UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Summer diet selection by snowshoe hares Seccombe-Hett, Pippa Elizabeth 1999

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

Item Metadata


831-ubc_1999-0605.pdf [ 5.11MB ]
JSON: 831-1.0228816.json
JSON-LD: 831-1.0228816-ld.json
RDF/XML (Pretty): 831-1.0228816-rdf.xml
RDF/JSON: 831-1.0228816-rdf.json
Turtle: 831-1.0228816-turtle.txt
N-Triples: 831-1.0228816-rdf-ntriples.txt
Original Record: 831-1.0228816-source.json
Full Text

Full Text

Summer Diet Selection by Snowshoe Hares by Pippa Elizabeth Seccombe-Hett B.Sc. University of British Columbia, 1996  THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Department of Botany) We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA September 1999 © Pippa Elizabeth Seccombe-Hett, 1999  f  In  presenting  degree freely  at  this  the  thesis  in  partial  fulfilment  University  of  British  Columbia, I agree  available for  copying  of  department publication  this or of  reference  thesis by  this  for  his thesis  and study. scholarly  or for  her  Department The University of British C o l u m b i a Vancouver, Canada  DE-6  (2/88)  I further  purposes  the  requirements that the  agree that  may  representatives.  financial  permission.  of  It  gain shall not  Library  an  granted  by  is  understood  advanced  shall make  permission for  be  be  for  the that  allowed without  it  extensive  head  of  my  copying  or  my  written  ABSTRACT  The primary objective of this study was to identify the plant species included in the diets of male snowshoe hares (Lepus americanus) during the summer months. Snowshoe hare selection for and against each plant species was assessed by comparing the relative use and availability of each plant species. The secondary objective was to examine several hypotheses of what influenced the selection of those species individually and within an integrated modeling framework. Three hypotheses were examined, i) Hares might select plants with high nutritional content (of energy or protein), ii) Hares might select plants to avoid or minimize deleterious plant secondary compounds (i.e. tannins and alkaloids), iii) Hares might select plants that minimize their risk of predation.  The hare diet during the summer consists offivemain plant species: Betula Festuca altaica, Lupinus arcticus, Salix  glandulosa,  spp. and Shepherdia canadensis, although a  number of other species were occasionally included. Nutritionally, hares select for plant species with high protein content and avoid toxic effectsfromsecondary compounds by ingesting a diverse diet. Hares are not consistently found associated with any particular vegetation types, although they prefer habitats with both a dense understory and an abundance of preferred food species.  Although some support was generated for all of the hypotheses of diet selection, no single hypothesis explained all of the observed patterns of diet selection. A linear programming foraging model combined with optimization techniques was thus used to examine the  interactions between the variables. Overall, the model is successful in integrating the conflicting hypotheses of hare foraging. It appears that hares change their diet selection response to conflicting goals such as reproductive condition and risk of predation. The model suggests that the interaction between plant protein and chemical defense compounds are the primary determinants of hare diet.  Table of Contents Abstract  11  Table of Contents  l v  List of Tables List of Figures Acknowledgements  v  w  n  V I U  Chapter 1  1  Chapter 2  4  Methods  8  Results  2  Discussion  ^1  Chapter 3  7  Methods  4  8  Discussion  The diet selection model  1  7  Results  Chapter 4  1  0  ^7 1  1  0  2  0  4  Results  1  Discussion  1^9  Chapter 5  1  2  4  1  7  References  iv  List of Tables Table 2.1  The relative proportion of species available in the hare enclosures  22  Table 2.2  Selection ratios for hares at time 1  29  Table 2.3  Comparison of species selection ratios at timet  32  Table 2.4  Selection ratios for hares at time 2  33  Table 2.5  Comparison of species selection ratios at time 2  34  Table 2.6  Mean crude protein content of plant tissues  37  Table 2.7  Comparison of mean consumed versus mean available plant nutritional properties  43  Table 2.8  Mean gross caloric content of plant tissues  Table 2.9  Secondary compounds identified in plant tissues  58  Table 2.10  ANOVA results for protein, energy, secondary compounds in diet selection Comparison of resultsfromstudies of summer diet selection  59  Table 2.11  46  62  Table 3.1  Mean habitat characteristics on three study sites  Table 3.2  Results from correlation of fecal pellet number against habitat variables... 82  Table 3.3  Summary of composite variable analysis  86  Table 3.4  Correlation between composite variables  87  Table 3.5  Mean vegetation characteristics within each cluster on each grid  90  Table 3.6  Resultsfromad hoc logistic regression to predict cluster membership  94  Table 3.7  Significant results from logistic regression on fecal pellet transects  Table 3.8  Correlations between tree cover, Betula glandulosa and Festuca  Table 4.1  81  95  altaica  96.  Parameter definitions for model  107  Table 4.2  Mean plant nutritional characters included in the model  109  Table 4.3  Time related parameters included in the model  114  Table 4.4  Comparison of predicted and observed diets using Belovsky's original model  123  Table 4.5  Summary of constraint equations violated by observed hare diets  136  Table 4.6  Results from sensitivity analysis of model constraint equations  137  Table 4.7  Summary of sensitivity analysis on time costs  138  Table 4.8  Assumptions of the diet selection model  146  vi  List of Figures Figure 2.1  Schematic diagram of sampling design within hare enclosure  12  Figure 2.2  Comparison of proportion of plant species consumed versus available within hare enclosures time 1 25  Figure 2.3  Comparison of proportion of plant species consumed versus available within hare enclosures time 2  27  Figure 2.4  Mean crude protein content of leaf tissue  Figure 2.5  Standardized selection ratios as a function of protein content time 1  39  Figure 2.6 Figure 2.7  Mean energy content of plant leaf tissue Standardized selection ratios as a function of energy content time 1  44 48  Figure 2.8  Correlation of protein and energy content of leaf tissues at both times... .50  Figure2.9  Meanfibrecontent of leaf tissue  Figure2.10  Mean water content of leaf tissues  Figure3.1  Correlations of fecal pellet number and distance to cover for three study sites 83  Figure3.2  Regression of composite food variables against distance to cover for each study site 88  Figure 4.1  Fecal index of hare daily activity  Ill  Figure 4.2  Optimization results from new diet selection model at time 1  125  Figure 4.3  Optimization resultsfromnew diet selection model at time 2  Figure 4.4  52 55  128  Optimization resultsfromnew model at time 2. Adjustments applied to secondary compounds of Betula glandulosa and Shepherdia canadensis  Figure 4.5  36  Comparison of time minimized, energy maximized and simultaneous maximization goals with the observed diet  131  134  vii  Acknowledgements This work was funded by NSTP, a graduate fellowshipfromNSERC and an NSERC operating grant for Dr. Turkington. I would like to thank those who helped me in the field: Patrick Carrier, Sarah Davidson, Leo Frid, Elvira Harms, Chris Lortie, Liz Gillis, Jes Logher, and Chris Wulff but most particularly Katie Breen and Jaroslav Welz who had to suffer me daily. This work could not have been completed without the lab support was provided by Tony Sinclair, Gilles Galzi and Elaine Humphrey. Summerfieldsupport camefromthe Arctic Institute of North America, particularly Jan, Sian and Andy Williams. A special thank you goes to Karen Hodges was supportive and helpful in directing my ideas in the early stages. An enormous thank you to Clive Welham who patiently convinced me that modeling and animal behaviour was not so bad; without his guidance this thesis would not have taken the same direction. A special thank you to Roy who guided me, supported me and patiently dealt with my difficulties in reaching the end of this. Finally I have to thank Matthew Evans who was involved in every step of the way whether he wanted to or not.  viii  CHAPTER 1  DIET SELECTION IN THE SNOWSHOE HARE CYCLE  INTRODUCTION Populations of snowshoe hares (Lepus americanus) fluctuate in numbers with population peaks occurring at periods of 8 to 11 years throughout most of their range in the northern boreal forest (Keith 1963, Krebs etal. 1986). Many long-term studies have investigated this periodicity in hare numbers (e.g. Keith and Windberg 1978, Krebs et al. 1995) and several hypotheses have been generated concerning the causes of these fluctuations. The two main hypotheses for this cause are predation, and over-winter food shortage (Krebs et al. 1995). As a result, research on the hare cycle has concentrated on investigating the interactions between the hares, their predators and winter food supply and consequently there has been relatively little research effort focused on the summer food supply. In this thesis I will examine summer diet selection by snowshoe hares and attempt to elucidate how plant qualities such as spatial distribution, nutritional content and defensive compounds affect hare food choice. The cycle in hare population numbers is more or less synchronous throughout most of their range (Sinclair et al. 1993 and Sinclair and Goslinel997) and when the hare population reaches peak numbers it is the dominant herbivore in the boreal forest system (Boutin etal. 1995). Changes in hare population numbers effect the whole boreal ecosystem. Concurrent changes (with a one-year lag) in population numbers have been documented in the predators of hares, (Keith et al. 1977, Keith 1990, Boutin et al. 1995),  1  both avian (e.g. great horned owls and goshawks) and mammalian (e.g. lynx and coyotes). Woody plant species, which comprise the majority of the winter diet of the hares, suffer heavy browsing in peak population years (Pease et al. 1979, Smith et al. 1988, Keith 1990) and subsequently undergo a period of regrowth during the time of low hare numbers. Many of these plant species contain chemicals which act as feeding deterrents (Bryant and Kuropat 1980, Sinclair and Smith 1984, Bryant et al. 1992) and the concentrations of these compounds change as plants recover from incidents of heavy browsing (Bryant 1981, Fox and Bryant 1984).  Three main hypotheses have been examined with respect to snowshoe hare winter diet. First, hares might select for high nutritional content (of energy, protein, sodium, calcium, phosphorous) (e.g. Miller 1968, Sinclair et al. 1982, Belovsky 1984, Sinclair et al 1988. Rodgers and Sinclair 1997). Second, hares might select diets to avoid or minimize ingesting deleterious plant chemical defenses (Bryant and Kuropat 1980, Sinclair and Smith 1984, Bryant etal. 1992, Schmitz etal. 1992). Third, hares might forage in ways which minimize their risk of predation, thus the spatial distribution of forage species is important (Wolff 1980, Hik 1995, Hodges 1998). While these have all yielded useful information, none fully explain the patterns that have been identified in hare diet selection, thus foraging models have been constructed to examine the interactions of the hypotheses (Belovsky 1984, Schmitz et al. 1992).  While most of the preceding hypotheses have been examined with respect to winter food supply, only two studies have investigated the interactions between snowshoe hares and summer vegetation. Grisley (1990) determined dietary preferences of hares in the  2  summer using cafeteria trials with caged animals, and Wolff (1978) collected stomachs from wild hares. The resulting data sets are lists and proportions of consumed species that make no attempt to analyze diet selection relative to species availability, or to investigate what factors influence selection. The primary objective of this study was to describe the summer diet of snowshoe hares in the field. The secondary objectives were to examine each of the three hypotheses used to analyze winter diet and test whether they determine diet selection of snowshoe hares during the summer. In chapter 2,1 will describe the diet composition of the hares in the summer and examine how both nutritional content and secondary compounds of plant species influence hare selection. In chapter 3,1 will examine spatial distributions of hares and examine whether hares are spatially associating themselves with particular habitats or food resources. In chapter 4,1 will construct a foraging model and attempt to predict hare diets using an optimal foraging approach.  3  CHAPTER 2 SUMMER DIET SELECTION OF SNOWSHOE HARES: A TEST OF NUTRITIONAL HYPOTHESES  INTRODUCTION The study of resource selection can reveal how organisms meet their fundamental survival requirements. When resources are used disproportionately to their availability, animals are said to be selective (Manly et al. 1993). Diet selection is the process by which an animal chooses food resources to consume and how much to consume (Johnson 1980, Hall et al. 1997). Resource selection operates on many scales, from the geographic range of a species all the way to particular food items within a particular habitat type, and the criteria for selection may be different at each level (Johnson 1980).  Attempts to understand and identify the processes that determine diet selection in mammalian herbivores have typically presented field biologists with many difficulties. A large number of factors need to be examined in each investigation. Considerations of plant quality, availability, abundance and palatability must be addressed both for a number of plant species and plant parts (i.e. leaves, stems and flowers), all of which change seasonally.  4  One common approach to identify the plant species consumed by herbivores is a cafeteria trial, where animals are held in cages and simultaneously offered a number of food types. Preference for individual food items is assessed by the relative amounts of each plant species consumed. This type of experiment has received much criticism and is thought to bear little relation to food choice in the field (Norbury and Sanson 1992). Thus, the results from Grisley (1990) which showed that hares preferred Equisetum spp. over all other food items presented to them in a cafeteria trial, may tell us little about what hares select in the field. Wolff (1978) collected stomachs from hares in the field, yet neglected to measure plant availability. Because diet selection can only be estimated when measures of both use and availability of food resources have been estimated, Wolff s (1978) results provide only a list of plant species consumed by hares.  Thus, the main objective of this chapter is to identify the plant species that snowshoe hares include in their diet during the summer, determine which species are being selected, and examine how the diet composition changes throughout the season. Food - plant selection will be analyzed by measuring both use and abundance of all plant species.  Nutritional Selection Few studies of herbivore diets have identified clear and consistent trends in diet selection which has led to the generation of several hypotheses to account for the number and diversity of species included in the diet. It has been demonstrated that generalist herbivores grow better on a mixed diet (Freeland et al. 1985, Bernays et al. 1994). One commonly used explanation to account for this increased growth is that dietary mixing  5  allows herbivores to dilute the deleterious effects of any one individual toxin (Freeland and Janzen 1974, Jung and Baztli 1981, Lindroth etal. 1986, Freeland 1992, Bernays et al. 1994). An alternative explanation for mixed diets is that different plant species are selected on the basis that they fulfil different nutritive requirements (Belovsky 1978, Belovsky 1984). Thus mixing the species ingested allows herbivores to attain a balance of nutrients in their diet (Pulliam 1975, Westoby 1978, Clark 1982).  Numerous studies have investigated nutrient levels in plants and their selection by hares, although the majority of these have concentrated on shrub selection in the winter (Pease etal. 1979Pehrsone?a/. 1981, Sinclair  al. 1988, Rodgers and Sinclair 1997). Two  hypotheses have been repeatedly examined to explain the physiological aspects of food choice; selection for plant items based on nutritional content (i.e. energy or protein), or avoidance of plant chemical defenses (Bryant and Kuropat 1980, Sinclair et al. 1982, Sinclair and Smith 1984, Sinclair et al. 1988, Schmitz et al. 1998, Rodgers and Sinclair 1997, Hodges 1998).  Identifying physiological requirements of hares in the summer is important for analyzing diet selection. Whittaker and Thomas (1983) identified that snowshoe hares maintain low metabolic reserves of both fat and protein which led them to suggest that hares must adopt a foraging strategy that maintains energy and nitrogen on a very short-term basis. Studies on the snowshoe hare have shown that in the winter they select twigs that are high in protein content, and low in plant secondary compounds. However, neither is selected exclusively (Sinclair and Smith 1984, Schmitz et al. 1992). Mountain hares  6  (Lepus timidus) have been shown to select food items with high protein and phosphorous content (Miller 1968, Lindloff et al. 1974). Preferred foods have higher apparent digestibilities of dry matter and energy than nonpreferred foods (Rodgers and Sinclair 1997). Belovsky (1984) reported results that snowshoe hares in Michigan forage as energy maximizers and select for plants with high levels of sodium, a mineral he hypothesized to be limiting. Lindloff etal. (1978) identified that female mountain hares select plant species with high concentrations of calcium, particularly during the breeding season. Bryant etal. (1992) suggest that plant defense compounds, not plant nutritional characteristics are the main determinant of winter food selection, or avoidance, by hares.  The large variety in summer browse species available to hares, and the relatively low levels of secondary defense chemicals in them suggests that secondary compounds are of less importance in summer diet selection. However, the large number of plant species with different growth strategies (i.e. both evergreen and deciduous shrubs, forbs, and grasses) in terms of both defense compounds and nutritional allocation, should help identify trends in selection simply by the species hares exclude. The specific hypotheses that I wish to test are:  1) Preferred dietary items contain higher concentrations of protein and energy and lower fibre content relative to concentrations in less preferred species. 2) Preferred food items change seasonally and the changes are positively correlated with changes in the concentrations of protein and energy in the plant tissues. 3) Hares select food items with low concentrations of plant defense chemicals.  7  METHODS Study Area Location This research was conducted near Kluane Lake, in the southwest Yukon Territory (61°N, 138° 30'W). The area was the site of the ten-year Kluane Boreal Forest Ecosystem Project (KBFEP) and is described in Douglas (1974) and Krebs et al. (1986). The study site is located within the Shakwak Trench, flanked to the southwest by the Kluane mountain range and to the northeast by the Ruby ranges. The climate is predominantly cold continental and, because the study area is in the rain shadow of the St Elias Mountains, it is very dry. Mean monthly precipitation at the nearest weather station, in Burwash Landing (61° 22 N 139° 03W), 50 kilometers to the northwest, is 12.6mm in May, 48.4mm in June, 65.5mm in July and 42.3mm in August. The mean daily temperatures are 5.2°C in May, 10.3°C in June, 12.5°C in July and 10.6°C in August (Environment Canada 1998).  Vegetation The vegetation is dominated by a white spruce (Picea glauca Voss) forest, with pockets of aspen (Populus tremuloides Michx.) and poplar (Populus balsamifera L.). The major shrubs and woody species in open areas and the forest understory are willow (Salix glauca L. and Salix alaxensis Anderss.), birch (Betula glandulosa Michx.), soapberry (Shepherdia canadensis (Potentillafruticosa  (L.) Nutt.) and occasional patches of both shrubby-cinquefoil  L.) and rose (Rosa acicularis Lindl.). The understory is dominated  by a mixture of prostrate shrubs (Arctostaphylos uva-ursi (L.) Spreng., Arctostaphylos  8  rubra (Rehd. & Wilson) Fern, and Linnaea borealis L.), grasses (Festuca altaica Torr.), forbs {Achillea millefolium L . ssp. borealis (Bong), Anemoneparviflora  Michx.,  Delphinium glaucum S.Wats., Epilobium angustifolium L . , Gentianapropinqua Lupinus arcticus Lindl., Mertensiapaniculata multiradiata  Plant  Richards,  (Ait.) G. Don var. paniculata, Solidago  Ait. and Senecio lugens Rich.) and a variety of both mosses and lichens.  Phenology  The growing season is short, approximately 12 weeks long, snow melt is generally complete by mid-May. Leaves begin to emerge on both the deciduous shrubs and herbaceous plants by the end of May and plants then invest in growth until peak biomass is reached by mid-July. The majority of the herbaceous species flower between mid-June and mid-July. Plant senescence starts anytime from early to mid-August  Herbivores The snowshoe hare {Lepus americanus) is the most abundant herbivore in the system (Krebs et al. 1995). Their population numbers are monitored in the area and this research was conducted during the increase phase of the hare cycle in 1997. Other herbivores include red squirrels {Tamiasciurus hudsonicus), and various species of voles {Clethrionomys  ground squirrels {Spermophilus  parryii)  rutilus and Microtus spp.) (Boutin et al.  1995).  9  Experimental Design Study Site  Twelve study sites were chosen between one and three kilometers to the south of the Alaska Highway at Boutelier Summit (Mile 1050 Alaska Highway). Sites were chosen to include a large variation in habitat types and vegetation. Four sites were located in each of three habitat types: areas dominated by shrubs and both open and closed spruce forests. Each site contained a variety of herbaceous plant species. No attempt was made to standardize the composition of the vegetation between sites because habitat characteristics were incorporated into the data collection. At each site a hare enclosure (10m x 10m) was constructed using chicken wire fencing with 2.5 cm mesh and 1.5m high. The fence was secured to the ground to prevent hares escaping. Within each enclosure both vegetation availability and habitat descriptions were recorded. One male hare was added to each enclosure and left for 24 hours at which time their stomachs were collected and the contents removed. All 12 enclosures were used synchronously at each of two separate time periods (early June and late July).  Hare  Trapping  The hares were trapped (Tomahawk Live Trap Co., Tomahawk WI) the night before they were put into the enclosure. Hare traps were set at approximately 11p.m., baited with alfalfa and apple, and then checked at 6 a.m. Only male hares were used for the experiments. Hares were ear-tagged in therightear using Monel #3 tags and the following measurements were made to obtain an index of the condition of the animals: i) Weight ii) Length of right hind foot  10  iii) Reproductive condition (abdominal or scrotal).  Data collection in enclosures  Within each enclosure, the following measurements were made: Biomass Clips- To obtain estimates of the vegetation composition within the enclosures, biomass clips were done in 12 quadrats (each lm x 0.1m) immediately before introducing the hares. The quadrats were systematically placed 1.5 meters apart along three transects which traversed the enclosure (Fig. 2.1). In each of these quadrats all of the herbaceous plants were clipped at ground level and sorted by species into bags. Leaves were collected from the shrubs which hung over the quadrat to a height of 1 meter (the maximum height that hares are able to reach). For prostrate dwarf shrubs such as A. uvaursi and L. borealis, the plants were clipped along the edges of the quadrat because it is difficult to determine where these plants are rooted. All samples were placed in the drying oven at 60°C for four days and weighed.  Stomach Contents- Hares were placed in the enclosures for approximately 24 hours. They were released in the enclosures immediately after trapping (by 8a.m.) and then left alone until the following morning at which time they were sacrificed, their stomachs collected, the contents removed, placed in plastic vials and frozen.  Stomach Content Analysis The stomach contents were analyzed following Norbury's method (1988) of systematic sampling that includes correction factors.  11  Figure 2.1: Schematic diagram of sampling design within hare enclosures. Large rectangles delineate locations of clip quadrats. Squares show location of fecal pellet count quadrats and dots within them identify the placement of the tagged plants.  12  13  Reference  Collection  Because hares masticate their food into tiny pieces, the stomach contents had to be identified microhistologically. Thus, a reference collection of plant epidermal structures was prepared. Leaf samples were collected of all species present in the enclosures and used to make reference slides. Clear nail polish was applied to fresh plant tissues and left to dry. The nail polish adhered to the epidermal layer allowing it to be peeled from the leaf and mounted between two slides for reference. The dry nail polish preserved an epidermal print of the leaf. These reference slides were then photographed and catalogued.  Correction  Factors  The proportion of identifiable fragments of epidermal tissue has been shown to vary substantially between species (Griffiths and Barker 1966, Westoby et al. 1976, Norbury 1988) using microhistological techniques. Therefore, correction factors are used to correct for this source of error. Thus, for each species that was included in the diet I determined the ratio of identifiable to unidentifiable fragments. This ratio was used as a correction factor to obtain an accurate estimate of the dry weight of each species in the stomach contents. When the number of identifiable fragments in a stomach sample is multiplied by this correction factor, the dry weight of each plant species in the stomach can be estimated.  A 1.5g sample of dried plant matter of each species was blended in 150ml of distilled water for five minutes, to break the plant tissue to a size that can be viewed under the  14  microscope. Bleach was then added to the solution to a concentration of 10%. The sample was stirred and left to bleach for two hours. After this time the sample was sieved (mesh size 118um) and rinsed thoroughly, and placed in a specimen jar in 100ml of water. Flow through water was checked for identifiable fragments. Each of the 34 species was analyzed separately under the microscope to determine the ratio of identifiable to unidentifiable tissue in the sample. A total of 308 sample points per species were viewed to estimate each correction factor. The slides were sampled systematically with the sample points 1-field-of view apart. At each point it was recorded whether there was afragmentat the cross hairs, and whether it was identifiable or not. The spatial layout of the sample points was recorded such that it can be verified whether the use of systematic sampling is valid (i.e. that there is no periodic variation in the distribution of the sample that coincides with the sampling distance). A second reference collection was preparedfromthe solution used to determine the correction factors. Semi-permanent slides were made by placing a few drops of each solution under a cover slip on a slide and then sealing the edges with nail polish to prevent the slide from drying. Stomach Samples  The stomach samples were prepared and sampled in the same way as the correction factors. The slides were sampled systematically with the sample points 1-field-of view apart. At each point it was recorded whether there was afragmentat the cross hairs, and whether it was identifiable or not. Positive identification of a plant camefrommatching the tissues with the reference epidermal prints (photographs and slides). The number of  15  sample points required to estimate the proportion of each species in a sample, was calculated according to the methods in Zar (1984, p 380). Point frequencies of each plant species in the stomach contents were adjusted by their respective correction factors to estimate a converted point frequency with the following equation: n  Converted point frequency = f;/ fHDi) / £ [ f;/*f(IDi)] i =  1  where fi = point frequency of epidermis for species i in the mixture,ftTDi)= point frequency of identifiable epidermis for species i on the reference slide, and n= total number of species in the mixture (equation from Norbury 1988). The converted point frequencies were then multiplied by the dry weight of each species to estimate the percent dry weight of each species in the stomach sample.  Plant Analysis Field  Collection  Fifteen samples lOg each (dry weight) were collected for each of fourteen species in the area, estimated to be included in the hare's diet. These samples were collected close to the enclosures and on both occasions when the hares were put in the enclosures, and at a third time at the end of the growing season, in mid-August.  Protein and Energy Content -  16  Five of the 15 samples of each species were collected at each of three times, and air dried for two weeks so as not to cause heat damage to the proteins. The samples were subsequently ground in a Wiley mill through a 40 mesh. The crude protein content was determined using the Dumas method (Tate 1994). Gross caloric content was measured using a bomb calorimeter. Fibre and Water Content  Water content was determined for the ten remaining samples of each species at each of three times by obtaining wet weights immediately after collection and dry weights after the samples were in the drying oven for four days at 60°C. Water content was calculated as: = 1 - (sample dry weight/sample wet weight) * 100% These samples were subsequently prepared for fibre content analysis by being ground through a Wiley mill mesh size 40. Fiber content was measured on four samples of each species at each time, with an additional four samples processed at each of the two times the hares were placed in the enclosures for the following species: L. arcticus, F. altaica, Shepherdia canadensis, Salix  spp. and B. glandulosa. The acid-pepsin digestibility  technique used follows the procedure outlined by Larter (1992). A 0.2g sample was inoculated with 20ml solution of 2.0g pepsin in IL of 0.1 M HCL. The samples were placed in a heated water bath at 37°C for 48 hours and swirled to aid the mixing of the inoculant and the plant tissue at time 0, and subsequently at 1,6 and 24 hours after the digestion commenced. After 48 hours the samples were vacuum-filtered ontofilterpaper of known weight, then dried at 90°C for 48 hours and reweighed.  17  Fibre content was calculated as: = (sample mass after digest (g) / initial sample mass (g)) *100%  Statistical Analysis Stomach Contents Stomach contents were analyzed using selection ratios (w), following Manly etal. (1993 pp.66-67). Selection ratios in their simplest form are calculated as the ratio of used resource units (o), to available resource units (%) which in this case is the ratio of plant species in the stomach to the plant species abundance in the enclosures. w = o / 7C However, because the estimates of use and availability were measured separately for each hare in each enclosure, a modified calculation is used for the selection of each category (/') by each animal (/'): Wij=  Uij/(7tjj  *U+j)  where uy is the number of category / resource units used by animal j, u+j is the number of category / resource units used by all sampled animals and TTy is the proportion of the resources available to animal j that are in category /'. Results are combinedfromseparate animals to estimate the selection ratio for the population of animals for each i resource category with the equation:  Wj = Ui+ / £  Ttij . U+j  18  where Ui+ is the number of category i resource units used by all sampled animals (i.e. total consumed) and £ Tty * u +j is the expected consumed by the null hypothesis.  Variance for each estimate is calculated as follows: Var(w ) = {I(u .(w *7t u )) /(n-l)} { 1/(n(7tijU ) )} 2  i  ij  i  ij  +j  2  +j  The larger the w value, the greater the selection of that resource relative to what was available for selection. Significant selection ratios are identified by w-values greater than 1 and confidence intervals do not include 1. Negative selection is identified by values less than 1 with confidence intervals below and not including 1. Selection ratios are presented in the standardized format (B): i  Bi  = Wi  fE wi i=l  where Bj is the standardized selection ratio, Wi is selection ratio for species i  The results from all animals were pooled for each time to calculate standard errors and Bonferroni confidence intervals. The Bonferroni confidence intervals were used to allow multiple comparisons without increasing the probability of error. The 'available' resource units include only the species that were identified in the stomach contents at one of two times, which includes only a subset of the available biomass estimated in the biomass clip quadrats.  19  Differences between selection ratios at, and between, each time were calculated along with standard errors and Bonferroni confidence intervals to compare changes in selection ratios through the season.  Nutrient Selection  Mean nutritional content values were compared over time to identify significant differences between species and times. The mean nutritional values were then compared to the selection ratios as mean values for each time. Diet composition was converted to nutritional units. The proportion of each plant species in the diet was multiplied by its respective nutritional value (e.g. protein, energy,fibreor water) at each sampling time. Nutritional availability was calculated by multiplying the (  availability of each species by its nutritional value. This allowed the calculation of mean nutritional values in the diet and mean availability within the enclosures at each sampling time. Thus comparisons could be made for each nutritional characteristic included in this study.  An analysis of variance was used to compare the effects of the nutritional variables on selection. Protein, digestible energy and secondary compounds were included as the independent variables and the standardized selection ratio was the dependent variable.  20  RESULTS Diet Composition Of the 38 plant species recorded in the enclosures only 12 were identified in the stomach contents of the hares (Tables 2. la, b). At the first sampling time L. arcticus, F. altaica, Shepherdia canadensis and Salix spp. were the main components identified in the stomach contents each individually comprising 34, 26, 20 and 12% of the diet, respectively (Table 2.1; Fig. 2.2). In the second sampling time, L. arcticus, F. altaica, Salix spp. remained substantial components of the stomach contents. However, Shepherdia canadensis decreased to only 6% and B. glandulosa increased to comprise 16% of the diet (Table 2.1; Fig. 2.2). The breadth of the diet remained constant with a total of 9 species identified in 10 stomachs at the first sampling time and 8 identified in 12 stomachs collected at time 2. Species occur in the diet at frequencies that differ from those of availability in the enclosures (Figs. 2.2 and 2.3). Hares are demonstrating selection at both sampling times. The perception of whether a species is identified as preferred or avoided depends on the relative abundance of species included in the analysis, thus, for the purposes of this investigation, all results are calculated only with species that were included in the diet from at least one of the two sampling times.  21  Table 2.1a: Average proportion (± 1 SE) of each plant species available within the enclosures at each of two sampling times. Means are presentedfroma sample of 10 enclosures at time 1 and 12 enclosures at time 2.  Plant Species Achillea millefolium Anemone parviflora Antennaria spp. Arctostaphylos rubra Arctostaphylos uva-ursi Arnica spp. Artemesia spp. Aster spp. Betula spp. Carex spp. Cornus canadensis Delphinium glaucum Empetrum nigrum Epilobium angustifolium Equisetum ssp. Festuca altaica Gentiana spp. Juniper communis Linnaea borealis Lupinus arcticus Mertensia paniculata Moneses uniflora Unidentified grasses Picea glauca Polemonium spp. Pyrola spp. Rosa acicularis Salix spp. Senecio lugens Shepherdia canadensis Solidago multiradiata  Time 1 Mean Available 0.01 0.00 0.00 0.01 0.18 0.00 0.01 0.00 0.03 0.00 0.01 0.00 0.02 0.01 0.04 0.23 0.00 0.00 0.32 0.03 0.00 0.01 0.00 0.03 0.00 0.00 0.00 0.02 0.00 0.01 0.02  SE  Time 2 Mean Available  SE  0.03 0.01 0.02 0.03 0.14 0.00 0.03 0.00 0.06 0.02 0.03 0.01 0.05 0.03 0.07 0.16 0.00 0.01 0.19 0.06 0.02 0.03 0.02 0.06 0.01 0.02 0.02 0.01 0.01 0.03 0.04  0.01 0.00 0.01 0.01 0.11 0.00 0.00 0.01 0.07 0.00 0.01 0.00 0.01 0.03 0.02 0.22 0.00 0.00 0.25 0.07 0.01 0.00 0.00 0.03 0.00 0.00 0.00 0.03 0.04 0.04 0.02  0.00 0.00 0.00 0.01 0.04 0.00 0.00 0.00 0.03 0.00 0.00 0.00 0.01 0.01 0.01 0.04 0.00 0.00 0.06 0.03 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.02 0.02 0.02 0.01  22  Table 2. lb: Average proportion (± 1 SE) composition of the hares diet at the two sampling times, in early June and late July. Means are presented from a sample of 10 stomachs at time 1 and 12 at time 2.  Plant Species Achillea millefolium Anemone parviflora Arctostaphylos rubra Betula glandulosa Cornus canadensis Epilobium angustifolium Festuca altaica Lupinus arcticus Mertensia paniculata Salix spp. Shepherdia canadensis Solidago multiradiata  Time 1 SE 0 3.5 0.7 2.0 0.2 0 26.3 34.4 0 12.1 20.4 0.4  3.7 0.8 1.4 0.2 7.2 7.6 3.9 6.5 0.5  Time 2 0.2 0 0 16.2 0 3.2 19.2 37.7 0 9.1 6.3 0.6  SE 0.2 3.0 1.8 3.9 5.6 2.8 2.1 0.7  The selection ratios for the first sampling period reveal that hares demonstrate significant selection for both L. arcticus and Shepherdia canadensis (Table 2.2a). There is negative selection for F. altaica, Solidago multiradiata, Arctostaphylos rubra, C. canadensis and B. glandulosa  (i.e. each species is included in the diet of the hares far less than what  would be expected by random chance). Anemone parviflora and Salix spp. usage did not differ significantlyfromtheir availabilities. Although, Anemone parviflora has a high forage ratio value of 12.29, its confidence intervals are large and include 1. Thus it is not significantly selected. Anemone  parviflora  was identified in only one sample. Thus the results were reanalyzed without this sample to examine how the selection ratios of the remaining species change (Table 2.2b). Only the value of Salix spp. changesfromneutral to preferred. When comparisons are made between species, the larger the w value, the greater the selection for that species. Shepherdia canadensis, L. arcticus and Salix spp. are all significantly selected for and significantly differentfromthe ratios of all other species (Table 2.3). The remaining species were not significantly differentfromone another. The resultsfromthe second sampling time indicate that hares demonstrate positive selection forL arcticus (w = 3.2) and B. glandulosa (w=1.42) late in the season (Table 2.4). Hares demonstrate negative selection for Shepherdia canadensis, multiradiata, M. paniculata, F. altaica  and Achillea  Solidago  millefolium ssp. borealis.  Both Salix  24  Figure 2.2: Comparison of the proportion (± 95% C.I.)of species available to the hares within the enclosures and the proportion included in the stomach contents from sample collected at time 1. Means are calculated with samples from 10 male hares.  Figure 2.3: Comparison of the proportion ( ± 95% C.I.)of species available to the hares within the enclosures and the proportion included in the stomach contents from sample collected at time 2. Means are calculated with samples froml2 male hares in late July.  uouJodoJd  28  8  a  c £ <2  e  •S c o  > o  (A  •a  x s*  O  rs  D  o<  o  in vO O  o  0\  «/•>  C*»  fi  ©  00  00  (J  •a CU .S c cu o c fa eu c  1  as  r©  o  cs  o PQ  S -SP o .— S fc  vU  j  3 « c  C/3  V"!  m  o  CN  O  o  if) so  rs  CXI  ,cu  t  00  —>  ©  o  fa "O o +•»  m 3 s « CO  «  o  •a  o o .2 O  JS a. O e  oe cu £  s o  o  S "S — o o  o 1/1  —i  CN  CN r*>  2  °  o o  00  00  OS  vo CN  <N  o o  00 9 ©  CN  CN  o o  o  C*»  rs  ©  ©  C/}  o n e  IJ 8  8  «  e CO  CU  CO  CO {_} .—<  *1 I 3 X  g  a *  00  o  © r-i  rs  i« in o  ii  80 }> T3  .—  o  •3  88  C g C  «  o  jd l>  6  CN  AO  o  2  rs  -H  rH  ©  cj\  m  rs  ro ©  <n o  1-H  »H  ©  V5  c*»  vo.  u-  ^ o cd  I H  1? co  "8  <u  cd O co  2  3 ! » cd o 2 >  E2  e o  O  .i  m m ©  CN  «7\  rs  3  CO  O  OH  £ 91  "S  .2 II § eu  co  •j > o « o <s 5 o  eu "3 .2 H o w  a  «  •9  « I .2 I  9 « an  S B  eg ca  a. uCU  mm*  xfi  x: «3  o 60  •3 o Ol  r-~  o. Cd  OJ ^  —  fi  «  o v «8 o a. <-» o o  c »  IT/  ©  35  4>  h  * .2  1/3 V l t- vc «s ao o  vo  0  •a  T t  g  .3  o  M  in  vo ©  9v va  VO  >r>  0  .1 § 1 C  C eS  1 B  4> O .  83 C  c o  05  o ©  r> ©  P o  ©*  VO  f»l  vo  C/2  t  e o  CQ — » " co "O  O  •3  «« .2 .2 Wi  3  CO  O  "o c 0>  8 g CO  +-»  ON  6 C o o •0 4>  co  c o  a,.a  3  u.  «> 8  o  -t_j c  CO  co  o c -o  •a  o  CN  o <*> o  11  c">  0  o  o  o  "* !C rH  S  •<*  ©  © ' © O  0  00  o  ©  ©  H  N  ©\ ©  »  ri  F!  in  o  — ©  oe  ON  «  t- —;  co  §  S  I-S •ed  CN  0  O CO  e  CJ  a. E «5 co cd CO 'o III  cd  O  <N ©'  Wl  -f  «")  0>  ©  oe1 00  •>*  ©'  ©'  <2 °)  m  v>  © ' © ' © '  ©  W  co  0\ »H  ©  00 ©  - S .0)  o 4> " .£3 to 60  "5  |.2 £ a> P o CO  O OH  4) >  i  3  "8  i  o  o  u-i £ CJ e P -5 c a> o o11 a > fd "+-» cd  o  <N T3 cN  3 o  t  4) ,™ co O  1  00  ©  ©  ©'  ©  ©  ©  t^ ri  ©  v©  VO  —;  ir> ©'  spp.  and Epilobium angustifolium are included in the diet in proportion to their  availability, as can be seen by the breadth of the confidence intervals (i.e. 95 % confidence intervals include 1). The only selection ratios that were significantly different from all other species were B. glandulosa, L. arcticus and Salix spp. (Table 2.5). Lupinus arcticus, Salix spp.  and F. altaica all comprise consistent amounts of the average  hare diet at both sampling times (Table 2.1a and compare Figs. 2.2 and 2.3). The percentage of both Shepherdia canadensis and B. glandulosa in the average diet changes significantly between sampling times (Table 2. la). Small amounts of other species were consumed, although none were consistently included in the diet at either sampling time.  31  E s  •8 W  0 J> C "o  4? *  I  3  i  a  e 8  ©  ©  ©  3  If  I  o  ©  o  o  "3  ©  +  +  +  "3  8? C CO  11 18 1§  1.1 ' H co  a.  3  oo <-  ©  o  ©  ©  o  o  ©  ©  o  ©  + . +  +  +  +.+  +  +  *<v5 §  s '3 3 .2  > e3^ CO  .2  (A lb  ^  +  8 >» a. *j  CO  C o  CO  %  4> % "5 ~ •2 c e  to g> o  a. s  -J  .a •8  1  M  eu co —• eu eu . CO CM  O  1  5  C L CO CO CO  .2 <8 C O  3  co  Si o  £  5  "S8. CN s  s  J» 2 S n  CO  « ^  «  C  1 1 CO  8  1 i  60 CO  I  c2 3 6  a I  1  3  +  +  4) O  4> <£  c  C  u c • * .9  2  CO  4»  •8  J2  §  §  m vo r-* ©  OH  ao «r> t- o oo © •—' vr> Ov CN O  CN O  O  £ 8 c  •IS """ C C D  c  •"" * !a  S  O  g  03  o 03  c D  c  8  4> c  'c p o  CO u .—  .2  a  C o  4) CN o o  1  41 o  C/3  o  4> *•* C cN  £ o c J£ o ° §  1 3 v  (NJ  cd  O  cd  4>  55  8  •o -.5 e u  0  8 8S 4> "O  .2 o  4>  4) 4>  2  o •£  O. c  co  w  co  C  '55 «  03  «2 — .2 §  > •£| •3 4>  _  m  o © © o ©  .—i O t~~ CN VO Os vo t"~ O CN O O O 1 O O °. **i o o o o © o o o  «vi  H  *H  o o  H  08  *0 «  mO  —' o  ">  " o  \fl  M  ' t  o o ^ o » ^ ^ °. ^  C  8  "EL *  «  O o o  00 O  —  3  as  C  ce)  CN  fS  O  o  8 8 *o J;  C  O  vo t— CN  r> m 00 cN 0O Ov »Q O CN O © — o o © © O O O ©  m ©  Ou £ co  8  rm •rt O  © C/i  pa  S_  CN t -  td  4>  0  C  O  -a  t« —  3  E  * o W CJ  t-^  Ov O  oo  VO  vo  O  00 o CN M  ^  VO 00 C">  i/J  w  M  p  001 t >  ^-  ^'  T3  4)  <j  m U o oo o vo oo »So — X- N -. — ^«*> o^ o o o oo o —• 2 S ° i n m  r»- o  N  vo CN  ©  73 3 g..~ > cd  p  CO  a  £  2 Id © . -S •*-» c c p O  E <B e eS «o  .S  4)  P  4) CN  4>  O  co  4>  '5 H  S  4>  CO*  3  • s e c  ~_ >  O.  C/3  cd  4>  V  C O  2S  CO  _3  co C  'I  4) So  3  S .a  "2 ^  S ed  C .o  E © — 2 E CO  3  V  O  CO CJ  iP s  3 +^  Ou  CO  ii  W ti-  2  c e3  T3 - a cd E 73 u o. C ed "So o to st  "co  ed  4) ;=  £  ed  ed  s  c l.is S  . .2 E  . £ 4> o- "g p  S. c  S-J  ef o> i C O C/3 C/J C/3  "3  51  ed  J3  I  8 o  ©  +  o  +  o  +  +  +  o  +  +  +  +  +  +  Xi  "8 ^ •j§'—  IH 2  S  3  o  w  2 o  CO efe  .n * CO  4}  c  8 je  J-l to  o  1  H  i2 C 3 £ 8 o  ••3 §  co 5.2 C  i n -2 . o  ©  u  •C  Q .  u  ex " 5 co c cy  ©  o  +  o  +  +  +  +  +  ©  +  +  +  +  +  +  +  +  +  +  +  +  +  +  +  '+  I  2 o  E  IB _o '5. W  co <*2  .2 ^  +  JS C/l  3  ||  '  u  CQ  8 £ £ .2P .2  I  c ™ 2 3  8  cd  +  +  +  +  +  +  •+•.+  CO  3  C  '5.  +  3  CO  co '(j O CL E  c £  3  ScO  '5-8  3  ret co O  3  CO  i |  a  'a  a> o.  Ui  CO CO  3 00  £  CO  •8 'a,  03  W  £ 3  3  i  o  I"o on  s  U  <3  o.  a.  _o  s  a  !  1)  CO  o £  ca  a.  34  Nutritional Elements Crude Protein  Protein content was highest in leaf tissues of most species early in the season (Fig. 2.4). All species show a trend of decreasing protein content over the summer except for C. canadensis, which demonstrated a slight increase in mid-July. Lupinus arcticus contained a significantly higher concentration of crude protein than any other species during the first two sampling periods (35% and 21% respectively; Table 2.6). At both sampling times, the species that were significantly selected were the species with the highest mean crude protein content (Figs. 2.5a and b). However, the average crude protein content of the diet is not significantly different from the average crude protein content in the environment (Table 2.7). Energy  Energy content showed the same general trend as protein content (Fig. 2.6, Table 2.8). The highest energy content of leaf tissue was early in the season for all species except C. canadensis, which was the only evergreen included in this analysis. This species showed an increase at the second sampling time. Leaf material of B. glandulosa consistently had the highest energy content of 5200 and 5000 kcal/g at times 1 and 2 respectively, which is significantly higher than all other species sampled at all times. There is a clear relationship between the highly selected species and energy content at the first sampling time. Shepherdia canadensis, Salix spp. and L. arcticus are all selected at time 1 (when Anemone parviflora is excluded from the analysis), and contain high levels of energy, lower only than B. glandulosa, which was selected against. When both  35  Figure 2.4: Mean crude protein content (± 95% C.I., n = 4) of leaf tissues collected at three sampling times (early June, late July and mid-August).  36  »-  CM <*>  0 0 0)  E E E  P P r• • E3  luaiuoo uiaioid epoio }uaaJ3d  Table 2.6: Mean (n=4) crude protein content (% crude protein/ gram dry weight) of leaf tissues collected at three separate sampling times (early June, late July and mid-August); time 1 and time 2 correspond to the times stomachs were collected. Species  Time 1 SE  Time 2  SE  Time 3 SE  Achillea millefolium  24.6  0.7  16.2  0.5  16.5  0.6  Anemone parviflora  19.7  0.2  14.2  0.5  14.0  0.3  Arctostaphylos rubra  20.7  1.1  12.8  0.9  11.4  0.6  Betula glandulosa  27.8  2.1  20.5  0.4  17.8  0.7  Cornus canadensis  9.6  0.6  12.4  0.5  10.7  0.7  Delphinium glaucum  25.6  1.3  14.1  0.6  10.7  1.4  Epilobium angustifolium  27.9  1.6  17.2  0.8  15.1  1.6  Festuca altaica  20.1  2.3  11.1  0.2  12.3  0.7  Lupinus arcticus  34.6  1.0  21.1  1.4  15.8  1.2  Mertensia paniculata  25.2  1.2  14.9  0.4  13.4  1.5  Salix spp.  30.9  2.0  15.6  0.8  15.5  0.8  Senecio lugens  22.4  1.6  13.7  0.7  12.1  0.8  Shepherdia canadensis  33.2  1.7  19.7  0.3  24.5  1.1  Solidago multiradiata  28.2  1.3  17.7  0.3  16.9  0.4  Carex spp.  8.1  0.2  Equisetum spp.  7.2  0.8  38  Figure 2.5a: Comparison of standardized selection ratios (B) against crude protein content of leaf tissues of individual species at the first sampling time in early June. Values for A. parviflora have not been used in the analysis.  co O CO  c cu  in co  (0  CO  c to o CO  •  E  <5  * o co  Q .  a.  3  co X  E  co  ~4>  3 TJ  C . « CD  cd  co ;> u  "£  in  CO • CL  S  CO  = to  o co to  1 0O TO 3 mm  2 •  o u c  3  £  m  5e  o  u  CM CP  Cl-  IL.'  in  in  —i—  in co  co  d  to  CM d  CM d  in  io o  d  (a) oaea uonoaias pez!PJepue»S  40  Figure 2.5b: Comparison of standardized selection ratios (B) against crude protein content of leaf tissues of individual species at the second sampling time, in late July.  41  CN CN  (0  3  a CD  ••*»•  O  (9  3  '5 c«9  2 • co CD  TJ co  o  CM CO  •  •o S  s 3  E E  CO  3 O  «  E  ui  o "5  c eg  •1  a. a. is 4> x  § o e 5g co  s  a  o**  1  «  S  3  O  CO  a.  c  •  CO  a  S o  a.  it  ca a. 2  CO  55  c  CO T3 CO  c eg o  CO  o  O  CO  — i —  in CO  in CM  CM ©  m o  (e) <W3J uoaoaps pazipjepueiS  42  Table 2.7: Average dietary nutritional proportions compared to available nutritional proportions. Values are calculated at time 1 averaged over 10 replicates for each estimate and at the second sampling time 12 replicates were used.  Energy content (Kcal/g)  Fibre Content (%) Protein Content (%) Water content (%)  Diet  SE  Available  SE  Significance  Time 1  4720  28  4479  42  p< 0.005  Time 2  4218  146  4396  115  -  Timet  46.5  1.6  36.5  0.5  p< 0.001  Time 2  44.0  1.9  51.8  1.7  p < 0.005  Time 1  29.0  0.9  29.2  0.6  -  Time 2  16.8  0.7  15.1  0.7  -  Time 1  78.2  0.9  80.3  0.7  -  Time 2  71.0  2.4  71.7  1.6  -  43  Figure 2.6: Mean energy content (± 95% C.I., n = 4) of leaf tissues collected at sampling times (early June, late July and mid-August).  (B/IBOM) »ua*uoo ARiaug  Table 2.8: Mean (n=4) gross caloric content (Kcal /gram dry weight) of leaf tissues collected at three sampling times; early June, late-July and mid-august.  Time 1  SE  Time 2  SE  Time 3  SE  Achillea millefolium  4304  10  4289  75  4167  16  Anemone parviflora  4509  19  4231  44  4228  10  Arctostaphylos rubra  4699  25  4574  30  4498  22  Betula glandulosa  5182  18  5226  15  5123  11  Comus canadensis  4201  34  4260  27  4266  29  Delphinium glaucum  4459  35  4245  130  4410  67  Epilobium angustifolium  4443  29  4490  56  4458  19  Festuca altaica  4408  49  4328  29  4304  62  Lupinus arcticus  4761  38  4263  62  3937  90  Mertensia paniculata  4349  40  4004  25  3901  65  Salix ssp.  4960  21  4864  36  4734  24  Senecio lugens  4062  58  3950  35  3938  29  Shepherdia canadensis  4930  14  4855  18  4781  18  Solidago multiradiata  4620  34  4427  45  4472  49  Carex spp.  4375  7  Equisetum spp.  3755  7  Species  46  Anemone parviflora and B. glandulosa are excluded from the analysis it appears that hares are selecting species with high energy content (Fig. 2.7). The average quality of the leaf tissues available in the environment is significantly lower than the mean energy found in the diet at time 1 (Table 2.7). At time 2 no trends were evident. Energy content and protein content are positively correlated at both sampling times (Fig. 2.8).  Fibre  Festuca altaica contained the highest amount offibrein its leaf tissues at all three sampling times, which accounted for between 55 and 60 % of the dry weight (Fig. 2.9). Most of the species sampled showed a trend towards an increase infibrethrough the summer, notable exceptions are Arctostaphylos rubra, Shepherdia canadensis, Salix spp.  and C. canadensis. The deciduous shrubs had relatively highfibrecontent at the first sampling time (>45%) whereas most of the herbaceous plants were below 40%. The averagefibrecontent in the diet remains constant over time whereas the average amount offibreavailable to the hares increases by 15% (Table 2.7). At time 1 the mean fibre content of the diet was significantly higher than what was available to them, yet this trend reverses at time 2 where the average hares' diet was significantly lower than the fibre in the environment.  Water Content  All of the deciduous species showed a decline in water content, through the season, between sampling times, whereas the evergreen species, (Arctostaphylos uva-ursi, C. canadensis, Empetrum nigrum and L. borealis) all showed a large increase in water  47  Figure 2.7: Comparison of standardized selection ratios (B) against leaf energy content individual plant species at the first sampling time, in early June.  o o  CM m  o o o m •  • -2  icn o cco  U  8  4- oo § +  to 3 O  1 <  •  to o  3  (0  ia  o CO  E CO  E  §  c  o O >. 0)  c  3  UJ 4B —  to •  •  re  o o  3  c 2. E  <  o o CN  -+-  CM d  oo  co  d  d  •* *~ d  o  CM 'T  1-;  O  oo o o  CO o o  o d  CM o  o  o o o *  (g) oney uo!J03|9S pazipjepuejs  49  Figure 2.8: Regression of crude protein content and energy content. Each symbol represents a single plant species.  B. Time 2  5500 B. glinduloai Stlbt «pp. 5000  E . angustifollum .8.4500 -\  + tt. ptnlcultti A . millefolium  C. canadantis  1 o  S. lugant  y s 30.106X * 3S09.I  ^4000  R  1  » 0.38  S  UJ  3500  3000  20  2*  C r u d e Protein Content  (%)  51  Figure 2.9: Fibre content (± 95% C.I.) of leaf tissues collected at three sampling times (early June, late July and mid-August). Means are calculated over four samples for all species except B. glandulosa, F. altaica, L. arcticus, Salix ssp. and S. canadensis where  means were calculated using 8 samples.  52  — O  fN U  fl U  see H  H  •  •  tS.  content between times 1 and 2 (Fig. 2.10). Leaf tissues of herbaceous plants have a higher water content (> 80%) than grass (F. altaica 74%) and both evergreen and deciduous shrubs (53-75%). No trends were identified between diet selection and water content, at both sampling times the average water content of the diet was not significantly different to that in the environment.  Secondary  Compounds  Data from Jung and Batzli (1979) were included to examine if there are links between secondary compounds and the species included in the diet (Table 2.9). At both times there are alkaloids and tannins included in the diet, in concentrations decreasing in that order. It should be noted that high (+++) concentrations occur only in species that are significantly selected by hares. Analysis of Variance with all nutritional  characters  No significant results were obtainedfromthe analysis of variance (Table 2.10). However, at both times the protein content explained the largest amount of the variation in the model.  55  Figure 2.10: Mean water content ( ± 95% C.I., n = 10) of leaf tissues of 14 species collected at three sampling times (early June, late July and mid-August).  T-  CM CO  Table 2.9: Secondary compounds identified in leaf tissues of plant species.  Species ** Anemone parviflora * Arctostaphylos alpina * Betula nana * Epilobium latifolium * Festuca brachyphylla Lupinus arcticus Salix alaxensis Salix glauca * Senecio atropurpens *** Shepherdia canadensis  Tannins  Alkaloids +  ++ +++ + + ++ +++  +  +++ + +  +++  Data takenfromJung, Batzli and Seigler 1979. + cloudiness or <lmm ppt., ++ >5mm ppL after sitting overnight, +++ > 5mm ppt. formed immediately. * Plants sampled are in same genera but different species than those included in this study. ** Stem andflowersused as well as leaves. *** Data takenfrom(Ayresl992).  59  Table 2.10: Results from the analysis of variance using all plant qualities to predict the selection ratios at each time. Timet Variable Protein Secondary Compounds Digestible Energy  F Ratio  Prob > F  2.417 0.062 0.054  0.22 0.82 0.83 0.30  Whole Model  Time 2 Variable Protein Secondary Compounds Digestible Energy Whole Model  F Ratio  Prob > F  2.347 0.740 0.038  0.20 0.44 0.85 0.35  DISCUSSION The snowshoe hare is typically faced with an abundant supply of low nutritive quality food items whose availability is further reduced by structural and chemical defence mechanisms. It remains unclear which plant qualities determine diet selection, although secondary defence chemicals, and both protein and energy content, have been proposed as important factors in winter diets. No single species satisfies all of these requirements and all of these plant characteristics are subject to change with plant phenology. Nutritional hypotheses for diet selection in hares were examined. It seems that no single hypothesis explains the results. Diet selection is a result of a combination of several interacting factors. Early in the season, though May and early June, it appears that hares select for species which contain high concentrations of protein, energy, and fibre. At the end of the summer hares select for species with both high protein and lowfibrecontent. All plant species that are selected by the hares contain high levels of defensive compounds. Diet Composition The summer diet of the snowshoe hare in the summer included a surprisingly low number of species relative to what was available in the enclosures; only five of 38 species comprised 85% of the diet (L. arcticus, Salix spp., F. altaica, B. glandulosa and Shepherdia canadensis). Early in the season, L. arcticus, Shepherdia canadensis and  Salix spp. were all included in the diet in quantities greater than what would have been  61  expected by chance. Late in the season, hares continue to demonstrate significant selection for L. arcticus, and start to select for B. glandulosa. The hares demonstrate avoidance of F. altaica at both sampling times, yet the abundance of this species is such that it still comprises a significant portion of the diet (between 20-26%). It is difficult to directly compare these results with those of other studies of summer diet selection by hares because of substantially different methodologies. Wolff (1978) identified that hares consume a large amount of leaves from deciduous shrubs, many of which are the same species as have been identified in this study (Table 2.11). Grisley (1990) identified Equisetum spp. as the most preferred species in two of three preference trials and although it was available to the hares, it was not included in any of the stomach samples. The results from this study overlap with those of Grisley (1990) because 10 of 14 species identified in the diet of hares were common to both studies in the Kluane area (Table 2.11). However, the relative preference for each species was not in agreement. This supports the criticism that the resultsfromcafeteria trials do not reflect what hares would select in the field.  Valid estimation of selection ratios requires that the available resource units are correctly identified (Manly et al. 1993). The available resource units for the enclosures are defined as the species that have been identified as edible to the hare (i.e. they have been identified in at least one of the stomach samples). This limited definition of availability may be a source of bias in the estimates of the selection ratios because it excludes some species that hares may occasionally include in their diet, yet were not identified in these samples.  62  Table 2.11: Summary of results from studies of diet selection in snowshoe hares in the summer. All of the results presented are takenfrommicrohistological identification in stomach contents. Species included on list were available to hares in all three studies except Alnus crispa and Polygonum alaskanum, which were available only to the hares studied by Wolff. Plant Species  Wolff  Grisley  1978  1990  This Study  Shrub Alnus crispa  x  Arctostapylos rubra  x  Betula spp.  x  Ledum groenlandicum  x  Picea mariana  x  x  x  x  x  x  x  x  x  x  x  Populus tremuloides Rosa acicularis  x  Rubus ideaus  x  Salix spp. Shepherdia canadensis Vaccinium spp.  x  Herbaceous Achillea millefolium Anemone parviflora Calamagrostis spp.  x  Cornus canadensis  x  Epilobium angustifolium  x  Equisetum spp.  x  x x  x  x  Festuca altaica  x  x  Lupinus arcticus  x  x  Mertensia paniculata  x  x  x  x  Polygonum alaskanum Solidago multiradiata  x  However, the estimation of these ratios is sensitive to the inclusion of large 'available' groups that are not used, such as Arctostaphylos uva-ursi and L. borealis, that comprise approximately 50% of the biomass available to the hares in the enclosures, yet are not included in the diet. Protein It was hypothesized that hares would select plant species with high concentrations of protein. At both sampling times hares demonstrate significant selection for those species with the highest crude protein content. This supports the hypothesis that hares are selecting for high protein content and is consistent with the results of hare diet selection in the winter (Miller 1968, Lindloff et al. 1974, Sinclair and Smith 1984, Angerbjorn and Pehrson 1987, Sinclair etal. 1988, Schmitz etal. 1992). However, the mean protein content of the diet is not significantly different from the mean protein content available to the hares in the enclosures at either sampling time. There is a fairly simple resolution to this apparent discrepancy, which is consistent with the idea of a nutritionally balanced diet (Pulliam 1975, Westoby 1978, Clark 1982). Hares might be selecting Shepherdia canadensis, L. arcticus and B. glandulosa on account of their elevated protein content thus satisfying their protein requirement and might be selecting other species to satisfy other nutritional demands. Thus, the protein content of the diet does not differfromthat of the environment because of the lower protein content of other species selected for different nutritional qualities.  The protein content of the available leaf tissues decreased by approximately half between the two sampling times,from29% to 15% dry weight. Yet at both sampling times the  64  average leaf contains significantly greater than the 11% crude protein required for body weight maintenance as suggested by Sinclair et al. (1988). Thus it is possible that hares are not protein limited and as such are not selecting for protein, but for other plant characteristics that are correlated with protein content such as energy content in early June.  Fibre  Thefibrecontent of the leaf tissues of all species showed an increase from early to late summer, as would be expected with the increase in leaf structural components associated with growth (Bryant et al. 1983, Bryant and Kuropat 1992). I hypothesized that hares would select species with high digestibility and thus lowfibrecontent. Although the meanfibrecontent of the plant tissues available to the hares increased between sampling times, the meanfibrecontent of the diet remained constant at 45%. Early in the season the hares have a diet high in fibre, yet at the end of the summer the hares' diet contains significantly lessfibrethan the average available in the environment. The importance of fibre has been documented by Cheeke (1983) who suggested that a basal threshold, between 15 and 20% dry weight is required for normal digestive function in rabbits. Snowshoe hares may have evolved a higher tolerance for poor quality food items than closely related species such as mountain hares (Lepus timidus, Bryant et al. 1989) and change their cecal length with the season (Smith et al. 1980). Although, the diet of the snowshoe hares contains significantly morefibrethan rabbits, this could reflect their basal digestive requirements as an adaptation to the poor quality of forage in their environment.  65  Most winter diet studies of hares indicate that the twigs selected by hares have a higher digestibility than the average twig available in the environment (Sinclair and Smith 1984, Sinclair et al. 1988, Rodgers and Sinclair 1997). Thus when availability is greater than 45%fibre,hares select against it, which is consistent with the pattern of selection that hares demonstrate late in the summer. Energy The energy content of the hares' diet is significantly higher than the energy available to the hares early in the season which supports the hypothesis that hares were selecting for plant tissues with relatively high energy content. Of the four species with the highest gross caloric content, three were selected; L. arcticus, Shepherdia canadensis and Salix spp. Both of these results support the idea that hares were selecting dietary items on the basis of energy content. Betula glandulosa is not included in the diet despite its high caloric content at thefirstsampling time. This may be explained by the fact that early in the season the leaves of B. glandulosa are covered with a sticky resinous layer (pers. obs.) which contains high concentrations of tannins (Bryant and Kuropat 1980, Bryant et al. 1992) and is a potential deterrent to hares. Energy selection was not significant late in the season yet B. glandulosa, which contains the highest energy content, is now included in the diet after the resin from the leaf surfaces has evaporated. This inconsistent result makes it difficult to identify whether energy could influence summer diet selection. Thefirstdiet samples were collected at a time when the female hares were in estrous and all of the males were in breeding condition whereas the second samples were collected after the females had dropped their litters and were no longer  66  receptive. The different sexual states could potentially alter the foraging goals of the male hares and explain the discrepancy in energy selection between the two times (discussed further in chapter 4).  Secondary Compounds All of the plant genera selected by the hares at some point in the summer diet contain high levels of secondary compounds, whereas all genera not significantly selected contain low levels (plant chemical results taken from Jung and Batzli 1979). This provides an interesting example of co-evolution in plant herbivore dynamics. Previous studies have clearly documented the numerous plant defense compounds within the woody species available to herbivores in the winter within the boreal forest and their potential deterrence to hare feeding (Bryant et al. 1980, Bryant et al. 1985, Sinclair and Smith 1984, Bryant et al. 1992). It is thus significant to identify that hares continue to ingest heavily defended species (e.g. Shepherdia canadensis, B. glandulosa and L. arcticus) in the summer when many less defended plants are available to the hares (e.g. F. altaica, E. latifolium and A. parviflora). It is interesting to note that at both sampling times the species that are positively selected concurrently contain high levels of different groups of secondary compounds. During thefirstsampling time, the hares selected L. arcticus, Salix spp. and Shepherdia canadensis that contain high levels of alkaloids and tannins. At the second sampling time hares selected B. glandulosa and E arcticus which again differ by containing high levels of tannins and alkaloids respectively.  67  Secondary defence compounds have dosage-dependent toxic properties (Freeland 1992) and it has been shown that herbivores can select a variety of dietary items such that they minimize adverse effects of any single toxic secondary defense chemical (Freeland and Saladin 1989, Bernays et al. 1994). This trend could suggest that hares are selecting a variety of species with different defensive compounds and avoiding both the accumulation and detrimental effects of any individual toxins that would support the dilution hypothesis. Rabbits do not suffer any negative effects from Lupinus alkaloids when they comprise up to 29% of the diet (Johnston and Uzcategui 1989). Mice have demonstrated the ability to select a diet including both tannins and sapponins in proportions that nullify the toxicity of either compound (Freeland et al. 1985). Therefore it is possible that hares are using this same strategy in diet selection, because of the consumption of small amounts of many different phytochemicals. The complex issues surrounding the different modes of action of the diverse plant defense compounds is beyond the scope of this analysis yet the possibility that these compounds may have a significant impact on the processes must be acknowledged and deserves a more complete examination. Whole Plant Analysis Prior to the analysis, I was unable to hypothesize what plant trait would be most important when food items contained both positive and negative characters, since so little is know about diet selection during the summer. The hares were ultimately presented with two options i) plants that contain high protein with high doses of defense compounds and ii) plants with low protein content and low concentrations of defense compounds. All of the plants that were identified as positively selected were heavily defended, the plant  68  species that were avoided or taken on proportion to their abundance both comprised the latter group. Ultimately the analysis of variance suggests that protein content is the plant character most important to selection. Although, it is clearfromthe avoidance of Betula glandulosa  early in the summer extremely high concentrations of secondary compounds  will alter this pattern and deter hare browse.  Summary The summer diet of snowshoe hares is composed predominantly offivespecies; Lupinus arcticus, Salix spp. Shepherdia canadensis, Betula glandulosa and Festuca altaica. The  selection for these species changes between sampling times yet there is no single explanation for the relative changes in their inclusion in the diet of hares. The selection patterns displayed by the hares may reflect the need to balance many conflicting plant qualities. No single dietary item can satisfy all of the hypothesized nutritional requirements (i.e. high energy, high protein, lowfibre,and low defense compounds) and this investigation has not successfully identified any clear trends in what dictates diet selection, Of all the forage characteristics examined, it appears that protein content consistently explains the largest amount of variation in diet selection, although, protein selection is modified by extremely high concentrations of defense compounds (as seen by the avoidance of B. glandulosa at time 1).  At this point, progress in understanding nutritional influences on hare diets should follow either a more detailed biochemical approach examining interactions of phytochemical defense or concentrate on more manipulative work similar to that of Schmitz et al.  69  (1992). This latter technique will help to elucidate patterns and identify operative offs in nutritional foraging decisions.  CHAPTER 3 SUMMER HABITAT SELECTION BY SNOWSHOE HARES  INTRODUCTION All prey organisms are faced with the dilemma of obtaining sufficient nutritional resources to grow and reproduce while at the same time avoiding death by predation. Organisms have various ways of balancing these two pressures, one of which is to adopt behavioral strategies that increase theirfitnessin face of these selective forces. One way in which we can examine how animals balance these selective forces is through their habitat selection. The distribution of animals then becomes important in understanding diet selection processes. In this chapter I will attempt to identify patterns in summer habitat selection by snowshoe hares by comparing how they are distributed in the environment relative to both preferred food species and cover from predators.  To the herbivore, different habitats vary in terms of foraging profitability and predation risk. The way in which organisms select habitats according to these vegetation characters can reveal the behavioral strategies employed by that group. This trade-off has been examined in many small mammals. Prairie vole distribution is influenced by both food and risk of predation (Desy and Batzli 1989), whereas tundra voles are influenced primarily by high quality food items (Batzli and Lesieutre 1991). Predator induced habitat selection has been demonstrated for many small herbivores (Lima and Dill 1990), and small rodents have been shown to increase their use of dense habitats in the presence  71  of predators and increased illumination (Brown et al. 1988, Kotler et al. 1988, Gilbert and Boutin 1991, Longland and Price 1991).  Investigations of diet selection often ignore the spatial distribution of forage species relative to cover, yet it is important to consider when trying to quantify forage availability (Norbury and Sanson 1992). Wolff (1980, 1981) reported that hares select habitats that contain an abundance of both food and coverfrompredators at low population densities, yet as hare population numbers increase, they expand their range to include areas of lower quality in terms of both resources. An examination of the spatial distribution of winter plant species near Kluane, in the Yukon, revealed that there was a larger amount of forage available in the open areas and less in the closed habitats, yet, hares spent a greater amount of time in closed habitats during peak hare population numbers (Hik 1995). Thus, estimates of availability of forage over the area as a whole would be greater than what would be perceived as available or encountered by a hare. However, Hodges (1998) couldfindno evidence to support this and also showed that during low hare population densities, there was typically no food - predation risk trade-off.  Understory density has been identified as the principal feature influencing habitat use by hares (Wolff 1980, Orr and Dodds 1982, Pietz and Tester 1983, Litvatis et al. 1985 Rogowitz 1988, Litvatis 1990). Habitat use during the winter does not correlate well with food availability (Rogowitz 1987), yet patterns in association with preferred food species have not been examined. During the summer, hares are thought to migrate to more open areas, with more choice of herbaceous food, when cover is increasing during the growing  72  season (Wolff 1980, Wolfe et al. 1982). However, foraging behavior and habitat use by hares in the summer has not been well studied.  Avian predators of hares have higher hunting efficiency in open areas (Hik 1995, Rohner and Krebs 1996), and a greater proportion of hare deaths occur in open habitats than in closed habitats (Hik 1994, Murray et al. 1994, Rohner and Krebs 1996). This suggests that hares are more vulnerable to predation when foraging in open habitats. Other studies have also shown that survival of snowshoe hares is higher in dense habitats (Litvatis et al. 1985, Sievert and Keith 1985). Clearly, there is a greaterriskto foraging for plants located in open areas, which could alter a hare's perception of their abundance.  My primary objective in this chapter is to identify which habitat characteristics hares are selecting for during the summer. The secondary objectives are to examine whether there are habitats that provide hares with an abundance of both food and cover and if the spatial distribution of hares in the environment alters their encounter rates of preferred food species.  I examine the distribution of hares using fecal pellet plots and compare the presence or absence of fecal pellets to the distribution of food plant species and cover. The hare population was at peak density at the time of the work, which allows us to examine the hypotheses of Wolff (1980) and Hik (1995) and identify whether hare habitat selection in the summer compromises food and risk of predation. If the hypotheses of Wolff (1980) and Hik (1995) are supported, we predict that  73  1) Hares will select habitats with an abundance of cover. 2) Hare will avoid open habitats with little or no cover. 3) Food plant species in closed habitats will be encountered more frequently than their measured abundance, whereas plant species located in open habitats will be encountered less frequently than their measured abundance.  METHODS Location  All of the work described in this chapter was conducted near Kluane Lake in the southwestern Yukon Territory (61°N, 138° 30' W) during the summer months, from May though to September 1998. This was the site of the Kluane Boreal Forest Ecosystem Project (KBFEP), and as such is described in detail in both Chapter 2 and in Krebs et al. (1986).  Experimental Design Study Site  Three study sites were selected along the Alaska Highway, located between mile 1030 and 1054. Each site is 36ha and previously used as control grids by the KBFEP. The sites were each separated by approximately 10 km and have been labeled sites A, B and C, locally known as Silver, Sulfur and Chitty respectively.  74  Fecal Pellet Plot Construction  At each of these three study sites 200 fecal pellet plots were selected and a total of 600 plots sampled. Each pellet plot was lm x 0.05m. However, on site A, 80 of the plots were 3m x 0.05m. Pellet plots were distributed systematically over each of the three study sites.  Fecal Pellet Plot Surveys  Fecal pellet plots were selected in mid-May 1998. All plots and surrounding 10cm were cleared of all hare fecal pellets. The number of new pellets deposited on each of these plots was counted at the end of August 1998.  At each of the 600 pellet plots, habitat variables were recorded to describe the vegetation for the area surrounding the plot within a radius of 5m. The following measurements were made: i) Tree cover and species; recorded as percent cover in size groups of 0-5, 6-25, 26-50, 51-75, 75+%cover ii) Shrub cover and species; recorded as percent cover in size groups of 0-5, 6-25, 26-50, 51-75, 75+%cover iii) Deadfall; recorded as percent ground cover in groups of 10%, i.e. 0-10, 11-20, 21-30. iv) Standing dead; recorded as percent cover in groups of 10%, i.e. 0-10, 11-20, 21-30. v) Cover at a height of l-2m above the plot and species; recorded as presence or absence of cover  75  vi) Distance to cover and cover type was recordedfromboth end markers of the fecal pellet plots. Distances were measured up to a distance of 5mfromthe plot. Cover was defined as a location that encloses a hare on a minimum of three sides.  Plant abundance was recorded within a 5m radius surrounding the plot for the following plants groups: a) Shrubs Abundance was recorded for Betula glandulosa, Salix spp., and Shepherdia canadensis as the amount of vegetation within reach of the hares (i.e. under lm in height) in size classes of percent cover 0-5, 6-25, 26-50, 51-75, 75+.  b) Herbaceous plants Abundance was recorded for Achillea millefolium, Anemone parviflora, rubra, Epilobium angustifolium, Equisetum spp., Lupinus arcticus, Solidago multiradiata  Arctostaphylos  Mertensiapaniculata,  and Senecio lugens. The number of individuals within a 5m radius  of the pellet plot was recorded in groups of 0, 1-10, 11-50, 50+.  c) Grasses  The abundance of Festuca altaica was classified in size classes of 10%.  Fecal Pellet  Transects  The number of fecal pellets is an index of the amount of time a hare spends at a location, and as such, time spent associated with different habitat variables can be assessed. At 76  each of three study sites 50 transects (dimensions 10m x Im) were surveyed for fecal pellet piles that consisted of greater than 10 fecal pellets and had been deposited over the summer months. The 50 locations were selected systematically over the 36ha and were different stations than the fecal pellet plots. Only summer fecal pellets were counted which were identified both by their color and texture. A pile was defined as a minimum of 10 fecal pellets located within a small area such that no pellet was separatedfromthe pile by a distance greater than 10cm.  Within a 5m radius of the fecal pellet transects, habitat descriptions were recorded. The following measurements were made using the same scales as outlined in the previous section; tree cover and species, shrub cover and species, deadfall cover, amount of standing dead trees, abundance of Salix spp., B. glandulosa, Shepherdia canadensis, L. arcticus and F. altaica.  Once a fecal pellet pile was identified, some additional variables were measured from that pile; cover at a height of l-2m above the pile, distance to cover, cover type, and distance to closest source of each of four preferred food items; Salix spp., B. glandulosa, Shepherdia canadensis, and L. arcticus.  11  Statistical Analysis Fecal Pellet Plots  Three different statistical methods were used to identify patterns in habitat use by hares; simple correlations, analysis of variance with composite variables, and a cluster analysis followed by a logistic regression.  Pearson  correlations  Pearson correlations were applied only to the plots on which fecal pellets were deposited. Each of the three sampling sites was analyzed separately.  Analysis of Variance with composite variables  Habitats variables were condensed into three composite variables, which describe food abundance, understory density, and canopy density. These variables are an index of food abundance, visibility to ground predators, and aerial predators, respectively. Food abundance was calculated as the sum of abundance measures for four preferred food species; B. glandulosa, L. arcticus, Salix spp. and Shepherdia canadensis. Understory  density is the sum of shrub abundance within lm of the ground, deadfall cover and the distance to cover (= 3- (distance to cover/ 2)). Canopy density is the sum of tree cover and cover above the plot at a height of l-2m. A logistic regression was performed with these composite variables to predict the presence or absence of fecal pellets in each plot.  78  Cluster analysis and logistic regression  A cluster analysis was performed to identify patterns in the data assemblages. All strongly correlated variables were removed and strongly skewed variables with greater than 50% zeros were treated as binary variables (i.e. presence or absence). A logistic regression was then done on each set of clusters with the presence or absence of fecal pellets as the dependent variable to identify clusters (habitat types) that hares were either selecting, avoiding or using in proportion to their availability.  An ad hoc logistic regression was done to predict cluster membership. This allowed the identification of the variables that accounted for the largest amount of the variation between clusters. Once the important variables were identified, descriptive statistics were used to compare cluster means and thus identify the parameters relevant to hare habitat selection within each group of variables.  Fecal Pellet Transects  A logistic regression was conducted separately for each sampling site in an attempt to predict the presence or absence of a fecal pile on the transect with respect to habitat characteristics. A stepwise approach was used to obtain the most simple significant model. All redundant variables were removed and strongly skewed abundance variables were treated as binary (presence or absence) variables.  79  RESULTS Study Sites  The three sites surveyed had fundamentally different habitat characteristics (Table 3.1). Sites B and C were most similar in their stand characteristics. Site A habitats generally had a low canopy density with abundant standing dead trees and deadfall, high shrub cover and an abundance of Salix spp., B. glandulosa and L. arcticus. The majority of the plots on site B had a dense spruce canopy (between 25 - 75% spruce cover), and a sparse understory which contained low amounts of preferred diet items. Salix spp. was the mosit abundant shrub. Site C is characterized by having a young dense spruce canopy with low shrub cover and with Shepherdia canadensis being abundant in the understory.  Fecal Pellet Plots  Pearson correlations identified the most consistent and interpretable trends for the fecal pellet plot data (Table 3.2). On all three sites, hares selected areas that were close to cover. As the distance to cover increased there was a decrease in the number of fecal pellets on each plot (Fig 3.1).  On sites B and C, hares selected areas with abundant deadfall which also provides cover in the forest understory. Plots with extremely low canopy cover and those located in areas where the only source of cover is white spruce were avoided. Food plants that were selected include Achillea millefolium, Arctostaphylos rubra, L. arcticus and Shepherdia  canadensis, whereas E. angustifolium was avoided.  80  Table 3.1: Habitat characteristics for the three sites. Numbers are the percentage of plots that are described by each character level.  Site A  Site B  SiteC  N  193  194  197  Tree Cover Low cover (0-25%) Moderate Cover (26-50%) Dense Cover ( >50%)  54 43 3  34 40 26  16 36 48  Shrub Cover Low cover (0-25%) Moderate Cover (26-50%) Dense Cover ( >50%)  9 49 42  38 51 11  65 32 3  6 62 31 1  42 46 11 1  75 22 2 1  89 9 1 1  87 13 0 0  100 0 0 0  Scarce (0-5%) Low cover (6-25%) Moderate Cover (26-50%) Dense Cover ( >50%)  81 19 0 0  77 21 2 0  42 30 24 4  abundance Absent Scarce (1-10 indiv's) Frequent (11-50 indiv's) Abundant (>50 indiv's)  5 6 27 62  97 3 0 0  64 11 20 5  40 45 15  47 24 29  36 35 29  Habitat Characteristic  Salix spp. abundance  Scarce (0-5%) Low cover (6-25%) Moderate Cover (26-50%) Dense Cover (>50%) Betula glandulosa abundance  Scarce (0-5%) Low cover (6-25%) Moderate Cover (26-50%) Dense Cover (>50%) Shepherdia canadensis abundance  Lupinus arcticus  Cover l-2meters Absent Cover 50% Covered 100%  81  CU  c o  8 CO  08 eu  u  CJ  il  B Q  >  O  eu  ; x  eu C  1 •*-» co  CO O  "o.  "O  O.  eu  "o3 o ,eu  O  i-  8 £ 2 |  if  +2 -is  •2 ^ eu  CO  CU  1 cd  43  es  43 O  15 cd  4S  to  p  •S .S eu "Z  S co  cd  v. g> 8 « cU r" T3 .5  eu  cu g  eu eu  2 « 52 a co 5 co 2 £ g 8 g g  8 .S .S •o  8  «  •8  .s  e  I? t. cd  fO O 00 i— O —"  v  o o  M M  N ft  O «  5  M «  0\  o o o o o o o o o o  eu «> eu eu S  gg cd  So  8  ° g £  cd  CO ed  jg  J?  « CO cd  £ £  1.11.1  00 «/"> t-» VO <N ^ 00 ON ON © © © © © © O  o' © o ©" o ©  1 C Ol ON  O.  eu  n  ©  f-  CN  oi  t- £j —  C7s  —  —  »N  CS  o ° o" o  £2  ^£s a  o ©' © ©" ° ©  u  c2 CM  eu  O  co C  o  co  '£ "S *  1 co CO b. cd  8  cd  CU  JS  "S  "O  e  3 43  e  cd  c —  cu  2— tfa  CU  co . 3  •s «> 3  ii  .2  es  CO 4)  2 2  si cd  5  §  ^  <N =  "> S. cu _T  1 8 H  cu  •*mi  <  II  c  3  eu > Q  o  tcu  >  S  o o R o > is 5 eu «—  C <4—I ed - O  .3 g  Q Q ca  cu  8 2 CO  f  cd  e  •s  1  cd  "O  3  •8 3 s  c  c  c c cd  cd  5  eu o  eu c cd o •o  eu  co  o  eu o  3 43  ^1 8 I8  ed  5  8 3  eu  cd  -C3  >  u.  O  o ^ o  /-N  ? ° 8  §1 C C*H cd T3  .2 g  C cd  II ^ 3  o C  .2  to  Si  Q Q ON  II  e  c2  82  Figure 3.1: Correlations of fecal pellet number on each plot against distance to cover for three sites. Site A: y=5.67 - 0.84x; p<0.0830 Site B: y = 3.24-0.347x; p<0.0383 Site C: y = 4.31 - 0.62x; p<0.0077.  83  A. Silver  B. Sulfur 14 12 0  a 10 cu  •a 1 & «•* o u  a  8  H  6 4 2 0 0  C Chitty  18  •  16  a 8. •«  14  S3  •  12  il w  10  o  8  2  6  •  • •  •*  4 2 0  •  •  •  •  •  «*  •  *» •  4MMMM» • « 3  * • •  «a» < »» • ~  «»  •  •  4  Distance from cover (m)  84  Composite Variables for Fecal Pellet Plots When the habitat variables were grouped into three composite variables of food availability, understory cover, and canopy cover, a range of results emerged. Understory density was identified as the variable most related to fecal deposition on two of three sites, B and C (Table 3.3). Hares select areas with a medium to high understory density, composed of both shrubs and deadfall although neither exclusively. On site A none of the composite variables were sufficient to predict fecal pellet deposition.  Neither of the composite variables for food abundance or canopy explained the fecal pellet distribution pattern. It should be noted that understory density was positively correlated with food abundance on both sites B and C (Table 3.4). Distance to cover was negatively correlated (p<0.05) with food abundance on all three sampling sites (Fig 3.2).  Cluster Analysis The resultsfromthe cluster analyses are presented separately for each study site. Three main habitat types were identified on each site once the outliers were excludedfromthe analysis (Table 3.5). The overall hare habitat selection of each cluster is indicated by the chi-square value. No single variable was consistently selected across all three sites. On both sites A and B fecal pellets were found more than expected by random chance in habitats with relatively low tree cover, high shrub cover, low deadfall and high Salix spp. abundance. Whereas on site C, the fecal pellets occurred in areas with relatively high tree cover, high deadfall, low shrub abundance and high L. arcticus abundance.  85  Table 3.3: Summary of results from the composite variables analysis. The variables are used to predict the occurrence of fecal pellets in a plot at each site. The variables for canopy include measures of tree canopy and cover at l-2m above the transect. Understory density includes measures of shrub abundance, deadfall abundance and distance to cover. The food variable includes abundance measures of Salix spp., Betula glandulosa, Shepherdia canadensis and Lupinus arcticus. Site A  Grouping Variable Canopy Understory Food  P  -  P (fecal pellet) increases with: -  B  Canopy Understory Food  0.102 -  increase in understory density -  C  Canopy / Understory Food  0.059 -  increase in understory density -  86  Table 3.4: Correlations between composite variables at each site. Values in bold are significant at p<0.05.  Site  Variable 1  Variable2  A  Understory density Understory density Canopy Cover  B  C  Correlation  P<  Food abundance Canopy cover Food Abundance  -0.004 -0.021 -0.078  0.954 0.777 0.278  Understory density Understory density Canopy Cover  Food abundance Canopy cover Food Abundance  0.462  0.000  -0.104  0.150  -0.430  0.000  Understory density Understory density Canopy Cover  Food abundance Canopy cover Food Abundance  0.344 0.210  0.000 0.003  0.032  0.660  87  Figure 3.2: Standard regression lines of composite food variables against distance to cover for each study site. Site A: y = 6.3-0.23x; p<0.014. Site B: y 2.9 - 0.18x; p<0.003 Site C: y=3.4 - 0.16x; p<0.023.  A. Silver 12 10  s^  «  •  • • •  6  «»«»tt — • • «•• • •  »*»  •  4  •  •  •  4>  • •«  •  •  •  •  •  ** *~  • *  • •  •  *  2 0 0  B. Sulfur 7  • • «• •  •  •  C.Chitty 7 0)  n  *  6  •  •  •  •  5  •  •  •  •  •  4  • • • «•  4> «>«>  •  1U l l  •  •  • •  •  3  "«• # «  *> •  2  E o o 1  » •• 1 1 1 1 1  • «*> •  »  «M»  • 4>  4*  •  « 4M*«  •  •  • •  * •  "  «•*>  *• •  •  *>**» «* •  •*> *>«>•«•>•  •  0  ««M>•«  «*> « * * > •  •  4>  0  t tt •> •  -  • •  •  •  •  •  •  1  —1  1  1  1  2  3  4  5  6  7  D i s t a n c e to C o v e r (m)  89  co  A  3 TS  3 8;  2  _ © o O v V  a.  0  o © V V  3S  J2  - 2 - f> 2 2 00 00  <u «3 .  co  '5  Tf  o» c i vo' vo en  c  c  o  .•2 U  <N — — r -  o  o> .ti g> o 4>  ^  o  o  <N —  —" — o  0  0 0 vo r p i V9 VO N  _ o  _ o  o  o  — — <scs  0  o  m n ©  s -°  N  —  o  o  o  ^  — — —. — - !  o  o  o  ie  H  d o  <s o o  o  o  o  <1  o  o  N  o  o  fi o  <~l CO  £ o "3 o.  •i * S «  ^  _ —* o* o  © o o  o  o  8 '55 H <*< G S 0  H  «> £  to _3  3 C  8  E  <s — — r~ «N — — — — CN CN — -<. — . - - o ^ ^ o ©' 0 o © © o o © o o © —  le *.-a 3  0  00  >  r, ^ « t ^ . N t w t <*> 9 . p i 6 - 2 2 « o ' o ' N * ' n - ' 0 ® w  so *-  |  •c * o>  10  i§ U CO  •a| >  w'ii OO 1—*  4> ecj  in  tt  +-*  "O c  2^ *  *£ •c  H  • cd  CO  0  u  u  s  co I  u  CA  X  08  O  H  *•  '  -  e  O  ft.  "cu  •8  11  — r-  11  A JS  gS  o  8 f.S  «" d  V  ^ o  (S _ „ 2 P o o o oo o «r> *T n v o «*> . -  5.  loo  3  O vo  •8 3 £ C S f>  £1  aj 'co  C o « o.  CA  co ii CO  ^ _ _ — o — O — o ' o ' d o ' d d o o  cs — - ; — CN —; — I O O O O O O O O  VO 0 0 O CN CN —| • * CN , o-i — « ' © © © ' — < O O  CN «* — CN ( N —> C N O © — — f\  f\ CN ©  ©  1  ©  ."2 1 O vo |<  '•5 *  u  cd <D c s  c  >  S  CA  £  «j co J>  ¥ 1 t - © » - « © e N ro © —' © © ©' ©  C ^  H  cu  3  c  O  «  J3  -  is i  Ed  a <t3 3>  <s, ^ — — © — © © © ' © ' © * d ©" d  -* ©  t  <N - « -H — © — © — © ©' d © © ©' ©' d  eu  ~\ •C  in * CN  d  m m o v ^ - O t * f i © © © O CN © ' © '  co  co _  CN — — CN — — • © © © © o © © o  i  ,  <t ^  --> © d  « » PI —< i d  CN — —| —m r-\ CN —< —I d o © © o © o © ew oo oo o — © • * «* — CN © © © CN — ©  •8  S  i  o  i  vo |  vo  rt  5  co  Q  CU  *i •— cd  co  s*-° O cd cd C M S O  2? .2 W * CO t o -H  -p cu  C «> cd cu H  xi -o in c  cd  e s  CO I  5  -  .cd CO  H  ed  ca eu  in  •c 8  8  ea J= U  •a  i  u  I  w  I-  v  it-  S  S  Si 8 * ZH © w e o _ ex  i  2  fi H  « 1 =3  •J  a.  £ -9  Illli  (A  0  m  •I «  tfl  9  2  A  o 5  e  -o  CN NO  CN •*  th  ©  25  C  © © ?  a.  73  e  S  co  •a ° ctj  oo cn J v£> « s  t  n  • *  ~ S %» s  • 8* 5  t  a n — 00 VO c*> 00 © 00 CN  «  P>  ;  O.  4>  -O  "H S  « 2 *& »  co  £  CN — — >n  _ O  - CN ° . CN CN © ^  — ©  CA  O _ — CN — — O CN —. CN © " © ' © ' © © ©'O © © © ©  p r« ^ « t~ oo q ov. * * © ^ ' J r i M' - o ^ o <s  |  § sc £ cd  co  ea q> "co £ o H u:  «> to  3  CA  d  © © CN . © • © • © © ' © ' © < = .  vo © r~_ ©. r - o . —; oo © ©' © ' ©' © c i © © © *N ©  O 't. I . ro © © ' © -  5 ^  © © © © <=>  £  3  C  73 g  8 a » o .S *  w  CA  3  •f  _ _ _ o  _  N *n oo  H  . ©  CN-©*"*  Of*  -<  0  -  -  -«  © -  -  —  ©©'©'-«' . © ' © ' © © © < = > o ° ° ° ° >r\ «N o> vp CN e>, -  -  t- ©  ° t-;  ' - © - © © C N C N O O C N - ©  c* 4> co  § 3  73  3  .2I <l*-  <u o oo  8  «.a  8 a w  "I  »T3  •5  >•—*-•  c3 J= u  •c  ea H  ea .fi I  *  H  CO  co  C/5  eg H  Im  I •5 «B 2 -  - >s |  * w w > © — oe w 5 .a «~ 3  1£ s i l 9  I  6"*  G  s  E  t- a. 3 f S 1  jlllllllll  C A & 3 GQ  O  ^  u e u  The composite variables for the clusters indicate that high food species abundance was the variable most associated with habitat use by hares on site A. All of the clusters from site A have high understory density and low tree cover. Habitat selection in site B is associated with increased food abundance, high understory density and avoidance of dense tree canopies. The composite variables for site C indicate that there are no major differences in food species abundance between habitats, thus, hares select habitats with both a dense understory and canopy.  The variables that account for the largest amount of variation between cluster groups were identified; both deadfall and Salix spp. abundance were consistently significant (p<0.05) in differentiating between cluster groups (Table 3.6).  Fecal Pellet  Transects  Significant (p<0.05) and clear models were obtainedfromthe logistic regressions done for each sampling area (Table 3.7). Hares avoided areas with an abundance of F. altaica on all three sites and B. glandulosa on sites B and C. On all three sites both F. altaica and B. glandulosa  are associated with decreasing tree cover (Table 3.8), thus, it appears that  hares are avoiding unforested areas. Hares selected areas with L. arcticus on site A and Salix  spp. on site B.  93  Table 3.6: Resultsfromthe logistic regression used to predict cluster membership. Only significant are included in the tables. Site A - Silver Variable Shrub cover Deadfall Standing dead Salix spp. abundance Salix spp. cover Deadfall cover  Chi-Square 13.55 6.65 4.81 8.38 11.51 11.83  Whole model Site B -  Probability 0.0011 0.036 0.0904 0.0151 0.0032 0.0027 <0.0001  Sulfur Variable Tree cover Deadfall Standing dead Salix spp. abundance Spruce atl-2m Salix at l-2m Spruce cover Salix spp. cover  Chi-Square 19.2 17.21 12.53 12.7 16.57 12.66 11.66 9.56  Whole model  Probability 0.0001 0.0002 0.0019 0.0017 0.0003 0.0018 0.0029 0.0084 <0.0001  SiteC - Chitty Variable Tree cover Shrub cover Deadfall Salix spp. abundance Lupinus arcticus abundance Spruce cover Deadfall cover Whole model  Chi-Square  Probability  34.84 9.08 15.31 9.36 20.04 5.29 13.57  0.0000 0.0107 0.0005 0.0093 0.0000 0.0071 0.0011 <0.0001  94  oo  i i o. 3 •2 8  XI • •M  U 8  •a  1  US* 8  u  — o  fo  ©  •©  Os fO  o d  m  00  O  o  d  d  WI  VO <N O  d  VO  /") •< o d  o  «/i CN o  d  d  o  o o  o o  d  d  o  o  A  s a.  •S 2 cd  •c  <3 C3  5  es  > ©  I  CO 43  s  2 • CJ •S 3 •a.  ^5  43  "3  1f CA  18  3 8  CO CO -2 a  I  S ? 8  | 31 *a  43 CA  JB  8  s I  § I  1 JU  "3  -S  a  -S  §  1 "o  cq fa.  * cn co oq it.  CQ  95  Table 3.8: Correlations between tree cover and abundance of Betulaglandulosa and Festuca altaica. The result in bold are significant (p<0.05) Correlation  P<  B. glandulosa abundance F. altaica abundance B. glandulosa abundance  -0.26 -0.11 0.36  0.0002 0.1133 0.0000  Tree Cover Tree Cover F. altaica abundance  B. glandulosa abundance F. altaica abundance B. glandulosa abundance  -0.26 -0.25 0.27  0.0003 0.0005 0.0002  Tree Cover Tree Cover F. altaica abundance  B. glandulosa abundance F. altaica abundance B. glandulosa abundance  -0.34 -0.44 0.57  0.0000 0.0000 0.0000  Variable 1  Variable 2  A  Tree Cover Tree Cover F. altaica abundance  B  C  Site  96  DISCUSSION  Scale is an important issue in studies of habitat selection. It is difficult to argue what scales hares might be sensitive to in their habitat selection, yet with differences in predation risk, forage species abundance and cover having been detected at microscales in the study area (Hik 1995, Rohner and Krebs 1996, Hodges 1998) it seemed appropriate to examine small scale selection during the summer. Hares' summer home ranges are about 1 lha for females and 20-24ha for males, yet half of their activity is confined to only 20 to 30% of the total area (Hodges 1998). The median diameter of overstory patches in Kluane falls within 79-188m (ODonoghue 1997, Hodges 1998), yet hares home ranges are sufficiently larger than this, thus, hares move between habitat types. Studies of habitat selection in hares have examined a range of scales from small local measurements (up to a distance of 15m i.e. Hik 1995, Hodges 1998) to several hectares (Wolff 1980, Orr and Doods 1982, Wolfe et al. 1982). Differences in hares' survival have been detected at a scale of 15 - 20m (Hik 1995, Murray et al 1994, Rohner and Krebs 1996). This suggests that selection needs to be considered first at small scales and whole stand analysis may not be appropriate in Kluane.  What habitat characteristics are hares selecting for? Hares were selective in their choice of habitats through the summer and were not distributed randomly in their environment. However, the vegetation characters that influenced this habitat selection were not consistent between the three sites. It is difficult to identify vegetation characters that are both relevant to snowshoe hare choice and that are consistent across habitat types.  97  The resultsfromthis analysis are inconsistent and not easy to interpret. The one habitat character that emergedfromthe analysis is that hares locate themselves such that they minimize their distance to cover within habitats in which they are found. Also, on two of the sites, hare distribution correlates with dense understory characters. This is consistent with results of previous studies demonstrating that hares select habitats with cover close by (Hik 1995, Hodges 1998).  The composite variables are an attempt to summarize the general vegetation traits involved in hare habitat selection. This approach facilitates comparisons between sites because the selection depends on general availability conditions of food and cover and does not depend on changes in individual species abundances. It is particularly useful since each of the three sites is at a substantially different stage in forest maturity. Site A is located in a mature forest, composed largely of standing dead trees, with an abundance of canopy gaps and deadfall, which ultimately provide low tree cover. Relative to the other two sites, site A has an abundance of both preferred food species and cover. At this site, when both cover and food are abundant, hares select habitats with an abundance of food.  Sites B and C were located within areas with moderate to dense canopies. Site B contains a mature healthy forest whereas site C has a young dense white spruce canopy. Both sites have low shrub cover and preferred food plant species relative to site A. The composite variables indicate that in both locations habitat selection is influenced by understory characteristics. Hares in site B select locations with abundant food and avoid  98  dense overstory, whereas hares in site C select areas with a dense understory. It appears that when food and cover are low, hare habitat selection is primarily related to understory density and thus predation risk. This is consistent with the many studies which have identified this correlation (Orr and Dodds 1982, Pietz and Tester 1983, Litvatis 1985, Rogowitz 1988, Litvatis 1990, Hik 1995, Hodges 1998).  Are food and cover spatially separate?  Hik (1993, 1995) hypothesized that hares are faced with a food - predation risk trade-off during the winter. One clear trend that emerges at all three sites is that the Hik (1995) hypotheses are not supported; during the summer months, preferred food species and cover are not spatially separate. Therefore hares are not faced with this trade-off. At each site the distance to cover increased with decreased food abundance. This can be attributed to the fact that preferred food species are often a source of cover for hares. Of these, Salix spp. is used most frequently. At both sites B and C the composite variables for food abundance increase with an increase in understory density.  Are there costs associated with any of the preferredfood species?  The fecal pellet transects revealed that hares avoid both F. altaica and B. glandulosa. Since the abundance of both these species is negatively associated with tree cover, it suggests that hares avoid unforested, open areas. The B. glandulosa data demonstrate the only situation identified in this study where a preferred food item is spatially separated from cover and located in a habitat which hares avoid. It has previously been reported that hare survival rates are higher in closed habitats (Sievert and Keith 1985, Litvatis et  99  al. 1985) and that avian predators of hares demonstrate higher hunting efficiency in open areas (Hik 1995, Rohner 1996), thus there are clearly costs to the hares foraging within them. This suggests that there might be a cost to hares feeding on B. glandulosa and indicates that abundance measures of both B. glandulosa and F. altaica are likely to overestimate a hares' perception of availability of these two species.  One notable exception to this trend is amongst the clusters on site A where the habitat clusters that hares are selecting contain a significantly higher amount of B. glandulosa than the areas that were avoided. Betula glandulosa is typically an early successional species, found in open meadow areas with high light levels (Douglas 1974, Dale and Zbigniewicz 1996). In younger forest stands, it inhabits open meadows and clearings, and is not found beneath dense, young canopies typical of the forests on both sites B and C. Therefore B. glandulosa is only identified in meadows at these sites and is spatially separatefromcover. In contrast, on site A, the habitat heterogeneity resultingfromthe mature dead canopy with an abundance of canopy gaps allows sufficient light to penetrate to the shrub layer and thus B. glandulosa is relatively abundant in the understory. The occurrence of B. glandulosa in this heterogeneous habitat, allows hares access to a preferred food source, high in both protein and energy, while reducing the risk associated with more open environments.  Because the study sites provide such fundamentally different habitats, any consistent patterns in habitat selection between them would help to identify characteristics that are important to their choices. All of the study sites have large populations of snowshoe  100  hares. Spring densities recorded in April 1998 indicate densities of 1.78 hares/ha on site A and 2.25 hares/ha on site B (Krebs unpublished data). The difficulty in obtaining consistent trends between the sites highlights the necessity of maintaining several replicate sites for the extrapolation of these data beyond the sampling area. Many differences were detected between sites that are more a reflection of habitat availability than selection. Caution is required in data interpretation because selection at this scale is a question of examining relative versus absolute differences in habitat characteristics.  Summary  Snowshoe hares are selective in their habitat use in the summer. When both food and cover are abundant, hares select areas with an abundance of preferred food species. However, when habitats are poor in both food and cover, hares select areas with a dense understory for cover. Generally there was little evidence to support the idea that preferred food species and cover are spatially separated, for all species except B. glandulosa. The spatial separation of B. glandulosa and F. altaica from cover at sites B and C indicates that forage estimates of abundance overestimate the hares' perception of these species, particularly in young white spruce forests. In heterogeneous mature forests hares have increased access to these species and select areas with an abundance of B. glandulosa. Despite a large sample size, and number of habitat measurements, I am able to explain very little of the observed variability in hare habitat use.  101  CHAPTER 4 THE DIET SELECTION MODEL  INTRODUCTION A number of hypotheses have been proposed to explain diet selection by snowshoe hares. These include differences between plants in nutritional characteristics such as protein and energy (Belovsky 1984, Sinclair et al. 1988, Rodgers and Sinclair 1997), secondary metabolites (Bryant and Kuropat 1980, Bryant et al. 1986, 1992, Schmitz et al. 1992), and the distribution of plants in relation to hare predators (Wolff 1980, Rogowitz 1988, Hik 1995, Hodges 1998). Although there is support for each of these ideas, no single hypothesis is sufficient to explain diet selection over more than a narrow range of circumstances.  Optimal foraging theory has been used to generate hypotheses of diet selection in mammals by developing models based upon linear programming methods (e.g. Belovsky 1986, Belovsky and Schmitz 1992, Schmitz et al. 1998). This technique involves constructing a series of physiological constraint equations and then solving these equations to predict the diet that maximizes or minimizes certain pre-defined foraging goals. This approach has been applied to the diet of snowshoe hares foraging in Isle Royale, Michigan (Belovsky 1984) and in selection between two shrub species during winter in the southern Yukon (Schmitz et al. 1992).  The objective in this chapter is to develop a foraging model that predicts the composition  102  (i.e. identity and abundance of species) of the summer diet of snowshoe hares in the Kluane region of the southern Yukon using the hypotheses of diet selection examined in chapters 2 and 3 plus a few physiological constraints adapted from Belovsky (1984). The model can be solved to predict the composition of the diet for a number of pre-determined foraging goals (e.g. maximizing energy or protein content) and compared to the observed field diets of hares. Similarities between predicted and observed diets will allow us to identify hare foraging strategies.  In Kluane, the summer diet of snowshoe hares is composed largely of herbaceous plants, grasses, and the leaves of some deciduous shrubs. This range in plant material may be the result of hares selecting a diet high in both protein and energy content while avoiding the toxic effects of plant metabolites because hares alter the diet in response to changes in plant nutritional and defensive compounds (chapter 2). During the summer, male hares are in reproductive condition from early April to mid-July; females are synchronously receptive for short intervals during this time. Snowshoe hares are serially polygynous so there is intense competition between males for access to females, and those males that maintain high energy and protein levels are likely to have the best chance of multiple matings. Hares also avoid foraging in open habitats, however, and while this reduces their risk of predation they incur a cost in terms of the encounter rates of nutritionally valuable plants within these areas (chapter 3).  Although there is an abundance of plant species available to hares during the summer, the majority are nutritionally poor or heavily laden with defensive compounds. Hence, no  103  single plant species is likely to supply an individual with all of its nutritional requirements without the hare also ingesting toxic levels of a defensive compound. Here I develop four hypotheses to predict the diet of breeding male snowshoe hares and use a linear programming technique to solve for the optimum balance of prey items that vary in nutritional content and toxicity, given a particular objective function (see below). The hypotheses are that breeding male snowshoe hares select a diet that serves to: 1. Minimize time spent foraging (the time minimization hypothesis) 2. Maximize energy intake (the energy maximization hypothesis), 3. Maximize protein intake (the protein maximization hypothesis), and 4. Minimize the intake of secondary defensive compounds (the secondary compound minimization hypothesis).  THE DIET SELECTION MODEL The model structure is based upon the foraging models developed for moose (Belovsky 1978) and hares on Isle Royal, Michigan (Belovsky 1984). Belovsky's models were constrained by a minimum intake of sodium. Sodium is not thought to be limiting in Kluane (Schmitz et al. 1992) and this constraint has been removed. Hares are thought to maintain low protein reserves (Whittaker and Thomas 1983) so a new constraint was constructed (minimum daily protein requirement). A series of adjustments was also made to accommodate differences between the Yukon hare population and the southern hares of Isle Royale. These include a plant defensive compound limit, and time costs associated with foraging which depends on different plant species distribution and abundance. My model considers plant species individually (as opposed to functional groups; see Belovsky  104  1984) and only those species that comprise greater than 5% of the observed diet are included; Betula glandulosa, Festuca altaica, Lupinus arcticus, Salix spp. and Shepherdia  canadensis. The basic structure of the model is developed here and is subsequently used to predict the composition of hare diets at two times during the summer (early June and late July).  1) Maximum Digestive Capacity Constraint The model assumes that there is an upper limit to the amount of food a herbivore can process during a day and this limit is set by the capacity of the digestive system and the turnover time of food within the gut. The equation was adapted from Belovsky (1984) but modified with values for the size of the digestive organs of hares in the Yukon. Calculation of the digestive constraint equation has 3 components:  i)  Digestive capacity, scaled to body mass of Yukon hares, is given as:  Content limit (g -wet weight) = 5.524W  04171  (1)  where all parameters are defined in Table 4.1.  ii)  Turnover time of food in the digestive system. Belovsky (1984) suggested that the  turnover time during the summer months was 3.5 hr. I incorporated a more conservative estimate of 4 hr.  105  The maximum digestive capacity (grams of wet weight) can now be calculated from i and ii:  Maximum digestive capacity = (24 hr d' / 4 hr) (5.524 W °- ) 1  4m  (2)  I am assuming that the volume occupied by each plant species in the diet is set by the amount ingested and the bulk of the plant matter, as was also done by Belovsky (1984).  iii)  Plant bulk values are the inverse of the water content (Table 4.2). Thus the bulk of  the diet can be calculated as:  Volume of ingested food = £ Bi dw;  (3)  The digestive constraint then becomes  maximum digestive capacity > volume of ingested food  (4)  When the five plant species characters bulk volumes are inserted, the equation becomes:  (24 hr d" / 4 hr) (5.524 W 1  0 4 1 7 1  ) > 3.4b + 3.7f +7.11 +4.3s + 3.6sh  (5)  106  Table 4.1: Definitions of parameters used in model equations Parameter  Definition  Values taken from  b f 1 s sh  Betula glandulosa Festuca altaica Lupinus arcticus Salix spp. Shepherdia canadensis  consumption of each species predicted by model  W  weight of hares dry weight of species i total dry weight of all species energy content of species i digestibility of species i protein content of species i tannins concentration of species i bulk of species i  Chapter 2 hare weights Chapter 2 enclosures Chapter 2 enclosures Chapter 2 nutrient analysis Chapter 2 nutrient analysis Chapter 2 nutrient analysis TakenfromJung and Batzli 1980 Chapter 2 nutrient analysis  dw\ tdw Ei Di Pi ti Bi  In the late summer, when the water content of the leaves declines, the bulk values change and the equation becomes:  (24 hr d" / 4hr) (5.524 W 1  04171  ) > 3.2b + 3.2f + 6.61 + 3.7s + 3.7sh  (6)  2) Feeding Time Constraint During the summer, hares partition their time amongst a number of activities such as feeding, reproduction, scanning for predators, and movements between feeding sites. Thus there is an upper limit to the time available for a hare to feed during a day. Using an approach based upon thermal energy demands, Belovsky (1984) calculated a total daily activity time during the summer of 8.7 hr (but see critique by Ferron and Ouellet 1992). Using fecal pellet counts as an estimate of hare activity, Hodges (1998) calculated that hares are active 16 hr per day during the winter . Using this method, I determined that during the summer, hares were active an average of 12.5 hr/day (± 0.9 hr) in 1997 and 16 hr/day (± 0.6hr) in 1998 (Fig 4.1). It is unclear why the two estimates are so different and may be a result of small sample sizes. Of this active time, Belovsky (1984) suggested that hares spend 26% actually foraging whereas my observations indicated that in Kluane, hares spend about half of their time foraging. The upper limit to the amount of time available for foraging is thus: Maximum time available for foraging = (12.5 hr. active/day)(0.5)  (7)  108  Table 4.2: Mean plant characteristics included in the model. Bulk is calculated as the plant wet weight/ dry weight. Digestible energy is calculated as the digestibility multiplied by the energy content of each species. Time 1 and 2 correspond to the time hares were placed in enclosures (chapter 2) and correspond to early June and late July. Protein Content (% dry weight) 28 21  SE  0.02 0.13  Digestible Energy (D,*E,)kcal/g 2584 2574  SE  Betula glandulosa  1 2  Bulk (ww/dw) 3.35 3.2  Festuca altaica  1 2  3.67 3.23  0.06 0.06  1879 1683  20 11  2.3 0.2  Lupinus arcticus  1 2  7.05 6.65  0.22 0.12  3124 2776  35 21  1 1.4  Salix spp.  1 2  4.3 3.67  0.09 0.11  2517 2103  33 16  2 0.8  Shepherdia canadensis  1 2  3.58 3.71  0.02 0.07  2533 2719  33 20  1.7 0.3  Plant Species  Time  109  2.1 0.4  During the period of active foraging, hares incur a time cost in consuming each food item. This cost is a function of the encounter rate of a given item and its associated handling time. The time cost associated with consuming each food item has three components: encounter rate, amount of available biomass per encounter, and cropping rate, i)  Encounter rate was determinedfromthe 600 fecal pellet transects (Chapter 3;  Table 4.3). This rate was scaled to values between 1 and 2 so that a relative decrease in abundance could be assigned an increased time cost and is calculated as follows:  Encounter rate = 1 + (proportion of stations where species i is absent)  ii)  (8)  The relative proportion of biomass for each plant species within each of the  enclosures (Chapter 2), was calculated as an estimate of the mean amount of edible biomass within reach of the hare (Table 4.3). Values were assigned on a relative scale from 0 to 1 as follows:  Biomass available = (1- dwj/ tdw)  (9)  Thus species with low biomass incur high time costs.  110  Figure 4.1: Mean (± 1 SE) hourly fecal deposition rate by hares. A. 1997 (n- 4) and B. 1998 (n=8). Time on the x-axis of both graphs begins at 1 lam.  Ill  B.  70  60  A  11  12  1  2  3  4  5  6  7  8  9  10  T i m e of  11  12  1  2  3  4  5  6  7  8  9  10  Day  112  iii)  Cropping rate was determinedfromobserving caged hares and measuring the  amount of time taken to ingest a particular food item of known weight. Food items of the same size and mass were dried and weighed and cropping times then converted to grams dry weight for each species (Table 4.3) using the following equation:  Cropping rate = dw; ingested / time required to ingest item  (10)  The structure of the time constraint equations becomes: Time active * Proportion of time spent foraging > ^(encounter rate * abundance)dwj + Z (cropping rate)dwj  (11)  When the values are inserted for each species (Table 4.3) the equation becomes:  (12.5 hr.)(0.5) > 1.51b + 0.71f + 1.341 + 0.92s + 1.21sh + 0.82b + 3.69f + 1.061 + 0.82s + 0.82sh  (12)  113  Table 4.3: Time related parameters included in the time constraint equation in the model. Cropping rate was calculated as the mean over 20 samples for each of three groups: shrubs, herbs, and grasses.  Plant Species  Betula glandulosa Festuca altaica Lupinus arcticus  Biomass Cropping Rate Encounter Rate min/g ( 1-proportion of (1- abundance) stations where species i absent) 0.9 0.97 1.7 3.7 0.41 1.17 1.1 0.89 1.55  Salix spp. Shepherdia canadensis  SE  0.1 0.5 0.1  1  0.94  0.9  0.1  1.36  0.96  0.9  0.1  114  3) Minimum Daily Energy Requirement A male hare requires some minimum level of energy intake to satisfy its basic metabolic requirements. The daily metabolic rate for snowshoe hare males during the summer can be calculated from 230.8 W° kcal/kg (Hart et al. 1965, Belovsky 1984). 75  Energy intake is determined as a function of the energy content of the leaf tissues, the digestibility of those tissues (values in Table 4.2) and the amount of each species ingested. The energy intake of the diet is calculated:  Energy Intake = E(E * D * dw.) ;  ;  ;  (13)  Clearly, the energy content of the diet must be greater than basic metabolic demands thus the constraint in early summer can now be written with the plant species:  230.8 W > 2584b + 1878f+ 31241 + 2517s + 2532sh 75  (14)  The energy content of plant tissues declines through the summer, thus in the late summer, the energy constraint is calculated:  230.8 VV > 2574b + 1683f+ 27761 + 2104s + 2719sh 75  (15)  115  4) Minimum Daily Protein Requirement The minimum daily protein requirement of snowshoe hares is estimated to be between 11% and 16% of the total dry mass of food ingested (Sinclair et al. 1982, 1988). I therefore used an intermediate value of 14%. The daily protein requirement for hares is:  Protein Intake > 0.14 t W i  (16)  Defensive compounds such as tannins are common to many plant species and may reduce the ability of herbivores to digest protein (Sinclair et al. 1984, Bernays et al. 1989, Dearing 1997). The protein content of each species containing tannins was therefore reduced in proportion to its estimated tannin concentration, according to the following equation:  Adjusted Protein Content - (1 -ti)(Pi)  i)  (17)  The protein content of each plant species was determined at two sampling times  early June and late July (chapter 2, values in Table 4.2).  ii)  Estimates of concentrations of defensive compounds were taken from Jung and  Batzli (1980), and for Shepherdia canadensis, from Ayres et al. (1997) (Table 2.10). I assigned relative concentrations of defense compounds that correspond to the scale of precipitate. Where -m-'s is converted to a concentration of 0.3 and + is converted to  116  0.1 units of defense compound. Betula glandulosa is recorded as having a concentration of only 0.3 units of tannins. This value was increased to 0.5 for the early summer period to reflect the thick resin coating it possesses on its leaf surface, and that may deter hares from feeding on this species at this time. By late summer the resin had evaporated and the value of 0.3 was used for the tannin concentration.  When the species specific parameters (Table 4.2) are inserted the protein constraint early in the summer can be written as:  0.14tw < 0.28b*t + 0.2f*tf+ 0.351 + 0.31s*t+ 0.33sh*U b  s  (18)  In late summer, as plant qualities change, the protein constraint becomes:  0.14tw < 0.20b*t + 0.1 If *t+ 0.211 + 0.16s*t, + 0.20sh*U b  f  (19)  5) Minimizing Plant Defensive Compounds The consumption of Lupinus alkaloids has been shown to inhibit the growth of juvenile rabbits (Johnston and Uzcategui 1989). In addition, large amounts of tannins can inhibit mammalian protein digestion (Mould and Robbins 1981, Sinclair et al. 1984, Robbins et al. 1989, Beraays etal. 1989, Dearing 1997), and also have toxic effects on lagomorphs (Dearing 1997). Therefore, I set an upper limit to the amount of plant defensive compounds that can be ingested because there are physiological limits to both the amount and rate at which these compounds can be detoxified (Reichhardt et al. 1984, Bryant et al.  117  1985, 1989). Unfortunately, it is difficult to determine precisely the concentration of defensive compounds for a given plant species that induces a measurable toxic effect within a particular species of herbivore (Bryant etal. 1985, 1992). Furthermore, all plants contain some defensive compounds and hares are likely to have evolved some physiological tolerance, at least to the more common ones (tannins, for example).  I set an upper limit to the ingestion of alkaloids and tannins separately, and a limit to total ingestion, each of which represents an approximation of the physiological tolerance of hares. The limit was set at (i) 20% dry weight of the total weight of food ingested for alkaloids, (ii) 30% dry weight for tannins, (iii) and the total amount of defensive compounds in the diet was limited to less than 25% dry weight ingested..  Constraints on the intake of defensive compounds are therefore written: i) Alkaloids 0.2 tw> 0.1b+ 0.31 +0.1s  (20)  0.3 tw> 0.5b+ 0. If + 0.3s+0.3sh  (21)  0.25 tw > Z alkaloids + tannins  (22)  ii) Tannins  iii) Total  Note that the concentrations of tannins and alkaloids associated with each species are expressed on a relative scale of activity (from Jung and Batzli 1979; Table 2.10).  118  The late summer model was modified during the late summer by altering the Shepherdia canadensis tannin concentration to 0.5 to reflect the hares' avoidance of this species at this time.  Model Objectives Four objective functions were examined in the model: 1. 2. 3. 4.  Minimize time spent foraging (time minimization), Maximize energy intake per unit time spent foraging (energy maximization), Maximize protein intake (protein maximization), Minimize the intake of secondary defensive compounds (secondary compound minimization).  The diet is predicted for each hare that was placed in an enclosure at each sampling time (chapter 2). Thus for each individual, we use physiological constraints specific to that hare, and plant availabilities that correspond to its enclosure. The model can then be used to predict that amount of each prey type that would be consumed to satisfy each of the four potential foraging goals. This is done by solving thefiveconstraint equations simultaneously in accordance with the objectives outline above (see Belovsky 1978 for further details). Predictions were made at two time periods (early June and late July) during the summer.  The predicted and observed diets were then compared (chapter 2). To assess the fit of one model relative to an alternative, the following sum of squares were then calculated for each time period:  119  Sum of Squares = I (observedy - predictedy )  2  (23)  where the diets are summed over j hares and i food items.  Sensitivity Analysis A sensitivity analysis was conducted to identify the equations or parameters to which the model's predictions were most sensitive. The general approach I used was to sequentially alter each parameter by 25% of its value and then solve the model to examine how this altered the models' predictions once the new optimal solution was then identified. Only model equations and parameters were examined; other plant characters such as protein content, water content, and abundance, that have errors associated with their measurements were not altered. The constraint equation was modified as follows: i)  The total secondary compound limit was altered from 0.25tw to 0.19tw for the lower limit and to 0.3 ltw for the upper limit.  ii)  The model's sensitivity to the digestive constraint was examined by changing the turnover ratefrom4 hr to 3 hr (to increase digestive capacity) and 5 hr (to decrease digestive capacity).  iii)  The sensitivity of the model's predictions to tannin ingestion was examined by altering the limit of 0.3tw to an upper limit of 0.375tw, and to a lower limit of 0.225tw.  iv)  The sensitivity of the models' predictions to alkaloid ingestion was examined by altering the limit of 0.2tw to an upper limit of 0.25tw, and to a lower limit of 120  0.15tw. Because of the large difference in time costs of foraging on different plant species, the 25% alteration was insufficient to examine the sensitivity of the model to this constraint. Therefore, the constraint was altered such that the time costs became so high that a plant species was excludedfromthe diet, or such that there was no time cost at all to ingesting a particular plant species, or specifically: v)  The sensitivity of the model to the time costs of individual species was examined. A high time cost was applied to each species by altering the (encounter * availability) function to a value of 5dw; as the upper limit, and to 0.2dw; as the lower limit.  Because of the large error associated with the estimates of active foraging time for the hares, the model was examined without this constraint. vi)  The time constraint was removed.  After each separate modification, the model was again solved for each objective function and the difference between the observed and predicted diets calculated using the sum of squares. The sums of squares between the original and modified models were then compared.  RESULTS Initially the model was solved using the equations and parameters of Belovsky (1984), modifying only the plant nutritional valuesfromhis original model (Table 4.4). The model consistently predicted a diet composed exclusively of shrubs as the optimal solution for  121  both the time minimizing and energy maximizing objectives (Table 4.4). The constraint equations were subsequently modified with parameters specific to the Kluane snowshoe hare population by altering the digestive capacity and cropping rate. Again, the model indicates that a diet composed of 100% shrubs is the optimal solution for both goals (Table 4.4). Both attempts are completely unsuccessful in predicting the observed diet.  When the new model, developed in this chapter, was solved to predict the optimal diet, the energy maximization function best predicted the observed diet during early summer (Fig 4.2), although the protein maximization solution provides a closefit.This likely reflects the fact that protein and energy content are positively correlated in the plant tissues included in the model (Fig 2.8) and thus provide similar solutions. In late summer the male hares appeared to select a diet that maximized protein intake, although again the energy maximization solution provides a very similar result (Fig. 4.3). When the Shepherdia canadensis constraint was modified late in the summer, thefitof the model improved and the protein maximizing function again provides the best predicted diet (Fig 4.4), but as with the other result energy maximization is also a close fit.  It is useful to look at the model's solutions in terms of the objective values. The results from early summer show the model's optimal solutions for time minimization (A), energy maximization (B) and an intermediate strategy that simultaneously optimizes both constraints (C) (Fig 4.5a). This is compared with the mean observed hare diet at that time.  122  Table 4.4: Comparison of predicted and observed diets using Belovsky's (1984) model. The modified Belovsky model includes parameters for cropping rate and digestive capacity that have been calibrated for Yukon snowshoe hares.  Model  Plant Group  Timel Time 2 Predicted Observed Predicted Observed  Belovsky i) Energy maximized Shrubs Herbaceous Grass  1 0 0  0.36 0.36 0.28  1 0 0  0.32 0.49 0.19  Shrubs Herbaceous Grass  1 0 0  0.36 0.36 0.28  1 0 0  0.32 0.49 0.19  Modified Belovsky i) Energy maximized Shrubs Herbaceous Grass  1 0 0  0.36 0.36 0.28  1 0 0  0.32 0.49 0.19  Shrubs Herbaceous Grass  1 0 0  0.36 0.36 0.28  1 0 0  0.32 0.49 0.19  ii)Time minimized  ii)Time minimized  123  In the early summer, the observed diet falls along the time minimizing - energy maximizing trade-offline which suggests that, although hares are maximizing energy intake, they might also be balancing a conflicting time goal. A similar graph for the late summer reveals that hares fall along the protein maximizing - time minimizing trade-offline and are close to the simultaneous maximization point (Fig 4.5b).  To verify the success of the model, the observed diets and the weight of the hares were both programmed into the model. This facilitated a detailed analysis of the models' constraints so that I could identify the constraints that did not apply to snowshoe hares in the Kluane region (Table 4.5). The constraint equation for digestive capacity was violated most frequently during the entire season, and the overall secondary compound constraint is too limiting for hares in the late summer.  Resultsfromthe sensitivity analysis indicate that the model is particularly sensitive to changes in the secondary compound limit early in the summer and both digestive capacity and time spent foraging in late summer (Table 4.6). The model is most sensitive to high time costs associated with both Lupinus arcticus and Shepherdia canadensis early in the summer (Table 4.7). The sensitivity analysis of low time costs on the late summer model indicates that the predictions are most variable in response to changes in time costs, particularly of Shepherdia canadensis and Festuca altaica.  124  Figure 4.2: Optimization solutions for the model during the early summer. SS is the sum of squares when comparing the predicted and observed diets. Each point represents a different species. Error bars are ± 2SE, calculated using data from 9 hares. The line has a slope of 1 and indicates a one to onefitbetween observed and predicted diets.  125  A. Time Minimized SS - 2.54 0.60 0.50  Lupinus  0.00  0.10  0.20  0.30  0.40  0.70  0.60  0.50  0.80  Mean PredictedTime Minimized  B. Energy Maximized SS = 1.57 0.60  n  0.50  A  0.00  0.05  0.10  0.15  0.20  0.25  0.30  0.35  0.40  0.45  0.50  Mean Predicted Energy Maximized  126  C. Protein Maximized SS = 1.71  D. Secondary Compound Minimized SS = 2.53 0.60  T  0.50 ]  0.00  0.10  0.20  0.30  0.40  0.50  0.60  Mean Predicted Secondary Compound Minimized  127  Figure 4.3: Optimization solutions for the model during the late summer. SS is the sum of squares when comparing the predicted and observed diets. Each point represents a different species. Error bars are ± 2SE, calculated using datafrom12 hares. The line has a slope of 1 and indicates a one to onefitbetween observed and predicted diets.  128  A . Time Minimized SS = 5.28  B. Energy Maximized SS = 5.11  0.6!  Mean Predicted Energy Maximized  129  D. Secondary Compounds Minimized SS = 3.30  130  Figure 4.4: Optimization solutions for the adjusted model during the late summer. SS is the sum of squares when comparing predicted and observed diets. Each point represents a different species. Error bars are +2SE, calculated using datafrom12 hares. The line has a slope of 1 and indicates a one to onefitbetween observed and predicted diets.  131  A. Time Minimized SS = 2.30  B. Energy Maximized SS = 1.76  132  C. Protein Maximized SS= 1.70 0.6  0  0.1  0.2  0.3  0.4  0.5  0.6  0.7  Mean Predicted Protein Maximized  D. Secondary Compound Minimized SS = 1.76  133  Figure 4.5: The mean characters of the hare diets are compared with the predictions of the model during both times. The points on the early summer are as follows; Point A is the time minimized solution, B. is the energy maximized solution, and C is the multi-objective solution for simultaneously maximizing energy while minimizing time foraging. During the late summer point B is the protein maximized solution.  134  A. Early Summer  •Predicted • Observed  200  600  500  400  300  Time (min)  B. Late Summer  25 T  23  h-f l - H ^ .  '  3 21 § 19  I 17  A  ^  B  c  •Observed  15 J 200  300  400 Time (min)  • Predicted  500  600  Table 4.5: Summary of constraint equations violated by observed hare diets. Time lvalues are the sum of violations over 9 animals and the time 2 results are tabulated over 12 animals. Time 1 corresponds to early June and time 2 corresponds to late July.  Constraint Equation  Time 1 Time 2 Number of Violations Number of Violations  Alkaloids Digestive Constraint Energy Protein Secondary Compounds  2  1  3  3  1  0  0  2  0  4  Tannins  0  0  136  Table 4.6: Summary of results from sensitivity analysis. The sum of squares values was calculated as the deviation over all hares observed at each time. Time 1 and 2 corresponds to early June and late July.  Timel  Parameter Changed  New Parameter Value Model with Best Solution  Sum of Squares  Alkaloids  0.15 0.25  Time Minimized Energy Maximized  1.6 1.6  Tannins  0.23 0.34  Energy Maximized Energy Maximized  1.6 1.6  Total Secondary Compound limit  0.19 0.31  None Secondary Compound Minimized  2.6  Turnover Time  3 5  Time, Protein, Energy Time Minimized  1.6 1.5  Time Minimized  1.6  Time limit removed  Time 2  Parameter Changed  New Parameter Value Model with Best Solution  Sum of Squares  Alkaloids  0.15 0.25  None Energy/ Protein Maximized  1.8  Tannins  0.23 0.34  Energy/ Protein Maximized Energy/ Protein Maximized  1.8 1.8  Total Secondary Compound limit  0.19 0.31  None Energy/ Protein Maximized  1.8  Turnover Time  3 5  Time, Protein, Energy None  2.3  Time Minimized  2.3  No time limit  137  Table 4.7: Summary of sensitivity analysis on time costs associated with individual species. High time costs indicate the cost function for each species was increased to 5dw and for low time costs the cost function was decreased to 0.2dw; of species i. ;  Time 1 Parameter Changed High Time Costs  Model with Best Solution Sum of Squares  Betula glandulosa Time Minimized Time Minimized Festuca altaica Lupinus arcticus Energy Maximized Salix spp. Time Minimized Shepherdia canadensis Time Minimized  1.95 1.95 2.99 1.61 2.26  Low Time Costs Betula glandulosa Festuca altaica Lupinus arcticus Salix spp. Shepherdia canadensis  Time 2 Parameter Changed High Time Costs  Time Minimized Time Minimized Time Minimized Time Minimized Time Minimized  Model with Best Solution Sum of Squares  Betula glandulosa Defence minimized Festuca altaica Time Minimized Lupinus arcticus Defence minimized Time Minimized Salix spp. Shepherdia canadensisDefence minimized  Low Time Costs  1.95 1.95 1.27 1.27 1.15  Betula glandulosa Time Minimized Festuca altaica Time Minimized Lupinus arcticus Time Minimized Salix spp. Defence minimized Shepherdia canadensis Time Minimized  3.16 2.43 3.99 2.47 2.25 3.63 4.08 2.47 3.59 4.08  138  DISCUSSION Belovsky Model The original model proposed by Belovsky (1984) for snowshoe hares foraging during summer was not successful in predicting the diet of this more northerly population of males. There are several possible explanations for this result. First, Belovsky's model required the inclusion of a variety of plant species to satisfy both a hare's sodium and energy requirements. Thus the consumption of each group involved a tradeoff in terms of either energy or sodium intake. This contrasts with the diet considered here in that a snowshoe hare can satisfy all of its minimum requirements (of protein and energy in this model) by foraging on a single group (shrubs). Hence, the predicted diet was determined by digestive and time constraints, and because the leaves of shrubs are both the fastest items to consume and the least bulky, they are always the optimal solution regardless of which single objective function is used. Second, Belovsky's model did not include constraints that are important to this particular northern hare population. An example of this in my case is the presence of toxic secondary metabolites that have an influence upon prey selection (Chapter 2). Third, Belovsky used time costs that depended only on the speed of ingestion of a given prey type rather than its distribution and abundance relative to all prey types. This approach seems to have been applicable to Isle Royale, however, the large difference in the rate of ingestion in Kluane caused a large bias against all groups other than shrubs, that again facilitates the single group dominance in the model predictions. When the plants are treated as groups, it is reasonable to assume equal accessibility of prey types. However, in the new model, the extremely skewed  139  distributions and abundance of the prey types in Kluane incurs a time cost that must be addressed.  The early summer diet The energy maximizing function best explained prey selection in early summer although the protein maximizing solution also provided a closefit.This result is consistent with previous models of hare diets (Belovsky 1984, Schmitz et al. 1992), even though the time minimizing strategy was predicted for male herbivores during the breeding season (Belovsky 1986). By minimizing time spent foraging, males can maximize the time available to pursue females. Hares maintain very small energy and protein reserves that are sufficient to maintain activity for only a short time (Whittaker and Thomas 1983). Given, the short reproductive season and that breeding chases represent brief periods of intense physical activity, maximizing energy intake while foraging allows hares to maintain breeding activity for longer periods before foraging is necessary to restore nutritional reserves.  When the diet is examined form the perspective of the constraint units this suggests that hares are simultaneously trying to balance two conflicting demands. Although, energy maximization is the main foraging goal early in the summer, the observed diet falls short of the potential energy maximization point. This suggests that hare energy intake in their diets may be limited by some other factor that was not identified in this model or that they might be maximizing energy while balancing some other conflicting goal such as time. This would be consistent with the ideas put forward by Hik (1993, 1995) that hare foraging  140  behavior is sensitive both to food quality (energy) and distribution (time).  The late summer diet The failure of the initial model to predict the hare diet in late summer is due largely to the hares' avoidance of Shepherdia canadensis. Why hares avoid this species in late summer is paradoxical since it has high nutritional value and requires relatively little time to consume. One possibility is that there is a change in the concentration of its defensive compounds. The twigs of Shepherdia canadensis are defended during the winter with high concentrations of tannins (Ayres et al. 1997, Sinclair et al. unpublished data) and during the summer, the berries contain high concentration of saponins (Willard et al. 1992). However, little is known about the defense allocation to the leaves (that hares consume during the summer) or how this allocation changes seasonally. This needs to be investigated.  When the Shepherdia tannin concentration was increased in the model, and the Betula glandulosa  tannin concentration decreased (to reflect the fact that resin no longer coats  the leaf surfaces of this latter species), its predictions were much closer to the observed diet. This is consistent with the results from chapter 2 which indicate that hares are selecting plant species predominantly because of their high protein content and that the plant defense compounds may alter species selection within the protein maximizing objective. Personalfieldobservations support the change in the value of the parameters for the Betula glandulosa defense compounds (change in resin coating on leaf surfaces). The model suggests there is a similar change in the Shepherdia canadensis defensive allocation  141  that needs to be examined.  Although protein maximization is the dominant strategy used by hares, again when examined from the perspective of the constraint units, the observed diet falls short of the potential protein maximization diet. This suggests that hare foraging involves a protein time tradeoff which is consistent with the suggestion of Hik (1995) that hares forage in ways that both minimize their time exposed to predators and maximize their protein gain. The consistent discrepancy between nutritional and time goals should be examined. This suggests that the summer diet selection by hares is sensitive to either predation risk or that the model is overestimating the maximum nutritional goals.  The solutions for the protein and energy maximizing objectives are similar in all three models. This likely reflects the fact that protein and energy are positively correlated within these plant tissues which suggests that maintaining both equations may in fact be redundant. It is possible that hares alter their selection between these two, although it is equally possible that hares are selecting food items on overall plant quality (i.e. both protein and energy), which we are artificially able to distinguish in the model.  General Applicability of Model The model has identified several issues that require further investigation. The main weakness in the model structure revolves around secondary compounds. The major deficiency in a large majority of diet selection studies centers around plant defensive compounds and their effects on herbivores. Although, general patterns have been  142  identified for the effects of tannins on herbivores, the majority of the work for boreal plant species concentrates on shrub twigs, and not leaves, and little is known about temporal variance in the concentrations in the tissue. Further work on secondary compounds is required, particularly for Shepherdia canadensis and the Salix genus.  This model concentrated on male hares to avoid the complications of changing female reproductive demands in the modeling. Because the females were pregnant and lactating during both times, both their energy and protein demands would be highly variable. I would expect that during this time, females would forage as protein maximizers. An examination of the female diet during the summer would provide a good test of the model, particularly of the physiological constraint equations because during this time the nutritional demands on the females are higher.  All of the hypotheses for winter diet selection have been included in this model, thus a winter trial with hares, both male and female, would provide an adequate test of the broader applicability of the model. This would be particularly useful in examining the nutritional minima set by the model, because of the reduced plant quality when hares begin foraging on the twigs of the deciduous shrubs. During the winter all of the plants available to hares are defended with tannins, so this would provide an opportunity to explore the tannin maxima set by the model. Studies of hares during the winter show that hares forage as energy maximizers (Belovsky 1984) and differentiate between secondary compound concentrations (Schmitz et al. 1992), thus it would be useful to examine the whole winter diet in a multi-species context.  143  The most useful product of this model is in applying these equations to the species that are excluded from the summer diet. The simple approach of simultaneously examining several plant characters with hare physiological constraints can reveal why certain species are not included in hare diets. This will be examined in chapter 5.  Model Sensitivity The model is sensitive to the digestive capacity constraint, which is itself dependent on the turnover time within the digestive organs. The value for the turnover time was taken from Belovsky (1984), that was measuredfromthe passage time through the gut of hares ingesting cotton string, a method likely to underestimate passage time and thus increase the digestive capacity.  Neither the early nor late summer models were sensitive to the limit set for tannins, though both were sensitive to decreases in the alkaloid constraint. The number of violations of the alkaloid constraint at both times suggests that snowshoe hares can ingest a large quantity of alkaloids without incurring negative consequences. It should be noted that the species of snowshoe hare studied here has been shown to tolerate higher concentrations of tannins in both Betula and Salix species than their European relatives, the mountain hares (Lepus timidus) (Bryant etal. 1989).  Assumptions Several assumptions are included in this model (Table 4.8). Many of these have some  144  evidence to support them and the appropriate references are listed. The digestive constraint has been criticized (Hobbs 1990,Owen-Smithl996) because of the assumptions of constant turnover rate and the digestive volume limits are dependent on water content. Although this may be challenged by digestive physiologists, it is sufficient in this context. The linear approach has been criticized for being too simplistic and circular (Owen-Smith 1993, 1994,1996). Models are frequently criticized in this context because of being oversimplistic, and although I have acknowledged the potential sources of weakness, this model may be deconstructed into its components and improved.  Summary Overall, the model is successful in integrating the conflicting hypotheses of hare foraging. The most significant resultfromthe model is that the hares' diet selection is sensitive to conflicting demands. Hares change their diet selection in response to conflicting goals such as reproductive condition and risk of predation. Within each of the behavioral strategies, diet selection is consistent with the hypothesized trade-off between energy content, protein content and plant defense compounds.  145  Table 4.8: Assumptions of the diet selection model  Digestive Constraint (Belovsky  1984) 1 Digestive capacity is limited by the size of the stomach and caeca.  2 Plants with greater water contents occupy more space 3 All plant species turnover in the digestive organs at the same rate  Time constraint 1 Fecal deposition indicates level of hare activity (Hodges 1998). 2 Hares encounter plant species in proportion to their abundance. 3 Hares are not territorial  Energy Constraint 1 Digestible energy of plant tissues is diminished byfibrecontent.  Protein Constraint 1 Protein digestion is diminished by tannin concentration  Secondary Compound Constraint 1 Hares have upper limits to the amount of defense compounds they can tolerate (Bernay etal. 1989, Dearing 1997). 2 Effects of plant metabolites in the same group are cumulative (Bernay et al. 1989).  Assumptions relevant to whole model 1 Thefivespecies included in model comprise the majority of the summer diet of hares. 2 Assumes linear interactions.  146  CHAPTER 5  SUMMARY OF DIET SELECTION OF MALE SNOWSHOE HARES DURING THE SUMMER  In this thesis, I have tested three main hypotheses of hare foraging that pertain to 1) plant nutritional characteristics, 2) defensive chemical compounds and 3) spatial distribution of plants. The general approach was to examine each factor independently in chapters 2 and 3 and subsequently as interacting factors in a foraging model (chapter 4).  Hares select plants with high protein content in early summer. Betula glandulosa is an exception, because leaves are coated with a high-tannin resin in early summer. Late in the summer the protein content of the plants was almost sufficient to predict which plants would be included in the diet except at this time Shepherdia canadensis is significantly avoided, despite its high protein content. Clearly something deters the hares from browsing Shepherdia canadensis during this period, probably a higher lever of defensive chemical compounds. The ability to identify patterns in diet selection with ad hoc explanations for aberrant results is typical of the difficulties in understanding plantherbivore dynamics. In essence this result invokes the question of how the concentrations of individual plant chemicals change through the growing season and how they affect selection by hares. This requires further investigation for a more complete understanding of diet selection.  147  In both early and late summer the prey options available to the hares were i) plants with high protein and high concentrations of defensive compounds or ii) plants with low protein and low concentrations of defensive compounds. While it appears that diet selection in winter may be influenced primarily by plant defensive compounds (Bryant et al. 1992), I suggest that selection for high protein content accounts for the majority of the variation observed in the summer. If hare diet selection during the summer was influenced most by plant defense chemicals we would predict that hares would select species with low protein and low concentrations of defensive compounds. The only species actually includedfromthis low protein group was Festuca altaica. Although Festuca comprised a large proportion of the diet, the inclusion of this species reflects the abundance of the biomass available to hares, because this species was actually avoided by hares (i.e. hares consumed proportionally less Festuca than they encountered), throughout the summer. Clearly the hares are not adopting a defense minimization foraging strategy. The failure of hares to forage in this way could be interpreted two ways: 1) hare diet selection is not influenced by plant defense chemicals or 2) the negativefitnessconsequences of foraging on plants within the low protein - low defense group are high.  There is ample evidence to reject thefirstinterpretation that hare browse is not deterred by plant chemicals (i.e. Bryant 1981, Bryant et al. 1983, Sinclair and Smith 1984, Reichardt et al. 1984,Clausene/a/. 1986, Sinclair et al. 1988, Jogia etal. 1989, Reichardt et al. 1990a, 1990b, Schmitz et al. 1992, Williams et al. 1992). It can be demonstrated that thefitnesscosts associated with the secondary compound minimization diet are extremely high in the context of the model. Those plant species are predominantly herbaceous and as  148  such have high water content, which limits digestive capacity (Fig. 2.10). The majority of these species incur high time costs because of their slow ingestion rates and low relative abundance (Tables 2.1a, 4.2). Therefore, the net diet composedfromthese minimally defended species, would increase risk of exposure, by increasing foraging time and decreasing net gain due to low protein and energy content. Ultimately this strategy falls into what Schmitz et al. (1998) refer to as the non-dominated set of solutions and would result in an extremely inefficient and evolutionarily unfavorable diet.  Although plant defensive chemicals are not the primary determinant of hare diet selection, they are an important secondary consideration. Within the protein maximizing strategy, hares avoid plants with extremely high concentrations of tannins, and consequently, are often avoiding plants with high protein content. However, this is also consistent with the protein maximizing goal, because tannins are thought to inhibit protein digestion. This was demonstrated by their avoidance of Betula glandulosa early in the summer and is suggested by their avoidance of Shepherdia canadensis late in the summer, both highprotein species.  The results of this research suggest that diet selection by male hares is a function of a number of characters of both snowshoe hares and plants. Ultimately, diet selection is an expression of a snowshoe hares' ability to maximize theirfitness.Early summer, when the hares are reproductive, the selected diet appears to maximize their nutritional intake while minimizing their foraging time. This foraging strategy might allow hares to maintain their condition and maximize their breeding success. When reproductive demands are removed,  149  hares appear to be balancing time foraging and protein intake.  Within the context of thefitnessgoals in the model, it appears that plant nutritional characters are the most important determinants of diet selection. However, when the model and the observed diets are examined in terms of the potential objective space, the observed diet falls short of the potential nutritional maximizing goals. Although, the maximization points may be an artifact of high parameter estimates used in the model, the lower values of the observed diets suggests that hares are simultaneously balancing time within these foraging goals. This result is consistent with the ideas of Hik (1993, 1995), that hares forage in ways that maximize their nutritional gain while minimizing their time foraging (exposure). During the summer months, food and cover from predators are not necessarily spatially separated, the spatial distribution and abundance of each forage species are relevant to diet selection, particularly with respect to how they incur time costs in the model.  All three hypotheses examined tested in this thesis were identified as important in diet selection. Hares forage primarily to maximize their nutritional gain of protein and energy, however, selection of food types is modified by both the spatial distribution and abundance of each plant species. Plant metabolites act secondarily to alter potential nutritional gain. All of these plant characteristics need to be examined simultaneously because of conflicting factors in their selection. Particular emphasis is required to understand the effects of the plant defensive compounds.  150  REFERENCES Alldredge, J.R. and J.T. Ratti 1992. Further comparison of some statistical techniques for analysis of resource selection. Journal of Wildlife Managements 56:1-9. Atsatt P.R. and D.J. O'Dowd 1976. Plant Defense Guilds. Science 193:24-29. Ayres, MP., T.P. Clausen, S.F. Maclean, A.M.Redman and P.B. Reichardt. 1997. Diversity of structure and antiherbivore activity in condensed tannins. Ecology 78:1696-1712. Batzli G.O. and C. Lesieutre. 1991. The influence of high quality food on habitat use by arctic microtine rodents. Oikos 60:299-306. Belovsky, G.E. 1978. Diet optimization in a generalist herbivore: the moose. Theoretical Population Biology 14:105-134. Belovsky, G.E. 1981.Food plant selection by a generalist herbivore: the moose. Ecology, 62:1020- 1030. Belovsky G.E. 1984a. Herbivore optimal foraging: a comparative test of three models. The American Naturalist 124:97-115. Belovsky, G.E. 1984b. Snowshoe hare optimal foraging and its implications for population dynamics. Theoretical Population Biology, 25:235-264.  Belovsky G.E. 1986. Optimal foraging and community structure: implications for a guild of generalist grassland herbivores. Oecologia 70:35-52. Belovsky G.E. and O.J. Schmitz. 1991. Mammalian herbivore optimal foraging and the role of plant defenses. In R.T. Palo and C T . Robbins (eds.) Plant chemical defence and mammalian herbivory pp. 1-28. Belovsky G.E. and O.J. Schmitz. 1994. Plant defenses and optimal foraging by mammalian herbivores. Journal of Mammology 74:816-832. Bemays, E.A., G. Cooper and M. Bilgener 1989. Herbivores and plant tannins. Advances in Ecological Research 19:263-302. Bernays, E.A., K.L. Bright, N. Ganzalez and J. Angel. 1994. Dietary mixing in a generalist herbivore: tests of two hypotheses. Ecology 75:1997-2006. Boutin, S. and C.J. Krebs, R. Boonstra, M.R.T Dale, S.J. Hannon, K.Martin, A.R.E. Sinclair, J.N.M. Smith, RTurkington, M.Blower, A.Byrom, F.I. Doyle, C. Doyle, D.Hik, L.Hofer, A. Hubbs, T. Karels, D.L. Murray, V.Nams, M. O'Donaghue, C. Rohner and S. Schweiger. 1995. Population changes of the vertebrate community during a snowshoe hare cycle in Canada's boreal forest. Oikos 74:69-80. Bryant J.P. 1981. Phytochemical deterrence of snowshoe hare browsing by adventitious shoots of four Alaskan trees. Science 213:889-890.  151  Bryant J.P. 1981. The regulation of snowshoe hare feeding behaviour during winter by plant antiherbivore chemistry. In K. Meyers and C D . Maclnes (eds.). Proceedings of the World Lagomorph conference. Guelph University Press, Guelph Canada, pp 720-731. Bryant, J. P. and P.J. Kuropat. 1980. Selection of winter forage by subarctic browsing vertebrates: The role of plant chemistry. Annual review of Ecology and Systematics 11:261-285. Bryant, J.P., G.D. Weiland, P.B. Reichardt, V.E. Lewis, and M.C. McCarthy. 1983. Pinosylvin methyl ether deters snowshoe hare feeding on green alder. Science 222: 1023-25. Bryant, J.P., J.Tahvanainen, M. Sulkinoja, R. Julkunen-Tiitto, P. Reichardt and T. Green. 1989. Biogeographic evidence for the evolution of chemical defense by boreal birch and willow against mammalian browsing. American Naturalist 134:20-34. Bryant, J.P., P.B. Reichardt and T.P. Clausen. 1992. Chemically mediated interactions between woody plants and browsing mammals. Journal of Range Management 45:18-24. Bryant, J.P., P.B. Reichardt, T P . Clausen F,D, Provenza and P.J.Kuropat. 1992. Woody plantmammal interactions, pp.343-470. in Herbivores: Their Interactions with Secondary Metabolites, eds. G.E. Rosenthal and M.B. Berenbaum. Academic Press, Inc., London. Cheeke, P.R. 1983. The significance of fibre in rabbit nutrition. Journal of Applied Rabbit Research 6:103-106. Clark, D.A. 1982. Foraging behaviour of a vertebrate omnivore (Rattus rattus): meal structure, sampling and diet breadth. Ecology 63:763-772. Dale, M.RTand M.W. Zbigniewicz. 1997. Spatial pattern in boreal shrub communities: effects of a peak herbivore density. Canadian Journal of Botany 75: 1342-1348. Dearing, M.D. 1997. Effects of Acomastylis rossii tannins on a mammalian herbivore, the North American pika, Ochontaprinceps. Oecologia 109:122-131. Desy E.A. and G.O. Batzli 1979. Effects of food and predation on density of prairie voles: a field experiment. Ecology 70: 411-421. Douglas, G.W.1974. Montane zone vegetation of the Alsek river region, southwest Yukon. Canadian Journal of Botany 52:2505-2532. Ferron, J. and J-P. Ouellet. 1992. Daily partitioning of summer habitata and use of space by die snowshoe hare in southern boreal forest. Canadian Journal of Zoology 70:2178-2183. Fox, J.F and J.P. Bryant. 1984. Instability of the snowshoe hare and woody plant interaction. Oecologia 63: 128-135. Freeland, W.J., P.H. Calcott, amd L. Anderson. 1985. Tannins and sapponin: interaction in herbivore diets. Biochemical Systematics and Ecology 13:189-193. Freeland, W.J. and D. H. Janzen 1974. Strategies in herbivory by mammals: the role of plant secondary compounds. The American Naturalist 108:269-289.  152  Freeland W.J. and L.R. Saladin 1989. Choice of mixed diets by herbivores: the idiosyncratic effects of plant secondary compounds. Biochemical Systematics and Ecology 17:493497. Freeland, W.J. 1992. Plant secondary metabolites: biochemical coevolution with herbivores. pp.61-81 in Herbivores: Their Interactions with Secondary Metabolites, eds. G.E. Rosenthal and M.B. Berenbaum. Academic Press, Inc., London. Griffiths M. and R. Barker. 1966. The plants eaten by sheep and by kangaroos grazing together in a paddock in south-western Queensland. CSIRO Wildlife Research. 11:145-167. Grisley, K.J. 1990. Summer food preference of snowshoe hares (Lepus americanus). and its implications on juvenile growth. B.Sc. Honors Thesis, University of British Columbia. Hall, L.S., P.R Krausman and M.L. Morrison. 1997. The habitat concept and a plea for standard terminology. Wildlife society bulletin 25: 173-182. Hart, J.S., H. Pohl and J.S. Tener. 1965. Seasonal acclimatization in varying hare (Lepus americanus). Canadian Journal of Zoology. 43:731-744. Hik, D.S. 1993. Predationriskand the 10-year snowshoe hare cycle. Ph.D. thesis. University of British Columbia. Hik, D.S. 1995. Does theriskof predation influence population dynamics? Evidencefromthe cyclic decline of snowshoe hares. Wildlife Research 22:115-129. Hobbs, N.T. 1990. Diet selection by generalist herbivores: a test of the linear programming model. In Hughes R.N. (ed) Behavioral mechanisms of food selection Springer, Berlin Heidelberg New York, pp395-413. Hodges K.E. 1998. Snowshoe hare demography and behaviour during a cyclic population low phase. Ph.D. Thesis. University of British Columbia. Jogia, M.K., A.R.E. Sinclair and R.J. Andersen. 1989. An antifeedant in balsam inhibits browsing by snowshoe hares. Oecologia 79:189-192. Johnson, D.H. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology 61:65-71. Johnston N.P. and M.E. Uzcategui. 1989. The effect of soybean meal, soybeans, bitter lupine, fava bean and peas on the growth and lactation of rabbits. Journal of Aplied Rabbit Research. 12:42-44. Jung H.G. and G.O. Baztli. 1981. Nutritional ecology of microtine rodents: effects of plant extracts on the growth of arctic microtines. Journal of Mammology 62:286-?. Jung, H, G., G.O. Batzli and D.S. Seigler. 1979. Patterns in phvtochemistry of arctic plants. Biochemical Systematics and Ecology 7:203-209. Keith, L.B. 1963 .Wildlife's ten year cycle. University of Wisconsin Press, Madison.  153  Keith, L.B. 1990. Dynamics of snowshoe hare populations, pp.119-195 in Current Mammology, e.d.H.H. Genoways. PlenumPress, New York. Keith, L.B., A.W. Todd, C.J. Brand, RS. Adamcik and D.H. Rusch. 1977. An analysis of predation during a cyclic fluctuation of snowshoe hares. Proceedings of the International Congress of Game Biologists 13:151-175. Keith, L.B. and L.A. Windberg 1978. A demographic analysis of the snowshoe hare cycle. Wildlife Monographs 58, 1-70. Klein D.R. 1977. Winter food preferences of snowshoe hziQs(Lepus americanus) in interior Alaska. Proceedings of the International Congress of Game Biologists 13:266-275. Krebs, C.J., B.S. Gilbert, S. Boutin, A.R.E. Sinclair, and J.N.M. Smith 1986. 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., S. Boutin, R. Boonstra, A.RE. Sinclair, J.N.M. Smith, M.T. Dale, K. Martin and R. Turkington 1995. Impact of food and predation on the snowshoe hare cycle. Science 269:1112-1115. Larter, N.C. 1992. Foragefiberanalyses: a comparison of two techniques. Wildlife Research 19:289-93. Lima, S.L. and L.M. Dill. 1990. Behavioural decisions made under the risk of predation: a review and prospectus. Canadian Journal of Zoology 68: 619-640. Lindlof, B. A. Pehrson and A. Johanson. 1978. Summer food preference by penned mountain hares in relation to nutrient content. Journal of Wildlife management 42:928-932. Lindlof, B. E.Lindstrom and A. Pehrson. 1974. Nutrient content in relation to food preferred by mountain hare. Journal of Wildlife Management 38:875-879. Lindroth, R. 1989. Mammalian herbivore-plant interactions. - In Abrahamson, W.G. (ed.), Plantanimal interactions. McGraw-Hill, New York, pp. 163-206. Lindroth, R.L., G.O. Baztli and S.I. Avildsen, S.I. 1986. Lespedeza phenolics and Penstemon alkaloids; effects ondigestion efficiency and growth of voles. Journal of Chemical Ecology 12:713Litvatis, J.A., Shepburne, J.A. and Bissonette, J.A. 1985. A comparison of methods used to examine snowshoe hare habitat use. Journal of Wildlife Management 49:866-873. Manly, B.J.F., L.L. McDonald and D.L. Thomas. 1993. Resource Selection by animals: statistical design and analysis for field studies. Chapman and Hall, London. Miller, G.R 1968. Evidence for selective feeding on fertilized plots by Red Grouse, hares, and rabbits. Journal of Wildlife Management 32:849:853 Murray, D.L., S. Boutin and M. O'Donoghue. 1994. Winter habitat selection of lynx and coyotes in relation to snowshoe hare abundance. Canadian Journal of Zoology 72:1444-1451. 154  Norbury, G.L. 1988. Microscopic analysis of herbivore diets- a problem and a solution. Australian Wildlife Research 15:51-57. Norbury, G.L. and G.D. Sanson 1992. Problems with measuring diet selection of terrestrial, mammalian herbivores. Australian Journal of Ecology 17:1-7. O'Donoghue, M. 1997. Responses of coyotes and lynx to the snowshoe hare cycle. Ph.D. Dissertation, University of British Columbia. Orr, C D . and D.G. Dodds. 1982. Snowshoe hare habitat preferences in Nova Scotia spruce-fir forests. Wildlife Society Bulletin 10: 147-150. Owen-Smith, N. 1993. Evaluating optimal diet models for and African browsing ruminant, the kudu: how constraining are the assumed constraints? Evolutionary Ecology7: 499-524. Owen-Smith N. 1994. Foraging responses of kudus to seasonal changes in food resources: elasticity in constraints? Evolutionary Ecology 7: 1050-1062. Owen-Smith N. 1996. Circularity in linear programming model of optimal diet. Oecologia 108:259-261. Pease, J.L., R.H. Vowles and L.B. Keith, 1979. Interaction of snowshoe hares and woody vegetation. Journal of Wildlife Management 43:43-60. Pehrson, A. 1981. Winter food consumption and digestability in caged mountain hares. Pages 732-742. in K. Myers and C D . Machines (ed.). Proceedings of the World Lagomorph Conference. Guelph, Ontario. August 12-16. 1979. Guelph University Press, Guelph, Ontario. Petrides, G.A. 1975. Principal foods versus preferred foods and their relations to stocking rate and range condition. Biological Conservation 7:161-169. Pietz, P.J. and Tester, J.R. 1983. Habitat selection by snowshoe hares in north central Minnesota. Journal of Wildlife Management 47:686-696. Pulliam, H.R. 1975. Diet optimization with nutrient constraints. American Naturalist 109: 765768. Pyke, G.H., H.R. Pulliam, and E.L. Chamov 1977. Optimal foraging: A selective review of theory and tests. Quarterly Review of Biology, 52:137-154. Reichardt, P.B., J.P. Bryant, T.P. Clausen and G.D. Wieland. 1984. Defence of winter-dormant Alask paper birch against snowshoe hares. Oecologia 65:1941-1959. Reichardt, P.B., J.P Bryant, B.J. Anderson, D.Philips, T.P. Clausen, M.Meyer and K. Frisby. 1990. Germacrone defends labrador teafrombrowsing by snowshoe hares. Journal of Chemical Ecology. 16:1961-1970. Robbins, C T . 1983. Wildlife feeing and nutrition. Academic Press, New York.  155  Rohner, C. and C.J. Krebs. 1996. Owl predation on snowshoe hares: consequences of antipredator behaviour. Oecologia 108: 303-310. Rodgers, A.R. and A.R.E. Sinclair. 1997. Diet choice and nutrition of captive snowshoe haiQsfLepus americcmus): Interactions of energy, protein, and plant secondary compounds. Ecoscience4: 163-169. Rogowitz, G. L. 1988. Forage qulaity and use of reforested habitats by snowshoe hares. Canadian Journal of Zoology 66:2080-2083. Schmidt, K.A., J.S. Brown and R.A.Morgan. 1998. Plant defenses as complementary resources: a test with squirrels. Oikos 81:130-142. Schmitz, O.J., D.S. Hik, and A.R.E. Sinclair 1992. Plant chemical defense and twig selection by snowshoe hares: an optimal foraging perspective. Oikos 65:295-300. Schmitz, O.J., J.L. Cohon, K.D. Rothley and A.P. Beckerman. 1998. Reconciling variability and optimal behaviour using multiple criteria in optimization models. Evolutionary Ecology 12:73-94. Sievert P.R. and L.B. Keith. 1985. Survival of snowshoe hares at a geographic range boundary. Journal of Wildlife Management 49:854-866. Sinclair, A.R.E. and J.M. Gosline. 1997. Solar activity and mammal cycles in the Northern Hemisphere. American Naturalist 149:776-784. Sinclair, A.R.E. and J.N.M. Smith. 1984. Do plant secondary compounds determine feeding preferences of snowshoe hares? Oecologia 61: 403-410. Sinclair, A.R.E., C.J. Krebs and J.N.M. Smith. 1982. Diet quality and food limitation in herbivores the case of the snowshoe hare. Canadian Journal of Zoology 60:889-897. Sinclair, A.R.E., C.J. Krebs, J.N.M. Smith and S. Boutin 1988. Population biology of snowshoe hares III: Nutrition, plant secondary compounds and food limitation. Journal of Animal Ecology 57:787-806. Sinclair, A.R.E., J.M. Gosline, G. Holdsworth, C.J. Krebs, S. Boutin, J.N.M. Smith, R. Boonstra and M.Dale. 1993. Can the solar cycle and climate synchronize the snowshoe hare cycle in Canada? Evidencefromtree rings and ice cores. American Naturalist 141:173-198. Smith J.N.M., C.J. Krebs, A.R.E. Sinclair and R.Boonstra. 1988. Population biology of snowshoe hares. II. Interactions with winter food plants. Journal of Animal Ecology 57:269-286. Smith R.L., D.J. Hubartt and R.L. Shoemaker. 1980. Seasonal changes in cecal length and pancreatic function of snowshoe hares. Journal of Wildlife Management 44:719-724. Tate, D.F.1994. Determination of nitrogen in fertilizer by combustion: collaborative study. Journal of AOAC International 7:829-839.  156  Westoby M., G.R. Rost and J.A. Weis. 1976. Problems with estimating herbivore diets by microscopically identifying plant fragments from stomachs. Journal of Mammology 57:167-172. Westoby, M. 1978. What are the biological bases of varied diets? American Naturalist 112:627631. Whittaker, M.E. and V.G. Thomas 1983. Seasonal levels of fat and protein reserves of snowshoe hare in Ontario. Canadian Journal of Zoology 61:1339-1345. Willard T. 1992. Edible and medicinal plants of the rocky mountains and neighbouring territories. Wild rose college of natural healing, Ltd. Canada. Williams, D.E., A.R.E. Sinclair and R.J. Andersen. 1992. Triterpene constituents of the dwarf birch, Betula glandulosa. Phytochemistry 31:2321-2324. Wolfe, M.L., N.V. Debyle, C.S. Winchell and T.R. McCabe 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 42:148-153. Wolff, J.O. 1980. The role of habitat patchiness in the population dynamics of snowshoe hares. Ecological Monographs 50:111-130. Zar, J.H. 1984. Biostatistical analysis. Prentice-Hall, Inc. Englewood Cliffs, New Jersey.  157  


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