UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Home range behavior of Roosevelt elk in Strathcona Park Sovka, David G. 1993

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

Item Metadata

Download

Media
831-ubc_1993_spring_sovka_david.pdf [ 4.5MB ]
Metadata
JSON: 831-1.0086167.json
JSON-LD: 831-1.0086167-ld.json
RDF/XML (Pretty): 831-1.0086167-rdf.xml
RDF/JSON: 831-1.0086167-rdf.json
Turtle: 831-1.0086167-turtle.txt
N-Triples: 831-1.0086167-rdf-ntriples.txt
Original Record: 831-1.0086167-source.json
Full Text
831-1.0086167-fulltext.txt
Citation
831-1.0086167.ris

Full Text

HOME RANGE BEHAVIOR OF ROOSEVELT ELK IN STRATHCONA PARK by DAVID GRANT SOVKA B.Sc., The University of Calgary, 1990 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Department of Animal Science) We accep this thesis as conforming e requ' ed standard  THE UNIVERSITY OF BRITISH COLUMBIA April 1993 © David Grant Sovka, 1993  In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission.  Department of  /40/44/-1  SC:-/ '5771/  The University of British Columbia Vancouver, Canada  Date .4/E7/4-  DE-6 (2/88)  .  -II-  Abstract This thesis investigates the population, group composition, and home range sizes of 5 groups of Roosevelt elk (Cervus elaphus roosev -It) in 4 regions of Strathcona Provincial Park on Vancouver Island, British Columbia. Ten elk were successfully radio-collared and re-located for 12-21 months, beginning in March 1991, and ending in January 1993. Eight of the 10 study animals were nonmigratory. Sizes of cumulative and seasonal ranges, and core use areas were determined for each study animal using the harmonic mean estimator of Program Home Range. Three seasonal models (summer, mild winter, and severe winter) of habitat suitability were tested using a non-migratory elk group. A Geographic Information System (GIS) was used to create a digital map of the study area. The models were applied to the map to generate habitat suitability values for each of the polygons comprising the study area. This study provides additional support for a correlation between habitat suitability as predicted by the model and elk habitat selection. Elk were not randomly selecting habitats in the study area with respect to modeled habitat suitability values. Habitat suitability of elk cumulative range was higher than portions of the study area unused by elk. Core use areas had higher habitat suitability than cumulative range. No one model best predicted the habitat suitability of the home range selected by the non-migratory study animals. Use of the model as a predictor of suitable elk habitat in areas currently unused by elk is validated, but not warranted given the time and money required to collect the information necessary to use the model.  Table of Contents Abstract^  ii  Table of Contents^  iii  List of Tables^  v  List of Figures^  vi  Acknowledgement^  vii  GENERAL INTRODUCTION ^ Classification and Distribution^ Current Elk Population Status on Vancouver Island ^ Roosevelt Elk Management on Vancouver Island^ Roosevelt Elk Management in Strathcona Park ^  1 1 2 3 4  CHAPTER 1. POPULATION AND HOME RANGE CHARACTERISTICS FOR ROOSEVELT ELK IN STRATHCONA PARK^ Introduction^ Population and Home Range Concepts^ Measuring Population and Home Range ^ Study Area^ Study Area Boundaries ^ Elk River Valley^ Thelwood Valley^ Heber River-Camel Creek Valleys^ Ucona River Basin^ Methods^ Trapping and Radio-Collaring Animals^ Population and Group Composition Data ^ Animal Locations^ Home Range Delineation^ Animal/Season Selection Criteria ^ Home Range Analysis^ Core Use Areas^ Statistical Analyses^ Results^ Mortalities^ Animal Locations^ Group Composition and Population^ Home Range Characteristics and Movements^ Elk River Valley^ Thelwood Valley^  7 7 7 9 10 10 13 14 15 16 17 17 18 18 22 27 30 31 33 35 35 35 35 38 38 46  -ivHeber River-Camel Creek Valleys^ 48 Ucona River Basin^ 49 Discussion^ 50 Mortalities^ 50 Home Range Characteristics and Movements ^ 51 Elk River Valley^ 52 Thelwood Valley^ 53 Heber River-Camel Creek Valleys ^ 54 55 Ucona River Basin^ Migration and Non-migration as Movement Strategies^56 Home Range Estimation^ 62 Summary^ 65 CHAPTER 2. TESTING SEASONAL MODELS OF ROOSEVELT ELK HABITAT SUITABILITY^ 67 Introduction^ 67 Habitat Suitability Index (HSI) Model Development and Testing ^68 GIS Applications in Wildlife Habitat Modelling ^ 70 Study Background - Brunt's (1991) Thesis^ 72 75 Study Area^ Study Area^ 75 Historical Overview^ 76 Methods^ 77 Study Area Map Development^ 77 Model Development and Application^ 78 Forage^ 79 Cover^ 82 Interspersion^ 83 Aspect/Elevation^ 85 Habitat Suitability Calculations ^ 87 Home Range Suitability Calculations^ 88 Analyses of Modeled Habitat Suitability ^ 89 A) Random Habitat Selection Test^ 89 B) Habitat Suitability Within Study Area ^ 90 C) Comparisons Among Seasonal Model Prediction ^92 Results^ 93 Habitat Suitability Model Validation Tests^ 93 A) Random Habitat Selection Test^ 93 B) Habitat Suitability Within Study Area ^ 93 C) Comparisons Among Seasonal Model Output^97 Discussion^ 97 Conclusions^ 105 LITERATURE CITED^  109  -vList of Tables Table 1.^Dates used to delineate seasonal home ranges for each study animal. 28 Table 2.^Home ranges analyzed using the harmonic mean contour method. 29 Table 3.^Core use area delineation methods and core area and home range sizes.^ 34 Table 4.^Classification counts for summer and winter seasons. ^36, 37 Table 5.^Sizes (km 2 ) of seasonal home range and core use areas for 2 nonmigratory elk from the Heber River-Camel Creek valleys.^44 Table 6.^Sizes (km 2 ) of cumulative (year-round home range using 100% of utilization distribution) and core use areas for all study animals. ^45 Table 7.^Habitat types used by the Elk River valley group during summer 1992 (May 21-August 30).^ 47 Table 8.^Forage modifier values.^  65  Table 9.^Cover and potential forage suitability values by understory type. ^66 Table 10.^Interspersion modifier values.^  69  Table 11. Aspect/elevation modifier values used in the mild and severe winter models.^ 71 Table 12.^Frequencies of expected (given random habitat selection) and observed habitat suitability class values for the summer model. ^80 Table 13.  Habitat suitability values for total study area, unused portion of the total study area, cumulative range, and core use area for summer, mild winter, and severe winter models.^ 81  Table 14.  Mean habitat suitability values for animal locations in the core use area, and random locations in the unused portion, of the Elk River valley study area. 83  Table 15.^Mean habitat suitability values of summer, winter, and cumulative animal locations estimated by each of the 3 seasonal models. ^84 Table 16.^Results of all possible two-tailed paired t-test comparisons among summer, winter, and cumulative animal location datasets using the 3 habitat suitability models.^ 85  -viList of Figures Figure 1.^The 4 study areas in Strathcona Provincial Park on Vancouver Island, British Columbia.^ 11 Figure 2.^Cumulative range and core use area for elk it's 120, 179, 218, and 259, in the Elk River valley.^ 39 Figure 3.^Cumulative range and core use areas for elk #580 in the Thelwood valley.^ 40 Figure 4.^Summer and winter (1991-92) seasonal ranges and core use areas for elk #499 in the Heber River-Camel Creek valleys. ^ 41 Figure 5.^Summer and winter (1991-92) seasonal ranges and core use areas for elk #688 in the Heber River-Camel Creek valleys. ^ 42 Figure 6.^Cumulative range and core use areas of elk it's 159, 300, and 339 in the Ucona River basin.^ 43 Figure 7.^Total study area, unused portion of the study area, cumulative range, and core use area for elk it's 120, 179, 218, and 259 in the Elk River valley, as delineated for seasonal habitat suitability model testing. 94  -viiAcknowledgements This research was supported financially by the Ministry of Environment, Lands and Parks; the Canadian Wildlife Service; the Vancouver Island Habitat Island Enhancement Fund; and David Shackleton. I would like to thank Rik Simmons of BC Parks for administrative support, and also Ron Quitter and Joanne McLeod of the Strathcona Zone Office. Kim Brunt provided assistance throughout my study, including darting and collaring the study animals, data analysis, and helpful suggestions in the field. I would like to thank David Shackleton, Fred Bunnell, and Mike Pitt of the University of British Columbia, and Rik Simmons of BC Parks for participating on my graduate committee. Strathcona Park Ranger Rob Robertson assisted in data collection several times. Mike Scott of VideoWave Productions, Inc. (Courtenay, BC) accompanied me on several backcountry trips to search for elk, and videoed a short documentary of the study. Thanks to Jerry Maedel and Peter Murtha of the University of British Columbia's FIRMS/GIS Laboratory for allowing me to use the equipment. I would like to thank my beloved for her support and encouragement throughout this study. She tolerated my numerous absences over the past two years; missed me; loved me; and even accompanied me several times in the field up steep, dangerous logging roads in our 1972 Mercury, and all the while 6 months pregnant with our good and perfect gift. I would like to thank him too, for the great joy that he is. Above all, though, thanks to Him who's creation I studied with wonder and fear; who protected and cared for me in spite of so much.  GENERAL INTRODUCTION Classification and Distribution Currently, 2 subspecies of elk occur in British Columbia: Roosevelt elk  (Cervus elaphus roosevelti) and Rocky Mountain elk (C. e. nelsoni). Rocky Mountain elk are the more abundant and widespread, occurring in greatest numbers in the Rocky Mountains and the Rocky Mountain trench of southeastern British Columbia. Roosevelt elk occur in Canada only on Vancouver Island, which is the northernmost limit of their present-day natural distribution; in valleys at the head of several mainland inlets adjacent to northern Vancouver Island; and on the Sechelt Peninsula as a result of several transplants in the late 1980's (Brunt 1990). Historically, elk were more widely distributed in coastal British Columbia than they are at present (Cowan and Guiguet 1965, Bryant and Maser 1982). Local extinctions were caused by human settlement and excessive hunting, particularly on southern Vancouver Island. The present reduced numbers and limited distribution of elk on the Island result from a number of factors, including habitat quality, historical hunting and predation pressures, geographical barriers to colonization, transplant efforts, and a certain amount of chance patterns of dispersal (Brunt 1990). Roosevelt elk population numbers are considered to be far below their maximum sustainable level (Nyberg and Janz 1990). A discussion of the classification and historical distribution of elk in North America can be found in Bryant and Maser (1982), together with a thorough  -2physical description of the Roosevelt elk subspecies. Geist (1982) has reviewed elk social behavior. Current Elk Population Status on Vancouver Island An accurate census of ungulate populations in the coastal forests found on Vancouver Island is virtually impossible because cover is so dense. Recent estimates of elk populations prepared for the regional wildlife and habitat plans of the British Columbia Ministry of Environment, Lands, and Parks (MOELP) indicate that Roosevelt elk number between 2170 and 2770 (Nyberg et al. 1990). The Wildlife Species Evaluation List describes the status and prognosis of the wildlife species in British Columbia. In 1991, the MOELP placed Roosevelt elk on the blue list (sensitive/vulnerable species). The reasons for blue-listing species include the following: i) a significant current or predicted downward trends in population numbers or density; and ii) significant current or predicted downward trends in habitat suitability that would further reduce the species' existing distribution. Elk populations on Vancouver Island have declined over the past fifteen years. Climate, habitat changes, and hunting have most affected elk on Vancouver Island. Effects of the severe 1968-69 winter were most pronounced on southern Vancouver Island, where because of past logging history, few old-growth winter ranges remained, while there were many large areas dominated by young, regenerating forests. Elk populations declined during this period because of deep snow accumulations, resulting from the loss of canopy cover (Nyberg et al. 1990).  -3Over the next decade, elk numbers gradually recovered in response to a series of mild winters, but increased illegal hunting and unregulated native hunting have slowed the elk recovery (MOELP 1989a). Nyberg et al. (1990) describe the present status of Roosevelt elk on northern Vancouver Island as "more encouraging," where most herds are stable or increasing because of the recent mild winters, a favorable habitat mosaic in many watersheds, and relatively low levels of illegal hunting. Wolf predation affects some herds in watersheds where deer numbers are low and wolves are switching to elk as an alternative prey (Janz and Becker 1986). In such areas, the effects of wolf predation on elk populations are likely to increase during more severe winters as elk congregate on small winter ranges and become less mobile and less healthy. Roosevelt Elk Management on Vancouver Island Management of the provincial wildlife resource is the responsibility of the Wildlife Branch, Ministry of the Environment (MOELP). The mission of the wildlife program is "...to manage the province's wildlife resources for the benefit and enjoyment of British Columbians, by maintaining an optimal balance between ecological, cultural, economic, and recreational needs" (MOELP 1989b). A major goal is to maintain and enhance wildlife and their habitats to ensure an abundant, diverse, and self-sustaining wildlife resource throughout British Columbia. The present management objective for Roosevelt elk on Vancouver Island is to increase elk numbers to approximately 3000-3800 by 1996, which should generate 2350-3200 hunter-days of recreation and provide increased viewing  -4opportunities (Nyberg et al. 1990). Management activities required to meet population objectives include the curtailment of illegal and unregulated hunting; habitat management focusing on protection of important old-growth winter ranges and special habitats such as wetlands and vegetated slides; provision of forage and cover in regenerating forests; and manipulation of young stands to provide snow interception cover. At present, the status of Roosevelt elk in British Columbia is listed as "stable to increasing (2,500)" (BC Wildlife Branch 1991). However, significant losses to traditional elk habitat have occurred, even within supposedly protected areas such as Strathcona Provincial Park on Vancouver Island (Strathcona Park Elk Management Plan, 1990). Roosevelt Elk Management in Strathcona Park Within the MOELP, the Parks branch (BC Parks) is responsible for the designation, management and conservation of a land and water-based system of parks, recreation areas and ecological reserves, containing the best representative elements of British Columbia's natural and cultural heritage for the inspiration and recreational use of British Columbians and their visitors (BC Parks' Mandate, 1990). As such, in 1990 BC Parks developed a management plan for Strathcona Park which set out objectives outlining the government's stance on the future of elk in the Park. The Strathcona Park Elk Management Plan states that "...maintaining a healthy viable population of elk in the park is the primary goal of the park  -5management plan as well as satisfying park system objectives." One objective of the park system is to represent the biodiversity of the province. BC Parks determined that Roosevelt elk, as a keystone species of the province of British Columbia, "...must be maintained not only in their present but also their historic distribution and abundance," (Strathcona Park Elk Management Plan, 1990). At the same time, however, it was realized that "...the present state of knowledge is limited and does [not) include a complete understanding of existing habitat, utilization, distribution and abundance by elk or allow an assessment of other areas of the park's significance to elk." In light of the Ministry's goal "...to provide an environment in which the continued presence of elk is assured in Strathcona Park in perpetuity," (Strathcona Park Elk Management Plan, 1990), a set of objectives was drafted, which include the following goals: protecting and enhancing elk habitat, increasing elk abundance and distribution, and increasing opportunities for viewing elk. Within the 3 broad goals identified in the Strathcona Park Elk Management Plan (1990), broad action steps were defined. For the goal of protecting and enhancing elk habitat, the defined action steps included determining the present elk distribution and abundance in the Park, and developing "...a park specific understanding of elk ecology and capability." For the goal of increasing elk abundance and distribution, the defined action steps included enhancing the level of knowledge of habitats and habitat utilization in the park, reintroducing extirpated  -6populations to the Park, identifying unused habitat, and designing and implementing a transplant of elk into unused habitat. In the interests of meeting the first and second goals, BC Parks, in cooperation with the BC Wildlife Branch, trapped and radio-collared 10 elk in different areas within and adjacent to the Strathcona Park boundary. Applicable results obtained in the present study form Chapters 1 and 2 of this thesis. My study represents the first phase in a multi-level plan addressing the defined objectives of the Strathcona Park Elk Management Plan (1990). The specific objectives of my study were to collect baseline ecological data on the population and home ranges of Roosevelt elk of Strathcona Provincial Park on Vancouver Island; and ii) to validate seasonal habitat suitability models developed by Brunt (1991), and which may be a useful park management tool for predicting high quality elk habitat throughout the Park. The main purpose of my study was to provide the baseline ecological and practical information that will allow Park managers to better understand the current elk-habitat relationship in Strathcona Park.  -7CHAPTER 1 Population and Home Range Characteristics for Roosevelt Elk in Strathcona Park  INTRODUCTION Population and Home Range Concepts Individual elk do not live naturally in isolation, but rather in a particular social environment, and with other elk, as members of a group. "Group" can refer to a province-wide group, the population of a management unit, the population of a watershed, or a distinct group within a watershed. For the purpose of this thesis, an elk group is defined as > 1 Roosevelt elk which occupy a similar environment, and inhabiting a contiguous region, such as a watershed. As Taber et al. (1982) noted, such a population has 3 general characteristics which may be described mathematically, including those related to production, to mortality, and to movement. Analysis of these characteristics and their interactions can provide estimates of life history attributes which may not be measured readily, such as life expectancy. Gathering and analyzing population data is fundamental to deriving wildlife-habitat management guidelines and strategies. A concept closely related to the characteristics defining an animal group, and equally important to wildlife-habitat management decisions is that of home range. Burt's (1943) classic conceptual definition of home range is "...that area traversed by the individual in its normal activities of food gathering, mating and caring for young." Home range may be defined over daily, seasonal, or lifetime time frames (Harested 1979).  -8The size of an animal's home range is related to its energy requirements, and the productivity of the habitat it is utilizing (Bunnell 1978). Larger animals, for instance, are capable of covering a greater area in the course of a day, as well as requiring more absolute energy than do smaller animals. Other factors affecting or influencing home range size include population density, season of year, interspecific interactions, age, latitude, terrain, and precipitation (Janz et al. 1980b). Habitat quality and terrain are other key factors affecting home range size. Harper (1971) concluded that terrain was the most important factor regulating the home range size of Roosevelt elk in Oregon. Population characteristics have been described for Roosevelt elk on Vancouver Island (Brunt et al. 1989), and for other North American elk sub-species (Taber et al. 1982, Morgantini 1988, Woods 1991). Seasonal and cumulative home ranges have been estimated for several migratory and non-migratory Roosevelt elk groups on Vancouver Island (Janz and Lloyd 1977, Janz et al. 1980b, Brunt et al. 1989, Brunt 1991). Important to understanding the relationship between population characteristics and home range selection in terms of size, stability, and seasonality is the concept of migration. Migration is generally defined as the annual return movement of individuals between two geographically separate areas (Woods 1991). It follows implicitly that migratory animals establish a degree of philopatry to seasonal ranges and make relatively little use of areas between these ranges. While many species of birds and mammals have been classified as being either  -9migratory or non-migratory, intraspecific variation in seasonal movement patterns is common (Baker 1978). Elk are often classified as migratory, despite observations of co-existing non-migratory groups, and groups utilizing several concurrent movement strategies (Woods 1991). Morgantini (1988) suggested that the variation found in elk movement strategies is an example of the species' flexibility in coping with a varied environment. Woods (1991) concluded that the occurrence of different space-use patterns in elk populations was an evolutionary stable strategy (ESS). Measuring Population and Home Range There are 2 levels of population structure analysis, including the structure that can be detected through group composition counts (viz. recognition and enumeration of age and sex classes in a population), and that which is developed through the use of mathematical models of the population (Taber et al. 1982). The first level is employed in the present study, where group composition counts consisted of direct tallies of individual elk in the study area. Twenty years ago Craighead et al. (1973) noted that the techniques of radiotelemetry make possible the delineation of land areas vital to individual animals, to determining seasonal use of those areas as the animal's seasonal needs are met, and to observing behavior and activity that has spatial and temporal significance in the life of the animal. The use of radio-telemetry to study space and habitat use patterns of animals is now considered standard practice in wildlife research and management (Skovlin 1982, White and Garrot 1990, Samuel and Kenow 1992).  -10Radio-triangulation is the most common technique used to determine an animal's location (Springer 1979), and was used in the present study to identify animal locations, which were then imported into a home range estimating program. In a survey of recent studies presenting home range data, Boulanger and White (1990) commented that the application of different home range estimators produces confusion in the interpretation of home range estimates because some of the differences observed between studies are due to the estimators themselves, and not to the behavior of the study animals. All home range estimations in the present study were made using the harmonic mean option in Program Home Range, second edition (Ackerman et al. 1990), a method of home range estimation rapidly gaining popularity in wildlife ecology (Boulanger and White 1990, Harris et al. 1990, Brunt 1991). In this chapter of the thesis, I describe the composition and home range characteristics of 5 distinct Roosevelt elk groups within or adjacent to the boundary of Strathcona Provincial Park on Vancouver Island.  STUDY AREA Study Area Boundaries The study area (Figure 1) includes regions of Strathcona Provincial Park, and areas immediately adjacent to the Park's boundaries. Strathcona Park, British Columbia's oldest provincial park, is centrally located on Vancouver Island, and covers an area of approximately 210,000 ha. Most of the Park is mountainous,  Highway 28  Upper Campbell Lake  HEBER RIVER - CAMEL CREEK VALLEYS  ELK RIVER VALLEY  Study Areas 0  5.0  Buttle Lake  UCONA RIVER BASIN  STRATHCONA PROVINCIAL PARK  THELWOOD VALLEY VANCOUVER ISLAND  Strathcona Provincial Park  • Victoria Figure 1. The 4 study areas in Strathcona Park on Vancouver Island, British Columbia.  10.0 km  -12forming a large part of the Vancouver Island Ranges, which are generally composed of a heterogeneous group of pre-Cretaceous sedimentary and volcanic rocks folded by numerous granite batholiths (Holland 1964). Elevations in the Park range from sea level to 2228 m at the peak of the Golden Hinde, the highest point on Vancouver Island. Heavy snowfalls are common on the mountain slopes and alpine plateaus from November through March. Snow remains all year on the mountain peaks, and may last into July at lower elevations. A detailed biophysical description of the Park can be found in Kojima and Krajina (1975). In Strathcona Park's 82-year history, much resource exploitation has occurred within its confines, including logging and mining. Mining is presently occurring in the south-east region of the Park, near the south end of Buttle Lake. Much of the area surrounding the Park has been, or is currently in the process of being, clearcut logged. At present, 79%, (172,000 ha) of the Park has been zoned either for preservation as wilderness conservation, or for wilderness recreation, thus controlling mechanized access and facility development (Strathcona Provincial . Park Master Plan Draft, 1992). All land within the Park is closed to hunting, while the surrounding areas have a limited-entry hunt for elk. The illegal hunting of elk is known to occur in all areas surrounding and within the Park (BC Parks; pers. comm.). Janz and Lloyd (1977) reported that elk are primarily distributed in the northern and western regions of Strathcona Park and adjacent areas outside the park. The southern and eastern regions of the park support very few elk. Four  -13regions within or near Strathcona Park's boundaries were chosen as study areas by the Ministry of Lands, Environment, and Parks (Figure 1). Each of the 4 regions was deemed important to the study of elk for various reasons, including supporting a past population of elk; proximity to major roadways and heavy public-use zones; easy access for trapping, radio-collaring, and subsequent monitoring of elk; and being politically and/or ecologically sensitive (BC Parks; pers. comm.). The following are brief descriptions of the 4 study areas, including suspected elk population numbers. Elk River Valley The Elk River valley study area lies entirely within the northern region of Strathcona Park. It covers an area of approximately 4450 ha, following along the course of the lower Elk River, from the Drum Lakes to the western shores of Upper Campbell Lake. Two roadways running parallel to one another, transect the study area. Highway 28 travels along the south side of the Elk River, and the ERT gravel logging road travels along the north side of the Elk River. Both roads are open year-round, and are heavily used. Jones (1983) reported that the lower Elk River valley likely provided winter range for approximately 30 elk, 10 of which he believed to summer in the elk winter range area of the valley. He thought that in summer the other elk likely spread out into the upper Elk River valley, Cervus Creek, the upper portions of Idsardi and Tlools Creeks, and possibly other, unknown areas.  -14Janz and Lloyd (1977) observed 2 distinct groups of elk in the Elk River valley; a small group of 7 adult females in the western portion of the valley, and a larger but unknown number of elk in the eastern portion of the valley up to the western shore of Upper Campbell Lake. The elk winter ranges in the lower Elk River valley were believed to be in largely 30-40 year-old second-growth resulting from logging in the Park. The productivity of these habitats is relatively poor when compared to the natural oldgrowth forest (Jones 1983). In addition, the 1957-58 flooding of Upper Campbell Lake permanently destroyed part of the original elk winter range. Thelwood Valley The Thelwood valley study area, at the southern end of Buttle Lake, is completely within Strathcona Park's boundaries, and consists of approximately 655 ha of flood plain, alluvial fan, and valley slopes (Blood 1988). The area once supported a small herd of approximately 15 elk, which were last seen in 1973 (Jones 1983). A series of natural disasters and human-caused impacts likely resulted in the extirpation of elk in the Thelwood by the late 1970's. Hydroelectric development in 1957-58 raised the level of Buttle Lake 9 m, resulting in flooding and permanent loss of flood plain habitat in the lower Thelwood valley. The 1958 Thelwood fire drastically affected habitat on the entire elk winter range. Most herbaceous and browse forage produced in the summer of 1958 was destroyed, resulting in an extremely poor range for elk in the winter of  -151958-59. In addition, during the first several years following the fire, there was no cover in the lower Thelwood valley. The Westmin Mine Co. has been in operation since the late 1950's in the area, and currently occupies much of the Myra Creek valley to the west of the Thelwood valley. Westmin has constructed a hydroelectric powerhouse station in the Thelwood, and a dam at Jim Mitchell Lake, above the valley. Noise and disturbance associated with mine development, operation, and traffic in the Thelwood and nearby Myra Creek valleys may also have had a negative impact in past elk populations. It is believed that easy access resulting from mine activity led to the illegal hunting of any remaining elk in the area (BC Parks; pers. comm.). Heber River-Camel Creek Valleys The Heber River and Camel Creek watersheds are found on the periphery of the western boundary of Strathcona Park. The Heber River valley acts as summer range and the Camel Creek valley as winter range for the population of migratory elk in the area. Both watersheds are criss-crossed by logging roads, most of which are open due to the ongoing active logging in some regions of the valley. The vast majority of the watersheds have already been clearcut logged, as part of the tree farm license managed by Fletcher-Challenge Ltd. Jones (1983) stated that the Heber River area had approximately 100 or more elk, the majority of which wintered and summered outside the Park. Because much of the elk summer range in the upper Heber River valley is adjacent to the Park boundary, some of the elk may summer in tributaries of the Heber River,  -16within the Park. Observations by MOELP personnel indicated that during severe winter seasons, some elk from the Heber River area may winter in the Elk River valley, within the Park (Jones 1983; Janz and Lloyd 1977). Ucona River Basin The Ucona River region forms a large basin on the periphery of the centralwestern Park boundary, at the confluence of several valleys historically important to elk. These valleys include Quatchka Creek, and the Ucona Basin itself, both outside of the Park, and Kunlin Lake, Donner Lake, Pamela Creek, and an unnamed hanging valley inside the Park boundary. All of the winter range for the elk in the Ucona study area, including those animals which summer in the Park, lies outside the Park. The winter ranges have been adversely affected by logging (Jones 1983), and the majority of the Ucona River basin has been clearcut as part of the tree farm license managed by Canadian Pacific Forest Products Ltd. The Ucona River elk winter range is currently in an early vegetation sere created by logging during the mid-1960s (Janz and Lloyd 1977). There are only a few small pockets of mature timber within the area to provide thermal cover during extreme winters. Logging is currently occurring in the Quatchka Creek valley, and in the Pamela Creek valley on the Park boundary. Aside from the Donner Lake region which is old-growth timber and consequently a difficult area to access, the entire Ucona River basin is criss-crossed by an extensive network of well-  -17maintained logging roads. Janz and Lloyd (1977) noted a plentiful supply of winter elk forage in the area. Jones (1983) noted that the Ucona River area supports a group of approximately 100 or more elk, most of which live outside Strathcona Park. Of these, approximately 20 spend at least a portion of the summer within the Park, in the Pamela Creek valley.  METHODS Trapping and Radio-Collaring Animals To determine the seasonal movements and home ranges of elk in Strathcona Provincial Park, 10 adult (> 2 years) female elk were captured and fitted with LMRT-4 radio-collars (Lotek Engineering, Newmarket, Ontario) transmitting in the 150-151 KHz range. In March of 1991, prior to the start of my study, a total of 4 animals were successfully tranquilized from a helicopter a fitted with radio-collars by MOELP personnel; 1 in the Thelwood valley, 1 in the Ucona River area, and 2 in the Heber River valley. In January of 1992, a corral trap was constructed in the Elk River valley, baited with alfalfa hay, and used to capture 13 elk, 4 of which were radio-collared. In the same year, on February 17 and on March 11, 2 more elk in the Ucona River Basin were tranquilized from a helicopter and fitted with radio-collars. An upper canine tooth was extracted from each of the 6 tranquilized animals for later age determination by cementum analysis (Matson's Laboratory, Milltown, Montana).  -18Population and Group Composition Data For group composition counts to be representative of the entire population, all sex and age classes should be equally observable (Taber et al. 1982). This likely is never true, but it is more likely to occur at some periods of the year than others; because of this, group composition counts were conducted throughout the year. Whenever elk (radio-collared or not) were located visually, records were made of the number of elk present in each of the following age-sex classes: adult male ( > 2 years, based in antler development), juvenile mate ( < 2 years, but not young of the year), adult female (not young of the year), young of the year, and unidentified. Ground observations of elk groups were made through 8X30 binoculars, and a 20-40X60 spotting scope. Aerial observations of elk were made without visual aid. In addition to my observations, Park personnel were asked to note elk numbers and age-sex classes observed during the study period, and were interviewed monthly for this data.  Animal Locations Field work during the study consisted of 2 intensive and 2 less-intensive periods. Intensive periods occurred during the summers of 1991 and 1992 (May August). Less-intensive periods occurred during the winter seasons of 1991-92 (September-April), and 1992-93 (September-January). During the second lessintensive study period, animal locations were obtained by BC Parks personnel. Locations of the 10 radio-collared elk were determined by triangulation using a three-element directional Yagi antenna, a Lotek SRX-400 receiver (Lotek  -19Engineering, Newmarket, Ontario), and aviator headphones. The majority of animal locations were obtained from the ground or from the air during daylight hours, over the course of a 2-year period commencing in the spring of 1991, and ending in the winter of 1992-1993. Radio-locations of radio-collared elk obtained from the ground were determined in 1 of 2 ways. First, the area the radio-collared elk was initially believed to be in (from the previous location, or from the "best guess") was scanned with the telemetry equipment on a high gain setting from logging and/or primary roads. If a signal was not detected, adjacent areas were scanned in the same manner until a signal was detected. If still no signal could be heard, the area was noted and left, until an aerial search for the radio-collared animal could be conducted. If a signal could initially be detected, the gain on the receiver was decreased incrementally until the signal was just barely audible in the headphones. A field compass was used to take a bearing in the direction from which the signal was detected, and the bearing line was then drawn on a 1:50,000 topographic field map of the area. At least 2 more bearings were also taken from different positions along the logging and/or primary roads. When the bearing lines did not intersect at a common point, the mid point of the polygon defined by the bearing lines was taken to be the location of the radio-collared animal. The triangulation process was conducted as rapidly as accuracy and travel between successive telemetry "fixes" permitted.  -20For each bearing taken, the observer's exact location on the map had to be first calculated; this involved taking a series of at least three bearings from prominent geophysical features, and/or man-made structures such as road intersections, and then plotting the location on the field map. Because error is associated with each bearing taken, and with plotting each telemetry location on the topographic map, permanent telemetry "stations" were set up in some areas. These stations were of known exact location, from which accurate bearings on radio signals could be quickly made. The second, and more commonly followed, process to determine animal locations from the ground was to complete the above procedure, and then to go in on foot to the location determined by triangulation to attempt to visually observe the animals. This added process allowed me to confirm the animal's location, test the accuracy of the triangulation technique, determine sex and age classification data for each elk group observed, determine the habitat type the animals were found in, and observe elk behaviour. A Bell Jet Ranger Ill or Bell 206 helicopter with a single 3-element Yagi antennae was used to determine animal locations aerially at the start of the field season, and when an animal occasionally could not be located from the ground. On a high gain setting, radio signals from up to approximately 10 km away could be received and used to direct the helicopter to the correct watershed, and the general location of the radio-collared animal. Repeated slow passes were made over the radio-collared animal's general location with the receiver on a low gain  -21setting. When the radio-collared animal and/or any associated elk were visually observed, the location was plotted on 1:50,000 topographic maps of the study areas. When visual contact with the animal could not be made due to dense vegetative cover, the point above where the loudest signal was detected without the antenna, or with a wooden pencil used as an antenna, was taken as the radiocollared animal's location. A Cessna 172 fixed-wing aircraft was also used for aerially determining animal-locations, primarily during the winter months when ground accessibility to some areas was not feasible, and time and manpower were constrained. During fixed-wing flights, 2 three-element directional Yagi antennae were mounted on the aircraft, one on each wing, and each at an angle of approximately 45 degrees from the vertical axis. The procedure to locate radio-collared elk was similar to that used in the helicopter assisted locations. The inherent errors associated with all two-dimensional representations of three dimensional space, coupled with those of the "approximate" nature of a small scale map, means that each time a location was plotted on the 1:50,000 topographic maps, whether from the ground or aerially, a certain amount of error was incurred. The 1:50,000 topographic maps were the most accurate maps available for the entire study area. In addition, because these maps were used at all times, any errors associated with plotting location data were assumed to be uniformly distributed throughout the study. As such their use was felt to be justified.  -221 tested the accuracy of ground telemetry 4 times throughout the study by having someone else place a radio-collar in one of the study areas, and then using the above described triangulation procedures, searched for and located it on a map. The actual location of the radio-collar was generally found to be within 50-100 m of the triangulated location. Without visual confirmation, accuracy of radio-collar location from the helicopter was thought to be at least this accurate. From the fixed-wing aircraft, without visual confirmation, accuracy was found to be less than from the helicopter; the actual location of the radio-collar was generally within 75125 m of the aerially determined location. Due to the relatively large number of radio-locations and direct sightings of the Elk River valley study animals (elk #'s 120, 179, 218, and 259), these observations were recorded by habitat during the 1992 summer season, based on simple features of structure and composition, and by physiographic features of the selected sites (habitat types modified from Janz and Lloyd 1977).  Home Range Delineation Program Home Range (second edition, Ackerman gt al. 1990) was used to generate home ranges for all radio-collared animals. Program Home Range allows estimation of home range size using 5 different methods: bivariate normal ellipse (Jennrich and Turner 1969), weighted bivariate normal ellipse (Samuel and Garton 1985), Fourier transformation (Anderson 1982), convex polygon (Hayne 1949, Michener 1979, Bowen 1982), and harmonic mean (Dixon and Chapman 1980). The program is capable of testing underlying assumptions such as independence of  -23observations (Swihart and Slade 1985). Harmonic mean core areas can also be calculated (Samuel gt al 1985, Samuel and Green 1988). Individual location data points can be weighted to decrease or eliminate the effect of outliers (Samuel and Garton 1985), to reduce serial correlation (Swihart and Slade 1985), or to include proportion of time in different activities (Samuel and Garton 1987). Both the convex polygon and harmonic mean methods were used to delineate radio-collared elk home ranges. The convex polygon method was initially used to determine home ranges for all study animals because of its historical prominence as one of the first developed home range estimators, and one of the most widely used. The convex polygon estimator is a simple and intuitive method for calculating the area enclosed by a set of locations (Beckoff and Mech 1984). In it, the most peripheral locations in a set of location data are connected by a line in such a way that the internal angles of the generated polygon do not exceed 180 degrees. However, the convex polygon method is strongly biased at relatively small sample sizes; is severely affected by outliers, and may include large areas not used by the animal; provides information solely on the area used by the animal during the exact periods of observation and not on the area potentially used by the animal; and provides no information concerning how the area within the home range is used (see Ackerman At  a  1990).  After initial home range estimation, the minimum convex polygon method was not used in this study primarily because I felt that it did not accurately reflect the actual home ranges of the study animals. In particular, the home ranges  -24generated by this method often included large areas of mountain top not used by the animals, especially when an animal made use of 2 or more parallel valleys, or a large bowl and an adjoining small watershed, but not the area between them. In addition, the sample bias of the minimum convex polygon method, where the calculated home range tends to increase as the number of locations increases (Jennrich and Turner 1969), would have made comparisons between animals and seasons invalid. The inaccessibility of certain study areas resulted in large differences in sample sizes of animal locations among both seasons and study animals. In comparing currently popular home range estimation techniques, including the Fourier series, minimum convex polygon, harmonic mean, and 2 - 95% ellipse home range estimators, Boulanger and White (1990:314) concluded that overall, the harmonic mean estimator shows the "best performance." The harmonic mean method of home range estimation is a non-parametric procedure based on the volume under a fitted three-dimensional utilization distribution (Ackerman el al. 1990). The distribution is based on harmonic mean values calculated at grid points systematically located throughout the animal's home range (Dixon and Chapman 1980). The probability of use at any location in the home range is estimated, and the utilization distribution is then determined by calculating the harmonic values at each grid point. When their harmonic values exceed the highest harmonic value for any animal location, grid points are considered outside the animal's home range, and thus excluded from the utilization distribution. Thus harmonic mean estimation  -25is less affected by outliers (Ackerman et al. 1990), and can in fact be used to identify outliers. The utilization distribution is based on a rectangular grid of square cells. This grid is used to calculate the non-parametric harmonic mean distribution from the animal's location data. The size of each grid cell in the rectangle is measured in real world units (e.g. metres), and is dependent on the number of cells and the scale parameter. Each of the animal locations, and all of the generated harmonic contours, must fit on the harmonic grid, or the estimates will be invalid. The choice of the scale and number of grid cells in the rectangle are crucial to ensuring that the size of plot area is large enough to accommodate all of the data. Ackerman  et at. (1990) suggest choosing the scale parameter so that it is  approximately the larger of 1 /8 th of the range of the Y coordinates, or 1 120t h of the range of the X coordinates. However, these suggested levels were found to most often result in a "plot scale too small to determine total utilization volume" error message, and so were only used as a starting point for determining the optimum scale value. The scale value was then adjusted by increments of 1.0 metres until the error message was no longer generated. The grid density parameter should achieve an average of 1 observation per grid cell (Ackerman et al. 1990), and this was accomplished automatically using the "optimization of grid density" option in the program. The harmonic mean method of home range estimation was criticized by Spencer and Barrett (1984) and by Worton (1987, 1989), as being overly sensitive  -26to the choice of scale parameters. I also noticed, as did Brunt (1991), that each animal's home range size and shape could be severely influenced by changing the scale and grid density parameter values in the home range calculations. The harmonic mean estimator method allows for the use of weighting factors to be applied to each animal location, to account for the relative amount of time the animal spent at each location (Braun 1985, Samuel and Garton 1987). Without weighting, the program assumes that each location represents an equal amount of time spent there by the animal, and so tends to overemphasize the relative importance of areas sampled more often. I chose to use a weighting factor when more than one location per animal was obtained within a 24-hour period. In those cases, the weighting factors summed to 1.0, and so equalled the weighting factor given to all independent location observations. Program Home Range requires selection of a minimum distance between observations, which is essentially the typical accuracy of the animal locations (Ackerman et al. 1990). I set the minimum distance between observations at 50 m for all home range calculations, because this was considered to be the minimum error value of triangulated animal locations. Dates for delineating seasons were based on study animal movements between seasonal ranges. Not all study animals exhibited obvious migratory behavior or seasonal shifts in areas of use. For those elk demonstrating migratory behavior, the timing of migratory movements differed between years, and between  -27study animals, so seasons delineated for the purpose of home range calculations are presented in Table 1. Animal/Season Selection Criteria Over the course of the 2-year field study, some radio-collared animals associated with others for varying time periods. For the purpose of home range estimation, elk were considered to be associated with the same group whenever they were located within approximately 200 m of each other (Brunt 1991). When 2 or more elk were considered to be associated together during part or all of a particular season, a degree of association was assessed to determine which animals had sufficiently different seasonal range selection patterns to warrant calculation of individual animal/season home ranges. If the proportion of a given season spent with other study animals was > 90%, one member of the associated animals was randomly selected for seasonal home range calculations (Brunt 1991). By contrast, individual home ranges were calculated only for animals whose association with other study animals was < 90%, and for groups consisting of 2 or more study animals whose association with each other was > 90%. Table 2 outlines the 13 animal/seasons for which home ranges were calculated. The 4 radio-collared adult females in the Elk River valley (#'s 120, 179 , 218, and 259) were found to be closely associated with each other > 90% of the study period commencing with their collective capture in January of 1992. For brief periods during the calving season in the spring of 1992, 1 or more of the 4  -28-  Table 1. Dates used to delineate seasonal home ranges for each study animal.  ERV Groupe  Cumulativeb  Jan 24/92 - Jan 6/93  159  Cumulativeb  Mar 11/92 - Jan 6/93  300  Cumulativeb  Mar 3/92 - Jan 6/93  339  Cumulativeb  Jun 15/91 - Jan 6/93  Cumulative  Jun 14/91 - Dec 10/92  Summer 1991  Jun 14/91 - Nov 9/91  Winter 1991-92  Nov 10/91 - Jun 4/92  Summer 1992  Jun 5/92 - Sep 9/92  Winter 1992-93  Sep 10/92 - Dec 10/92  Cumulativeb  Jun 13/91 - Jan 6/93  Cumulative  Jun 17/91 - Jan 6/93  Summer 1991  Jun 17/91 - Dec 21/91  Winter 1991-92  Dec 22/91 - May 29/92  Summer 1992  May 30/92 - Nov 7/92  Winter 1992-93  Nov 8/92 - Jan 6/93  499  580  688  "ERV = Elk River valley, and includes elk #'s 120, 179, 218, and 259 (Note: #218 mortality on December 20, 1992). b Because the elk did not demonstrate migratory behavior, seasonal ranges could not be delineated.  Table 2. Home ranges analyzed using the harmonic mean contour method. Home Range Estimated  Study Animals  Total Ranges Analyzed  Cumulative'  ERV Group b , 159, 300, 339, 499, 580, and 688  7  Summer 1991  499, 688  2  Winter 1991-92  499, 688  2  Summer 1992  499, 688  2  aYear-round home range using all animal locations b ERV= Elk River valley, and includes elk #'s 120, 179, 218, and 259  -30study elk were located in areas distinctly separate from the main group. Elk #259 left the main group for a short period in the fall of 1992. The 2 adult females radio-collared in the Heber River watershed were associated with each other for portions of both winter and summer seasons, but not for periods > 90% of any 1 season. Therefore, the 2 Heber study animals were treated as non-associated elk, and separate home ranges were calculated for each. Two of the 3 study animals in the Ucona River basin (#159 and #339) were closely associated with the same group during the summer of 1992. However, whereas elk #339 was radio-collared in the spring of 1991, elk #159 was radiocollared in the spring of 1992, so it is not known whether the 2 were associated prior to the summer of 1992. No other study animals (elk #580 in the Thelwood valley, and elk #300 in the Ucona River basin) was observed In close association" with other radio-collared animals. Home Range Analysis Three types of home range were calculated and compared for most animal/seasons: i) cumulative home ranges were determined from all animal locations obtained during the study period; ii) seasonal home ranges were estimated from all non-outlier locations obtained during the seasons listed in Table 1; and iii) core use areas were delineated as areas of concentrated use (see below) within each seasonal range. Cumulative home ranges were delineated for each study animal as the area included by 100% of the locations obtained during the study. The cumulative  -31range provides a means of defining all areas available to the animal for seasonal range selection (Brunt 1991). However, animals may occasionally and temporarily leave their normal activity areas, producing extreme locations (outliers) that can seriously affect home range estimates. Outliers may be the result of mapping and coding errors in the location data (Ackerman et al. 1990), and can also represent transitional locations between seasonal ranges. In accordance with Burt's (1943) classic definition of home range, these occasional excursions outside the normal area of use were not considered part of the home range. Program Home Range allows for the calculation of contours which contain a specified percentage of the animal's utilization distribution, and thus identifies outliers. To calculate seasonal home ranges, I used the 95% contour to reduce the effects of potential outlier values, in accordance with Burt's (1943) home range definition and the convention followed by others (Jennrich and Turner 1969, Schoener 1981). Core Use Areas The concept of core use areas has received considerable use in the ecological literature, and has generally been used to define central areas of intense use, in an effort to describe the internal anatomy of the home range (Ackerman et al. 1990). Core use areas are important because they are areas of particularly high home range usage, and thus they often may provide a more clear measure of the changing pattern of range use than does the total home range area (Harris et al. 1990). Core use areas are often more useful for understanding both intraspecific  -32and interspecific patterns of home range use, than are the more peripheral contours (e.g. 100% minimum convex polygon contour, 99% harmonic mean contour, etc.). Harris et al. (1990) cited an example of this phenomenon in a study comparing roe deer (Capreolus capreolus) and muntjac (Muntiacus reevesi) in the same habitat. -  The number and relative position of core use areas differs between the species, but such differences are masked when comparing total home ranges. In addition, while male and female core use areas do overlap, they do not occupy the same space. In the case of male muntjac ranges, which may overlap to some extent, their core use areas are mutually exclusive. An option in Program Home Range allows core use areas to be automatically identified by comparing the utilization distribution from harmonic mean calculations with a uniform use model (Samuel  ex t. 1985). Core areas are defined as the  maximum area where the observed utilization distribution (based on harmonic values) exceeds a uniform distribution (Ackerman et al. 1990). Program Home Range then uses a chi-square test on the ordered cumulative distribution of observed data, compared to the uniform model (Samuel and Green 1988). An illustration of the statistical test and further description of the above described method are presented in Samuel et al. (1985) and Samuel and Green (1988). The average utilization distribution contour of the 10 home ranges calculated using this method was 60.0% (SD= 5.8%). Dixon and Chapman (1980) also described the core use area as the 50% harmonic mean contour. If Program Home Range's automatic core use area  -33estimator was unable to detect a significant core use area present in a study animal's home range, the 50% harmonic mean contour was used instead. The harmonic mean contour actually plotted was the closest one to 50% that could be calculated given the study animal's location data. For the 3 core use areas for which Program Home Range's automatic method could not determine the presence of a significant core use area, the average of the contours plotted was 69.7% (SD =10.209%). This was not significantly different from the average utilization distribution of 60.0% for core use areas determined by the automatic method (t = 2.098, tcritical = 2.201, df =11, a = 0.05). Core use areas are not considered reliable when S 12 animal locations are used to delineate a home range (Ackerman et al. 1990). Therefore, core use areas determined for the following 4 seasonal ranges are not considered reliable: summer 1992 and winter 1992-93 for elk #'s 499 and 688. The 13 core areas that were estimated and plotted are presented in Table 3, along with the method used to calculate them (Program Home Range's automatic core use area calculation, or the 50% harmonic mean contour). Statistical Analyses A probability level of 0.05 (95% confidence level) was used in all statistical analyses. Walpole (1982), Zar (1984) and A. Kozak (pers. comm.) were consulted to determine appropriate statistical procedures. Statistical tests were conducted on a hand calculator.  -34-  Table 3. Core use area delineation methods and core area and home range sizes. Home Range Estimated  Study Animal  Core Area Calculation Method  Proportion Utilization Distribution  Seasonal Range Size (km2)  Core Use Area Size (km2)  Summer 1991  499  50% Contour  0.64  31.82  6.53  688  Automatic  0.68  36.97  13.12  Winter 1991-92  499  Automatic  0.57  61.16  17.56  688  50% Contour  0.61  18.59  1.37  Summer 1992  499  50% Contour  0.84  7.24  0.43  688  Automatic  0.64  63.25  11.86  ERVa  Automatic  0.65  32.38  10.14  159  Automatic  0.67  41.53  13.71  300  Automatic  0.54  28.15  6.21  339  Automatic  0.63  26.14  10.90  499  Automatic  0.57  95.98  32.28  580  Automatic  0.51  15.93  3.51  688  Automatic  0.54  155.67  29.11  Cumulative  aERV= Elk River Valley Group, and includes elk #'s 120, 179, 218, and 259.  -35RESULTS Mortalities One of the 2 adult female elk first radio-collared in the Ucona River Basin died within a month after trapping and radio-collaring activities. Cementum annuli analysis indicated that elk #199 was 4 years old (accuracy + I- 1 year). It was clear from the mortality site that elk #199 had been preyed upon by an undetermined number of wolves, whose tracks and bite marks on the radio-collar were obvious. A second study animal mortality occurred on December 20, 1992, when one of the Elk River valley study animals, elk #218, was shot by a limited entry hunt ticket holder within 50 m of the Strathcona Park boundary (BC Parks, pers. comm.). Animal Locations A total of 535 locations was obtained during 5 seasons from June 13, 1991 through January 6, 1993. Ground radio-telemetry was used to obtain 279 (52.15%) of the locations, and aerial radio-telemetry was used to obtain the remainder (256 or 47.85%). Group Composition and Population Group composition counts were conducted throughout the year in the Thelwood valley, Heber River-Camel Creek, and Ucona River study areas; the ratios and percentages of group sex-age classes were determined for both summer (1991-92 combined) and winter (1991-92 combined) seasons (Table 4). Group composition counts for the Elk River valley were also conducted throughout the  -36-  Table 4. Classification counts for summer and winter seasons in study area.  I  Elk River Valley - Summer° 1992 Juvenile Male  Adult Male  Adult Female  YOYb  Total  #  2  3  9  4  18  Ratio  22.2  33.3  100  44.4  %  11.1  16.7  50  22.2  100 (  Elk River Valley - Winter' 1992 Juvenile Male  Adult Male  Adult Female  YOY  Total  #  1  2  9  1  13  Ratio  11.1  22.2  100  11.1  %  7.7  15.4  69.2  7.7  100  The!wood Valley - Summers 1991-92 Juvenile Male  Adult Male  Adult Female  YOY  Total  #  0  2  1  0  3  Ratio  0  200  100  0  %  0  66.7  33.3  0  100^I  Thelwood Valley - Winters 1991-92 Juvenile Male  Adult Male  Adult Female  YOY  Total  #  0  2  1  1  4  Ratio  0  200  100  100  %  0  50.0  25.0  25.0  aSummer season from May through October Y0Y = young of the year (calves) °Winter season from November through April b  100  -37-  Table 4 (continued). Classification counts for summer and winter seasons. Heber River Watershed - Summers' 1991-92 Juvenile Male  Adult Male  Adult Female  YOYb  Total  #  14  27  216  51  308  Ratio  6.5  12.5  100  23.6  4.5  8.8  70.1  16.6  100  Heber River Watershed - Winters` 1991-92 Juvenile Male  Adult Male  Adult Female  YOY  Total  #  0  0  49  7  56  Ratio  0  0  100  14.3  %  0  0  87.5  12.5  100  Ucona River Basin - Summers 1991-92 Juvenile Male  Adult Male  Adult Female  YOY  Total  #  1  19  40  8  68  Ratio  2.5  47.5  100  20  %  1.5  27.9  58.8  11.8  100  Ucona River Basin - Winters 1991-92 Juvenile Male  Adult Male  Adult Female  YOY  Total  #  0  2  9  1  12  Ratio  0  22.2  100  11.1  %  0  16.7  75.0  8.3  aSummer season from May through October b Y0Y =young of the year (calves) `Winter season from November through April  100  -38year (1992); however, I believe the total sex-age class numbers presented are accurate, and not just indices of group composition, as in the other study areas. Home Range Characteristics and Movements Home ranges and core use areas were estimated for all 10 radio-collared study animals (Figures 2-6). Location data indicated that the 2 Heber River elk (#499 and #688) were migratory, and the remaining 8 radio-collared elk were nonmigratory. Seasonal home ranges and core use area sizes were determined for the 2 migratory elk for summer 1991, winter 1991-92, and summer 1992 (Table 5). Although data continue to be collected at the time of this writing, seasonal range and core use area size could not be estimated for the winter 1992-93 season, due to insufficient locations having been obtained by the end of my study period. Cumulative home range and core use area sizes using all location data obtained over the study period were calculated for all study animals and are presented in Table 6. Elk River Valley The core group of 16 elk in the Elk River study area consisted of 9 adult females, 4 of which were radio-collared, 4 calves, 3 adult males, and a juvenile male (Table 4). This group was directly sighted over the 1992 summer period on average once every 4 days, and numerous times throughout the winter season; consequently the group size and composition is believed to be accurate. The largest number of elk in the Elk River valley during the summer season was found to be 18 animals. From the distribution of sightings in the valley, it is  N CUMULATIVE RANGE glIM OM OM 4PB  0  Highway 28 ERT Logging Road 1.5^3.0 km  -T1 EC  -  CD  co  Drum Lakes  0  ...,  CORE USE AREA  CD CD X  Figure 2. Cumulative range and core use area for elk #'s 120, 179, 218, and 259, in the Elk River valley.  Figure 3. Cumulative range and core use areas for elk #580 in the Thelwood valley.  WINTER RANGE  SUMMER RANGE WINTER CORE USE AREA  SUMMER CORE USE AREA  STRATHCONA PARK BOUNDARY  est  - N^ — — -- Logging Road  Crest Lake  ek ■  •^  SMA N  s.  ^-  •  -  CQ A PARK  ' -- BOUND ARY  Highway 28 2  4km  Figure 4. Summer and winter (1991-92) seasonal ranges and core use areas for elk #499 in the Heber River-Camel Creek Valleys.  Figure 5. Summer and winter (1991-92) seasonal ranges and core use areas for elk #688 in the Heber River-Camel Creek Valleys.  Figure 6. Cumulative range and core use areas of elk #'s 159, 300, and 339 in the Ucona River Basin.  Table 5. Sizes (km 2 ) of seasonal home rang& and core use areas for 2 nonmigratory elk from the Heber River-Camel Creek valleys. Season  Dates  Study Animal  n Locations  Seasonal Range Size  Core Use Area Size  Summer 1991  Jun 14-Nov 9  499  5\11b  31.82  6.53  Jun 17-Dec 21  688  9\6  36.97  13.12  Nov 10-Jun 4  499  13\4  61.16  17.56  Dec 22-May 29  688  9\4  18.59  1.37  Jun 5-Sep 9  499  5\7  7.24  0.43  May 30-Nov 7  688  7\4  63.25  11.86  Winter 1991-92 Summer 1992  aHome range estimated using 95% of utilization distribution b Aerial locations/ground locations  -45-  Table 6. Sizes (km 2 ) of cumulative (year-round home range using 100% of  utilization distribution) and core use areas for all study animals. Study Animal  n Locations  Cumulative Range Size  Core Use Area Size  120  24/478  32.38  10.14  179  24/47  32.96  9.77  218  22/47  29.70  9.54  259  23/47  32.26  9.58  159  21/6  26.14  10.90  300  21/7  28.15  6.21  339  32/14  41.53  13.71  Thelwood  580  31/28  15.93  3.51  Heber RiverCamel Creek  499  29/22  95.98  32.28  688  28/14  155.67  29.11  Study Area  Elk River  Ucona River Basin  aAerial locations/ground locations  -46possible that several more elk, likely in small "bachelor" groups of 2-3 adult males make use of the valley in the early spring and late fall. If this is indeed the case, the maximum population of elk in the Elk River valley approaches 20 individuals. In the winter season, the maximum number of elk observed was 13 animals. The Elk River valley group was found to use a total of 7 habitat types during the summer period of May 21 - August 30, 1992 (Table 7). Riparian willow (Salix spp.) flats, wet meadows or bogs, the grassy right-of-way below the power transmission fines, and second growth coniferous forest (predominantly Douglas fir Pseudostugas menziesii) and western hemlock (Tsuga heterophylla), with some lodgepole pine (Pinus contorta latifolia)) were the habitat types most frequently used by this group. While the Elk River valley group made repeated use of their entire home range throughout the summer, I observed that as the water levels in the Elk River dramatically decreased over the course of the summer season, the elk made greater use of wet meadows or bogs and the river estuary, and less use of riparian willow flats. Thelwood Valley The Thelwood valley presently supports a known population of 4 elk, consisting of 1 adult female, 1 calf of undetermined sex, and 2 adult male elk. The calf was born in the spring of 1992. The majority of locations in the Thelwood valley of elk #580 were nearly evenly spread between 2 areas; low elevation (223 m), 30-year old coniferous  Table 7. Habitat types' used by the Elk River valley group b during summer 1992 (May 21-August 30). Habitat Type  # Observations`  %Use  Second growth conifer forest  4  8  Mature conifer forest  1  2  Power transmission lines  12  24  River estuary - lakeside  2  4  Bog - wetland  8  16  Riparian willow flats  5  10  Riparian second-growth deciduous forest  18  36  Totals  50  100  *modified from Janz and Lloyd (1977) b Elk #'s 120, 179, 218, and 259 `Direct sightings or locations later confirmed by presence of elk sign  -48second growth and adjacent regenerating riparian alder (Alnus rubra) flats, and higher elevation (450-700 m) riparian areas and vegetated slides. Young alder bushes, salmonberry (Rubus spectabilis), thimbleberry (Rubus parviflorus), and bracken fern (Pteridium aquilinum oubescens) were the among most common understory vegetation associated with animal locations. Heber River-Camel Creek Valleys The Heber River watershed likely supports a population of approximately 100 elk during the summer season. These 100 elk are found in several groups of 10-30 individuals which repeatedly congregate over the summer in groups of up to 70 elk. In the winter season, the lower Heber River valley supports few elk, probably no more than 30 individuals; the upper valley, with a higher elevation and almost no snow-interception or thermal cover, supports even fewer. It is not known how many elk use the Camel Creek valley during the summer season. During the winter, I believe that between 20 and 40 use the area. In the summer season, elk in the Heber River watershed (#499 and #688) made use of the 20-30 year-old dense Douglas fir and western hemlock regeneration along the valley bottom, and the clearcut sideslopes above. Vaccinium spp. dominated the understory layer. Summer range elevation varied from 750-1400 m. Elk #688 left the Heber River watershed and spent approximately half of the summer seasons outside of the study area. In the winter season, both elk #499 and #688 were associated together with a small group which migrated to the Camel Creek valley, a distance of  -49approximately 43 km (measured from the northernmost point in elk #499's summer range to the southernmost point in #499's winter range). Winter range elevation varied from 350 m in the lowermost Heber River watershed to 750 m in the Camel Creek drainage system. Most of the Camel Creek valley has been clearcut logged, the lowermost portions of which consist of Douglas fir regeneration forest. Ucona River Basin The Ucona River Basin elk population likely numbers approximately 100 individuals, which spend both winters and summers in the basin and its associated tributaries. During the winter season this number may increase to approximately 125-140 elk, which migrate to their winter range in the Ucona River Basin from surrounding watersheds. The Ucona River Basin study animals made similar use of habitat over the course of the study period, although in different areas of the river basin. In the first year of the study, elk #339 made extensive use of the moist riparian areas in the Quatchka Creek drainage, and the thick Douglas fir and western hemlock regeneration forests on the sideslope. Almost all locations were found in areas of very thick cover, where the predominant understory vegetation was thimbleberry, salmonberry, and sword fern (Polystichum munitum). In the second year of the study, elk #339 was closely associated with elk #300 during the winter, and elk #159 in the summer. In the summer season, elk it's 339 and 159 were found in a clearcut "hanging valley" recently added to Strathcona Park (1987), grazing on the  -50sideslopes. Elk #300 made use of moist riparian habitat in the Pamela Creek drainage, and of the vegetated slides adjacent to the riparian areas.  DISCUSSION Mortalities Wolves are the most important natural predators of elk on Vancouver Island (Janz and Becker 1986), and are known to live in the Ucona River Basin (BC Parks; pers. comm). The death of elk #199 was likely a natural occurrence of predation, and unrelated to earlier trapping and radio-collaring activities. Janz at al. (1980b) noted that predation, especially by wolves, on ungulate populations has in some cases been shown to be a major limiting factor on population growth, although relatively little is known of their impact on elk populations, particularly on Roosevelt elk. However, in a study of wolf food habits on Vancouver Island, Scott (1979) found that Roosevelt elk made up 28% by relative weight of wolf scat content during her May - October study period. In Banff National Park, in addition to collisions with trains and automobiles, predation by timber wolves is the major source of mortality for Rocky Mountain elk (Woods 1991). The Elk River valley group, including elk #218 which was legally hunted, was located within the Park on December 10, 1992, and later on December 23 approximately 2 km east of the Park boundary, where they were using the rocky outcrop habitat on the south-facing slopes above Upper Campbell Lake. I believe that the elk had moved out of the Park because of deep snowfall accumulation  -51which had occurred in the latter half of December. Although the Elk River group had not previously been observed outside the valley, Janz and Lloyd (1977) indicated that this area outside the Park was part of the group's winter range. Home Range Characteristics and Movements Brunt et al. (1989) found that on Vancouver Island the home range sizes of migratory elk were approximately 6 times larger than those of non-migratory elk. This difference is consistent with my findings, in which the cumulative home range sizes of migratory elk were 2.3 - 9.8 (mean = 6.1) times larger than those of nonmigratory elk. Brunt gt al. (1989) estimated the winter home range size of migratory Roosevelt elk to vary from 2.2 - 43.8 km 2 , and the summer home range size to range from 3.5 - 29.2 km 2 . The present study estimated the winter home range size to vary from 18.6 - 61.2 km 2 , and the summer home range size to vary from 7.2 - 63.3 km 2 . Graf (1955) reported that non-migratory Roosevelt elk in the Coast Range of Oregon occupied home ranges of approximately 2.6-5.2 km 2 , and in California, Franklin Di al. (1975) estimated the home ranges of non-migratory Roosevelt elk to be 2.9 km 2 . These values are smaller than those I estimated for all study animals in the present study except for the Thelwood elk. However, operational differences concerning the nature of "home range" described in each study precludes realistic comparisons among home range sizes listed in these 2 published studies; neither study reported how the home ranges were calculated, nor what kind of home range was estimated. And as Arcese (1989) observed,  -52home range sizes may vary with population size, and therefore do not provide a universal scale for inter-specific comparisons. Elk River Valley The elk population estimate of the Elk River valley in the present study is reasonably close to Jones' (1983) estimate, which was "...winter range for about 30 elk... probably about 10 of these summer in the winter range area." The discrepancy may be a result of Jones' (1983) underestimation of the elk carrying capacity of the area. Given the amount of highly suitable elk habitat in the area (see Chapter 2), I believe that the Elk River valley could sustain at least 50% more elk that it presently does. Janz and Lloyd (1977) determined that movements of the elk in the lower Elk River valley were confined to the valley bottom and low elevation, windward sidehills. This is consistent with the findings of my study in which all 4 radiocollared elk showed a high degree of philopatry to the lower Elk River valley over the course of the study period. The tenacity with which an animal returns to specific areas or groups of like animals is generally referred to as philopatry (Greenwood 1980). Philopatry, and its opposite - dispersal, are relative rather than absolute terms. Several studies have shown adult elk to have high degrees of philopatry (Knight 1970, Adams 1982, Morgantini 1988, Woods 1991), however, as Woods (1991) noted, theoretical criteria for defining philopatry (or dispersal) are lacking.  -53Population trend is one of the major indicators of elk response to habitat changes (Janz et al. 1980a), and as such is important to understanding the present wildlife-habitat situation, and for forming the context in which future management decisions can be made. During Janz and Lloyd's (1977) 1976 winter study, the "A-frame" group and the "Campbell River" group (comparable to the "Elk River valley" in the present study) were the only elk found to inhabit the Elk River valley study area. The "A-frame" group consisted of 7 adult females, and confined its activity to an area composed of moss and lichen-covered rock bluffs and the adjacent second growth (predominantly hemlock and lodgepole pine) in the western-most region of the Elk River valley outside of the "Campbell River" group's home range. This area was repeatedly searched for signs of elk presence in the summer, and although some fresh and old sign of deer, and old bear scat was found, no sign or observations of elk were made either in the summer or winter. I believe that this area is not currently used by elk, and may only be used in severe winters where the requirements of a small group of elk can be met in a comparatively small area with high structural and compositional diversity of vegetation. The'wood Valley The very small number of elk (4) in the Thelwood valley is of concern given the apparent ease with which the previous population of at least 15 animals were extirpated (BC Parks; pers. comm.). Paternity of the calf is presently unknown, as is the possible maternal or sibling relationship between the adult female and each  -54adult male. It is unlikely, however, that the female is the mother of either male because she is 4 years old ( + /- 1 year accuracy) based on cementum annuli analysis, and the males are probably at least as old, based on the size and development of their antlers. The ultimate survival of the current population has been questioned (BC Parks; pers. comm.), and an immediate transplant of nonmigratory elk into the valley recommended (Blood 1988). I believe that such a transplant is warranted. The 3 distinct core use areas in the Thelwood valley used by the elk were not seasonally divided; movement between each area occurred throughout the year. Some of the movement between the areas may be related to the noise and other disturbances associated with the Westmin Mine Co.'s test-drilling operations in the 1991-92 and 1992-93 winters. It was observed that when the drilling process began, in which heavy machinery and equipment generating loud noises was moved into the lower area used by the elk, the animals began using the higher area. However, animal locations during the winter seasons were not obtained frequently enough to confirm the hypothesis that the elk moved in reaction to disturbance by humans. Heber River-Camel Creek Valleys During 2 years of study, primarily in the summer season, elk #688 spent approximately 1/2 of the season outside of the study area. It is believed that she made use of the ridge tops north and west of the Heber River valley, on the side slopes and moist valley bottoms approximately 8 km to the north, near Gold Lake,  -55and areas >10 km north of Gold Lake (BC Parks; pers. comm.). Due to the inaccessibility of the regions this animal used, and because she could not be aerially located for much of the time, location data for #688 are limited and severely underestimate the animal's cumulative and summer season home range, and her core use areas, and seasonal movements. This must be considered when any comparisons are made between the home ranges occupied by the 2 migratory study animals. Ucona River Basin While elk #339 repeatedly used different areas of the Ucona River Basin regardless of season, I believe that it is likely the animal shifted its primary use of the Quatchka Creek valley to the hanging valley as a result of clearcut logging of the Quatchka area. However, the relatively few locations obtained between the first and second summer seasons when the shift occurred cannot substantiate this hypothesis. Additional years of animal location monitoring will determine the degree of association between the 3 study animals, and may determine the number of distinct elk groups found within the Ucona River Basin study area. Franklin et (1975) observed that precise boundaries between distinct Roosevelt elk groups did not exist, and concluded that the absence of outside animals in core use areas and their infrequent use of other parts of the home range showed that spacing between populations does occur. This confirms Graf's (1955) observation that closely  -56adjoining home ranges seldom overlap, and that there is little trespassing by one group on the range of another. Dasmann and Taber (1956) noted that in areas of dense cover, cervids often form small groups or are solitary, and form larger groups when inhabiting open areas. Franklin  at al. (1975) concluded that Roosevelt elk population size appears  to be affected by habitat in a similar way. Observations made in this study are consistent with these findings. During the summer season in open, cut over areas such as those found on the clearcut logged slopes in the Heber River and Ucona River watersheds, elk were more often found in large groups. In areas of more dense cover within the above watersheds, smaller elk groups and sign of smaller elk numbers were observed. Migration and Non-migration as Movement Strategies  Cervus elaphus have the widest natural distribution of any wild ungulate in the world (Clutton-Brock et  al. 1982), and have a range of movement behaviors  including migration, non-migration, and partial migration (Adams 1982). Sinclair (1983) defined migration in vertebrates as the repetitive seasonal movements of individuals between distinct areas. Non-migration refers to the movement behavior of individuals which use the same area throughout the seasons. Seasonal movements of elk are population-specific, thus it is important to recognize inter-population differences. Significant variation was seen in the movement strategies of the elk in the present study. In the case of the 2 Heber River-Camel Creek study animals, each elk spent a portion of both summer and  -57winter seasons together on the same ranges, but migrated at different times and via potentially different routes. In the Elk River valley, all 4 study animals were observed to be closely associated with the rest of their group throughout the year, save for very brief periods during the calving and rutting seasons. The Thelwood valley study animal made use of 3 separate areas in its home range, but not seasonally, and the 3 Ucona River Basin elk shifted their core use areas somewhat over the course of a year, but not distinctly in that all regions of the study area still received use. The literature is full of similar variations in migratory, non-migratory, and a mix of the 2 types of movement strategies. While Rocky Mountain elk at Jackson Hole, Wyoming, migrated in large groups (Boyce 1991), elk in a Montana population moved individually or in matriarchal groups (Knight 1970), and elk in Banff National Park, Alberta, migrated singly or in small groups (Woods 1991). Although most migratory elk winter at low elevations, and summer at high elevations, individual behaviors such as multiple migration cycles per year and specific migrations to rutting areas illustrate that migration patterns vary widely among individuals (Woods 1991). Harper et al. (1967) suggested that historically, 2 kinds of Roosevelt elk were found in California, migratory and non-migratory. Most groups of Roosevelt elk in Washington, Oregon, and California did not demonstrate definite migratory movements, although they do make seasonal movements in response to forage conditions (Graf 1943, Franklin et al 1975). Adams (1982) noted that such movements resemble migration in that they generally are to higher elevations in  -58summer and to valleys and lower elevations in winter. These movements differ from migration in that they are not periodical. If the right combination of environmental variables are found in an area, the elk will remain there year-round; for example, if forage becomes less available, they may move to or in search of a more suitable supply area. However, some groups of elk, such as those of the Olympic Peninsula, Washington (Schwartz and Mitchell 1945), and on Afognak Island, Alaska (Troyer 1960), are migratory in response to severe winter weather conditions. Madison (1966) observed that the Tule elk (C. e. nannodes) in California is non-migratory, and that there is no evidence that it ever was. The Tule elk utilize various portions of their range in response to seasonal variations in food availability. McCullough (1969:47) stated that: ...such seasonal movements are not regarded as migrations for the following reasons: i) summering areas are not inaccessible because of weather during the winter period; ii) movements are not consistent among herds; and iii) the timing of movements differs from herd to herd. McCullough (1969) considered these movements to be solely local shifts in response to local conditions. Considering the wide range of movement strategies utilized by elk throughout western North America, why then are some of the elk in the present study utilizing a non-migratory strategy, while others a migratory strategy? Migration should be favored when the benefits of moving outweigh the benefits of not moving (Baker 1978). Woods (1991) listed the following potential benefits of  -59seasonal migration: i) access to better quality and quantity of food; ii) reduced competition for food; iii) escape from predation and other forms of harassment; iv) access to increased mating opportunities; and v) avoidance of extreme seasonal weather. Potential costs to migration include: i) increased energy requirements for travel; ii) time lost from other activities; and iii) increased risk of mortality due to travel hazards and predation. Within a species, or within a population, all individuals may not face the same costs and benefits of migration. Woods (1991) cited the example that the cost/ benefit ratio evaluation may vary with the size of the individual, and that individuals may vary in their genetic tendencies to migrate. By far the most favored proposal for the benefit of altitudinal migration in North American ungulates living in mountainous areas is that of a nutritional advantage over non-migration (Hebert 1973, Shackleton 1973, Morgantini and Hudson 1983). Because forage quality generally peaks in young, rapidly growing plants and then declines as the plants mature (Nelson and Leege 1982), animals could benefit from tracking early growth stages as spring advances from lower to higher elevations. If this is the case, then at any time during the growing season, altitudinal migratory animals should have access to more nutritious forage than if they had remained at low elevations. However, Woods (1991) concluded that empirical data for various North American ungulates regarding forage quality, animal condition, and animal movements, do not consistently support the hypothesis that altitudinal migratory animals have access to better quality food. Using a model, Boyce (1991) showed that seasonal fluctuations in food availability  -60on high, summer Rocky Mountain elk ranges could determine the fitness of migratory and non-migratory elk. In areas with low seasonality, non-migratory individuals had higher fitness over migratory animals because they had no migration costs. At moderate levels of seasonality, fitness of both migratory and nonmigratory animals decreased, although the fitness of non-migratory individuals decreased more sharply. Migrants do not universally gain access to better quality food, even by moving to higher elevational ranges. When one considers the large geographical range of Cervus elaphus (CluttonBrock in al 1982) and the diverse habitats occupied by the species (Geist 1982), it is likely that the "nutritional payoff" for migration will vary widely among populations. In areas like the upper Heber River valley, where summer conditions on low elevation winter ranges are relatively hot and dry, large differences in forage quality may exist between low and high elevation sites. In more mesic and cooler areas like the Elk River valley, the nutritional benefits of migration could be reduced or non-existent. Murie (1951), Baker (1978), and McCullough (1985) have presented an alternative explanation for choice of movement, based on learning. The suggestion is that young animals develop patterns of home range use through individual experience as well as association with others (particularly maternal association). Woods (1991) observed that in Banff National Park, where a mother-young pair was continuously tracked, the daughter adopted the movement strategy of the mother. However, although Clutton-Brock g al. (1982) demonstrated a close  -61association between mother-young pairs in red deer during the first years of the daughter's life, Woods (1991) notes that this trait has not been well documented in elk, or in other ungulates. Life history traits are often phenotypically plastic; that is, a single genotype produces a range of phenotypes depending on the environment (Lessells 1991). Morgantini (1988) commented that flexibility in movement strategies and patterns, allows elk to adjust to a variable environment, and Boyce (1991) noted that such phenotypic plasticity could be shaped by natural selection. Bergerud (1974) suggested that such adaptability explained variation in caribou (Rangifer tarandus) movements, and was a major adaptation in that species. Woods' (1991) observations of the movement patterns and variations in a Rocky Mountain elk population supports these views. Woods (1991) noted that such flexibility allows elk to make a conditional assessment of density and environmental variables, and to either migrate or stay accordingly. As Morgantini (1988) and Woods (1991) concluded, migration may not be a species characteristic of elk, as has been implied by many authors, but rather an adaptive and versatile response to a specific environmental situation. This is consistent with Geist's (1982) "opportunism" theory of elk behavior, and Woods' (1991) observation that no one variable (e.g. diet quality, previous experience, harassment by predators) is likely to explain the range of foraging and movement behaviors seen in the species.  -62Home Range Estimation The basic purpose of a home range estimator is to provide a quantifiable area describing the area traversed by an individual animal in the course of its normal activities over a given time period. Because each home range estimator defines this quantity differently, home range estimates should be viewed only as a general measure of animal activity areas, and comparisons between studies of home ranges should be taken only as general guidelines given the limitations of estimators (Boulanger and White 1990). Even within the harmonic method of home range estimation in a program such as Program Home Range, an animal's home range size and shape can be severely influenced by changing the scale and grid density parameter values in the home range calculations. Ackerman et al. (1990) suggested that in order for harmonic mean home range estimates to be most comparable between animals, the scale and grid cell parameters should be the same. Their suggestions for accomplishing this were followed in the present study, where each animal's location data was run separately with a "reasonable" choice of scale parameter value, and the automatic grid density calculation option. An "average or typical" value for the scale and grid density values, based on the initial runs for each animal, was chosen and used to re-run all of data for each animal. Scale and grid density values are only comparable between study animals of similar movement patterns; that is, the home ranges of non-migratory study animals are only strictly comparable with those of other non-migratory study animals using the same average scale and grid density values. Changing the home range size and  -63shape in this manner reinforces the notion that quantified home ranges should be accompanied by detailed descriptions and maps of range areas to supplement calculated areas (Harris et al. (1990). Harris et al.'s (1990:97) review of 93 papers on home range analysis using radio-tracking data, and published in the 5-year period to the end of 1988, showed that even 25 years after the first radio-tracking studies, in the majority of papers there was still "insufficient attention given to accurate and sufficient data collection, and to using appropriate analytical techniques to assess home range size and configuration." Less than 33% of the papers Harris et al. (1990) reviewed included an assessment of the accuracy of the radio-locations obtained (fixes), <10% considered whether or not locational fixes were autocorrelated, < 33% stated how many fixes were used to calculate home range size, and < 25% considered whether sufficient locations had been obtained for the home-range estimation to reach an asymptote. The number of radio-fixes required to estimate a home range size must be known before the majority of field work is completed (Harris et al. 1990). This can be accomplished by initiating a trial radio-tracking study period, where a large number of fixes can be collected from a range of animals in order to encompass variations due to sex and age, and to determine at what point home range size reaches an asymptote. This is determined by plotting home range size vs. number of locations, and defined as the point after which additional locations result in a minimal increase in range size. Because asymptotes for individual home range  -64estimates are reached at different values and with a variety of curves depending on the pattern of home range utilization and size, it is important that the selected time interval for the home range estimation is based on a sound assessment of the animal's biology (Harris et al. 1990). In the present study I did not initially determine the number of radio-fixes required; however, the data here presented may act as such information for any subsequent studies of the Strathcona Park Roosevelt elk population. Swihart and Slade (1985) showed a bivariate test of independence using empirically derived critical values for the ratio of mean squared distance between successive observations (t 2 ) to mean squared distance from the centre of activity (r 2 ). Using this, and 2 other tests of independence, Program Home Range automatically determined whether or not each animal location dataset was autocorrelated. The bivariate test of independence showed that over half of the datasets were indeed autocorrelated in the present study. Harris at  al. (1990)  noted that strict adherence to the collection of non-autocorrelated data is worthwhile when the estimation of home range size is the primary aim of the study and/or a large time interval between successive locations is possible. However, they concluded that most radio-tracking studies require the collection of data which are dependent to some degree. So, even with a very large database, it may not be possible to translate autocorrelated data into an independent form and still retain a sample size that is adequate for home range estimation. This was the case in the present study, where it was not possible to translate autocorrelated data into  -65independent observations (see Swihart and Slade 1985, Samuel and Garton 1987, Ackerman ei al. 1990) and still retain a large enough sample size for home range estimation to occur.  SUMMARY The main purpose of my study was to provide the baseline ecological and practical information that will allow Park managers to better understand the current elk-habitat relationship in Strathcona Park before being able to manage for another more "desirable" one. Home ranges and core use areas were estimated using Program Home Range for all 10 radio-collared study animals. Location data for all study animals indicated that the two Heber River elk (#499 and #688) were migratory, while the other 8 radio-collared elk were non-migratory. Population trend is one of the major indicators of elk response to habitat changes (Janz g al. 1980a), and as such is important to understanding the present wildlife-habitat situation, and for forming the context in which future management decisions can be made. Population and age-class composition estimates were presented for each of the 4 study areas. Of concern is the low number of elk (4) in the Thelwood valley. Brunt At al. (1989) estimated the home range sizes of migratory and nonmigratory elk on Vancouver Island. Other studies (Graf 1955, Franklin etal. 1975) and reviews (e.g. Adams 1982) have reported a wide variation in home range size  -66within and among elk supspecies in western North America. Home range sizes estimated in the present study are within the variation reported in the literature; however, as Arcese (1989) observed, home range sizes may vary with population size, and therefore do not provide a universal scale for inter-specific comparisons. The different movement strategies used by the elk in the present study is consistent with notion that migration is likely not a species characteristic of elk, but rather a versatile response to a given environmental situation.  -67CHAPTER 2 Testing Seasonal Models of Roosevelt Elk Habitat Suitability  INTRODUCTION Central to the study of wildlife ecology is how an animal uses its environment; in particular the kinds of foods it consumes and the types of habitats it occupies. Harris et al. (1990) noted that measuring an animal's home range size, shape, and pattern of utilization, are important for most ecological and/or behavioral studies, and particularly for those concerned with habitat selection. The ability to accurately evaluate habitat and to predict wildlife habitat values from previously or readily gathered mapped data, would be of significant value to park, wildlife, and resource extraction managers. One of the first major steps taken toward such evaluative and predictive capabilities was the development of habitat suitability models (U.S. Fish and Wildlife Service 1981). The ability of most models to provide accurate quantitative predictions is limited, but they are useful for understanding and evaluating how systems operate, and can facilitate the development of useful resource management plans (Bunnell 1989). Models of seasonal habitat suitability for Roosevelt elk in forested watersheds were first developed during phase I of the Integrated Wildlife-Intensive Forestry Research (IWIFR) Program (Brunt et al. 1989). These habitat suitability models were designed to provide a reliable means of assessing impacts of forestry development on potential Roosevelt elk habitat (Brunt 1991). The purpose of testing the validity these models in my study, was to provide park, wildlife, and  -68forest managers with a reliable tool for assessing Strathcona Park for areas for general elk habitat suitability, and for determining the impacts of recreational and industrial development on elk habitat. In particular, this should assist park managers in evaluating the land within Strathcona Provincial Park for the presence of elk, for identifying areas suitable for the future placement of elk, and in determining areas important to elk which require special protection or habitat enhancement. Johnson (1980) identified a 4-level ordering of the habitat selection process. First-order selection involves the physical or geographical range of a species. The home range of an individual animal or a social group within that range is determined by second-order selection. Third-order selection refers to the usage made of habitat components within the home range, and fourth-order selection determines the actual procurement of food items from those available at that site. In this chapter, I examine seasonal range selection by a non-migratory Roosevelt elk population in the Elk River valley of Strathcona Provincial Park (second-order selection) in relation to the relative suitability of the range's habitat components (third-order selection). Habitat Suitability Index (HSI) Model Development and Testing The habitat suitability index (HSI) models developed by the U.S. Fish and Wildlife Service (U.S. Fish and Wildlife Service 1981) are based on assessment of the physical and biological attributes of habitat, under the assumption that habitat suitability is proportional to carrying capacity (K. Berry 1986). Habitat use  -69documents a species' use of, or preference for, particular areas within its home range. Two assumptions underlie this notion: i) an individual selects and uses areas that are best able to satisfy its life history requirements; and because of this, ii) greater use should occur in higher quality habitats (Schamberger and O'Neil 1986). These 2 assumptions are not always valid, either because factors other than habitat characteristics may affect an animal's use of a site, or because of our perception of that use. While there has been a recent increase in the number of wildlife habitat models being published in refereed journals, few models have been tested (K.Berry 1986, Bunnell 1989). Most wildlife HSI models have been constructed using the literature and opinions of professionals (K. Berry 1986). Because models are simplifications of the systems, depicting and requiring numerous assumptions, they can never completely mimic the real world (Maynard Smith 1974), and so are incomplete pictures of reality (Bunnell 1973). Bunnell (1989:6) made the important distinction between model validation and model verification, where "validate" means "sufficiently supported by actual fact," and "verify" means "to establish the truth, accuracy, or reality of." The present study proposes to validate Brunt's summer season habitat suitability model. Lancia ?I al. (1982) considered habitat use by individual animals to be the most reliable method of model validation. According to Schamberger and O'Neil (1986), model testing serves 2 important purposes: i) to provide information concerning model performance and  -70reliability; and ii) to provide data that may lead to model improvement. Beyond these, model tests are intended to determine how well a model meets its stated objectives. Models should be developed and tested using 2 or more different sets of the same type of data (Schamberger and O'Neil 1986). The present study represents the second set of data used to validate Brunt's (1991) seasonal habitat suitability models. The first testing of the validity of the models was presented in Brunt (1991). A third testing is currently underway on northern Vancouver Island (K. Campbell, unpubl. data). In testing a model, hypotheses can be formulated at different levels within a model, including tests of assumptions, variables, components, or overall output. Most wildlife habitat model tests have been at the overall output level, which provides little information for improving model performance; testing individual model variables provides information for determining and improving model reliability (Schamberger and O'Neil 1986). This study was designed to test Brunt's (1991) summer season habitat suitability model at the overall output level. The primary means of testing the model was accomplished through linking a Geographic Information System (GIS) with Brunt's (1991) seasonal models. Throughout the process, Brunt's (1991) warning, which cautioned researchers to constantly beware of the "black box" syndrome of computer modeling of ecological relationships was heeded (see also Bunnell 1973, 1989).  GIS Applications in Wildlife Habitat Modeling  -71Although the term GIS is relatively new, the use of spatial information for management decisions has existed since the first map was drawn. A GIS allows the efficient and useful organization, storage, retrieval, and display of information for management decisions (Devine and Field 1986). GIS technology is similar to conventional map processing, yet provides advanced analytic capabilities which can enable managers to address complex issues in entirely new ways (J. Berry 1986). GIS technology was introduced to park, wildlife, and forest managers as a tool for storing, retrieving, and analyzing map and tabular information (Schwaller and Dealy 1986). The main purpose of a GIS is to process spatial information; its main power is that the relationships of the map data can be summarized (database inquiries), or manipulated (analytic processing). These procedures rely on the storage abilities of GIS, which can efficiently organize and search large sets of data for frequency statistics and coincidence among variables (J. Berry 1986). Existing maps can be quickly and efficiently edited with GIS techniques. Predetermined attributes can be entered into the GIS for each polygon (a defined area of particular importance), independent of map creation, so that when complete, an attribute file exists for every polygon (Consoletti 1986). Spatial information is represented numerically, rather than by analog means, such as in the inked lines of a map. This digital representation has the potential for quantitative as well as qualitative processing (J. Berry 1986). A key feature of this study is the importance of assessing habitat interspersion. The use of a GIS allows the determination of spatial interspersion  -72values while retaining the spatial integrity of the habitat data. Eng et al. (1990) noted that previous attempts to incorporate habitat interspersion into wildlife planning have failed to adequately represent wildlife habitat because interspersion indices were added up over large areas. This has resulted in a failure to represent the relationships among individual habitat polygons. Study Background - Brunt's (1991) Thesis Brunt's (1991) study was based on the general movements and seasonal habitat use of a transplanted group of migratory elk. A transplanted group used primarily for 3 reasons: i) habitat use by transplanted animals is more likely to be in response to existing habitat conditions than it is to long-established patterns; ii) for the first several years after the transplant, elk densities in the study area would be lower than in most other Vancouver Island watersheds, thus minimizing the possible confounding effects of density-dependent habitat selection patterns; and iii) a more precise test of model performance may be evaluated because exploratory movements of the transplanted elk can be closely followed to assess which areas within the watershed the animals were or were not familiar with. The seasonal models of elk habitat suitability that Brunt (1991) developed and refined, were thus based on a low-density population of elk in an area completely new to the animals. Brunt's models may in fact primarily reflect (or be confounded by) these 2 factors. In order to further validate the models, it is necessary to test model performance and reliability with a group of elk  -73unconstrained by these factors; namely, a group of elk in a more densely populated watershed with which they are completely familiar. The specific objectives of Brunt's (1991) study were to develop models quantifying Roosevelt elk habitat suitability, and to test these models against elk seasonal range selection patterns in a different area from where they were developed. The ultimate function of these models was to provide wildlife and forest managers with a reliable means of assessing the impacts of forestry development on elk habitat suitability (Brunt 1991). While habitats impose selection pressures which cause particular life history traits to evolve, a complete description of the components of a habitat relevant to the evolution of life histories would probably involve huge numbers of variables, and would become unwieldy for comparing different species (Lessells 1991). Habitat classification is an attempt to summarize the kinds of selection pressures imposed by different habitats in just a few variables. The main factors affecting habitat suitability in Brunt's (1991) models are forage, cover, and the interspersion of forage and cover. Seasonal forage and cover quality for a polygon were calculated from information provided by the understory types, which are similar to ecosystem associations as described by Klinka gt al. (1984). A polygon's forage value was calculated from potential forage values modified according to overstory characteristics of the polygon. In testing Brunt's (1991) habitat suitability models, all potential cover and forage suitability values by understory type, and the modifier values of cover,  -74forage, and interspersion of cover and forage used by Brunt, were used here. I felt that these values are based on sound reasoning and professional experience, and so assumed that they reflect the true (sensu Popper 1959) relative values of model components in each identified habitat polygon. Brunt (1991) developed and tested 3 models of seasonal habitat suitability (summer, mild winter, severe winter seasons) in the context of migratory Roosevelt elk habitat selection patterns on Vancouver Island. The question of seasonal range selection patterns of elk that do not migrate to higher elevations during the summer season as a matter of habit was not addressed in the IWIFR program research in which the habitat relationships used in the models were initially developed. In Brunt's (1991) study, this question was only addressed in a recommendation for further study. Contrary to the importance placed on migratory elk in the IWIFR program and in Brunt's (1991) study, it is known that the majority of Roosevelt elk are non-migratory, as has been previously discussed in Chapter 1. The primary aim of my study was to validate the seasonal habitat suitability models, as applied to the non-migratory Elk River valley elk population, and to determine which model provides the most accurate prediction of seasonal range selection patterns by nonmigratory elk. The Elk River valley elk population was chosen as the study group for habitat model testing because: i) the area is within Strathcona Provincial Park and thus of interest to BC Parks; ii) funding was available for the required habitat mapping of the area; iii) a stable elk population was believed to be present; iv) the area was  -75relatively small and easily accessed; and v) the Elk River valley has somewhat less dense vegetation than other areas inhabited by radio-collared elk, which facilitated frequent direct sightings of the study animals. STUDY AREA Study Area The Elk River valley study area (Figure 1) is an approximately 4450 ha area, fallowing the course of the lower Elk River. The elevation gradient along the entire Elk River changes dramatically from its upper valley headwaters (731 m at Landslide Lake) to its terminus (221 m at Upper Campbell Lake). The Elk River valley may be divided into 2 sections, the upper and lower. The lower valley currently begins at the confluence of the Elk River and the Drum Lakes effluent (diverted water from the Heber River and Crest Creek; Kellerhals 1992), and flows east into its terminus, Upper Campbell Lake. The above described lower Elk River valley is considered as the "study area" in the present study. The upper valley, which flows into the lower, begins approximately 8.5 km south of Drum Lakes, at Landslide Lake. Vegetation in the Elk River valley is characterized by the Coastal Western Hemlock zone (CWH), and above approximately 700-900 m elevation, the Mountain Hemlock zone (MH), both as described by Krajina (1965). A detailed biophysical description of the area can be found in Kojima and Krajina (1975). The most recent description of habitat types along the Elk River valley can be found in Russell (1979).  -76Historical Overview Janz and Lloyd (1977) noted that the major historical land use practice within Strathcona Park influencing present elk distribution is the logging of old Crown-granted lands adjacent to the Elk River. They reported that the majority of the low elevation mature forest was logged during the early 1940's, from valley bottom (221 m elevation) to approximately 230-365 m elevation on the side hills. Although the entire Elk River drainage has been part of Strathcona Provincial Park since its creation in 1911, the lower Elk River, in particular, has been extensively modified by direct and indirect, primarily man-made interferences. Early photographic records and written accounts dating back 50 years indicate a much more stable, productive, and aesthetically pleasing environment previously existed (Tredger et al. 1980). Kellerhals (1992) notes that the main impacts on the study area are thought to be: i) clear-cut logging of the flood plain (mid 1940's); ii) flooding due to a major landslide into a headwaters lake (1946); iii) road construction on the flood plains (mid 1940's - present); iv) flooding of the lowermost 10 km of river by BC Hydro's Strathcona Dam; v)  complete diversion of Crest Creek into the Elk River; and  vi) diversion of parts of the Heber River into the Elk River, which, together with item 5 above, increases the low to intermediate water flows by 6080%. The overall effects of these human-caused interferences has been a dramatic change in river morphology along the lowermost alluvial reaches of the Elk River.  -77The active un-vegetated channel zone is now 2-3 times as wide as it was before 1940, and this process of flood plain erosion is continuing (Kellerhals 1992). During this process, much fish and wildlife habitat has been irrevocably lost (Tredger et al. 1980). METHODS Study Area Map Development Using the Terrasoft 9c GIS (Digital Resources Inc., Nanaimo, B.C.), a digital base map of the Elk River valley was created from 1:20,000 understory association maps, 1:15,000 (approximate) air photographs, and the existing 1:50,000 series 92F/13 - Upper Campbell Lake topographic map. Features digitized into the GIS included biogeoclimatic subzones, understory association, canopy cover, aspect and elevation class, watercourses, roads, and the powerline transecting the study area. The 1:20,000 biogeoclimatic subzone and understory association map was created by Gartner Lee Ltd. (Burnaby, B.C.), under contract to the MOELP, and was based on the biogeoclimatic codes from Green et al. (1984) and understory associations described in Nyberg and Janz (1990). Mapped understory associations were delineated by the contractor from existing topographic and forest cover maps, color air photographs, and 40 field plots and ground truthing lines in the study area. I then digitized the polygons delineating unique habitat types, based on biogeoclimatic subzone and understory type from the 1:20,000 map, into the GIS, in addition to the above listed features.  -78Eight biogeoclimatic subzone units were delineated in the area, and 21 unique habitats based on understory association types. In addition to the 40 field plots and ground truthing lines established by Gartner Lee, Ltd., I placed 10 - 10 m X 10 m field plots in different habitat types within the study area to further test the accuracy of the understory association map. However, the limited number of field plots and the their highly biased placement relative to accessibility, cannot confirm the accuracy of understory association map. Without more extensive verification, the map should be considered unconfirmed. Model Development and Application The main factors affecting habitat suitability in each of the 3 models developed by Brunt (1991) are forage, cover, and the interspersion of forage and cover. The latter is important because elk is an ectone species whose habitat use is concentrated along the edge of more open areas which provide forage, and more dense areas which provide cover (Skovlin 1982). All 3 factors have long been recognized as the basic, seasonally fluctuating, habitat requirements which determine elk abundance (Nyberg and Janz 1990). The model also considers aspect and elevation to be important determinants to winter habitat selection because of their influence on snow accumulation on elk winter range (Brunt 1991), and are combined to form an additional factor used to calculate winter habitat suitability in the mild and severe winter models. Brunt (1991) did not consider the location and density of roads to be important factors affecting habitat suitability in the 3 models. He believed that the  -79potential negative influences of the presence of roads, such as increased hunting pressures, may likely be balanced by the potential positive benefits, which include an abundance of preferred forage along roadsides, and the provision of relatively easy travel routes. The location and density of roads were not considered in my study in accordance with the above argument, and also because the 2 roads and powerline which occur in the study area completely transact the valley, likely affecting the habitats within the entire area in a consistent manner.  Forage Forage values used in the model were based on understory types (Nyberg and Janz 1990) modified by overstory conditions. Potential forage values by understory type for summer and winter seasons were developed based on Vancouver Island Roosevelt elk diet information (Janz 1983, Brunt et al. 1989). These potential forage values were then modified depending on overstory conditions; the underlying notion being that an increase in canopy closure results in a decrease in understory forage production (Nyberg and Janz 1990). Potential forage values ranged from 0 to 0.99, with 0.99 being optimal. Forage and cover values of 0.99 were used rather than 1.0 as optimal to avoid division by 0 in the calculation of overall habitat suitability, as discussed in detail later under "Habitat Suitability Calculations." Forage modifier values are presented in Table 8; potential forage values for the study area are presented in Table 9. A  Table 8. Forage modifier valuesa. —^Habitat  Forage Modifier Value  Logged 5 2 years previously  0.40  Logged 3- 5 years previously  0.75  Logged 5- 15 years previously  1.00  Unlogged - deciduous dominated overstory b  0.75  Unlogged - conifer dominated overstoryc  0.50  Bog/Wetland; Rock outcrop; Slide complex  1.00  4^/  From Brunt (1991) 'Deer fern, Salmonberry, and Swordfern understory types °All other understory types with the exception of Bog/Wetland, Rock outcrop, and Slide complex  -81Table 9. Cover and potential forage suitability values by understory type'. Primary 1^Understory  •  • •  •  • •  Secondary Understory  Ab  B  C  D  -  0  0  0.99  0.90  -  0.99  0.99  0.20  0.30  Moss  0.99  0.99  0.18  0.26  Rock outcrop  0.99  0.99  0.22  0.28  Salmonberry  0.99  0.99  0.32  0.40  Slide complex  0.99  0.99  0.26  0.44  Sword fern  0.99  0.99  0.32  0.38  -  0.99  0.99  0.10  0.10  Rock outcrop  0.99  0.99  0.14  0.12  Dull Oregon grape  0.99  0.99  0.2  0.1  -  0  0  0.30  0.20  -  i 0.99  0  0.30  0.20  -  0.99  0  0.40  0.20  Huckleberry-Moss  0.99  0  0.36  0.22  Lichen-salal  0.99  0  0.36  0.18  Rock outcrop  0.99  0  0.38  0.20  -  0.99  0  0.80  0.80  -  0  0  0.50  0.99  l 0.99  0.99  0.80  0.70  Salmonberry  0.99  0.99  0.80  0.72  -  0  0  0  0  -  'Modified from Brunt (1991) b A = Security Cover; B = Thermal/Snow Interception Cover; C= Potential Winter Forage; D= Potential Summer Forage  -82more detailed explanation of the rationale used in potential forage and modifier cover value selection may be found in Brunt (1991). Cover Nyberg and Janz (1990) discussed 3 types of cover as being important for elk: security, thermal, and snow interception cover. Only 2 types of cover are addressed in the habitat suitability models, security cover, and thermal/snow interception cover. Security cover is defined as vegetative or topographic cover capable of obscuring 90% of a standing adult elk at a distance of <61 m from a human observer (Thomas et al. 1979), although this relatively arbitrary definition is open to question (Rahme 1991). Nyberg and Janz (1990) defined the standard for thermal cover in British Columbia coastal forests as stands > 10 m in height with a mean canopy closure of > 70%; stands > 10 m in height with a mean canopy closure of 60-90% provide snow interception cover for elk in coastal British Columbia forests. Habitat plots established in the study area (n = 50), examination of air photos, consultation with a vegetation ecologist, and a detailed knowledge of the area, formed the basis for decisions on which habitats satisfied elk cover requirements. Cover suitability values are presented in Table 9. Similar to forage values, understory types form the basis for the assignment of cover values in the 3 models (Brunt 1991), however, habitat polygons are not differentiated by their relative cover value; they either do, or do not, satisfy cover requirements. Brunt (1991) felt that there is insufficient information currently available to rank the relative ability of different habitats to satisfy elk cover  -83requirements, and that given our present level of understanding, this "all-ornothing" approach is the most realistic (and conservative). Habitats were assigned values of either 0 or 0.99 for security and thermal/snow interception cover suitability values depending on whether or not they satisfied these cover requirements. Interspersion Interspersion of forage and cover is recognized as a critical component of wildlife habitat suitability (Nyberg and Janz 1990). Numerous studies (e.g. Witmer  In al. 1985, Hunter 1990; see also Nyberg and Janz 1990) have noted the decline in elk use of both forage and cover areas as the distance from their common edge increases. Although elk are an highly mobile species, their use of a particular habitat is strongly influenced by the forage and cover characteristics of nearby habitats. Because of this, one half of the total habitat suitability of a particular polygon is derived from its forage and cover characteristics, and the other half determined from distance from cover and high quality forage (forage suitability value 0.5) (Brunt 1991), Edge-to-edge distances from areas of high forage suitability (z 0.5) and from security and thermal/snow interception cover were determined for the entire study area, and the resulting interspersion factor was determined for each model using the GIS' corridor analysis function. Interspersion modifier values for the various distances from cover and high forage value areas (Table 10) were developed by Brunt (1991) from distance-to-  Table 10. Interspersion modifier valuesa. Distance from Cover or Preferred Forage Areas (m)  Modifier Value  Ob  1.0  <140  1.0  141 - 249  0.6  250 - 300  0.4  >300  0.1/0.01c  aFrom Brunt (1991) b Site qualifies as cover or preferred forage area c0.1 for cover; 0.01 for food  1  -85forage/cover relationships generated in previous elk/habitat research (Brunt et al. 1989, Nyberg and Janz 1990). Modifier values assigned to distances > 300 m from cover or from high quality forage differed because Brunt (1991) considered elk distribution to be more affected by extreme distances from high quality food than from cover. He argued that cover availability was high throughout his study area, and because cover availability was also high in the present study area, I used the same modifier values.  Aspect/Elevation The limiting factor to habitat suitability in mild winter and severe winter models is a combination of site aspect and elevation, which together influence snowpack depth and persistence, both of which are critical components of elk winter range suitability (Brunt 1991). Snow buries forage and increases the costs of locomotion at a time of the year when food availability is limited and energetic demands are high relative to those in summer (Nyberg and Janz 1990). Aspect/elevation modifiers (Table 11) are used in the models because high elevation, north aspect areas receive and retain more snow than do lower elevation, south aspect areas. These modifier values were developed based on relationships among aspect and elevation and snowpack accumulation and melt rates "...from [Brunt's] general experience and observations, and from consultation with other deer, elk, and habitat biologists working on Vancouver Island" (Brunt 1991:56).  Table 11. Aspect/elevation modifier values used in the mild and severe winter models'. Elevation (metres)  Aspect 290-700 (north)  71-1100 (east)  110-250° (south)  250-290 0 (west)  Flat  0-350  0.6  0.8  1.0  0.8  1.0  351-550  0.4  0.6  0.8  0.6  0.8  551-1050  0.2  0.4  0.6  0.4  0.6  >1050  0  0  0  0  0  'From Brunt (1991)  -87Habitat Suitability Calculations To calculate a habitat suitability value for each polygon in the study area for each of the 3 models, I followed Brunt's (1991) methods. Habitat polygon data were digitized into the GIS in a vector (line) format, and subsequently gridded into a raster format for overlay analyses and area calculations. Data levels containing forage, cover, interspersion, and aspect/elevation values were overlaid, resulting in the formation of 492 polygons for the entire study area. Each resultant polygon was assigned summer, mild winter, and severe winter habitat suitability values based on the respective model's algorithm. Within the model algorithms, the relative weight values applied to the variables and component relationships used to calculate seasonal range suitability "...were judgement based estimates of the relative contribution of the various factors to habitat suitability derived from 10 years of elk-forestry studies on Vancouver Island" (Brunt 1991:58). The basic assumptions used in the development of these weighting values were the following (Brunt 1991): i) forage quality/availability is more important than cover during summer and mild winters compared to severe winters; ii) snow interception cover is not required on summer ranges and is less important during mild winters than during severe winters; and iii) the influence of aspect/elevation on snowpack accumulation and persistence is more important during severe winters than mild winters.  The algorithms (Brunt 1991) used to calculate seasonal range habitat suitability values were:  -88-  Summer Season Habitat Suitability: exp (0.5*ln (1-exp(0.7*In (1-SUMM) + 0.15*In (1-THERMAL) + (0.15*In (1SECURITY)))) + 0.25*In (SUMMDIS) + 0.125*In (TDIS) + 0.125*In (SDIS))  Mild Winter Season Habitat Suitability: exp (0.5*In (1-exp(0.7*In (1-WINT) + ((0.2*In (1-THERMAL)) + (0.1*In (1SECURITY)))) + 0.25*In (WINTDIS) + 0.125*In (TDIS) + 0.125*In (SDIS))*ASEL• 5  Severe Winter Season Habitat Suitability: exp (0.5*In (1-exp(0.5*In (1-WINT) + ((0.4*In (1-THERMAL)) + (0.1*In (1SECURITY)))) + 0.20*In (WINTDIS) + 0.20*In (TDIS) + 0.1*In (SDIS))*ASEL L° Where:^In = natural logarithm; exp(x) = e' x '; WINT = winter forage value; SUMM = summer forage value; THERMAL = thermal cover value; SECURITY =security cover value; WINTDIS =distance to winter forage of value ....0.5; SUMMDIS =distance to summer forage of value ....0.5; TDIS = distance to thermal/snow interception cover; SDIS = distance to security cover; and ASEL = aspect/elevation modifier value.  A more detailed explanation of algorithm development and choice of mathematical computations is presented in Brunt (1991:56-61).  Home Range Suitability Calculations The cumulative ranges and core use areas generated from the home range program were imported into the GIS to generate habitat suitability values for the ranges. The seasonal suitability values of the individual polygons within the ranges were weighted by the area of the polygon in question relative to the total Elk River valley study area. The sums of the weighed habitat suitability values represented the overall suitability of the range.  -89Analyses of Modeled Habitat Suitability Three main analyses were conducted on the habitat suitability data generated from application of the 3 seasonal models. Where appropriate, the habitat suitability of individual animal locations and/or the habitat suitability of animal home ranges generated by the home range program from the animal location data, were used for model validation tests. The 3 validation tests of the habitat suitability models' output were the following: A) testing for random habitat selection with respect to modeled habitat suitability; B) examining modeled habitat suitability of the total study area, unused portions of the study area, cumulative range, and core use areas; and C) assessing which of the 3 seasonal habitat suitability models best reflects the range selection of non-migratory elk.  A) Random Habitat Selection Test Brunt's (1991) habitat suitability models would have no predictive capability if the elk were selecting habitats at random with respect to modeled habitat suitability. In order to provide "statistical evidence" that no random habitat selection occurred within the study area, a chi-square goodness-of-fit test was applied to the summer season animal location data. Because it was immediately obvious that random habitat selection was not occurring, and because modeled mild winter and severe winter habitat suitability values for the animal location data were very similar to those of the summer location data, goodness-of-fit tests were not applied to the 2 winter models.  -90Four classes of habitat suitability were defined in an attempt to follow the suggestions made by Cochran (1954:417) that no expected frequency should be less than 1.0, and no more than 20% of the expected frequencies should be less than 5.0. Expected frequencies were determined by calculating the proportion of the entire study area having habitat suitability within the 4 classes chosen multiplied by the number of locations obtained (n = 216) for the summer season animal location data. Observed frequencies were identified as the number of animal locations occurring within each of the 4 suitability classes. B) Habitat Suitability Within Study Area If the 3 seasonal habitat suitability models accurately reflect elk habitat selection patterns, areas of higher use should have higher suitability values. The principle validation test of the models' ability to predict elk seasonal range selection was to compare the habitat suitability of the following 4 calculated regions: the total study area; portions of the total study area unused by elk; the cumulative elk range within the study area; and the core use area within the cumulative elk range. All 3 seasonal range suitability models were applied to each of the 4 range types. The seasonal suitability values of the individual polygons within each of the 4 regions were weighted by the relative area that the polygon represented of the total area of the region. The sums of these weighted values represent overall suitability of the 4 regions. The total study area was somewhat arbitrarily delineated west-to-east by the maximum height of the Elk River valley watershed at the Drum Lakes, to Upper  -91Campbell Lake, and north-to-south by the approximate height of land of the study area. The cumulative range and core use area were estimated by Program Home Range (Ackerman Dt DI. 1990) using all animal locations (n = 284). The portion of the study area unused by elk was determined by repeated ground and aerial surveys of those regions of the total study area exclusive of the cumulative range throughout the season of interest. Although Janz and Lloyd (1977) reported that these areas supported a very small group of elk during more severe winter seasons, no elk sign could be found during the study period, and I feel confident that these areas did not support any elk during the study period. Although comparisons of total study area, unused total area, cumulative range, and core use area habitat suitability values cannot be accomplished in a statistically rigorous fashion (essentially because there is only one of each range type available for comparison with the other types), if the models adequately predict habitat selection, a trend should be evident, indicating habitat suitability to be proportional to the elk's use of an area. A more rigorous statistical analysis of the habitat suitability values between animal locations (n = 284) in the intensively used (core use area), and 284 randomly generated locations in the unused portions of the study area was accomplished by using a one-tailed t-test. Habitat suitability was weighted by the area of the polygon relative to the area of the core use area in which all of the animal locations fell. Each animal location and randomly generated location was given the habitat suitability value of the polygon in which it fell, for each of the 3 models. Means  -92and standard deviations of the habitat suitability values were then calculated for each model. If the models adequately predict habitat selection, the mean habitat suitability value of the core use area animal locations should be significantly higher than that of the randomly generated locations in the unused portion of the study area.  C) Comparisons Among Seasonal Model Prediction As discussed earlier, if the models accurately reflect elk habitat selection, areas of higher use should have higher suitability values. But because the Elk River elk are non-migratory, it is expected that at least 2 of the models (summer and winter) should be equally good at predicting year-round habitat selection; no one model should predict higher suitability values for areas of higher use. In order to determine if there is any difference in the abilities of the 3 models to predict habitat selection for a non-migratory elk population, habitat suitability values generated by all 3 models were compared using animal location data. Habitat suitability were weighted by the area of the polygon relative to the area of the core use area in which all of the animal locations fell. Each animal location (n = 284) was given the habitat suitability value of the polygon in which it fell, for each of the 3 models. Means and standard deviations of the habitat suitability values were then calculated for each model. D'Agostino's test for normality (D'Agostino 1971 a, b) indicated that the distribution of habitat suitability values for the animal location data was non-normal (D = 0.191495, critical values = 0.2705, 0.2866, a = 0.05). However, because  -93n = 284 was large enough to approximate a normal distribution, two-tailed paired ttests were performed to determine statistical difference between mean habitat suitability value output of each model (A. Kozak; pers. comm.). RESULTS Habitat Suitability Model Validation Tests Figure 7 presents the Elk River valley study animals' (elk #'s 120, 179, 218, and 259) cumulative range and core use area within the total study area as estimated by Program Home Range (harmonic mean estimator). A) Random Habitat Selection Test Summer habitat suitability values of locations used by elk in summer were significantly different than expected values based on availability (Table 12); elk used habitats in summer with the highest suitability values ( > 0.90) more often than expected. This indicates non-random habitat selection with respect to summer model output, and suggests that further analyses to test model validity are warranted. B) Habitat Suitability Within Study Area A consistent trend of increasing habitat suitability value is obvious in all 3 models from the unused portion of the study area, to the total study area, to the cumulative range, to the core use areas (Table 13). Mean habitat suitability of animal locations (n = 284) in the core use area was significantly higher than that of randomly generated locations (n = 284) in the  TOTAL STUDY AREA ^ CUMULATIVE RANGE CORE USE AREA  UNUSED PORTION OF TOTAL STUDY AREA  , ___  _  -- — - - Logging Road _  Highway  ..—....„-------------._,„,-----  0^  28 2.5^5 km  Figure 7. Total study area, unused portion of the study area, cumulative range, and core use area for elk #'s 120, 179, 218, and 259 in the Elk River Valley, as delineated for seasonal habitat suitability model testing.  Table 12. Frequencies of expected (given random habitat selection) and observed habitat suitability class values for the summer model.  Summer Model Suitability Class  "  Summer' Animal Locations (n = 216) Observed  Expected'  24  36.12  0.60 - 0.87  0  79.54  0.87 - 0.90  60  62.71  132  37.64  <0.60^t  >0.90  Chi-square = 320.28; critical value = 7.815; a = 0.05; degrees of freedom = 3  aSummer = May 1, 1992 - September 15, 1992 'Proportion of the study area in the suitability class multiplied by 216 locations  Table 13. Habitat suitability values for unused portion of the total study area, total study area, cumulative range, and core use area for summer, mild winter, and severe winter models. Unused Portion of Total Study Area  Cumulative Range  Core Use Area  0.76  0.71  0.81  0.88  Mild Winter  0.73  0.67  0.77  0.86  Severe Winter  0.76  0.71  0.81  0.89  Model Summer  Total Study Area  -97unused portion of the study area (Table 14), for all 3 models (summer t =10.551, mild winter t = 9.056, severe winter t= 9.736; for all tests  t criti„, =  1.645, df = 566,  a =0.05). C) Comparisons Among Seasonal Model Output The mean habitat suitability values of summer, winter, and cumulative (summer and winter) animal location datasets calculated by all 3 models showed little variation (Table 15). Of the 9 comparisons possible between mean habitat suitability values of summer, winter, and cumulative animal location datasets for each of the 3 models (Table 16), only that between summer and winter values from the mild winter model, was significantly different (t = 2.180,  t-critical =  1.960,  df = 282, a = 0.05) (Table 16).  DISCUSSION Brunt's (1991) models are models of understanding; they represent a first step towards the development of a predictive tool capable of making realistic, quantitative predictions about the response of elk populations to habitat management activities. While the results of my study validate the models' simplification of the real-world elk-habitat relationship, one must keep in mind Bunnell's (1989:8) caution that "validity and simplicity are not synonyms of veracity," and that although a researcher can quantify the accuracy of a predictive model for a manager, he cannot quantify the degree to which the model has been strengthened.  Table 14. Mean habitat suitability values for animal locations in the core use area, and random locations in the unused portion, of the Elk River valley study area. Mean Habitat Suitability Value of Model  Core Use Area Animal Locations (n= 284)  Unused Portion of Study Area Random Locations (n= 284)  Summer  0.87^(0.15)a  0.66 (0.30)  Mild Winter  0.87 (0.17)  0.68^(0.31)  Severe Winter  0.89^(0.19)  0.67 (0.33)  "Sample standard deviation  Table 15. Mean habitat suitability values of summera, winter b and cumulative animal locations estimated by each of the 3 seasonal models. ERV Group` Model  Summer Locations (n= 216)  Winter Locations (n= 68)  Cumulative Locations (n= 284)  Summer  0.88^(0.14) d  0.84 (0.17)  0.87^(0.15)  Mild Winter  0.89^(0.16)  0.84 (0.18)  0.87^(0.17)  Severe Winter^0.90 (0.18)  0.86 (0.22)  0.89 (0.19)  'May 1, 1992 - September 15, 1992 January 11, 1992 - April 31, 1992; September 16, 1992 - January 6, 1993 `Includes elk #'s 120, 179, 218, and 259 d Sample standard deviation b  -100-  Table 16. Results of all possible two-tailed paired Mesta comparisons among summer b , winter, and cumulative d animal location datasets using the 3 habitat suitability models.  Model Used  Summer  Mild Winter  Severe Winter  )  Comparisons Between Animal Location Datasets  t-value  summer and winter  + 1.948  summer and cumulative  + 0.760  winter and cumulative  -1.443  summer and winter  + 2.180  summer and cumulative  + 1.336  winter and cumulative  -1.292  summer and winter  +1.512  summer and cumulative  +0.596  winter and cumulative  -1.133  aln all cases a = 0.05, and t-critical =^1.960 Summer = May 1 - September 15, 1992 'Winter = January 11 - April 30, 1992; and September 16, 1992 - January 6, 1993 d Summer and winter animal location datasets combined b  -101The models of seasonal habitat suitability tested in the present study were validated in that they appear to reflect elk range selection patterns. While no true seasonal ranges were selected by the non-migratory Elk River valley elk population, modeled suitability of the habitats selected by elk were significantly different from what would be expected if selection had been random. This was the case for all 3 models, using both the animal location data and the home ranges estimated from those location data. For all 3 models, validation tests further indicated that the cumulative elk ranges were of significantly greater suitability value that the total study area and, as a consequence, those portions of the total study area unused by elk. In addition, the elk core use areas had significantly greater suitability values than the cumulative ranges. Core use areas are especially important because they denote areas of particularly high home range usage, and thus they often may provide a more clear measure of the changing pattern of range use than does total home range area. Core use areas are often more useful than the more peripheral contours delineating an animal's home range for understanding both intraspecific and interspecific patterns of home range use (Harris et al. 1990). As expected, the animal locations within the core use area had the greatest mean habitat suitability values within the entire study area. The relative increase in predicted habitat suitability values from unused to core use areas indicates the ability of the models tested in this study to reflect elk habitat selection patterns. The assumption being that elk select cumulative areas  -102to satisfy certain requirements from all areas available to them (the total study area =the lower Elk River valley), and that within the cumulative range, the elk "key in" on core use areas of particularly high habitat suitability. Similar consistent trends in Brunt's (1991:132) study led him to likewise conclude that they provided "...what I consider the best corroborative evidence" of the models' validity. Habitat suitability values for both home range and animal location datasets in the Elk River valley were notably higher than those calculated in Brunt's (1991) study of migratory elk, where cumulative range habitat suitability values ranged from 0.27-0.55; seasonal range values from 0.36-0.74; and core use area values from 0.24-0.85. The difference in habitat suitabilities calculated in my study and Brunt's (1991) is probably related to the fact that Brunt's study animals were transplanted into a watershed with which they were unfamiliar, and consequently unaware of all areas of high habitat suitability value. In addition, Brunt's study animals were all migratory. One of the reasons the Elk River valley elk are nonmigratory may be related to the higher habitat suitability values found in the area they occupied. In the present study, no one model best predicted the habitat suitability of the home range selected by the non-migratory study animals. Studying migratory elk, Brunt (1991) found that the 2 winter models predicted statistically higher suitability values for winter ranges than for summer ranges, as was expected. However, there was no statistical difference between the summer model's predicted suitability value for the summer and winter ranges. Brunt postulated that  -103the winter ranges within his study area could provide areas of similar suitability to those occupied during the summer (winter ranges had the highest suitability predicted by each model), but elk introduced to the study area continue to migrate out of habit. In support of this concept that elk movement behavior becomes relatively fixed and unchanged despite translocation, Brunt (1991) noted that in previous transplants of non-migratory elk on Vancouver Island, the animals remained non-migratory even though high quality summer range was available at higher elevations. Likewise, migratory elk may continue to migrate even when adequate summer range is available without migrating. In the present study I found that the 3 models showed little variation in calculated habitat suitability values using summer, winter, and cumulative animal location datasets. There was no statistical difference between model outputs except between summer and winter using the mild winter model. In view of the other 8 comparisons for which no differences exist (Table 16), I believe the difference between the summer and winter model is unimportant, and that there is virtually no difference in habitat suitability values predicted by the 3 models. Russell (1979) concluded that the amount of available forage in the Elk River valley is the greatest limiting factor on the number of elk using this area. In addition to forage as a limiting factor, the vast majority, if not the entirety, of the study area has been significantly and routinely disturbed from its pre-colonial state (Tredger et al. 1980, Kellerhals 1992). Unquestionably, this must have had numerous negative effects on local fish and wildlife populations, including on the  -104Roosevelt elk. Two potential threats particular to the elk in the Elk River valley include illegal hunting and accidental road kill, both of which are facilitated by the 2 roads running through the area. In spite of these threats, I agree with Brunt's (1991) recommendation that the location and density of roads need not be incorporated into the models. Illegal hunting and accidental road kill are unpredictable factors, and thus difficult to incorporate as variables of habitat suitability models on Vancouver Island. Eng et al. (1990) noted that Vancouver Island exhibits a cyclical pattern of snow depths; severe winters occur approximately every 18 years. Thus, long term maintenance of elk populations requires the availability of both mild and severe winter habitats. Because the winter season during the study period provided conditions under which the mild winter model might be applied (MOELP; pers. comm.), the severe winter model remains untested with respect to both migratory and non-migratory elk populations. Brunt (1991) noted that since the initiation of the IWIFR research project which provided the majority of information for the development of the models, winter weather conditions on Vancouver Island have been relatively mild. Until the habitat relationships that drive the severe winter model can be tested during winters of prolonged, deep snow accumulations, the severe winter model cannot be considered more than preliminary, and as Brunt (1991:133) cautioned, "...applied with prudence." Brunt (1991) recommended conducting further model validation tests in different areas during years of different weather conditions; this provides the  -105necessary context for more intensive research in the future. As discussed earlier, most tests of wildlife-habitat models, including Brunt's (1991), the present study, and the ongoing study on northern Vancouver Island, examine only model output. Schamberger and O'Neil (1986) noted that additional tests of the individual model variables or assumptions of the model provide information for determining and improving model reliability. Intensive research into the individual model components and relationships would be the next logical step in further assessments of the validity of the models, and to improve their predictive capabilities (Brunt 1991). This is particularly appropriate in light of the findings of the present study, in terms of the extent to which the models apply to the habitat selection patterns of non-migratory Roosevelt elk. Brunt (1991) noted that previous transplants of non-migratory elk on Vancouver Island to an area where higher elevation summer range was available did not result in the animals adopting migratory behavior to take advantage of these areas. Further model validation tests using such a transplanted population may provide additional insight into the dynamics of nonmigratory elk habitat selection patterns.  CONCLUSIONS After testing and validating the 3 habitat suitability models, Brunt (1991) still concluded that the models may not be embraced by forest and wildlife managers. The principle reason for possible reluctance to make extensive use of the models centers on the understory mapping which is necessary for determination of both  -106forage and cover suitability indices used in the models. Unfortunately, information to generate an understory map is not readily available from existing map bases. An extensive background is required in vegetation ecology, expertise in air photo interpretation, and extensive ground checking to prepare a validated understory map (Brunt 1991). This will likely prohibit ready acceptance of the models into general forest resource and wildlife-habitat planning. Bunnell (1989) considered 4 ways by which modeling can waste time, talent, and funding. They include misuse, where models are used for the wrong purpose; disuse, where models are no longer used; lack of model evaluation; and legal challenges to the validity of the model and its use in making management decisions. The models tested in the present study are most likely to fall victim to the second fate: disuse. Disuse of a model is obviously wasteful if developing the model was expensive in terms of time and money, and the model is not used. Bunnell (1989:2) noted 4 broad, related reasons why wasteful disuse of models occurs: i) the model does not address the question or purpose of the intended user; ii) the intended user never existed, "just someone who might be convinced"; iii) although it asks the right question, the model is too complex to be used and/or requires too much data; and iv) model output is not sufficiently accurate. At the present time and state of model accuracy, all 4 reasons apply to BC Parks potential disuse of the models.  -107As was previously discussed, the primary purpose of initially developing the models was to provide forest managers with a reliable means of assessing the impacts of forestry development on elk habitat suitability. However, it has been suggested that BC Parks might apply the models as a management tool with a slightly different goal: to make predictive use of the models in terms of identifying high quality elk habitat throughout and adjacent to Park lands. Such identified areas would then be targeted for a particular management action such as special protection, reduced human access, transplant activities, etc. I believe that the cost-benefit ratio for this use of the model is not warranted, given the time and money required to collect the information necessary to generate an understory map of the entire Strathcona Park. A systematic identification of high quality elk habitat in the Park by qualified elk-habitat experts could be accomplished in less time, for far less money, and would not necessitate further ground verification. However, in addition to identifying high quality elk habitat, the baseline data gathered in the completion of a Park-specific understory map would allow for future studies of other species wildlife-habitat relationships. This study provides additional support for a correlation between predicted habitat suitability and elk habitat selection. Like Brunt's (1991) initial study of model development and testing, it is, in effect, an unreplicated experiment with no controls. Further research in watersheds throughout Vancouver Island and on nonmigratory as well as migratory elk is needed to gain confidence in the results from  -108application of the models. The models still must not in themselves be considered as decision-making tools. Romesburg (1981) stated that modeling was never intended to function as a means to scientific knowledge, and that the use of modeling in science is limited because it cannot predict to within established tolerances. However, its continual use is assured as a planning tool that can integrate scientific knowledge and common sense, as well as theory, hunches, and expert opinion to forecast alternative future images: "Planning is inaccurate by the standards of science, but that is no reason to abandon planning," (Romesburg 1981:310). The corroborative evidence obtained to date suggests that Brunt's (1991) models may indeed be useful in helping managers make more informed decisions about elk habitat suitability, and the effects of man-centered activities on this suitability.  -109Literature Cited Ackerman, B.B., F.A. Leban, M.D. Samuel, and E.O. Garton. 1990. User's manual for Program Home Range: second edition. For., Wildl. and Range Exp. Stn. Tech. Rep. 15, Contrib. No. 259. University of Idaho, Moscow. Adams, A.W. 1982. Migration. In Elk of North America: ecology and management. Edited 12 y J.W. Thomas and D.E. Toweill. Harrisburg. Stackpole Books. pp. 301-321. Anderson, D.J. 1982. The home range: a new nonparametric estimation technique. Ecology 63(1):103-112. Arcese, P. 1989. Intra-sexual competition, mating system and natal dispersal in song sparrows. Anim. Behay. 38:958-979. Baker, R.R. 1978. The evolutionary ecology of animal migration. New York. Holmes and Meier Publishers. Bekoff, M., and L.D. Mech. 1984. Simulation analyses of space use: home range estimates, variability, and sample size. Behavior Research Methods, Instruments and Computers. 16(1):32-37. Bergerud, A.T. 1974. The role of the environment in the aggregation, movement and disturbance behavior of caribou. In The behavior of ungulates and its relation to management. Edited by V. Geist and F. Walther. I.U.C.N. Publication 24, Volume 2. pp. 552-584. Berry, J.K. 1986. GIS: learning computer-assisted map analysis: report part III. J. For. 84(10):39-43. Berry, K.H. 1986. Introduction: development, testing, and application of wildlifehabitat models In Wildlife 2000:modeling habitat relationships of terrestrial vertebrates. Edited by J. Verner, M.L. Morrison, and C.J. Ralph. pp. 3-4. Blood, D.A. 1988. Elk transplant feasibility study, Thelwood valley, Strathcona recreation area. Min. Environ., Lands, and Parks, Parks and Outdoor Rec. Div., South Coast Region, North Vancouver, B.C. 54 pp. Boulanger, J.G. and G.C. White. 1990. A comparison of home range estimators using Monte Carlo Simulation. J. Wildl. Manage. 54(2):310-315. Bowen, W.D. 1982. Home range and spatial organization of coyotes in Jasper National Park, Alberta. J. Wildl. Manage. 46(1):201-216.  -110Boyce, M.S. 1991. Migratory behavior and management of elk (Cervus elaphus). Appl. Animal Behay. Sci. 29:239-250. Braun, S.E. 1985. Home range and activity patterns of the giant kangaroo rat, Dipodomys ingens. J. Mammal. 66(1):1-12. British Columbia Wildlife Branch. 1991. Managing wildlife to 2001:a discussion paper. British Columbia's environment, planning for the future. ISSN 11835931. Brunt, K.R. 1990. Ecology of Roosevelt elk. Ch. 3 In Deer and elk habitats in coastal forests of southern British Columbia. Edited by J.B. Nyberg and D.W. Janz. B.C. Mins. For. and Environ., Res. Br. Spec. Rept. Ser. No. 5. Victoria, B.C. Brunt, K.R. 1991. Testing models of the suitability of Roosevelt elk (Cervus elaphus roosevelti) seasonal ranges. M.Sc. Thesis, University of Victoria, Victoria, B.C. Brunt, K.R., D.Q. Decker, and J.A. Youds. 1989. Vancouver Island Roosevelt elk/intensive forestry interactions. Phase I (1981-1986). Job completion report. Research, B.C. Ministries of Environment and Forestry. IWIFR-33, Victoria, B.C. Brunt, K. R. and C. A. Ray. 1986. Seasonal range habitat suitability index models for Vancouver Island Roosevelt elk. In Proceed. 1986 Western States and Provinces Elk Workshop, Edited by D. L. Eastman. Oreg. Dept. Fish and Wildl., Portland. pp. 117-144. Bryant, L.D., and C. Maser. 1982. Classification and distribution. In Elk of North America: ecology and management. Edited by J.W. Thomas and D.E. Toweill. Harrisburg. Stackpole Books. pp. 1-60. Bunnell, F.L. 1973. Theological ecology, or models and the real world. The Forestry Chronicle. 49(4):167-171. Bunnell, F.L. 1978. Snow, tress, and ungulates. Report to B.C. Fish and Wildl. Br., CARP. 72 pp. Bunnell, F.L. 1989. Alchemy and uncertainty: what good are models? Gen. Tech. Rept. PNW-GTR-232. U.S.D.A. For. Ser., Pac. N. W. Res. Sta., Portland, Oregon. pp. 1-27.  Burt, W.H. 1943. Territoriality and home range concepts as applied to mammals. J. Mammal. 24(3):346-352. British Columbia Ministry of the Environment, Lands, and Parks. 1989a. Provincial wildlife strategy. Wildlife Branch, Victoria, B.C. In preparation. British Columbia Ministry of the Environment, Lands, and Parks. 1989b. Regional wildlife plan for the Vancouver Island region. Wildlife Branch, Victoria, B.C. In preparation. Clutton-Brock, T.H., F.E. Guinness, and S.D. Albon. 1982. Red deer: behavior and ecology of two sexes. Chicago. University of Chicago Press. Cochran, W.G. 1954. Some methods for strengthening the common chi-square tests. Biometrics 10:417-451. Consoletti, W.L. 1986. GIS in industrial forest management. In GIS applications: report part II. Journal of Forestry 84(9):40-41. Cowan, I.McT. and C.J.Guiget. 1965. The mammals of British Columbia. B.C. Prov. Mus. Handb. No. 11. Victoria, B.C. Craighead, J.J., F.C. Craighead, Jr., R.L. Ruff, and B.W. O'Gara. 1973. Home range patterns of non-migratory elk of the Madison drainage herd as determined by biotelemetry. Wild!. Monogr. No. 33. Washington, D.C.: The Wildlife Society. 50 pp. D'Agostino, R.B. 1971 a. An omnibus test of normality for moderate and large samples. Biometrika 58:341-348. D'Agostino, R.B. 1971 b. Tables for the D test of normality. Department of Mathematics, Boston University. Research Report. April 1971. Dasmann, R.F. and R.D. Taber. 1956. Behavior of Columbian black-tailed deer with reference to population ecology. J. Mammal., 37:143-164. Devine, H. and R. Field. 1986. The gist of GIS: report part I. Journal of Forestry 84(8):17-22. Dixon, K.R., and J.A. Chapman. 1980. Harmonic mean measure of animal activity areas. Ecology 61(5):1040-1044.  -112Eng, M.A., R.S. McNay, and R E. Page. 1990. Integrated management of forestry and wildlife habitat with the aid of a GIS-based habitat assessment and planning tool In GIS '90 - Vancouver. Sympos. Proceed. pp. 185-190. Franklin, W.L., A.S. Mossman, and M. Dole. 1975. Social organization and home range of Roosevelt elk. J. Mammal. 56(1):102-117. Geist, V. 1982. Adaptive behavioral strategies. in Elk of North America: ecology and management. Edited by J.W. Thomas and D.E. Toweill. Harrisburg. Stackpole Books. pp. 219-278. Graf, W. 1943. Natural history of the Roosevelt elk. Ph.D. Thesis. Oregon State University, Corvallis. 222 pp. Graf, W. 1955. The Roosevelt elk. Port Angeles Evening News, Port Angeles, Washington. 105 pp. Green, R.N., P.J. Courtin, K. Klinka, R.J. Slaco, and C.A. Ray. 1984. Site diagnosis, tree species selection and slashburning guidelines for the Vancouver Forest Region. B.C. Min. For., Victoria, B.C. Land Manage. Handb. 8. Greenwood, P. 1980. Mating systems, philopatry and dispersal in birds and mammals. Anim. Behay. 28:1140-1162. Harper, J.A. 1971. Ecology of Roosevelt elk. Oregon State Game Comm. 44 pp. Harper, J.A., J.H. Harn, W.W. Bentley, and C.F. Yocum. 1967. The status and ecology of the Roosevelt elk in California. Wildl. Monogr. No. 16. Washington, D.C.: The Wildlife Society. 49 pp. Harestad, A.S. 1979. Seasonal movements of black-tailed deer on northern Vancouver Island. B.C. Ministry of the Environment. Fish and Wildlife Report No. R-3. 98 pp. Harris, S., W.J. Cresswell, P.G. Forde, W.J. Trewhella, T. Woollard, and S. Wray. 1990. Home-range analysis using radio-tracking data -- a review of problems and techniques particularly as applied to the study of mammals. Mammal. Rev. Volume 20, Nos. 2/5, 97-123. Hayne, D.W. 1949. Calculation of size of home range, J. Mammal. 30(1):1-18. Hebert, D.M. 1973. Altitudinal migration as a factor in the nutrition of bighorn sheep. Ph.D. Thesis. University of British Columbia, Vancouver.  -113Holland, S.S. 1964. Landforms of British Columbia. B.C. Dept. of Mines and Petro. Res. Bull. No. 48. Victoria, B.C. 138 pp. Hunter, M.L. 1990. Wildlife, Forests. and Forestry - Principles of Managing Forests for Biological Diversity. Prentice-Hall, Inc. Engelewood Cliffs, New Jersey. 370 pp. Janz, D.W. 1980. Preliminary observations on seasonal movements and habitat use by Vancouver Island Roosevelt elk. In Proc. of the Western States Elk Workshop. Edited by W. MacGregor. Prov. of B.C. Ministry of the Environment. 174 pp. Janz, D.W. 1983. Seasonal composition and quality of Roosevelt elk diets on Vancouver Island. MSc. Thesis, University of British Columbia, Vancouver, B.C. Janz, D.W., and K. Lloyd. 1977. Elk investigations in Strathcona Park and adjacent area. Progress report: Jan. 1976-June 1977. B.C. Fish and Wildlife Branch, Nanaimo, B.C. 59 pp. Janz, D., K. Brunt, and J. Youds. 1980. Vancouver Island Roosevelt elk/intensive forestry interactions: working plan 1981-1985. Research, Ministries of Environment and Forests. IWIFR-12. Victoria, B.C. 71 pp. Janz, D.W., and D. Becker. 1986. Vancouver Island elk - animal and use characteristics. pp. 279-305 In Proc. 1986 West. States and Prov. Elk Workshop. D.L. Eastman (ed). Oregon Dep. Fish Wildl., Portland, Oregon. Jennrich, RA., and F.B. Turner. 1969. Measurement of non-circular home range. J, Theoretical Bio. 22(2):227-237. Johnson, D.H. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology, 61(1):65-71. Jones, G. 1983. Roosevelt elk in Strathcona Park. Survey Report. B.C. Fish and Wildl. Kellerhals, R. 1992. Strathcona dam and reservoir, Elk River channel stability and mitigative measures. B.C. Hydro and Power Authority, Vancouver, B.C. Klinka, K., R.N. Green, P.J. Courtin, and F.C. Nuszdorfer. 1984. Site diagnosis, tree species selection and slashburning guidelines for the Vancouver Forest Region. Land Manage. Handbk. No. 8. B.C. Ministry of Forests, Burnaby.  -114Knight, R.R. 1970. The Sun River elk herd. Wild!. Monogr. 23. 66 p. Kojima, S., and V.J. Krajina. 1975. Vegetation and environment of the coastal western hemlock zone in Strathcona Provincial Park, British Columbia, Canada. Syesis 8:1-123. Krajina, V.J. 1965. Ecology of western North America. Dept. of Botany, University of British Columbia. 112 pp. Lancia, R.A., S.D. Miller, D.A. Adams, and D.W. Hazel. 1982. Validating habitat quality assessment: an example. Trans. 47 th N. Amer. Wild!. and Nat. Res. Conf. 47:96-110. Lessells, C.M. 1991. The evolution of life histories. In Behavioral Ecology. Edited bs J.R. Krebs and N.B, Davies. Blackwell Scientific Publishers. Oxford. pp. 32-68. Madison, J. 1966. The elk. Conserv. Dept., Winchester-Western Div., Olin Corp. East Alton, Ill.: Olin Corp. 125 pp. Maurer, A.L. 1989. Strathcona Provincial Park Wildlife viewing development plan. Min. Environ., Lands, and Parks. South Coast Region, North Vancouver, B.C. Maynard Smith, J. 1974. Models in ecology. Cambridge University Press, Cambridge, England. 146 pp. McCullough, D.R. 1969. The Tule elk: its history, behavior, and ecology. Univ. Calif. Publ. Zool., Vol. 88. Berkeley and Los Angeles: Univ. California Press. 209 pp. McCullough, D.R. 1985. Long range movements of large terrestrial mammals. In Migration: mechanisms and adaptive significance. Edited by M.A. Rankin. Contrib. Marine Sci. Suppl. 27:444-465. Michener, G.R. 1979. Spatial relationships and social organization of adult Richardson's ground squirrels. Can. J. Zool. 57(11:125-139. Morgantini, L. 1988. Behavioral adaptive strategies of wapiti in the Canadian Rockies. Ph.D. Thesis. University of Alberta, Edmonton. Morgantini, L. and R.J. Hudson. 1983. Nutritional significance of altitudinal migrations for wapiti. Agriculture and Forestry Special Issue 62:109-112.  -115Murie, O.J. 1951. The elk of North America. The Stackpole Co., Harrisburg, Pennsylvania and Wild!. Manage. Inst., Washington, D.C. 376 pp. Nelson, J.R., and T.A. Leege. 1982. Nutritional requirements and food habits. In Elk of North America: ecology and management. Edited by J.W. Thomas and D.E. Toweill. Harrisburg. Stackpole Books. pp. 323-367. Nyberg, J.B. and D.W. Janz, eds. 1990. Deer and elk habitats in coastal forests in southern British Columbia. B. C. Min. For., B. C. Min. Environ., Res. Br. Spec. Rept. Ser. No. 5. Victoria, B. C. Nyberg, J.B., L.O. Peterson, and L.A. Stordeur. 1990. Responses of understory vegetation to thinning and fertilization of forests on Vancouver Island. B.C. Min. Environ. and Min. For., Victoria, B.C. IWIFR 37. Popper, K.R. 1959. The logic of scientific discovery. London, England: Hutchinson and Co. Ltd. 480 pp. Rahme, A.H. 1991. Visual cues and site selection by Mule Deer. M.Sc. Thesis. Simon Fraser University, Burnaby, B.C. 102 pp. Romesburg, H.C. 1981. Wildlife science: gaining reliable knowledge. J. Wild!. Manage. 45(2):293-313. Russell, J. 1979. Elk habitat inventory and enhancement consideration on lower Elk River (Strathcona Park), Vancouver Island. B.C. Fish and Wild!. Branch, Victoria, B.C. Samuel, M.D., and E.O. Garton. 1985. Home range: a weighted normal estimate and tests of underlying assumptions. J. Wildl. Manage. 49(2):513-519. Samuel, M.D., and E.O. Garton. 1987. Incorporating activity time in harmonic home range analysis. J. Widl. Manage. 51(1):254-257. Samuel, M.D., and R.E. Green. 1988. A revised test procedure for identifying core areas within the home range. J. Anim. Ecol. 57(3):1067-1068. Samuel, M.D., and K.P Kenow. 1992. Evaluating habitat selection with radiotelemetry triangulation error. J. Wildl. Manage. 4:725-784. Samuel, M.D., D.J. Pierce, and E.O. Garton. 1985. Identifying areas of concentrated use within the home range. J. Anim. Ecol. 54(3):711-719.  -116Schamberger, M.L. and L.J. O'Neil. 1986. Concepts and constraints of habitatmodel testing. In Wildlife 2000: modeling habitat relationships of terrestrial vertebrates. Edited by, J. Verner, M.L. Morrison, and C.J. Ralph. pp. 5-10. Schoener, T.W. 1981. An empirically based estimate of home range. Theoretical Pop. Bio. 20(3):281-325. Schwartz, J.E., and G.E. Mitchell. 1945. The Roosevelt elk on the Olympic Peninsula, Washington. J. Wildl. Manage. 9(4):295-319. Schwaller, M.R. and B.T. Dealy. 1986. Landsat and GIS: two friends grow up together In GIS applications: report part II. Journal of Forestry 84(9):40-41. Scott, B. 1979. The Vancouver Island wolf (Canis lupus crassodan): an initial study of food habits and social organization. M.Sc. Thesis. University of British Columbia, Vancouver, B.C. 109 pp. Shackleton, D.M. 1973. Population quality and bighorn sheep. Ph.D. Thesis. University of Calgary, Alberta. 227 pp. Sinclair, A.R.E. 1983. The functions of distance movements in vertebrates. In The ecology of animal movement. Edited by I.R. Swingland and P.J. Greenwood. Oxford University Press, Oxford. pp. 240-258. Skovlin, J.M. 1982. Habitat requirements and evaluations. In Elk of North America: ecology and management. Edited by J.W. Thomas and D.E. Toweill. Stackpole, Harrisburg, Pennsylvania. pp. 369-413. Spencer, W.D., and R.H. Barrett. 1984. An evaluation of the harmonic mean measure for defining carnivore activity areas. Acta. Zool. Fennica. 171:255-259. Springer, J.T. 1979. Some sources of bias and sampling error in radiotriangulation. J. Wildl. Manage. 43:926-935. Strathcona Park Elk Management Plan 1990. 1990. British Columbia Ministry of Lands, Parks, and Environment. South Coast Region, North Vancouver, B.C. Swihart, R.K., and N.A. Slade. 1985. Influence of sampling interval on estimates of home range size. J. Wildl. Manage. 49(4):1019-1025. Taber, R.D., K. Raedeke, and D.A. McCaughran. 1982. In Elk of North America: ecology and management. Edited by J.W. Thomas and D.E. Toweill. Harrisburg. Stackpole Books. pp. 279-300.  -117Thomas, J.W, H.Black Jr., R.J. Scherzinger, and R.J. Pederson. 1979. Deer and elk. In Wildlife habitats in managed forests -- the Blue Mountains of Oregon and Washington. Edited b_y J.W, Thomas. U.S.D.A. For. Serv. Agric. Handbk. No. 553. Washington, D.C. pp. 104-126. Tredger, C.D., J.R.L. Russell, and E.J. Karanka. 1980. Potential strategies for fish and wildlife habitat rehabilitation of the Elk River (Strathcona Park), Vancouver Island. B.C. Min. Environ., Victoria. B.C. Troyer, W.A. 1960. The Roosevelt elk on Afognak, Island, Alaska. J. Wild!. Manage. 24(1):15-21. United States Fish and Wildlife Service. 1981. Standards for the development of habitat suitability models. 103 ESM. Div. Ecol. Serv., U.S. Fish and Wildl. Serv., Washington, D.C. Walpole, R.E. 1982. Introduction to Statistics: third edition. Macmillan Publishing Co., Inc. New York. 521 pp. White, G.C., and R.A. Garrott. 1990. Analysis of wildlife tracking data. Academia Press, Inc. San Diego. 383 pp. Witmer, G.W., M. Wisdom, E.P. Harshman, R.J. Anderson, C. Carey, M.P. Kuttle, I.D. Lyman, J.A. Rochelle, R.W. Scharpf, and D. Smithey. 1985. Deer and elk. In Management of wildlife and fish habitats in forests of western ^Brown. Publ. No. R6-F&WL-192Oregon and Washington. Edited^E.R. b_y 1985. U.S.D.A. For. Serv., Pac. Northwest Reg., Portland, Oregon. Woods, J.G. 1991. Ecology of a partially migratory elk population. Ph.D. Thesis. University of British Columbia, Vancouver, B.C. 149 pp. Worton, B.J. 1987. A review of models of home range for animal movements. Ecological Modelling. 38(3/4):277-298. Worton, B.J. 1989. Kernel methods for estimating the utilization distribution in home range studies. Ecology. 70(1):164-168. Zar, J.H. 1984. Biostatistical Analysis: second edition. Prentice-Hall, Inc. Englewood Cliffs, New Jersey. 718 pp.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

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"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0086167/manifest

Comment

Related Items