Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Factors limiting moose numbers and their interactions with elk and wolves in the central Rocky Mountains,… Hurd, Thomas E. 1999

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

Item Metadata


831-ubc_1999-0527.pdf [ 6.83MB ]
JSON: 831-1.0099324.json
JSON-LD: 831-1.0099324-ld.json
RDF/XML (Pretty): 831-1.0099324-rdf.xml
RDF/JSON: 831-1.0099324-rdf.json
Turtle: 831-1.0099324-turtle.txt
N-Triples: 831-1.0099324-rdf-ntriples.txt
Original Record: 831-1.0099324-source.json
Full Text

Full Text

FACTORS LIMITING M O O S E N U M B E R S A N D THEIR INTERACTIONS WITH E L K A N D W O L V E S IN T H E C E N T R A L R O C K Y M O U N T A I N S , C A N A D A .  by Thomas E. Hurd B.Sc, University of British Columbia, Vancouver, 1983  A THESIS SUBMITTED FN P A R T I A L F U L F I L L M E N T OF T H E REQUIREMENTS FOR T H E D E G R E E OF M A S T E R OF SCIENCE  in T H E F A C U L T Y OF F O R E S T R Y (Department of Forest Sciences)  We accept this thesis as conforming to the required standard  T H E UNIVERSITY OF BRITISH C O L U M B I A July 1999 © Thomas E. Hurd 1999  UBC Special Collections - Thesis Authorisation Form  In p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t o f the r e q u i r e m e n t s f o r an advanced degree a t the U n i v e r s i t y o f B r i t i s h Columbia, I agree t h a t the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r r e f e r e n c e and s t u d y . I f u r t h e r agree t h a t p e r m i s s i o n f o r e x t e n s i v e c o p y i n g o f t h i s t h e s i s f o r s c h o l a r l y purposes may be g r a n t e d by the head o f my department o r by h i s o r her r e p r e s e n t a t i v e s . I t i s u n d e r s t o o d t h a t copying or p u b l i c a t i o n of t h i s t h e s i s f o r f i n a n c i a l gain s h a l l not be a l l o w e d w i t h o u t my w r i t t e n p e r m i s s i o n .  Department o f The U n i v e r s i t y o f B r i t i s h Columbia Vancouver, Canada  1 ofl  18/08/98 11:29 AM  FACTORS LIMITING MOOSE NUMBERS AND THEIR INTERACTIONS W I T H E L K A N D W O L V E S IN T H E C E N T R A L R O C K Y MOUNTAINS, CANADA  ABSTRACT  Numbers of moose declined in Banff National Park between the mid-1940's and present. This coincided with increases in abundance of both elk and wolves. To explain the decline, I tested several predictions related to exploitative competition with elk, and apparent competition mediated by wolves. The relationship I found between moose and elk winter pellet abundance was consistent with the inverse density pattern expected if exploitative competition and/or apparent competition were operative. Interactions between the two cervids and a shared food resource - willow - supported several of the key criteria required to demonstrate exploitative competition. The two cervids overlapped in resource use, and use by elk probably reduced resource availability to moose. The large asymmetries in distribution, abundance, diet breadth, diet overlap, and browse utilization provided the conditions necessary for elk to dominate competitive interactions. Interactions between predators (wolf and grizzly bear) and the two cervids revealed the potential for apparent competition where the impact on secondary prey, moose, was greater than on primary prey, elk. Predation rates on radio-collared moose appeared unsustainable. The two forms of "competition" operated at different scales, and appeared additive in their negative effect on moose. Exploitative competition was localized in areas of high elk density and apparent competition (predation) operated throughout the study area. My findings contrast with the usual study of mixed competition-predation systems where the two factors provide a mechanism for coexistence.  Table of Contents  Abstract Table of Contents List of Tables List of Figures Acknowledgments 1.0 2.0  3.0  INTRODUCTION  ii iii v vi vii 1  METHODS 2.1 STUDY AREA 2.2 CERVID A B U N D A N C E A N D DISTRIBUTION 2.2.1 Pellet group surveys 2.2.2 Variation in pellet abundance 2.3 INTERSPECIFIC ASSOCIATION A N D C O V A R I A T I O N OF CERVIDS 2.3.1 Interspecific association 2.3.2 Interspecific covariation 2.4 CERVID-RESOURCE O V E R L A P A N D BROWSE RELATIONSHIPS 2.4.1 Diet and diet overlap 2.4.2 Cervid-browse relationships 2.4.3 Ungulate-vegetation overlap 2.5 CERVID-WOLF RELATIONSHIPS 2.5.1 Moose capture and monitoring 2.5.2 Moose population parameters 2.5.3 Elk dominance, survival, and population trends  5 5 8 8 10  RESULTS 3.1 INTERSPECIFIC ASSOCIATION A N D C O V A R I A T I O N OF CERVIDS 3.1.1 Interspecific association 3.1.2 Interspecific covariation 3.1.3 Underlying univariate relationships 3.2 C E R V I D - R E S O U R C E O V E R L A P A N D BROWSE RELATIONSHIPS 3.2.1 Diet 3.2.2 Diet overlap 3.2.3 Browsing rates 3.2.4 Ungulate-vegetation overlap 3.3 CERVID-WOLF RELATIONSHIPS 3.3.1 Moose mortality and reproduction 3.3.2 Wolf population trend 3.3.3 Elk dominance and population trend 3.3.4 Wolfpredation  21  11 11 11 14 14 15 16 17 17 17 19  21 20 22 27 32 32 32 35 40 42 42 45 47 49  iii  4.0  DISCUSSION 4.1 E V I D E N C E OF E X P L O I T A T I V E COMPETITION 4.1.1 Interspecific association 4.1.2 Interspecific covariation: the inverse density pattern 4.1.3 Diet overlap and resource limitation 4.1.4 Vegetation overlap and the importance of willow 4.1.5 Theoretical outcomes of competition 4.1.6 Other impacts on browse resources 4.1.7 Variation in diet, diet overlap, and snowpack 4.2 E V I D E N C E OF A P P A R E N T COMPETITION 4.2.1 Moose and elk survival 4.2.2 Predator-cervid relationships 4.2.3 Contributing mortality factors 4.3 INTERACTIONS B E T W E E N COMPETITION A N D A P P A R E N T COMPETITION 4.3.1 Scale 4.3.2 Competitive ability and predation risk  51 51 51 52 53 54 55 56 57 58 58 60 61 63 63 64  5.0  M A N A G E M E N T IMPLICATIONS  66  6.0  R E F E R E N C E S CITED  68  Appendices  79  List of Tables Table 3.1  Table 3.2 Table 3.3 Table 3.4  Table 3.5  Classification and regression tree (CART) analysis of the relationship between moose abundance, and environmental and biological predictor variables, Banff National Park, 1978-1979 and 1995-1997 Winter diets of moose and elk on sympatric range in the Bow Valley, Banff National Park, 1995-1996 Areal extent of 7 vegetation types sampled in the 1996 pellet survey, Banff National Park Annual mortality and survival calculations for 45 radio-marked moose in Banff National Park and Peter Lougheed Provincial Park, 1994-1997 Observation and survival of moose calves, Banff National Park and Peter Lougheed Provincial Park, 1994-1997  24 33 40  43 46  V  List of Figures  Figure 2.1 Figure 3.1  Figure 3.2  Figure 3.3  Figure 3.4  Figure 3.5  Figure 3.6  Figure 3.7 Figure 3.8 Figure 3.9 Figure 3.10  Figure 3.11  Figure 3.12  Study area Classification and regression tree analysis of the relationship between moose abundance and environmental and biological predictor variables, Banff National Park: 1978-1979 pellet survey model Classification and regression tree analysis of the relationship between moose abundance and environmental and biological predictor variables, Banff National Park: 1996 pellet survey model Classification and regression tree analysis of the relationship between moose abundance and environmental and biological predictor variables, Banff National Park: 1995/97 browse plot model , One-cycle partitioned regression of moose and elk covariation based on pellet counts in willow-bearing vegetation types, Banff National Park, 1978-1979 and 1995-1997 Mean ( + SE) and median moose pellet density on willowbearing vegetation types in the Bow, Cascade, and Red Deer Valleys, 1978-1979 and 1996 Percentage mean ( ± SE) relative density of plant fragments per individual, and percentage relative frequency of plant genera per species, in moose and elk winter diets, Bow Valley, Banff National Park, 1995-1996 Percent willow twigs browsed relative to elk pellet density, Banff National Park, 1995 and 1997 Percent willow twigs browsed relative to moose pellet density, Banff National Park, 1995 and 1997 Mean willow twig diameter at the bite-point relative to browsing intensity, Banff National Park, 1995 and 1997 Mean pellet group density ( + SE), and median rank of ecosite importance, for 7 vegetation types, Banff National Park, 1996 pellet survey Mean ( ± SE) and median moose, elk, and deer abundance on willow-bearing vegetation types in the Bow, Cascade, and Red Deer Valleys, Banff National Park, 1978-1979 and 1996 Approximate distribution of wolf pack territories, elk winter concentrations, and radio-collared moose kill sites in the study area, 1994-1997  6  25  26  27  29  30  34 37 38 39  41  48  50  vi  ACKNOWLEDGMENTS  I am indebted to Cliff White and Bob Haney of Parks Canada for permitting me to undertake this study. Dave Dalman, Perry Jacobson, Terry McGuire, and Jillian Roulet of Parks Canada also provided key support. I thank Fred Bunnell, my supervising professor, for allowing me to pursue a broad topic of interest (some might say too broad). He is a superb teacher and I am grateful to him. Committee members Peter Achuff, David Shackleton, and Tony Sinclair each provided important input at critical junctures in the study. In addition to my committee, Alton Harestad of Simon Fraser University reviewed and substantially improved the thesis. Helpful reviews were also furnished by Mark Edwards, David Hamer, Bronwen Jones, Paul Paquet, Ian Pengelly, Nigel Waltho, and Cliff White. Funding and in-kind support was provided by Parks Canada, Alberta Environmental Protection, Alpine Helicopters, and Human Resources Development Canada. Martin Urquhart provided valuable assistance in every aspect of the field work. This study benefited from his dedication. Mike Vassal, Gail Moir, Joanne Saher, and Thea Mitchell braved the elements, including some of the severest winters on the record, to collect data. Additional field assistance was provided by Danah Duke, Jattinder Gill, Cliff Nietvelt, John Paczkowski, Melanie Percy, Dan Tomlinson, and a number of other capable souls. Steve Donelon supported and administered the study in Peter Lougheed Provincial Park. Todd Shury of Shury Veterinary Services provided expert capture and necropsy assistance. Chief Pilot Lance Cooper of Alpine Helicopters made moose capture in rugged terrain appear routine, and Mike Dupuis of Wildlife Observation Air Service safely flew his telemetry aircraft in a full range of mountain weather conditions. GIS assistance was provided by Scott Jevons of Geo works, and Darrel Zell of Parks Canada. I am grateful to my wife, Bronwen, and my children, Eric and Gavin, for their encouragement and support in this lengthy endeavour. I couldn't have done it without them.  1.0  INTRODUCTION  Moose, Alces alces, were relatively common in Banff National Park (BNP) from the time of the first formal game surveys in the 1940's (Cowan 1943, 1946; Green 1950, 1955; Mair 1952) to the early 1970's (Holroyd and Van Tighem 1983; Woods 1990). They began to decline during the 1970's (Holroyd and Van Tighem 1983) to the point where they were observed rarely in annual aerial and ground surveys between 1985 and 1998 (Woods 1990; Parks Canada files). The period of moose decline coincided with an increase in the number of elk, Cervus elaphus, (Woods 1991; Skjonsberg 1993; Morgantini 1995; Parks Canada files) and wolves, Canis lupus, (Gunson 1992; Paquet 1993) following the cessation of elk culling in 1969, and of province-wide wolf control in 1956. The period of moose decline also coincided with: 1) increased prevalence of giant liver fluke (GLF), Fascioloides magna, infection, a parasite well tolerated by elk but potentially harmful to moose (Butterworth and Pybus 1993); 2) habitat change caused by fire suppression (Holroyd and Van Tighem 1983; Achuff et al. 1996); and, 3) increased road and railway mortality (Holroyd 1979; Woods 1990). Elk may be directly or indirectly implicated in the decline of moose through the effects of competition, predation, and parasite infection. In this study, I focus on moose interactions with elk and wolves as a potential explanation of moose decline, but I also attempt to assess the importance of the alternative factors listed above.  Competitive interactions between elk and moose, or between elk and other ungulates, have long been a topic of interest in B N P (Green 1949; Tanner 1950; Cowan 1950; Mair 1952; Flook 1964) and in other Rocky Mountain national parks (Jasper, Cowan 1947; Yellowstone, McMillan 1953, Singer and Norland 1994; Glacier, Singer 1979, Jenkins and Wright 1988). Earlier studies examined competitive interactions involving pairs or guilds of Rocky Mountain ungulates, and their exploitation of shared foods and habitats. However, the influence that predators may have had on potential competitors was rarely  1  considered. This was due, in part, to the previous reduction or elimination of major predators from many of the study areas.  Prey species that share predators may interact in an indirect manner known as "apparent competition", a predator-mediated form of competition (Holt 1977). The process occurs when a food-limited predator responds numerically to an abundant prey and causes an alternate prey to decline. Competition and apparent competition need not be considered mutually exclusive. Holt et al. (1994) suggested that when the two effects do mix, the outcome often promotes coexistence between competitors because of trade-offs between a species' competitive ability and its vulnerability to predation.  Understanding the importance of competitive effects between herbivores is important for park managers because there is debate about the importance of the ecological effects of high elk densities in many of the Rocky Mountain national parks (e. g., Kay 1990; Wagner et al. 1995; White et al. 1998; Huff and Varley 1999; Wright 1999). The debate is most pronounced in parks where major predators have either been eliminated or suffer disrupted access to prey through the effects of landscape fragmentation. Many authors have suggested that effects related to intensive elk herbivory are an appropriate outcome of national park policies, while others have argued that herbivory-related impacts to animal and plant communities contradict park policy. White et al. (1998) characterized the positions according to two well known conceptual models of large mammal community organization. A "bottom-up" food-regulation model and a "top-down" predatorregulation model. Each alternative implies different assumptions about the carrying capacity of dominant herbivores, like elk. Should this study demonstrate negative interactions between moose and elk, their future coexistence may depend, to some extent, on the model that park managers choose to accept.  I examined two potential outcomes of high elk density and intensive herbivory on the relationship between moose and elk. I hypothesized that moose interact with elk through the mechanisms of exploitative competition, mediated by winter browse resources, and by  2  apparent competition, mediated by wolves. Because of the elk's generalist feeding behaviour (Houston 1982) and numeric dominance, I further hypothesized that elk dominate the outcome of such interactions. Competitive interactions were examined in the winter season when resources are most limiting, ungulates are spatially confined by snow (Telfer 1970, 1977), and competitive effects are expected to be greatest (Jenkins and Wright 1988).  To test for the presence of exploitative competition, I used a subset of the criteria listed by Wiens (1994) in order of increasingly supportive evidence of competition. If species were competing one should expect to find the following general predictions to be true (after Wiens 1994): 1) a negative (inverse) relationship exists in the distribution or abundance patterns between species; 2) species overlap in their use of resources; and, 3) a hrnitation in resources available to one species is caused by another. Accordingly, I tested the following specific predictions: la) moose and elk are found in association (co-occur) on a shared habitat type less often than would be expected by chance (an inverse association pattern); lb) moose and elk densities on a shared habitat type are negatively correlated (an inverse density pattern); 2) the two species overlap in their use of winter food resources; 3 a) their overlap in food resource use is asymmetrical, the generalist elk overlaps the specialist moose to a greater extent than moose overlap with elk; and, 3b) browsing by elk limits the browse resources available to moose.  Apparent competition may result in the same spatial patterns that are suggestive of exploitative competition (predictions l a and lb above). Other general predictions associated with this process are as follows (after Holt 1977; Holt et al. 1994): 1) predators respond numerically to prey; and, 2) the impact of predation is sufficient to cause one prey species to decline relative to another. The numerical response of wolves to prey density has been well documented in other studies (Packard and Mech 1980; Fuller 1989; Messier 1994). In multi-prey systems, such as in this study, wolf numeric responses are most likely linked to the most profitable prey species, where profitability depends on prey size, density, and vulnerability (Messier 1995). Elk prey are considered highly  3  profitable to wolves in the Rocky Mountains because they are usually abundant, and their vulnerability is considered similar to that of smaller prey species like deer, and greater than that of larger more defensive prey like moose (Huggard 1993a; Weaver 1994; Kunkel 1997). Consequently, wolf numbers should be expected to respond primarily to elk density where elk dominate a prey guild. If apparent competition were an important moose-elk interaction, the following predictions, in addition to the inverse spatial patterns predicted in l a and lb (above), would be expected to be true: 1) elk are the most abundant prey, 2a) wolf predation is the primary cause of moose mortality; and, 2b) wolf predation is sufficient to cause moose to decline relative to elk.  4  2.0  2.1  METHODS  STUDY A R E A  Field work was conducted between 1994 and 1997 in Banff National Park (BNP) and Peter Lougheed Provincial Park (PLPP), Alberta (Figure 2.1). These parks are characterized by rugged mountain slopes, steep sided ravines, and flat valley bottoms. Elevations range from 1350 to 2800 m and generally increase from east to west. The climate is predominantly continental with long cold winters and short cool summers (Janz and Storr 1977). Average annual precipitation is 45 cm in the lower montane regions and 76 cm in the higher subalpine regions (McGregor 1984). Winter snow accumulation is directly related to elevation. Snow depths at a centrally located subalpine weather station (Mt. Norquay, 1740 m, Appendix II) reached over 100 cm in 1995 and 1996, and remained just under 75 cm in 1994 and 1997.  The vegetation of B N P has been described and mapped (Holland and Coen 1982) and is similar to types found in PLPP. Vegetation in the montane zone (< 1500 m) is characterized by lodgepole pine forest, aspen and Douglas fir stands, and dry grasslands. Several extensive wetlands occur along floodplains and are characterized by white spruce forest, willow stands, and sedge meadows. Shallow snow conditions (usually < 50 cm), and mild temperatures in this zone permit important winter ungulate habitat. Vegetation in the lower subalpine (1500 - 2000 m) is characterized by extensive lodgepole pine forests and by Engelmann spruce and subalpine fir forests. Snow cover is frequently shallow, permitting ungulates to winter in this zone, however, it is highly variable and can cause die-offs in severe winters (Holland and Coen 1982; Holroyd and Van Tighem 1983) The upper subalpine (2000 - 2300 m) is characterized by open-canopied Engelmann spruce and subalpine fir forests. Snow accumulation is heavy and winters are too severe  5  Figure 2.1. Study area. The study area was the east slope of the Central Rocky Mountains mainly within Banff National Park and Peter Lougheed Provincial Park. The dark-green shaded areas represent intensive study nodes where diet analyses, browse assessments, and pellet-counts were conducted.  for ungulates, except on steep, wind-blown slopes (Holroyd and Van Tighem 1983). Floodplain complexes in the subalpine are less prevalent, and wetlands less productive, than in the montane zone.  The ungulate-predator assemblage in the study area is complex and typical of the Rocky Mountain region. The ungulate community consists of moose, elk, white-tailed deer, Odocoileus virginianus, mule deer, O. hemionus, bighorn sheep, Ovis canadensis, mountain goat, Oreamnos americanus, and a small isolated population of woodland caribou, Rangifer tarandus. Moose habitat overlaps most extensively with the habitat of elk and deer, thus the study was limited primarily to these 3 cervids. Important ungulate predators include the wolf, grizzly bear, Ursus arctos, black bear, U. americanus, and mountain lion, Puma concolor. Human use in most of the study area is restricted to nonmotorized backcountry recreational activities. However, intensive human use occurs in the Bow Valley, and an elaborate irifrastructure includes a major highway, railway, two towns (Banff and Lake Louise), and several outlying visitor facilities.  Four sub-areas were selected for intensive study (Figure 2.1) based on relative differences in the number of wintering elk. They are listed in descending order of elk abundance as follows: 1) Bow Valley, 2) Panther Valley, 3) Red Deer Valley; and, 4) Cascade Valley. The Bow Valley is characterized as montane, whereas each of the other sub-areas is classified as subalpine. Because of the position of the Red Deer and Panther valleys in the eastern-most "front" ranges, winter conditions are milder than in other subalpine areas.  7  2.2  CERVID A B U N D A N C E AND DISTRIBUTION  2.2.1  Pellet Group Surveys  Pellet group surveys were used to assess the relative use of willow-bearing vegetation types by moose, elk, and deer. Two survey methodologies were employed, with each representing a different spatial scale. The 1996 pellet survey included long (1-km) transects and was representative of pellet abundance within each study sub-area. The 1995/97 browse plot survey involved shorter (270-m) transects and was representative of pellet abundance at browse feeding sites.  1996 Pellet Survey One hundred and fifty-one 1 -km transects were completed, 96 in the Bow Valley, 25 in the Cascade Valley, and 30 in the Red Deer Valley. Each 50-m segment of a transect in a distinct vegetation type was recorded as a separate sample. Transect segments were truncated when a road, river, or lake was encountered. Transects were stratified by subarea (3 valleys, proportional allocation of sampling; Krebs 1989) and by vegetation type (7 types, optimal allocation based on initial estimates of variance and sampling cost in each stratum; Krebs 1989) as described by Huggard (1993a). Sample allocation was confined to forest habitats below the alpine zone where lighter, more predictable, snow accumulations result in more consistent use by cervids (Section 2.1; Holroyd and Van Tighem 1983).  Transect surveys were conducted between mid-April and mid-June, as soon as possible after the snow had melted. A l l transects ran north-south, a bearing neither perpendicular nor parallel to the main orientation of the valleys, and each began at stratified random map coordinates. A l l pellet groups with centers within 1 m of the transect line were recorded as moose, elk, deer, or sheep. Pellet groups of the two deer species could not be distinguished. Elevation above sea level, and aspect were recorded for each 50-m segment. To analyse cervid abundance relationships within similar habitats, contiguous  8  transect segments within willow-bearing vegetation types (Appendix I) were aggregated into 236 transect segments of a mean length of 212 m (range: 121-250; mean segment area: 424 m ). 2  1995/97 Browse Plot Survey Nineteen pellet samples were obtained in association with a browse assessment survey (Section 2.4, below) as follows: 7 in the Bow Valley, 5 in the Cascade Valley, 3 in the Panther Valley; and, 4 in the Red Deer Valley. Each pellet sample consisted of the aggregation of 3 parallel 90-m pellet transects spaced 15-20 m apart. The samples were stratified by sub-area (4 areas, proportional allocation of sampling). Sample site allocation was confined to valley bottom flood plains or low benches less than 100 m above the valley bottom, and to previously mapped willow-bearing vegetation types (Appendix I).  Both the enumeration method and seasonal timing was as described in the 1996 pellet survey (above), except this survey was conducted during the springs of 1995 and 1997. Transect orientation was parallel to the long-axis of each willow-shrub meadow, or in the case of large meadows (> 90 m x 90 m), transects were placed perpendicular to any observed gradient in vegetation or soil moisture. Sample sites were randomly selected from willow-shrub meadows as identified on aerial photographs. Elevation and aspect were recorded for each sample (as above). In addition, variables representing genera- or species-specific shrub height, shrub density, relative twig density, and browsing intensity were recorded (Section 2.4).  1978-1979 Pellet Survey I used winter pellet-group data collected by Holroyd and Van Tighem (1983) between 1974 and 1980 to compare historical herbivore abundance and distribution to my 1996 pellet survey data. The selection of historical records was confined to sampling that took place within a confined period (1978-1979), in willow-bearing vegetation types, and vrithin the boundaries of the study sub-areas (Figure 2.1).  9  Holroyd and Van Tighem's (1983) initial sample plots consisted of 25 quadrats, each measuring 2-m by 5-m spaced 50 m apart, in 5 rows of 5 quadrats. Later, they doubled the area sampled by placing a second set of quadrats between quadrats of the first sample. They stratified their samples by ecosite-type (equal allocation) and by watershed (proportional allocation and subjective estimates of cost). They sampled through the summer months, but distinguished winter and summer pellet deposits based on whether they were congealed (summer) or separate (winter). They selected sample sites by using a random stratified procedure (G. Holroyd pers. comm.). The mean area of 40 historical sample plots used in my analysis was 468 m (range: 250 m - 500 m ). 2  2.2.2  2  2  Variation in Pellet Abundance  Pellet count methods have been criticized where they are used to derive population estimates (Fuller 1991a; White 1992; Timmerman and Buss 1997) and habitat use estimates (Collins and Urness 1979, 1981). Others have supported the use of the technique to measure relative abundance (Neff 1968; Oldemeyer and Franzmann 1981; Freddy and Bowden 1983), relative distribution (Edge and Marcum 1989), and relative use of habitats (Leopold et al. 1984; Loft and Kie 1988). A key assumption is that defecation rates are similar within and between species. Although reported defecation rates appear variable (moose: Franzmann et al. 1976a, 1976b; Joyal and Ricard 1986; elk: Neff et al. 1965; Collins and Urness 1979; deer: Neff 1968; Smith 1968; Collins and Urness 1981), the variation is frequently dependent on season (Rogers 1987; Collins and Urness 1979; Sawyer et al. 1990). Accordingly, I confined my analysis of pellet data to the relative distribution and abundance of ungulates in winter.  10  2.3  INTERSPECIFIC ASSOCIATION AND COVARIATION OF CERVIDS  The pellet survey data was used to examine the affiriity of moose, elk, and deer in a twostep analysis (Ludwig and Reynolds 1988) as follows: 1) the presence and absence of pellets on sample transects were examined to determine if species were associated, and, if so, the degree of strength of the association; and, 2) the relation between cervid pellet densities was examined to determine if species abundance co varied, and, if so, the character of the relationship (direct or inverse). Pellet count data from each of the three survey-types (described above) were converted to common units of pellets per hectare prior to the analysis.  2.3.1  Interspecific Association  Following the approach of Ludwig and Reynolds (1988), simultaneous associations among moose, elk, and deer, were determined using the variance ratio (VR) method of Schluter (1984) and the test statistic, W. In addition, the presence of pairwise associations were determined by using 2 x 2 presence/absence tables with the chi-square test and Yates' continuity correction for low cell expectations. If species occurred together more often than expected under the assumption of independence, the association was positive; if their joint occurrences were less than expected, the association was considered negative. The degree of strength of pairwise associations were measured using the Ochai index of association (after Ludgwig and Reynolds 1988). The index varies between 0 (species never co-occur) and 1 (species always co-occur).  2.3.2  Interspecific Covariation  Multivariate Analysis Covariation in pellet density was examined using a multivariate "classification and regression tree" technique (CART; Brieman et al. 1984; SYSTAT 7.0, SPSS Chicago, IL.). I used the C A R T procedure because mixed sets of categorical and continuous  11  variables are more easily interpreted and discussed than in linear models, and because it has the inherent ability to detect contingencies in effects (interactions; Clark and Pregibon 1992). The objectives of the C A R T analysis were to: 1) distinguish whether moose abundance was dependent on elk abundance, among other species and habitat variables; and, 2) permit interpretation of the interactions between predictor variables.  Moose pellet density was the dependent variable in each of three C A R T analyses, one for each type of pellet survey (Section 2.2.1). The set of predictors common to each model included elk and deer pellet density, slope-aspect, elevation, and location. Several additional predictors were available for the browse survey model (Section 2.4, below). The C A R T process repeatedly divides and subdivides the set of observations into groups, choosing the best available explanatory variable at each decision point to yield maximum between-group diversity. The first variable splits the sample into two sub-groups each more homogeneous than possible than with any other split. Each of the sub-groups is split in turn, continuing until the stopping criteria are satisfied (O'Connor et al. 1996).  Splits were restricted to those that resulted in at least a 0.05 improvement in the proportional reduction in error (PRE), a "r -like" statistic (SPSS 1997). Stopping criteria 2  were invoked before any split resulted in a sub-group containing less than four samples. Pellet density variables were rank-order, or log transformed, if either transformation resulted in a better fit of the model to an a priori classification of the data (V. Lemay pers. comm.). Otherwise, the variables were left in raw form. SYSTAT displays C A R T results graphically, in the form of a decision tree, with accompanying diagnostics. The modelling process often stops before the complete set of predictors is used, a result characteristic of C A R T (Brieman et al. 1984).  Statistical inference for C A R T models is poorly understood (Clark and Pregibon 1992), thus there are no procedures to determine probabilities. However, each of the models was subject to a cross-validation procedure in which 90% of the original data were randomly re-sampled, without replacement, to produce a set of 20 test models. The resulting set of  12  test trees was assessed, to determine the stability of each split, with the intention of "praning back" to the most parsimonious selection of predictor variables necessary to obtain consistent results (sensu Clark and Pregibon 1992). The results of the crossvalidation procedure were expressed in terms of the proportion of splits, identical to those in the original model, that occurred in the set of test models.  Univariate Analysis To further explore relationships between moose pellet abundance and each predictor variable, I used univariate statistical procedures to compare the character and strength of each relationship, and to infer statistical significance.  The moose-elk density relationship was tested using the partitioned regression method of Thomson et al. (1996). They recommended the procedure as one possible method to test for the presence of correlations in bivariate scattergrams when the data appear as a "point cloud" proscribed by a discrete edge. Such distributions do not lend themselves to traditional linear or curvilinear correlation analysis, but their edges may contain useful ecological information (Thomson et al. 1996). Species abundance relationships often fall into this pattern (Ausperger 1996) because animal distributions are often spatially patchy, resulting in many points near the lower limits of one or both axes of a bivariate scattergram.  The points well inside the edge of the point scatter may represent a spatial pattern of competition in poor quality habitat, where a few individuals might compete for scarce resources. Alternatively, these points may represent habitats that are not fully utilized because of the action of predators or some other disturbance. At the outer edge of the point scatter, an inverse density pattern may represent competition where both species are more abundant because of higher quality habitat. Thus, points near the origin may be difficult to interpret, however, a discrete upper edge is less ambiguous, because it implies that species reach their maximum densities only when they are separate from one another and never when they are together.  13  Thomson et al. (1996) refer to point scatters with boundaries as "factor-ceiling distributions" (FCD). Using their method, I partitioned the F C D resulting from the relationship between moose and elk pellet counts into upper and lower subsets using linear regression. Similarly, a linear regression was applied to each subset. A Spearman's rank correlation coefficient (r ) was calculated for the upper subset of points to test for the s  presence of a correlation between species densities near the F C D boundary. Samples where both species were absent were excluded from the analysis because they did not contribute to the understanding of the bivariate density relationship (after Krebs 1989).  Relationships between moose abundance and other predictor variables were assessed using non-parametric tests (Kruskal Wallis, H; Mann Whitney, U ; Spearman's rank correlation, r ), as appropriate (Zar 1984). Moose were rare in the study area, consequently s  distributions of sample data were skewed toward zero values. Attempts to normalize the data using standard transformation procedures (Krebs 1989) were ineffective.  2.4  CERVID-RESOURCE O V E R L A P AND BROWSE RELATIONSHIPS  2.4.1  Diet and Diet Overlap  Moose and elk diets were approximated by microhistological analysis of fecal samples collected on sympatric range in the Bow Valley, from December through to early April in 1995 and 1996. Each sample consisted of 5-g of fresh fecal material from a single pellet group. The botanical composition of the diet was expressed in terms of frequency of occurrence, and percent relative density of identifiable plant fragments, based on 20 random microscope fields per sample (Composition Analysis Laboratory, Ft. Collins, Colorado). I used Levin's Standardized Measure (Krebs 1989) to measure niche breath (based on diet) and I tested the difference in the number of diet items per species using a Kruskal Wallis test. Diet overlap was calculated using several indices (Morisita's,  14  Simplified Morisita's, Horn's, and MacArthur and Levin's; after Krebs 1989) to enable direct reference to the variety of indices used in other diet overlap studies (cited below). Comparisons were restricted to principal forage species comprising more than 2% of a species' diet.  2.4.2  Cervid-Browse Relationships  Winter cervid-browse relationships were determined to assess the impact of cervid abundance on browse resources. Nineteen willow-shrub meadows were selected using a stratified random procedure (described in Section 2.2.1). Each sample consisted of 21 quadrats (3 rows of 7 quadrats) each measuring 1 m by 1 m spaced 15-20 m apart. The woody stems in each 1- m quadrat were classified according to species or genus, and 2  enumerated. The maximum height attained by each species also was recorded. A 3-m long stadia pole was placed upright at the center of each quadrat to systematically select twigs by first choosing the twig closest to the pole, then working outward to the next closest twig. This process was repeated twice in each of 6 height classes (20-49 cm, 5099 cm, 100-149 cm, 150-199 cm, 200-249 cm, and 250-299 cm), to obtain a 12-twig sample. If a height class did not contain any twigs, twig samples were re-allocated evenly across the remaining height classes until 12 twigs were obtained.  Each twig was classified as browsed or unbrowsed. A browsed twig was defined as a terminal branch that showed evidence of being eaten. A n unbrowsed twig was defined as an intact teirninal branch at least 3 cm long. The diameter at the basal node was measured on each twig, as was the tip diameter of unbrowsed twigs, and the diameter at the bitepoint of browsed twigs. Temporal replicates were obtained by repeating seven samples two years after the initial sampling. Stem density was calculated for each woody species or genus, and expressed in relative (percent of total stems) and absolute (stems/m ) terms. 2  Browse rates (browsed twigs/total browsed and unbrowsed twigs) were also calculated. The relationship between willow browse rates, bite-diameter and cervid abundance was  15  tested using linear regression (SYSTAT 7.0) or cundlinear regression (TableCurve 2D 3.0, Jandel Scientific, San Rafael CA.) as appropriate (Zar 1984). Percentages were arcsine transformed prior to analysis (Krebs 1989).  2.4.3  Ungulate-Vegetation Overlap  To determine the relative use of vegetation resources by each ungulate species, mean and median pellet densities were compared for each vegetation type in the 1996 pellet survey (7 types, Section 2.2.1). In addition, the median importance rank of each vegetation type was determined using the species-specific ecosite importance ranking of Holroyd and Van Tighem (1983). Holroyd and Van Tighem (1983) derived importance ranks by classifying ecosites according to 6 categories of relative use based on winter track counts and pellet surveys. I used the ecosite rank of each transect segment to calculate the median rank for each vegetation type.  Differences in median pellet densities, and importance ranks, were analyzed with a Kruskal Wallis one-way A N O V A , H, and Dunn's multiple comparison procedure (Sigmaplot 2.0, Jandel Scientific, San Rafael CA). Contiguous transect segments were aggregated (as for willow, Section 2.2.1) to obtain similar segment lengths for analysis (mean length: 208.6 m; range: 121-250 m; n = 673). Overlap in vegetation use among ungulates was determined using Horn's index, as is recommended when resource use is expressed as proportions (pellet-groups per hectare), rather than as the number of individuals (Krebs 1989). In addition, the spatial extent of each vegetation type was tabulated (Arcview 3.1, ESRI, Redland CA).  16  2.5  CERVID-WOLF RELATIONSHIPS  2.5.1  Moose Capture and Monitoring  To estimate moose mortality and reproduction rates, 45 adult moose were radio-collared and monitored between 1994 and 1997 (22 females and 23 males). Moose were immobilized by darting, or net-gunning, from a helicopter, or by darting from the ground. Individuals were usually captured as they were encountered, however opportunities to capture some males were ignored so that extra effort could be spent searching for females. Six late autumn and winter capture sessions, of 2-4 days each, were conducted between March 1994 and December 1996. Captures were made throughout the study area.  Moose were chemically immobilized using a mixture of carfentanil citrate (Wildlife Laboratories, Ft. Collins, Colorado) and xylazine hydrochloride (Rompun, HaverLockhart, Mississauga, Ontario), or net-gunned and physically immobilized with hobbles. Each animal was fitted with a radio collar (Lotek, Newmarket, Ontario) equipped with a 4-year battery and a 6-hour delay mortality sensor. A numbered ear tag was applied to each animal, and a lower incisor was extracted (from anaesthetized moose only) for cementum aging (Sergeant and Pimlott 1959). Pregnancy was determined either by palpation or by extracting blood samples for later analysis of progesterone levels (Haigh et al. 1982; Stewart et al. 1985). Naltrexone hydrochloride, an anesthetic antagonist, was administered after handling to speed recovery. Radio signals were monitored approximately 3 times per month from telemetry-equipped aircraft, or by ground telemetry.  2.5.2  Moose Population Parameters  Calf Survival and Recruitment Aerial calf-count surveys of collared cows were flown annually, usually in mid-June following the calving period (mid-May to mid-June). Negative calving results were  17  cordirmed with at least one subsequent aerial or ground observation in June or July. Visual observations of each cow-calf pair were attempted once a month to track calf survival.  Adult Mortality Mortality signals or reports were investigated, usually within 1-3 days, by a thorough examination of the carcass and kill-site. I assumed that predation was the probable cause of death if there was evidence of a chase and struggle, if blood was observed on the ground, or if predators had consumed the carcass within one week of the last observation of the live animal. Predators were distinguished by the following criteria (after Larsen et al. 1989): 1) direct observation of predators at the kill site, 2) tracks, scat, or hair characteristics; and, 3) kill site characteristics including carcass burial, and the size of area over which the remains were spread. Bear kills in subalpine habitats were attributed to grizzly bears. The "meat" scats of black and grizzly bears cannot be reliably distinguished, but black bears rarely kill adult moose in western North America (Ballard 1992) and are considered uncommon in subalpine habitats in the study area (Holroyd and Van Tighem 1983).  When carcasses were found intact the cause of death was determined by field necropsy, usually with veterinary assistance (Dr. T. Shury, Canmore, Alberta). In one instance, a whole carcass was retrieved for lab necropsy. Tissue specimens were submitted for forensic pathology tests as necessary (Canadian Cooperative Wildlife Health Centre, Saskatoon, Saskatchewan).  Moose survival and mortality rates were estimated using the program M I C R O M O R T , and were compared using the Z-statistic (Heisey and Fuller 1985). For the purposes of analysis, the biological year began on 1 May. Because of the small number of collared moose, I pooled 4 years of telemetry data to obtain a single estimate of annual survival  18  and mortality. Subsequent poolings of sex classes or mortality classes were carried out if log-likelihood ratio analysis suggested that the rates were not significantly different (Heisey and Fuller 1985).  Contributing Factors Femur sections containing marrow were collected when present and kept frozen until analysed. A 5-g marrow sample was removed from each femur, weighed, then oven-dried at 60° C for 48 hours, and reweighed. Percent fat was estimated as the dry-weight of marrow expressed as a percentage of the fresh weight (Neiland 1970).  Livers were collected and kept frozen until they were analysed to determine the presence and intensity of giant liver fluke infection. Livers were visually inspected, weighed, and sectioned into 1 cm thick slices to count flukes (after Pybus 1990). Livers from hunterkilled moose on adjacent provincial lands in Alberta were sampled in 1997 with the assistance of Dr. M . Pybus (Department of Environmental Protection, Alberta).  In addition to collections from radio tagged animals and hunter kills, femur and liver collections, and age estimates, were conducted whenever possible for all moose kills that were reported or discovered during the study period. Adult moose deaths attributed to predation were considered additive to other forms of natural mortality if femur marrow fat was > 20% (Peterson et al. 1984), and if miivimum sex-specific age criteria were satisfied (females < 15, males < 12; following Gasaway et al. 1992). In cases where femur marrow could not be retrieved, the determination was based solely on age criteria.  2.5.3  Elk Dominance, Survival, and Population Trends  To determine if elk were the dominant prey in each of the sub-areas, I calculated and compared mean and median cervid pellet abundances in each sub-area. I also compared cervid pellet abundances relative to early (1978-1979) and late (1996) periods of wolf reestabhshment using a Mann-Whitney test. To determine the relative importance of each  19  cervid prey in the diet of wolves, I reviewed published and unpublished data pertaining to local wolf-prey ecology (Paquet 1993; Huggard 1993a, 1993b, 1993c; Weaver 1994; Paquet et al. 1996; Central Rockies Wolf Project files). Similarly, I reviewed data relevant to local elk populations (Woods 1991; Skjonsberg 1993; Morgantini 1995; Woods et al. 1996; Parks Canada files) to determine elk survival rates and population trends in the presence of wolves.  20  3.0  RESULTS  3.1  INTERSPECIFIC ASSOCIATION AND COVARIATION OF CERVIDS  3.1.1  Interspecific Association  1996 Pellet Survey Cervid pellets were observed on 81.8% (193/236) of the sample transects in willowbearing vegetation types. Elk, moose, and deer pellets were present on 71.2% (168/236), 14.8% (35/236), and 24.1% (57/236) of the transects, respectively. When moose or deer pellets were observed, they were associated with elk in 71.4% (25/35) and 71.9% (41/57) of samples, respectively. In contrast, elk pellets occurred exclusive of the other two species more often than in association. They were associated with moose in 14.8 % (25/168) of samples and with deer in 29.8% (50/168) of samples. The pattern of association was similar in each watershed (Bow, Cascade, and Red Deer). Deer and moose were found separately more often than together. Eighty percent (28/35) of the sites that contained moose pellets were exclusive of deer, and 87.7% (50/57) of the sites where deer were present were exclusive of moose. Neither the simultaneous test of association among the three cervids (W, n = 236, P > 0.10), nor any specific pairwise associations (Yate's chi-square, n = 236, P > 0.10), yielded significant results. Thus, the null hypothesis that species occurred independently of one another was accepted.  1978-1979 Biophysical Pellet Survey In the historical data, cervid pellets were observed in 100% (40/40) of the samples. The frequency of occurrence of elk, moose, and deer pellets was 97.7% (39/40), 70.0% (28/40), and 57.5% (23/40), respectively. Similar to our observations in 1996, moose or deer pellets were nearly always associated with elk (100% (28/28) and 95.7% (22/23) respectively). Elk pellets were associated with the other cervids less often, but more  21  frequently than in 1996. Elk were associated with moose in 71.8% (28/39) of samples and with deer in 56.4% (17/39) of samples. The pattern of association was similar between watersheds.  In contrast to the 1996 data, deer and moose were found together more often during 1978-1979 than separately. Nearly sixty-one percent (60.7%, 17/28) of the sites that contained moose pellets were associated with deer, and 73.9% (17/23) of the sites where deer were present were associated with moose. Simultaneous and pairwise tests of association (W (n = 40) and Yate's chi-square (n=40), respectively) were not significant (P > 0.10). Again, the null hypothesis that species occurred independently of one another was accepted.  Changes in Cervid Dominance and Association Elk pellet frequency was greater than other cervid species in both study periods. However, elk were less dominant historically because moose and deer pellets occurred in comparatively higher frequencies. The degree of interspecific association decreased between study periods. Mean values of the Ochai association index (after Ludwig and Reynolds 1988), stratified by watershed, declined between 1978-1979 and 1996 as follows: 1) moose-elk, 0.83 to 0.36; 2) moose-deer, 0.65 to 0.11; and, 3) deer-elk, 0.73 to 0.30.  3.1.2  Interspecific Covariation  In the C A R T analysis, pellet density variables were identified by species, with a prefix to indicate the type transformation used. For example, the log transformation of moose pellet density was " L N M O O S E " , and the rank-order transformation of deer pellet density was " R A N K D E E R " . Environmental variables included the following: 1) aspect (ASPECT) described as N , S, E, W or F (flat); 2) metres above sea level (ELEVATION); and, 3) location (WATERSHED). Vegetation variables included the following: 1) willow stem density (SXDENSITY); 2) birch stem density (BGDENSITY); 3) total stem density  22  (TOTDENSITY); 4) percent willow twigs (PERCENTSX); and, 5) percent birch twigs (PERCENTBG). E L E V A T I O N was removed from the analysis in favour of W A T E R S H E D because the two variables were highly correlated. C A R T Results Moose abundance was consistently greater where elk abundance was lower (Table 3.1 and Figures 3.1, 3.2, and 3.3). Besides elk abundance, additional predictors of moose abundance were as follows: 1) 1978-1979 Pellet Survey Model: ASPECT and R A N K D E E R ( R A N K M O O S E and R A N K D E E R were positively related); 2) 1996 Pellet Survey Model: W A T E R S H E D ; and, 3) 1995/97 Browse Plot Model: W A T E R S H E D and S X D E N S I T Y (LNMOOSE and S X D E N S I T Y were positively related). The results of the cross-validation process for the three models, expressed in terms of the selection frequency of predictor variables, were as follows: 1) 1978-1979 pellet survey model: ASPECT 100% (20/20), R A N K D E E R 30% (6/20), R A N K E L K 25%(5/20); 2)1996 pellet survey model: R A N K E L K 100% (20/20), W A T E R S H E D 100% (20/20); 3) 1995-97 Browse Plot Model: SXDENSITY 100% (20/20), W A T E R S H E D 75% (15/20), L N E L K 55% (11/20). The results suggest that the 1996 pellet survey model was the most stable (least overfit) followed by the 1995/97 browse plot model and the 1978-1979 pellet survey model. Variation in moose pellet abundance was best explained by 1995/97 browse plot model (PRE = 0.753) followed by the 1996 pellet survey model (PRE = 0.479) and the 1978-1979 pellet survey model (PRE = 0.449).  23  Table 3.1. Classification and regression tree ( C A R T ) analysis of the relationship between moose abundance and environmental and biological predictor variables, Banff National Park, 1978-1979 and 1994-1997. DATA SOURCE  DEPENDENT VARIABLE  "PRE MODEL  PREDICTOR VARIABLES S E L E C T E D BY CART  PRE (PARTIAL)  PREDICTOR VARIABLES NOT SELECTED BY CART  1978-1979 Biophysical Survey n=39  RANKMOOSE  0.449  ASPECT RANKDEER(+) ASPECT RANKELK(-)  0.271 0.057 0.058 0.064  WATERSHED  1996 Pellet Survey (this study) n=178  RANKMOOSE  0.471  RANKELK(-) WATERSHED RANKELK(-)  0.237 0.152 0.089  ASPECT RANKDEER  1995-1997 Browse Plots (this study) n=19  LNMOOSE  0.753  SXDENSITY(+) WATERSHED LNELK(-)  0.621 0.066 0.065  ASPECT BGDENSITY TOTDENSITY PERCENTBG PERCENTSX  b  "Proportional reduction in error, an "r -like" statistic (SPSS, 1997). "Tnverse (-) or direct (+) relationship to moose abundance. 2  24  RANKMOOSE n=39 y =79.5 pellets/ha  n = 39 P R E = 0.449  a  i !.. . i  t_ i -- t ttt A S P E C T p North  ASPEC"/ = Flat, South, East"  L  • ii  n=15 y= 124.0 pellets/ha  n=24 y=51.7 pellets/ha  •• • •• t.ttm  M  ASPECT = South, East  I  n=ll y=20.0 pellets/h^  I t .  R A N K E L K  ASPECT!  RANKDEER< 26.00  Flat  I  n=13 y=78.5 pellets/ha  n=6 y= 116.0 pellets/h;  •• •...j  I  > 28.00  RANKEL.|K<  I  RANKDBER > 26.00 n=9 y= 140.0 pellets/h^  L  28.00 n=7 y=l 12.5 pellets/ha  n=6 y=24.0 pellets/ha  • • •  •  •••in  Figure 3.1. Classification and regression tree analysis of the relationship between moose abundance and environmental and biological predictor variables, Banff National Park: 1978-1979 pellet survey model. Moose abundance was greatest on north aspects where it varied directly with deer abundance. Moose were least abundant on south and east aspects and moderately abundant on flat aspects. Onflataspects moose abundance varied inversely with elk abundance. "Mean moose pellet abundance. ''Only 2 of 39 samples were on east aspects.  25  RANKMOOSE  n = 178 P R E = 0.479  n=178  y^lOJpellets/ha  RANKELK < 20.50  RANKELkk > 20.50 n=168 y=7.8 pellets/ha  n=10 y=52.2 pellets/ha  WATERSHED = CASCADE AND RED DEER  WATERSHED = BOW n=96, y=.63 pellets/ha  RANKELKp 96.50 n=25 y=3.2 pellets/ha  n=72, y=17.4 pellets/ha  RANKEUK S9E.50 n=47 y=25.0 pellets/ha  Figure 3.2. Classification and regression tree analysis of the relationship between moose abundance and environmental and biological predictor variables, Banff National Park: 1996 pellet survey model. Moose abundance was inversely related to elk abundance and dependent upon watershed.  Mean moose pellet abundance  a  26  LNMOOSE  n = 19 P R E = 0.753  SXDENSI  n=19 y =79.9 pellets/ha a  095 s t e m s / m  2  SXDENSITY > 13.095 s t e m s / m  n=8 y= 166.7 pellets/ha  n=ll y=16.8 pellets/ha  WATERSHED = BOW  •• » • n=6 y=6.2 pellets/ha  WATERSHED =lcASCADE, RED DEER ,PANTHER LNELK > 5.69  • n=5 y=29.6 pellets/ha  2  #> m m n=4 y=92.6 pellets/ha  LNELK <'5.69 i  • n=4 y=240.7 pellets/ha  Figure 3.3. Classification and regression tree analysis of the relationship between moose abundance and environmental and biological predictor variables, Banff National Park: 1995/97 browse plot model. Moose abundance was greatest in areas of high willow stem density where it varied inversely with elk abundance. Otherwise, moose abundance was dependent on watershed.  'Mean moose pellet abundance.  27  3.1.3  Underlying Univariate Relationships  Moose and Elk Covariation Abundances of moose and elk pellets were inversely related in the 1996 pellet survey (r = s  -0.450, n = 178, P < 0.001) and the 1995/97 browse plot survey (r = -0.447, n = 19, P < s  0.05), but were not significantly related in the 1978-1979 pellet survey (r = 0.145, n = 39, s  P > 0.5). The strength of the correlations increased following the application of partitioned regression analysis (Figure 3.4), and although r was still not significant in the s  1978-1979 data, the sign changed indicating a negative relationship. The trend toward increasing regression slopes in each of the data sets, as the edge of the point scatter was approached, suggested the presence of a factor ceiling distribution.  Moose and Deer Covariation The C A R T model constructed with the 1978-1979 pellet survey model suggested a positive relationship between moose and deer on north aspects (n = 15, Figure 3.1). In subsequent univariate tests, moose and deer pellet abundance was positively related but the relationship was not significant (r = 0.295, n = 39, P > 0.05). The moose-deer s  relationship in the 1996 pellet survey data was weakly negative (r = -0.036, n = 178, P > s  0.05).  Moose Abundance Across Watersheds The C A R T analyses (Figures 3.2 and 3.3) show the general absence of moose from the Bow watershed compared to the Cascade and Red Deer watersheds. Non-parametric tests confirmed the difference in moose abundance between watersheds (H = 55.47, n = 236, P < 0.0001). A similar dependence on watershed was found in the 1978-1979 pellet survey (H = 6.813, n = 40, P < 0.05) however, the pattern was different (Figure 3.5). In 19781979, moose abundance was greater in Bow in than in either the Cascade or Red Deer Watersheds.  28  1 2000  1  1  1  1  1—  Pellet Survey, 1978-1979 13,r = -0.183, P> 0.5. upper  1500 1000  500  Partitioned data subsets  i_ <C  100  200  i  +•>  o  -L 300  1  400  • UPPER • LOWER  J L 500 600  1  1  r  Pellet Survey, 1996  0)  nu er=60, r=-0.646, P < 0.0001  CL  PP  CO  o a> CL LU  0' 0  L 100  200  300  400  500  Moose Pellets per Hectare Figure 3.4. One-cycle partitioned regression of moose and elk covariation based on pellet counts in willow-bearing vegetation types, Banff National Park, 19781979 and 1995-1997.  29  10001=  rm 1978-1979  OJ i_  CO  I—I  o  1996  I —I Median  OJ X  OJ Q.  in  ~v  100  a3  Q_  B  C  R  Watershed Figure 3.5. Mean ( ± SE) and median moose pellet density on willow bearing vegetation types in the Bow (B) Cascade (C) and Red Deer (R) Valleys, Banff National Park, 19781979 and 1996.  30  Moose Abundance and Willow Stem Density Moose pellet abundance and willow stem density were positively correlated (r = 0.676, n s  = 19, P < 0.005). A linear regression of moose pellet abundance (log-transformed) on willow stems/m was significant (F = 8.58, n = 19, P < 0.009), but explained only a small 2  amount of the total variation in moose abundance (r - 0.335). 2  Moose Abundance and Aspect Aspect was an important predictor variable in the 1978-1979 survey data, but not in the 1996 data. In the 1978-1979 C A R T model, moose abundance was greatest on north aspects, and less on flat, south, and east aspects. A second split suggested that after the north aspect had been accounted for, the next most abundant moose densities occurred on flat aspects. Using a Kruskal Wallis test, I found the same pattern of differences in median moose abundances between the three major aspect classes; north, flat, and south (H = 10.44, n = 37, P < 0.005; east aspect was excluded because of the small sample size, n=2). In the 1996 pellet survey, low moose pellet densities precluded the detection of differences between aspects because medians moose abundance were zero in each aspect class. In the 1995/97 browse survey, the limited number of samples was not adequately distributed across aspect classes to test for an effect.  31  3.2  CERVID-RESOURCE O V E R L A P AND BROWSE RELATIONSHIPS  3.2.1  Diet  Elk used 14 plant genera frequently (> 2%), and moose used six (Table 3.2, Figure 3.6). The median number of diet items per individual elk and moose was 5 and 3, respectively. The difference was significant (H = 24.16, P < 0.001). Moose diets were restricted almost completely to browse (trees and shrubs). On average 83.2 ± 11.94% (SE) of individual moose diets consisted of willow and subalpine fir, Abies lasiocarpa. Elk diets, on the other hand, were distributed more evenly across tree, shrub, and herb genera (Figure 3.6).  3.2.2  Diet Overlap  Twenty five percent (4/16) of frequently used plant types were common to both moose and elk diets (willow; pine, Pinus; buffaloberry, Shepherdia canadensis; Douglas fir, Pseudotsuga menziesii; Table 3.2). The heavy use of several of the shared plant types resulted in a high diet overlap (Simplified Morisita's index = 0.549). Willow and pine each comprised > 10% of individual species' diets (Figure 3.6). On average, 41.6 ± 6.0% (SE) of the moose diet was composed of willow compared to 16.5 ± 5.7% for elk. Thirteen percent (13.2 ± 3.8%) of individual moose diets was composed of pine compared to 29.6 ± 7.4% for elk. Individual moose consumed about 2.5 (41.6% /16.5%) times more willow than did elk.  The diet overlap was asymmetrical. Of the 6 plant genera used frequently by moose, they shared 67% (4/6) with elk. In contrast, only 29% (4/14) of the plant genera used by elk were shared with moose. I quantified the asymmetry with MacArthur and Levin's Index. The index value for moose overlap with elk was 0.313, and for elk overlap with moose it was 0.871 (Table 3.2).  32  Table 3.2. Winter diets of moose and elk on sympatric range in the Bow Valley, Banff National Park, 1995-1996". Moose diet was dominated by browse and contained significantly fewer plant foods than elk diet. The latter included herbs and forbs as well as browse. Moose and elk diet overlapped, and the overlap was highly asymmetrical. Winter Niche Breadth (based on diet) Index or Measure  Moose (n=39)  Elk (n=23)  10  18  0.360  0.636  6  14°  Total number of plant genera detected Levin's Standardized Measure Number of frequently used plant genera  B  (Cutoffvalue = 0.02) Mean number of diet items detected per  2.80  5.30  3  5  individual Median number of diet items detected per individual (central 5 0 % of observations)  D  D  (4.00-6.75)  (2.00-3.00)  Winter Diet Overlap Total number frequently used plant genera  16  (Cutoffvalue = 0.02) Proportion of diet common to both moose and  0.25(4/16)  elk Morisita's Original Index  0.482  Simplified Morisita's Index  0.549  Horn's Index  0.614  MacArthur and Levin's Index Moose overlap with elk Elk overlap with moose — . r  0.313 0.871  "Elk pellets were collected in 1995 only. Moose pellets were collected in both winters ( 1 9 9 5 and 1996, 2 0 and 19 pellet samples, respectively). Tree and shrub genera: Salix, Abies, Pinvs, Shepherdia, Pseudotsuga, Betula. T r e e and shrub genera: Salix, Pinus, Shepherdia, Pseudotsuga, Picea. Herb genera: Juncus, Carex, Poa, b  Equisetum, Stipa, Koeleria, Eriophorum, Bromus, Agropyron. d  Medians were significantly different: H = 2 4 . 1 6 , P < 0.001.  33  co CD  60  I  o  CD QCO i_ CD CL  I  I  I  I  I  I I  I  I  I  I  I I  Herbs  Trees and Shrubs  40  S Moose (n=39) • Elk (n=23)  ^ O  c CD =5  CT CD  £ 20 CD >  ra  1 nn n  CD  0  n  n  i  '  Plant Genera  CO  -o  60  I  >  I  I  I  I  I  I  I  I  I  I  I  Trees and Shrubs  TJ  I  I  i  i  Herbs  c 1—  CD Q-  • Moose (n=39) o Elk (n=23)  40  c CD  Q 0  > 20 *ro CD  c ro  1  CD  0  f  i  t l  u  1* k I  A  i  fft  1  £  ip  * E  A  Plant Genera  Figure 3.6. Percentage mean ( ± SE) relative density of plant fragments per individual, and percentage relative frequency of plant genera per species, in moose and elk winter diets, Bow Valley, Banff National Park, 1995 -1996.  3.2.3  Browsing Rates  Most stems on browse plots were willow, they averaged 67 ± 5% (SE) of the total stems, and a mean density of 11.6 ± 1.3 stems/m . The next most numerous genus was birch, 2  with a mean relative stem frequency of 18.0 + 3.4% of the total stems per plot, and a mean density of 4.1 ± 0.8 stems/m . Shrubby cinquefoil, Potentilla fruticosa, accounted for a 2  mean frequency of 6.3 ± 1.8% of the total stems per plot and a mean density of 1.2 ± 0.4 stems/m . Other genera and species that occurred in mean frequencies of < 1% of the 2  total stems were: juniper, Juniperus spp.; lodgepole pine, Pinus contorta (> 3-m tall); balsam poplar, Popuius balsamifera; wild rose, Rosa acicularis; buffaloberry; spruce (< 3m tall); and spruce (> 3-m tall).  Of the 6,149 twigs examined, 2,069 (33.6%) showed evidence of being browsed. The overall browse rates on willow, birch, buffaloberry and shrubby cinquefoil twigs were 39% (1832/4648), 20% (207/1053), 11% (11/123), and 4% (9/209), respectively. Sample sizes for the remainder of genera were too small to analyze (< 10 twigs sampled).  Willow browsing intensity was directly related to elk pellet abundance (Figure 3.7, F = 68.7, n = 26, P < 0.001, r = 0.730), but was not well related to moose pellet abundance 2  (Figure 3.8, F = 7.89, n = 26, P < 0.009, r = 0.216). The pattern of scatter suggests that 2  moose were more abundant where browsing intensity was less (Figure 3.8). The elkbrowse rate relationship was approximately asymptotic (note cube-root scaled x-axis), suggesting that an upper limit of browse intensity was being approached. Browse rates did not exceed 80% of available twigs. Bite diameters approached an asymptote as browsing intensity increased (Figure 3.9, F = 17.7, n = 18, P < 0.001, r = 0.525) 2  suggesting that an upper limit to edible twig diameters was approached. Mean bite diameters did not exceed 0.28 cm.  35  Deer pellets were only found on three browse plots, in low numbers, consequently no browse relationship was detected. Similarly, snowshoe hare feces occurred in only 4 samples and none of the observed twig bites displayed the characteristics described for that species (Krebs et al. 1986).  90 h  ro  U  d)  s  o GQ C <D O  i_  <X>  D_  Elk P e l l e t s per H e c t a r e  Figure 3.7. Percent willow twigs browsed relative to elk pellet density, Banff National Park, 1995 and 1997. Samples were obtained in the Bow, Cascade, Panther and Red Deer River Valleys. Arrows indicate the direction of change between 1995 and 1997 in 7 repeated samples. Arrow vectors are generally parallel to the regression line further validating the correlation.  "Arcsine transformed.  37  90 80  Non-repeated sample: n = 19, F = 1.83, P = 0.194, r = 0.044 Including 7 repeat samples: n = 26, F = 7.89, P < 0.009, r = 0.216 2  70  2  60 50 40 30 20 10  0  0  1  1  120  240  i.  360 480  M o o s e P e l l e t s per H e c t a r e  Figure 3.8. Percent willow twigs browsed relative to moose pellet density, Banff National Park, 1995 and 1997. Samples were obtained in the Bow, Cascade, Panther and Red Deer River Valleys. Arrows indicate the direction of change between 1995 and 1997 in 7 repeated samples. Arrow vectors form acute angles suggesting a poor relationship with corresponding weak regression statistics.  'Arcsine transformed.  38  30 E E  ojj  $  & <p  o£  cf)  Percent Browsed  Figure 3.9. Mean willow twig diameter at the bite-point relative to browsing intensity, Banff National Park, 1995 and 1997. Seven sites were excluded because the mean twig diameters at bite-point were based on small samples (n < 30). The remainder of mean twig diameters at bite-point are based on a mean sample size of 89 bites (range: 32 - 192).  39  3.2.4  Ungulate-Vegetation Overlap  Willow-bearing vegetation types were the most extensive type in the study area, excluding alpine and upper-subalpine open herb and low-shrub types (Table 3.3). Willow also was the most highly ranked type for moose, and one of the second-highest ranked types for elk (Figure 3.10b). Spruce/fir and pine vegetation types were also extensive (Table 3.3), and used by moose (Figure 3.10), but they were significantly less important to moose than willow (Figure 3.10b; H = 493.9, P < 0.001, n = 673; Dunn's, P < 0.05). Moose abundance appeared highest on the spruce/fir vegetation type, next highest on willow, and lower on every other type (Fig. 3.10a, H = 24.7, P < 0.001, n = 673). However, no significant differences were detected between vegetation types in pairwise comparisons of moose pellet density (Figure 3.10a; Dunn's, P > 0.05).  Table 3.3. Areal extent of 7 vegetation types sampled in the 1996 pellet survey, Banff National Park. Vegetation Type  Aspen Drywoods Grassland Mixed Pine Spruce/fir Willow Total  Percent Area  Area (km ) 2  10.27 225.11 12.29 278.73 475.07 709.40 836.92 2590.61  Percentage of A l l Vegetated Terrain in B N P "  0.40 8.84 0.48 10.94 18.64 27.84 32.85 100.00  0.24 5.35 0.29 6.63 11.30 16.86 19.90 60.57  "Includes open herb and open low-shrub types in upper-subalpine and alpine zones (not sampled in 1996).  Elk were the most dominant ungulate in each vegetation type (Figure 3.10). Pairwise analyses of ungulate overlap across vegetation types (Figure 3.10a), suggested that moose overlapped primarily with deer (Horn's Index (HI) = 0.736), secondarily with elk (HI = 0.587), and least with sheep (HI = 0.401). Elk-deer overlap (HI = 0.705) was greater than elk-sheep overlap (HI = 0.615), and deer-sheep overlap (HI = 0.478).  40  Very High h -  Habitat Type Figure 3.10. Mean pellet group density ( ± SE), and the median rank of ecosite importance, for 7 vegetation types, Banff National Park, 1996 pellet survey. Differences within species categories in b were significant (H, P < 0.001, n=673; Dunn's, P < 0.05), with the following exceptions: elk and deer, aspen-grassland, aspenmixed; deer, drywoods-grassland, mixed-drywoods, aspen-pine; sheep, pine-willow, spruce-drywoods. "Sample size distribution and species legend were the same in a and b.  41  3.3  CERVID-WOLF RELATIONSHIPS  3.3.1  Moose Mortality and Reproduction  Mortality Twenty eight of 45 adult radio-collared moose died between 1994 and 1997 (13/22 females, 15/23 males). Fifteen kills were attributed to wolves, 2 to grizzly bear, 4 to starvation or disease, and 3 to unknown natural causes. Human related mortality included 2 moose killed on the highway and 2 hunted. Moose were monitored for 82.3 radio-years (37.7 female-years and 44.6 male-years). The mean annual survival rate was 71.3% (64.6 82.3%, 95% C.L; Table 3.4) with no significant difference between sexes (% = 1.7, 3 d.f., 2  P > 0.05). The mortality rate attributed to wolves and grizzly bears was 17.4% (9.9 25.0%). Starvation, disease, and deaths from unknown cause accounted for 7.2% (2.1 12.3%) of deaths, and human-caused mortalities contributed another 4.1% (0.2 - 8.2%). Four collars dropped off, thus the fate of the animals could not be determined. However, data from these individuals were used as radio-days survived up to the time of the collar drop. More moose were killed by wolves during late winter and spring than during summer, autumn, and early winter (P < 0.05). Forty-six percent of wolf-caused mortalities (7/15) z  were concentrated in the first three months of the biological year (May-July) with little difference between sexes (4 females, 3 males). Another 33% of wolf-caused mortalities (5/15) occurred in the last quarter of the biological year (February-April) and were predominantly males (1 female, 4 males).  42  £w y  VO I  -r 5'  VO VO  3 ^ 3 SL 3 3 ~ o' 3 § o O  -  Q  -  O 3  o o_ ft" o.  VO VO  4^ i VO VO  3 O 3 i 3*  -4  TJ o o_  VO VO  VO VO  ST* CL  I  4^  VO VO  VO VO  I  -o.  ft X  TJ o a.  TJ  o o, ft"  -J  C  00  3  3 3 E. ON UI  3 § 3 o  oo  00  a  to  to  U J  UJ  UI  1.1  ON  — a  VO  oo  oo  oo  00  oo  oo  00  00  UJ  Pi O  o to oo  3-~ 2 T> *  VO ON  to  * !o  03 tn  3 25  ON  3 ? 3 ?  ©  0Q  ra ra  o to  o to  Si o> ON ra u to Cu i  o to  1  ON  <  S* 25  ra ^ •5  o p p £ b " O (j, oo UJ " © » to to ON VO to  VO UI  ^  o  O O N  W  >  3 o o-  CO  O  to  00  VO  A  VO UI  -J  to  ~4  CL  3  a.  5  25  3' 5'  to ui  to  5  a. £• 5? » a  U) ON UI  UJ Os  UJ ON U")  to  S-  ra <25  ra  3  to 4^ -4 Cu  C P  11  2  to  o g 7 > 9 CO o o  to  to  UJ  VO VO  1  ui  i i  o  o  Ui  CL  V TJ V  ra , so  p  11 p b oo o  P b  sit V S3 o 00 o  b 2 © g, ° vo 2 0 0 9 w 9 © p b ui to UJ  8* 25 <t  U>  Contributing Factors Predation rates on adult moose during heavy snow years, 1995 and 1996 (13.8%, 4.9 22.6%), were not significantly less than during years of light snow, 1994 and 1997 (22.8%, 9.8% - 35.8%; P = 0.134). z  The mean femur fat content of 14 moose killed by wolves was 64.4%, (range: 24-97%), and the mean age of 20 adult moose killed by wolves was 9.1 years (range: 2-18). Of the two moose recorded killed by grizzly bears, one was a 9-year-old female with 89.9% marrow fat content, the other was a 13-year-old male (no marrow data available). Overall, 88% of adult female moose (7/8) and 66% of adult male moose (8/12) killed by wolves or bears were considered additive to other forms of natural mortality (see Methods).  Livers were recovered from 18 moose kills usually following accidents on the highway and railway (13/18), most often in the Bow Valley (12/13). Seven of 13 (53%) moose in the Bow Valley were infected by giant liver fluke (GLF), and 1 of 5 (20%) moose in Peter Lougheed Provincial Park was infected. Five of 8 infected moose were females. Only one liver was recovered from a predator kill (in PLPP and it was infected). G L F infection was diagnosed as the cause of death in 1 of 4 carcasses that were recovered intact (a 10-yearold moose from the Bow Valley, male, 72.4% femur fat). Only 1 of 23 (4.3%) bull moose killed by hunters on provincial land east of the study area was infected.  Sixty five moose deaths were recorded in or immediately adjacent to Banff and Peter Lougheed parks during the study period. Mortalities in order of decreasing proportions were as follows: 1) predation, 40.0%; 2) highway, 27.7%; 3) unknown, 13.8%; 4) railway 9.2%; 5) disease and starvation, 6.1%; and, 6) hunting, 3.1%. Thirty five percent (35.4%) were adult females, 36.9% were adult males, and 27.7% were calves.  44  Reproduction Nine of 12 mature cows were pregnant at the time of capture (75%). Pregnancy could not be determined at the time of capture for another 10 cows, usually because the capture occurred during the rut, or soon afterward. In late June calf surveys, 66.7% (21/31) of cows had calves at heel. Twin calves were observed only once (4.7%, 1/21). Thirty three percent (7/21) of June calves survived to yearling age, and overall, only 7 of 31 (22.5%) mature females successfully reared a yearling (Table 3.5).  3.3.2  Wolf Population Trend  Between 1994 and 1997, 6 wolf packs were active in the study area and most of them denned there as well. There was a pack in or near each of the three main watersheds, the Bow, Cascade, and Red Deer, and a pack in each of the northern and southern extremes, the North Saskatchewan Valley, and Peter Lougheed Park respectively (Paquet et al. 1996; Central Rockies Wolf Project files). Paquet et al. (1996) reported that 6-7 packs were located at least partially within B N P in 1996 with an approximate mean winter pack size of 6 wolves (about 42 wolves in total). In contrast, between 1976 and 1980, the entire B N P wolf population was estimated to be 25 animals, consisting mainly of solitary individuals or pairs (R. Chapman pers. comm. via P. Paquet).  Wolves were eliminated from B N P in the 1950's by an extensive predator control program (Gunson 1992; Paquet 1993). By 1978, the only confirmed pack re-establishment was well north of the study area, in the lower Clearwater drainage, but solitary wolves and small groups were occasionally seen in the lower Red Deer, the Panther and the middle Cascade Valleys (Holroyd and Van Tighem 1983). The only evidence of denning in B N P was recorded after the 1978-1979 pellet survey, in the Panther Valley in 1980 (Holroyd and Van Tighem 1983). Wolves did not re-establish in the Bow Valley until ca. 1986 (Mickleera/. 1986).  45  Table 3.5. Observation and survival of moose calves, Banff National Park and Peter Lougheed Provincial Park, 1994-1997. Observation of Calves Moose ID  1994 No No (Mar. ,F ) Unknown  74 75 76 — 83 85 86 ~ 89 91 93 — 94 — 96 — 98 — 101 — 103 60 62 — 63 — 66 — 68 — 69 — 106 — 108 2 Number of mature cows surveyed in June. 0.5 (1/2) Proportion of cows with calves at heel in June. 0.0 (0/1) Proportion of June calves surviving to yearling age. 0.0 (0/2) Yearlings recruited per mature female. Cow dead or not yet captured. Month that a calf was last seen. M = male calf, F = female calf. Possible juvenile~not included in tally. b  -  c  1995  1996  -Unknown  —  a  Yes (M)  No  —  ~ No (Oct., F) ~  Yes (M) No  —  -  No No Yes (M) No No No (July) No ~  —  -~ No - (Feb) ~  11  — —  —  No (May) ~  —  Yes (F) Yes (M) No (Nov., M) No (Sept.) No (Mar., M.)  No (Feb., M) Yes (F)  No -  — —  Yes (M) No No No (July)  -  Unknown  No (July) Unknown  d  —  No (Feb., F) No (Aug.) 13  ~  — — —  No (Jan.) 5  0.45 (5/11)  0.77 (10/13)  1.0 (5/5)  0.6 (3/5)  0.3 (3/10)  0.2 (1/5)  0.27 (3/11)  0.23 (3/13)  0.2 (1/5)  a  b  c  d  46  3.3.3  Elk Dominance, and Population Trend  Elk were abundant and widely distributed, relative to moose and deer, in both 1978-1979 and 1996 (Fig. 3.10). Elk pellets outnumbered moose pellets by ratios of 8.75 ± 1.55 1  (SE), 9.39 ± 3.75, and 18.46 ± 12.72 based on the 1996 pellet survey, the 1978-1979 pellet survey, and the 1995/97 browse plot survey, respectively.  Median cervid pellet abundance declined between 1978-1979 and 1996 (Figure 3.11; moose, U = 7398, P < 0.0001; elk, U=6921, P < 0.0001; deer, U = 6270, P < 0.001). However, in a review of three discrete periods of annual elk censuses in the Bow Valley > 6 years each, Woods (1991) reported an increasing trend based on aerial and ground counts (sightability-corrected). His results were as follows: 1944-53, 648 ± 66 SE; 195968, 725 ± 92; and, 1985-90, 978 ± 54. Between 1986 and 1989, he estimated that survival rates of adult radio-collared elk were 0.80-1.0 (95% C L , n=33 females) and 0.64-1.0 (n= 14 males).  More recently, Woods et al. (1996) reported that the elk population in the Bow Valley remained stable, or decreased slightly, between 1985 and 1995, during the period of wolf re-establishment (Paquet 1993; Paquet et al. 1996). The mean number of elk observed in annual aerial elk counts (uncorrected), for the 11-year period, was 774 ± 24.7 SE (Woods et al. 1996). Woods et al. (1996) also noted that although the elk population remained relatively stable, elk distribution contracted sharply into areas near the town of Banff.  1  Mean of ratios, stratified by watershed with equal weighting. 47  1000b  100L  B  C Location  C Location  R  R  c. Deer  1000 U CO  r-a 1978-1979  o cu X  I—I 1996  i _  CD  | CD  CL  100l_  n=15  i °-i Median  a  n=12 n=136  n=12 n=22  _J_  =78  n  C Location  Figure 3.11. Mean ( ± SE) and median moose, elk, and deer abundance on willowbearing vegetation types in the Bow (B), Cascade (C), and Red Deer (R) Valleys, Banff National Park, 1978-1979 and 1996.  "Sample size distributions in Figures a, b, and c were the same.  48  Skjonsberg (1993), in a series of elk counts between 1983 and 1987, reported that Red Deer, Clearwater, and Panther watershed elk populations increased by 7% per annum, in the presence of wolves. Morgantini (1995) observed that the increasing trend continued in the lower Red Deer Valley, through to the period of his review in 1995. In a joint aerial survey in 1995, Parks Canada and Alberta Environment staff counted 2,993 elk wintering in, and immediately east of, the study area between the Bow and Red Deer valleys (Parks Canada files).  3.3.4  Wolf Predation  Between 1988 and 1997, elk were the primary prey of wolf packs that occupied the Bow and Cascade watersheds. Mule and white-tailed deer were secondary prey, and bighorn sheep, moose and mountain goat were killed in lesser numbers (Huggard 1993a, Paquet 1993; Paquet et al. 1996; Paquet et al. 1998). Elk were probably the main prey of the wolf pack that occupied the Panther and Red Deer watersheds as well (C. Callaghan, pers. comm.) because wolves there had access to a large and growing elk population in the lower Red Deer and Panther (Skjonsberg 1993; Morgantini 1995). Elk, white-tailed and mule deer were approximately equal in prominence in the diet of the Peter Lougheed Pack at the south end of my study area. Moose and bighorn sheep were secondary prey (Paquet 1993; Paquet et al. 1998).  Fourteen of the moose kills attributed to wolves were distributed across 4 of the 5 pack territories in my study area. A n additional kill occurred in a neighbouring pack territory in British Columbia. Wolves preyed upon moose throughout the study area, regardless of the proximity of elk concentrations (Figure 3.12).  49  Figure 3.12. Approximate distribution of wolf pack territories, elk winter concentrations, and radio-collared moose kill sites in the study area, 1994-1997. "W", " G " , and " U " denote wolf and grizzly kills, and kills of unknown cause, respectively (compiled from data of Paquet et al. 1996; Central Rockies Wolf Project files; White et al. 1995; this study).  50  4.0  4.1  DISCUSSION  EVIDENCE OF EXPLOITATIVE COMPETITION  The observed relationship between moose and elk winter pellet abundance (Figures 3.13.4) was consistent with the inverse density pattern expected if competition were operative (Roughgarden 1983; Wiens 1994). Interactions between the two cervids and a shared food resource—willow—supported several of the key criteria required to demonstrate exploitative competition (Reynoldson and Bellamy 1971; MacNally 1983; Wiens 1994). I found that the two cervids overlapped in resource use, and that resource use by elk probably reduced availability to moose. The large asymmetries observed in distribution, abundance, diet breadth, diet overlap, and browse utilization, provided the conditions necessary for elk to dominate competitive interactions.  4.1.1  Interspecific Association  The lack of significant cervid associations (Section 3.1.1) suggests that moose and elk occur independently of one another and that competition may be unlikely. This finding contrasts with the inverse density relationship supported in the C A R T analyses. Different results between the two types of analyses are not uncommon and their comparison helps sharpen the interpretation of interspecific relationships (Hurlbert 1969; Ludwig and Reynolds 1988). The results may suggest that the cervids respond to one another on the basis of density rather than simple co-occurence. This interpretation is supported by the partitioned regression analysis, where inverse density effects were strongest when either, or both, species' abundances were high; that is, a factor ceiling distribution (Figure 3.4).  Alternatively, the association analysis (VR, and pairwise chi-square) may have been biased because elk were present in nearly all of the sample sites where ungulate activity was recorded (1996, 87%, 168/193; 1978-1979, 98% , 39/40). This would leave few available  51  sites for moose and deer to avoid elk, even if they attempted to do so. The nearly ubiquitous presence of elk would have the effect of reducing the magnitude of the chisquare test statistic that was used in the analyses of pairwise associations. The value of the test statistic is largest when most observations are distributed either between single species occurrences (negative association), or between joint occurrences and joint absences (positive association; Ludwig and Reynolds 1988). Elk dominance resulted in the majority of observations being distributed between single elk occurrences, and joint elk-moose or elk-deer occurrences, thereby limiting the size of the test statistic. The result may have masked significant associations (positive or negative) if there were any.  4.1.2 Interspecific Covariation: the Inverse Density Pattern  The C A R T modeling process statistically selected elk abundance variables in each of the three data sets, supporting the hypothesis that moose decline as elk increase. Crossvalidation of the C A R T models showed that the consistency of selection of elk variables ( R A N K E L K and L N E L K ) was dependent on the particular data set. For example, R A N K E L K appeared in only 25% of the cross-validation sets of the historical data (Section 3.1). I chose to accept all three models without "pruning back" (sensu Breiman et al. 1994; Clark and Pregibon 1992) because: 1) elk abundance variables were more stable in the two models based on my field data (55% and 100%, Section 3.1.2); 2) I was most interested in the selection of predictors, and not necessarily in the quality of prediction; and, 3) the relationship between moose and each predictor variable, except R A N K D E E R , was found to be significant in follow-up univariate tests.  I also recognized that, given the possibility of a factor-ceiling distribution (Thomson et al. 1996), there was relatively little power to test the elk-moose relationship in the 1978-1979 model. The historical data had relatively few points in the upper boundary area (Figure 3.4) thus, the lack of consistency in the C A R T model, and the weak result in the  52  partitioned regression analysis, could have been due to insufficient sampling at the upper thresholds of elk and moose densities, rather than the result of a weak interspecific relationship.  In the C A R T analysis, variables representing elk abundance were prominent among several other biological and physical predictors. The results support the probability that the inverse density pattern resulted from a real interaction between elk and moose, but the possibility that the result was created by differences in variables that were not measured cannot be entirely dismissed (Abramsky et al. 1986; Caughley and Sinclair 1994). Jenkins and Wright (1988) found a similar inverse density pattern in northwestern Montana. In their study, moose abundance was inversely correlated with a combination of elk and white-tailed deer. They interpreted the pattern as limited evidence of diffuse competition, dominated by elk and deer.  4.1.3  Diet Overlap and Resource Limitation  The diet overlap found between moose and elk (e. g., Simplified Morisita index = 0.55, Table 3.2) was similar to values previously reported for the two cervids by Jenkins and Wright (1986, Simplified Morisita index = 0.57 - 0.71). If exploitative competition were occurring, overlap in the use of willow would be expected to be more important than overlap in the use of pine because: 1) the two cervids generally prefer willow to pine (elk, Nelson and Leege 1982; moose, Peek 1974, Pierce 1984); and, 2) willow is much less prevalent than pine in the study area (Holroyd and Van Tighem 1983; Achuff et al. 1  1996).  Although willow was 2.5 times more prevalent in the diets of individual moose than in those of elk (Section 3.2.2 and Figure 3.6), the numeric dominance of elk on willow vegetation types probably resulted in the elk population consuming more willow in  'Refers to relative plant cover relationships rather than to the generalized vegetation types designated in my study.  53  absolute terms. The strong relationship between elk abundance and willow browse rates, in contrast to the weakly negative or null relationship between moose and willow browse rates (Figures 3.7, 3.8), supports this contention. Elk had a large measurable effect on the willow food supply, but moose density appeared to have no effect.  The relationship between willow browse rates and elk abundance suggested that where elk were abundant, they browsed most of the available willow twigs. The apparent ceiling in browse rates (ca. 80%) indicated that not all available twigs were consumed, however, the remaining twigs may have been unavailable because they were: 1) sandwiched between large stems and inaccessible (personal observation); 2) too widely scattered for efficient feeding; or, 3) possibly unpalatable. In a study of elk-beaver interactions in the Bow Valley, Nietvelt (1999, M . Sc. thesis, Univ. of Alberta, submitted) found that willow twig consumption exceeded twig production, where elk were abundant. Elk appeared capable of consuming all of the current annual growth of willow.  The asymptotic relationship between willow twig bite diameter and browsing rates (Figure 3.9) suggests that favourable twigs became increasingly scarce as feeding patch is used. The ceiling in mean bite diameters (ca. 0.27 cm) probably occurred because the nutritional content of twigs declines as diameter increases. For example, Shipley and Spalinger (1995) found that the consumption of large diameter twigs resulted in significant declines in the digestible energy intake of moose, because larger twig diameters contain higher proportions of indigestible fibrous plant parts (Risenhoover 1987; Vivas et al. 1991). Willow food resources appeared limited by elk in nutritional quality (twig diameter), as well as quantity, as elk density increased.  4.1.4  Vegetation Overlap and the Importance of Willow  Exploitative competition in the willow vegetation type would not be expected to be biologically important to moose (a negative effect sensu Wiens 1994) if such types represented only a small proportion of the total area or vegetation resources used by  54  moose. Nor would it be expected to be important if moose were well segregated from elk in other habitats. Elk dominated each vegetation type, and they overlapped with moose wherever moose were present (Figure 3.10, HI = 0.587). In addition, willow was the most extensive vegetation type in the study area (Table 3.3). Thus, competition in willow vegetation types could be biologically important. This is supported by results suggesting that: 1) moose abundance varied directly with willow density (Sections 3.1.2, 3.1.3); 2) willow was prevalent in both species diets (especially moose, Figure 3.6); and, 3) the willow vegetation type is highly important to both moose and elk (Figure 3.10).  The spruce/fir vegetation type represented the lowest elk pellet density, and the least important vegetation type for elk (Figure 3.10). In contrast, this type represented relatively high moose pellet density, and was of medium importance to moose (Figure 3.10). The spruce/fir vegetation type may represent an alternative habitat for moose that is partially segregated from elk. However, pellet densities in spruce/fir and willow types were not signifcantly different (Dunn's, P > 0.05; Figure 3.10a), and the importance rank of the willow type was greater than that of the spmce/fir type, for both moose and elk (Dunn's, P < 0.05; Figure 3.10b).  4.1.5  Theoretical Outcomes of Competition  Elk would be expected to be relatively tolerant of intensive browsing of willow because their diet is composed of several other food items (Table 3.2 and Figure 3.6). The other foods may be more abundant, or less sensitive to intense herbivory. In contrast, moose abundance was directly related to willow abundance (Figure 3.3), a result explained, at least in part, by the dependence of moose on willow for a large part of their winter diet (Figure 3.6), as well as their narrow diet niche (Table 3.2). Tilman (1990), in his analysis of mechanistic models of competition, argued that the outcome of resource (exploitative) competition is often predicted by a simple rule: "the winner is the species that depresses the equilibrial resource abundance to the lowest level consistent with its own maintenance, relative to the levels required by competing species"; which he called the " R * rule". A  55  numerically dominant and generalist herbivore such as elk should, in theory, be able to win resource competition interactions by depressing and mamtaining willow abundance below that required by the more specialized moose.  Vegetative forage is generally of low quality, consequently herbivores must make a timeintensive investment to consume thousands of bites per day to survive (Bunnell and Gillingham 1985). Moen et al. (1997), in a spatial model of moose foraging and energetics, found that the simulated body mass of moose was most sensitive to browse digestibility and least sensitive to dry matter (digestible plus fibrous) intake rates. Sand (1996) linked female moose fecundity directly to body mass, a common observation among ungulates (White 1983; Caswell 1989). Moose are an obligate browser in winter, their reproductive output would be expected to be highly susceptible to limited availability of important browse species like willow.  4.1.6  Other Impacts on Browse Resources  In the study area, willow communities are maintained by wetland and riparian conditions, by climatically imposed cold air sinks, or by snow-avalanching (Holroyd and Van Tighem 1983). Fire can also impact willow abundance and vigor. Oldemeyer and Regelin (1987; cited in Schwartz and Franzmann 1989) found a pronounced peak in the biomass of browse resources (willow, paper birch, and aspen) about 10 years post-fire on the Kenai Peninsula, Alaska. Forest fire frequency has declined in BNP since the late 1800's (White 1985) and few forest stands are younger than 100 years (Rogeau and Gilbride 1994). Van Wagner (1995) concluded that the current fire-free period is unique over the span of the last 500 years in the mountain parks. The absence of fire is most likely due to the reduction in burning by Native peoples (Achuff  al. 1996; White 1985) and is not well  explained by any long term changes in climate, fire prevention, or fire suppression (see review in Achuff et al. 1996).  56  AchuSet al. (1996) modeled forest succession in BNP spatially, using the current altered fire regime. They found that older closed forest vegetation has increased at the expense of herb and low shrub communities and young conifer forests. They forecast that in the next 50 years, understory shrubs and grasses will be replaced by mosses as lodgepole pine stands age and are replaced by more shade tolerant spruce. I interpret that the exclusion of fire is probably causing a decline in willow biomass, especially in montane systems where the special climatic factors that maintain willow communities are less prevalent than at higher elevations. The fire-related decline of willow, other browse types, and grasses may intensify competitive interactions between elk and moose over the long term.  4.1.7  Variation in Diet, Diet Overlap, and Snowpack  The collection of fecal pellets for diet estimation during two deep-snow winters (1995/96 and 1996/97) may have biased the analyses of both diet and diet overlap. Deep snow may cause elk to use browse resources more than in milder winters (e. g., Jenkins and Wright 1986, 1988), and both species may switch their diet to understory browse that is protected by mature forest canopies (Pierce 1984; Jenkins and Wright 1986, 1988). The timing of my field sampling for browse assessments spanned both a both a light-snow (1994/95) and a deep-snow (1996/97) winter. The intensity of willow browsing was well explained by elk abundance in both winters and appeared independent of snowpack depth (Figure 3.7). Furthermore, functional changes in browse rates may be small relative to large changes in snowpack. For example, Jenkins and Wright (1986) measured only a small increase in the proportion of deciduous browse in the diet of elk (36% - 46%) relative to a 100% increase in snowpack (50-cm - 100-cm). I conclude that neither diet overlap nor browse exploitation relationships were appreciably biased by snowpack variation.  57  4.2  EVIDENCE OF APPARENT COMPETITION  The inverse density pattern between moose and elk (Figures 3.1-3.4) provided evidence of apparent competition (Holt 1977), as it did for exploitative competition. The relationship between wolves and moose further supported the possibility of apparent competition, where the impact on moose, a secondary prey, was greater than on elk, a primary prey.  4.2.1  Moose and E l k Survival  Moose Survival Predation rates on adult moose appeared unsustainable. The mortality rate attributed to predation was larger than any other mortality category (P < 0.05, Table 3.4), and yearling z  recruitment was insufficient to compensate for the losses. For example, assuming a calf sex-ratio of parity, and that all 18-month-old females were reproductively mature (the latter assumption less likely to be true; Bunnell 1987; Schwartz 1998), approximately 57 yearling recruits per 100 females would be needed to compensate for the mortality from all causes (28.7%; Table 3.4). If the predation rate of 17.4% (Table 3.4) was applied to adult females, with no other mortality, dispersal, or immigration assumed, at least 34.8 yearling recruits per 100 females would be needed to compensate for predation alone. M y recruitment estimate of 22.5 yearlings per 100 females (Table 3.5) suggests that the moose population is currently unable to compensate for predation, much less the additional causes of mortality.  Elk Survival Elk were both the dominant cervid and principal cervid-prey. They fared well under the conditions of re-estabUshing wolf populations (Section 3.3.2). Female elk survival rates estimated prior to this study were high (0.90), and elk populations throughout the study area were reported to be increasing or nearly stable up to 1996.  58  My observation of an elk decline between 1978-1979 and 1996 (Figure 3.11) appears at odds with positive trends reported in the literature review (Section 3.3.2), however, my observations were complicated by wolf-free refugia created by human activity (Paquet et al. 1996) and exploited by elk (Woods et al. 1996; and see predictions of White et al. 1998). The concentration of elk in human-dominated habitats has been described as an "anti-predator strategy" in Jasper N.P. where elk congregate near busy roads and Jasper townsite (Dekker et al. 1995), and in Riding Mountain N.P. where elk cows and calves were frequently found near the edges of farmland where wolves were few relative to the interior of the park (Carbyn 1980).  Two of the reported increases in elk numbers (Panther and Red Deer; Skjonsberg 1993; Morgantini 1995) occurred in areas straddling the national park boundary (Figure 3.12), beyond which wolves are frequently trapped in winter, and are hunted year-round. In the Central Rockies, including the Red Deer and the Panther watersheds, shooting comprises 35.1% of annual wolf mortality, and mean annual wolf mortality rates from all causes are considered high (e. g., > 25.5% in the Bow Valley, 1987-1995, Paquet et al. 1996). M y pellet survey data were obtained in the subalpine portion of the Red Deer Valley inside the park, where human-caused wolf mortality and disturbance is minimal. The decline in pellet density there may reflect both the greater security of wolves, and the lower quality of cervid habitat, afforded in the park. I acknowledge that elk have increased in the lower Red Deer and Panther drainages beyond the park boundary. I also stress that my results pertain to winter distributions only.  The relatively stable elk trend reported in the Bow Valley between 1987 and 1996 was accompanied by a redistribution of the majority of wintering elk to a small area surrounding the town of Banff (Woods et al. 1996). In the townsite area, high levels of human activity and infrastructure displace wolves (Paquet et al. 1996). The 1996 pellet survey in the Bow Valley straddled the area of high elk density near the townsite, and the  59  adjacent areas where elk numbers have declined markedly (Woods et al. 1996). The estimated decline in elk pellet density in the Bow Valley is a reflection of their present highly clumped distribution.  4.2.2  Predator-Cervid Relationships.  The importance of elk in the diet of wolves during the period of wolf re-establishment, suggests that the numeric response of wolves following the cessation of predator control (Holroyd and Van Tighem 1983; Paquet 1993; Gunson 1992) was primarily a response to abundant and increasing elk (Woods 1991; Skjonsberg 1993; Morgantini 1995). Secondary predation on moose, however, represented a substantial impact to the moose population.  I found higher predation rates (0.174 ± 0.075, 95% C.I.) and lower survival rates (0.713 ± 0.089) for adult female moose than have been reported in other multiple-predator, multiple-prey systems where moose are low ranking prey. In a study area near the Montana-British Columbia border, where deer and elk were the primary prey, Kunkel (1997) reported annual combined bear and wolf mortality rate of approximately 7% on adult female moose. The relative contribution of the two predators was approximately equal. Kunkel (1997) found that predation was the most important source of mortality for moose, but that it was not sufficient to cause a decline. He concluded that a predator dilution effect (Messier 1994), caused by comparatively abundant and vulnerable whitetailed deer, contributed to the stability of the moose population. M y results are suggestive of Messier's (1994) alternative prediction, where a predator's numerical response towards its primary prey exacerbates its effects on lower ranked and less abundant prey species. Results similar to ours have been reported in different predator-cervid assemblies. For example, Seip (1992) and Bergerud and Elliot (1986) concluded that wolves responding numerically to moose, caused woodland caribou to decline.  60  4.2.3  Contributing Mortality Factors  Winter Snowpack Less wolf predation on moose during years of heavy snowpack, though not significantly different from light-snow years (P=0.132), suggested that snow depth may influence moose mortality. The relationship may reflect a wolf functional response to kill proportionately more elk and deer in deep snow, and less alternate prey like moose. Cervid prey are usually considered more vulnerable in deep snow (e. g., Mech and Frenzel 1971; Peterson 1977; Fuller 1991b; Boyd et al. 1994), however, moose are more snowtolerant than deer and elk (Telfer 1978). Both Huggard (1993c) and Kunkel (1997) found that wolves preyed proportionately more on elk and deer in deep snow because they became more vulnerable as the snow-pack increased. The results suggest that deep snow conditions offer moose a temporary reprieve from predation.  Age and Condition Eighty-eight percent of adult female moose (7/8) and 66% (8/12) of adult male moose that were preyed on by wolves or bears were of prime age classes, and all had femur fat content above the reported threshold where severe malnutrition was likely (Peterson et al. 1984; Gasaway et al. 1992). M y results suggest that predation, age, and condition were primarily non-compensatory, especially for female moose.  Giant Liver Fluke Infection The proportion of infected moose observed in the Bow Valley (53%) was similar to the 7year mean of 52% (n=22, 1985-1991) reported in the same area by Butterworth and Pybus (1993). Similarly, the low prevalence of infection beyond the east boundary of the study (4.5%) corresponded to a broader survey of the area (4%) by Pybus (1990, n=191). Because of the inherent difficulties in recovering intact livers from dead moose, it was impossible to test for compensation. However, I interpreted the high prevalence of G L F in moose in the Bow Valley as evidence that parasite-mediated competition between  61  moose and elk was possible. G L F infection and predation may be compensatory but not in the usual direct sense, rather, through the mechanism of apparent competition.  GLF rarely mature in moose (Lankester 1974) thus infections are only expected when definitive hosts like elk, white-tailed deer, or caribou are present (Wobeser et al. 1985; Lankester and Luttich 1988). Infections in definitive hosts are relatively benign (Pybus 1990) compared to the tissue damage that can be caused in moose (Lankester and Samuel 1998). Although detrimental effects on moose have not been clearly demonstrated, (Lankester and Samuel 1998), sub-lethal effects produced by heavy infections, if accompanied by malnutrition, may be fatal (Fenstermacher and Olsen 1942; Cowan 1946; Cheatum 1951).  Swales (1935; cited in Butterworth and Pybus 1993) noted that where definitive hosts were crowded, and intermediate snail hosts were abundant, infestations of large numbers of GLF would occur. Trends in G L F abundance in the Bow Valley correlate directly with the historical increases in elk abundance reported by Woods (1991) and Woods et al. (1996; Section 3.3.2 above). GLF prevalence in elk has increased from mean annual rate of 13.1% between 1958 and 1967 (Flook and Stenton 1969) to 64.6% between 19841991 (Butterworth and Pybus 1993). Butterworth and Pybus (1993) also reported a significant increasing trend within the period of their review (1984-1991) from 43% to 79%. More recently, Shury (1995) reported a 6-year mean annual prevalence of 79.8% (1989-1995). Under the conditions of abundant wetland habitat in the Bow Valley, G L F prevalence has apparently tracked elk abundance. The importance of G L F infection in causing moose mortality remains unknown, however, it is probable that any mortality effects caused by G L F are directly linked to elk abundance. Under conditions of apparent competition (predator- and parasite-mediated), G L F infection and predation may be compensatory.  62  4.3  INTERACTIONS BETWEEN COMPETITION AND APPARENT COMPETITION  Competition and apparent competition (predation) are often viewed as counter-balancing processes that facilitate coexistence between competitors (Holt et al. 1994). I found that the two "competition" processes operated at different scales and were apparently additive in their negative effect on moose. I examined some potential reasons for the behaviour of the system in the study area. The current relationship between the two large cervids may have been the result of their disparate abundances at the time wolf re-establishment began. Their future coexistence may be largely dependent on the character of the wolf-elk relationship, and on current human influences in the system.  4.3.1  Scale  My interpretation regarding the exclusion of moose by competitive exploitation is restricted to a subset of available habitats (willow-bearing vegetation types) and applies only when elk are abundant at a particular site. The process might be considered localized in scale. I did not test whether or not the process is operative across a full range of resources and habitats. M y findings regarding apparent competition apply to a much broader geographic scale. Wolves, sustained primarily by elk, killed moose throughout the study area regardless of the proximity of elk concentrations (Section 3.3, Figure 3.12). Wolves in BNP (Huggard 1993b) and elsewhere (e. g., Carbyn 1983) range throughout their territories rather than staying exclusively in areas of high prey density. Fritts and Mech (1981) observed that wolf kills tend to be more uniformly distributed than are prey. The combination of spatial scales and competitive processes (exploitative and apparent) appear additive with respect to their negative effect on moose. However, the reduced association between cervids, following the re-estabUshment of wolves (Section 3.1.1), supports the contention that predation has the potential to reduce the intensity of exploitative competition.  63  4.3.2  Competitive Ability and Predation Risk  Coexistence in mixed competition and predation systems requires that each species be able to increase when rare, and it is most likely to be observed when there is a clear trade-off between competitive superiority and predation risk (Holt et al. 1994). I have described the competitive superiority of elk, a result of their generalist feeding behavior and numeric dominance. Others have shown that wolves selectively prey on elk (especially elk living in groups), relative to other prey (Huggard 1993a, Weaver 1994), presumably because of the their greater vulnerability and numbers (profitability).  Although some important aspects of elk behaviour as prey are not fully understood, wolf predation appears capable of strongly affecting elk numbers (Huggard 1993a, Weaver 1994) . Given a sufficient numeric response, the wolfs sharp functional response towards elk (Huggard 1993a, Weaver 1994) suggests that selective predation could continue to the point where elk are limited at lower densities (currently under study in B N P ; M . Hebblewhite, U. of Montana). The pronounced range contraction of elk in both Banff and Jasper National Parks following wolf re-establishment (Woods et al. 1996; Dekker et al. 1995) , provides some empirical evidence of wolf effects on elk. In general however, conditions of abundant elk, and limited predation effects, have persisted in the study area despite wolf re-establishment.  Wolves were slow to re-establish their former range in the Central Rockies (Paquet 1993). Pack activity was not documented until approximately 30 years after wolf control had ended (mid 1950's to mid 1980's) in both the Bow Valley (Mickle et al. 1986) and Kootenay National Park (Poll et al. 1984). Reasons for the delay are unknown, but, they are most likely attributable to: 1) wolf displacement from areas of increasingly intense human use (Pacas 1996; Paquet et al. 1996), 2) localized predator control that occurred in Alberta until 1966 (Gunson 1992), 3) trapping and hunting that continued after the control programs ended (Gunson 1991; Paquet et al. 1996); and, 4) heavy highway and railway  64  mortality (Paquet et al. 1996). Consequently, either few wolves, or impaired wolf access to elk, and a strong potential for competition and apparent competition, continue to exist across many areas of the landscape.  If, under conditions of unfettered access to elk and low wolf mortality, wolves do limit elk at low density, coexistence between moose and elk would appear more probable. Moose are regarded as less vulnerable than elk and deer and they should be expected to derive a benefit from sympatry (a "dilution" effect, Messier 1994). However, the domination of the system by elk, appears to have exacerbated, rather than diluted predator impacts on moose. The strong effect of apparent competition, sustained by the presence of wolf-free refugia for elk, appears sufficient to overwhelm any differences in prey vulnerability and prey-specific predation rates.  65  5.0  M A N A G E M E N T IMPLICATIONS  My findings regarding the competitive relationship of moose and elk have implications for the management of large herbivore systems in parks. Competition studies are conducted usually in the context of the coevolution of species. Tests of niche overlap are often used to search for limiting similarities that result in habitat partitioning between species, a prediction of competition theory. In this study, I tested for the presence of competition as a diagnostic procedure to determine whether the large herbivore system in BNP may have diverged from the expected niche partitioning. I was less interested in the evolutionary importance of competition, than in the effects of humans on current relationships between moose and elk.  The results supported the hypotheses of competition and apparent competition. M y findings suggest that willow is an important food of moose, and that it is limited by elk browsing where elk occur in high numbers; that is, most of the montane habitats in the study area. Competition mediated by predators (apparent competition), especially wolves, also appears to be an important factor limiting the moose population. Wolves, sustained primarily by elk, exerted a strong impact on the moose population. Future coexistence between moose and elk appears unlikely given the current additive nature of the two effects.  The role of wolves could change, from that of an agent of apparent competition, to one that mediates coexistence between moose and elk, if elk were as vulnerable to wolf predation as suggested by Huggard (1993a, 1993b), Weaver (1994), and Kunkel (1997). However, a greater predation impact on elk is unlikely as long as wolves continue to avoid human-dominated refugia that are inhabited by large elk aggregations.  Given the support I found for my hypotheses, neither food nor predators appeared to be partitioned in a manner that nrinimized competition between moose and elk. Thus, the cervid-predator system appears divergent from the state predicted by competition theory.  66  The potential of wolf predation to restore niche partitioning seems substantial, and it may indicate the primacy of predation in the evolution of the system. M y findings suggest that the large herbivore system may have evolved from an alternative state of community organization, such as the top down predator regulation model described by White et al. (1998). Several other studies currently in progress with respect to elk-beaver relationships (C. Nietvelt, Univ. of Alberta), predator-prey modeling (M. Hebblewhite, Univ. of Montana), elk-aspen relationships (C. White, Univ. of British Columbia), and elk-refuge relationships (J. McKenzie, Univ. of Guelph) will allow further inferences regarding the past organization of the terrestrial community in the study area.  67  6.0  REFERENCES CITED  Abramsky, Z., M.A. Bowers, and M.L. Rosenzweig. 1986. Detecting interspecific competition in the field: testing the regression method. Oikos 47:199-204. Achuff, P.L., I. Pengelly, and J. Wierzchowski. 1996. Vegetation: cumulative effects and ecological futures outlook. Chapter 4 in Green, J., C. Pacas, L. Cornwell, and S. Bayley, eds. 1996. Ecological Outlooks Project: A Cumulative Effects Assessment and Futures Outlook of the Banff Bow Valley. Prepared for the Banff Bow Valley Study. Dept. of Canadian Heritage, Ottawa, O N . Ausperger, C. 1996. Editor's note. Ecology 77:1698. Ballard, W.B. 1992. Bear predation on moose: a review of recent North American studies and their management implications. Alces (Suppl.)l:l-15. Bergerud, A.T. and J.P. Elliot, 1986. Dynamics of caribou and wolves in northern British Columbia. Can. J. Zool. 64:1515-1529. Boyd, D.K., R.R. Ream, D.H. Pletscher, and M.W. Fairchild. 1994. Prey taken by colonizing wolves and hunters in the Glacier National Park area. J. Wildl. Mange. 58:289-295. Brieman, L., J.H. Freidman, R.A. Olshen, and C.J. Stone. 1984. Classification and regression trees. Wadsworth, Belmont, C A . Bunnell, F.L. 1987. Reproductive tactics of Cervidae and their relationships to habitat. Pp. 145-199 in Wemmer, C M . , ed. Biology and Management of the Cervidae. Smithsonian Institution Press, Washington, D.C. Bunnell, F.L. and M.P. Gillingham. 1985. Foraging behavior: dynamics of dining out. Pp. 53-79 in Hudson R.J., R.G. White, eds. Bioenergetics of Wild Herbivores. CRC Press, Boca Raton, F L . Butterworth, E. and M.J. Pybus. 1993. Impact of the giant liver fluke on elk and other ungulates in Banff and Kootenay National Parks. Canadian Parks Service, Calgary. Carbyn, L.N. 1980. Ecology and management of wolves in Riding Mountain National Park, Manitoba. Canadian Wildlife Service report for Parks Canada's Large Mammal Systems Study. Rep. No. 10:1-184.  68  Carbyn, L . N . 1983. Wolf predation on elk in Riding Mountain National Park, Manitoba. J. Wildl. Manage. 47:963-976. Caswell, H. 1989. Matrix population models. Sinauer, Sunderland, M A . Caughley, G. and A.R.E. Sinclair. 1994. Wildlife Ecology and Management. Blackwell Scientific Publications, Cambridge, M A . Cheatum, E.L. 1951. Disease in relation to winter mortality of deer in New York. J. Wildl. Manage. 15:216-220. Clark, L.A. and D. Pregibon. 1992. Tree-based models. Pp. 377-419 in Chambers, J.M. and T J . Hastie, eds. Statistical models in S. Wadsworth and Brooks/Cole. Advanced Books and Software, Pacific Grove, C A . Collins, W.B. and P.J. Urness. 1979. Elk pellet group distributions and rates of deposition in aspen and lodgepole pine habitats. Pp. 140-144 in M.S. Boyce and L.D. Hayden-Wing, eds. North American Elk: Ecology, Behavior, and Management. The Univ. of Wyoming Press, Laramie, Wyoming. Collins, W.B. and P J . Urness. 1981. Habitat preferences of mule deer as rated by pelletgroup distributions. J. Wildl. Manage. 45:969-972. Cowan, I. McT. 1943. Report on game conditions in Banff, Jasper, and Kootenay Parks. National Parks Bureau, Ottawa. Cowan, I. McT. 1946. Report of wildlife studies in Jasper, Banff and Yoho National Parks, 1944, and parasites, diseases and injuries of game animals in the Rocky Mountain national parks, 1942-1944. Can. Wildl. Serv., Ottawa. Cowan, I. McT. 1947. Range competition between mule deer, bighorn sheep and elk in Jasper National Park, Alberta. Trans. N. Am. Wildl. Conf. 12:233-247. Cowan, I. McT. 1950. Some vital statistics of big game on overstocked mountain range. Trans. N. Am. Wildl. Conf. 15:581-588 Dekker, D., W. Bradford, and J.R. Gunson. 1995. Elk and wolves in Jasper National Park, Alberta, from historical times to 1992. Pp. 85-94 in Carbyn, L.N., S.H. Fritts, and D.R. Seip, eds. Ecology and conservation of wolves in a changing world. Canadian Circumpolar Institute, Occasional Publication No. 35. Edge, W.D. and C.L. Marcum. 1989. Determining elk distribution with pellet-group and telemetry techniques. J. Wildl. Manage. 53:621-624.  69  Fenstermacher, R. and O.W. Olsen. 1942. Further studies of diseases affecting moose, III. Cornell Vet. 32:241-254. Flook, D.R. 1964. Range relationships of some ungulates native to Banff and Jasper national parks, Alberta. Pp. 119-128 in D.J. Crisp, ed. Grazing in Terrestrial and Marine Environments. Blackwell Scientific Publ, Oxford, England. Flook, D.R. and J. E. Stenton. 1969. Incidence and abundance of certain parasites in wapiti in the national Parks of the Canadian Rockies. Can. J. Zool. 47:795803. Franzmann, A.W., P.D. Arneson and J.L. Oldemeyer. 1976 a. Daily winter pellet groups and beds of Alaskan moose. J. Wildl. Manage. 40:374-375. Franzmann, A.W., J.L. Oldemeyer, P.D. Arneson and R.K. Seemel. 1976b. Pellet-group count evaluation for census and habitat use of Alaskan moose. Alces 12:127142. Freddy, D.J. and D.C. Bowden. 1983. Sampling mule deer pellet-group densities in pinyon-juniper woodland. J. Wildl. Manage. 47:512-516. Fritts, S.H. and L.D. Mech. 1981. Dynamics, movements and feeding ecology of a newlyprotected wolf population in north-western Minnesota. Wildl. Monogr. 80:179. Fuller, T.K. 1989. Population dynamics of wolves in north-central Minnesota. Wildl. Monogr. 105:1-41 Fuller, T.K. 1991a. Do pellet counts index white-tailed deer numbers and population change? J. W M . Manage. 55:393-396. Fuller, T.K. 1991b. Effect of snow depth on wolf activity in north central Minnesota. Can. J. Zool. 69:283-287. Gasaway, W.D., R.D. Boertje, D.V. Grangaard, D.G. Kellyhouse, R.O. Stephenson, and D.G. Larson. 1992. The role of predation in limiting moose at low densities in Alaska and Yukon and implications for conservation. Wildl. Monogr. 120:159. Green, H.U. 1949. The bighorn of the Banff National Park. Banff Warden Service, Banff, A B . Unpubl. Green, H.U. 1950. The occurrence, distribution, and population of mammals, Banff National Park. Banff Warden Service, Banff, A B . Unpubl.  70  Green. H.U. 1955. The moose of Banff National Park. Banff Warden Service, Banff, A B . Unpubl. Gunson, J.R. 1991. Management plan of wolves in Alberta. Wildlife management planning series. No. 4. Forestry, Lands and Wildlife, Fish and Wildlife Division. Gunson, J.R. 1992. Historical and present management of wolves in Alberta. Wildl. Soc. Bull. 20:330-339. Haigh, J.C., E.W. Kowal, W. Runge and G. Wobeser. 1982. Pregnancy diagnosis as a management tool for moose. Alces 18:45-53. Heisey, D.M. and T.K. Fuller. 1985. Evaluation of survival and cause-specific mortality rates using telemetry data. J. Wildl. Manage. 49:668-674. Holland, W.D. and G.M. Coen. 1982. Ecological (Biophysical) Land Classification of Banff and Jasper National Parks. Volume II: Soil and vegetation resources. Alberta Institute of Pedology, University of Alberta, Edmonton, A B . Holroyd, G.L. 1979. The impact of highway and railroad mortality on the ungulate populations in the Bow Valley, Banff National Park. Canadian Wildlife Service, Banff, A B . Unpubl. Holroyd, G.L. and K.J. Van Tighem. 1983. Ecological (Biophysical) Land Classification of Banff and Jasper National parks. Volume III. The wildlife inventory. Canadian Wildlife Service, Edmonton, A B . Holt, R.D. 1977. Predation, apparent competition, and the structure of prey communities. Theoretical Population Biology 12:197-229. Holt, R.D., J. Grover, and D. Tilman. 1994. Simple rules for interspecific dominance in systems with exploitative and apparent competition. Am. Nat. 144:744-771. Houston, D.B. 1982. The Northern Yellowstone Elk: Ecology and Management. Macmillan, New York. Huff, D.E. and J.D. Varley. 1999. Natural regulation in Yellowstone National Park's northern range. Ecol. Appl. 9:17-29. Huggard, D.J. 1993a. Prey selectivity of wolves in Banff National Park. I. Prey species. Can. J. Zool. 71:130-139. Huggard, D.J. 1993b. Prey selectivity of wolves in Banff National Park. II. Age, sex and condition of elk. Can. J. Zool. 71:140-147.  71  Huggard, D J . 1993c. Effect of snow depth on predation and scavenging by gray wolves. J. Wildl. Manage. 57:382-388. Hurlbert, S.H. 1969. A coefficient of interspecific association. Ecology 50:1-9. Janz, B. and D. Storr. 1977. The climate of the contiguous mountain parks; Banff, Jasper, Yoho, Kootenay. Env. Can. Atmos. Env. Serv., Met. Applic. Br., Proj. Rep. No. 30. Jenkins, K.J. and R.G. Wright. 1986. Dietary niche relationships among cervids relative to winter snowpack in northwestern Montana. Can J. Zool. 65:1397-1401. Jenkins, K.J. and R.G. Wright. 1988. Resource partitioning and competition among cervids in the Northern Rocky Mountains. J. Appl. Ecol. 25:11-24. Joyal, R. and J-G. Ricard. 1986. Winter defecation output and bedding frequency of wild, free-ranging moose. J. Wildl. Manage. 50:734-736. Kay, C E . 1990. Yellowstone's northern elk herd: A critical evaluation of the "natural regulation" paradigm. Ph. D. Dissertation, Utah State Univ., Logan, UT. Krebs, C.J. 1989. Ecological Methodology. Harper Collins, New York. Krebs, C.J., B.S. Gilbert, S. Boutin, A.R.E. Sinclair, and J.N.M. Smith. 1986. Population biology of snowshoe hares I. Demography of a food supplemented population in southern Yukon, 1976-1984. J. Anim. Ecol. 55:963-982. Kunkel, K.E. 1997. Predation by wolves and other large carnivores in northwest Montana and southeast British Columbia. Ph. D. Dissertation. Univ. of Montana, Missoula, M T . Lankester, M.W. 1974. Parelaphonstrongylus  tenuis (Nematoda) and  Fascioloides  magna (Trematoda) in moose in southeastern Manitoba. Can. J. Zool. 52:235-239. Lankester, M.W. and S. Luttich. 1988. Fascioloides magna (Trematoda) in woodland caribou (Rangifer tarandus caribou) of the George River Herd, Labrador. Can. J. Zool. 66:475-479. Lankester, M.W. and W . M . Samuel. 1998. Pests and diseases. Pp. 479-518 in Franzmann, A.W. and C C . Schwartz, eds. Ecology and Management of the North American Moose. Smithsonian Institution Press, Washington, D.C. Larsen, D.G., D.G. Gauthier, R.L. Markel. 1989. Causes and rate of moose mortality in the Southwest Yukon. J. Wildl. Manage. 53:548-557.  72  Lautenschlager, R.A. and P. A. Jordan. 1993. Potential use of track-pellet group counts for moose censusing. Alces 29:175-180. Leopold, B.D., P.R. Krausman, and J.J. Hervert. 1984. Comment: the pellet-group census technique as an indicator of relative habitat use. Wildl. Soc. Bull. 12:325-326. Loft, E.R. and J.G. Kie. 1988. Comparison of pellet-group and radio triangulation methods for assessing deer habitat use. J. Wildl. Manage. 52:524-527. Ludwig, J.A., and J. F. Reynolds. 1988. Statistical Ecology: a Primer on Methods and Computing. John Wiley and Sons, New York. MacNally, R.C. 1983. On assessing the significance of interspecific competition to guild structure. Ecology 64:1646-1652. Mair, W.W. 1952. The impact of an introduced population of elk upon the biota of Banff National Park. M . Sc. Thesis, Univ. of British Columbia, Vancouver, B.C. McGregor, C.A. 1984. Ecological land classification and evaluation: Kananaskis Country. V o l 1: Natural resource summary. Alberta Energy and Nat. Res., Resource Evaluation and Planning Div., Edmonton, A B . McMillan, J.F. 1953. Measures of association between moose and elk on feeding grounds. J. Wildl. Manage. 17:162-166. Mech, L.D. and L.D. Frenzel. 1971. Ecological studies of the timber wolf in northeastern Minnesota. For. Serv. Paper NC-52, North Central For. Exp. Stn., St. Paul, MN. Messier, F. 1994. Ungulate population models with predation: a case study with the North American moose. Ecology 75:478-488. Messier, F. 1995. On the functional and numerical responses of wolves to changing prey density. Pp. 187-197 in Carbyn, L.N., S.H. Fritts, and D.R. Seip, eds. Ecology and conservation of wolves in a changing world. Canadian Circumpolar Institute, U. of Alberta, Edmonton, A B . Mickle, D., G. Fortin, and R. Kunelius. 1986. Status of wolves: Banff National Park. Parks Canada, Banff, A B . Unpubl. Moen, R., J. Pastor, and Y . Cohen. 1997. A spatially explicit model of moose foraging and energetics. Ecology 78:505-521.  73  Morgantini, L.E. 1995. The Y a Ha Tinda: A n ecological overview. A report prepared for Canadian Heritage, Parks Canada, Calgary, A B . Neff, D.J. 1968. The pellet-group count technique for big game trend, census, and distribution: a review. J. Wildl. Manage. 32:597-614. Neff, D.J., O.C. Wallmo, and D.C. Morrison. 1965. A deteraiination of defecation rate for elk. J. Wildl. Manage. 29:406-407. Neiland, K.A. 1970. Weight of dried marrow fat in caribou femurs. J. Wildl. Manage. 34:904-907. Nelson, J. R. and T.A. Leege. 1982. Nutritional requirements and food habits. Pp. 323367 in J. W. Thomas and D.E. Toweill, eds. Elk of North America: Ecology and Management. Stackpole Books, Harrisburg, Pennsylvania. O'Connor, R.J., M.T. Jones, D.White, C. Hunsaker, T. Loveland, B. Jones and E. Preston. 1996. Spatial partitioning of environmental correlates of avian biodiversity in the conterminous United States. Biodiversity Letters 3:97-110. Oldemeyer, J.L. and A.W. Franzmann. 1981. Estimating winter defecation rates for moose, Alces alces. Can. Field-Nat. 95:208-209. Oldemeyer, J.L. and W.L. Regelin. 1987. Forest succession, habitat management, and moose on the Kenai National Wildlife Refuge. Swedish Wildlife Res. Suppl. 1:163-183. Pacas, C. 1996. Human use of the Banff Bow Valley: past, present and future. Chapter 3 in Green, J., C. Pacas, L. Cornwell and S. Bayley, eds. Ecological Outlooks Project. A Cumulative Effects Assessment and Futures Outlook of the Banff Bow Valley. Prepared for the Banff Bow Valley Study. Department of Canadian Heritage, Ottawa, O N . Packard, J.M. and L.D. Mech. 1980. Population regulation in wolves. Pp. 135-150 in Cohen, M.N., R.S. Malpass, and H.G. Klein, eds. Biosocial Mechanisms of Population Regulation. Yale Univ. Press, New Haven, CT. Paquet, P.C. 1993. Summary reference document - ecological studies of recoIonizing wolves in the Central Canadian Rocky Mountains. Prepared by John/Paul and Associates for Parks Canada, Banff National Park Warden Service, Banff, AB. Paquet, P . C , J. Wierzchowski, and C. Callaghan. 1996. Effects of human activity on gray wolves in the Bow River Valley, Banff National Park, Alberta. Chapter 7 in Green, J., C. Pacas, L. Cornwell and S. Bayley, eds. Ecological Outlooks  74  Project. A Cumulative Effects Assessment and Futures Outlook of the Banff Bow Valley. Prepared for the Banff Bow Valley Study. Department of Canadian Heritage, Ottawa, O N . Paquet, P.C., C. Callaghan, C. McTavish, M . Hebblewhite. 1998. 670 Wolf kill observations in the Central Rocky Mountains, 1988-1997. Central Rockies Wolf Project, unpubl. data. Peek, J.M. 1974. A review of moose food habit studies in North America. Nat. Can. 101:195-215. Peterson, R.O. 1977. Wolf ecology and prey relationships on Isle Royale. U.S. Natl. Park Serv. Sci. Monogr. Ser. 11:1-210 Peterson, R.O., J. D. Woolington, and T.N. Bailey. 1984. Wolves of the Kenai Peninsula, Alaska. Wildl. Monogr. 88:1-52. Pierce, J.D. 1984. Shiras moose forage selection in relation to browse availability in northcentral Idaho. Can. J. Zool. 62:2404-2409. Poll, D.M., M . M . Porter, G.L. Holroyd, R.M. Wersheler, L.W. Gyug. 1984. Ecological land classification of Kootenay National Park, B.C. Vol. II. Wildlife Resources. Canadian Wildlife Service report to Parks Canada, Calgary, A B . Pybus, M.G. 1990. Survey of hepatic and pulmonary parasites of wild cervids in Alberta, Canada. J. Wildl. Dis. 26:453-459. Reynoldson, T.B. and L.S. Bellamy. 1971. The establishment of interspecific competition in field populations, with an example of competition in action between Polycelis nigra (Mull.) and P. Tenuis (Ijima) (Turbellaria, Tricladida). Pp. 282-297 in den Boer, P.J. and G.R. Gradweel, eds., Dynamics of Populations. Proceedings of the Advanced Study Institute on dynamics of numbers in populations, Oosterbeek, The Netherlands, 7-18 September 1970, Wageningen: Centre for Agrciultural Publishing and Documentation. Risenhoover, K.L. 1987. Winter foraging strategies of moose in subarctic and boreal forest habitats. Ph. D. Dissertation. Michigan Technical Univ., Houghton. Rogers, L.L. 1987. Seasonal changes in defecation rates of free-ranging white-tailed deer. J. Wildl. Manage. 51:330-333. Rougeau, M-P. and D. Gilbride. 1994. Forest stand origin mapping of Banff National Park. Parks Canada, Banff National Park, Banff, A B .  75  Roughgarden, J. 1983. Competition and theory in community ecology. Am. Nat. 122:583601. Sand, H. 1996. Life history patterns in female moose (Alces alces): the relationship between age, body size, fecundity and environmental conditions. Oecologia 106:212-220. Sawyer, T.G., R.L. Marchinton, and W . M . Lentz. 1990. Defecation rates of female whitetailed deer in Georgia. Wildl. Soc. Bull. 18:16-17. Schluter, D. 1984. A variance test for detecting species associations, with some example applications. Ecology 65:998-1005. Schwartz, C C . and A.W. Franzmann. 1989. Bears, wolves, moose and forest succession, some management considerations on the Kenai Peninsula, Alaska. Alces 25:110. Schwartz, C C 1998. Reproduction, natality, and growth. Pp. 141-171 in Franzmann, A.W. and C C Schwartz, eds. Ecology and Management of the North American Moose. Smithsonian Institution Press, Washington, D.C. Seip, D. R. 1992. Factors limiting woodland caribou populations and their interrelationships with wolves and moose in southeastern British Columbia. Can. J. Zool. 70:1494-1503. Sergeant, D.E. and D.H. Pimlott. 1959. Age detennination in moose from sectioned incisor teeth. J. Wildl. Manage. 23:315-321. Shipley, L.A. and D.E. Spalinger. 1995. Influence of size and density of browse patches on intake rates and foraging decisions of young moose and white-tailed deer. Oecologia 104:112-121. Shury, T. K. 1995. A n update of giant liver fluke (Fascioloides magna) infection in ungulates of Banff National Park 1989-1994. Parks Canada, Banff, A B . Unpubl. Singer , F. J. 1979. Habitat partitioning and wildfire relationships of cervids in Glacier National Park, Montana. J. Wildl. Manage. 43:437-444. Singer, F.J. and J.E. Norland. 1994. Niche relationships within a guild of ungulate species in Yellowstone National Park, Wyoming, following release from artificial controls. Can. J. Zool. 72:1383-1394. Skjonsberg, T. 1993. Eastern slopes wildlife study, 1984-1987. Final technical report. Parks Canada, Banff National Park, Banff, A B .  76  Smith, A.D. 1968. Defecation rates of mule deer. J. Wildl. Manage. 28:435-444. SPSS. 1997. SYSTAT 7.0: New Statistics. SPSS Inc., Chicago, IL. Stewart, R.R., E. H. Kowal, R. Beaulieu, and T.W. Rock. 1985. The impact of black bear removal on moose calf survival in east central Saskatchewan. Alces 21:403418. Swales, W.E. 1935. Researches on liver fluke in deer. Pp. 406-411 in Transactions of the 2 l'st Game Conference. New York, N Y . Tanner, H.C. 1950. An investigation into the competition between elk and big-horned sheep in the Cascade Valley, Banff National park, Alberta. Canadian Wildlife Service, Edmonton, A B . Unpubl. Telfer, E.S. 1970. Winter habitat selection by moose and white-tailed deer. J. Wildl. Manage. 34:553-559 Telfer, E.S. 1978. Cervid distribution, browse, and snow cover in Alberta. J. Wildl. Manage. 42:352-361. Tilman, D. 1990. Mechanisms of plant competition for nutrients: the elements of a predictive theory of competition. Pp. 117-142 in Grace, J.B. and D. Tilman, eds. Perspectives on Plant Competition. Academic Press, San Diego, C A . Timmermann, H.R. and M.E. Buss. 1998. Population and harvest management. Pp. 559615 in Franzmann, A.W. and C C . Schwartz, eds. Ecology and Management of the North American Moose. Smithsonian Institution Press, Washington, D.C. Thomson, J.D., G. Weiblen, B.A. Thomson, S. Alfaro, and P. Legendre. 1996. Untangling multiple factors in spatial distributions: lillies, gophers, and rocks. Ecology 77:1698-1715. Van Wagner, C.E. 1995. Analysis of fire history for Banff, Jasper, and Kootenay National Parks. Report to Parks Canada, Ottawa. Vivas, H.J., B.E. Saether, and R. Andersen. 1991. Optimal twig-size selection of a generalist herbivore , the moose, Alces alces: implications for plant-herbivore interactions. J. Anim. Ecol. 60:395-408. Wagner, F.H., R. Foresta, R.B. Gill, D.R. McCullough, M . R. Pelton, W.F. Porter, and H. Salwasser. 1995. Wildlife Policies in the U.S. National Parks. Island Press, Washington, D.C.  77  Weaver, J.L. 1994. Ecology of wolf predation amidst high ungulate diversity in Jasper National Park, Alberta. Ph. D. Dissertation. Univ. of Montana, Missoula, M T . White, C A . 1985. Wildland fires in Banff National Park 1880-1989. Parks Canada Occasional Paper 3:1 -106. White, C.A., D. Gilbride, M . Scott-Brown, and C. Stewart. 1995. Atlas of the Central Rockies Ecosystem, towards an ecologically sustainable landscape. A report to the Central Rockies Interagency Liaison Group (CREILG), Komex International, Calgary, A B . White, C.A., C E . Olmsted, and C E . Kay. 1998. Aspen, elk, and fire in the Rocky Mountain national parks of North America. Wildl. Soc. Bull. 26:449-462. White, G . C 1992. Do pellet counts index white-tailed deer numbers and population change?: a comment. J. Wildl. Manage. 56:611-612. White, R.G. 1983. Foraging patterns and their multiplier effects on productivity of northern ungulates. Oikos 40:377-384. Wiens, J.A. 1994. The Ecology of Bird Communities. Volume 2: Processes and variations., Cambridge University Press, N Y . Wobeser, G., A . A . Gajadhar and H.M. Hunt. 1985. Fascioloides magna: occurrence in Saskatchewan and distribution in Canada. Can. Vet. J. 26:241-244 Woods, J.G. 1990. Effectiveness of fences and underpasses on the Trans-Canada Highway and their impact on ungulate populations in Banff National Park, Alberta. Canadian Parks Service, Calgary. Woods, J.G. 1991. Ecology of a partially migratory elk population. Ph. D. Dissertation, Univ. of British Columbia, Vancouver. Woods, J.G., L.Cornwell, T. Hurd, R.Kunehus, P. Paquet, and J. Wierchowski. 1996. Elk and other ungulates. Chapter 8 in Green, J., C. Pacas, L. Cornwell, and S. Bayley, eds. Ecological Outlooks Project: A Cumulative Effects Assessment and Futures Outlook of the Banff Bow Valley. Prepared for the Banff Bow Valley Study. Dept. of Canadian Heritage, Ottawa, O N . Wright, R.G. 1999. Wildlife management in the national parks: questions in search of answers. Ecol. Appl. 9:30-36. Zar, J.H. 1984. Bio statistical Analysis. Prentice-Hall, New Jersey.  78  APPENDIX I Willow-bearing Vegetation Types Banff National Park  O 0" s  CO < CO <  3,  3,  bir ch-  birch-  birCh-  CA  CA  en  3,  o  O. 3  nc  2 o CT •?  I Ic r CT >< >< §. O  §L  9  l-wil  -willow  -willow  &  CT  &  o.  CT S  p  CD CO CD  sr o 3  O  o o  < CD CQ  o.  O  CT  3  •o  -3  CD CA ><  3  CT O  P  r  - 2  to  &  s-1  5'  O.  3  §L n  CO u  3 <  o  o  O  § 2  O CD 3 CD  -m  CT  1=2  ss  «  5  w  n °  o _ ai  B  8 p  w  rtl  8  2.  P  n>  < s.: »< * >  " CT o  5" N O  o  O  -S  m  3 CO  2. 3  0) 3 3 CA T3  2  5-  O  2 ° 25 a o (D •D 3'  3! IB O  ff •2  E  =S CO  I (O c cn  ff •2  <n c CT 0). T J  •a (D CO "< CD ca  o CT O T3  CT  o 3  (D (A •<  3  CT o  o  a  EJ.  O  1' £CT«3 B  CL  <D A )  O 3  o  O EJ ST CT X  ft  M m o 5  00  era  3  H •<  CT C  o  O oo  o  CD  V)  •2  s  CD  •3  c  < CD CQ  a  CO  3  65  8 CT E3 w  9L c co  Bi rt  n  8*8* i-t  cr C/>  A P P E N D I X II Snowpack Depth Mount Norquay, Banff, Alberta 1969-1997  


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