UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A Study of factors influencing a non-cyclic, island population of snowshoe hares Zimmerling, Todd 1993

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

Item Metadata


831-ubc_1993_spring_zimmerling_todd.pdf [ 3.81MB ]
JSON: 831-1.0086194.json
JSON-LD: 831-1.0086194-ld.json
RDF/XML (Pretty): 831-1.0086194-rdf.xml
RDF/JSON: 831-1.0086194-rdf.json
Turtle: 831-1.0086194-turtle.txt
N-Triples: 831-1.0086194-rdf-ntriples.txt
Original Record: 831-1.0086194-source.json
Full Text

Full Text

A STUDY OF FACTORS INFLUENCING A NON-CYCLIC, ISLANDPOPULATION OF SNOWSHOE HARESbyTODD ZIMMERLINGB.Sc., The University of Alberta, 1990A THESIS SUBMITTED IN PARTIAL FULFILMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinTHE FACULTY OF GRADUATE STUDIESDEPARTMENT OF ZOOLOGYWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAMARCH 1993© Todd ZimmerlingIn presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(SignatureDepartment of Zoo/o;/The University of British ColumbiaVancouver, CanadaDate /114i-a 0'3DE-6 (2/88)ABSTRACTIn this study I examined potential causes of theobserved demographic differences between Jacquot Island andmainland snowshoe hare (Lepus americanus) populations, inthe southwestern Yukon. The population dynamics of snowshoehares on Jacquot Island have been observed to differ fromthe dynamics of a population undergoing a "10-year cycle,"such as that on the adjacent mainland.Hare densities, predation rates, reproduction andjuvenile survival were all monitored on both Jacquot Islandand mainland study areas. Habitat on the study sites wasalso examined for cover and browse availability.Jacquot Island juveniles had survival rates 3-15 foldhigher than mainland juveniles. The difference in leveretsurvival was attributed to the lack of small mammalianpredators on the island.The Jacquot Island study grids had habitat with moredense understory than that of mainland areas. The densityof cover at 10 cm above the ground was positively associatedwith juvenile survival, and may have also contributed to theobserved differences in juvenile survival rates.Differences in predation pressure on adult hares wereobserved between the island and mainland. I examined theinfluence of stochastic predation on the Jacquot Island harepopulation through the use of simulation models. I foundthat stochastic predation alone could not account for theiiobserved dynamics of the Jacquot Island population.Through the use of a simulation model I was also ableto show that the Jacquot Island hare population would cyclein a manner similar to the mainland if it were exposed topredation which followed a Type II functional response and a1-year-delayed density-dependent numerical response. Thismodel produced cycles with a period of 9 years and a maximumhare density of 3.15 hares/ha.I concluded from this study that the observeddifferences in population dynamics between the mainland andJacquot Island hare populations are caused in part by:1) Differences in predation pressure between the mainlandand Jacquot Island, where Jacquot Island experiencesstochastic variation in predation pressure from year toyear, and the mainland experiences delayed density-dependentpredation pressure.2) Differences in the numbers of predators of juvenilehares, where Jacquot Island has only red squirrels as asmall mammalian predator, while the mainland has redsquirrels, Arctic ground squirrels and weasels.3) Differences in habitat, resulting in Jacquot Islandhaving more dense "refuge habitat" than mainland areas.iiiTABLE OF CONTENTSAbstract^Table of Contents^ ivList of Tables viiList of Figures^ viiiAcknowledgements ixGENERAL INTRODUCTION^ 1Literature Cited 9CHAPTER 1: DEMOGRAPHIC CHARACTERISTICS OF A NON-CYCLIC,ISLAND POPULATION OF SNOWSHOE HARES^ 111.1) Introduction^ 111.2) Study Area 131.3) Methods^ 14Hare trapping 14Adult hare survival^ 16Reproduction and juvenile survival^ 17Dispersal rates 18Predator numbers 191.4) Results^ 19Population density^ 19Adult hare survival 22Problems with reproduction and juvenile survival data ^ 26Male spring body weights^ 28Pregnancy rates^ 30Litter size 30Body weights of newborn hares 32Juvenile survival on Jacquot Island^ 32Proximate causes of mortality of juveniles on JacquotIsland^ 34Comparison of Jacquot Island juvenile survival withmainland 40Overwinter mortality on Jacquot Island^ 40Dispersal rates^ 41Predator numbers 411.5) Discussion 45Predation on adult hares^ 45Predation on juvenile hares 46ivExposure as a mortality factor for leverets^ 47Dispersal^ 48Causes of the Jacquot Island population decline^ 481.6) Conclusions^ 521.7) Literature Cited 54Chapter 2: THE INFLUENCE OF HABITAT CHARACTERISTICS ON ANON-CYCLIC, ISLAND POPULATION OF SNOWSHOE HARES^ 562.1) Introduction^ 562.2) Study Area 572.3) Methods^ 58Habitat analysis 58Understory cover density^ 58Understory cover at predation sites^ 60Browse diameters^ 602.4) Results 61Tree habitat 61Shrub habitat^ 64Understory cover density^ 67Understory cover at predation sites^ 67Browse diameter 712.5) Discussion^ 71Tree habitat 71Shrub habitat 72Understory cover 73Browse diameters^ 752.6) Conclusions 762.7) Literature Cited^ 78CHAPTER 3: THE USE OF SIMULATION MODELS TO EXAMINE FACTORSINFLUENCING A NON-CYCLIC, ISLAND POPULATION OF SNOWSHOEHARES^ 803.1) Introduction^ 803.2) Model 1: Delayed Density-Dependent Predation^ 823.3) Model 1: Results^ 863.4) Model 2: Stochastic Predation^ 89V3.5) Model 2: Results^ 913.6) Model 3: Stochastic Weather Patterns^ 923.7) Model 3: Results^ 973.8) Discussion^ 100Model 1^ 100Model 2 101Model 3 1033.9) Conclusions^ 1033.10) Literature Cited^ 106GENERAL CONCLUSIONS^ 108Literature Cited 114APPENDIX 1.1) Jolly-Seber Estimates of Hare Numbers onJacquot Island and Mainland Study Areas^ 115APPENDIX 1.2) Jacquot Island Litter Sizes and Dates of BirthObtained from Females Confined in Maternity Cages^ 116viLIST OF TABLESTable 1.1. 28-day survival rates of radio-collared, adultsnowshoe hares (95% confidence limits in parentheses andtotal hares collared beneath)^ 24Table 1.2. Proximate causes of mortality of radio-collared,adult snowshoe hares on Jacquot Island and mainland... .25Table 1.3. Pregnancy rates of adult female hares on JacquotIsland and mainland study areas (sample size inparentheses)^ 27Table 1.4. Mean body weights (± S.E.) of adult malesnowshoe hares on Jacquot Island in May, 1991 and May1992 (sample size in parentheses) ^ 29Table 1.5. Mean litter sizes (± S.E.) for Jacquot Islandand mainland study sites in 1991 (sample size inparentheses)^ 31Table 1.6. Mean newborn weights at birth (± S.E.) forJacquot Island and mainland study sites, in 1991 (numberof litters in parentheses) ^ 33Table 1.7. Proximate causes of mortality of radio-taggedjuvenile snowshoe hares on Jacquot Island in 1991^ 39Table 1.8. Sighting index of predators on Jacquot Islandand mainland grids during the summer of 1991 and 1992 ^ 43Table 2.1. Mean understory cover density (± S.E.) onJacquot Island and mainland study areas at 10 cm and 100cm above ground (sample size equals 30 in all cases)...68Table 2.2. Mean understory cover density (± S.E.) atpredation sites and at random locations on JacquotIsland grids at 10 cm and 100 cm above ground (samplesize in parentheses)^ 69Table 2.3 Mean diameter at point of clipping of willow(Salix spp.) browse (± S.E.), measured in the spring of1992 (sample size in parentheses) ^ 70Table 2.4. Mean understory cover density (±S.E.) at 10 cmabove ground and survival rates of leverets to 14 daysof age, in 1991 74Table 3.1. The effects of increasing and decreasing A andH values by 25% intervals, on the maximum hare densitiesreached and period of cycles (in parentheses) in Model 1simulations 88viiLIST OF FIGURESFigure 1. Jolly-Seber population estimates of snowshoehare populations on Jacquot North (Jacquot Island) andSilver (mainland) study areas during the winter(February-March)^ 5Figure 1.1. Jolly-Seber population estimates of snowshoehare populations on four study grids in the spring(April-May), fall (Aug.-Sept.) and winter (Feb.-Mar.)..21Figure 1.2. Survivorship curves for first litterradio-tagged juveniles on Jacquot Island, in 1991 (withS.E. bars) ^ 36Figure 1.3. Survivorship curves for second litterradio-tagged juveniles on Jacquot Island, in 1991 (withS.E. bars) ^ 38Figure 2.1. Distribution of spruce tree habitats on thefour study grids (percentage of grid area covered)^ 63Figure 2.2. Distribution of shrub habitats on the fourstudy grids (percentage of grid area covered)^ 66Figure 3.1. Jolly-Seber population estimates of snowshoehares on Jacquot North grid during the winter (February-March) 1978 to 1992^ 94Figure 3.2. Examples of hare densities produced byModel 2 when a dispersal factor is added^ 96Figure 3.3. Examples of hare densities produced byModel 3 simulations^ 99viiiACKNOWLEDGEMENTSI would like to thank a number of people for their helpin allowing me to complete this thesis. I would especiallylike to thank my wife, Linda, for her love and support andher ability to deal with her husband being stranded on anisland 2,100 km away, for 5 of the first 12 months of ourmarriage. My supervisor, Charles J. Krebs, always providedme with insightful comments, "The best way to destroy aphenomenon is to study it," and ensured that I not onlylearned from doing my M.Sc., but that I also enjoyed it.He, and the rest of my research committee, A.R.E. Sinclair,J.N.M. Smith and T.P. Sullivan, as well as D. Chitty wereall invaluable to me for advice and guidance in planning andwriting my thesis, and I thank them for that. I would liketo thank C. Jardine, L. Preston, S. Brown, M. Keller, and P.Vikman for their help as field assistants, and F. Doyle, C.Doyle and M. O'Donoghue for their advice in the field. Ithank I. Wingate who was of great help to me, ensuring thatI got the supplies I needed and reminding me of the suppliesI did not know I needed! I thank J. Sovell and B. Kull forallowing me to use reproductive data they collected onmainland areas in 1991 and 1992. I would also like to thankDr. K. Larsen, who, although not directly involved with thisthesis, inspired me as an undergraduate to continue mystudies at the graduate level. This study was funded by agrant from the Natural Sciences and Engineering ResearchixCouncil of Canada to C.J. Krebs, and by Northern ScienceTraining grants.xGENERAL INTRODUCTIONThe well known "10-year" cycle of the snowshoe hare(Lepus americanus) has been documented for many decades.The regularity of this cycle and its occurrence in mostsnowshoe hare populations have led to studies of a widerange of possible regulatory factors (Keith, 1987).As of yet, no clear answer has arisen as to the causeor causes of these cycles; however, a variety of hypotheseshas been put forth. Extrinsic factors such as weather(Moran, 1953) and solar activity (Elton, 1924; Sinclair etal., 1993) have been examined as factors which may influencethe cycle. Chitty (1967) proposed that genotypic shifts inthe population may influence the cycle. Christian and Davis(1964) felt changes in the endocrine feedback system may bethe key. The importance of food shortage has also beenstudied (Keith et al., 1984; Fox & Bryant, 1984; Krebs etal., 1986). A number of researchers have also examined therole of predation or vegetation-hare-predation interactions,in regulating the cycle (Keith, 1974; Keith & Windberg,1978; Keith et al., 1984; Boutin et al., 1986; Krebs et al.,1986; Trostel et al. 1987; Sinclair et al., 1988).Most work on the snowshoe hare cycle has been on cyclicpopulations. Few studies have examined non-cyclic harepopulations to explore the factors that differ between non-cyclic and cyclic hare populations. The majority of studies1that have examined non-cyclic hare populations have dealtwith populations near the southern range boundary ofsnowshoe hares (Wolff, 1981; Sievert and Keith, 1985). Inthese southern areas, factors such as competition with otherrabbit species, and the absence of obligate predators, mayexplain the lack of a cycle.No study has dealt with a detailed demographiccomparison of a cyclic and a non-cyclic hare population inclose proximity in the northern regions of the snowshoehare's range. In this study I examined a non-cyclicsnowshoe hare population on Jacquot Island in Kluane Lake,Yukon and compared it with a cyclic mainland population.Previous work by Krebs et al. (1986) and Trostel(1986), has shown that the Jacquot Island hare populationdoes not go through a "10-year cycle" that is directlycomparable to mainland hare populations. The hare densitiesobserved on a Jacquot Island study site and a mainland studysite are shown in Figure 1. It appears from these data thatthe Jacquot Island hare population is non-cyclic in nature.However, with the limited data available it is difficult toseparate non-cyclic dynamics from a dampened or loweramplitude cycle. It is clear that the demography of theJacquot Island hare population differs to some degree fromthe mainland hare population and it is the cause of thesedemographic differences that I am interested in. Throughoutthis study I will refer to the Jacquot Island hare2population as being non-cyclic in nature, however, I do notdiscount the possibility of a dampened or lower amplitudecycle occurring on the island. I suggest only that theJacquot Island hare population does not cycle in the samemanner as mainland populations do.Live trapping data collected since 1977 on JacquotIsland has shown a fluctuating density of hares from year toyear (Trostel, 1986). Densities have ranged from a high of6 hares/ha in the fall of 1978 to a low of 0.16 hares/ha inthe spring of 1985. However, there is no indication of acycle with an amplitude such as that observed on the nearbymainland areas.The objective of this study was to examine populationparameters and habitat availability on Jacquot Island andcompare these findings with data collected in mainland areas(approximately 40 km SE) currently being studied by theKluane Boreal Ecosystem Project. Comparisons between thetwo study areas has allowed me to identify differencesbetween cyclic and non-cyclic populations and to postulatenew hypotheses on the causes of the cycle. At the time ofthis study, the mainland hare populations in the Kluaneregion were declining, after reaching peak densities in 1989and 1990 (Kluane Boreal Ecosystem Project, unpublished).In this thesis I examine five main hypotheses toexplain the differences between Jacquot Island and themainland hare populations.3Figure 1. Jolly-Seber Population Estimates of Snowshoe HarePopulations on Jacquot North (Jacquot Island) and Silver(mainland) Study Areas During the Winter (February-March).4100.01^IIIIIIIii^1^1^1^1 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92Year-- Jacquot North^SilverHI: Predation pressure differs between the mainlandand Jacquot Island areas. Predation pressure follows apredictable pattern on the mainland, with predator numberstracking the hare population with a 1-3 year lag. OnJacquot Island predation pressure is less predictable.Predators hunt on the island during the winter as they do onthe mainland, but ice melt in the spring restricts themovement of mammalian predators to and from the island. Asa result, in some summers there may be no large mammalianpredators, such as coyotes (Canis latrans) or lynx (Lynxcanadensis), on the island, while in other summers there maybe several individuals. This fluctuating predation pressureresults in the observed differences in hare populationdynamics.H2: Predators of juvenile hares are at lower densitieson Jacquot Island as compared with the mainland. O'Donoghue(1991) found that red squirrels (Tamiasciurus hudsonicus)and arctic ground squirrels (Spermophilus parryii) were amajor cause of mortality for hares less than 30 days old onmainland study areas at Kluane. Red squirrels and arcticground squirrels are at lower densities on the island thanthe mainland. As a result, juvenile hare survival is higheron the island owing to reduced predation.H3: Owing to the distance between Jacquot Island andthe mainland, immigration of hares to the island occurs lessoften than emigration. Hares from the island may venture6across the 1.5-5 km of ice to reach mainland habitat, buthares on the mainland are reluctant to cross the samedistance to go to a small area like the island. The netmovement of hares away from Jacquot Island allows themainland to act as an effective dispersal sink during timesof high densities on the island. Thus the island harepopulation never reaches the peak numbers recorded for thecyclic mainland population.H4*. Hares living in a habitat with dense cover sufferlower predation rates and Jacquot Island has more such"refuge habitat" than mainland areas. Jacquot Island andmainland areas differ enough in the amount of dense coverhabitat available to hares to result in differences in haresurvival. The differences in survival rates of the haresresults in the different demographies observed.H5: Food availability is a limiting factor on JacquotIsland. Browse species are less available on the islandthan on the mainland, and this causes higher wintermortality, higher predation rates or lower reproductiveoutput of the hares. Thus, the Jacquot Island harepopulation never reaches the peak numbers seen on themainland.In this thesis I deal with each of these hypotheses.In Chapter 1 I compare the demographic characteristics ofthe Jacquot Island population with mainland areas. I alsoexamine differences in reproduction, survival and predation7rates between Jacquot Island and mainland areas. In Chapter2 I examine the habitat structure of Jacquot Island andmainland grids using vegetation analysis to consider boththe cover density and food availability hypotheses. InChapter 3 I present data on stochastic events and theirimpact on the Jacquot Island hare population. I also usedata and information from previous studies to produce adynamic population model of the Jacquot Island harepopulation.8LITERATURE CITEDBoutin, S., C.J. Krebs, A.R.E. Sinclair and J.N.M. Smith.1986. Proximate causes of losses in a snowshoe harepopulation. Canadian Journal of Zoology 64:606-610.Chitty, D. 1967. The natural selection of self-regulatorybehaviour in animal populations. Proceedings of theEcological Society of Australia 2:51-78.Christian, J.J. & D.E. Davis. 1964. Endocrines, behaviour,and populations. Science 146:1550-1560.Elton, C. 1924. Periodic fluctuations in the numbers ofanimals: their causes and effects. Journal ofExperimental Biology 2:119-163.Fox, J.F. & J.P. Bryant. 1984. Instability of the snowshoehare and woody plant interaction. Oecologia 63:128-135.Keith, L.B. 1974. Some features of population dynamics inmammals. International Congress of Game Biology 11:17-58.. 1990. Dynamics of snowshoe hare populationsin Current Mammalogy 2:119-195. H.H. Genoways editor,Plenum Press, N.Y.Keith, L.B. and L.A. Windberg. 1978. A demographicanalysis of the snowshoe hare cycle. WildlifeMonographs 58:1-70.Keith, L.B., J.R. Cary, O.J. Rongstand & M.C. Brittingham.1984. Demography and ecology of a declining snowshoehare population. Wildlife Monographs 90:1-43.Krebs, C.J., B.S. Gilbert, S. Boutin, A.R.E. Sinclair &J.N.M. Smith. 1986. Population biology of snowshoehares. I. Demography of populations in the southernYukon, 1976-84. Journal of Animal Ecology 55:963-982.Moran, P.A. 1953. The statistical analysis of the Canadianlynx cycle. II: synchronization and meteorology.Australian Journal of Zoology 1:291-298.O'Donoghue, M. 1991. Reproduction, juvenile survival andmovements of snowshoe hares at a cyclic populationpeak. M.Sc. Thesis, University of British Columbia,Vancouver B.C.9Sievert, P.R. and L.B. Keith. 1985. Survival of snowshoehares at a geographic range boundary. Journal ofWildlife Management 46:854-866.Sinclair,A.R.E., C.J. Krebs, J.N.M. Smith & S. Boutin.1988. Population biology of snowshoe hares.III. Nutrition, plant secondary compounds and foodlimitation. 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 synchronizethe snowshoe hare cycle in Canada? Evidence from treerings and ice cores. The American Naturalist 141:173-198.Trostel, K. 1986. Investigation of causes of the 10-yearhare cycle. M.Sc. Thesis, University of BritishColumbia, Vancouver B.C.Trostel, K., A.R.E. Sinclair, C.J. Walters and C.J. Krebs.1987. Can predation cause the 10-year cycle?Oecologia 74:185-192.Wolff, J.O. 1981. Refugia, dispersal, predation, andgeographic variation in snowshoe hare cycles. InProceedings of the World Lagomorph Conference,University of Guelph, Guelph, Ont. K. Myers and C.D.MacInnes editors.10CHAPTER 1: DEMOGRAPHIC CHARACTERISTICS OF A NON-CYCLIC,ISLAND POPULATION OF SNOWSHOE HARES.1.1) INTRODUCTIONA previous study by Trostel (1986) examined trappingdata collected on Jacquot Island from 1977 to 1986. Trostel(1986) found that recruitment rates did not differ betweenthe mainland and island. Overwinter survival, however,differed between the island and mainland. Using trappingdata, Trostel (1986) determined that adult hares experiencedhigher overwinter survival on the island as compared withthe mainland, in all but the first year of the study. Shealso found that overwinter survival of juveniles wassignificantly lower on the mainland than on the island,during a 3-year period centred on the peak in hare numberson the mainland.Without the use of telemetry, it is difficult todetermine the effects of hare dispersal on survival rates.In Trostel's (1986) study, only a single year of data onsurvival rates was collected using radio telemetry ratherthan trapping estimates, and these data were for adult haresonly. During this year the only difference between theisland and mainland was that male hares on Jacquot Islandsuffered higher predation rates during the summer thanmainland males.The purpose of my study was to continue to collect11demographic information on Jacquot Island and provideadditional data through the extensive use of radiotelemetry. Predation has been the primary proximate causeof mortality in cyclic snowshoe hare populations that havebeen monitored by telemetry (Brand et al., 1975; Keith etal., 1984; Boutin et al., 1986; Trostel et al., 1987). Inthis study I collected data on adult survival and cause ofdeath over two summers using radio telemetry. The datacollected on Jacquot Island were compared with datacollected by the Kluane Boreal Ecosystem Project from twomainland control grids, Silver Creek and Sulphur.Reproductive rates and juvenile survival rates havebeen documented to change throughout the hare cycle (Greenand Evans, 1940; Meslow and Keith, 1968; Keith and Windberg,1978; Krebs et al., 1986). During the low and earlyincrease phase of the cycle juvenile recruitment is high.At the peak and early decline juvenile recruitment dropsoff. These changes in recruitment are thought to be a majorcause of the decline in the snowshoe hare population (Meslowand Keith, 1968; Keith and Windberg, 1978; Krebs et al.,1986).For the non-cyclic population on Jacquot Island theonly available information on reproduction and juvenilesurvival has come from the number of juveniles trapped(Trostel, 1986). I attempted to obtain more detailedinformation on reproduction and early juvenile survival (0-1230 days of age), using techniques described by O'Donoghue(1991) for capturing pregnant hares and holding them untilparturition. These techniques allowed a more directestimate of reproductive rates and enabled the radio-taggingof young snowshoe hares. The reproductive data I collectedwere compared, with those collected in 1991, by J. Sovell(unpublished) and in 1992, by B. Kull, working for theKluane Boreal Ecosystem Project.1.2) STUDY AREAJacquot Island is approximately 5 km2 in area and islocated near the north end of the long, narrow Kluane Lake,in the southwestern Yukon. The island consists of twoseparate sections joined by a thin (30 m wide) isthmus. Thenorthern portion of the island (approximately 400 ha) isseparated from the mainland by a minimum of 1.5 km. Thesouthern end (approximately 54 ha) is separated from theadjacent mainland by a minimum of 3 km. Owing to the largedistance and cold water temperature of Kluane Lake, I assumethat Jacquot Island is completely isolated from migration ofhares and most terrestrial predators, during the ice-freemonths. From late November to late May ice covers the lake,and during this time it terrestrial animals can move to andfrom the island.The understory habitat of Jacquot Island is dominatedby shrub willow (Salix spp.) with small patches of birch13(Betula glandulosa) present at the far northern end of theisland. Spruce (Picea glauca) trees dominate the canopywith a few additional hillside groves of balsam poplar(Populus balsamifera) and aspen (Populus tremuloides).Jacquot Island is relatively hilly, with a number oftreeless, windswept hill tops covered with only grasses andlow shrubs. For a more thorough description of the habitatsee Krebs et al. (1986) and Chapter 2 of this thesis.The two mainland areas used in this study (Silver Creekand Sulphur) serve as unmanipulated control grids for theKluane Boreal Ecosystem Project. Silver Creek and Sulphurgrids are approximately 40 and 60 km southeast of JacquotIsland, respectively. The understory habitat of themainland grids is dominated by grey willow (Salix glauca)and Shepherdia canadensis. The dominant tree species isspruce with small amounts of balsam poplar also present. Adescription of these grids can be found in Krebs et al.(1986) and Smith et al. (1988).Throughout this study I will refer to the mainlandSilver Creek grid as Mainland 1 and to Sulphur grid asMainland 2. In some instances the two grids have beencombined in order to increase sample size, and in thesecases I refer simply to the mainland.1.3) METHODSHare Trapping14Two trapping grids were set up on Jacquot Island. Onewas located on the North end and the other on the South endof the island. The North grid consisted of 180 stations ina 15 x 12 grid, with 30 m between the stations. NinetyTomahawk livetraps were placed at alternate stations in acheckerboard pattern. This resulted in an effectivetrapping area of approximately 34 ha (based on radiotelemetry; J. Boulanger, unpublished).The South grid consisted of 160 stations in a 10 x 16grid, with 40 in between the stations. One Tomahawk livetrapwas placed at each station. The South grid was arrangeddifferently than the North in order to accommodate asimultaneous study by J. Boulanger on snowshoe haretrappability. Although the traps were arranged differentlythe effective trapping area was 35 ha for the South grid,(J. Boulanger, unpublished) and thus almost equal to thesize of the North grid.Traps were baited with apple and alfalfa and were leftlocked open between trapping sessions. One trapping sessionconsisted of three nights of trapping within a five-dayperiod. One session per month, from May to August wasundertaken on each grid in order to gain a monthly densityestimate. Upon capture each hare was given a numberedeartag, weighed on a spring scale, had its sex determinedand right hind foot measured as an index of body size; andthe reproductive condition of females was noted as lactating15or non-lactating.Adult Hare SurvivalA number of hares (15-20) on each grid were fitted withradio collars. Approximately half of the collars consistedof Biotrack, model SR-2 transmitters attached to a nylonwebbing collar (weight = 25-35 g). Individual hares withthese collars were tracked every 1-2 days for visualsighting to determine that they were still alive. The otherhalf of the collars were Lotek, model SMRC-3RB mortalitytransmitters, with leather collars (weight = 45-50 g).These transmitters doubled in pulse rate if the collar wasnot moved within a period of four hours. The mortalitycollars were monitored every 1-2 days, but tracking was notnecessary unless the hare had died.Survival rates for the radio-collared adults werecalculated using the Kaplan-Meier method (Kaplan & Meier,1958; Pollock et. al. 1989a), a procedure which allowscensoring of data (owing to loss or failure of thetransmitter or animals living beyond the duration of thestudy) and staggered entry of animals (Pollock et al.,1989b).The cause of death of hares was determined through siteanalysis. If predation occurred the predator involved wasdetermined using criteria set out by the Kluane BorealEcosystem Project (C. Doyle, pers. comm.) The kill site wasexamined for evidence of the predator, such as feathers,16whitewash, scat or tracks. As well, some predators havecharacteristic ways of dealing with prey. For example alynx (Lynx canadensis) may bury any uneaten parts of thehare. Raptors may leave entrails in trees and great-hornedowls (Bubo virginianus) often sever prey at the neck andcarry the head away.Reproduction and Juvenile SurvivalFollowing the procedure outlined by O'Donoghue (1991), Iattempted to capture 10-15 pregnant females, shortly beforeparturition, on each island grid. The females were placedin 60 x 60 x 120 cm, mesh wire cages, next to their point ofcapture. The back half of each cage was covered withburlap. The bottom of each cage was lined with moss andspruce branches, and spruce branches were also placed alongthe sides and woven into the top wire to create a concealedarea for the female. The females were fed half an apple and30-40 willow twigs every morning and evening. Uponparturition the cage was opened to allow the female toleave.Once the female had departed (usually only 2-3minutes), each leveret was tagged with a numbered eartag andweighed; its right hind foot was measured and its sex wasdetermined. Up to four leverets from each litter werefitted with a radio transmitter to allow survival rates tobe determined. The radios, (Biotrack, model SR-1; weight =2-3 g), were attached directly to the back with Krazy Glue.17The fur was removed from a small patch on the back to allowcontact with the skin. This procedure kept the radiotransmitters attached for 3-4 weeks. Materials from thematernity cage were then removed from the cage and placed onthe ground and made into a small nest in which the leveretswere placed. The maternity cage was then removed from thesite so as not to attract or deter predators and to avoidhaving leverets crawling into the mesh and becoming caught.Radio-tagged leverets were tracked every 1-2 days forvisual sighting to determine if they were still alive. If aleveret was found dead, the cause of death was determined asdescribed previously for the adults.I calculated survivorship curves for the radio-taggedleverets using the Kaplan-Meier procedure, as describedearlier for radio-collared adults. Survivorship curves werecompared using the log-rank test (Savage 1956, as describedin Pollock et al., 1989b).Dispersal RatesDispersal rates from Jacquot Island were examinedthrough the use of radio-collared animals. In the fall of1991, 11 juvenile (born spring/summer 1991) and 23 adulthares were collared. Twice during the winter and again inthe spring each animal was located to determine if any haddispersed to the mainland. This method gave an indicationof the rates of dispersal from the island. However, I wasunable to determine movement to the island or if individuals18moved to the mainland and returned to the island betweentelemetry sessions.Predator NumbersA sighting index of predator numbers was obtained.Each worker on the island recorded the number of hours inthe field each day and the number of each species ofpredator seen during that time. The total number of hoursand total number of sightings by all workers during eachmonth were compiled and an index of sightings per 100 hrs inthe field was calculated. The same procedure was used onthe mainland sites by the workers of the Kluane BorealEcosystem Project, to allow for comparisons of predatorsightings between the mainland and island areas.Red squirrel numbers were estimated on each of theJacquot Island grids by active midden counts and visualsightings of squirrels using middens. Abundance of arcticground squirrels was estimated through observation of activeburrows. On the mainland, red squirrel and arctic groundsquirrel numbers were estimated through live-trapping by theKluane Boreal Ecosystem Project.1.4) RESULTSPopulation DensityHare population densities declined overall from springof 1991 to the fall of 1992 on both island and mainlandstudy sites. The two island grids maintained higher19Figure 1.1. Jolly-Seber Population Estimates of SnowshoeHare Populations on Four Study Grids in the Spring (April-May), Fall (Aug.-Sept.) and Winter (Feb.-Mar.).201 ,FallFall1991 1992SpringWinter0.1 ^Spring10Mainland 1^x^ Mainland 2^° Jacquot North^—G-- Jacquot Southdensities than the mainland grids at all times (Figure 1.1).In 1991 hare numbers on Jacquot North and Jacquot Southshowed an increase from spring to fall due to reproduction.Mainland 1 and Mainland 2 grids also increased in numbers,but the increase was small, from approximately 0.9 hares/hain the spring to 1.1 hares/ha in the fall (Jolly-Seberestimates, Figure 1.1).In the fall of 1991 Jacquot South hare densities weretwice that of Jacquot North and 4 times the densities foundon the mainland grids. One cause of the higher numbers onJacquot South may be reduced dispersal rates from the grid.Jacquot South is connected to the much larger northern endof the island by a very thin (30 m wide) isthmus, as aresult, dispersal is restricted during the summer. Thisrestriction may result in a seasonal fence effect (Krebs et.al., 1969), where reduced natural dispersal increaseddensities of hares on the Jacquot South grid.From the fall of 1991 until the end of the study (fall1992) numbers on all four grids declined. In contrast tothe summer of 1991, none of the study areas showed anyincrease in densities due to reproduction in 1992. By thefall of 1992 all four populations had declined, but Jacquotisland grids still maintained slightly higher densities(Figure 1.1).Adult Hare SurvivalIn 1991, summer survival rates were significantly22higher on both Jacquot Island grids than on mainland areas(log-rank test, P < 0.001). On Jacquot Island, nodifference was found between grids in either summer (Table1.1). However, there was a significant decrease in 28-daysurvival rates on Jacquot South between summer 1991 andsummer 1992 (log-rank test, P < 0.01). On the mainlandgrids (Mainland 1 and Mainland 2 combined) 28-day survivalrates increased significantly between summer 1991 and summer1992 (log-rank test, P < 0.001).Table 1.2 displays the proximate causes of mortality ofradio-collared hares. Owing to small sample size the datafrom Jacquot North and Jacquot South have been combined. Noclear difference can be seen in the proximate causes ofmortality between the island and mainland, but there doesappear to be a difference between years on both the islandand mainland. On Jacquot Island in 1991 no incidents oflynx predation were observed, but in 1992 lynx predationmade up the largest single percentage of proximate causes ofmortality. In 1991, both goshawks (Accipiter gentilis) andgreat-horned owls were involved in predation. However, in1992 no predation by either of these avian predators wasobserved. It should be noted that in both years there werea number of predation events for which a proximate causecould not be positively identified. It is possible thatlynx may have accounted for some of the "Unknown Predator"events in 1991 or that goshawks or great-horned owls23Table 1.1. 28-day Survival Rates of Radio-Collared, AdultSnowshoe Hares (95% confidence limits in parentheses andtotal hares collared beneath).Summer31991Summer1992Jacquot 0.81 0.87North (0.76^-^0.88) (0.71^-^0.93)54 18Jacquot2 0.94 0.76South (0.87^-^0.99) (0.50^-^0.89)47 24Mainlandl 0.65 0.83(0.55 - 0.75)^(0.61 - 0.96)53 301Significant difference between years for Mainland (log-ranktest, P < 0.05).Significant differences between years for Jacquot South(log-rank test P < 0.01).3Mainland rate significantly lower than Jacquot North andJacquot South in 1991 (log-rank test, P < 0.001).24Table 1.2. Proximate Causes of Mortality of Radio-Collared,Adult Snowshoe Hares on Jacquot Island and Mainland.Percentage of MortalitiesSummer 1991 Summer 1992Mortality Factor Jacquot Mainland Jacquot^MainlandLynx 0.0 4.8 35.7^9.1Coyote 29.4 38.1 14.3 0.0Unknown Mammal 0.0 9.5 14.3 27.3Goshawk 11.8 0.0 0.0 27.3Great-horned Owl 5.8 0.0 0.0 0.0Unknown Avian 11.8 4.8 7.1 0.0Unknown Predator 29.4 42.9 21.4 27.3Non-predator 11.8 0.0 7.1 9.1Total Number Killed 17 21 14 1125accounted for some of the "Unknown Avian" or "UnknownPredator" events in 1992. However, sighting indices ofpredator presence on the island (see Predator Numbers,below) correspond well to the observed mortality factorspresented in Table 1.2.Problems with Reproduction and Juvenile Survival DataIn 1991, collection of reproduction and juvenilesurvival data for first and second litter went very well.Poor trappability and low pregnancy rates (Table 1.3) inAugust resulted in small sample sizes for the third litter.Owing to small sample size, limited data were collectedon reproduction and juvenile survival in 1992. On themainland, low pregnancy rates combined with low densities ofhares in the spring of 1992 (Figure 1.1) made it difficultto capture enough pregnant females for the first litter.The problem became worse throughout the summer as harenumbers continued to decline and trappability decreased. Asa result, the only useful data from the mainland in 1992were first litter pregnancy rates (Table 1.3).Problems also arose in 1992 on Jacquot Island.Pregnancy rates for the first litter were down to 40% (Table1.3). Also, pregnant females that were captured and placedin maternity cages did not survive more than 3 days.Although the same methods were employed in 1991 as in 1992,for some reason in 1992, after 3 days in the maternity cagesfemales were either dead or had stopped eating and had to be26Table 1.3. Pregnancy Rates of Adult Female Hares on JacquotIsland and Mainland Study Areas (sample sizes inparentheses).Pregnancy RatesJacquot^JacquotYear^Litter^North South^Mainland 111991 1 100 (17) 100 (21) 100 (15)2 87 (15) 85 (13) 85 (20)3 50 (10) 56 (16)1992 1 40 (10) 40 (15) 71 (7)2 100 (3) 100 (8)3 50 (6) ----1Mainland values in 1991 are from Sovell (unpublished, M.Sc.Thesis); mainland values for 1992 are from Kull(unpublished).27released. The cause for this response in 1992 is not known.During second and third litters on the Jacquot Islandin 1992 there were very few pregnant females left on thegrids. In addition, a lynx on the island repeatedlyattacked females in maternity cages resulting in death orinjury to the females. Maternity cages had not beendisturbed by any predators on the island in 1991. As aresult, sample sizes were too small to be of use for secondand third litters.Male Spring Body WeightsSpring body weights were examined to determine ifanimals were in poorer physical condition on Jacquot islandin 1992 than in 1991. To avoid the confounding effects ofweight change due to pregnancy, I used only males. Springweights of males did not show any statistically significantdifference between years (Table 1.4; Two-way ANOVA, P >0.25). However, there was a decrease in mean body weight ofadult males of 71 g on Jacquot North grid between May 1991and May 1992 (Table 1.4). This is a drop in body weight ofapproximately 6%, which, if it also occurs in females, maybe biologically important given that females must beginusing energy for reproduction in April. The same bodyweight trend was not observed on Jacquot South.A significant difference in spring body weights ofmales was found between grids (Table 1.4; Two-way ANOVA, P <0.01); males on Jacquot North were significantly lighter28Table 1.4. Mean Body Weights (± S.E.) of Adult MaleSnowshoe Hares on Jacquot Island in May, 1991 and May, 1992(sample size in parentheses).Body Weights (g)May 1991 May 1992'Jacquot North 1215 ± 24 1144 ± 37(19) ( 9 )Jacquot South 1264 ± 22 1296 ± 13(13)^( 9 )1Significant difference between Jacquot North and JacquotSouth in 1992 (Two-way ANOVA, P < 0.01).29than males on Jacquot South in 1992. No differences werefound in right hind foot lengths (Two-way ANOVA, P > 0.50)indicating that there was no shift in average skeletal bodysize that may account for weight differences.Pregnancy RatesIn 1991, pregnancy rates were equivalent between theJacquot Island grids and the mainland site (Table 1.3).Pregnancy rates on the island declined steadily from firstto third litter in 1991 and remained low for the firstlitter of 1992. The mainland showed the same trend in 1991,with pregnancy rates declining from 100% for first litter to85% for second litter. Low sample size prevented apregnancy rate calculation for the third litter. However,for the first litter of 1992 pregnancy rates were higher onthe mainland than on the island grids.Litter SizesIn 1991 females were held in maternity cages prior toparturition from 1 to 13 days with a mean of 6.2 days.Regression analysis revealed no effect of days in cage onlitter size (ANOVA, P > 0.25).The only statistically significant differences found inmean litter sizes between grids was for the second litterwhere Jacquot South had a larger mean litter size thanMainland 1 (Table 1.5, 2-way ANOVA, P < 0.05). On bothJacquot North and Jacquot South there was a trend for littersize to increase from first to second litter, however, the30Table 1.5. Mean Litter Sizes (± S.E.) for Jacquot Islandand Mainland Study Sites in 1991 (sample sizes inparentheses).Mean Litter SizeJacquot^JacquotLitter^ North South^Mainland 111 3.6 ± 0.4^3.5 ± 0.3^3.9 ± 0.3^(12) (10) (8)2^ 4.6 ± 0.4^5.3 + 0.42^3.9 ± 0.4(14) (8)^(10)3 5.3 ± 0.6(3)1Mainland values adapted from Sovell (unpublished, M.Sc.Thesis).Significantly different between litters and between grids(Two-way ANOVA, P < 0.05).31trend was only statistically significant for Jacquot South(Table 1.5, 2-way ANOVA, P > 0.05). Mean litter sizes in1991 did not show the same pattern on Mainland 1.Body Weights of Newborn HaresThe number of days a female was held in a maternitycage before parturition had no significant effect on meanneonate weights (ANOVA, P > 0.10). Newborns born on JacquotSouth were significantly heavier than those born on eitherof the other two grids (Table 1.6; 2-way ANOVA, P < 0.001).While both Jacquot North and Mainland 1 newborns increasedin mean weight between first and second litter, only theincrease on Jacquot North grid was statistically significant(Two-way ANOVA, P < 0.001). Newborn weights on JacquotSouth stayed relatively constant over all three litters.Juvenile Survival on Jacguot IslandJuvenile survival data for Jacquot Island in 1991 werebased on the 55 leverets that were radio-tagged during thefirst litter and 50 leverets radio-tagged during the secondlitter. No survival data were gathered for the third litterowing to the small sample of juveniles tagged (n = 8). Ofthe 105 transmitters put out in the first and secondlitters, 18 disappeared and were never recovered. I haveattributed the disappearance of four litters of 4, 3, 3 and3 juveniles to predation, because: 1) all leveretsdisappeared before seven days of age (well before the age ofdispersal), and 2) all siblings of each litter disappeared32Table 1.6. Mean Newborn Weights at Birth (± S.E.) forJacquot Island and Mainland Study Sites in 1991 (number oflitters in parentheses).Mean Juvenile Weights (g)LittersJacquot2NorthJacquotSouth Mainland 111 47.8 ± 1.6 54.3^+ 2.53 47.3^±^1.2(10) (9) (8)2 57.3^± 2.9 54.1 ± 2.8 52.6 ±^3.2(13) (9) (10)3 56.1 ± 3.2(3)1Mainland values adapted from Sovell (unpublished, M.Sc.Thesis).Significant difference between litter 1 and litter 2(ANOVA, P < 0.001)Significantly heavier than Jacquot North and Mainland(ANOVA, P < 0.001).33the same day. These four litters account for 13 of themissing transmitters. The fate of the other five leveretsthat disappeared is unknown, and I have treated them ascensored (unknown fate) individuals for survivalcalculations.For both the first and second litters, Jacquot Southleverets had significantly higher 30-day survival rates thanthose on Jacquot North (Figure 1.2 and Figure 1.3, log-ranktest, P < 0.001). Survival rates on both gridssignificantly increased between first and second litter(log-rank test, P < 0.001). During both first and secondlitters most mortality occurred within the first five daysof life. During this time leverets were most often foundtogether, very close to their place of release from thematernity cage. In most cases the leverets began to splitup and find separate hiding spots between five and sevendays of age.In 1992, 12 leverets were radio-tagged over all threelitters. Of these 12 leverets, four were killed bypredators, four died from exposure and one disappeared, allwithin the first 14 days of life.Proximate Causes of Mortality of Juveniles on Jacquot IslandThe major cause of mortality for the leverets duringfirst litter was coyote predation (Table 1.7). Allpredation attributed to coyotes occurred in the first fivedays of life and consisted of all the siblings of a single34Figure 1.2. Survivorship Curves For First Litter Radio-tagged Juveniles on Jacquot Island, in 1991 (with S.E.bars).35Figure 1.3. Survivorship Curves For Second Litter Radio-tagged Juveniles on Jacquot Island, in 1991 (with S.E.bars).37_7--a.0 0.8>0I I^I^I^I5^10 15 20 25^30Age (days)—6— Jacquot North^1=1^ Jacquot SouthTable 1.7. Proximate Causes of Mortality of Radio-tagged,Juvenile Snowshoe Hares on Jacquot Island in 1991.Percentage of MortalitiesJacquotNorthJacquotSouthMortality Factor Litter 1 Litter 2 Litter 1 LitterSuspected Red Squirrel 0.0 25.0 0.0 50.0Suspected Coyote 47.4 33.3 0.0 0.0Unknown Avian 0.0 0.0 9.1 0.0Unknown Predator 0.0 0.0 9.1 0.0Exposure 26.3 16.6 63.6 0.0Unknown Non-predator 26.3 16.6 9.1 50.0Total Mortalities 19 12 11 2Total Radio-tagged 27 30 28 20239litter disappearing the same day. It appeared that thecoyotes were able to find the leverets at a young age whenthey were still hiding together and unable to move well.Exposure to bad weather was also important during thefirst litter on both grids and during the second litter onJacquot North. In most cases bad weather resulted in entirelitters dying. The most common cause was a severe snow orrain storm occurring within five days of birth of a litter (atthis time the leverets move very little), soaking exposedlitters and resulted in death from hypothermia.Comparison of Jacquot Island Juvenile Survival with MainlandAlthough data on juvenile survival collected by Sovell(unpublished) in 1991 on the mainland were based on only thefirst 14 days of life, large differences can still be observedbetween Jacquot Island grids and the Mainland grid. Survivalrates to 14 days of age for first and second litters, were 20%and 5%, respectively, on the Mainland grid. These rates arewell below 30-day survival rates calculated for Jacquot Northand Jacquot South grids of 27% and 63%, respectively, forfirst litter, and 56% and 90% , respectively, for secondlitter.Overwinter Mortality on Jacquot IslandOverwinter mortality was higher for juveniles thanadults. Trapping data indicated that only 15% of juveniles onJacquot North and 3% on Jacquot South survived from fall 1991to spring 1992. For adults the numbers were 21% and 20% on40Jacquot North and Jacquot South, respectively, for the sametime period. The ratio of juveniles/adult decreased from 1.37to 1.0 from fall 1991 to spring 1992 on Jacquot North, andfrom 1.03 to 0.2 on Jacquot South. Trapping data can notdifferentiate between mortality and dispersal. However,overwinter telemetry data indicated the same trend inmortality. Of 23 adults radio-collared in August, 1991, 11(48%) were killed by predators before May 1992. Of the 11juveniles collared in August, 1991, seven (64%) were killed bypredators before May, 1992.Dispersal RatesOf the 34 hares radio-collared during the winter of 1991-92, 18 were found dead in March 1992, 12 disappeared and 4were found alive on Jacquot Island in May. Although therewere no confirmed cases of hares leaving Jacquot Island duringthe winter, the possibility exists that some of the missinghares may have dispersed to the mainland and out of telemetryrange. During the first week of May 1992, an adult haredispersed from the island, crossing 2 km of ice to reach themainland shore. This hare survived only two days on themainland before being killed by an unknown predator. At thetime when this dispersal occurred, all the snow had been gonefrom the island for approximately two weeks and willow browseappeared to be readily available as a food source.Predator NumbersSix of eight avian predator species were sighted less on41Jacquot Island in 1992 than in 1991. This trend occurred bothon Jacquot Island and the mainland. The only exceptions weremarsh hawks (Circus cyaneus) which increased in sightings, andkestrels (Falco sparverius) which increased between years onthe mainland, but decreased on the island (Table 1.8). Birdssuch as hawk owls (Surnia ulula), marsh hawks, kestrels, grayjays (Perisoreus canadensis), and ravens (Corvus corax), havebeen included in Table 1.8 because of their possible impact ofpredation on juvenile hares.Sightings of coyotes (Canis latrans) and lynx (Lynxcanadensis) differed between years and between areas. Duringthe summer of 1991 no sign of a lynx was found on JacquotIsland. However, in March, 1992, a lynx was sighted on theisland and several sets of tracks were also found (J.Boulanger pers. comm.). During the summer of 1992 a lynx wassighted several times every month. The majority of sightingsand signs occurred on Jacquot South.Sightings of a single coyote occurred on Jacquot Islandduring both summers after ice melt. In 1991 a number ofcoyotes (2-3) were heard howling several times throughout thesummer; however, in 1992, howling was only heard early in thespring when the ice could still be crossed by a coyote. AfterMay no howling by more than one coyote was heard.During the winter of 1991-92 J. Boulanger (pers. comm.)observed a number of different predators on the island,including coyotes, lynx, a great-horned owl and a wolverine42Table 1.8. Sighting Index of Predators on Jacquot Island andMainland Grids During the Summer of 1991 and 1992.Sightings/100 hrsJacquot Island MainlandPredator 1991 1992 1991 1992Coyote 0.18 0.08 0.28 0.06Lynx 0.0 0.81 0.31 0.13Great-horned owl 0.13 0.23 0.35 0.06Hawk owl 0.63 0.0 0.05 0.0Red-tailed hawk 0.88 0.0 2.85 2.4Goshawk 0.78 0.0 3.05 1.0Marsh hawk 0.80 0.85 1.20 2.38Kestrel 4.1 2.65 0.83 1.09Gray jay 29.93 15.60 19.63 9.52Raven 19.83 1.98 10.98 5.0743(Gulo luscus). The great-horned owl appeared to hunt on theisland throughout most of the winter and early spring, but bymid-June it had left the island.Small mammalian predators such as arctic ground squirrels(Spermophilus parryi) and weasels (Mustela spp.) were neverobserved on the island although both occur on the mainland.Red squirrels (Tamiasciurus hudsonicus) were found on theisland, but at much lower densities than the mainland. OnJacquot South, red squirrel spring densities were estimated tobe 0.30/ha, and on Jacquot North densities were estimated at0.25/ha. Red squirrel densities stayed relatively constantbetween years. On the mainland, red squirrel spring densitiesranged from 0.5/ha - 2.6/ha on control grids (Kluane BorealEcosystem Project, unpublished).One important factor to note is that vole numbers reacheda peak on Jacquot Island in 1991. Although no extensivetrapping was done, a population high was obvious from thelarge number of tunnels found in grassy areas after snowmeltas well as continuous sightings of voles while researcherswere walking through the grids. A number of voles were foundcrossing the ice (approximately 1-2 km from the island) inMay. Tracking these individuals revealed they had dispersedfrom the island. The large number of voles in 1991 and theirsubsequent decrease in 1992 may have influenced the number andspecies of predators sighted on the island.441.5) DISCUSSIONPredation on Adult HaresData collected on predators and predation events supportthe hypothesis that predation pressure differs between JacquotIsland and the mainland (Hypothesis 1).The predator sighting index indicated a difference inpredator densities between the island and mainland in bothyears of the study (Table 1.8). However, the trend ofdecreasing numbers of most avian predators sighted between1991 and 1992 was similar between the island and mainland.Mammalian predator sightings differed much more betweenyears, and between the island and mainland areas, than didsightings of avian predators (Table 1.8). The mostsignificant difference was the presence of a lynx on JacquotIsland during the summer of 1992. The lynx was the majorproximate cause of predation on the island in 1992 (Table 1.2)and was most likely responsible for the significant decreasein survival rates of adult hares on Jacquot South, where thelynx was sighted most often. Although a coyote was sighted onthe island in both 1991 and 1992, the impact of coyotes aspredators on the study areas appeared to be lower in 1992.One reason for this occurrence may have been the presence ofmore than one coyote on Jacquot Island in 1991. Although onlya single coyote was ever sighted at any time over bothsummers, tracks and howling indicated that at least twocoyotes inhabited the island in 1991. No evidence of more45than one coyote was found in 1992.Predation on Juvenile HaresEvidence was also found to support hypothesis 2:Predators of juvenile hares are at lower densities on JacquotIsland than on the mainland. No arctic ground squirrels werefound on the island, nor have any ever been sighted on Jacquotisland in the past (C.J. Krebs, pers. comm.) The lack ofground squirrels constitutes a major difference between theisland and mainland, as ground squirrels are found throughoutthe Kluane Lake area and are important as prey for a varietyof avian and mammalian predators (Kluane Boreal EcosystemProject, unpublished). Ground squirrels are also significantpredators of juvenile snowshoe hares on the mainland studyareas (O'Donoghue, 1991; J. Sovell, unpublished).A lack of suitable habitat over much of Jacquot Islandprobably resulted in red squirrels, a possible predator ofjuvenile snowshoe hares (O'Donoghue, 1991), being present atlower densities than those observed on the mainland. Redsquirrels depend heavily on mature spruce forests as primehabitat in the Kluane area. As will be discussed in Chapter2, both Jacquot North and Jacquot South lack this type ofhabitat.Although no information has been collected on weaseldensities on the mainland, researchers sighted weasels severaltimes during the summer in both years of this study. Incomparison no weasels were sighted on Jacquot Island, nor were46any predation events attributed to weasels. As a result, Iconclude that weasel densities were lower on Jacquot Islandthan on the mainland, even though the actual density onmainland study sites was unknown.The lack of predators of juvenile hares on Jacquot Islandmay in fact be a major cause of the higher survival rates ofradio-tagged juveniles, on the island than on the mainland(see p. 40).Exposure as a Mortality Factor for LeveretsIt is impossible to know how my choice of nest siteaffected the risk of mortality of leverets due to exposure.Females that give birth naturally may be more discriminatingin their choice of parturition site. However, care was takento choose sites that would protect the leverets from normalrains. As stated previously, it was severe storms thatresulted in mortalities. Whether or not a naturally bornlitter would have adequate protection from such storms isunknown, but it seems unlikely.Chance appeared to be the most important factor indetermining exposure mortalities. I found that, if less than3 days old when a severe rain or snow storm occurred, thelitter would die of exposure. As the leverets became olderthe probability of dying from exposure appeared to decrease.These data indicate an important link between weather andjuvenile survival which may be capable of influencingpopulation demographics in a stochastic manner.47DispersalMy information did not support hypothesis 3 (immigrationto Jacquot Island is much lower than emigration). Of thehares that were radio-collared overwinter, none were found todisperse from Jacquot Island. The one dispersal event thatoccurred from the island in the spring did, however, provethat dispersal is possible and may be occurring at low rates.Boutin et.al . (1984) found that dispersal rates from gridson the mainland near Kluane Lake peaked during the winterswhen hare populations were at peak and early declinedensities. The Jacquot Island population appeared to reachpeak densities in the fall of 1991 (Figure 1.1) and declinedthereafter. As a result, I would have expected dispersal tobe highest during the winter of 1991-92 if the dynamics of theisland population were similar to that on the mainland. Itappears that this was not the case, as dispersal accounted forthe loss of only a single known animal from the island overthe course of the study. This compares with 8% to 28% loss todispersal from study grids observed by Boutin et. al. (1984).Causes of the Jacquot Island Population DeclineThe most interesting result found during the two years ofthis study was the apparent crash of hare population onJacquot Island between fall 1991 and spring 1992, and thedramatic reduction in reproduction for the first litter of1992.Mainland snowshoe hare populations in the Kluane Lake48area reached peak numbers in 1989-90 and were in the declinephase of the cycle during the two years of this study (KluaneBoreal Ecosystem Project, unpublished). In 1991 hares onJacquot Island increased significantly throughout the summeron both grids, whereas Mainland 1 and Mainland 2 grids showedvery little increase over the summer as would be expected froma declining population. During the winter of 1991-92, harenumbers on the island grids decreased substantially. By theend of the study in 1992, population estimates on all fourgrids were relative similar (Figure 1.1). Upon initialexamination of these data it appears that Jacquot Island hasgone through a cyclic peak and decline during 1991 and 1992,and that, in fact, Jacquot Island does not differ from themainland in its population dynamics. Previous studies,however, suggest that the Jacquot Island population does notcycle (Krebs et. al., 1986; Trostel, 1986). During 1980-81the mainland population reached a cyclic peak and then beganto decline. The island population fluctuated in numbers fromyear to year, but showed no indication of a cyclic peak anddecline (Krebs et. al., 1986; Trostel, 1986) (Figure 1).I do not believe that the decline in the populationobserved over this study is an indication of a "10-year"cycle. Instead, I feel that a combination of factorscontributed to the decline that were unrelated to thepopulation cycle on the mainland. As previously mentioned, anumber of different predators were observed on Jacquot Island49during the winter of 1991-92. Owing to the high numbers ofhares on Jacquot Island in the fall of 1991 and the lownumbers on mainland areas, the island became a "hot spot" forpredators over the winter. Overwinter telemetry indicatedthat a minimum of 53% of radio-collared hares were killed bypredators on the island between September 1991 and May 1992.In addition to the increase in predation pressure in thewinter of 1991-92, there is evidence that the hares were at ahigher risk of predation due to malnutrition. In a study inRochester, Alberta, Keith et al. (1984) found thatmalnutrition predisposed hares to an increased risk ofpredation. Keith et al. (1984) found that juveniles were moresusceptible to malnutrition than adults, and that inpopulations suffering from food shortage juveniles suffereddisproportionately higher overwinter mortality than didadults. These results were also found by Vaughan and Keith(1981) and Windberg and Keith (1976) in studies onexperimental populations where juvenile snowshoe hare survivalwas found to be more markedly reduced by food shortage thanwas adult survival. Data collected on Jacquot Islandindicated a large difference in juvenile and adult overwintermortalities. The difference is comparable to that found byKeith et al. (1984).Another piece of evidence pointing towards malnutritionis a study by Vaughan and Keith (1981) in which all majorcomponents of reproduction were affected by malnutrition,50including pregnancy rates. Malnutrition may explain thedramatic drop in pregnancy rates found on Jacquot Islandbetween the first litter in 1991 and the first litter in 1992(Table 1.3).The inability of females to survive in maternity cagesalso points to the possibility of malnutrition. Boonstra(unpublished) found that poor nutrition reduced a snowshoehare's ability to cope with stress in their environment.Live-trapping (a stressful experience) was handled much betterby hares with more nutritive diets. Being placed in amaternity cage must also be stressful for a hare, and sincethe same procedures were used in both 1991 (when there were noproblems) and 1992, one explanation for the deterioration inhealth observed in the females in 1992, is a reduced abilityto deal with the stress. Sapolsky et al. (1987) hypothesizedthat repeated exposure to chronic stress (e.g. placing afemale in a maternity cage) can compromise the animal'sability to maintain homeostasis.One possible cause of the food shortage on the JacquotIsland, and, as a result, the ultimate factor in the declinein the population, may have been snowfall. Keith et. al.(1984) found that snowfall could either increase or decreaseavailable food for hares, depending on the angle of thevegetation (whether it would be bent down by the snow) and thedepth of the snow (whether the vegetation would be buried).Keith et al. (1984) found that snow buried available food on51his study sites and that hares would not dig in the snow toget to the food. Snowfall data collected by EnvironmentCanada at Burwash Landing (approximately 10 km from JacquotIsland) indicated a very high snowfall in the winter of 1991-92. Accumulated snowfall from October 1 until May 1 for 1991-92 equalled 145 cm. This value is more than two standarddeviations greater than the 17 year mean (1975 - 1992) of 87.1cm. This amount of snow could easily bury the majority ofbrowse available to the hares over winter and as a resultmalnutrition would become an important factor. If snowfallwas responsible for food shortage then it would explain whyspring weights of adult males (Table 1.4) did not show anyclear indication of overwinter weight loss. In May 1992, whenI first arrived on the island and started to trap the hares,snow cover had completely disappeared from the island. As aresult, hares most likely had a month or more during whichsnow levels were dropping and exposing ample food, whichallowed the hares to increase body their weights before Iarrived.1.6) CONCLUSIONSIn this study I have found that several differences existbetween Jacquot Island and mainland study sites. Largemammalian predator numbers appear to vary from year to year,and show no consistent relationship with hare densities on theisland. These results support the hypothesis that stochastic52predation pressure is a major cause of the observed haredynamics on Jacquot Island.Only one small mammalian predator, the red squirrel, isknown to inhabit the island and its densities were lower thanon the mainland sites. As a result, juvenile hare survival ismuch higher than that observed on the mainland.Weather may have been an important influence on theJacquot Island hare population. Evidence suggests that foodshortage over the winter of 1991-92, was a factor in thepopulation decline. This food shortage may have resulted fromheavy snowfall, covering a large proportion of browse.Weather also resulted in a large number of newborn deaths dueto exposure.Although hare densities on Jacquot Island during thisstudy seem to indicate a declining population such as thatobserved on the mainland, I do not believe that the ultimatecause of the decline on Jacquot island is the same as that onthe mainland. I predict that as long as the winter of 1992-93is an average winter, in terms of snowfall, the population onJacquot Island will rebound during the summer of 1993.531.7) LITERATURE CITEDBoutin, S., B.S. Gilbert, C.J. Krebs, A.R.E. Sinclair andJ.N.M. Smith. 1984. The role of dispersal in thepopulation dynamics of snowshoe hares. Canadian Journalof Zoology 63:106-115.Boutin, S., C.J. Krebs, A.R.E. Sinclair and J.N.M. Smith.1986. Proximate causes of losses in a snowshoe harepopulation. Canadian Journal of Zoology 64:606-610.Brand, C.J., R.H. Vowles and L.B. Keith. 1975. Snowshoe haremortality monitored by telemetry. Journal of WildlifeManagement 39:741-747.Green, R.G. and C.A. Evans. 1940. Studies on a populationcycle of snowshoe hares on the Lake Alexander area.I. Gross annual census 1932-1939. Journal of WildlifeManagement 4:220-238Kaplan, E.L. and P. Meier. 1958. Nonparametric estimationfrom incomplete observations. Journal of the AmericanStatistical Association 53:457-481.Keith, L.B. and L.A. Windberg. 1978. A demographic analysisof the snowshoe hare cycle. Wildlife Monographs 58:1-70.Keith, L.B., J.R. Cary, O.J. Rongstand and M.C. Brittingham.1984. Demography and ecology of a declining snowshoehare population. Wildlife Monographs 90:1-43.Krebs, C.J., B.L. Keller and R.H. Tamarin. 1969. Microtuspopulation biology: demographic changes in fluctuatingpopulations of M. ochrogaster and M. pennsylvannicus insouthern Indiana. Ecology 50:587-607.Krebs, C.J., B.S. Gilbert, S. Boutin, A.R.E. Sinclair andJ.N.M. Smith. 1986. Population biology of snowshoehares. I. Demography of food supplemented populationsin southern Yukon, 1976-84. Journal of Animal Ecology55:963-982.Meslow, E.C. and L.B. Keith. 1968. Demographic parameters ofa snowshoe hare population. Journal of WildlifeManagement 32:812-834.O'Donoghue, M. 1991. Reproduction, juvenile survival andmovements of snowshoe hares at a cyclic population peak.M.Sc. Thesis, University of British Columbia, Vancouver,B.C.54Pollock, K.H., S.R. Winterstein and M.J. Conroy. 1989(a).Estimation and analysis of survival distributions forradio-tagged animals. Biometrics 45:99-109.Pollock, K.H., S.R. Winterstein, C.M, Bunck and P.D. Curtis.1989(b). Survival analysis in telemetry studies: Thestaggered entry design. Journal of Wildlife Management53:7-15.Sapolsky, R., M. Armanini, D. Packan and G. Tombaugh. 1987.Stress and glucocorticoids in aging. Endocrinologyand Metabolism Clinics of North America 16:965-980.Sievert, P.R. and L.B. Keith. 1985. Survival of snowshoehares at a geographic range boundary. Journal ofWildlife Management 46:854-866.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 AnimalEcology 57:269-286.Trostel, K. 1986. Investigation of causes of the 10-yearhare cycle. M.Sc. Thesis, University of BritishColumbia, Vancouver, B.C.Trostel, K., A.R.E. Sinclair, C.J. Walters and C.J. Krebs.1987. Can predation cause the 10-year cycle?Oecologia 74:185-192Vaughan, M.R. & L.B. Keith. 1981. Demographic response ofexperimental snowshoe hare populations to overwinter foodshortage. Journal of Wildlife Management 45:354-380.Windberg, L.A. and L.B. Keith. 1976. Experimental analysisof dispersal in snowshoe hare populations. CanadianJournal of Zoology 55:2061-2081.Wolff, J.O. 1981. Refugia, dispersal, predation, andgeographic variation in snowshoe hare cycles. InProceedings of the World Lagomorph Conference,University of Guelph, Guelph Ont. K. Myers and C.D.MacInnes editors.55CHAPTER 2: THE INFLUENCE OF HABITAT CHARACTERISTICS ON ANON-CYCLIC ISLAND POPULATION OF SNOWSHOE HARES.2.1) INTRODUCTIONThe habitat in an area may influence the demography ofsnowshoe hare populations in a variety of ways. Snowshoehares show a distinct preference for habitats with denseunderstory (Wolff, 1980; Sullivan and Sullivan, 1983;Sullivan and Moses, 1986). If the understory is removedfrom these areas, snowshoe hares will vacate the area insearch of suitable habitat (Sullivan and Moses, 1986). Theamount of understory cover within a habitat is an importantcharacteristic in determining the ability of hares to avoidpredation. Sievert and Keith (1985) found that hares inhabitats with little understory cover suffered higherpredation rates than hares in habitat with greaterunderstory cover.Food availability may also be determined by habitatcharacteristics. The overwinter diet of snowshoe haresconsists primarily of twigs and stems <5 mm in diameter.However, the availability of such browse differsconsiderably even between neighbouring areas (Keith et al.,1984; Smith et al., 1988). A lack of suitable browse canresult in malnutrition, which has been found to predisposehares to an increased rate of predation (Keith et al.,1984). An increase in overwinter mortality due to56malnutrition has been postulated to initiate the cyclicdecline in snowshoe hares (Keith, 1974; Vaughan and Keith,1981).The question then arises as to whether or notdifferences in habitat between Jacquot Island and themainland can result in the observed differences inpopulation dynamics. This study was designed to examineHypothesis 4 and Hypothesis 5 (described in the GENERALINTRODUCTION).H4: Hares living in a habitat with dense cover sufferlower predation rates and Jacquot Island has more such"refuge habitat" than mainland areas. Jacquot island andmainland areas differ enough in the amount of habitat withdense cover available to result in differences in haresurvival. The difference in survival rates of hares resultsin the different demographies observed.H5: Food availability is a limiting factor on JacquotIsland. The availability of browse species is lower on theisland, and this causes higher winter mortality, higherpredation rates or lower reproductive output of hares.Thus, the Jacquot Island hare population never reaches thepeak numbers seen on the mainland.2.2) STUDY AREAThe grids used in this chapter are the same as thosedescribed in Chapter 1.572.3) METHODSHabitat AnalysisThe habitat on each grid was classified using modifiedcriteria developed by M. Nams in 1987 and 1988 for theKluane Boreal Ecosystem Project. At each grid station aresearcher would first determine the dominant tree and shrubspecies located within a 15-m radius circle centred at thestation. The percent cover of the dominant tree species wasthen classified according to the area of the circle covered,as dense (>50%), sparse (5%-50%) or no trees (<5%). Therelative age of the trees in the circle was also estimatedusing average height and average DBH as criteria forclassification as mature or immature. For spruce (Piceaglauca), if >50% of the trees were taller than 10 m and DBH> 20 cm, the site was classified as mature. For aspen(Populus tremuloides) and balsam (Populus balsamifera), if>50% of the trees were taller than 6 m and DBH > 13 cm, thesite was classified as mature. The percent cover of thedominant shrub species within the 15-m radius circle wasthen classified as closed (>75%), open (5%-75%), or noshrubs (<5%).Understory Cover DensityUnderstory cover was measured at two heights, 10 cm and100 cm above the ground. Cover at 100 cm above the groundshould dictate the amount of concealment an area offers tohares from avian predators. Cover at 10 cm above the ground58should be important as concealment from terrestrialpredators, especially for juvenile hares.A relative index was used to determine the density ofunderstory cover. On each grid 30 locations were randomlychosen for cover density measurements. The measurementswere done using a sighting stick. The sighting stick was 1m long and 2 cm wide. In the middle of the stick, twenty-five 1-cm wide black and orange alternating stripes werepainted, side by side.At each location the sighting stick was heldhorizontally at 10 cm and 100 cm above the ground. Theobserver stood 5 m away and observed the number of stripesthat were blocked from view by vegetation or woody debris.The observations were done while eye level was at a heightequivalent to the height of the sighting stick (10 cm or 100cm above the ground).^If any part of a stripe was obscuredfrom view it was considered blocked and counted.Four measurements at each height (10 cm and 100 cm)were done at each location. Each measurement was done at aright angle to the previous measurement. In this way anaverage cover density at each site could be determined,which eliminated problems such as single trees obstructingsighting from one direction. The understory cover densityfor a site was then calculated as the mean number ofobstructed bars. A site with no understory would then havean index of 0 and a site with a very dense understory would59have an index of 25.Understory Cover at Predation SitesIn order to determine whether or not understory coveraffected predation risk for hares, measurements of coverwere taken at each predation site on Jacquot North andJacquot South. The mean understory cover density atpredation sites on each grid was then compared to the meanunderstory cover density of that grid.Browse DiametersWolff (1980) and Pehrson (1981) found that thenutritional value and digestibility of hare browse decreaseswith increasing diameter. As a result, the size of browsetaken by a hare becomes an important factor in determiningthe adequacy of the diet (Keith et a/., 1984). Incollecting browse diameter data, I have assumed that anincrease in browse diameter indicates a decrease in foodavailability, since only hares with a reduced selectionshould be taking large diameter browse.In the spring of 1992, 30 locations on each of the fourgrids were randomly chosen. A 5-m radius circle around eachlocation was visually estimated and the diameter at point ofclipping of all willow (Salix spp.) twigs browsed by hares,within the circle, was recorded. Approximately 200 browsepoints per grid were collected. These data allowedcomparison of average diameter of willow browse takenthroughout the winter and early spring on Jacquot Island and60mainland grids.2.4) RESULTSTree HabitatTree habitat on Jacquot North and Jacquot South gridswas similar. Immature, sparsely spaced spruce trees made upthe largest proportion of the tree habitat on both grids(Figure 2.1). One important difference between JacquotNorth and Jacquot South was the presence of a largerproportion of habitat without trees on Jacquot South. Theabsence of trees is reflected in the shrub habitat datawhich will be discussed later (Figure 2.2).The mainland grids differed from one another in treehabitat and neither was similar to the Jacquot Island grids(Figure 2.1). On the island grids, immature, sparse sprucewas the most common habitat (48% on Jacquot North and 55% onJacquot South). However, Mainland 1 had no immature, sparsespruce habitat, and Mainland 2 had only 14% of this habitat.Another difference between the island and mainland grids wasthe amount of dense spruce habitat. Mainland 1 had 44%mature, dense spruce, and Mainland 2 had 21% mature, denseand 55% immature, dense habitat. Therefore, a largeproportion of the mainland grids had a closed canopy whereasa large proportion of the island grids had an open canopy.There was a large amount of post-fire debris observedon both Jacquot Island grids, but particularly on Jacquot61Figure 2.1. Distribution of Spruce Tree Habitats on theFour Study Grids (percentage of grid area covered).62M,Sp55%M,D21%Mainland 1 Mainland 2I,D31%I,Sp0100°M,SpNo4%M,D6%Jacquot NorthM,Sp9%Jacquot South48%I,D - Immature, dense spruceI,Sp - Immature, sparse spruceNo - < 5% tree coverM,D - Mature, dense spruceM,Sp - Mature, sparse spruce0 - Trees other than spruce I,D55%No1% 410/4 o%I,Sp14%M,Sp9%63North. Downed, burned logs and standing, dead trees were acommon sight. Information from the Canadian Forest Serviceindicated that a fire had affected a large part of JacquotIsland sometime in the last 60 - 80 years. There was noevidence of fire disturbance found on either of the mainlandgrids.Shrub HabitatThe two Jacquot Island grids differed considerably fromone another with respect to shrubs (Figure 2.2). JacquotNorth had the least amount of willow habitat out of the fourgrids. Almost one quarter of the area of Jacquot North had<5% shrub cover. Another 18% of the area was shrubs otherthan willow (Salix spp.) or birch (Betula glandulosa); forthe most part, this shrub was Shepherdia canadensis.Jacquot North also differed from the other three grids inthat it had no closed willow habitat. In contrast JacquotSouth was quite similar to Mainland 2 in shrub habitat.Jacquot South had very little area of no shrubs (5%) andonly 1% of the area was shrubs other than willow or birch.The mainland grids also differed from one another.Closed willow made up 77% of the habitat on Mainland 1 butonly 23% on Mainland 2. Mainland 1 differed from the otherthree grids in its lack of areas with no shrubs or areaswith shrubs other than willow. As well, it also had a muchsmaller proportion of open willow habitat. Mainland 2 wasunique in that it was the only grid which had areas where64Figure 2.2. Distribution of Shrub Habitats on the FourStudy Grids (percentage of grid area covered).6532%OW62%AA \CWNo24%OW58%No5%1%Jacquot North^Jacquot SouthOW - Open Willow^CW - Closed Willow^ CB - Closed Birch0 - shrubs other than Willow or Birch^No - <5% of ground covered by shrubsOW70% CW77%23%Mainland 1^Mainland 266birch was the dominant shrub.Understory Cover DensityOn both Jacquot Island grids understory cover was moredense at 10 cm above the ground than it was at 100 cm abovethe ground (Table 2.1). This trend was reversed on themainland grids. At 10 cm above the ground, all grids hadsignificantly different mean cover densities (ANOVA, P <0.001) from one another. Jacquot North and Jacquot Southdensities were approximately twice as high as Mainland 1,and 3-4 times as high as Mainland 2. Mainland 2 had asignificantly lower mean cover density at 100 cm above theground (ANOVA, P < 0.05) than the other three grids.The majority of low ground cover (10 cm above theground) on Jacquot North consisted of grass, low shrubs anddowned woody material. There was a large amount of post-fire debris on Jacquot North. It was this debris thatcreated most of the obstructions observed during the coverdensity measurements.Understory Cover at Predation SitesOverall there was a trend for hares to be killed inareas of less dense understory as compared to mean gridcover densities on Jacquot North and Jacquot South (Table2.2). However, the only statistically significantdifference was for Jacquot South at 10 cm above the ground(Wilcoxon test, P < 0.001).67Table 2.1. Mean Understory Cover Density (± S.E.) onJacquot Island and Mainland Study Areas at 10 cm and 100 cmAbove Ground (sample size equals 30 in all cases).Cover Density10 cm2 100 cmJacquot North 17.8 ± 1.1 12.7 ± 1.3Jacquot South 23.1 ± 0.7 13.8 ± 1.2Mainland 1 10.8 ± 1.1 13.8 ± 1.3Mainland 2 5.4 ± 0.9 8.3 +^1.411Significantly lower than other grids (ANOVA, P < 0.05).2Significant difference between all grids (ANOVA, P <0.001).Note: 25 = most dense cover, 0 = no cover.68Table 2.2. Mean Understory Cover Density (± S.E.) atPredation Sites and at Random Locations on Jacquot IslandGrids at 10 cm and 100 cm Above Ground (sample size inparentheses).Jacquot NorthCover Density10 cm 100 cm15.9 ± 2.9 11.3^± 2.2Predation Sites (12) (12)Jacquot North 17.8 ± 1.1 12.7^±^1.3Random Locations (30) (30)Jacquot South 13.5^+^3.41 11.7^±^3.9Predation Sites (8) (8)Jacquot South 23.1 ± 0.7 13.8^± 1.2Random Locations^(30)^ (30)1Significantly lower than mean understory cover density forJacquot South random locations (Wilcoxon, P < 0.001).69Table 2.3. Mean Diameter at Point of Clipping of Willow(Salix. spp) Browse (± S.E.), Measured in the Spring of 1992(sample size in parentheses).Mean Browse Diameter (mm)Jacquot NorthJacquot SouthMainland 1Mainland 22.8 ± 0.8(217)3.0 ± 0.9(238)2.7 ± 0.9(200)3.0 ± 0.9(205)70Browse DiameterThe mean diameter of willow browse measured in thespring of 1992 indicated no significant difference betweenthe four grids (ANOVA, P > 0.1)(Table 2.3).2.5) DISCUSSIONTree HabitatThe data suggest that habitats differ between JacquotIsland and the mainland study areas. In particular, thelack of mature spruce on both Jacquot Island grids has anumber of implications (Figure 2.1). Snowshoe hares in theKluane area feed upon the lateral branches of spruce duringthe winter (Smith et al., 1988, D. Hik, pers. comm.);however, juvenile spruce (<150 cm in height) contain highconcentrations of camphor, which is a deterrent to harebrowsing (Sinclair et al., 1988a). Sinclair et al. (1988b)found that four of eight snowshoe hares refused to eatjuvenile spruce during an experimental feeding trial.Although not all the immature spruce identified on JacquotIsland would be considered juvenile (<150 cm in height) asignificant proportion of the dense spruce habitat was inthis category. As a result, snowshoe hares on JacquotIsland may be limited in the availability of spruce browseover winter, not only owing to the large amount of sparsespaced spruce habitat, but also owing to a reducedpalatability of juvenile trees.71The lack of mature spruce may also affect theaccessibility of other browse species during winters of highsnowfall. Mature trees provide important winter range for anumber of ungulate species (Edwards, 1956; Gilbert et al.,1970; Jones, 1975; Bunnell, 1979). The large canopies ofmature trees intercept snowfall, thereby reducing snow depthon the ground, facilitating travel and making food moreaccessible. In a similar way, mature trees may also beimportant for snowshoe hares. In years of high snowfall alarge amount of browse can be buried and become inaccessibleto hares (Keith et al., 1984). On the mainland sites, wheremature trees are common, snowfall may be less of a problemfor hares since snow depth may be reduced owing to largetrees intercepting the snow. Hares on Jacquot Island,however, may be more influenced by snowfall since there arefew trees with canopies large enough to intercept asignificant amount of snow. Although it appears that inmost years snow accumulation on Jacquot Island is less thanon the mainland (Krebs et al., 1986) and therefore largetrees are less important for snow interception, in yearssuch as the winter of 1992, when snowfall was exceptionallyhigh (see Chapter 1) the large canopies of mature trees maybecome extremely important.Shrub HabitatAlthough I took no direct measurement of food availableto hares, it appears that Jacquot North has less willow72browse per unit area than the other grids. Habitat analysisindicated that approximately 1/4 of the area of JacquotNorth grid had <5% shrub cover (Figure 2.1). This result isin striking contrast to the other three grids, which havebetween 23% and 77% closed willow habitat. This differencein willow shrubs may influence the quality of habitat forthe hares. From the standpoint of shrub growth JacquotNorth seems to be sub-optimal habitat (Keith, 1966; Wolff,1980; Sievert and Keith, 1985). If this is true, then morework is required to examine all habitat on Jacquot Island todetermine if a large proportion of the habitat would beconsidered sub-optimal. Such a result could explain theobserved demographic differences between the island andmainland hare populations.Understory CoverUnderstory cover density at 10 cm above ground was muchhigher for both Jacquot Island grids compared to themainland areas (Table 2.1). Understory densities appearedto be associated with juvenile hare survival presented inChapter 1 (Table 2.4). Juvenile survival rates were highestfor Jacquot South, and lowest for Mainland 1. This trendsuggests that cover density at 10 cm affects predation rateson juvenile hares. This result makes intuitive sense giventhat the majority of deaths of juveniles (1-30 days of age)occurred in the first 5 days of life (see Chapter 1).During this time the leverets are not very mobile, and must73Table 2.4. Mean Understory Cover Density (± S.E.) at 10 cmAbove Ground and Survival Rates of Leverets to 14 Days ofAge, in 1991.Study Grid Cover DensitySurvival Rate1st Litter^2nd LitterJacquot South 23.1 ± 0.7 63% 90%Jacquot North 17.8^± 1.1 27% 70%Mainland 1 10.8 ± 1.1 20% 5%74depend completely on staying hidden to avoid predators.These results are the first indication that ground cover mayinfluence juvenile (<30 days of age) snowshoe hare survival.Sievert and Keith (1985) found that understory coverinfluenced predation rates on adult hares. Owing to thelimited number of adult kills that took place on the JacquotIsland grids, I could not compare predation rates betweenJacquot North and Jacquot South to determine if the averagecover density on a grid affected the predation rates.However, cover densities at predation sites on JacquotIsland indicated the same trends as found by Sievert andKeith (1985). Kills occurred in areas where understorycover density was less than the average density of thegrids.Browse DiametersNo indication of differences in overwinter foodavailability were found among the four grids. The meandiameter of willow browse was very similar among all fourgrids. In addition, the means were well below the 5 mmdiameter assumed to be the upper limit of optimal harebrowse in the area (Smith et al., 1988). These data areconsistent with the results of a feeding experiment by Krebset al. (1986), who found no response of the hare populationon the south end of Jacquot Island to the addition of rabbitchow as a winter food source.752.6) CONCLUSIONSJuvenile snowshoe hares may have had a higher survivalrate on Jacquot Island grids, owing in part, to a more denseunderstory compared with that on the mainland. As well,predation events on the island tended to occur in areas ofless than average understory cover. These data supportHypothesis 4: Hares living in a habitat with dense coversuffer lower predation rates and Jacquot Island has moresuch "refuge habitat" than mainland areas. The data are alsoconsistent with results presented by Wolff (1981) andSievert and Keith (1985), who showed that predation rates onhares were higher in areas of less understory cover.Spruce habitat differed considerably between the twoisland grids and the mainland areas. The lack of maturetrees may have considerable impact on the effects ofsnowfall on the hare population. An increased influence ofweather on the Jacquot Island hare population may explain anumber of the results presented in Chapter 1. Thepossibility of differences in habitat resulting in thedifferential impact of weather on hare populations warrantsfurther work in the future.Although spruce browse may be less available on JacquotIsland than mainland areas, there was no evidence ofoverwinter food shortage to support Hypothesis 5: Foodavailability is a limiting factor on Jacquot Island. Therejection of the food limitation hypothesis is a puzzling76result, as data from Chapter 1 indicated that reproductionin the spring of 1992 was reduced significantly from theprevious year. I hypothesized that these results werecaused by a food shortage, however, data from Chapter 2 donot support this hypothesis. The possibility still existshowever, that a short-term food shortage near the time ofbreeding may have occurred. A short-term food shortage maynot be detected by the crude methods employed in this study.A second possibility is malnutrition. Rather than femalehares being short of total biomass of browse, they may havelacked a certain browse species. These possibilitiesdeserve further research in the future.It appears from these data that habitat differencesalone are not responsible for the observed differences inthe population dynamics of the snowshoe hares. Habitatdifferences, however, are evident and may, in combinationwith other factors, contribute to the differences inpopulation fluctuations between Jacquot Island and themainland.772.7) LITERATURE CITEDBunnell, F.L. 1979. Deer-forest relationships on northernVancouver Island. Pp. 86-101 in Sitka black-taileddeer. USDA Forest Service, Alaska Region, series No.R10-48. Wallmo O.C. and J.W. Schoem editors.Edwards, R.Y. 1956. Snow depths and ungulate abundance inthe mountains of western Canada. Journal of WildlifeManagement 20:159-168.Gilbert, P.F., O.C. Wallmo and R.B. Gill. 1970. Effect ofsnow depth on mule deer in Middle Park, Colorado.Journal of Wildlife Management 34:15-23.Jones, G.W. 1975. Aspects of winter ecology of black-taileddeer (Odocoileus hemionus columbius Richarson) onnorthern Vancouver Island. M.Sc. Thesis, University ofBritish Columbia, Vancouver B.C.Keith L.B. 1966. Habitat vacancy during a snowshoe haredecline. Journal of Wildlife Management 30:828-832.. 1974. Some features of population dynamics inmammals. Proceedings of the International Congress onGame Biology 11:17-58.Keith, L.B., J.R. Cary, O.J. Rongstad and M.C. Brittingham.1984. Demography and ecology of a declining snowshoehare population. Wildlife Monographs 90:1-43.Krebs, C.J., B.S. Gilbert, S. Boutin, A.R.E. Sinclair andJ.N.M. Smith. 1986. Population biology of snowshoehares. I. Demography of food-supplemented populationsin the southern Yukon, 1976-84. Journal of AnimalEcology 55:963-982.Pehrson, A. 1981. Winter food consumption anddigestibility in caged mountain hares. Pages 732-742in Proceedings of the World Lagomorph Conference,University of Guelph, Guelph Ont. Myers, K. and C.D.MacInnes, editors.Sievert, P.R. and L.B. Keith. 1985. Survival of snowshoehares at a geographical range boundary. Journal ofWildlife Management 49:854-866.Sinclair, A.R.E., M.K. Jogia and R.J. Anderson. 1988a.Camphor from juvenile white spruce as an antifeedantfor snowshoe hares. Journal of Chemical Ecology14:1505-1514.78Sinclair, A.R.E., C.J. Krebs, J.N.M. Smith, S. Boutin.1988b. Population biology of snowshoe hares III.Nutrition, plant secondary compounds and foodlimitation. Journal of Animal Ecology 57:787-806.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 ofAnimal Ecology 57:269-286.Sullivan, T.P. and D.S Sullivan. 1983. The use of indexlines and damage assessments to estimate populationdensities of snowshoe hares. Canadian Journal ofZoology 61:163-167.Sullivan, T.P. and R.A. Moses. 1986. Demographic andfeeding responses of a snowshoe hare population tohabitat alteration. Journal of Applied Ecology 23:53-63.Vaughan, M.R. and L.B. Keith. 1981. Demographic responseof experimental snowshoe hare populations to overwinterfood shortage. Journal of Wildlife Management 45:354-380.Wolff, J.O. 1980. The role of habitat patchiness in thepopulation dynamics of snowshoe hares. EcologicalMonographs 50:111-130.• 1981. Refugia, dispersal, predation, andgeographic variation in snowshoe hare cycles. Pp.441-449 in Proceedings of the World LagomorphConference, University of Guelph, Guelph, Ont. Myers,K. and C.D. MacInnes, editors.79CHAPTER 3: THE USE OF SIMULATION MODELS TO EXAMINE FACTORSINFLUENCING A NON-CYCLIC, ISLAND POPULATION OF SNOWSHOEHARES.3.1) INTRODUCTIONTrostel et al. (1987), using data collected from theKluane Lake area, Yukon, was able to produce a simulationmodel with 8-11 year cycles of snowshoe hares (Lepusamericanus). Trostel et al. (1987) used only a Type IIfunctional response (Holling, 1959) and a delayed density-dependent numerical response (Nicholson and Bailey, 1935) ofpredators to produce the cycles. However, a number ofstudies have produced evidence for a predator-hare-vegetation interaction as a driving force of the 10-yearcycle in snowshoe hares (Keith, 1974; Vaughan and Keith,1981; Keith et al., 1984; Smith et al., 1988).Smith et al. (1988) and Sinclair et al. (1988) foundthat the hare population in the Kluane area had asignificant impact on the vegetation during the 1980-81 harepeak. Sinclair et al. (1988) also found that a proportionof the hare population suffered from malnutrition during thepeak. Work by Keith et al. (1984) and Windberg and Keith(1976) have shown that food limitation can affect a harepopulation through increased risk of predation, decreasedreproductive output and decreased juvenile survival.In this chapter I have incorporated both food80limitation and predation into a simulation model to testhypotheses about factors influencing the non-cyclic JacquotIsland hare population. In my model I have used informationfrom Smith et al. (1988), Sinclair et al. (1988) and Peaseet al. (1979) to estimate browsing rates of hares, browsespecies preference and browse growth rates.The predation parameters used were those estimated byTrostel et a/. (1987) for a Type II functional response anda 1-year-delayed density-dependent numerical response ofpredators. These parameters were estimated by Trostel etal. (1987) for mainland areas being studied by the KluaneBoreal Ecosystem Project.The vegetation and predation estimates were thencombined with snowshoe hare reproductive and survival data Icollected on Jacquot Island, to produce a simulation model.This model allowed the test of the hypothesis that theJacquot Island hare population would become cyclic if itexperienced the same predation pressures (Type II functionalresponse and delayed density-dependent numerical response)as observed on the mainland areas.I then removed the delayed density-dependent numericalresponse of predators from the model and replaced it withstochastic predation pressure. I then tested the hypothesisthat stochastic predation pressure (on a yearly basis) isthe main cause of the observed dynamics of the hares onJacquot Island. I compared the resulting simulated81population dynamics to the observed dynamics of JacquotIsland over the past 15 years. I then added the effects ofstochastic weather patterns and the increased juvenile haremortality which results from severe storms (see Chapter 1:Exposure as a Mortality Factor for Leverets) to the modeland examined the effects on the Jacquot Island harepopulation.3.2) MODEL 1: DELAYED DENSITY-DEPENDENT PREDATIONPredation rate was modelled using parameters estimatedby Trostel et al. (1987). I assumed a Type II functionalresponse (Bolling, 1959) and a linear delayed density-dependent numerical response (Nicholson and Bailey, 1935) todetermine predation rate in this model, as did Trostel eta/. (1987).Predation effects are calculated as instantaneouspredation rate in the Nicholson-Bailey model and are equalto:Predation rate = kill/N(t)^ (1)where N(t) is the hare numbers at the beginning of the timeinterval in which the predation rate is measured. The killcomponent of equation 1 is the product of predator numbersand kill rate per predator. Predator numbers areproportional to hare numbers the year previous and arecalculated as follows:82Predator numbers = c * (N(t-1))^ (2)where c represents the constant proportional change inpredator numbers in relation to hare numbers in the previousyear, N(t-1).The kill rate per predator is modelled with a Type IIfunctional response equation:Kill rate = a * N(t)/(H + N(t))^ (3)where a is the maximum kill rate per predator and H is thehare density at which predators can achieve 1/2 of a.Combining equation 2 and equation 3 then gives aninstantaneous rate of predation:a * N(t)/(H + N(t))) * (c * N(t-1))^(4)N(t)This simplifies to:a * c * N(t-1) / H + N(t)^(5)Trostel et al. (1987) further simplified equation 5 to:predation rate = A * N(t-1) / H + N(t)^(6)where A is the combined effects of numerical and functionalpredation responses (A = a*c).Trostel et al. (1987) estimated A and H values for theKluane area to be 0.198 kills/hare/month and 0.313 hares/hafor the winter months, and 0.072 kills/hare/month and 0.06983hares/ha for the summer months.Reproduction for the hares was assumed to take place inMay, June and July as per data collected on Jacquot Island.The litter sizes were 3.6, 5.0 and 5.0 leverets per femalein each of the months, respectively. Number of youngproduced was calculated as:Young = N(t) * 0.5 * B * LS^ (7)where N(t)*0.5 calculates the number of females, B is thelitter size and LS is the 30-day survival rate for theleverets (Chapter 1).Owing to a lack of data on predation rates on leveretsover 30 days of age, I estimated predation during the summerupon leverets older than 30 days to be 2.5%/month.Juveniles were graduated to the adult age class and beganto experience the same predation pressures in September.In this model hares fed upon Salix glauca, Piceaglauca, and Shepherdia canadensis, which were the three mostcommonly browsed species in the Kluane area (Smith et al.,1988). Adult hares were assumed to consume 300 g/day (wetweight) of browse (twigs < 5 mm in diameter; estimated fromPease et al., 1979) all year long. Juvenile hares consumed1/4 of the adult intake (75 g/day) during the summer, butconsumed 300 g/day during the winter months. The amount ofeach of the three browse species eaten by hares wasestimated according to preferences shown by the hares (Smith84et al., 1988; Sinclair et al., 1988). For the winter monthsI estimated 60% of the diet to be Salix glauca, 25% to bePicea glauca and 15% to be Shepherdia canadensis. Duringthe summer months the diet changed to 30% Salix glauca, 17%Picea glauca, 3% Shepherdia canadensis and 50% herbs andgrasses.The survival rate of hares was calculated for bothadults and juveniles as a simple linear relationship withupper and lower thresholds (0.985 and 0.5 respectfully):Monthly survival = Browse Available / Browse Required (8)where browse available was the combined biomass of Salixglauca, Picea glauca, Shepherdia canadensis and herbs andgrasses in each month. Browse required was total browserequired by all hares in the population during the month.The requirement for adult hares was estimated by Pease etal. (1979) to be 3000 g/day per adult hare. The 3000 g ofbrowse represents the standing crop biomass which must beavailable to an adult hare in order for it to acquire the300 g of choice browse it needs to maintain body weight(Pease et al, 1979). During the summer months I estimatedjuvenile hare requirements to be 1/4 of adult harerequirements (750 g/day). From September to April juvenilesand adults both required 3000 g/day of available browse.The biomass of different browse species available tohares was estimated from Smith et al. (1988) who measured85browse availability and twig growth of Salix glauca, Piceaglauca and Shepherdia canadensis over an 8 year period on anumber of study sites in the Kluane area. Using thesevalues I estimated starting biomass of browse for the modelto be 1200 kg/ha (wet weight) of Salix glauca, 265 kg/ha(wet weight) of Picea glauca, and 265 kg/ha (wet weight) ofShepherdia canadensis. Herbs and grasses also becameavailable to the hares during the summer months at 400kg/ha.Browse growth calculations were simplified to a singlemonth, where all browse growth took place in May. Thegrowth of the browse was estimated from data collected bySmith et al. (1988). The growth of browse was calculatedas:Plant(i) available * (1 + GR(i))^(9)where GR(i) is the growth rate of each Plant(i). In thismodel the growth of Salix glauca was 25%/yr, Picea glaucawas 21%/yr and Shepherdia canadensis was 35%/yr.Using these parameters I incorporated predation andfood limitation effects on snowshoe hares into a simulationmodel.3.3) MODEL 1: RESULTSUsing A and H values for summer and winter predation(eq. 6) estimated by Trostel et al. (1987), the model86produced hare cycles with a period of 9 years and peakdensities of 3.15 hares/ha. This compares to mainland peakdensities of 2.5-5.6 hares/ha (Krebs et al., 1986, 1992).I examined the sensitivity of the model to A and Hestimates by increasing and decreasing summer and wintervalues of A simultaneously and summer and winter values of Hsimultaneously. I increased A and H estimates 25%, 50%, 75%and 100% over the estimated values of Trostel et al. (1987)(A summer = 0.072 kills/hare/month, winter = 0.198kills/hare/month; H summer = 0.07 hares/ha, winter = 0.313hares/ha). I also examined the effect of reduced A and Hestimates by reducing them by 25%, 50% and 75%. The modelproduced 8-9 year cycles with peak densities between 2 and 4hares/ha over a wide range of H values, but only a narrowrange of A values (Table 3.1).For A values set at 0.072 kills/hare (summer) and 0.198kills/hare (winter), cycles with periods of 9 years and peakdensities of approximately 3.0 hares/ha were produced forall H values examined (Table 3.1). Changes in H had littleaffect on the period or amplitude of the cycle for all butthe largest values of A (+100%, Table 3.1). With A valuesset at 0.144 kills/hare (summer) and 0.396 kills/hare(winter), the period of the cycles ranged from 6 to 10 yearsand the maximum hare densities reached varied from 0.48hare/ha up to 10.7 hares/ha, over the range of H values used(Table 3.1).87Table 3.1. The Effects of Increasing and Decreasing Aand H Values by 25% Intervals, on the Maximum Hare DensitiesReached and Period of Cycles (in parentheses) in Model 1Simulations.0.48(6)0.51(6)0.54(6)0.59(6)10.7(9)9.91(10)10.1(10)10.8(10)0.30(6)0.35(6)0.39(6)0.41(6)0.44(6)0.47(6)0.49(6)0.55(6)11.1(10)10.6(10)6.87(10)9.75(10)9.20(10)8.71(10)8.78(10)9.27(10)-_8.30(10)3.13(9)2.98(9)8.34(10)3.10(9)2.40(9)9.02^5.72(10)^(9)^5.63^8.37(9)^(9)8.78^3.56(9)^(9)3.18(9)3.13(9)3.15(9)1.65(8)3.03(9)1.42(8)3.02(9)1.36(8)3.02(9)1.38(8)3.02(91.38(8)0.82(8)0.83(8)1.07(8)1.04(8)1.01(8)0.98(8)0.82(8)0.73(8)** ** ** ** ** ** ** **-75% -50% -25% 00 +25% +50% +75% +100+1 0 0+75+50+25Change inEstimated 0A Values-25%-50%-75%Change in Estimated H ValuesNote: Highlighted area indicates cycles with period andmaximum hare densities within the range observed inthe Kluane area.00 - represents the values of A and H estimated byTrostel et al. (1987) (A winter = 0.198 kills/hare, Asummer = 0.072 kills/hare; H winter = 0.313 hares/ha,H summer = 0.07 hares/ha).** - dampening cycles.88The lowest A values used (summer = 0.018 kills/hare,winter = 0.0495 kills/hare) in combination with all H valuesresulted in dampening cycles. There appeared to be no cleartrend through changing A values on the period or maximumhare densities of the cycle. Between A values that had beenreduced by 50% (-50%, Table 3.1) increased by 50% (+50%,Table 3.1), maximum hare densities increased as A valueswere increased. Increases of A above +50% resulted in adrop in maximum hare densities reached (Table 3.1).3.4) MODEL 2: STOCHASTIC PREDATIONAll parameters for this model are the same as thosedescribed for Model 1, except for the way in which predationwas calculated. In testing the stochastic predationhypothesis I have assumed that the existence of aterrestrial predator on Jacquot Island during the summer isa random event governed by the probability of a lynx (Lynxcanadensis) or coyote (Canis latrans) being on the islandwhen ice melt occurs. I assumed that once ice melt occurs,in late May, terrestrial predator numbers can not increaseuntil November when the ice forms again. Terrestrialpredator numbers can decrease due to starvation if harenumbers are too low. The only avian predator whichI included in this model was the great-horned owl (Bubovirginianus). Owing to the small size of the island (5 2 km)I have assumed that the maximum number of coyotes, lynx or89great-horned owls that could be present on the island wastwo. I also assumed the presence of a great-horned owl onthe island to be a random event dependent on the probabilityof an owl or pair of owls making Jacquot Island theirterritory. As a result of these assumptions the numericalresponse of predators was eliminated from the predationequation.I created ten different combinations of the threepossible predators which could be found on the island,including a possibility of no predators. The combination ofpredators found on Jacquot Island each year was thendetermined in May by a random number, which corresponded toone of the ten predator choices.Predation effects of each type of predator (coyote,lynx or great-horned owl) were calculated as a Type IIfunctional response (as in Model 1). The values used wereestimated from unpublished data from the Kluane BorealEcosystem Project and M. O'Donoghue (pers. comm.). Duringthe spring and summer the maximum kill rate of each predatorwas assumed to be 1 adult hare/day. During the fall andwinter months the maximum kill rate of coyotes was increasedto 1.5 hares/day, owing to surplus killing which has beenobserved to be done by coyotes in the fall at Kluane (KluaneBoreal Ecosystem Project, unpublished).Using these estimated values I examined if this form ofyearly stochasticity in predation pressure could account for90observed demographics of the Jacquot Island snowshoe harepopulation.3.5) MODEL 2: RESULTSSimulations with stochastic predation alone producedhare populations that crashed to very low densities (<0.1/ha) and remained there. I found that because of therandom nature of the predation pressure in the model, thehare population was able to escape the effects of predatorregulation if there were 2 or more years of low predationpressure. In these situations hare numbers increasedrapidly and peaked between 16-24 hares/ha. The high densityof hares resulted in all vegetation being eaten and thehares then crashed the following year. Hare densitiesthereafter remained low (< 0.1 hares/ha) and were limited bybrowse growth.In order to prevent the biologically unrealistic peakdensities of hares from occurring in the simulations I addedbrowse dependent dispersal to the model. Windberg and Keith(1976) and Boutin et al. (1985) found that snowshoe haredispersal rates increased during winters of food shortage.I used a continuous function equation to estimate thedispersal rate from Jacquot Island in relation to browseavailability:D * (1/Foodratio)^ (11)1/F + (l/Foodratio)91where D is the maximum dispersal rate and F is the foodratioat which half the maximum dispersal rate occurs. Foodratiois the ratio of browse available to browse required by thehares. The values of D for juvenile and adult hares wereestimated from Boutin et al. (1985) to be 0.20 and 0.05,respectively. For both juveniles and adults I estimated Fto be 6.With a dispersal parameter added to the model the harepopulation could be maintained for more than 200 years atdensities above 0.1 hares/ha. However, peaks up to 13hares/ha still occurred during some simulations. Haredensities of this magnitude are well above densitiesobserved on Jacquot North grid over the past 15 years(Figure 3.1).Figure 3.2 shows 3 simulations as examples from Model 2with dispersal added. The yearly fluctuations in densitiesobserved on Jacquot North (Figure 3.1) can also be seen inthe simulations (Figure 3.2). The simulations, however,often reach hare densities higher than those observed onJacquot North.3.6) MODEL 3: STOCHASTIC WEATHER PATTERNSThe effect of stochastic weather patterns was added toModel 2. I found that on Jacquot Island a number of younglitters died during severe rain and snow storms (Chapter 1).In this model a random number in each of the months during92which leverets were born determined the effect of weather onFigure 3.1. Jolly-Seber Population Estimates of SnowshoeHares on Jacquot North Grid During the Winter (February-March) 1978 to 1992.931 00.1^[IIIIIIiiiiii 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92YearFigure 3.2. Examples of Hare Densities Produced by Model 2When a Dispersal Factor is Added.95101O. 110Simulation 11^2^3^4^5^6^7^8^9 10 11 12 13 14 15Simulation 212^3 4^5^6^7^8^9 10 11 12 13 14 15Simulation 3101 2^3^4^5^6^7^8^9 10 11 12 13 14 1596Yearsurvival. Weather was able to cause 0%, 10%, 30% or 50%mortality to leverets in their first month of life.Mortality caused by weather was added into the reproductionequation (eq. 7):N(t) * 0.5 * B * LS * (1 - W)^(10)where /-W is the mortality caused by weather. The combinedeffect of stochastic predation pressure and stochasticleveret mortality, caused by weather, was then examinedusing this model.3.7) MODEL 3: RESULTSThe addition of the weather effects to Model 2 (withdispersal added) reduced the number of high densityoutbreaks of hares that occurred. The weather effects actedessentially as another predation factor upon young hares.As a result, in years when no predators were present, (thetime during which hare numbers increased quickly in Model2), weather could still have an impact on the population, bykilling up to 50% of the newborns. Weather was, therefore,capable of preventing large increases in hare numbers in asingle year.Overall hare densities appeared to be slightly lower onaverage, in Model 3 simulations as compared to Model 2simulations (Figure 3.2 and Figure 3.3). A comparison of97Figure 3.3. Examples of Hare Densities Produced by Model 3Simulations.9810 Simulation 11102^3^4^5^6^7^8^9 10 11 12 13 14 15Simulation 21102^3^4^5^6^7^8^9 10 11 12 13 14 15Simulation 312 3 4 6 8 799Year8^9^10 11^12 13 14 15sample simulations of Model 3 (Figure 3.3) and the observeddata from Jacquot North (Figure 3.1) reveal a great deal ofsimilarity between the simulations and the observeddensities. For the most part hare densities from the Model3 simulations were within the range of densities observed onJacquot North (0.2-2.0 hares/ha). The one obvious exceptionwas in simulation 3 (Figure 3.3) where in year 15 haredensities reached 7.2 hares/ha. Densities this highoccurred in both Model 2 and Model 3 simulation, but thefrequency of occurrence was very much reduced in Model 3simulations.3.8) DISCUSSIONModel 1Using predation parameters estimated by Trostel et a/.(1987) I have been able to produce a simulation model thatproduces 9 year cycles of snowshoe hares with peak densitiesfrom 2.4-5.7 hares/ha. These densities are within the rangeobserved in the Kluane area during the 1980-81 and 1990-91peaks (Krebs et al., 1986, 1992). In creating this model Ihave assumed a simple relationship between hares andvegetation, in which hare survival is related to browseavailability. I have not incorporated any of the proposedhypotheses dealing with changes in plant secondary compoundsor browse structure (Bryant et al., 1983; Fox and Bryant,1984) into my model. Smith et al. (1988) and Sinclair et100al. (1988) working in the Kluane area found no evidence tosupport the secondary compound or browse structurehypotheses. As a result, I have not concerned myself withincorporating these aspects into my model.In all simulations I have assumed that the reproductionand juvenile survival rates I observed on Jacquot Island in1991 (see Chapter 1) were constant from year to year. Caryand Keith (1979) found that reproduction varies with thepopulation density of hares in a cyclic population. Whetherthese changes were a result of hare densities or a result offood availability was unclear. In my model, reproductionwas reduced if browse availability was reduced, but I haveadded no density dependent reproductive parameter. Furtherwork examining the effects of hare density on reproductionwould greatly enhance our ability to create accuratesimulation modelsModel 2I was unable to create a simulation model that closelyapproximated the observed dynamics of Jacquot Island usingonly stochastic predation. It appears from the results thatsome kind of regulation is required in order to maintain theJacquot Island population. The addition of browse dependentdispersal helped to create simulations that more closelyapproximated the observed dynamics. However, large peaks ofhares still occurred occasionally (Figure 3.1 and Figure3.2). These peaks, of up to 13 hares/ha, were well above101the maximum observed to occur on Jacquot Island (see Chapter1 ).Boutin et al. (1985) found that dispersal rates frommainland study grids during the cycle were not large enoughto account for significant changes in demography. Inaddition, Boutin et al. (1985) found that the highestdispersal rates occurred during peak and early declinephases of the cycle. It follows, then, that increasingdispersal rates from Jacquot Island in order to decrease themaximum densities reached in the simulation may bebiologically flawed. It appears from the results of Model 2simulations that some other form of regulation is requiredin order to simulate the Jacquot Island population moreaccurately. This regulation may take the form of some typeof density-dependent behavioural interaction, which I didnot include in the model. Boutin (1984) found that thepresence of conspecifics had a negative impact on juvenilehare survival rates. Graf (1985) observed dominancehierarchies of snowshoe hares at experimental feedingstations. Results of Boutin (1984) and Graf (1985) suggestthat behavioural interactions may play an important role inthe dynamics of a population. The results of Model 2simulations also suggest that social behaviour may need tobe added in order to explain the observed hare demographicson Jacquot Island.102Model 3The impact of weather on juvenile hares in the modelwas estimated arbitrarily, based solely on the exposuredeaths I observed on Jacquot Island in 1991 (see Chapter 1).With the addition of these crude estimates, however, I wasable to produce simulations that closely approximated theobserved population dynamics on Jacquot Island (Figure 3.1and Figure 3.3).Model 3 supports my speculation in Chapters 1 and 2about the importance of weather in influencing the JacquotIsland hare population. With the addition of the stochasticweather effects, fewer high density hare outbreaks occurred.It makes intuitive sense that if additional stochasticfactors such as winter weather (snowfall, temperature), andtime of spring snowmelt, were also added to the model theprobability of a high density hare outbreak occurring wouldbe reduced further.If enough stochastic factors were added to the modelthe probability of a high density outbreak occurring wouldsoon become zero. The addition of stochastic factors wouldthen eliminate the need for density-dependent factors.3.9) CONCLUSIONSGiven the assumptions made in Model 1, it appears thatthe resulting simulations support the hypothesis that theJacquot Island hare population would become cyclic if it103experienced the same predation pressure (Type II functionalresponse and delayed density-dependent numerical response)as observed on the mainland areas. This result supports theview that predation is a necessary part of the hare cycle(Keith et al., 1984; Boutin et al., 1986; Trostel et al.,1987; Sinclair et al., 1988). Specifically the results ofthis study show that a delayed density-dependent numericalresponse of predators may be a necessary part of the 10-yearcycle of snowshoe hares.Results from Model 2 simulations do not support thehypothesis that stochastic predation pressure is the maincause of the observed yearly fluctuations in hare numbers onJacquot Island, unless a dispersal function is added. Whendispersal is added the simulations of Model 2 correspondmuch better to the observed densities on Jacquot Island.However, high density outbreaks still occur from time totime in the simulations. With the addition of weather intothe model (Model 3), the simulations conform very well toobserved densities on Jacquot Island.These models have made a number of assumptions aboutthe Jacquot Island hare population that have yet to betested:1) Reproductive and juvenile survival rates remainrelatively constant with respect to hare densities.2) Dispersal of hares from Jacquot Island is relatedto browse availability only, not to hare densities.1043) Deaths of newborns caused by weather are inaddition to deaths that would be caused by predation and arenot compensatory.4) Browse growth and availability are the same onJacquot Island as on mainland study areas.A much better understanding of the factors influencingthe Jacquot Island hare population will be gained once theseassumptions have been properly tested.1053.9) LITERATURE CITEDBoutin, S. 1984. The effects of conspecifics on juvenilesurvival and recruitment of snowshoe hares. Journal ofAnimal Ecology 53:623-637.Boutin, S., B.S. Gilbert, C.J. Krebs, A.R.E. Sinclair,J.N.M. Smith. 1985. The role of dispersal in thepopulation dynamics of snowshoe hares. CanadianJournal of Zoology 63:106-115.Boutin, S., C.J. Krebs, A.R.E. Sinclair, J.N.M. Smith.1986. Proximate causes of losses in a snowshoe harepopulation. Canadian Journal of Zoology 64:606-610.Bryant, J.P., F.S. Chapin III, D.R. Klein. 1983.Carbon/nutrient balance of boreal plants in relation tovertebrate herbivory. Oikos 40:357-368.Cary, J.R. and L.B. Keith. 1979. Reproductive changes inthe 10-year cycle of snowshoe hares. Canadian Journalof Zoology 57:375-390.Fox, J.F. and J.P. Bryant. 1984. Instability of thesnowshoe hare and woody plant interaction. Oecologia63:128-135.Graf, R.P. 1985. Social organization of snowshoe hares.Canadian Journal Of Zoology 63:468-474.Holling, C.S. 1959. The components of predation asrevealed by a study of small-mammal predation of theEuropean pine saw-fly. Canadian Entomology 91:293-320.Keith, L.B. 1974. Some features of population dynamics inmammals. Proceedings of the International Congress onGame Biology 11:17-58.Keith, L.B., J.R. Cary, O.J. Rongstand and M.C. Brittingham.1984. Demography and ecology of a declining snowshoehare population. Wildlife Monographs 90:1-43.Krebs, C.J., B.S. Gilbert, S. Boutin, A.R.E. Sinclair andJ.N.M. Smith. 1986. Population biology of snowshoehares. I. Demography of food-supplemented populationsin the southern, Yukon. Journal of Animal Ecology55:963-982.106Krebs, C.J., R. Boonstra, S.Boutin, M. Dale, K. Martin,A.R.E. Sinclair, J.N.M. Smith, R. Turkington.^1992.What drives the snowshoe hare cycle in Canada's Yukon?in Wildlife 2001: Populations. McCullough, D.R. andR.H. Barrett, eds.Nicholson, A.J. and V.A. Bailey. 1935. The balance ofanimal populations. Part I. Proceedings of theZoological Society of London, pp. 551-598.Pease, J.L., R.H. Vowles and L.B. Keith. 1979. Interactionof snowshoe hares and woody vegetation. Journal ofWildlife Management 43:43-60.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 foodlimitation. Journal of Animal Ecology 57:787-806.Smith, J.N.M., C.J. Krebs, A.R.E. Sinclair, and R. Boonstra.Population biology of snowshoe hares. II.Interactions with winter food plants. Journal ofAnimal Ecology 57:269-286.Trostel, K., A.R.E. Sinclair, C.J. Walters and C.J. Krebs.1987. Can predation cause the 10-year hare cycle?Oecologia 74:185-192.Vaughan, M.R. and L.B. Keith. 1981. Demographic responseof experimental snowshoe hare populations to overwinterfood shortage. Journal of Wildlife Management 45:354-380.Windberg, L.A. and L.B. Keith. 1976. Experimental analysisof dispersal in snowshoe hare populations. CanadianJournal of Zoology 55:2061-2081.107GENERAL CONCLUSIONSIn this study I have found a number of interestingdifferences between Jacquot Island and mainland study areas,and have proposed a number of new hypotheses to explain theobserved demographic differences.Survival rates of juvenile hares (<30 days of age) werefound to be much higher on Jacquot island as compared withthe mainland (Chapter 1). I hypothesize that the majorcause of the survival differences is the lack of smallmammalian predators on the island (Chapter 1). No Arcticground squirrels (Spermophilus parryi) and no weasels(Mustela spp.) have been observed on Jacquot Island, and redsquirrels (Tamiasciurus hudsonicus) were found only at lowdensities. O'Donoghue (1991) found that a minimum of 48.8%of radio-tagged juveniles that died in his study were killedby small mammals. The lack of small mammalian predators onJacquot island constitutes a major difference between islandand mainland study areas.Large mammalian and avian predation pressure appearedto vary between years on Jacquot Island. However, with onlytwo years of data it is impossible to know if this variationis a common occurrence on the island. I hypothesized thatstochastic variation in predation pressure from year to yearon Jacquot Island is the major cause of the non-cyclicfluctuations in hare abundance. In Chapter 3 I tested this108hypothesis along with the hypothesis that if Jacquot Islandexperienced the same kind of predation pressure as observedon the mainland (Type II functional response and 1-year-delayed density-dependent numerical response) it wouldresult in a cyclic hare population. Using simulation modelsI found that stochastic predation alone was unable toexplain the observed dynamics of Jacquot island. It appearsthat some form of density-dependent behaviour may beregulating the hare population. A number of previousstudies have indicated the importance of social behaviour ininfluencing hare populations (Christian and Davis, 1964;Boutin, 1984; Graf, 1985). Further research examiningsocial interactions of hares at different densities onJacquot Island would provide insight into the demography ofthe hares.Through the use of simulation models I found thatJacquot Island would cycle if it experienced the samepredation pressure as mainland areas (Chapter 3). Increating the model I made a number of assumptions aboutreproduction and survival rates of the hares, as well asassumptions about browse availability, requirements andgrowth on Jacquot Island (Chapter 3). If the assumptions ofthe model are not too far from reality, the results of thesimulations support the view that predation is a necessaryand important part of the hare cycle (Keith et al., 1984;Boutin et a/., 1986, Trostel et al., 1987). Future research109to test those assumptions would be valuable in increasingthe usefulness of computer models in predicting the factorsinfluencing hare populations.A number of habitat differences were found betweenJacquot Island and mainland study sites. Among the moreimportant differences were the lack of mature spruce trees(Picea glauca), and the increased density of understory onJacquot Island sites. Studies by Jones (1975) and Bunnell(1979) have shown that mature trees are very important inthe winter range of black-tailed deer (Odocoileus hemionuscolumbius). The large canopies of mature conifers act tomoderate the effect of snowfall, wind, rain and coldtemperatures. To what extent mature spruce trees may servethe same function for snowshoe hares is unknown. Hares inthe Kluane area have been observed using the depression inthe snow near the base of mature trees (A.R.E. Sinclair,pers. comm.). Whether the hares use this depression forcover from predators or thermal protection or both, is notknown.Weather effects resulted in a number of newborn deathsin 1991, and I have speculated that heavy snowfall in thewinter of 1992 may have influenced reproduction in 1992(Chapter 1). Without the moderating effects of maturespruce trees on the island, weather may be having a greaterimpact on the hare population than observed on the mainlandareas.110The possibility of weather affecting the Jacquot Islandhare population in a significant way was examined using asimulation model (Chapter 3). I found that stochasticweather effects were capable of influencing the populationdynamics of hares on Jacquot Island. Owing to thestochastic nature of weather patterns, it is unlikely thatdifferences in weather patterns from year to year couldexplain the hare cycle. However, the degree of impactweather has on a population could explain the lack of acycle.Jacquot Island study sites were found to haveunderstories with more dense cover than mainland study sites(Chapter 2). The differences in understory were found to beassociated with survival rates of juvenile hares. Leveretsborn in areas of more dense understory appear to have higher30-day survival rates. The importance of understory coverfor juveniles less than 30 days of age has not been examinedpreviously. A more detailed study of the associationbetween cover and leveret survival would provide informationon the importance of habitat characteristics in influencinghare demography.I undertook this study in order to examine five mainhypotheses dealing with the causes of the observeddifferences in demography between Jacquot Island andmainland hare populations.H 1 : Predation pressure differs between the mainland111and Jacquot Island areas.H2•- Predators of juvenile hares are at lower densitieson Jacquot Island as compared with the mainland.H.3' Owing to the distance between Jacquot Island andthe mainland, immigration of hares to the islandoccurs less often than emigration.H4: Hares living in a habitat with dense cover sufferlower predation rates and Jacquot Island has moresuch "refuge habitat" than the mainland.H5: Food availability is a limiting factor on JacquotIsland.No evidence was found to support the dispersalhypothesis (113) or the food limitation hypothesis (H5).However, I do not feel that the dispersal hypothesis wasappropriately tested. In order to examine dispersalproperly a winter study is required. Winter work wouldallow radio-collared hares to be monitored more closely toreduce the likelihood of losing contact with transmitters.The rejection of the food limitation hypothesis issupported by previous work by Krebs et al. (1986), who foundthat the hare population on the south end of Jacquot Islanddid not respond in the same manner as mainland areas to theaddition of rabbit chow as a source of winter food.However, the possibility of short-term food shortage or theshortage of important trace elements in the diet still needsto be addressed. Little information is known on what112natural foods are required by hares and in what quantitiesto allow normal reproduction and survival.Hypotheses 1, 2 and 4 were all supported by results ofthis study. These results are some of the first differencesto be documented between Jacquot Island and mainland areas(see Trostel, 1986), and lay the ground work for futurestudy. The influence of dispersal and food limitation onthe Jacquot Island hares must still be examined. As well, along-term study of Jacquot Island, on predation,reproduction and juvenile survival would provide importantinformation which must be obtained to gain a betterunderstanding of factors influencing this non-cyclic harepopulation.1133.10) LITERATURE CITEDBoutin, S. 1984. The effects of conspecifics on juvenilesurvival and recruitment of snowshoe hares. Journal ofAnimal Ecology 53:623-637.Boutin, S., C.J. Krebs, A.R.E. Sinclair, J.N.M. Smith.1986. Proximate causes of losses in a snowshoe harepopulation. Canadian Journal of Zoology 64:606-610.Bunnell, F.L. 1979. Deer-forest relationships on northernVancouver Island. Pp. 86-101 in Sitka black-taileddeer. USDA Forest Service, Alaska Region, series No.R10-48. Wallmo O.C. and J.W. Schoem editors.Christian, J.J. and D.E. Davis. 1964. Endocrine, behaviorand population. Science 146:1550-1560.Graf, R.P. 1985. Social organization of snowshoe hares.Canadian Journal of Zoology 63:468-474.Jones, G.W. 1975. Aspects of winter ecology of black-tailed deer (Odocoileus hemionus columbius Richarson)on northern Vancouver Island. M.Sc. Thesis, Universityof British Columbia, Vancouver, B.C.Keith, L.B., J.R. Cary, O.J. Rongstand and M.C. Brittingham.1984. Demography and ecology of a declining snowshoehare population. Wildlife Monographs 90:1-43.O'Donoghue, M. 1991. Reproduction, juvenile survival andmovements of snowshoe hares at a cyclic populationpeak. M.Sc. Thesis, University of British Columbia,Vancouver, B.C.Trostel, K. 1986. Investigation of causes of the 10-yearhare cycle. M.Sc. Thesis, University of BritishColumbia, Vancouver, B.C.Trostel, K., A.R.E. Sinclair, C.J. Walter and C.J. Krebs.1987. Can predation cause the 10-year cycle?Oecologia 74:185-192.114APPENDIX 1.1. Jolly-Seber Estimates of Hare Numbers onJacquot Island and Mainland Study Areas (95% C.I. inparentheses).JacquotNorth1991 1992SpringFallWinterSpring36363260(34-38)(35-37)(31-33)(57-63)1811(17-19)(4-18)22 (21-23)JacquotFall 147 (142-152) 11 (3-19)SouthWinter 26 (25-27)Spring 54 (46-62) 21 (17-35)Mainland 1 Fall 70 (60-80) 6 (5-7)Winter 15 (14-16)Spring 48 (36-60) 14 (13-15)Mainland 2 Fall 58 (43-73) 7 (6-8)Winter 14 (13-15)Effective Trapping Area of GridsJacquot North 35 haJacquot South 34 haMainland 1 60 haMainland 2 60 ha115APPENDIX 1.2. Jacquot Island Litter Sizes (number dead atbirth in parentheses) and Dates of Birth Obtained fromFemales Confined in Maternity Cages.Jacquot North Jacquot South2^May 19 3^May 244 May 23 4(2) May 244 May 23 5 May 245 May 24 4 May 264 May 25 3 May 26Litter 1 1(1) May 26 2 May 261991 3(3) May 26 4 May 282 May 27 3(3) May 284 May 29 5(5) May 282 May 31 3 May 284 May 31 3 May 293 Jan 16 June 25 6 June 284 June 28 3 June 285 June 29 6 June 297 June 29 6 July 25 June 29 6 July 35 July 1 5 July 3Litter 2 3(3) July 1 4 July 31991 3 July 1 6 July 66 July 1 6 July 84 July 31 July 55 July 64 July 77 July 84 Aug 8Litter 3^ 6 Aug 101991 3^Aug 115 May 20Litter 1^ 3(3) May 241992 5(5) May 26Litter 21992Litter 319924(4) June 274(4) July 16 June 294 July 11 July 1 5 Aug 6116


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