. and Slade, H. 1985. Testing for independance of observations in animal movements. Journal of Wildlife Management 49, 1019-1025. Taitt, M.J. and Krebs, C.J. 1985. Population dynamics and cycles. In Biology of New World Microtus (Ed. R.H. Tamarin). American Society of Mammalogists Special Publication 8, 567-620. Trostel, K., Sinclair, A.R.E., Walters, C.J. and Krebs, C.J. 1987. Can predation cause the 10-year hare cycle? Oecologia 74, 185-192. Underwood, R. 1982. Vigilence behaviour in grazing African antelopes. Behaviour 79, 81 -1 08. Vaughan, M.R. and Keith, L.B. 1981. Demographic response of experimental snowshoe hare populations to overwinter food shortage. Journal of Wildlife Management 45, 354-380. Whittaker, M.E. and Thomas, V.G. 1983. Seasonal levels of fat and protein reserves of snowshoe hares in Ontario. Canadian Journal of Zoology 61, 1339-1345. Wilkinson, L. 1988. Systat. Evanston, IL.  140 Williamson, C.E. 1993. Linking predation risk models with behavioural mechanisms: identifying population bottlenecks. Ecology 74, 320-331. Wolfe, M.L., Debyle, N.V., Winchell, C.S. and McCabe, T.R. 1982. Snowshoe hare cover relationships in northern Utah. Journal of Wildlife Management 46, 662-670. Wolff, J.O. 1978. Food habits of snowshoe hares in interior Alaska. Journal of Wildlife Management 42, 148-1 53. Wolff, J.O. 1980. The role of habitat patchiness in the population dynamics of snowshoe hares. Ecological Monographs 50, 111-130. Wolff, J.O. 1981. Refugia, dispersal and geographic variation in snowshoe hare cycles. In Proceedings World Lagomorph Conference 1979 (Eds. K. Myers and C.D. Mclnnes) pp. 441-449. University of Guelph, Ontario. Ylönen, H., Jedrzejewska, B., Jedrzejewki, W., and Heikkilã, J. 1992. Antipredatory behaviour of Clethrionomys voles: ‘David and Goliath’ arms race. Annales Zoologici Fennici 29, 207-216.  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items