Sheep on Ewin Ridge, February 2010; Photo: I. Teske

Habitat use, seasonal movements, and population dynamics of bighorn sheep in the Elk Valley

Prepared for:

BC Ministry of Forests, Lands and Natural Resource Operations

205 Industrial Road G. Cranbrook, BC V1C 7G5

and Teck Coal Limited

P.O Box 2003 Sparwood, BC V0B 2G0

Prepared by:

Kim G. Poole Aurora Wildlife Research 1918 Shannon Point Road, Nelson BC V1L 6K1 Tele. (250) 825-4063; e-mail: [email protected]

May 2013 Elk Valley bighorn sheep project

Executive summary Winter is a critical season for most mountain ungulates, which use a variety of strategies to cope with burial of preferred forage and increased cost of locomotion in snow. Rocky Mountain bighorn sheep (Ovis canadensis canadensis) inhabit the east side of the Elk Valley in southeastern British Columbia where forestry and 4 large, open-pit coal mines are in operation. Sheep in this area generally winter at high elevation on windswept, south-facing native grasslands, with some sheep also wintering on mine properties. Expansion of coal mining is proposed in portions of the valley which may result in direct loss of high-elevation winter range, which in some instances can result in the creation of winter habitat through reclamation of mine disturbance. Winter range may be the single most important factor limiting sheep populations in the area. The primary objectives of this study were to describe seasonal movements, winter habitat selection, and use of mine properties by this population. A concurrent companion study examined winter range plant communities and plant production, range condition, and winter diet. We obtained ~54,000 GPS locations from 41 sheep (19 ewes, 22 rams) between March 2009 and May 2011. Winter severity differed markedly between winter 2009-10 (very low snow) and winter 2010-11 (deep snow). Survival of collared sheep dropped from 0.93 (annual rate) during the first year to 0.78 during the second, more severe winter. Summer and winter range sizes did not differ between sexes, but was roughly one-third the size during winter 2010-11 (3.2 km2) compared with winter 2009-10 (9.5 km2). Most (79%) of the sheep monitored for a summer to winter season were migratory (non- overlapping seasonal ranges), and all non-migratory sheep – mostly ewes – were associated with the northern 2 adjacent coal operations. Fidelity to winter ranges among years was high and equal between sexes (88%); some segregation of ranges between sexes was apparent. Although differences among individuals and mine areas were apparent, use of mine properties by the population varied seasonally, and showed low use (~10–18%) between November-December and April, followed by increased use to peak at about 60–65% in September-early October. We used a 2-stage approach to examine habitat selection, by first modelling individuals using Resource Selection Function analysis (multivariate logistic regression) and then by averaging parameter estimates across individuals. We examined resource selection at 2 scales: winter use to home range and within the winter range. Selection at both scales was dominated by terrain variables, with slightly less influence by cover class variables. There were limited differences in use and selection between sexes. At both home range and winter range scales, wintering sheep were positively associated with high elevations, shorter distance to escape terrain, and warmer aspects (solar duration). Terrain ruggedness was not a strong variable in models. Relative to coniferous forests (strongly avoided by all individuals), high-elevation, native grasslands (combined with exposed lands for modelling) was the highest ranked cover class at both home and winter range scales. The industrial cover class (mine-altered properties) was ranked lower in importance compared with high-elevation grasslands. We used 2006-11 mine and government survey data for independent model validation; fit of the winter range model was very high (rs = 1.00). All mine properties showed areas of high probability winter range, yet only Greenhills Operations and the South Pit area of Elkview Operations had significant use during winter. Most sheep that used mine properties during winter used reclaimed habitats, primarily reclaimed spoils and pits.

Aurora Wildlife Research Page ii

Elk Valley bighorn sheep project

We identified lambing areas using collar data from 18 ewes and 31 ewe-years. Median date of lambing was 19 May in 2009 and 26 May in 2010. For both years combined, 60% of estimated births occurred between 19-27 May (full range 10 May – 4 June). Of 31 lambing areas, 13 (42%) were on active mine properties of the 3 northern mines. Fidelity to lambing areas was high. When compared to native habitats, lambing areas on mine properties were at significantly lower elevations, on shallower slopes, further from escape terrain, and with lower proportions of conifer, shrub, both types of grasslands, and rock-rubble cover classes, and higher proportions of industrial cover class. Ewes that lambed in native areas selected more strongly for warmer aspects and higher proportion of conifer cover and high- elevation grasslands than ewes that lambed in mine areas. Ewes that lambed on mine properties selected more strongly for more rugged slopes. Rutting occurred through the study area, primarily on winter ranges and mainly in native habitats; 28% of rutting locations were on mine properties. Fidelity to rutting areas varied between years. We conducted sheep surveys of the study area in February 2010 and 2011, using collars to obtain an indication of sightability. Accounting for a sightability of 0.82, 645 sheep (90% CI 580–772) were estimated within the study area in 2010. In 2011, lower sightability (0.77) and fewer sheep observed resulted in an estimate of 555 sheep (90% CI 485–687). Bighorn sheep utilize a range of native and mine-altered habitats to varying degrees at different times of the year. Use of mine properties by this population is high during the growing season, which may have contributed to the observed population increase since the late 1980s, likely aided in large part through reclamation. Greater winter severity appeared to result in reduced winter range size and increased mortality, attesting to the importance of this season to sheep ecology. Considering winter distribution and population counts, Ewin Ridge had by far the highest wintering population, and Gill Peak, Brownie, and Ewin Ridge had the highest densities of sheep. Management to enhance sheep habitat could include controlling forest encroachment on seasonal ranges and movement corridors. On mine properties, sheep habitat can be developed by providing escape terrain adjacent to high quality forage and considering landform design (i.e., steep south-facing aspects where wind modeling could predict snow free potential) as an integrated component of mine design when bighorn sheep are a target species for reclamation activity. However, successful use of these areas during winter likely depends on snow depths and forage availability. Further surveys could be conducted to examine the relative demographic fitness of sheep wintering on native versus mine- altered habitats. Balancing sheep and elk numbers on winter ranges will be an important component of wildlife and habitat management. Main winter ranges comprise 2.7% of the study area (4.3% of the merged annual sheep ranges), emphasizing the limited amount of occupied winter ranges within the landscape. In addition to sheep grazing pressure, the number of elk on native sheep winter ranges during both summer and winter may already be negatively influencing range quality in some locations and could negatively influence other ranges. High fidelity to winter ranges and the apparent influence of winter severity on survival suggest that disturbance to native winter range resulting from development should be minimized or be conducted in a manner that effectively manages and/or mitigates the impacts. Large scale removal of main native winter ranges would likely result in a population decline and should be avoided.

Aurora Wildlife Research Page iii

Elk Valley bighorn sheep project

Table of Contents Executive summary ...... ii Glossary and acronyms ...... 1 Introduction ...... 2 Study area ...... 6 Study design and methodology ...... 9 Bighorn sheep capture and collaring ...... 9 Collar data handling ...... 11 Fate and survival rates ...... 12 Seasonal ranges ...... 12 Movements ...... 13 Winter range fidelity ...... 13 Seasonal use of mine habitat ...... 14 Resource selection ...... 14 Habitat data and use ...... 14 Habitat selection ...... 16 Lambing areas ...... 18 Rutting areas ...... 19 Elk Valley East bighorn sheep surveys ...... 19 Sightability correction ...... 20 Survey study area correction ...... 21 Sightability model ...... 21 Population summary ...... 21 Elk surveys ...... 21 Results ...... 22 Captures ...... 22 Fate and survival rates ...... 22 Seasonal ranges ...... 24 Movements ...... 26 Winter range fidelity ...... 28 Seasonal use of mine habitat ...... 29 Resource selection ...... 30 Winter habitat use ...... 30 Winter habitat selection ...... 35

Aurora Wildlife Research Page iv

Elk Valley bighorn sheep project

Lambing areas ...... 42 Rutting areas ...... 45 Elk Valley East bighorn sheep surveys ...... 48 Sightability of collared sheep ...... 51 Population estimate using Aerial Survey ...... 53 Elk surveys ...... 54 Discussion...... 56 Survival ...... 56 Movements ...... 57 Resource selection ...... 63 Use of mine properties ...... 64 Lambing areas ...... 65 Rutting areas ...... 66 Elk Valley East bighorn sheep surveys ...... 66 Survey sightability ...... 70 Conclusions ...... 70 Management recommendations...... 71 Acknowledgements ...... 73 Literature cited...... 73 Appendix 1. Capture data for bighorn sheep, Elk Valley, May 2009 – October 2010...... 82

Aurora Wildlife Research Page v

Elk Valley bighorn sheep project

List of Tables

Table 1. Summary of snow depths (from snow pillow data1) during winters 2008-09 to 2010-11, Elk Valley, BC. “+” or “–“ represent percent deviation from average levels (since 1983)...... 7 Table 2. Description of variables examined in modelling habitat selection by bighorn sheep in the Elk Valley, southeastern British Columbia, 2009 to 2011...... 15 Table 3. Fate of bighorn sheep that have died or dropped collars, Elk Valley study area, February 2009 – March 2011...... 23 Table 4. Seasonal range size (km2) of bighorn sheep, summer 2009 to winter 2010-11, Elk Valley...... 25 Table 5. Mean home range (random locations) and winter 2009-10 and 2010-11 use values and 90th percentile ranges for topographic and cover class variables used by bighorn sheep, Elk Valley, winters 2009 to 2011. Sample size of sheep – home range: n = 31; winter 2009-10: n = 30; winter 2010-11: n = 28)...... 31 Table 6. Mean variable values of bighorn sheep wintering in the Elk Valley, 2009-10 and 2010-11, separated by mining (n = 23 sheep-winters) and native areas (n = 60 sheep-winters)...... 34 Table 7. Variables predicting winter habitat selection at the home range scale by bighorn sheep based on GPS collar data, Elk Valley, southeastern British Columbia. Presented are parameter estimates (Coefficient), standardized coefficients (Std. Coefficient) and 95% confidence intervals, relative importance (RI; based on AIC w), and the proportion of sheep that showed positive selection for that parameter. Winter 2009-2010 is coded as 0 and 2010-11 is coded as 1, thus negative interaction terms with year signify reduced selection for that variable in the deep snow winter (2010-11). Numbers with the strongest or significant selection or avoidance are bolded...... 36 Table 8. Variables predicting winter habitat selection at the winter range scale by bighorn sheep based on GPS collar data, Elk Valley, southeastern British Columbia. Presented are parameter estimates (Coefficient), standardized coefficients (Std. Coefficient) and 95% confidence intervals, relative importance (RI), and the proportion of sheep that showed positive selection for that parameter. Winter 2009-2010 is coded as 0 and 2010-11 is coded as 1, thus negative interaction terms with year signify reduced selection for that variable in the deep snow winter (2010-11). Numbers with the strongest or significant selection or avoidance are bolded...... 38 Table 9. Mean variable values within 100-m radius buffer zones of suspected bighorn sheep lambing areas, Elk Valley, 2009 and 2010, separated by mining (n = 13) and native areas (n = 18)...... 44 Table 10. Mean variable values of suspected bighorn sheep rut areas, Elk Valley, 2009 and 2010, separated by mining (n = 17) and native areas (n = 43)...... 46 Table 11. Elk Valley East bighorn sheep surveys, February 2010 and 2011...... 48 Table 12. Summary of bighorn sheep observations during surveys of the Elk Valley East, February 2010 and 2011...... 49 Table 13. Proportion (%) of ewes and rams observed on bighorn sheep winter ranges during February 2010 and 2011 surveys, Elk Valley, British Columbia. Class I rams were excluded from analysis. 51

Aurora Wildlife Research Page vi

Elk Valley bighorn sheep project

Table 14. Bighorn sheep population observations and estimates (± 90% CI) corrected for sightability for the Elk Valley East survey, 2010, during a winter of low snowfall. The collar study area was surveyed with the addition of the Crowsnest Pass and Deadman Pass areas...... 52 Table 15. Bighorn sheep population observations and estimates (± 90% CI) corrected for sightability for the Elk Valley East survey, February 2011, during a winter of high snowfall. The collar study area was surveyed with the addition of the Aldridge area...... 52 Table 16. Bighorn sheep population statistics for the Elk Valley East surveys, 2008, 2010, and 2011, based on the Idaho bighorn sheep sightability model (Unsworth et al. 1998)...... 53 Table 17. Ungulates observed during summer survey of high elevation bighorn sheep winter ranges, Elk Valley, 3-4 August 2011. WR is bighorn sheep winter range identified in this study...... 55 Table 18. Bighorn sheep observed, and lamb:ewe and ram:ewe ratios during aerial surveys from BC FLNRO data for the Elk Valley East population, 1981–2011. Early data provided by BC FLNRO (Teske and Forbes 2002, I. Teske, unpubl. data). Note that data for the Elk Valley East population prior to 1991 likely did not cover all winter ranges...... 67 Table 19. Bighorn sheep observed on winter ranges within the Elk Valley East population during aerial surveys, 1975 to 2011. Herd numbers were obtained from digital BC FLNRO data (I. Teske, BC FLNRO, unpubl. data) and checked spatially using GIS. Blank values signify no survey of that area conducted...... 68

Aurora Wildlife Research Page vii

Elk Valley bighorn sheep project

List of Figures

Figure 1. The Elk Valley bighorn sheep study area in southeastern British Columbia, 2009 to 2011. Mine areas refer to the current and past disturbance footprint...... 5 Figure 2. Snow water equivalents from Morrissey Ridge snow pillow station (2C09Q), winters 1983-84 to 2010-11. Calculated from the mean monthly snow pillow value for November to May...... 7 Figure 3. Winter Severity Index (Baccante and Woods 2008) from temperature and snowfall data from the Cranbrook airport (940 m elevation), winters 1975-76 to 2010-11...... 8 Figure 4. Bighorn sheep capture locations, Elk Valley, southeastern British Columbia, 2009 to 2011...... 10 Figure 5. Timing of mortalities of collared sheep in the Elk Valley study area, March 2009 to April 2011. Data do not include 2 mortalities that were likely capture related. Samples sizes of collared ewes ranged from 18–20 for most of the study except for the last 3 month (14-17), and of rams ranged from 17–20 for most of the study except for the last 6 month (10-15)...... 24 Figure 6. Frequency of overlap of winter ranges by collared bighorn sheep, Elk Valley, winters 2009-10 and 2010-11...... 25 Figure 7. Average daily elevation (7-day moving average) of collared bighorn sheep ewes and rams in the Elk Valley, British Columbia, March 2009 – May 2011...... 26 Figure 8. Average daily movement (m/hr) (7-day moving average) of collared bighorn sheep ewes and rams in the Elk Valley, British Columbia, March 2009 – May 2011...... 27 Figure 9. Movement paths of collared sheep constructed by joining sequential collar locations, Elk Valley, British Columbia, March 2009 – May 2011. Number of sheep and number of movement paths through low-elevation areas shown...... 28 Figure 10. Proportion (%) use of mine properties by collared bighorn sheep ewes and rams in the Elk Valley, British Columbia, March 2009 – May 2011. Data summed by week...... 29 Figure 11. Proportion (%) of collar locations on mine properties by individual female (xxF) and male (xxM) bighorn sheep in the Elk Valley, British Columbia, March 2009 – May 2011. Only those sheep monitored for ≥9 months are shown. FRO = Fording River; GHO = Greenhills; LCO = Line Creek; EVO = Elkview...... 30 Figure 12. Distribution of topographic and cover class variables for the random home range (RHR), random winter range locations for winter 2009-10 (RWIN1) and 2010-11 (RWIN2), and sheep use on winter ranges for winter 2009-10 (SWIN1) and 2010-11 (SWIN2), by female (F) and male (M) bighorn sheep, Elk Valley, 2009-2011 (open symbols; box and whisker plots with outliers [solid symbols]). Y-axis values: elevation (m); slope (%); terrain ruggedness (units); distance to escape terrain (m); solar (units); cover classes (scaled from 0–1). See Table 2 for description of acronyms...... 33 Figure 13. Relationship between relative probability of occurrence of sheep within winter range and elevation (a) and distance to escape terrain (b). Selection was scaled to 1 for plotting. Elevation and distance to escape terrain coefficients taken from variables predicting winter habitat selection at the winter range scale (Table 8); other variables held constant...... 39

Aurora Wildlife Research Page viii

Elk Valley bighorn sheep project

Figure 14. Bighorn sheep winter habitat predicted probability of occurrence at the home range scale, Elk Valley. Light to dark gradient depicts low to high habitat quality; the lowest 2 rankings were removed to enable background topography to show. GPS collar locations for sheep during winter 2009-10 (purple) and winter 2010-11 (yellow). Winter range names are shown, and mine properties are outlined in brown polygons...... 40 Figure 15. Bighorn sheep winter habitat predicted probability of occurrence at the winter range scale, Elk Valley. Light to dark gradient depicts low to high habitat quality; the lowest 2 rankings were removed to enable background topography to show. GPS collar locations for sheep during winter 2009-10 (purple) and winter 2010-11 (yellow). Winter range names are shown, and mine properties are outlined in brown polygons...... 41 Figure 16. Location of bighorn sheep lambing sites as determined from GPS collar data, Elk Valley, British Columbia, May 2009 (n = 15) and May 2010 (n = 16)...... 43 Figure 17. Lambing analyses individual means (open symbols; box and whisker plots with outliers [solid symbols]) for terrain and land cover parameters for available (1,000 m radius) and used (100 m radius) buffers in mine (n = 13) and native (n = 18) habitat, Elk Valley, 2009-2010. Y-axis values: elevation (m); slope (%); terrain ruggedness (units); distance to escape terrain (m); solar (units); cover classes (scaled from 0–100). See Table 2 for description of acronyms...... 45 Figure 18. Locations of collared bighorn sheep during the rut during 21-30 November 2009 and 2010, Elk Valley, British Columbia. Colours represent different sheep...... 47 Figure 19. Location of bighorn sheep observed during February 2010 and 2011 surveys of the Elk Valley East population. Most of the collar study area was surveyed both years, with additional surveys of the Crowsnest Pass and Deadman Pass areas in 2010, and the Aldridge Creek area in 2011. . 50 Figure 20. Mean numbers of elk, deer (mainly mule deer) and bighorn sheep observed on or near high elevation winter ranges identified by this study during annual surveys conducted by Teck Coal during mid-January to mid-March 2006–11 (L. Amos, Teck Coal, unpubl. data)...... 54 Figure 21. Bighorn sheep winter ranges as mapped from collar data 2009-11, 2010 and 2011 government survey data, and 2009–2011 Teck Coal survey data, Elk Valley East, British Columbia...... 59 Figure 22. Bighorn sheep density (sheep/km2) and mean count between 2010 and 2011 FLNRO surveys on winter ranges within the Elk Valley East, British Columbia...... 60 Figure 23. Groupings of collared bighorn sheep based on movement patterns, Elk Valley, 2009-2011. .. 62 Figure 24. Number of sheep observed during winter surveys in the main area of the Elk Valley East population (from Henretta to Sheep Mt. and including Greenhills), 1975–2011...... 69

Aurora Wildlife Research Page ix

Elk Valley bighorn sheep project

Glossary and acronyms

Term or acronym Definition or clarification Active workings Any place in a mine where miners are normally required to work or travel and which are ventilated and inspected regularly. Bench One of two or more divisions of a coal seam separated by slate or formed by the process of cutting the coal. CCR Coarse coal rejects, the “waste” product generated from processing coal DEM Digital elevation model EVO Elkview Operations Face The exposed area of a coal bed from which coal is being extracted. Footwall The interface between the base of a coal seam and the surrounding rock formation FRO Fording River Operations GHO Greenhills Operations GIS Geographic Information System GPS Global positioning system Heteroscedasticity A collection of random variables; often corrected in statistical comparisons. Highwall The unexcavated face of exposed overburden and coal in a surface mine or in a face or bank on the uphill side of a contour mine excavation. LCO Line Creek Operations LiDAR Light Detection and Ranging, an optical remote sensing technology MFLNRO Ministry of Forests, Lands and Natural Resource Operations Multicollinearity Where two or more predictor variables in a multiple regression model are highly correlated. Pit Used to describe the whole coal mine. Pseudoreplication Best described in this context where replicates (>1 sheep) are not statistically independent. Reclamation The restoration of land and environmental values to a surface mine site after the coal is extracted. RSF Resource selection function SE Standard error Spoil Waste rock dump; waste rock spoil; an accumulation of rock mined during the release of coal; a structure developed by depositing mined over burden

Aurora Wildlife Research Page 1

Elk Valley bighorn sheep project

Introduction Most mountain ungulates exhibit complex spatial and temporal use patterns, in large part because of the effects of winter and seasonality on forage production, availability and quality. Snow and winter weather influences mountain ungulates by increasing costs of thermoregulation and energetic costs of locomotion and burial of preferred forage species (Burles and Hoefs 1984, Parker et al. 1984, Daily and Hobbs 1989, Pauley et al. 1993). Ungulates that typically reside at high elevations cope with winter and potentially deep snow cover by using a variety of strategies. For example, Abruzzo chamois (Rupicapra pyrenaica ornata) selected wind-protected, very steep slopes, but moved to forested lower slopes when snow depth increased (Lovari and Cosentino 1986). Sympatric alpine chamois (R. rubicapra) and Himalayan tahr (Hemitragus jemlahicus) selected vegetated shrubland (both species) and rock bluffs (tahr) during winter (Forsyth 2000). Both Asiatic ibex (Capra ibex sibrica) and blue sheep (Pseudois nayaur) preferred habitat close to escape terrain and moderate slopes, but snow-free areas were not preferred (Namgail 2006). Mountain goats (Oreamnos americanus) utilized open, high-elevation habitats in shallow snow zones, but did not seek reduced snow levels in mature forest stands in deep snow areas (Poole et al. 2009). Mountain caribou (Rangifer tarandus) in some areas select shallow snow in thick cover in early winter for ground forage, then as winter progresses move up in elevation to areas of deeper snow for access to arboreal forage (Apps et al. 2001). In western North American, seasonal range selection by bighorn sheep (Ovis canadensis) varies among populations (Geist 1971; Festa-Bianchet 1986a, b; Shackleton et al. 1999; Demarchi et al. 2000; Krausman and Bowyer 2003). Habitat selection is thought to be a trade-off between nutritional and anti-predation constraints (Festa-Bianchet 1988a, Hebblewhite and Merrill 2011). Nearness to escape terrain is an important facet of most sheep habitat selection (e.g., Tilton and Willard 1982, DeCesare and Pletscher 2006, Bleich et al. 2009). The Elk Valley in southeastern British Columbia (BC) is home to a healthy population of Rocky Mountain bighorn sheep (O. c. canadensis). The Elk Valley sheep population is of provincial significance; the Ewin Ridge sheep range has been considered the most important bighorn sheep winter range in BC (Demarchi 1968). While sheep typically winter at lower elevations to avoid deep snows that accumulate in most parts of their range (e.g., Tilton and Willard 1982), sheep in the Elk Valley are unique in BC as they generally winter at high elevation on windswept, south facing native grasslands. Lower elevation habitat in this area is primarily forested with heavy snow accumulation. The Elk Valley supports important industrial activities, primarily coal mining and timber harvesting. Four large, open-pit coal mines occur within the east side of the Elk Valley north of Highway 3 (Fig. 1). Together these 4 operations have impacted 137 km2 of terrain (current and past disturbance), ranging from valley bottoms to mountain tops. Coal mines can create bighorn sheep habitat where none existed previously, as has occurred at mines in the Alberta foothills (MacCallum 1991, 1997; MacCallum and Geist 1992). At these Alberta mines, bighorn sheep from the adjacent occupied alpine habitat responded to the mine reclamation by occupying the new habitats (primarily as winter range, but also for lambing, summer range and during the rut), expanding their range, and exhibiting rapid population growth. These Alberta mines experience Chinook winds that melt snow at lower elevations and clear it from higher slopes, a phenomenon not found in the Elk Valley. The Greenhills Operation

Aurora Wildlife Research Page 2

Elk Valley bighorn sheep project population in the Elk Valley is another example of sheep utilizing new habitats during winter (this report). Bighorn sheep also utilize urban environments, which can provide higher diet quality (Rubin et al. 2002, Dibb et al. 2008). While human disturbance (e.g., noise, activity) from mines do not appear to deter use by sheep per se (Oehler et al. 2005, Jansen et al. 2006, Bleich et al. 2009), expansion of coal mining may occur in portions of the valley where direct loss of winter range could occur. Winter range is likely the most critical habitat for bighorn sheep (Demarchi et al. 2000), and has been suggested as the single most important factor limiting sheep populations in the area (Schuerholz 1984). The late winter/early spring (green-up) period is likely especially important, as delayed green-up may cause sheep to die after the long winter (Burles and Hoefs 1984). Incomplete knowledge about bighorn sheep ecology in the area of interest, including important winter use areas, habitat use and migration corridors, was identified as one obstacle to coordinating sheep management activities. Identified high value areas can be used to assess and manage the potential impacts of industrial activities. Previous radio-telemetry studies of sheep using standard VHF collars have been carried out in the area (Schuerholz 1984), but collaring was limited primarily to the Line Creek area, and was sporadic and incomplete; e.g., lambing grounds could not be determined from the data. Data on numbers and winter distribution of sheep are available. Independent winter wildlife surveys by the 4 coal mines in the region began in the early 1980s and are conducted annually. Also, BC Ministry of Forest, Lands and Natural Resource Operations (FLNRO; formerly Ministry of Environment), with funding from HCTF has conducted a valley-wide sheep survey in the Elk Valley East area approximately every 2 years since 2002. Lambing areas can be considered an important habitat to the population, as selection of lambing areas likely affect lamb survival (Festa-Bianchet 1988b). Lambing areas are generally located either on the winter range or in specific lambing ranges (Shackleton et al. 1999). Ewes usually seek steep, rugged terrain where they seclude themselves from other sheep for a number of days to give birth (Shackleton et al. 1999, Krausman and Boyer 2003). Escape terrain, which allows for predator avoidance, is especially important for ewes when giving birth. Forage quality and quantity at higher elevation often peaks later in the season, and ewes that move to high elevation for lambing may be compromising forage quality for decreased predation risk (Festa-Bianchet 1988b, Bleich et al. 1997). The addition of abundant and high-quality forage in proximity to escape terrain on reclaimed mine properties and post mining topographic features, coupled with possibly reduced predation risk closer to human activity (as predators may avoid areas of human activity), may influence habitat selection as well as ewe lambing strategies in altered habitats. In their status review of bighorn sheep in British Columbia, Demarchi et al. (2000) suggested a number of research needs, including habitat use patterns, movement and seasonal home range, impacts of open pit mining and effectiveness of mine reclamation. Here we report on the results of a research project conducted over a 27-month period from February 2009 to May 2011. Associated with this study, surveys of the Elk Valley East population were conducted in late February 2010 and 2011. The overall goal of this research project was to provide information to improve the management and conservation of bighorn sheep within the Elk Valley. Specific objectives were to:

Aurora Wildlife Research Page 3

Elk Valley bighorn sheep project

a. Describe and document movements and seasonal habitats – with a focus on winter range fidelity, habitat use and selection – of bighorn sheep occupying ranges along the east side of the Elk Valley from Highway 3 to the northern boundary of coal mines; b. Assess bighorn sheep seasonal use of mine sites; c. Examine lambing and rutting areas and contrast areas on and off mine properties; d. Estimate survival rates of collared ewes and rams; e. Use radio-collared sheep to quantify animal sightability using high-elevation habitats during winter surveys; Objectives related to winter range description and condition, standing crop production, and forage utilization and diet were addressed in a companion study headed by Clint Smyth (Summit Environmental Consultants Inc. 2012). The results of these 2 studies will be considered to provide recommendations regarding management of the health and size of the Elk Valley East bighorn population and habitats.

Aurora Wildlife Research Page 4

Elk Valley bighorn sheep project

Figure 1. The Elk Valley bighorn sheep study area in southeastern British Columbia, 2009 to 2011. Mine areas refer to the current and past disturbance footprint.

Aurora Wildlife Research Page 5

Elk Valley bighorn sheep project

Study area The overall study area is located in the Rocky Mountains of southeastern BC (ca. 50°0’ N, 114°45’ W; Fig. 1). The collaring study area focused on bighorn sheep populations wintering east of the Elk River from Highway 3 in the south to the upper reaches of the Fording River drainage in the north (Fig. 1). The approximately 830 km2 study area is within BC Wildlife Management Unit (WMU) 4-23, and overlaps into Alberta across the Continental Divide within Alberta WMUs 402 and 404. A number of sheep wintering areas occur within the study area. The larger aerial survey study area covered the Elk Valley East population, which includes the collaring study area and additional sheep within the Elk Valley East to the north (Aldridge, Tobermory) and south (Crowsnest Pass, Deadman Pass). Scattered herds of sheep occur west of the Elk River that were not covered during the 2010 or 2011 surveys. (Further reference to the “study area” refers to the collaring project study area.) Biogeoclimatic zones within the study area grade from Montane Spruce dry cool (MSdk) in valley bottoms, Engelmann Spruce–Subalpine Fir dry cool (ESSFdk, including subalpine parkland) at higher elevations, to small amounts of Alpine Tundra (AT) on ridge and mountain tops (Meidinger and Pojar 1991, Braumandl and Curran 1992). Krummholz subalpine fir (Abies lasiocarpa) with scattered whitebark pine (Pinus albicaulis) dominate the treeline, whereas closed to open stands of Engelmann spruce (Picea engelmannii), subalpine fir, lodgepole pine (Pinus contorta), and aspen (Populus tremuloides) are found at lower elevations. Cliffs of various sizes, and open scree slopes and meadows are located on most ridgelines and mountain tops. Sheep winter ranges are characterized by herb-rich, tall grass meadows that occur at higher elevation (most between 2,000–2,300 m elevation, but ranging from 1,800–2,500 m), and which retain low snow depths because of their south and southwest aspect and strong westerly winds (Summit Environmental Consultants Inc. 2012). Average mean daily temperatures at Sparwood, British Columbia, located at 1,140 m elevation in the bottom of the Elk River valley near the southern end of the study area, were –6.8°C and 15.4°C for January and July, respectively, and precipitation averaged 603 mm annually (Environment Canada 2011). On average 248 cm of snow falls at Sparwood annually. Colder temperatures and higher snowfalls would occur at higher elevations. Snow pillow data (snow water equivalent) were obtained for Morrissey Ridge, located at 1,966 m, 30 km southwest of the study area (http://bcrfc.env.gov.bc.ca/data/asp/realtime/index.htm). Snow water equivalent is significantly related to snow depth (r2 = 0.82; DelGiudice et al. 2001). Snow depths varied among winters during the study, with winter 2009-10 characterized by considerably lower snow depths, and winter 2010-11 with higher snow depths coupled with a very late melt (as a result of a cold and wet spring; Table 1). After generally deeper snow depths during the mid to late 1990s, most snow depths during the 2000s were low to moderate (Fig. 2). Snow depths during winter 2010-11 were the highest since 2001-02. A winter severity index that considered mean monthly temperature factored with total monthly snowfall (Baccante and Woods 2008) indicated that winter 2010-11 was 94% more severe than winter 2009-10, and the most severe winter since 1996-97 (nearest data from Cranbrook, 940 m elevation, 75 km southwest of the study area; Fig. 3).

Aurora Wildlife Research Page 6

Elk Valley bighorn sheep project

Table 1. Summary of snow depths (from snow pillow data1) during winters 2008-09 to 2010-11, Elk Valley, BC. “+” or “–“ represent percent deviation from average levels (since 1983).

Winter Early winter Mid-winter Late winter 2008-09 -- –10-15% –10-15%; normal melt 2009-10 Average snow depth –30-35% –10-15%, normal melt 2010-11 –5-10% +10-15% +50-80%; late melt 1 Data from station 2C09Q Morrissey Ridge, at 1,966 m elevation.

Figure 2. Snow water equivalents from Morrissey Ridge snow pillow station (2C09Q), winters 1983- 84 to 2010-11. Calculated from the mean monthly snow pillow value for November to May.

Aurora Wildlife Research Page 7

Elk Valley bighorn sheep project

Figure 3. Winter Severity Index (Baccante and Woods 2008) from temperature and snowfall data from the Cranbrook airport (940 m elevation), winters 1975-76 to 2010-11.

Potential predators of bighorn sheep in the study area include grizzly bear (Ursus arctos), black bear (U. americanus), cougar (Puma concolor), wolverine (Gulo gulo), wolf (Canis lupus), coyote (C. latrans), and golden eagle (Aquila chrysaetos). The area also supports a high density and diversity of other ungulates, including large numbers of elk (Cervus elaphus) and mule deer (Odocoileus hemionus), scattered mountain goat (Oreamnos americanus), and smaller numbers of moose (Alces alces) and white‐tailed deer (O. virginianus). Mule deer and especially elk were often observed during winter at mid- to high elevation on the same ranges utilized by sheep. The primary resource development activities within the study area are open pit coal mining and forestry. As noted, approximately 137 km2 of the study area (17%) is composed of mine-related infrastructure, pits, spoils, reclaimed areas, some logged areas, etc., from 4 operations: Elkview Operations (EVO) to the south (ranging from approximately 1,400–2,100 m elevation), Line Creek Operations (LCO) in the central area (1,500–2,200 m), Greenhills Operations (GHO; 2,000–2,150 m) and Fording River Operations (FRO; 1,600–2,200 m) to the north (obtained from “dissolveddisturbance” coverage obtained from Teck Coal, July 2012; Fig. 1). An additional approximately 180 km2 east of the Elk River is private land owned by Teck Coal. Back country roads to varying degrees of functionality occur within the study area, but vehicle traffic is generally light. Aircraft overflights are generally rare. Forestry activities are limited to lower to mid-elevations off main access roads.

Aurora Wildlife Research Page 8

Elk Valley bighorn sheep project

Study design and methodology

Bighorn sheep capture and collaring We deployed 40 GPS collars on bighorn sheep within the study area, spreading capture effort throughout our areas of interest (Fig. 4). Equal numbers of ewes and rams were collared. We attempted to deploy about 4 collars on each winter range (equal numbers of each sex if possible), with a greater emphasis on Ewin Ridge and Greenhills, and a slightly reduced emphasis on Elkview/Erickson Ridge based on relative herd numbers. Capture and handling protocol was conducted under BC Environment permit CB09-51173. Sheep were captured primarily by helicopter netgunning (Barrett et al. 1982). Once netted, we restrained the sheep using hobbles and blindfolds, and attached a radiocollar. A number of measurements were obtained in most cases: horn measurements (length and base), estimated age (using annuli on the horns and tooth eruption), total length, chest girth, and neck girth (Appendix 1). In addition, during initial capture efforts we sampled blood for pregnancy (progesterone levels) and selenium analysis, hair and a biopsy sample (ear punch) for DNA archives and genetic studies, and faecal pellets (for parasites and dietary analysis; see Summit Environmental Consultants Inc. 2012). To facilitate observation and identification from the ground, we generally placed a numbered, orange ear-tag on each animal.

Aurora Wildlife Research Page 9

Elk Valley bighorn sheep project

Figure 4. Bighorn sheep capture locations, Elk Valley, southeastern British Columbia, 2009 to 2011.

Aurora Wildlife Research Page 10

Elk Valley bighorn sheep project

Processing and release were accomplished as quickly as possible, generally within 15–20 minutes of netting. We avoided capturing rams with greater than ⅔ to ¾ curl because larger rams appear to have higher rates of capture myopathy after handling (H. Schwantje, BC Min. of Forest, Lands and Natural Resources Operations [FLNRO], pers. comm.). The GPS collars (model G2110B, Advanced Telemetry Systems, Isanta, Minnesota, USA) weighed approximately 450–500 g, and were equipped with a 12-channel Trimble GPS board. Fluorescent orange tape was affixed to the webbing to enhance sightability in the field. Collars were set to a 10- hour fix rate (2.4 locations/day) to allow >2 years of monitoring (2 full winters), and to allow complete sampling of the 24-hour period (by shifting fix attempt timing each day). The collars were equipped with motion-sensitive mortality sensors. Each collar was also equipped with a VHF transmitter. The collars were programmed to obtain a fix with 120-second maximum on time with no retry on failed attempts. The collars stored the following data: horizontal position in the WGS84 datum, altitude, date, time, satellite data, fix mode [2- or 3-dimensional (2D or 3D)], positional and horizontal dilution of precision (DOP), the time required to obtain a fix, ambient temperature, and tilt switch movements. An 810-day period was set for timed collar drop-off. We attempted to locate collared sheep monthly using a fixed-wing Super Cub to establish locations, monitor collars, and identify mortalities. Sheep locations were recorded using a hand held GPS. The general habitat and the presence of accompanying sheep were also recorded. We occasionally searched for collared sheep east of the Continental Divide into Alberta, when they could not be detected in the study area. Collars detected on mortality mode (rapid signal) were investigated as soon as possible, and the collar data downloaded. Collars were then refurbished and redeployed at the earliest opportunity. We redeployed collars opportunistically using ground and helicopter-based techniques. Ground-based capture involved darting with immobilizing drugs, and handled similarly as for helicopter netting. Sheep were less averse to human activity while in active mine areas, and could be approached for darting. We used 1.6–1.8 cc of Dormosedan/Telazol (5 cc of Dormosedan in 1 vial of Telazol), with 6 cc of atipamezole antidote to the Dormosedan given to aid recovery (as recommended by H. Schwantje, FLNRO, pers. comm.). We did not attempt to dart sheep after summer because of difficulties attaining full induction, especially with rams, possibly related to poor drug absorption related to good body condition and increased fat levels.

Collar data handling Prior to analysis we removed locations collected within 1 week of capture to reduce movements associated with capture-related disturbance and for the animals to acclimate to the transmitter (White and Garrott 1990, Morellet et al. 2009). No locations had a positional dilution of precision (a measure of location precision) >6.0, thus we did not correct for major outliers (D’Eon and Delparte 2005). We visually removed <10 locations considered errant outliers. Mean GPS location error was assumed to be 11 m, with 50% and 95% of locations within 6 and 31 m accuracy, respectively (D’Eon et al. 2002). Because of high fix success and location quality (see Results), we did not correct for habitat-induced GPS bias (Hebblewhite et al. 2007).

Aurora Wildlife Research Page 11

Elk Valley bighorn sheep project

Fate and survival rates To examine survival rates of sheep we used the staggered entry Kaplan-Meier survival estimator, calculated in program Ecological Methodology version 6.1. (Krebs 1999, Kenney and Krebs 2002). The Kaplan‐Meier formula used to calculate survival rate was:

S(t) = (1 – d1/r1)(1 – d2/r2) where:  S(t) is the probability of an individual in a population surviving t units of time from the beginning of the study;  d is the number of deaths at a point in time (subscripts 1 and 2 representing points in time);  r is the number of animals at risk at a point in time (Pollock et al. 1989) Capture-related deaths and those for which fate could not be determined (largely related to non- functional collars) were removed from the survival analysis. Since we had no indication that predation affected collar transmission or that long-distance dispersal occurred, and it appeared that rams were hard enough on the collars to break the VHF crystals (see Results), we assumed sheep were alive until censured when contact was lost, even if the collars were not retrieved by the end of the study. Survival rates were calculated separately for ewes and rams, and by year (beginning mid-May 2009 and mid- May 2010), and were presented as finite survival rates for the period of collaring, and converted to annual survival rates. We tested for differences between sexes and yearly differences using log‐rank tests (Garshelis et al. 2005).

Seasonal ranges Many populations of bighorn sheep are characterized by use of low-elevation ranges during winter, and migration to adjacent or distant alpine ranges during summer (e.g., Geist 1971, Festa-Bianchet 1988a, Demarchi et al. 2000). A maximum of 6 seasonal ranges may be used by rams and 4 by ewes (Geist 1971), but several of these ram ranges are occupied for short periods. We used changes in movement rates, elevation, and spatial distribution to define 4 broad seasons, with a focus on use of winter range and comparisons with summer range distribution. We examined each year independently because of differences in phenology and severity between winters. We averaged mean daily movement rates (distance moved/hour) and altitude among individual ewes and rams. The hourly movement rate for each location was calculated by dividing the distance moved by the time difference from the previous location; only locations with successful 10-hour fixes were used for calculations. Altitude (as recorded by the GPS collars) was selected from 3D locations only. We examined the pattern of movements plotted sequentially by date in ArcView (Environmental Systems Research Institute, Redlands, California, USA). Movement rate and changes in elevation by the population as a whole showed consistent annual patterns (see Results), but the extent and timing of seasonal movements was not fully synchronous among individuals. We set seasons based on population averages, and assigned winter as beginning 1 December 2009 and 15 December 2010 (generally low movement rates and stable use of elevation) and ending 30 April; spring as 1 May – 22 June (includes the bulk of lambing; use of low elevation [green-up]; increasing movement rates); summer as 23 June – 24 August (highest elevation use and

Aurora Wildlife Research Page 12

Elk Valley bighorn sheep project movement rates); and fall as 25 August – 30 November 2009/14 December 2010 (includes the rut in late November; variable use of elevation, declining movement rates [except for rams during the rut]). Where dates of individual arrival onto or departure from winter range did not fit these patterns, we modified season dates to accommodate these spatial shifts. Slightly later onset of winter in 2010-11 appeared to result in slower movement onto winter ranges compared with December 2009. To determine seasonal ranges we calculated 90% and 50% fixed kernels (Worton 1989; Seaman and Powell 1996, Girard et al. 2002) using the Home Range Extension (Rodgers and Carr 1998) for ArcView, with unit variance standardization, a user-defined smoothing factor of 0.70, and raster resolution set to 120. We defined a fixed smoothing parameter because least-square cross-validation produced under-smoothed data (Kie et al. 2010). Testing of data from various individuals suggested that the combination of a 90% kernel and 0.70 smoothing factor produced the most realistic depiction of seasonal ranges from bighorn sheep which often use the landscape in a clumped or linear fashion. We assumed 50% polygons identified core areas of intensive use (cf Kernohan et al. 2001). We used a 95% fixed kernel to define individual home ranges, using all locations for each individual sheep, as this best captured the extent and spatial pattern of annual sheep range use. We compared winter range sizes between sexes and years using t-tests on log-transformed data to correct for heteroscedasticity.

Movements We assumed that migration occurred if winter and summer ranges did not overlap (Brown 1992, Nicholson et al. 1997, Mysterud 1999) or if winter range encompassed <10% of summer range; we termed these sheep ‘migratory’. We termed sheep that did not migrate as ‘non-migratory’. Migration distance was defined as the horizontal distance between seasonal range centers of activity (centroids; Hayne 1949, Mysterud 1999, Petersburg et al. 2000), with the centroid calculated as the mean UTM coordinates of the locations for a sheep in each season using the Animal Movement Extension for ArcView (Hooge and Eichenlaub 2000). Migration distance between seasonal centroids was calculated between summer and winter for both years. Comparison of distances between summer and winter centroids was conducted using 2-tailed t-tests on log-transformed data. To examine overall movement patterns we joined sequential locations for each sheep and plotted the resulting line file. Movements through low-elevation habitat joining higher elevation habitat were documented for number of individuals and number of movements along that corridor. We did not consider movement corridors towards the Continental Divide because many of these occurred along higher elevations. Given the 10 hour fix rate, lines joining collar locations should be considered approximations of travel routes.

Winter range fidelity We examined winter range fidelity by determining the winter range sheep were located in during winters 2008-09 (as indicated by late February and March 2009), 2009-10, and 2010-11. We included sheep that were monitored but the collars were not recovered. We also summarized distances between consecutive winter use centroids.

Aurora Wildlife Research Page 13

Elk Valley bighorn sheep project

Seasonal use of mine habitat We used the “dissolveddisturbance”, “development”, “reclaimed”, and “facility” GIS layer files supplied by Teck Coal (D. Vasiga, Teck Coal, unpubl. data) to determine the proportion of sheep locations on mine properties. We removed “planned” (future) activities, and merged the layers to include all mine infrastructure, active mining areas, and reclaimed property, dated from July 2010 to January 2011. The proportion of locations on mine properties was summed weekly and plotted by sex to examine seasonal mine use. Individual variation also was examined by summing locations for each collared sheep.

Resource selection Habitat data and use To examine habitat use and selection we obtained digital data for habitat variables likely of importance to bighorn sheep during winter based on our observations and the literature (e.g., Festa-Bianchet 1988a, Sweanor 1996, MacCallum 1991, Dicus 2002, DeCesare and Pletscher 2006). We assembled the following digital geographic information system (GIS) databases: 1 m resolution LiDAR data flown in summer 2011 (D. Vasiga, Teck Coal, unpubl. data) covering the entire study area with the exception of a handful of locations in Alberta, where 1:50,000 scale data were used (Canadian Digital Elevation Data v1.1 2003; Natural Resources Canada 2007); land cover classification (Earth Observation for Sustainable Development of Forests – EOSD; Wulder et al. 2008); and mine disturbance footprints (D. Vasiga, Teck Coal, unpubl. data). The LiDAR data were degraded to 20 m resolution. From the digital data, we derived variables of interest covering terrain; slope was derived from triangulated irregular networks (TIN; 25-100 m on a side, with shorter sides in steeper or more rugged terrain) (Table 2). All databases were rasterized to ArcGIS 9.3 (Environmental Systems Research Institute, Redlands California, USA) grid format at 20 m resolution. We calculated terrain ruggedness using the vector ruggedness measure (VRM) of terrain, because it measures actual heterogeneity of terrain more independently of slope than other algorithms (Sappington et al. 2005). We calculated VRM using an ArcGIS script to calculate the 3-dimensional dispersion of vectors within the landscape at a 7 x 7 grid scale (script available online from the Environmental Systems Research Institute ArcScripts website: www.esri.com/arcscripts). We used computer-generated digital terrain modelling to calculate solar duration, a measure of solar radiation, which is the number of hours a pixel sees the sun per day based on latitude (sun angle) and the shading effects of nearby topography (Kumar et al. 1997). We used solar duration as a surrogate for aspect because of its strong correlation with aspect class (D’Eon and Serrouya 2005), because it has been proven a significant predictor of bighorn sheep habitat use (Dicus 2002, DeCesare and Pletscher 2006), and because it is a useful continuous variable for multivariate statistical applications. We calculated means for solar duration for 11 January to 31 March to match work completed elsewhere in the area (e.g., Poole et al. 2009). Although definitions of escape terrain for bighorn sheep vary (e.g., Tilton and Willard 1982, Sweanor et al. 1996, Singer et al. 2000, McKinney at al. 2003, DeCesare and Pletscher 2006), MacCallum (1991)

Aurora Wildlife Research Page 14

Elk Valley bighorn sheep project observed that bighorn sheep in the Alberta Rocky Mountains used steep terrain between 35° and 45° for escape, travel, bedding, and lambing. We subjectively used ≥37° (75%) slope to delineate escape terrain polygons, to ensure that mapped escape terrain was steep enough for bighorn sheep to avoid predators. From each grid cell we calculated the distance to the nearest escape terrain.

Table 2. Description of variables examined in modelling habitat selection by bighorn sheep in the Elk Valley, southeastern British Columbia, 2009 to 2011.

Category Acronym Variables Terrain variable ELEV Elevation (m) ELEV2 Squared term for elevation SLOPE Slope (%) VRM7X7 Index of ruggedness (Sappington et al. 2005) ESCPDST Distance to nearest escape terrain (m), defined as slope ≥75% SOLAR Solar duration value (hrs) Cover class1 CONIF Conifer-leading stands CUT Cutblocks, low to mid-elevation SHRUB >20% ground cover, with >33% shrub GHHIGH Grasses, forbs, graminoids; >33% herb; high zones GHLMID Grasses, forbs, graminoids; >33% herb; lower zones EXPLAND Barren, non-vegetated areas ROCRUB Bedrock, rubble, talus INDUST Disturbed property from mines OTHER Includes deciduous-leading stands and wetlands 1 Cover class modified from EOSD coverage (Wulder et al. 2008); see text.

EOSD data are derived from year 2000 classified Landsat-7 Enhanced Thematic Mapper Plus (ETM+) images with a pixel resolution of 25 m (Wulder et al. 2008). The original EOSD land cover data within BC were modified to remove shadows and small raster patches. The grid was run through a 3x3 majority filter 3 times and large shadow patches were inspected and coded by a vegetation specialist. Remaining small shadow patches were eliminated by merging them to the polygon with the longest shared boundary. Through visual inspection of ortho-imagery taken in summer 2010, wetland classifications were corrected, and forestry cutblocks and active mine boundaries were added. Additional classes were identified or modified based on data from the BC Freshwater Atlas (wetlands and waterbodies) (GeoBC 2010), Predictive Ecosystem Mapping (PEM: Resources Inventory Committee 1999), and Biogeoclimatic Ecosystem Classification (BEC) mapping (BC Ministry of Forests and Range - http://www.for.gov.bc.ca/hre/becweb/index.html). We used the “dissolveddisturbance”, “development”, “reclaimed”, and “facility” GIS layer files supplied by Teck Coal (as described above; D. Vasiga, Teck Coal, unpubl. data) to update the disturbed areas from the mines. The polygons were then transformed back into a raster grid. The original land cover classes in the study area were collapsed into 9 classes (Table 2) to remove very small classes and combine similar classes that would have little

Aurora Wildlife Research Page 15

Elk Valley bighorn sheep project impact on sheep. Original and unmodified EOSD data were used for areas or locations in Alberta (2.6% of the sheep locations). We used the digital data to characterize sheep habitat use within each full winter of monitoring (winters 2009-10 and 2010-11) within corresponding 90% fixed kernel winter ranges and 95% fixed kernel home ranges (see Seasonal ranges, above). Within each individual seasonal and home range we placed 1,000 random points, and calculated descriptive characteristics for each terrain and cover class variable. We determined the mean and 90th percentile range (i.e., lower 5% and upper 95% percentile values) for each terrain and cover class variable for each sheep based on these point data, and then calculated the mean of individual means and lower and upper limits to provide a sample mean and lower and upper limit for each variable. We used the 90th percentile as a measure of general use to avoid outliers and provide a more meaningful measure of the range of habitat use. We compared terrain and cover class variables among winter use and random home ranges using an analysis of variance (ANOVA) on log-transformed data to correct for heteroscedasticity; seasonal means were compared using Duncan’s multiple-range test. We also calculated mean and 90th percentile range values for terrain variables between sheep wintering on mine property and native ranges, and compared means using t-tests on log-transformed data. Within mine properties, we summarized cover type used by sheep: reclaimed lands, Habitat selection We considered that selection of home range within their geographic range (2nd order selection; Johnson 1980) would not be enlightening for this population of bighorn sheep because they select home ranges in areas with moderate to high elevation and generally open, rugged terrain within the landscape, and most valley bottoms and expanses of forest on non-steep slopes within the landscape are not truly “available” to bighorn sheep because of behavioural constraints (heavily forested and/or distant from steep escape terrain)(summarized in Shackleton et al. 1999, Krausman and Bowyer 2003). We examined winter habitat selection at 2 scales: 1) the home range scale (winter habitat use compared to random points within the home range) – assuming sheep winter where they choose within their home range – and 2) the winter range scale (winter habitat use compared with random points within the winter range) (3rd order selection; Johnson 1980). Both resource use and availability were identified by individual (type III study design; Manly et al. 2002). Comparisons between winters were conducted to examine the effects of differing snow accumulation (Table 1) on sheep habitat selection. Selection at the winter range scale was initially examined by sex (Geist 1971, Morgantini and Hudson 1981, Bleich et al. 1997, Shackleton et al. 1999). Although sheep likely select resources at different spatial-temporal scales, selection for resources at the winter range scale is important for predicting and understanding seasonal animal occurrences, and it is also at this scale (3rd order; stand-level) that land management decisions are frequently made (Nielsen et al. 2002). Here we define selection as the process in which an animal chooses a resource, and use the terms “selected” (use greater than availability) and “avoided” (use less than availability) to describe selection for or against a resource, respectively (Manly et al. 2002:3).

Aurora Wildlife Research Page 16

Elk Valley bighorn sheep project

We considered the individual collared bighorn sheep as the sample unit (Johnson 1980, White and Garrott 1990, Aebischer et al. 1993, Otis and White 1999). No 2 collared sheep shared the same herd year-round, therefore minimizing pseudoreplication (Hurlbert 1984, Millspaugh et al. 1998). Sample sizes of locations ranged from 275–325 per winter for most animals. We developed an a priori set of candidate models based on the literature and field observations to explain winter habitat selection at the 2 scales by bighorn sheep. We developed 3 model themes consistent with selection for topography and security (elevation, slope, terrain ruggedness, distance to escape terrain, and industrial cover (mine properties); Tilton and Willard 1982, MacCallum 1991, McKinney at al. 2003, Sappington et al. 2005), habitat and forage availability (cover type, especially open and grassland habitats, and solar duration; Festa-Bianchet 1988a), and thermoregulation (solar duration, conifer cover; Cook et al. 1998).We also examined several models that combined variables from themes. Linear and quadratic (squared) terms for elevation were considered in modelling to allow us to model for non-linear responses (Gross et al. 2002, Johnson et al. 2004). We included interaction terms with year (winter) with several variables because of hypothesized differences in selection between years due to major observed differences in snow depth. We evaluated multicollinearity among variables (lmer package in R; R Development Core Team 2008) with sheep as a random effect, which produces an output of correlations among fixed effects. However, we based the final coefficients on bootstrapped values of coefficients from individual sheep (see below). Between highly correlated pairs of variables (r > 0.7; Tabachnick and Fidell 1996) we chose the one with the greatest biological meaning according to the literature and our observations. We examined sheep habitat selection using multivariate logistic regression. Rather than using mixed effects models where individual animals are specified as a random effect (Gillies et al. 2006) we created a separate resource selection function for each individual. With mixed effects models the large number of data points per individual means that every variable is significant, revealing little about the strength among variables (preliminary analyses; R. Serrouya, unpubl. data). When models included the set of cover classes, we set the most abundant cover class (conifer-leading stands) as the reference category; conifer also was strongly avoided by all sheep (unpubl. data). Following suggestions by Anderson and Burnham (2002), we assessed the strength of competing models using Akaike’s Information Criteria (AIC) values; Anderson et al. 2000), differences in AIC values (ΔAIC), and Akaike weights (w). We calculated AIC weights for each variable to compare relative strength among variables (Burnham and Anderson 2002). Coefficients were weighted by w across all candidate models, and we determined the relative importance (RI) of individual variables based on AIC weights. Akaike’s Information Criteria values, ΔAIC and w for each model for each sheep are not shown (unpubl. data), but were required to calculate RI. We also calculated standardized coefficients (Menard 2004) that reflect the relative magnitude of each variable for each sheep. The modelling process was repeated for each sheep, but to obtain inference for the population we bootstrapped parameter estimates, weighted by the inverse of their unconditional (cross model weighted) standard errors, and presented mean values with 95% CIs using the percentile method. Relative importance for each variable was calculated by averaging across all sheep. Inconclusive statistical inference is expected from covariates with confidence intervals that overlap 0. To infer selection we used the RSF equation w(x) = exp(β1x1 +

Aurora Wildlife Research Page 17

Elk Valley bighorn sheep project

β2x2 … βnxn), where β1 . . . βn are coefficients and w(x) represents the relative probability of occurrence. RSF values range from 1 to infinity. At the home range scale, to evaluate the predictive accuracy of models we calculated the receiver operating characteristic (ROC) curve (Boyce et al. 2002) for each sheep and determined the mean ROC and corresponding 95% CIs. Models with a ROC value >0.9 were considered to reliably discriminate used from unused locations. Although this metric can be biased due to the use-available design employed (Boyce et al. 2002), biases would not compromise comparisons among models. For the winter range scale we combined 2 independent sources of data to validate models. The mines conduct aerial inventories for wildlife in the Elk Valley East each winter, using total counts and covering all areas within the BC portion of the study area (L. Amos, Teck Coal, unpubl. data). We used winter mine survey data between 2006 and 2011 (a recent period of complete coverage and uniform survey methods), covering 352 groups and totalling 2,588 individuals. Our 2nd source of model validation data was from BC government surveys of the Elk Valley East population conducted in 2008, 2010, and 2011, totalling 227 groups and 1,320 sheep (I. Teske, BC FLNRO, unpubl. data). We applied the averaged logistic regression model (i.e. parameters averaged across sheep) from the winter range scale and compared the fit of the data. We binned the validation data into 10 equal categories and summed the RSF within each bin. We then calculated the frequency of occurrence of the validation data (survey data) in each bin, which we divided by the spatial area of each category (i.e., the range of RSF values in each bin), estimated by theming the logistic regression model in a GIS. This is known as the area-adjusted frequency (Boyce et al. 2002), which was plotted on the y-axis. We then plotted the sum of each bin (x-axis) against the area-adjusted frequency for each bin (y-axis), and calculated the Spearman rank correlation between these 2 factors (sensu Boyce et al. 2002). Spearman values closer to 1 indicated a more robust model. Lambing areas Lambing sites are important to bighorn sheep populations, since both forage quality and reduced predation risk influence lamb health and survival (Festa-Bianchet 1988b, Bleich et al. 1997). Ewes may choose lambing sites at great distance from their winter range (Festa-Bianchet 1988a, MacCallum 1997), making sudden and rapid movements just prior to reduced movement for lambing and the immediate post-partum period (A. Dibbs, Park Canada, pers. comm.). We identified the lambing site for each ewe using changes in movement rates and spatial localization within the broader lambing period (early May to mid-June; cf Vore and Schmidt 2001, Poole et al. 2007). The mean UTM location was chosen based on the GPS locations at suspected parturition time. Since the location of actual lambing sites were not precisely known, we used a 100-m radius circle (3.14 ha) around the suspected lambing site, hereafter termed lambing area, to represent the habitat selected for lambing. To examine use of lambing areas within the broader landscape, we examined available habitat from a 1,000-m radius circle (314 ha) centred on the lambing site, hereafter termed extended lambing area, 100 times larger than the lambing area. We rationalized that ewes selected attributes of their lambing area from within their home range, and in the majority of cases the 1,000-m radius circle was entirely or mostly within each ewe’s home range which were much larger than 3 km2.

Aurora Wildlife Research Page 18

Elk Valley bighorn sheep project

For each lambing area and extended lambing area we determined the mean value of each topographic variable and the mean percent cover class. As described earlier, we used updated land cover classes using the mine coverages supplied by Teck Coal (D. Vasiga, Teck Coal, unpubl. data). Ewes lambed both on and off mine properties (see Results), and we presented data separated by mine or native lambing area. We compared individual variables between mine and native lambing areas using 2-tailed t-tests for unequal variance. We attempted to use a similar procedure as used in winter habitat selection to examine selection of habitat variables for lambing areas, contrasting selection between mine areas and native habitats. However, small sample size (n = 13–18) and extremes in binary land cover values resulted in even 3 or 4 parameter models not converging. We therefore displayed the importance of terrain and land cover variables and interpreted selection using box and whisker plots of available (1,000 m buffer extended lambing area) and use (100 m buffer lambing area) divided by mine and native areas. Rutting areas We mapped the distribution of suspected rutting areas. Because we were dealing with digital and not observational data, we used various sources to determine peak rut timing. Schuerholz (1984) and TAESCO (1985b) suggested that rutting peaked in November and occurred close to winter range. Ram daily movement rates were generally high from 1-30 November, but some of the earlier peaks in movements were likely related to pre-rut movements (Geist 1971). While most of the younger rams collared in this study may not have participated directly in the rut and may tend to roam more looking for breeding opportunities (Geist 1971), older rams and most ewes would presumably be in rutting areas in concentrated groups. Back-calculating 173-176 days (Krausman and Bowyer 2003) from assumed lambing dates places the peak of lambing during the last week of November. We therefore used the last 10 days in November (21-30 November) as the peak rutting period for mapping. We performed 2-tailed t-tests for unequal variance on individual variables between mine and native rutting areas. We conducted habitat analyses using R software (R Development Core Team 2008), and used SAS software (SAS Institute 2004) for all other analyses. Package MuMIn (Bartoń 2012) in R was used for model selection and model averaging within sheep. Unless otherwise noted, significant differences were assumed at α = 0.05; means were provided with associated standard errors (SE).

Elk Valley East bighorn sheep surveys BC FLNRO has conducted a valley-wide sheep survey in the Elk Valley East area sporadically since the mid-1970s, with fairly complete coverage of the central area every 1-3 years since 2002 (Teske and Forbes 2002, I. Teske, BC FLNRO, unpubl. data). These surveys have been conducted using relatively consistent methodology. They were flown by helicopter, generally during February or March when sheep are concentrated on their high-elevation winter ranges. During each survey a minimum count of sheep was obtained, and a sightability correction factor for the proportion of the population missed was applied to obtain a population estimate for the valley.

Aurora Wildlife Research Page 19

Elk Valley bighorn sheep project

A sightability of 80% has been assumed during recent valley-wide surveys carried out by BC FLNRO (I. Teske, BC FLNRO, pers. comm.), but this value has never been tested. In late February 2010 and 2011 we used the sample of collared sheep within the study area to help quantify sheep sightability (Bodie et al. 1995) to present some indication of the correction factor needed to provide more reliable population estimates. We conducted a fixed-wing relocation flight 3–5 days prior to each survey to simplify telemetry checks and obtain rough locations of collared sheep. Survey study design and methodology generally followed RISC standards (RISC 2002), and consisted of a total count survey. We established the census zone as the area within the Elk Valley East where we expected to observe wintering bighorn sheep, and covered from EVO and Erickson Ridge north to Henretta each year (Fig. 4). We surveyed using a Bell 206B or 206L helicopter. The pilot, front-seat navigator/recorder, and main rear-seat observer were the same between surveys. The second rear-seat observer was various environmental staff from Teck Coal who had varying degrees of experience at aerial surveys. We surveyed all high-elevation grasslands and known wintering areas for sheep. We flew roughly 125–150 m (400–500 foot) contour lines at 80–100 km/hr, 75–100 m from the hillsides. Animal locations and flight track were recorded with a hand-held GPS unit, which was later downloaded to a computer. We recorded elevation of groups observed from the helicopter’s altimeter (to the nearest 100 feet). We classified sheep to Level 4 classification (RISC 2002), which consisted of lambs, ewes, and Class I, Class II, Class III, and Class IV rams. During each survey the front-seat recorder employed blind telemetry (other crew members could not hear radio collar VHF signals) and monitored collar frequencies using a single H-antennae mounted on the helicopter. If a collar was observed in a group of sheep the survey was temporarily halted if needed to verify the collar frequency. Once an area (generally a winter range) was covered, any collars not identified during the regular survey of that area were located. We assumed that all marked (collared) and unmarked individuals within the population were independently distributed, had an equivalent probability of being observed, and that no errors were made differentiating marked and unmarked individuals. Sheep located outside of the census zone were censored for sightability correction, but were included for study area correction. We examined use of winter ranges between sexes for surveys from 2010 and 2011. We determined the proportion of ewes and rams on each winter range each year, and averaged the proportions between years. We removed Class I rams from the analysis because this age class may be associated with either nursery groups or rams groups (these surveys; Geist 1971). Sightability correction Following Heard et al. (1999), we estimated the fraction of collared sheep seen within the census zone to correct for sightability bias from the collars: p1 = m1/n1

Aurora Wildlife Research Page 20

Elk Valley bighorn sheep project

where m1 is the number of collared sheep seen by observers, and n1 is the number of collared sheep in the census zone. Because each collared sheep was an independent sample, the variance of p1 was based on the binomial distribution, var = pq/n (Sokal and Rohlf 1981). where q = 1-p Survey study area correction The proportion of collared sheep within the census zone was estimated as: p2 = m2/n2 where m2 is the number of collared sheep in the census zone, and n2 is the number of collared sheep in the Elk Valley East study area. The variance was also based on the binomial distribution. Population estimates were calculated using the joint hypergeometric estimator from program NOREMARK (White 1996). Sightability model To obtain data to fit into an existing sightability logistic regression model developed for California bighorn sheep in Idaho (Bodie et al. 1995, Unsworth et al. 1998), we classified groups when first observed for activity (moving or not moving) and broad habitat type. The only terms in the model were activity and broad habitat type (up to 8 habitat types could be recorded, which were collapsed into flats/open slopes or other habitats). Although not final parameters in the model, we also estimated percent snow cover in the general area and percent vegetation cover (perhaps best described as screening cover) around the first animal seen in the group (Unsworth et al. 1998). We estimated population size and sightability correction using the Idaho sheep model in program AERIAL SURVEY (Unsworth et al. 1998), and compared results from our collar sightability and AERIAL SURVEY for 2008, 2010 and 2011 surveys. Population summary In the discussion we provide a summary of sheep survey and ratio results since the mid-1970s and early 1980s, including observed counts by groupings of winter ranges. We used Kenney and Krebs (2002) to calculate average annual rate of increase.

Elk surveys Interspecific competition by elk, and to a lesser extent deer, may affect the capacity of the winter ranges to support bighorn sheep (Stelfox 1976). We examined elk numbers on winter ranges using 2 methods. First, we summarized elk, deer and sheep numbers from annual mine surveys conducted in mid-winter 2006-11 (L. Amos, Teck Coal, unpubl. data). Observations were extracted from a GIS database and were restricted to within 1 km of traditional sheep winter ranges. We also conducted a survey on 3-4 August 2011 of all winter ranges to enumerate ungulate use during summer. Surveys were conducted with 3 observers and a pilot in a Bell 206B helicopter from first light (6:30 am) to 11:00 am, ensuring complete coverage of all areas on and within 200-300 m of sheep winter ranges.

Aurora Wildlife Research Page 21

Elk Valley bighorn sheep project

Results

Captures We captured and fitted GPS collars on 39 bighorn sheep (19 ewes and 20 rams) in late February 2009, and subsequently re-deployed 11 collars (on 3 ewes and 8 rams) between late May 2009 and late October 2010 (Fig. 4; Appendix 1). Average age at collaring for ewes was 4.7 years (± 1.21 SD; range 3– 6 years; n = 22), and for rams was 2.9 years (± 1.04; range 1–5 years; n = 28). Re-deployment efforts were directed at rams as we became aware that proportionately more ram collars were dropping off or lost (see below). The ewe ages were likely an underestimate because annuli were less distinct in older animals; most (77%) were estimated to be 4 years or older. We obtained data from 41 collared sheep (19 ewes, 22 rams), monitored an average of 19.1 months (± 8.5; range 0.4–26.6 months). The final data set contained 54,304 locations (53,682 locations after censured locations were removed). GPS collar data quality was high, with 95.1% location success and 85.0% 3-dimensional fixes. Based on blood progesterone, 18 of 19 ewes (95%) 3+ years of age captured in February 2009 were pregnant. Serum selenium levels averaged 0.12 (± 1.21 SD ppm; n = 24), and did not differ between sexes (t = 1.19; P = 0.25).

Fate and survival rates Two mortalities occurred within 2 days of capture and handling, and were in all likelihood capture related and a result of changes in normal movements and behaviour (1 was caught in an avalanche, and 1 drowned after breaking through ice crossing a creek); these collars were censured from further analysis (Table 3). Three collars remotely released in mining areas were deemed too dangerous to collect from a personnel safety perspective and were left un-retrieved. The behaviour of the rams – fighting, butting heads – was detrimental to the collars. Five rams could not be located by the end of the project, and we suspect the VHF crystals in the collars were broken and not transmitting. In September 2011 we retrieved a non-transmitting collar from 1 of these rams and downloaded 8 months of data. Seven collars, all on rams, either pried off (n = 3) or prematurely released (n = 4) during the study, attesting to how hard rams were on these collars. No collared sheep were detected north or west of the study area or further than 1-2 km into Alberta during several long-distance scanning flights during and at the end of the field study.

Aurora Wildlife Research Page 22

Elk Valley bighorn sheep project

Table 3. Fate of bighorn sheep that have died or dropped collars, Elk Valley study area, February 2009 – March 2011.

Date dead or Sheep ID dropped Area Conclusion Comments 15F 26 Feb 09 Ewin Ridge Died in avalanche <6 hrs after Collar data censured capture; capture related 08M 28 Feb 09 Ewin Creek Drowned in pond 2 days after Collar data censured capture; capture related 04M 27 Apr09 Line Creek canyon above Natural mortality - unknown Mine property conveyor

41M 10 Jun 09 Castle Mt. – N end Natural mortality - grizzly bear Femur marrow was Chauncey predation firm, mostly white 38M 4 May 09 Greenhills South Collar fell/pried off sheep; sheep alive 39M 26 Sep 09 Greenhills South Collar fell/pried off sheep; sheep alive 24M 9 Dec 09 Ewin Ridge Collar released prematurely; sheep alive 19F 8 Dec 09 FRO Brownie free dump Natural mortality - unknown Mine property pile 28F 25 Dec 09 GHO South Natural mortality - unknown Mine property 44M 8 Jun 10 Horseshoe, E of LCO Collar fell/pried off sheep; sheep alive 30M 12 Oct 10 GHO South Collar released prematurely; sheep alive 34M 10 Nov 10 Avalanche chute on Mt. Collar released prematurely; sheep Lyne alive 20F 18 Jan 11 Pond in FRO active area Natural mortality - likely wolf Mine property; predation another ewe killed in same area 42F 28 Jan 11 Kilmarnock Ck., S side of Natural mortality - unknown Mine property FRO 14M 2 Feb 11 Todhunter Collar released prematurely; sheep alive 05F 13 Feb 11 Line Creek canyon Human-related mortality - likely Mine property vehicle collision 32F 28 Feb 11 Brownie Natural mortality - unknown 23F 6 Mar 11 FRO free dump area Natural mortality - unknown Mine property 35M 8 Mar 11 Castle Mt. Natural mortality - starvation Near 17M; 3rd uncollared ram dead in area 17M 13 Mar 11 Castle Mt. Natural mortality - starvation Near 35M; femur marrow red pasty

Aurora Wildlife Research Page 23

Elk Valley bighorn sheep project

Eleven sheep died during the study, 7 ewes and 4 rams (Table 3). Six mortalities (5 ewes, 1 ram) were attributed to unknown natural causes, 1 ram to grizzly bear mortality, 1 ewe to wolf mortality, and 2 rams to starvation. Only 1 known human-related mortality occurred, related to a vehicle strike on the Line Creek canyon road. Six of the mortalities occurred on mine properties (1 human-related, 5 natural causes). All but 1 mortality occurred during December to April (Fig. 5). We used 476 ewe-months and 443 ram-months of data for survival analysis. Finite survival rates were 0.66 (90% CI 0.49–0.83) and 0.75 (0.56–0.93) for ewes and rams, respectively, and did not differ between sexes (X2 = 0.43, 1 df, P > 0.05). Converted to an average annual survival rate, these figures were 0.83 (0.72–0.92) and 0.87 (0.77–0.97) for ewes and rams, respectively. Three mortalities occurred during mid-May 2009 – mid-May 2010 (survival rate 0.93 ± 0.85–0.99) and 7 during mid-May 2010 – mid-May 2011 (0.78 ± 0.66–0.90), but the difference in annual survival rates was not significant (X2 = 2.3, 1 df, P > 0.05) (Fig. 5).

3 Ewe Ram

2

1 Number of deaths ofNumber

0

Figure 5. Timing of mortalities of collared sheep in the Elk Valley study area, March 2009 to April 2011. Data do not include 2 mortalities that were likely capture related. Samples sizes of collared ewes ranged from 18–20 for most of the study except for the last 3 month (14-17), and of rams ranged from 17–20 for most of the study except for the last 6 month (10-15).

Seasonal ranges Range sizes did not differ between sexes for summers 2009 or 2010, or for winters 2009-10 and 2010- 11 (90% ranges, t < 1.37, 27-28 df, P > 0.18; 50% ranges t < 1.14, 27-28 df, P > 0.26). Sizes of summer ranges did not differ between years (90% ranges, t = 0.09, 58 df, P = 0.93; 50% ranges t = 0.09, 58 df, P = 0.93), but winter ranges were roughly one third the size during winter 2010-11 compared with the previous year (90% ranges, t = 2.59, 57 df, P = 0.01; 50% ranges t = 2.43, 57 df, P = 0.02) (Table 4).

Aurora Wildlife Research Page 24

Elk Valley bighorn sheep project

Table 4. Seasonal range size (km2) of bighorn sheep, summer 2009 to winter 2010-11, Elk Valley.

90% kernel 50% kernel Season n x SD SD Summer 2009 30 33.0 29.24 8.9 8.42 Summer 2010 30 37.0 43.18 9.3 10.04 Winter 2009-10 30 9.5 14.02 2.3 3.81 Winter 2010-11 29 3.2 3.15 0.8 0.88

Among collared sheep the greater overlap in frequency of winter range use occurred on Ewin Ridge, on ranges on the north side of FRO, and to a lesser extent on Sheep Mountain, Imperial Ridge, Todhunter, Chauncey, and GHO (Fig. 6). Much smaller winter range distribution was evident during 2010-11 compared with the previous winter (Fig. 6).

Figure 6. Frequency of overlap of winter ranges by collared bighorn sheep, Elk Valley, winters 2009- 10 and 2010-11.

Aurora Wildlife Research Page 25

Elk Valley bighorn sheep project

Movements The sheep population followed a fairly consistent pattern of use of elevation and changes in movement rates (Figs. 7, 8). Lowest elevations were used during late winter and spring (likely related to early green-up at lower elevations), and highest elevations during July and August. No low elevation spike was evident in spring 2009. Ewes used lower elevations than rams during fall. Sheep dropped to lower elevations earlier in late winter 2010-11 (Fig. 7), but significant movement off of traditional winter ranges was not detected by mid-May when collars were released. Movement rates were lowest during winter, and increased beginning in April-May to peak in late June to August (Fig. 8). Movement rates of ewes declined from late summer through to early winter, and rams increased movements during pre- rut and rut from mid-October through November.

2400 Ewe Ram

2300

2200

2100

Mean elevation (m) elevation Mean 2000

1900

1800

Figure 7. Average daily elevation (7-day moving average) of collared bighorn sheep ewes and rams in the Elk Valley, British Columbia, March 2009 – May 2011.

Aurora Wildlife Research Page 26

Elk Valley bighorn sheep project

120 Ewe Ram

100

80

60

40 Mean distance moved (m/hr)moveddistance Mean

20

0

Figure 8. Average daily movement (m/hr) (7-day moving average) of collared bighorn sheep ewes and rams in the Elk Valley, British Columbia, March 2009 – May 2011.

Most (79%) of the sheep monitored for a summer to winter session were migratory (both years considered; n = 53 sheep-years). A lower proportion of ewes were migratory (67%) than rams (96%). All of the non-migratory ewes resided on or near FRO (n = 8 sheep-years) or GHO (n = 2 sheep-years). The single male, a 2-year old ram at capture, that was non-migratory resided virtually year-round on GHO during winter 2010-11. Non-migratory sheep (10.5 ± 5.69 km; n = 6) had smaller individual ranges than migratory sheep (45.8 ± 38.68 km; n = 24; t = 2.05, 28 df, P < 0.001). For sheep that were migratory, distance between summer and winter centres of activity (centroids) did not differ between years for either ewes (t = 0.81, 18 df, P = 0.43) or rams (t = 0.25, 19 df, P = 0.80). Mean distance between centroids was greater for rams (7.7 ± 2.70 km) compared with ewes (5.8 ± 2.47 km; t = 2.25, 39 df, P = 0.03). We detected only 1 sheep, a 4-year-old ram at capture, which conducted extra-home range movements (movements beyond the normal range), a 3-week trip in June 2010 north to Tobermory (~38 km one way), and a 2.5 week trip from late November to mid-December 2010 east into Alberta (~20 km one way). The trip into Alberta crossed 8-10 km of relatively low elevation habitat to reach 2 fairly isolated alpine areas within the foothills. This animal returned to its normal home range on both occasions. Sheep movements followed high elevation mountain ranges and ridges (Fig. 9). No crossings of Alexander Creek to the east of Erickson Ridge were detected. Movements east to the Continental Divide were observed in the north end of the study area (from LCO north), but most occurred between

Aurora Wildlife Research Page 27

Elk Valley bighorn sheep project east of Imperial Ridge to east of Brownie. The largest number of individuals used lower-elevation corridors between Ewin Ridge and LCO (12 individuals; Fig. 9). Six individuals used areas between the north end of Erickson Ridge and Sheep Mountain, the Line Creek Canyon, and along the Ewin Ridge, Imperial Ridge, Todhunter, Chauncey, FRO corridors. Greatest use of corridors by number of movements occurred in the Line Creek Canyon, between Ewin Ridge and LCO, between Chauncey and FRO, and between FRO and GHO.

Figure 9. Movement paths of collared sheep constructed by joining sequential collar locations, Elk Valley, British Columbia, March 2009 – May 2011. Number of sheep and number of movement paths through low-elevation areas shown.

Winter range fidelity Fidelity to winter ranges among years was high. Of 25 sheep (15 ewes, 10 rams) where winter range was known in late February 2009, 2010, and 2011, only 3 did not use the same winter range each year. Two ewes used adjacent winter ranges in 1 of the years (Todhunter-Imperial, and Henretta-Turnbull), and a ram captured on Brownie subsequently wintered on Chauncey. Of 10 sheep (4 ewes, 6 rams) where winter range was known only in 2 winters, 6 used the same range in both years, a ewe moved from Brownie to Turnbull, a ewe moved from Ewin Ridge to Brownie/Turnbull (the longest shift in winter range within the study, ~21 km), a ram moved from Chauncey to Ewin Ridge, and another ram moved from Greenhills to Chauncey. On a sequential-year basis, this equates to 88% fidelity to

Aurora Wildlife Research Page 28

Elk Valley bighorn sheep project individual winter range (ewes 88%; rams 88%). Of the 7 cases that did not show fidelity, only 4 (7% of overall pairings) did not involve movements between adjacent winter ranges.

Seasonal use of mine habitat Sheep in the Elk Valley showed a pattern of low use (~10–18%) of mine properties between November- December and April, with increased use to peak at about 60–65% in September-early October (Fig. 10). Seasonal patterns of use were generally consistent between sexes and years. During the 3rd week of May 2009 and 2010 (roughly the peak of lambing) use by ewes of mine properties was roughly 50%, indicating ewes lambed within the confines of the properties. Differences among individuals and mine areas were apparent (Fig. 11). Sheep associated with the GHO property spent on average a larger proportion of their time on the property (65 ± 26.3%; n = 6) than sheep associated with FRO (27 ± 14.2%; n = 11), LCO (22 ± 10.7%; n = 10), and EVO (43 ± 9.4%; n = 2). Three sheep (2 ewes, 1 ram) spent 82–94% of their time on the GHO property, in essence almost year- round residency. Three sheep (2 ewes, 1 ram) also made almost no use of mine properties, occupying Imperial Ridge, Todhunter and east to the Divide. Differences in use of mine properties between sexes at some mines were also evident. Ewes on average spent a greater proportion of their time than rams on the FRO property (33 ± 10.8%, n = 8, vs 10 ± 4.6%, n = 3, respectively; t = 4.9, 8 df, P = 0.001) and on the GHO property (88 ± 8.3%, n = 2, vs 53 ± 23.9%, n = 4, respectively; t = 2.8, 4 df, P = 0.06), but not on LCO (23 ± 10.2%, n = 6, vs 20 ± 12.6%, n = 4, respectively; t = 0.5, 6 df, P = 0.64).

70

60

Ewes 50 Rams

40

30

20

Proportion (%) of locations on mine propertymine on locationsof (%) Proportion 10

0

Figure 10. Proportion (%) use of mine properties by collared bighorn sheep ewes and rams in the Elk Valley, British Columbia, March 2009 – May 2011. Data summed by week.

Aurora Wildlife Research Page 29

Elk Valley bighorn sheep project

100

90

80

70

60

50

40

30

20 Proportion (%) of locations on mine propertymine on locationsof (%) Proportion 10

0

22F 21F 33F 23F 19F 20F 32F 42F 28F 25F 05F 49F 06F 09F 10F 43F 01F 12F 13F

17M 18M 46M 37M 35M 27M 45M 47M 30M 34M 11M 44M 29M 03M 14M FRO GHO GHO/LCO LCO EVO NONE Sheep ID by mine area use

Figure 11. Proportion (%) of collar locations on mine properties by individual female (xxF) and male (xxM) bighorn sheep in the Elk Valley, British Columbia, March 2009 – May 2011. Only those sheep monitored for ≥9 months are shown. FRO = Fording River; GHO = Greenhills; LCO = Line Creek; EVO = Elkview.

Resource selection Winter habitat use Compared with home ranges, winter sheep locations were located at higher elevations (winter 2009-10 only), on warmer and steeper slopes, in more rugged terrain, and closer to escape terrain (Table 5, Fig. 12). Among main cover classes, winter sheep locations were more often found in high elevation grasslands (winter 2009-10 only) and exposed land, and less often found in coniferous forests, rock- rubble, and industrial lands compared with home ranges. Most (90%) winter sheep locations occurred at approximately 1,900–2,350 m elevation, on 35–105% slopes, and within 90–95 m of escape terrain. Overall sheep use of resources did not differ between winters, except for use of lower elevation, slightly less use of high elevation grasslands, and less use of rock-rubble during winter 2010-11 (deep snow) compared with winter 2009-10 (shallow snow). There were no differences in mean topographic values between ewes and rams within either winter (t-test, t < 1.95, P > 0.06). Among land cover variables, the only differences in use between sexes was ewes made significantly higher use of rock- rubble in both winters (2009-10: t = 2.89, P = 0.007; 2010-11: t = 2.23, P = 0.04), and during winter 2009-10 made lower use of logged habitats (t = 2.52, P = 0.02) and mid-elevation grasslands (t = 3.06, P = 0.005).

Aurora Wildlife Research Page 30

Elk Valley bighorn sheep project

Table 5. Mean home range (random locations) and winter 2009-10 and 2010-11 use values and 90th percentile ranges for topographic and cover class variables used by bighorn sheep, Elk Valley, winters 2009 to 2011. Sample size of sheep – home range: n = 31; winter 2009-10: n = 30; winter 2010-11: n = 28).

Home range Winter 2009-10 Winter 2010-11 90th percentile 90th percentile 90th percentile 1 Variable x 2 SE range3 SE range SE range Terrain Elevation (m)* 2029b 17.0 1644–2404 2165a 32.0 1929–2373 2090b 49.1 1870–2263 Slope (%)* 52b 1.4 8–99 69a 1.7 36–104 70a 2.2 34–105 Terrain ruggedness* 0.027b 0.0009 0.002–0.086 0.043a 0.0021 0.006–0.116 0.046a 0.0027 0.007–0.115 Dist. to escape terrain (m)* 90a 3.8 0–280 27b 2.0 0–95 27b 2.4 0–88 Solar duration (hrs)* 501b 7.8 152–696 622a 6.7 467–722 594a 17.2 457–710 Cover classes (%) Coniferous* 31a 2.1 12–49 5b 1.6 0–9 7b 2.1 0–31 Deciduous* 1a 0.1 0–2 0b 0.0 0–1 0b 0.0 0 Logged* 4a 0.6 0–12 2b 1.2 0–4 2b 2.1 0–2 Shrub* 3a 0.6 0–10 5b 1.5 0–20 3b 1.2 0–23 Wetlands* 1a 0.1 0–1 0b 0.1 0–1 1b 0.4 0–5 High elev. grasslands* 5ab 0.5 0–9 21a 2.4 0–40 20b 3.2 0–43 Medium elev. grasslands* 3a 0.4 0–9 2b 0.5 0–10 6b 1.4 0–18 Exposed land* 12b 1.6 2–30 43a 4.9 1–82 39ab 5.5 0–79 Rock-rubble* 11a 2.5 0–44 9b 2.5 0–37 5b 2.4 0–39 Industrial* 31a 3.7 1–73 13b 5.2 0–89 16b 6.2 0–97 1 Significant differences among season means (using log-transformed values; ANOVA, all: F > 3.7, 2,32 df, P < 0.01, are indicated with asterisks (*). Seasonal variables with different letters are significantly different (Duncan’s multiple range test, P > 0.05). 2 Mean values calculated as mean of means from individual sheep locations. 3 Upper and lower limits of 90th percentile ranges were calculated as mean upper and lower limits of 90th percentile ranges from individual sheep.

Aurora Wildlife Research Page 31

Elk Valley bighorn sheep project

Figure 12. Distribution of topographic and cover class variables for the random home range (RHR), random winter range locations for winter 2009-10 (RWIN1) and 2010-11 (RWIN2), and sheep use on winter ranges for winter 2009-10 (SWIN1) and 2010-11 (SWIN2), by female (F) and male (M) bighorn sheep, Elk Valley, 2009-2011 (open symbols; box and whisker plots with outliers [solid symbols]). Y- axis values: elevation (m); slope (%); terrain ruggedness (units); distance to escape terrain (m); solar (units); cover classes (scaled from 0–1). See Table 2 for description of acronyms.

Aurora Wildlife Research Page 32

Elk Valley bighorn sheep project

Figure 12. Distribution of topographic and cover class variables for the random home range (RHR), random winter range locations for winter 2009-10 (RWIN1) and 2010-11 (RWIN2), and sheep use on winter ranges for winter 2009-10 (SWIN1) and 2010-11 (SWIN2), by female (F) and male (M) bighorn sheep, Elk Valley, 2009-2011 (open symbols; box and whisker plots with outliers [solid symbols]). Y- axis values: elevation (m); slope (%); terrain ruggedness (units); distance to escape terrain (m); solar (units); cover classes (scaled from 0–1). See Table 2 for description of acronyms.

Aurora Wildlife Research Page 33

Elk Valley bighorn sheep project

Compared with native areas, sheep wintering on mine properties were at lower elevation, on less steep slopes and less rugged slopes, on cooler aspects, and further from escape terrain (Table 6). Use of aspect class differed between mine and native wintering locations (Gadj = 1217, P < 0.001), with less use of hot aspects by sheep using mine properties (57%) compared with native ranges (68%) but greater use of warm aspects (24% versus 6%, respectively). On mine property during winter, sheep primarily used reclaimed habitats and spoils, with lower use of pits and highwalls.

Table 6. Mean variable values of bighorn sheep wintering in the Elk Valley, 2009-10 and 2010-11, separated by mining (n = 23 sheep-winters) and native areas (n = 60 sheep-winters).

Mine Native 90th 90th 1 Variable x 2 SE percentile 2 SE percentile range3 range Topographic Elevation (m)* 1839 50.7 1340–2096 2156 24.3 1779–2355 Slope (%)* 48 2.4 31–65 72 1.1 61–88 Terrain ruggedness* 0.035 0.0032 0.020–0.058 0.046 0.0017 0.027–0.070 Dist. to escape terrain (m)* 51 4.3 18–79 23 1.4 8–44 Solar duration (hrs)* 563 25.2 256–663 608 8.8 500–675 Mine cover classes (%) Reclaimed lands 35 6.0 0–84 Spoil 29 5.5 0–74 Pit 8 2.4 0–31 Highwall 8 2.7 0–32 Footwall 1 0.3 0–3 Roads 1 0.6 0–4 Undifferentiated 18 3.5 0–47 1 Significant differences (t-test, P < 0.05) between native and mine means are indicated with asterisks. 2 Mean values calculated as mean of means from individual sheep locations. 3 Upper and lower limits of 90th percentile ranges were calculated as mean upper and lower limits of 90th percentile ranges from individual sheep.

Aurora Wildlife Research Page 34

Elk Valley bighorn sheep project

Winter habitat selection

Slope and distance to escape terrain were correlated (rs = –0.81), and we selected distance to escape terrain as the most causative variable for modelling (cf Sweanor et al. 1996, Dicus 2002, DeCesare and

Pletscher 2006). Exposed land was correlated with several land cover variables (rs = 0.75–0.79). Since exposed land and high-elevation grasslands often occurred simultaneously on several of the sheep winter ranges, we combined these 2 variables for modelling. At the home range scale, variables representing topographic/security and habitat/forage themes were most selected during winter (Table 7). The final model had excellent predictive accuracy (ROC median = 0.94, 95% CI 0.93–0.95). Elevation, distance to escape terrain, solar duration and high elevation grasslands/exposed lands were the strongest variables, and the majority of sheep showing consistent positive (elevation, solar, and grasslands) or negative selection (distance to escape terrain). Essentially all sheep selected for higher solar duration and shorter distance to escape terrain during winter at the home range scale. The negative quadratic term for elevation (ELEV2) indicated selection for mid- elevations within the home range. Selection for terrain ruggedness was positive but showed inconsistent selection among sheep. Among land cover variables, high-elevation grasslands/exposed lands were highest ranked and had the greatest positive selection by sheep. Mid-elevation grasslands also were selected by most sheep but were much lower in relative ranking. Rock-rubble cover showed moderately high ranking, but was equally selected and avoid by individual sheep. Industrial cover had low ranking and was equally selected and avoided by individual sheep. Except for distance to escape terrain which showed selection for closer to escape terrain during the high snow year, all terms with year interactions had confidence intervals that overlapped 0, and all had low relative importance; to simplify interpretation (Murtaugh 2007) we therefore removed year interaction terms from further mapping.

Aurora Wildlife Research Page 35

Elk Valley bighorn sheep project

Table 7. Variables predicting winter habitat selection at the home range scale by bighorn sheep based on GPS collar data, Elk Valley, southeastern British Columbia. Presented are parameter estimates (Coefficient), standardized coefficients (Std. Coefficient) and 95% confidence intervals, relative importance (RI; based on AIC w), and the proportion of sheep that showed positive selection for that parameter. Winter 2009-2010 is coded as 0 and 2010-11 is coded as 1, thus negative interaction terms with year signify reduced selection for that variable in the deep snow winter (2010-11). Numbers with the strongest or significant selection or avoidance are bolded.

Coefficient Std. Coefficient % sheep 2 Variable1 β 95% CI β 95% CI RI selected (n) ANALYR -0.79 -2.64–0.73 -0.81 -2.67–0.76 1.00 42 (13/31) Terrain ELEV 0.037 0.016–0.062 18.4 8.1–30.7 1.00 84 (26/31) ELEV_YR 2.92 -9.70–17.03 3.73 -12.19–21.24 0.86 52 (16/31) ELEV2 -0.000009 -0.00002– -0.00004 -18.1 -29.6– -8.0 1.00 19 (6/31) ELEV2_YR -3.78 -17.82–8.46 -4.95 -22.33–11.96 0.86 48 (15/31) VRM7X7 4.67 1.45–8.15 0.33 0.12–0.56 0.74 68 (21/31) ESCPDST -0.015 -0.018– -0.013 -2.59 -3.23– -2.14 1.00 0 (0/31) ESCAPE_YR -0.0039 -0.0073– -0.0005 -0.51 -1.00–-0.03 0.86 39 (12/31) SOLAR 0.0065 0.0055–0.0075 2.11 1.74–2.51 1.00 97 (30/31) SOLAR_YR -0.0002 -0.0018––0.0015 -0.10 -1.06–0.98 0.48 48 (15/31) Cover class2 GHHIGH3 2.14 1.85–2.42 1.93 1.59–2.26 1.00 100 (31/31) GHLMID 1.77 1.37–2.17 0.61 0.44–0.80 1.00 83 (24/29) SHRUB 1.50 1.17–1.85 0.51 0.33–0.71 1.00 73 (22/30) ROCRUB 1.43 0.87–1.89 0.99 0.61–1.32 1.00 48 (12/25) CUT 1.44 0.74–2.11 0.53 0.17–0.94 1.00 30 (8/27) OTHER 1.09 0.66–1.50 0.24 0.15–0.33 1.00 48 (12/25) INDUST 0.88 0.19–1.56 0.58 0.05–1.11 1.00 45 (14/31) CONIF 0 0 0 (0/31) 1 See Table 2 for description of acronyms. 2 Conifer is the reference value for all other cover classes (conifer was avoided by all sheep), thus coefficient values and whether sheep selected the cover class was relative to selection of conifer. 3 GHHIGH includes EXPLAND.

Aurora Wildlife Research Page 36

Elk Valley bighorn sheep project

At the winter range scale ewes and rams had similar selection of variables, with the exception of terrain ruggedness, where rams showed negative selection and ewes showed positive selection (Fig. 12). However, the importance of this variable was weak, and thus would have little influence on modelled output. Among cover classes, ewes showed negative selection and rams showed positive selection for logged habitats, but again at the winter range scale would have little bearing on outputs because of limited mid-elevation logged habitat in and near winter ranges (Table 5, Fig. 12). Essentially the only winter use of cutblocks was by 3 rams in 2010 on the lower slopes of the Chauncey winter range. We therefore combined sexes for modelling and mapping. At the winter range scale, sheep selected for higher elevations and warmer aspects (higher solar duration), and selected areas closer to escape terrain (Table 8). The negative quadratic term for elevation (ELEV2) indicated selection for mid-elevations within the winter range with a peak selection at about 2,150 m (Fig. 13). Selection for distance to escape terrain was highest immediately adjacent to escape terrain, dropped sharply to about 100 m distance and by 200 m was only 10% of the highest selection (Fig. 13). Terrain ruggedness did not factor as a strong and consistent variable, with equal selection and avoidance among individuals. There was little importance of year in habitat selection at the winter range scale; we removed these variables from further mapping. All cover classes showed positive selection. High elevation grasslands (combined with exposed land) and rock-rubble (where available) were selected by all sheep. The industrial cover class was selected by three-quarters of the sheep. The linear and quadratic terms for elevation were the most influential of these variables, with distance to escape terrain, solar radiation, high elevation grasslands, and rock-rubble next in importance. The Spearman correlation using the independent mine census and government survey data of the winter range model was very high (rs = 1.00), indicating a very robust model. Between home range and winter range scales, selection of the stronger and highest ranked model variables remained consistent, with slightly less pronounced selection for topographic variables at the winter range scale. At both scales selection for high-elevation grasslands/exposed lands and rock- rubble cover classes were consistent. Selection for mine properties (industrial cover class) was ranked higher at the winter range scale compared with the home range scale.

Aurora Wildlife Research Page 37

Elk Valley bighorn sheep project

Table 8. Variables predicting winter habitat selection at the winter range scale by bighorn sheep based on GPS collar data, Elk Valley, southeastern British Columbia. Presented are parameter estimates (Coefficient), standardized coefficients (Std. Coefficient) and 95% confidence intervals, relative importance (RI), and the proportion of sheep that showed positive selection for that parameter. Winter 2009-2010 is coded as 0 and 2010-11 is coded as 1, thus negative interaction terms with year signify reduced selection for that variable in the deep snow winter (2010-11). Numbers with the strongest or significant selection or avoidance are bolded.

Coefficient Std. Coefficient % sheep 2 Variable1 β 95% CI β 95% CI RI selected (n) ANALYR -0.20 -1.61–1.44 0.17 -1.87–1.67 0.96 52 (13/25) Terrain ELEV 0.027 0.013–0.041 12.2 6.7–17.9 0.98 85 (28/33) ELEV_YR -4.84 -12.96–5.10 -4.66 -14.18–7.26 0.83 36 (9/25) ELEV2 -0.000006 -0.000010– -0.000003 -11.7 -17.3– -6.1 0.98 18 (6/33) ELEV2_YR 4.58 -5.14–12.24 4.90 -6.15–14.19 0.83 64 (16/25) VRM7X7 0.57 -1.64–2.86 0.04 -0.15–0.24 0.77 52 (17/33) ESCPDST -0.012 -0.014– -0.010 -1.61 -2.02– -1.23 1.00 9 (3/33) ESCAPE_YR -0.0025 -0.0062–0.0014 -0.30 -0.67–0.02 0.85 40 (10/25) SOLAR 0.0054 0.0046–0.0062 1.84 1.52–2.20 1.00 97 (32/33) SOLAR_YR -0.0009 -0.0024–0.0009 0.12 -0.76–1.08 0.62 48 (12/25) Cover class2 INDUST 1.62 1.27–1.95 0.77 0.51–1.07 1.00 78 (21/27) GHHIGH3 1.60 1.41–1.80 1.57 1.34–1.81 1.00 100 (33/33) ROCRUB 1.52 1.18–1.85 1.03 0.70–1.37 1.00 100 (14/14) CUT 1.52 0.82–2.18 0.75 0.28–1.26 1.00 33 (7/21) GHLMID 1.48 1.18–1.80 0.68 0.51–0.87 1.00 81 (26/32) SHRUB 1.41 1.10–1.70 0.52 0.36–0.69 1.00 75 (24/32) OTHER 1.13 0.75–1.45 0.25 0.16–0.31 1.00 42 (13/31) CONIF3 0 0 0 (0/33) 1 See Table 2 for description of acronyms. 2 Conifer is the reference value for all other cover classes (conifer was avoided by all sheep), thus coefficient values and whether sheep selected the cover class was relative to selection of conifer. 3 GHHIGH includes EXPLAND.

Aurora Wildlife Research Page 38

Elk Valley bighorn sheep project

(a) (b)

1.0 1.0

0.9 0.9

0.8 0.8

0.7 0.7

0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2 Relative probability of occurrenceof probability Relative 0.1 0.1

0.0 0.0 1400 1600 1800 2000 2200 2400 2600 2800 3000 0 100 200 300 400 Elevation (m) Distance to escape terrain (m)

Figure 13. Relationship between relative probability of occurrence of sheep within winter range and elevation (a) and distance to escape terrain (b). Selection was scaled to 1 for plotting. Elevation and distance to escape terrain coefficients taken from variables predicting winter habitat selection at the winter range scale (Table 8); other variables held constant.

Although the modelling was primarily conducted to describe winter habitat selection, it can also be used to map and predict winter range. Modelled winter habitat was generally similar at both home range and winter range scales, and sheep winter collar locations were mainly located in highly ranked habitats (Figs. 14, 15). The main exceptions were sheep winter locations in the Line Creek Canyon (mainly winter 2010-11) and the Elkview South Pit (both winters) that were not within highly ranked habitats. Some areas with few collar locations also mapped as high habitat quality, including along the Continental Divide. However, areas of the Divide north of Imperial Ridge were used by sheep during winter 2009-10, and mine and government surveys have observed sheep in many of these high- elevation areas.

Aurora Wildlife Research Page 39

Elk Valley bighorn sheep project

Figure 14. Bighorn sheep winter habitat predicted probability of occurrence at the home range scale, Elk Valley. Light to dark gradient depicts low to high habitat quality; the lowest 2 rankings were removed to enable background topography to show. GPS collar locations for sheep during winter 2009-10 (purple) and winter 2010-11 (yellow). Winter range names are shown, and mine properties are outlined in brown polygons.

All mine properties showed areas of high probability winter habitat, yet only GHO and the South Pit area of EVO had significant use during winter, with some use of the FRO property also observed (Fig. 15). Most mine areas except GHO were at lower elevations than native winter ranges. Mean elevation of sheep winter use of mine properties varied from 1,575 m at EVO to 1,930 m at FRO and 2,060 m at GHO. When on mine sites, most sheep used reclaimed habitats during winter (92%, 42% and 28% of all collar locations on EVO, GHO and FRO, respectively), with underlying habitat dominated by spoils and pits. Almost all winter sheep locations at EVO were located on reclaimed habitats adjacent to the South Pit and South Dumps. On GHO, a large number of locations were at or near the interface between spoils and pits. Limited use of highwalls was detected, however, highwall use was observed off the west side of Turnbull in FRO and within GHO.

Aurora Wildlife Research Page 40

Elk Valley bighorn sheep project

Figure 15. Bighorn sheep winter habitat predicted probability of occurrence at the winter range scale, Elk Valley. Light to dark gradient depicts low to high habitat quality; the lowest 2 rankings were removed to enable background topography to show. GPS collar locations for sheep during winter 2009-10 (purple) and winter 2010-11 (yellow). Winter range names are shown, and mine properties are outlined in brown polygons.

Aurora Wildlife Research Page 41

Elk Valley bighorn sheep project

Lambing areas We obtained collar data from the lambing period for 18 ewes and 32 ewe-years. One ewe captured in winter 2008-09 was not pregnant (according to blood progesterone levels) and did not localize. Of the remaining 31 ewe-years of data, all appeared to localize within the broader lambing period. Twenty- four showed a good pattern of reduced movement rates and localization, and 7 had poorer data where the timing and location of lambing was more difficult to determine. Ewes generally localized from 2-5 days (most 3-4 days), but 1 ewe remained in the suspected lambing area for 14 days. Lambing was sometimes preceded by rapid movement. Mean (21 May) and median (19 May) estimated dates for lambing during 2009 (n = 15) were 5-7 days earlier than 2010 (25 May and 26 May, respectively, n = 16). For both years combined, 60% of estimated births occurred between 19-27 May (full range 10 May – 4 June). Of 31 lambing areas, 13 (42%) were on disturbed mine properties; 8 in the FRO/GHO area (6 ewes) and 5 on LCO property (3 ewes; Fig. 16). Sheep moved from native habitats to lamb on mine properties on 11 of 13 occurrences, moving on average 6.7 km (±0.6). On mine properties ewes appeared to lamb primarily on spoils, highwalls, and pits; reclaimed habitats were essentially not used. Areas of concentration of lambing in native habitats were Gill Peak north of FRO (7 lambings from 4 ewes), and the upper Chauncey Creek drainage (4 lambings from 2 ewes). For 13 ewes which lambed in both years, mean distance between successive lambing locations was 2.7 km (±0.9). Most ewes (85%) lambed within 3.8 km and 5 (38%) lambed within 1.1 km between years. When compared to native habitats, lambing areas on mine properties were at significantly lower elevations, on shallower slopes, further from escape terrain, on cooler aspects, and with lower proportions of conifer, shrub, both types of grasslands, and rock-rubble cover classes, and higher proportions of industrial cover class (Table 9). Use of aspect class differed between native and mine lambing areas (Gadj = 8.2, P = 0.017), with more use of south aspects by sheep using native areas (83%) compared with mine properties (38%). Ewes that lambed in native areas selected more strongly for warmer aspects and higher proportion of conifer cover and high-elevation grasslands than ewes that lambed in mine areas (Fig. 17). Ewes that lambed on mine properties selected more strongly for more rugged slopes and (obviously) for the industrial cover class. Lambing in native areas occurred at approximately the same elevation as winter ranges (Table 5), while those on mine properties occurred on average 250–300 m lower (Table 9). Nearly all lambing occurred within 40 m of escape terrain (Fig. 17).

Aurora Wildlife Research Page 42

Elk Valley bighorn sheep project

Figure 16. Location of bighorn sheep lambing sites as determined from GPS collar data, Elk Valley, British Columbia, May 2009 (n = 15) and May 2010 (n = 16).

Aurora Wildlife Research Page 43

Elk Valley bighorn sheep project

Table 9. Mean variable values within 100-m radius buffer zones of suspected bighorn sheep lambing areas, Elk Valley, 2009 and 2010, separated by mining (n = 13) and native areas (n = 18).

Mine Native 90th 90th 1 Variable x 2 SE percentile 2 SE percentile range3 range Topographic Elevation (m)* 1880 29.6 1738–2061 2162 37.4 1853–2479 Slope (%)* 60 5.8 20–95 82 3.4 65–114 Terrain ruggedness 0.049 0.00 0.006–0.104 0.038 0.0055 0.009–0.111 Dist. to escape terrain (m)* 56 28.975 8–382 15 3.3 1–58 Solar duration (hrs)* 496 31.0 319–650 549 23.0 235–665 Cover classes (%) Coniferous* 0 0.3 0–3 20 7.0 0–100 Deciduous 0 0.0 0–0 0 0.1 0–2 Logged 0 0.0 0–0 0 0.0 0–0 Shrub* 0 0.0 0–0 9 3.4 0–51 Wetlands 0 0.0 0–0 0 0.0 0–0 High elev. grasslands* 0 0.0 0–0 19 3.9 0–61 Medium elev. grasslands* 0 0.0 0–0 3 2.1 0–37 Exposed land 1 0.9 0–12 12 6.3 0–92 Rock-rubble* 0 0.0 0–0 37 8.9 0–94 Industrial* 99 1.0 88–100 0 0.0 0–0 1 Significant differences (t-test, P < 0.05) between native and mine means are indicated with asterisks. 2 Mean values calculated as mean of means from individual sheep locations. 3 Upper and lower limits of 90th percentile ranges were calculated as mean upper and lower limits of 90th percentile ranges from individual sheep.

Aurora Wildlife Research Page 44

Elk Valley bighorn sheep project

Figure 17. Lambing analyses individual means (open symbols; box and whisker plots with outliers [solid symbols]) for terrain and land cover parameters for available (1,000 m radius) and used (100 m radius) buffers in mine (n = 13) and native (n = 18) habitat, Elk Valley, 2009-2010. Y-axis values: elevation (m); slope (%); terrain ruggedness (units); distance to escape terrain (m); solar (units); cover classes (scaled from 0–100). See Table 2 for description of acronyms.

Rutting areas We documented 60 sheep-years of rut locations, representing 36 sheep (19 ewes, 17 rams).Forty-three (72%) rut locations were on native habitats and 17 were on mine properties (Fig. 18). With few exceptions, rutting areas in native habitats largely aligned with winter ranges. Among rutting areas in native habitats, most were on grasslands, exposed lands (often associated with grasslands), and rock- rubble habitats. Ewin Ridge (n = 11 sheep-years), Gill Peak (n = 6), Sheep Mountain (n = 6), and Todhunter (n = 6) made up two-thirds of rut occurrences in native habitats. One ram (at 4 years of age) was 12 km into Alberta during the 2010 rut, and another ram (5 years old) was along the Continental Divide during the rut in 2009. Among rutting areas in mine habitats, most occurred within the GHO property (n = 6), within active and previous mining areas within FRO off the west side of Turnbull (n = 5), and the edge of the inactive South Pit at EVO (n = 3), and included both reclaimed habitats and areas of active or recent mining. Compared with native areas, rutting areas on mine properties were at lower elevation, on less steep slopes and cooler aspects, further from escape terrain, and had lower proportions of both types of grasslands, exposed land, rock-rubble, and shrubs (Table 10).

Aurora Wildlife Research Page 45

Elk Valley bighorn sheep project

Table 10. Mean variable values of suspected bighorn sheep rut areas, Elk Valley, 2009 and 2010, separated by mining (n = 17) and native areas (n = 43).

Mine Native 90th 90th 1 Variable x 2 SE percentile 2 SE percentile range3 range Topographic Elevation (m)* 1903 46.5 1522–2126 2166 19.8 1972–2322 Slope (%)* 52 2.2 35–67 73 1.4 62–91 Terrain ruggedness 0.035 0.0034 0.011–0.064 0.040 0.0021 0.023–0.069 Dist. to escape terrain (m)* 50 5.6 26–104 29 8.1 5–53 Solar duration (hrs)* 566 18.8 398–673 601 6.5 534–663 Cover classes (%) Coniferous 3 1.2 0–17 7 1.6 0–30 Deciduous 0 0.0 0–0 0 0.0 0–0 Logged 0 0 0–0 1 0.5 0–0 Shrub* 1 0.4 0–5 7 1.8 0–27 Wetlands 0 0.0 0–0 0 0.1 0–0 High elev. grasslands* 1 0.7 0–11 26 2.5 8–65 Medium elev. grasslands* 0 0.3 0–4 4 1.0 0–17 Exposed land* 10 3.8 0–48 44 4.6 0–85 Rock-rubble* 1 0.3 0–4 9 2.6 0–48 Industrial* 84 4.7 30–100 2 1.1 0–14 1 Significant differences (t-test, P < 0.05) between native and mine means are indicated with asterisks. 2 Mean values calculated as mean of means from individual sheep locations. 3 Upper and lower limits of 90th percentile ranges were calculated as mean upper and lower limits of 90th percentile ranges from individual sheep.

Aurora Wildlife Research Page 46

Elk Valley bighorn sheep project

Figure 18. Locations of collared bighorn sheep during the rut during 21-30 November 2009 and 2010, Elk Valley, British Columbia. Colours represent different sheep.

Aurora Wildlife Research Page 47

Elk Valley bighorn sheep project

Rutting areas were identified for both 2009 and 2010 for 23 sheep (16 ewes, 7 rams). Eighteen of these sheep (78%) stayed with native or mine habitats between years, and 5 switched between habitats. Nine of the sheep (39%) rutted in the same area between years, and the remainder used rutting areas between 1.4 and 25.8 km apart. The 2 longest distances between sequential rut locations were from both a ram (25.8 km; Brownie in 2009 as a 3 year old to 12.5 km into Alberta in 2010), and a ewe (21.5 km; Ewin Ridge in 2009 as a 3 year old to Brownie in 2010).

Elk Valley East bighorn sheep surveys We surveyed the collar study area in both years, and in 2010 included Crowsnest Pass and Deadman Pass to the south, and in 2011 surveyed north to include the Aldridge Creek area. Snow coverage differed and survey effort was similar between years (Table 11). Elevations of observed sheep groups were similar between years (2010: x = 2,083 ± 27.8 m, range 1,340–2,560 m; 2011: = 2,079 ± 27.0 m, range 1,280–2,590 m; t = 0.1, P = 0.92). In 2010 we observed 596 sheep within 95 groups during the regular survey, and 29 additional sheep during telemetry checks on collared sheep (Table 12, Fig. 19). In 2011 we observed 454 sheep within 105 groups during the regular survey, and 27 additional sheep during telemetry checks. Mean group size in 2010 (6.3 ± 0.57 SE; range 1–30) was larger than in 2011 (4.3 ± 0.40 SE; range 1–23; t = 2.8, P = 0.006). Observed lamb:ewe ratios were lower and ram:ewe ratios were higher in 2011 (Table 12). When comparing the collar study area surveyed consistently between years, fewer sheep were observed in 2011. Similar numbers of rams were observed between years (131 and 135, respectively), but 24% fewer ewes were counted in 2011. Table 11. Elk Valley East bighorn sheep surveys, February 2010 and 2011.

Date Snow cover Survey time Census area Survey effort (hrs) (km2) (min/km2) 23-24 Feb 2010 Low; incomplete 7.7 228 2.0 21, 25 Feb 2011 High; complete 9.2 290 1.9

Aurora Wildlife Research Page 48

Elk Valley bighorn sheep project

Table 12. Summary of bighorn sheep observations during surveys of the Elk Valley East, February 2010 and 2011.

Study area Lambs:100 Rams:100 1 component Total Lambs Ewes U/C male Ram Cl 1 Ram Cl 2 Ram Cl 3 Ram Cl 4 U/C ewes ewes 2010 Survey 596 118 335 2 37 59 36 8 1 35 42 Telemetry checks 29 6 16 1 2 2 2 0 0 38 44 Total observed 625 124 351 3 39 61 38 8 1 35 43 Survey in 531 101 299 2 34 54 33 8 1 34 44 collared sheep study area

2011 Survey 454 67 233 1 38 70 35 10 0 29 66 Telemetry checks 27 4 20 0 0 3 0 0 0 -- -- Total observed 481 71 253 1 38 73 35 10 0 28 62 Survey in 427 65 227 1 29 60 35 10 0 29 60 collared sheep study area 1 U/C = Unclassified.

Aurora Wildlife Research Page 49

Elk Valley bighorn sheep project

Figure 19. Location of bighorn sheep observed during February 2010 and 2011 surveys of the Elk Valley East population. Most of the collar study area was surveyed both years, with additional surveys of the Crowsnest Pass and Deadman Pass areas in 2010, and the Aldridge Creek area in 2011.

Aurora Wildlife Research Page 50

Elk Valley bighorn sheep project

Some sexual segregation among winter ranges was evident (Table 13). Ewes comprised more than 95% of sheep observed on Henretta, Turnbull and Brownie, and rams comprised 91% of sheep observed on Chauncey. Ewes also tended to occur more often on Ewin Ridge and Sheep Mountain. Although sample sizes were smaller, ewes also were observed more often on Todhunter, Imperial Ridge, Elkview, and Crowsnest Pass, and rams comprised most sheep observed on Burnt Ridge Extension. Most of the sheep observed at higher elevations near the Divide in 2011 were also ewes. Sexes on Greenhills were more evenly split.

Table 13. Proportion (%) of ewes and rams observed on bighorn sheep winter ranges during February 2010 and 2011 surveys, Elk Valley, British Columbia. Class I rams were excluded from analysis.

Range Ewes Rams n 2010 n 2011 Aldridge 38 63 16 Henretta 96 4 36 22 Turnbull 98 3 20 15 Brownie 100 0 33 11 Greenhills 58 43 70 48 Burnt Ridge Ext. 5 95 10 10 Divide 92 8 26 Chauncey 9 91 33 24 Todhunter 80 20 18 13 Imperial Ridge 94 6 3 8 Ewin Ridge 74 26 98 106 Line Creek 90 10 8 5 Sheep Mt 82 18 49 38 Erickson Ridge 61 39 11 15 Elkview 85 15 25 14 Deadman Pass 60 40 5 Crowsnest 85 15 39

Sightability of collared sheep February 2010 In 2010, 31 collars were available for the sightability trial within the Elk Valley East. Two of these collared sheep were in high-elevation basins close to the Continental Divide and 1 sheep was in lower- elevation trees off of the Greenhills North range in areas that would not have normally been surveyed. Therefore, 28 collared sheep were within the census zone.

Twenty-three of the 28 collared sheep within the census zone were observed (p1 = 0.82 ± 0.072 SD), leading to an adjusted population estimate for the census zone of 645 (90% CI 580–772) (Table 14).

Aurora Wildlife Research Page 51

Elk Valley bighorn sheep project

Within the larger Elk Valley East area, the collar sightability resulted in an estimate of 725 sheep (90% CI

623–817). Three of the 31 collars were outside of the census zone but within the Elk Valley East (p2 = 0.90 ± 0.053 (SD)), therefore the corrected Elk Valley East population estimate was 803 sheep (90% CI 699–986). If the extrapolated totals from the Tobermory and Aldridge area are included (an additional 36 sheep to account for areas in the Elk Valley East not surveyed in 2010; see discussion), the estimate for the entire Elk Valley East population increases to about 840 sheep (90% CI 735–1,022).

Table 14. Bighorn sheep population observations and estimates (± 90% CI) corrected for sightability for the Elk Valley East survey, 2010, during a winter of low snowfall. The collar study area was surveyed with the addition of the Crowsnest Pass and Deadman Pass areas.

No. of sheep Estimate with sightability Estimate for collars outside Area observed applied (0.82) of census zone (0.90) Collar study area 531 645 (580–772) Survey study area 596 725 (623–817) 803 (699–986)

The proportion of collared ewes observed within the census zone (13/16; 81%) was similar to the proportion of collared rams observed (10/12; 83%). Mean size of groups with collared sheep that were observed (11.0, n = 17; range 1–30) tended to be larger than mean group size of collared sheep that were missed (5.0, n = 4; range 1–11; t = 2.3, P = 0.07), but the comparison suffered from small sample size. February 2011 We suspected up to 28 collared sheep were within the collar study area census zone during the survey. However, 2 collared sheep were not detected during the survey (and thus we were unable to verify if they were observed or not, or were within the census zone despite considerable search effort), leaving 26 collars available for the sightability trial. To our knowledge, none of the collared sheep were outside of the census zone.

Twenty of the 26 collared sheep were observed (p1 = 0.77 ± 0.083 SD), leading to an adjusted population estimate for the collar study area of 555 (90% CI 485–687) (Table 15). Within the Elk Valley East survey area (including the Aldridge area), the sightability correction resulted in an estimate of 590 sheep (90% CI 515–730).

Table 15. Bighorn sheep population observations and estimates (± 90% CI) corrected for sightability for the Elk Valley East survey, February 2011, during a winter of high snowfall. The collar study area was surveyed with the addition of the Aldridge area.

No. of sheep Estimate with sightability Area observed applied (0.77) Collar study area 427 555 (485–687) Survey study area 454 590 (515–730)

Aurora Wildlife Research Page 52

Elk Valley bighorn sheep project

The proportion of collared ewes observed within the census zone (11/14; 79%) was similar to the proportion of collared rams observed (9/12; 75%). Mean size of groups with collared sheep that were observed (7.5, n = 18; range 1–23) did not differ from mean group size of collared sheep that were missed (4.5, n = 6; range 1–13; t = 2.3, P = 0.26); however, the comparison again suffered from small sample sizes. Population estimate using Aerial Survey Sightability as calculated using the Aerial Survey model differed among surveys in 2008, 2010, and 2011, ranging from 0.68 to 0.81 (Table 16). Sightability was lower during the 2008 survey compared with 2010, likely because of more groups not moving (23% vs. 12%, respectively) and fewer groups in open slopes (31% vs. 61%, respectively). Survey effort was not calculated in 2008, but using rough hours spent in the central areas (BC FLNRO, unpubl. data), survey effort during 2010 may have been roughly 10–15% higher than used in 2008. Sightability estimated in 2011 was lower than observed in 2010, likely because of more groups not moving (32% vs. 12%, respectively) and more groups were observed in habitats with tree cover (20% vs. 9%, respectively). In 2011 mean estimated snow cover (87%) was considerably higher compared with 2010 (56%) and 2008 (61%). The modelled sheep population estimates for the collar survey area increased 24% between 2008 and 2010, and decreased 11% to 2011. The estimated lamb ratio was lower in 2011 and the ram ratio higher compared with previous years. Modelled sightability was nearly identical between ewes and rams in 2010 (0.80 vs. 0.81, respectively) and 2011 (0.74 vs. 0.72, respectively).

Table 16. Bighorn sheep population statistics for the Elk Valley East surveys, 2008, 2010, and 2011, based on the Idaho bighorn sheep sightability model (Unsworth et al. 1998).

Corrected Estimated Estimated No. of sheep population estimate lamb:100 ewe ram:100 ewe Survey area1 Year observed Sightability (± 90 CI) ratio ratio Entire survey 2008 412 0.68 609 ± 94 34 38 2010 596 0.79 748 ± 79 36 42 2011 454 0.74 616 ± 73 28 68 Collar area 2008 362 0.69 521 ± 87 33 36 2010 526 0.81 647 ± 70 34 44 2011 427 0.74 578 ± 72 28 61 1 Note that the area surveyed differed among years: 2008 – All except Crowsnest Pass and Deadman Pass; 2010 – all except Aldridge and Tobermory; 2011 – all except and Crowsnest Pass, Deadman Pass, and Tobermory. The collar area refers to the collar study area from Elkview in the south to Henretta in the north.

Aurora Wildlife Research Page 53

Elk Valley bighorn sheep project

Elk surveys During annual winter surveys, mine personnel observed the highest number of elk on EVO South Pit, Ewin Ridge, Imperial Ridge, Todhunter, Chauncey and Turnbull winter ranges (Fig. 20). Numbers of deer were generally low, with the exception of the Greenhills Ridge, north of GHO. A total of 5.5 hours was spent on survey during early August 2011. We observed highest numbers of elk on the Turnbull and Henretta winter ranges, with moderate numbers on several other ranges (Table 17). Fewer mule deer were observed, with highest numbers of mule deer observed on Chauncey and GHO property. Most sheep observed were associated with reclaimed habitats on LCO and GHO properties.

140

120 SE) ± 100

80

60 Elk Deer 40 Sheep

Mean no. of animals ( animalsno.of Mean 20

0

Figure 20. Mean numbers of elk, deer (mainly mule deer) and bighorn sheep observed on or near high elevation winter ranges identified by this study during annual surveys conducted by Teck Coal during mid-January to mid-March 2006–11 (L. Amos, Teck Coal, unpubl. data).

Aurora Wildlife Research Page 54

Elk Valley bighorn sheep project

Table 17. Ungulates observed during summer survey of high elevation bighorn sheep winter ranges, Elk Valley, 3-4 August 2011. WR is bighorn sheep winter range identified in this study.

Area Elk Mule deer Sheep EVO - South Pit WR 19 2 0 EVO – other areas 69 1 0 EVO - Erickson South WR 0 0 23 Sheep Mountain WR 0 0 0 West LCO 21 1 73 Ewin Ridge WR 23 3 0 Imperial WR 0 3 0 Todhunter WR 0 3 0 Chauncey WR 2 12 0 Brownie WR 25 0 0 Turnbull WR 135 0 2 Henretta WR 47 5 0 Greenhills Ridge WR 3 4 0 GHO WR 9 13 89 Burnt Ridge ext. WR 0 5 0 Total 347 51 187

Aurora Wildlife Research Page 55

Elk Valley bighorn sheep project

Discussion Elk Valley sheep wintered primarily on high-elevation grasslands, but utilized mine properties to some degree during winter and to a great extent during summer and fall. Our study indicates that sheep utilize a range of native and mine-altered habitats at different times of the year. Mine properties consisted mainly of pits and spoils, with lesser areas of highwalls and footwalls; reclamation – including planting with native and non-native mixtures – covered many of these features. Individual variability in habitat use and selection was evident, as were relatively subtle differences in habitat use selection between sexes. Sheep lambed and rutted in both mine-altered and native habitats. Winter was undoubtedly an important season for sheep survival, as indicated by the average 80 ha core winter range size observed during the more severe winter of 2010-11 (as compared to average 230 ha during the previous winter) and higher numbers of mortalities during that season.

Survival Annual survival rates of sheep in our study averaged 0.83 and 0.87 for ewes and rams, respectively. Jorgenson et al. (1997) found that in the absence of major predation or disease, adult ewes exhibited high survival rates up to 95% until approximately 7 years of age, but that survival of adult rams generally was lower, averaging 90% yearly (Festa-Bianchet 1989). However, Geist (1971) observed 96% survival for rams 2–7 years old, which was the age of most collared rams in our study. The lower survival rates observed in our study may have been related to the impacts of a severe winter on the population. During the 27-month study most mortality occurred during winter, as all but 1 of the 11 documented mortalities occurred between December and April. The calculated survival rates differed markedly between winters, with rates 3 times higher during the more severe winter (2010-11) compared with the winter of low snow accumulation (2009-10). Stelfox (1976) concluded that severe winters were accompanied by greater winter weight loss, and increased lamb mortality. Burles and Hoefs (1984) observed a decline in a Dall sheep (O. dall) population in the Yukon following the severe winter of 1981- 82, characterized by above average snowfall and below average temperatures and wind speeds; older sheep (≥7 years of age) and lambs were most affected. White et al. (2008) documented a population decline and recruitment of bighorn sheep following the severe winter of 1996-97. However, Jorgenson et al. (1997) found that winter severity had no effect on ewe or ram survival in Alberta populations. The thermoneutral zone in winter in bighorn sheep extends to –20°C (Chappel and Hudson 1978), thus only exceptionally cold and long winters may affect sheep survival. The apparent impact of a more severe winter on survival of Elk Valley sheep may be related to the high-elevation wintering strategy, leaving sheep with few options as snow depths increase and access to forage decreases (Stelfox 1976). Indeed, the 3 rams on Castle Mountain (north end of Chauncey winter range) that died in mid-March 2011 moved little in their last month of life, and effectively starved to death, despite apparent abundant forage under the snow. The 2 collared rams that died had wintered and survived in the same area in 2009-10. Thus the impact of winter severity may vary among wintering ecotypes. Although the importance of snow depth to over-winter survival cannot be overstated for this high- elevation wintering population of sheep, other climatic factors may also influence survival. Over-winter survival of northern ungulates is influenced not only by winter snowfall and restricted access to poor

Aurora Wildlife Research Page 56

Elk Valley bighorn sheep project quality forage, but the effects of summer climate – negatively by excessive heat or cool wet summers, and positively by high spring or summer precipitation – on forage quantity and quality; the resultant influence on body fat is also important (Stelfox 1976, Portier et al. 1998, Parker et al. 2009, White et al. 2011, Johnson et al. 2013). In a bighorn sheep population in eastern Alberta, Portier et al. (1998) found that winter lamb survival was not affected by winter weather, but that wet springs had a positive effect on neonatal and winter survival of lambs, and warm springs increased lamb survival the following winter. These factors may have acted on vegetation growth resulting in increased maternal nutrition and hence lamb survival. The lack of effect of winter weather on lamb survival may have been a result of a lack of exceptionally cold and long winters during the study, and the resulting Chinook winds that melt snow at lower elevations and clear it from higher slopes (Portier et al. 1998). Population density negatively affected lamb survival, with any effects of weather on lamb survival most evident at high density (density-dependence). Similarly, Johnson et al. (2013) observed that annual variation in pregnancy and recruitment trends in elk population in Oregon was most influenced by August precipitation. Winter severity, precipitation, and temperature were not significant in explaining variation in elk recruitment. Selenium deficiency has been linked to reduced fertility and disease/parasite resistance in domestic animals (Underwood and Suttle 1999, Higgs 2004). Mean serum selenium level detected from 24 sheep (0.12 ppm) sampled during capture was less than the mean level observed in a sample of 95 sheep from throughout the East Kootenay during 1978-2003 (0.16 ppm; Lemke and Schwantje 2005). Lower levels ( x = 0.06 ppm) were observed among ewes in a southwestern Alberta population (Jokinen et al. 2007). Selenium levels observed in Elk Valley sheep are within or near the lower bounds of established normal ranges for domestic sheep (0.08 – 0.50 ppm; Ministry of Agriculture and Lands Animal Health Centre, Abbotsford, BC) and California and Rocky Mountain bighorn sheep (0.13-0.20 ppm; Puls 1994), but are unlikely to influence population fitness. Selenium levels in forage grown on reclaimed coal waste on 3 of the mine properties were within the high range for consumption by domestic cattle and sheep, but were assessed to not likely be a risk to wildlife consuming this vegetation (C.E. Jones & Associates 2007). No large-scale die-offs have been reported in sheep in the Elk Valley (Beswick 1999), despite a die-off occurring in the early 1980s in adjacent areas of the Southern Rockies in both BC and Alberta (Davidson 1994). The die-off was believed to be a result of acute pneumonia, which was passed from domestic sheep to wild sheep. A small die-off among sheep in the Elk Valley East population occurred during 1998-99 (10 sheep died or collected due to poor condition), but although pneumonia (bronchopneumonia with abscessation) was the proximate cause of death, the ultimate cause was not determined (Beswick 1999).

Movements Winter ranges were roughly 10-30% of the size of summer ranges, with winter ranges during the more severe winter one-third the size of the low snow winter. During the more severe winter, collared sheep largely limited their activity to the known main winter ranges, largely forgoing habitats used to some extent in the previous (lower snow) winter towards and along the Continental Divide (Fig. 6). However, some use of the Divide area between Beehive Mountain and Mt. Farquhar was detected during the

Aurora Wildlife Research Page 57

Elk Valley bighorn sheep project

February 2011 FLNRO survey (Fig. 19) and Teck mines surveys (L. Amos, Teck Coal, unpubl. data) in areas where no collared sheep were found that winter. This suggests either greater use of habitats than indicated by the collared sample, or additional and unsampled groups of sheep present closer to the Divide that may have had different patterns of habitat use and movements. Indeed, our modelling did show high quality habitat in the vicinity of the Divide (Figs. 14, 15). Surveys suggested greater use of the Divide area by ewes. Of the 6 groups of sheep totalling 28 individuals observed at or near the Continental Divide during the February 2011 FLNRO survey, only 2 were Class II and higher rams, while 20 were ewes and 4 were lambs. Similarly, of 57 sheep in 7 groups observed at or near the Divide during early February 2011 mine surveys, only 2 were Class II and higher rams. Ewes may utilize these higher- elevation habitats to reduce predation risk (Festa-Bianchet 1988a, Bleich et al. 1997, Shackleton et al. 1999). In this population, as elsewhere (Geist 1971, Morgantini and Hudson 1981, Bleich et al. 1997), male and female bighorn sheep often occupied separate winter ranges, although spatial and temporal overlap did occur. Chauncey was used almost solely by rams, while Henretta, Turnbull, Brownie, Todhunter, and Sheep Mountain were dominated by ewes (Table 13). Approximately three-quarters of the sheep counted on Ewin Ridge in 2010 and 2011 were ewes. On an Alberta winter range, ram groups occurred on more steep and rocky ground and closer to forests than ewe groups (Morgantini and Hudson 1981). These differences in habitat use are most likely due to ewes selecting secure areas for raising their young, and rams choosing areas for maximizing body condition (Main et al. 1996). To examine the relative importance of winter ranges within the study area, we mapped the boundaries of winter distribution of sheep and summed observed sheep numbers within each range. This examination used collar locations and aerial survey data, and did not consider habitat mapping or features (Summit Environmental Consultants Inc. 2012). Winter distribution was free-hand digitized by tightly encompassing continuous areas of collar and survey locations, and was summed for 15 ranges identified. No attempt was made to identify scattered and sporadic winter distribution along the Continental Divide. Numbers of sheep observed within each range were averaged between the 2010 and 2011 FLNRO surveys, and the density calculated. We identified 29.3 km2 of habitat used by sheep during winter among 15 areas, approximately 22.6 km2 of which – 13 ranges – could be considered main winter range (Fig. 21). These main ranges comprise 2.7% of the study area (4.3% of the merged annual sheep ranges), emphasizing the limited amount of used winter ranges within the landscape. Gill Peak, Brownie, and Ewin Ridge had the highest densities of sheep, while Ewin Ridge had by far the highest mean count (Fig. 22). Sheep Mountain and Greenhills Ridge, north of GHO, also had relatively high counts and high densities of sheep, while Turnbull and Burnt Ridge extension had relatively high densities but lower overall counts. Chauncey had the highest count and density among the Imperial Ridge to Todhunter to Chauncey stretch. Considering the area of use and proportional use by collared sheep (Fig. 6) and the winter range population estimates (see below), as stated by Demarchi (1968) Ewin Ridge could still be considered the most important winter range within the study area.

Aurora Wildlife Research Page 58

Elk Valley bighorn sheep project

Figure 21. Bighorn sheep winter ranges as mapped from collar data 2009-11, 2010 and 2011 government survey data, and 2009–2011 Teck Coal survey data, Elk Valley East, British Columbia.

In the Canadian Rocky Mountain parks, Stelfox (1976) found that winter ranges stocked with 0.8 sheep- months/ha were in good condition, those stocked with 1.5 sheep-months/ha were in fair condition, and those with 2.0 sheep-months/ha were in poor, overgrazed condition. Assuming 5 months of winter residency (December–April), Gill Peak (2.8 sheep-months/ha) and Brownie (2.0) had the highest stocking rate, followed by Ewin Ridge (1.5), Turnbull (1.0) and Burnt Ridge extension (1.0). Forage levels by elk would also affect range condition (Stelfox 1976). Stelfox (1976) assumed an elk was 2.9 sheep units when calculating grazing capacity. White et al. (2008) observed increases in abundance and recruitment of bighorn sheep concurrent with a 50% decrease in numbers of elk.

Aurora Wildlife Research Page 59

Elk Valley bighorn sheep project

Density Mean count 50 140 45 120

40 ) 2 35 100 30 80 25 60 20

15 40 Sheep density (no./km densitySheep 10 20 5 surveys2011and 2010 of count Mean 0 0

Figure 22. Bighorn sheep density (sheep/km2) and mean count between 2010 and 2011 FLNRO surveys on winter ranges within the Elk Valley East, British Columbia.

Number of elk within the Elk Valley are high; a mid-winter 2013 census estimated 2,772 elk (90% CI: 2,682–2,862) within the valley (P. Stent, MFLNRO, unpubl. data). Survey conducted by mine personnel and this project suggest the impact from elk varies seasonally, but could contribute to reduced winter range conditions on Ewin Ridge, Imperial Ridge, Todhunter, Chauncey, Turnbull and Henretta (Fig. 20, Table 17). TAESCO (1985a,b) suggested that in the early 1980s there was heavy competition and temporal overlap between sheep and elk on lower elevation sheep ranges (i.e., Todhunter and Imperial Ridge), but no competition on the high elevation Ewin Ridge sheep winter range. Surveys during winters 1981-84 observed an average of 85 elk (range 42–143) on Imperial Ridge and Todhunter winter ranges and an average of 43 elk on Ewin Ridge (22–73; TAESCO 1985a). Teck mine surveys from 2006-11 observed an average of 58 elk (range 16–93) on Imperial Ridge and Todhunter, and 37 (11–81) on Ewin Ridge (L. Amos, Teck Coal, unpubl. data). These two sets of surveys are probably comparable in terms of methodologies, extent and effort, which suggests overlap between these species has continued for some time. Changes in elevation use and movement rates during the year followed predictable patterns (Figs. 7, 8), likely dictated by spatial and temporal changes in forage quality coupled with restrictions to movement imposed by winter snow depths. Snow restricts movements of mountain ungulates (Parker et al. 1984, Daily and Hobbs 1989), and mountain sheep generally avoid snow depths >25-30 cm (Stelfox 1976, Sweanor et al. 1996). Upward vertical migration in spring and summer enables populations to make

Aurora Wildlife Research Page 60

Elk Valley bighorn sheep project longer use of high-quality forage, as new growth proceeds up the mountain (Stelfox 1976, Shackleton et al. 1999). Collared sheep within this population, especially rams, were largely migratory between summer and winter ranges. However, nearly one-third of the collared ewes were non-migratory with overlapping summer and winter ranges, resulting in smaller individual (annual) range sizes. All non-migratory sheep resided on and adjacent to the GHO and FRO mine sites, suggesting these areas provided year-round habitat for sheep. The non-migratory behaviour of some Elk Valley sheep may have occurred because of availability of high quality forage supplied by reclamation. Non-migratory behaviour may instill reduced risk of predation (Festa-Bianchet 1988, Nicholson et al. 1997, Hebblewhite and Merrill 2011), but residency and increased long-term density may result in less access to new growth, high-quality forage (Shackleton et al. 1999) and higher potential for increased lungworm infections and disease transmission (Stelfox 1976). Observations of a population of elk in Banff National Park showed that while migrants experienced higher pregnancy rates and winter calf weights associated with higher forage quality, survival of migrant adult females and calves were lower than resident elk, as resident elk traded high quality food to reduce predation risk (Hebblewhite and Merrill 2011). The increased forage quality present on the reclaimed mine properties may negate a direct comparison with this elk study, as long as the forage is available during winter. Non-migratory sheep that concentrated their locations on the north side of FRO generally wintered on Henretta, Turnbull or Brownie, utilizing native, and reclaimed and non-reclaimed mine habitats, but sheep that focussed on GHO used the GHO property for wintering, using both reclaimed and non-reclaimed mine habitats. Migration events often occurred as rapid movements through forested, low-elevation valleys. Three of the 11 mortalities detected in this study occurred at low-elevation or within treed movement corridors. Considering a very low proportion of the population’s annual distribution occurred in these habitats, the risk involved in travelling through poor quality habitats is likely high. We did not delineate subpopulations within the study area. Demarchi et al. (2000: Table 1) defined a subpopulation as 2 or more wintering herds that share a common summer range. Thirteen to 15 spatially separated winter ranges occur within the study area, but the extent of overlap during summer is extensive, possibly influenced by re-vegetation on mine properties. Given the range of area over which these sheep occur (>70 km linear), it is doubtful they would share a common summer range as per the definition of Demarchi et al. (2000). However, a simple analysis suggests lower exchange among areas separated by greater distances and major valleys (Fig. 23). Sheep in the Elk Valley can be clustered into 5 main groups as follows:

 Collared sheep using Erickson Ridge and Sheep Mountain (2 ewes, 4 rams) also used Line Creek Canyon, but moved only as far north as the central and western portion of LCO;  Sheep that used Ewin Ridge (4 ewes, 4 rams) spent the vast majority of their time on Ewin Ridge and LCO, with limited excursions to Brownie, Turnbull and GHO by 5 of the individuals (1 ewe, 4 rams);  Sheep wintering on the Imperial Ridge, Todhunter, and Chauncey (2 ewes, 7 rams) moved mainly among these 3 winter ranges and into the southwestern portion of FRO property, Turnbull and Brownie, and made the greatest use of terrain closer to the Continental Divide;

Aurora Wildlife Research Page 61

Elk Valley bighorn sheep project

 Sheep using GHO (2 ewes, 6 rams) largely remained on GHO property, Burnt Ridge extension and the Greenhills Ridge;  Sheep that used Henretta and Gill Peak (7 ewes, 1 ram) also used FRO property and adjacent winter ranges of Turnbull and Brownie, but not GHO. One 2-3 year old ram that covered GHO, LCO and Ewin Ridge was excluded from this summary. Interchange among groups was evident, but the overall pattern largely held (Fig. 23). FRO appears to be the area of greatest overlap among adjacent groups, with LCO also showing overlap. Support for these groupings is provided by the high degree of fidelity within winter ranges, as also shown from the early 1980s (TAESCO 1985b). We observed only 4 instances (7%) of sheep moving outside of the groupings listed above during subsequent winters. This is not to suggest that genetic interchange among groups is limited in any way; 2 rams normally associated with the GHO mine property rutted on Ewin Ridge, and a ewe appeared to shift from Ewin Ridge to Brownie during the course of the study. However, annual movements by sheep are largely spatially limited, likely dictated by distance and broad valleys. Collaring studies from the early 1980s suggested the Ewin Ridge and Sheep Mountain sheep populations were completely separate but overlapped marginally within the Line Creek Canyon; that the Sheep Mountain population travelled as far south as Erickson and Alexander creeks; and populations on Ewin Ridge and Todhunter/Imperial Ridge were not connected (TAESCO 1985b), all of which largely agree with our assessment.

Figure 23. Groupings of collared bighorn sheep based on movement patterns, Elk Valley, 2009-2011.

Aurora Wildlife Research Page 62

Elk Valley bighorn sheep project

Fidelity to winter range was high (88%) and equal between sexes. Other researchers have suggested that ewes show high annual fidelity to their winter range, and some rams do not (Festa-Bianchet 1986a, b). Geist (1971) suggested fidelity was 90% for ewes and 75% for rams.

Resource selection Predictably, resource selection by sheep varied slightly by scale, with stronger selection and avoidance observed at the winter use to home range scale as compared with the within-winter range scale. Terrain variables (except for terrain ruggedness) and preference for high-elevation grasslands/exposed lands were the most important variables at both scales. Although some sheep did use mine properties during winter (Fig. 6), the industrial cover class ranked lower in importance during winter compared with native winter habitats, with much higher use (and likely selection) of mine properties by the population as a whole during summer and fall (Fig. 10). Limited differences in selection between sexes were evident. Security may be more important to ewes with lambs than to rams, because rams can better defend themselves against predators, and ewes may make trade-offs of reduced forage benefits to minimize predation risk to their offspring (Bleich et al. 1997, Shackleton et al. 1999). However, we did not detect differences in selection for distance to escape terrain between sexes. Previous studies have found slope and distance to escape terrain as important variables in sheep habitat selection, along with aspect or solar radiation index and low snow cover (<25-30 cm: Stelfox 1976, Sweanor et al. 1996, Dicus 2002, DeCesare and Pletscher 2006, Bleich et al. 2009). On low-elevation winter range in the East Kootenay, steep terrain, distance to steep terrain, and terrain ruggedness were important variables defining winter habitat selection by ewes (Jalkotzy 2000, Kinley 2007). Selection of high solar duration and open grasslands in our study coincided with areas where snow cover would be minimized. The majority of main “winter ranges” in our study area were mapped as various proportions of high elevation grasslands and exposed lands, likely depending upon range quality and the amount of vegetation cover (Summit Environmental Consultants Inc. 2012) and reflectance mapping. Good visibility has also been identified as a key component of bighorn habitat (Risenhoover and Bailey 1985, Zeigenfuss et al. 2000, DeCesare and Pletscher 2004). Rocky Mountain bighorn sheep use open habitats to facilitate detection of predators (Wakelyn 1987). Visibility is an important habitat feature for sheep because their predator-avoidance strategy involves foraging diurnally in relatively large dispersed groups on open habitat close to escape terrain. Habitats with greater visibility and larger group sizes increase foraging efficiency (Risenhoover and Bailey 1985). We could not directly evaluate visibility, but selected land cover classes were generally open, and coniferous forests were avoided. Sheep might avoid conifers because reduced visibility impairs predator detection (Geist 1971, Risenhoover and Bailey 1985), but open stands also have increased forage. However, DeCesare and Pletscher (2006) found no strong preference by bighorn sheep for areas with high horizontal visibility, but noted that visibility may be more important at coarse scale of selection. Terrain ruggedness and slope appear to distinguish 2 different, but biologically important, components of bighorn sheep escape terrain (Sappington et al. 2005). Terrain ruggedness was an overall weak variable at both scales; the direction and magnitude of selection varied greatly among individuals, rendering its value for modelling limited. Similar to our study, Bleich et al. (1997) found that bighorn

Aurora Wildlife Research Page 63

Elk Valley bighorn sheep project ewes consistently used more rugged terrain than rams. Sappington et al. (2005) showed that terrain ruggedness was an important, although not dominant factor in spring habitat selection by female desert bighorn sheep. However, after accounting for other variables, sheep in California were associated with less rough terrain than were random locations (Bleich et al. 2009). Poole et al. (2009) also found that terrain ruggedness was the overwhelming driver for winter habitat selection for mountain goats in the East Kootenay. Our 2-stage modelling approach considered individual variation in selection of parameters. The results appear to be superior to a mixed modelling approach that considers the individual as a term in model development (Gilles et al. 2006), because it explicitly highlights the range of responses among sheep. Bootstrapping parameter estimates with less weight given to individuals with high variance allowed more realistic models less influenced by outlier data. Given differences in use of mine properties by sheep, the individual modelling approach acknowledges different types of selection within the study area. Differences in selection among herds and areas are also likely. DeCesare and Pletscher (2006) found that resource selection models for ewes across sites gave mixed results, such that models developed for 1 herd were not well validated to the 2 other herds. These authors found nearness to escape terrain and grasslands were the only consistent variables in sheep winter resource selection among the 3 herds. Selection for cover types varied among herds, although high-visibility habitats were generally preferred. DeCesare and Pletscher (2006) recommended to externally validate local models in each area. While our modelling appeared to effectively capture used sheep winter ranges within the study area, areas lightly or not used by sheep were also captured in the models. These included the northern half of Erickson Ridge, much of the range west of LCO, ridges off the Continental Divide south of the level of LCO, and portions of all mine properties (Fig. 15). It may be that these areas are unused but suitable winter ranges, but it is more likely that our modelling did not capture some aspect of sheep winter range selection important to sheep. Snow cover and depth varied among ranges and between years (Summit Environmental Consultants Inc. 2012), and may affect traditional use of areas by sheep. Elevation and solar radiation can be considered surrogate variables to model snow but may not translate to the fine scale or accurately reflect true snow depths. Wind direction and strength would also influence snow cover and depth. Predation risk, distance among seasonal ranges, or other factors could be influencing selection of winter ranges (Festa-Bianchet 1988b, Shackleton et al. 1999).

Use of mine properties Sheep in the Elk Valley extensively utilized coal mine properties, especially during summer and fall (Fig. 10). We observed sheep in all habitats within mine properties, from adjacent to active mining and hauling and on highwalls, to reclaimed areas. During monthly telemetry flights to locate collars we regularly observed groups of 40–50 sheep during summer and fall on reclaimed habitats within LCO and GHO, and of 25–30 sheep on portions of FRO. Behaviour of sheep on the properties suggested high levels of habituation to machinery and personnel. The topographic characteristics of sheep wintering areas on mine properties differed markedly from native wintering areas, occurring on average 300 m lower in elevation, on less steep slopes further from

Aurora Wildlife Research Page 64

Elk Valley bighorn sheep project escape terrain (Table 6). Ninety-five percent of sheep winter locations on mine property occurred within 79 m of escape terrain, compared with 44 m for native habitats. Reclaimed lands and spoils were used to the greatest extent. Use of active and reclaimed mines by bighorn sheep has been previously observed. Jansen et al. (2006, 2007) found that desert bighorn sheep behaviour and activity at an active copper mine was generally similar to that recorded while outside the mine perimeter. These authors suggested modern mining might be predictable enough to allow sheep to habituate to human activities. At most of the Elk Valley mines sheep are regularly observed along active haul roads and around office complexes (K. Podrasky, LCO, pers. comm.). Oehler et al. (2005) observed no differences in size of annual home ranges, composition of diet, and lamb:ewe ratios between ewes inhabiting mined and non-mined areas. Similarly, Bleich et al. (2009) observed sheep being associated with areas closer to limestone mines than were random points. However, relative predation risk on mine properties compared with native habitats is unknown, and since reducing predation risk influences habitat selection (Festa-Bianchet 1988a), predation likely influences habitat selection at the mine versus native habitat scale. Whether occupancy of mine sites translates into benefits to sheep populations depends on demographic responses (Oehler et al. 2005), which can be difficult to quantify (Bleich et al. 2009). MacCallum and Geist (1992) documented use of a reclaimed open-pit coal mine on the east slopes of the Rocky Mountains in Alberta. There, reclaimed grasslands provided over twice the productivity of native ranges and likely contributed to large body mass of ewes and high lamb survival. Movement patterns were generally opposite of our study, in that Alberta sheep used the reclaimed areas primarily as winter range, but also for lambing, summer range and the rut, and in summer spent much of their time in nearby alpine areas. Differences in seasonal use of mine sites may be related to snow depths affected by winds and exposure, and wintering areas available.

Lambing areas During the lambing season, ewes are thought to seek solitude in rocky bluffs near their winter range to protect lambs from inclement weather (Geist 1971) and to avoid predation on lambs (Festa-Bianchet 1988b). Ewes and lambs are highly vulnerable for several days following birth. Shackleton et al. (1999) stated “…the only consistently effective anti-predator strategy at this stage seems to be the ewe’s selection of an isolated and precipitous area to give birth”. In an Alberta study, pregnant ewes moved from higher quality forage to areas of lower quality to provide better protection from predation (Festa- Bianchet 1988b). Collared ewes in this study used native and mine-affected habitats roughly equally during the study. DeCesare and Pletscher (2006) suggested that distance to escape terrain was the most important and consistent variable during the lambing period (defined as early May through late July). Active mine areas may provide areas of lower predation risk, as treed habitat conducive to cougar predation is largely absent and wolf activity (but likely not coyotes) may be reduced by the constant human activity. Thus, less vigilance – as shown by lambing on shallower slopes – may be needed to reduce predation risk. The lower elevations and reclaimed and re-vegetated habitats likely provide higher quality forage earlier in the spring than found in higher-elevation native areas. We were unable to assess fitness of ewes

Aurora Wildlife Research Page 65

Elk Valley bighorn sheep project following these different lambing habitat strategies, as we did not follow survival of lambs. At a coal mine development in the Alberta foothills, MacCallum (1992) found most ewes moved off site for lambing, but several stayed on site within 150 m of active rock dumping. Since that study the incident of lambing on mine sites has increased, with ewes selecting the highest and steepest walls on the mine to give birth (B. MacCallum, Bighorn Environmental Design Ltd., pers. comm.). Selection for higher solar duration and higher proportion of high-elevation grasslands by ewes lambing in native areas may have been a result of ewes seeking warmer micro-sites and areas of early green-up at these higher elevations. Timber may provide security against aerial predators (Shackleton et al. 1999); we observed a golden eagle harassing a ewe and lamb on Turnbull, but we observed little use and no selection for coniferous stands at native lambing areas. Based on limited standard VHF radio telemetry, Schuerholz (1984) suggested lambing occurred predominantly in conifer forests. Close proximity to water has been proposed as important to lambing (Sweanor et al. 1996, Zeigenfuss et al. 2000), but water is likely readily available in our study area. DeCesare and Pletscher (2006) found no consistent relationship between bighorn sheep use and proximity to water in any seasons.

Rutting areas Rutting occurred through the study area, primarily on winter ranges (as suggested by Schuerholz 1984) and mainly in native habitats. Rutting on mine properties was not limited to areas known to be used as winter range. Fidelity to rutting areas varied between years, possibly a result of the young age of many of the rams collared during this study.

Elk Valley East bighorn sheep surveys Numbers of Elk Valley East bighorn sheep have increased steadily since the late 1980s (Tables 18, 19; Stent et al. 2013). However, comparisons should be conducted with caution because not all areas were surveyed prior to 1991. Using only the area from Henretta to Sheep Mt. and including Greenhills, sheep numbers peaked in the early 1980s, declined to the late 1980s, and increased thereafter to levels slightly higher than the early 1980s (Fig. 24). Since 1998, the Elk Valley East population (all herds) increased at an average annual rate of 1.07, a doubling every 10 years. Higher rates of increase have been observed for bighorn sheep. Singer et al. (2000) reported that the average rate of increase for 3 healthy populations of bighorn sheep in western USA was 1.17, and rates as high as 1.30 have been observed, although primarily with introduced populations (Shackleton et al. 1999). It is unlikely that immigration played a role in the Elk Valley East increase because no exchange was detected with the Elk Valley West populations or into wintering ranges of the Alberta Southern east slopes populations (Alberta Conservation Association 2009). A portion of this increase could be attributed to population growth at GHO, which increased from very few sheep prior to the late 1980s, to nearly 100 sheep in 2010, concurrent with an increase in mine reclamation (Table 19). A similar increase to 40–45 sheep likely occurred in the South Pit area of EVO and adjacent Erickson Ridge.

Aurora Wildlife Research Page 66

Elk Valley bighorn sheep project

Table 18. Bighorn sheep observed, and lamb:ewe and ram:ewe ratios during aerial surveys from BC FLNRO data for the Elk Valley East population, 1981–2011. Early data provided by BC FLNRO (Teske and Forbes 2002, I. Teske, unpubl. data). Note that data for the Elk Valley East population prior to 1991 likely did not cover all winter ranges.

Date No. of sheep Lambs:100 ewes Rams:100 ewes

Mar 1981 343 54 42

Feb 1983 231 46 54

Mar 1985 213 41 62

Mar 1986 165 50 70

Jan 1987 196 32 67

Mar 1988 108 48 75

Mar 1990 158 36 50

Mar 1991 166 33 77

Feb 1998 273 50 61

Feb 1999 217 42 -

Feb 2002 327 47 49

Feb 2003 308 42 47

Feb 2005 376 33 46

Feb 2008 412 (4421) 35 39

Feb 2010 596 (6322) 35 42

Feb 2011 454 (5643) 29 66

1 Includes 30 sheep (average observed during 2002, 2003, and 2005 surveys) added to the total observed to account for Deadman Pass and Crowsnest Pass not surveyed due to inclement weather. 2 Includes 36 sheep (average observed during 2003, 2005, and 2008 surveys) added to the total observed to account for Tobermory and Aldridge not surveyed due to lack of funds. 3 Includes 45 sheep (2008 survey) added to the total observed to account for Tobermory not surveyed, and 65 sheep (2010 survey) added to the total observed to account for Deadman Pass and Crowsnest Pass not surveyed.

Aurora Wildlife Research Page 67

Elk Valley bighorn sheep project

Table 19. Bighorn sheep observed on winter ranges within the Elk Valley East population during aerial surveys, 1975 to 2011. Herd numbers were obtained from digital BC FLNRO data (I. Teske, BC FLNRO, unpubl. data) and checked spatially using GIS. Blank values signify no survey of that area conducted.

Chauncey, Brownie, Todhunter, Tobermory, Turnbull, Greenhills Imperial Ewin Ridge, Sheep Erickson Deadman Crowsnest Date Aldridge Henretta Range Ridge Line Creek Mountain Ridge Pass Pass Total Mar 1975 7 109 44 160 Mar 1976 21 87 53 161 Mar 1979 24 72 14 110 Mar 1981 20 70 179 74 343 Feb 1983 25 144 62 231 Mar 1985 19 76 100 18 213 Mar 1986 17 28 89 31 165 Jan 1987 14 40 95 47 196 Mar 1988 16 14 52 26 108 Mar 1990 19 30 69 40 158 Mar 1991 10 36 92 28 166 Feb 1998 41 26 32 62 53 11 27 1 253 Feb 2002 29 58 18 44 74 46 27 5 26 327 Feb 2003 30 70 29 43 63 33 13 7 20 308 Feb 2005 28 69 48 44 91 52 11 10 23 376 Feb 2008 50 65 77 47 110 43 20 412 Feb 2010 118 98 57 154 59 45 8 57 596 Feb 2011 27 (Aldridge) 59 63 92 128 46 39 454

Aurora Wildlife Research Page 68

Elk Valley bighorn sheep project

Figure 24. Number of sheep observed during winter surveys in the main area of the Elk Valley East population (from Henretta to Sheep Mt. and including Greenhills), 1975–2011.

Within the sheep collar area surveyed during 2010 and 2011, approximately 19% fewer animals were observed in 2011 compared with 2010 (Tables 18, 19), and 11% (modelled) to 14% (collar sightability) fewer sheep were estimated. The decline in observed and estimated sheep numbers between 2010 and 2011 surveys may in part be real, but it may also be an artefact of lower sightability during the 2011 survey (or very high sightability during the 2010 survey). Survey effort during 2010 and 2011 was similar (2.0 and 1.9 minutes/km2, respectively). Both collar and modelled sightability were lower in 2011 compared with 2010, and these differences may have been in part because of differences in sheep behaviour and fine-scale habitat selection between years. Snow coverage and depth was much higher in 2011, and more sheep may have been tucked into trees. Cold weather (-20 – -25°C) during the second day of the survey may also have affected sheep habitat selection and movements. The much higher ram ratio in 2011 (Table 18) suggests comparatively more ewes were missed during the survey. Alternatively, more severe winter conditions during winter 2010-11 may also have contributed to higher mortality rates and a decrease in population size. During the 2011 survey we observed a number of sheep kills, including a pair of coyotes on a sheep kill near the LCO office, evidence of wolf predation, and an apparent predation attempt by a golden eagle. In addition, snow cover affects bighorn sheep foraging efficiency and diet quality (Goodson et al. 1991), which may influence overwinter survival. Lamb ratios were below 30 lambs:100 ewes for the first time in recorded surveys (Table 18), suggesting higher lamb mortality. During the 1970s and early 1980s, Schuerholz (1984) February-March classification counts averaged 55 lambs: 100 ewes. Demarchi et al. (2000) suggest that

Aurora Wildlife Research Page 69

Elk Valley bighorn sheep project for a Rocky Mountain bighorn sheep population to remain stable, it must maintain an over-winter ratio of approximately 30 lambs:100 ewes. Ram ratios in the Elk Valley peaked in the late 1980s and early 1990s in the 65-75 rams:100 ewes range, and have subsequently declined to the 40-50 rams:100 ewes range (Table 18). The higher ratio observed in 2011 (66:100) may be an artefact of proportionately more ewes not observed compared with rams. Survey sightability In both 2010 and 2011 surveys, the sightability obtained from the collar data was similar to the sightability obtained from the logistic model (2010: 0.82 and 0.79; 2011: 0.77 and 0.74, respectively). Similar results from these 2 independent methods lend credence to the values. The logistic model was based on whether sheep were active or not, and if they were on flat/open slopes or not. The degree of vegetation cover may play more of a role in sightability in the Elk Valley than occurred in Idaho. During model development, Bodie et al. (1995) found estimated sighting probabilities for California bighorn sheep ranging from 0.33 for not moving in canyon habitats, to 0.90 for moving on flat/open slope habitats. During winter surveys, sightability of bighorn sheep in Colorado averaged 0.57 (range 0.35– 0.86; Neal et al. 1993), and of Stone’s sheep (Ovis dalli stonei) ewes in northern British Columbia ranged from 0.72 to 0.84 (Cubberley 2008). Although sheep sightability in the Elk Valley may average 75–85% depending upon conditions, it is more accurate to apply the sightability model to survey data, rather than apply an arbitrary value. We examined the influence of applying the collar-derived sightability based on the individual animal as compared with the group. The sightability correction factor from individual collars was similar to the correction factor determined from collared groups in 2010 (0.82 ± 0.072 and 0.85 ± 0.080, respectively) and 2011 (0.77 ± 0.083 and 0.78 ± 0.086, respectively). However, population estimates that use groups for calculations rather than by individuals may result in an overestimation of the target population (Neal et al. 1993, Cubberley 2008).

Conclusions Bighorn sheep numbers in the Elk Valley East population have increased since the late 1980s (Stent et al. 2013). This increase has occurred concurrent with active coal mining and reclamation within or near their main habitats. High elevation winter ranges are undoubtedly important to winter survival in the Elk Valley bighorn sheep population, as demonstrated by restricted distribution, high fidelity to winter ranges and the apparent influence of winter severity on survival. More severe winters undoubtedly stress individuals within the population, as exemplified by the starvation of 3 prime rams on Castle Mountain in mid-March 2011. However, the effects of spring weather on vegetation growth may be as important to winter survival of lambs as winter severity, presumably by affecting body reserves at the onset of winter (Portier et al. 1998). The main winter ranges as identified through collar and survey data comprise approximately 3% of the study area (4% of the merged annual sheep ranges), emphasizing the limited amount of utilized winter habitat within the landscape. However, reclamation on mine properties can provide high quality forage

Aurora Wildlife Research Page 70

Elk Valley bighorn sheep project year-round (MacCallum and Geist 1992), which in all likelihood affects overwinter survival and fecundity (Portier et al. 1998, Parker et al. 2009), and has likely contributed to the increase in the population since the late 1980s. In addition to sheep grazing pressure, the number of elk on native sheep winter ranges during both summer and winter may be negatively influencing range quality on these ranges (Summit Environmental Consultants Inc. 2012). Native sheep winter ranges in the Elk Valley occur within a specific elevation range (mainly 1,900– 2,300 m) on grasslands with warm aspects close to escape terrain. Sheep used mine-altered habitats during winter to a more limited extent; these areas were similar to native ranges in terms of access to escape terrain and generally warm aspects. Lambing and rutting occurred on both native and mine- altered ranges.

Management recommendations Bighorn sheep require a balance of habitats supplying adequate year-round forage and security from predation. Important components of winter habitat include juxtaposition of escape terrain and adequate and accessible forage, and warm aspects (Tilton and Willard 1982. Jansen et al. 2006).For the most part, the features characterizing winter habitat are static and in specific locations within the environment, with limited possibility to enhance through management. Mines can reproduce or mimic many of the static features of sheep range in general (MacCallum and Geist 1992, Bleich et al. 1997, Jansen et al. 2006), but may be unable to ensure low snow depths that are important for high quality winter range (Stelfox 1976). Avoidance of conifers and use of high visibility habitats suggest that forest encroachment would be detrimental to sheep (Wakelyn 1987), and thus could be controlled in areas where this is occurring using slashing or burning (Smith et al. 1999). Low-elevation movement paths (Fig. 9), although roughly located in this study given the 10 hr collar fix rate, could also be managed through maintenance of open habitats (Wakelyn 1987, Dibb et al. 2008) to reduce predation risk along these areas. On mine properties, sheep habitat can be developed by providing escape terrain adjacent to (ideally within 100 m of) high quality forage and by utilizing 3D GIS modeling and including use of wind modeling and solar radiation as a means of predicting snow cover. Appropriate design criteria for reclaimed habitats, as suggested by MacCallum and Geist (1992) for west-central Alberta and adapted to the Elk Valley, should be considered to ensure adequate escape terrain is interspersed with access to forage. Reclamation should include maintenance of highwalls (Jansen et al. 2006). Optimal design of features to enhance forage and security could reduce predation risk within these habitats. Security terrain is especially important for ewes and lambs, which may preferentially reduce predation risk at the expense of forage quality (Bleich et al. 1997). However, successful use of these areas during winter likely depends on snow depths and forage nutrient quality (Stelfox 1976) to provide conditions similar to native, high-elevation wind-swept grassland slopes. Problems could potentially arise if sheep are attracted to areas during winter that contain adequate physical structures but that otherwise lack some of the conditions or resources normally associated with those indicators under less than pristine conditions (i.e., they are ecological traps; Bleich et al. 2009). Reduced survival or reproduction in sheep could result during more severe than normal winters.

Aurora Wildlife Research Page 71

Elk Valley bighorn sheep project

It is unclear whether sheep wintering on native versus mine-altered ranges achieve comparable demographic fitness (i.e., population growth and health). The value of habitat selection models can be enhanced if demographic parameters, including recruitment rates, survival rates, and nutrient availability are assessed and incorporated into more complex models (Bleich et al. 2009). Demographic data from the Elk Valley are generally limited to winter (February to April) surveys. Therefore, limited information has been collected on determining causes and timing of lamb mortality. Surveys or studies at different times of the year (i.e., immediately post lambing and late summer) could provide information on neonatal survival and causes of lamb mortality. Similarly, to improve accuracy of age and sex ratio data, composition surveys could also be conducted during the November rut when rams are more integrated with ewe groups and the sexes are less likely to be segregated (Bleich et al. 1997). These data could then be used to examine the fitness and demographics of sheep populations wintering on native and mine-altered habitats. To provide more robust population estimates, government and mine personnel should at minimum collect the required parameters for the sheep sightability model during surveys (Bodie et al. 1995): activity (moving or not moving) and broad habitat type when first observed. Although not final parameters in the model, we also suggest recording estimated percent snow cover in the general area and percent vegetation cover (perhaps best described as screening cover) around the first animal seen in the group (Unsworth et al. 1998) in case the original model is modified to incorporate these parameters. Population size and sightability correction can be estimated using the Idaho sheep model in program AERIAL SURVEY (Unsworth et al. 1998). Elk numbers within the study area are high, in part because of restrictions to hunting imposed by controlled access onto Teck Coal lands because of safety concerns. Elk numbers and spatial and temporal distribution may be affecting range quality, the ability of winter range to support sheep, and sheep abundance and recruitment (White et al. 2008). The extent of this influence should be quantified, and measures to balance elk and sheep numbers within the study area considered. All species cannot be managed for maximum possible numbers. The mines could review their no hunting area boundaries and amend where appropriate to improve hunter access. High fidelity to winter ranges and the apparent influence of winter severity on survival suggest that disturbance to native winter range resulting from development should be minimized or be conducted in a manner that effectively manages and/or mitigates the impacts. Options for addressing loss of high quality winter range through mitigation may be limited. The degree to which disturbance to or removal of portions of existing winter range will affect either the carrying capacity of the range or population numbers is difficult to quantify, as many factors influence overall population health and survival, including the extent of summer and winter range, forage quantity and quality, forage production and access at other times of the year, winter severity, spring precipitation and temperature, and influence of foraging by competing species on winter ranges. However, it is likely that large scale removal of main high-elevation winter ranges in the absence of effective mitigation measures would result in population decreases, and should be avoided.

Aurora Wildlife Research Page 72

Elk Valley bighorn sheep project

Acknowledgements Volker Scherm (BearAir) piloted the Super Cub during all of the telemetry flights for this project up until June 2010. Sadly, Volker passed away in January 2011. His enthusiasm, expertise, and attention to detail will be sorely missed. This study was a joint effort developed through Teck Coal and BC FLNRO. Teck Coal, BC FLNRO, and the Habitat Conservation Trust Foundation (HCTF) provided primary funding for this project, with additional funding from The Wild Sheep Society and the East Kootenay Wildlife Association (BC Wildlife Federation). Teck Coal and BC FLNRO provided much in-kind support. I thank K. Podrasky (Teck Coal) and I. Teske (BC FLNRO) for administrative and logistic support, and the Elk Valley Bighorn Sheep Committee for their support of the project. Bighorn Helicopters expertly conducted aerial captures, and along with Ascent Helicopters conducted aerial surveys; the survey expertise of G. Goodison was greatly appreciated. BearAir carried out telemetry flights (A. Cyman and V. Scherm), with I. Teske doing the majority of telemetry. D. Lewis and L. Ingham (Fish and Wildlife Compensation Project – Columbia Basin) assisted with ground captures and collar collection, and H. Schwantje (BC FLNRO) provided exceptional veterinarian and capture support throughout the project. I thank the following for assistance with fieldwork or logistics: L. Amos, T. Barr, T. Caldwell, N. Caulkett, D. Charest, T. Chala, A. Fahlman, W. Franklin, K. Honeyman, N. Manklow, S. Medcalf, V. Naude, L. Ozeki, K. Penney, B. Phillips, G. Sword, T. Szkorupa, I. Teske, S. Thiel, J. Thorner, G. Wilson, D. Vasiga, and J. Volp, as well as various Teck staff who provided sightings and support. I thank biologists and GIS specialists in Golder Associates Ltd. and Matrix Solutions Inc. (Alberta) for modifications to the EOSD land cover database. W. Burt (BC FLNRO) skillfully conducted the GIS analyses and extractions, and map production. I am indebted to R. Serrouya, who expertly conducted the RSF habitat analyses. I also greatly appreciated his insights and discussions on the project and his comments on the draft manuscript. Finally, I thank K. Podrasky, G. Sword, and L. Amos (Teck), I. Teske, D. Martin, and G. Mowat (BC FLNRO), and B. MacCallum for comments on earlier drafts of this manuscript.

Literature cited Aebischer, N.J., P.A. Robertson, and R.E. Kenward. 1993. Compositional analysis of habitat use from animal radiotracking data. Ecology 74:1313–1325. Alberta Conservation Association. 2009. WMU 402 Goat Survey – ACA/ASRD D-AUS Report, July 2009. Alberta Conservation Association and Alberta Sustainable Resource Development, Fish and Wildlife, Blairmore, Alberta. Anderson, D.R., and K.P. Burnham. 2002. Avoiding pitfalls when using information–theoretic methods. Journal of Wildlife Management 66:912–918. Anderson, D.R., K.P. Burnham, and W.L. Thompson. 2000. Null hypothesis testing: problems, prevalence, and an alternative. Journal of Wildlife Management 64:912–923. Apps, C.D., B.N. McLellan, T.A. Kinley, and J.P. Flaa. 2001. Scale-dependent habitat selection by mountain caribou, Columbia Mountains, British Columbia. Journal of Wildlife Management 65:65– 77.

Aurora Wildlife Research Page 73

Elk Valley bighorn sheep project

Baccante, N., and R.B. Woods. 2008. Relationship between weather factors and survival of mule deer fawns in the Peace Region of British Columbia. Fish and Wildlife Section, BC Ministry of Environment, Peace Region Technical Report, Fort. St. John, BC. Bartoń, K. 2012. Package ‘MuMIn’. < http://cran.r-project.org/web/packages/MuMIn/MuMIn.pdf> Accessed 2 February 2012. Beswick, S. 1999. Elk Valley bighorn sheep health monitoring project 1998-1999. Unpublished report for Ministry of Environment, Wildlife Divisions, Cranbrook, BC. Bleich, V.C., R.T. Bowyer, and J.D. Wehausen. 1997. Sexual segregation in mountain sheep: resources or predation? Wildlife Monographs 134:1–50. Bleich, V.C., J.H. Davis, J.P. Marshal, S.G. Torres, and B.J. Gonzales. 2009. Mining activity and habitat use by mountain sheep (Ovis canadensis). European Journal of Wildlife Research 55:183–191. Bodie, W.L., E.O. Garton, E.R. Taylor, and M. McCoy. 1995. A sightability model for bighorn sheep in canyon habitats. Journal of Wildlife Management 59:832–840. Boyce, M.S., P.R. Vernier, S.E. Nielsen, and F.K.A. Schmiegelow. 2002 Evaluating resource selection functions. Ecological Modelling 157:281–300. Braumandl, T.F., and M. P. Curran. 1992. A field guide for site identification for the Nelson Forest Region. British Columbia Ministry of Forests, Victoria, British Columbia, Canada. Brown, C.G. 1992. Movement and migration patterns of mule deer in southeastern Idaho. Journal of Wildlife Management 56:246–253. Burles, D.W., and M. Hoefs. 1984. Winter mortality of Dall sheep, Ovis dalli dalli, in Kluane National Park, Yukon. Canadian Field-Naturalist 98:479–484. Burnham, K.P., and D.R. Anderson. 2002. Model selection and multimodel inference: A practical information-theoretic approach. Second edition. Springer, New York, New York, USA. C.E. Jones & Associates. 2007. 2006 annual reclamation research report, Fording River, Greenhills, and Line Creek operations. Prepared for Elk Valley Coal Corporation by C.E. Jones & Associates Ltd., Victoria, BC. Chappel, R.W., and R.J. Hudson. 1978. Winter bioenergetics of Rocky Mountain bighorn sheep. Canadian Jounral of Zoology 56:2388–2393. Cook, J.G., L.L. Irwin, L.D. Bryant, R.A. Riggs, and J.W. Thomas. 1998. Relations of forest cover and condition of elk: a test of the thermal cover hypothesis in summer and winter. Wildlife Monographs 141:1–61. Cubberley, J.C. 2008. Stone’s sheep demographics in the Sulphur / 8 Mile project areas, northern British Columbia, winter 2006/2007. Biennial Symposium of the Northern Wild Sheep and Goat Council 16:106–121. D’Eon, R., and D. Delparte. 2005. Effects of radio-collar position and orientation on GPS radio-collar performance, and the implications of PDOP in data screening. Journal of Applied Ecology 42:383– 388. D’Eon, R.G., and R. Serrouya. 2005. Mule deer seasonal movements and multi-scale resource selection using GPS radiotelemetry. Journal of Mammalogy 86:736–744.

Aurora Wildlife Research Page 74

Elk Valley bighorn sheep project

D’Eon, R.G., R. Serrouya, G. Smith, and C.O. Kochanny. 2002. GPS radiotelemetry error and bias in mountainous terrain. Wildlife Society Bulletin 30:430–439. Daily, T.V., and N.T. Hobbs. 1989. Travel in alpine terrain: energy expenditure for locomotion by mountain goats and bighorn sheep. Canadian Journal of Zoology 67:2368–2375. Davidson, P.W. 1994 East Kootenay Bighorn Sheep enhancement project: Completion report. Unpublished report prepared for B.C. Ministry of Environment, Cranbrook, British Columbia. 183 pp. DeCesare, N.J, and D.H. Pletscher. 2004. Bighorn sheep, horizontal visibility, and GIS. Proceedings of the Biennial Symposium of the Northern Wild Sheep and Goat Council 14:181–190. DeCesare, N.J, and D.H. Pletscher. 2006. Movements, connectivity, and resource selection of Rocky Mountain bighorn sheep. Journal of Mammalogy 87:531–538. DelGiudice, G.D., R.A. Moen, F.J. Singer, and M.R. Riggs. 2001. Winter nutritional restrictions and simulated body condition of Yellowstone elk and bison before and after the fires of 1988. Wildlife Monographs 147:1–60. Demarchi, D.A. 1968. An ecological study of Rocky Mountain bighorn sheep winter ranges in the East Kootenay Region of British Columbia. Progress Report. Fish and Wildlife Branch, Victoria, BC. 51pp. Demarchi, R.A., C.L. Hartwig, and D.A. Demarchi. 2000. Status of the Rocky Mountain bighorn sheep in British Columbia. BC Ministry of Environment, Lands and Parks, Wildlife Branch, Victoria, BC. Wildlife Bulletin No. B-99. 56 pp. Dibb, A., M.A. Tremblay, and M.S. Quinn. 2006. Modelling and management of bighorn sheep movement corridors. Biennial Symposium of the Northern Wild Sheep and Goat Council 16:221– 247. Dicus, G.H. 2002. An evaluation of GIS-based habitat models for bighorn sheep winter range in Glacier National Park, Montana, USA. M.S. thesis, University of Montana, Missoula. Environment Canada. 2011. Canadian climate normals 1971–2000. Environment Canada, Ottawa, Ontario. < http://climate.weatheroffice.gc.ca/climate_normals/index_e.html> Accessed 19 December 2011. Festa-Bianchet, M. 1986a. Seasonal dispersion of overlapping mountain sheep ewe groups. Journal of Wildlife Management 50:325–330. Festa-Bianchet, M. 1986b. Site fidelity and seasonal range use by bighorn rams. Canadian Journal of Zoology 64:2126–2132. Festa‐Bianchet, M. 1988a. Seasonal range selection in bighorn sheep: conflicts between forage quality, forage quantity, and predator avoidance. Oecologia 75: 580‐586. Festa-Bianchet, M. 1988b. Birth date and survival in bighorn lambs (Ovis canadensis). Journal of Zoology (London) 214:653–661. Festa-Bianchet, M. 1989. Survival of male bighorn sheep in southwestern Alberta. Journal of Wildlife Management 53:259–263. Forsyth, D. 2000. Habitat selection and coexistence of the Alpine chamois (Rupicapra rupicapra) and Himalayan tahr (Hemitragus jemlahicus) in the eastern Southern Alps, New Zealand. Journal of Zoology (London) 252:215–225.

Aurora Wildlife Research Page 75

Elk Valley bighorn sheep project

Garshelis D.L., M.L. Gibeau, and S. Herrero. 2005. Grizzly bear demographics in and around Banff National Park and Kananaskis Country, Alberta. Journal of Wildlife Management 69:277–297. Geist, V. 1971. Mountain sheep, a study in behaviour and evolution. The University of Chicago Press. 383 pp. GeoBC. 2010. Freshwater water atlas user guide, 12 January 2010. Prepared for Integrated Land Management Bureau, Victoria, BC., by Caslys Consulting Ltd., Saanichton, BC. Accessed 4 January 2012. Gillies, C.S., M. Hebblewhite, S.E. Nielsen, M.A. Krawchuk, C.L. Aldridge, J.L. Frair, D.J. Saher, C.E. Stevens, and C.L. Jerde. 2006. Application of random effects to the study of resource selection by animals. Journal of Animal Ecology 75:887–898. Girard, I., J. Ouellet, R. Courtois, C. Dussault, and L. Breton. 2002. Effects of sampling effort based on GPS telemetry on home‐range size estimations. Journal of Wildlife Management 66:1290–1300. Goodson, N.J., D.R. Stevens, and J.A. Bailey. 1991. Winter-spring foraging ecology and nutrition of bighorn sheep on montane ranges. Journal of Wildlife Management 55:422–433. Gross, J.E., M.C. Kneeland, D.F. Reed, and R.M. Reich. 2002. GIS-based habitat models for mountain goats. Journal of Mammalogy 83:218–228. Hayne, D.W. 1949. Calculation of size of home range. Journal of Mammalogy 30:1–18. Heard, D.C., K.L. Zimmerman, L.L. Yaremko, and G.S. Watts. 1999. Moose population estimate for the Parsnip River drainage, January 1998. Final report for Forest Renewal British Columbia, Project No. OP96004. Hebblewhite, M., and E.H. Merrill. 2011. Demographic balancing of migrant and resident elk in a partially migratory population through forage–predation tradeoffs. Oikos 120:1860–1870. Hebblewhite, M., M. Percy, and E.H. Merrill. 2007. Are all global positioning system collars created equal? Correcting habitat-induced bias using three brands in the central Canadian Rockies. Journal of Wildlife Management 71:2026–2033. Higgs, T. 2004. Trace element deficiencies in sheep and cattle, Department of Agriculture, Western Australia Farmnote 8/2004. Hooge, P.N, and B. Eichenlaub. 2000. Animal movement extension to ArcView. Version 2.0. Alaska Biological Science Center, U.S. Geological Survey, Anchorage, Alaska. Hurlbert, S.H. 1984. Pseudoreplication and the design of ecological field experiments. Ecological Monographs 54:187–211. Jalkotzy, M. 2000. Habitat selection by bighorn ewes on three winter ranges in the East Kootenays. Unpublished report for Columbia Basin Fish & Wildlife Compensation Program, produced by the East Kootenay Wildlife Association and Arc Wildlife Services Ltd. Jansen, B.D., P.R. Krausman, J.R. Heffelfinger, and J.C. Devos, Jr. 2006. Bighorn sheep selection of landscape features in an active copper mine. Wildlife Society Bulletin 34:1121–1126. Jansen, B.D., P.R. Krausman, J.R. Heffelfinger, and J.C. Devos, Jr. 2007. Influence of mining on behavior of bighorn sheep. The Southwestern Naturalist 52:418–423.

Aurora Wildlife Research Page 76

Elk Valley bighorn sheep project

Jokinen, M., D. Dorge, and P. Jones. 2007. Bighorn sheep survival and demography in the Yarrow‐Castle region of Alberta, Canada. Technical Report, T‐2007‐003, produced by Alberta Conservation Association, Blairmore, Alberta, Canada. 60 pp. + App. Johnson, B.K., P.K. Coe, and R.L. Green. 2013. Abiotic, bottom-up, and top-down influences on recruitment of Rocky Mountain elk in Oregon: A retrospective analysis. Journal of Wildlife Management 77: 102–116. Johnson, C.J., D.R. Seip, and M.S. Boyce. 2004. A quantitative approach to conservation planning: using resource selection functions to map the distribution of mountain caribou at multiple spatial scales. Journal of Applied Ecology 41:238–251. doi:10.1111/j.0021-8901.2004.00899.x. Johnson, D.H. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology 61:65–71. Jorgenson, J.T., M. Festa-Bianchet, J.-M. Gaillard, and W.D. Wishart. 1997. Effects of age, sex, disease, and density on survival of bighorn sheep. Ecology 78:1019–1032. Kie, J.G., J. Matthiopoulos, J. Fieberg, R.A. Powell, F. Cagnacci, M.S. Mitchell, J.-M. Gaillard, and P.R. Moorcroft. 2010. The home-range concept: are traditional estimators still relevant with modern telemetry technology? Philosophical Transactions of the Royal Society B 365:2221–2231. Kenney, A.J., and C.J. Krebs. 2002. Ecological methodology, version 6.1. [computer program]. Department of Zoology, University of British Columbia, Vancouver, British Columbia. Kernohan, B.J., R.A. Gitzen, and J.J. Millspaugh. 2001. Analysis of animal space use and movements. Pages 125–166 in Radio tracking and animal populations. J.J. Millspaugh and J.M. Marzluff, (eds). Academic Press, San Diego, CA. Kinley, T.A. (compiler). 2007. Rocky Mountain bighorn sheep habitat and population assessment for the East Kootenay Trench. Prepared for the East Kootenay Wildlife Association. Krausman, P.R., and R.T. Bowyer. 2003. Mountain sheep (Ovis canadensis and O. dalli). Pages 1095– 1115 in Wild mammals of North America: biology, management, and conservation. G.A. Feldhamer, B.C. Thompson, and J.A. Chapman, (eds). Baltimore: The John Hopkins University Press. Krebs, C.J. 1999. Ecological methodology, 2nd edition. Addison-Welsey Educational Publishers, Inc., Menlo Park, CA. Kumar, L., A.K. Skidmore, and E. Knowles. 1997. Modelling topographic variation in solar radiation in a GIS environment. International Journal of Geographic Information Science 11:475–497. Lemke, S.L., and H. M. Schwantje. 2005. British Columbia bighorn sheep trace minerals levels: a compilation of existing data. Prepared for Habitat Conservation Trust Fund and Wild Sheep Society of British Columbia. Lovari, S., and R. Cosentino. 1986. Seasonal habitat selection and group size of the Abruzzo chamois (Rupicapra pyrenaica ornata). Bolletino di zoologia 53:73–78. MacCallum, N.B. 1991. Bighorn sheep use of an open pit coal mine in the Foothills of Alberta. M. Sc. thesis, University of Calgary, Alberta. MacCallum, B. 1997. Bighorn sheep interaction with alpine habitat and reclaimed lands in Alberta’s Rocky Mountains. The Nikanassin Range bighorn sheep study. Prepared by Bighorn Environmental Design Ltd. for Alberta Department of Environmental Protection, Natural Resources Service, Wildlife

Aurora Wildlife Research Page 77

Elk Valley bighorn sheep project

Management Division; Cardinal River Coals Ltd.; Department of Canadian Heritage, Parks Canada, Jasper National Park; and Gregg River Resources Ltd. MacCallum, B., and V. Geist. 1992. Mountain restoration: soil and surface wildlife habitat. GeoJournal 27:23–46. McKinney, T., S. R. Boe, and J. C. deVos, Jr. 2003. GIS-based evaluation of escape terrain and desert bighorn sheep populations in Arizona. Wildlife Society Bulletin 31:1229–1236. Main, M.B., F.W. Weckerly,and V.C. Bleich. 1996. Sexual segregation in ungulates: new directions for research. Journal of Mammalogy 77:449–461. Manly, B.F.J., L.L. McDonald, D.L. Thomas, T.L. Mcdonald, and W.P. Erickson. 2002. Resource selection by animals: statistical design and analysis for field studies, second edition. Kluwer Academic Publishers, Dordrecht, The Netherlands. Meidinger, D., and J. Pojar. 1991. Ecosystems of British Columbia. Special Report Series 6, B.C. Ministry of Forests, Research Branch, Victoria, B.C., Canada. Menard, S. 2004. Six approaches to calculating standardized logistic regression coefficients. American Statistician 58:218– 223. Millspaugh, J.J., J.R. Skalski, B.J. Kernohan, K.J. Raedeke, G.C. Brundige, and A.B. Cooper. 1998. Some comments on spatial independence in studies of resource selection. Wildlife Society Bulletin 26:232–236. Morellet, N., H. Verheyden, J.-M. Angibault, B. Cargnelutti, B. Lourtet, and M.A.J. Hewison. 2009. The effect of capture on ranging behaviour and activity of the European roe deer Capreolus capreolus. Wildlife Biology 15:278–287. Morgantini, L.E., and R.J. Hudson. 1981. Sex differential in use of the physical environment by bighorn sheep (Ovis canadensis). Canadian Field-Naturalist 95:69–74. Murtaugh, P.A. 2007. Simplicity and complexity in ecological data analysis. Ecology 88:56–62. Mysterud, A. 1999. Seasonal migration pattern and home range of roe deer (Capreolus capreolus) in an altitudinal gradient in southern Norway. Journal of Zoology 247:479–486. Namgail, T. 2006. Winter habitat partitioning between Asiatic ibex and blue sheep in Ladakh, Northern India. Journal of Mountain Ecology 8:7–13. Natural Resources Canada. 2007. GeoBase - Canadian Digital Elevation Data, Level 1 Product Specifications. Edition 3. Government of Canada, Natural Resources Canada, Sherbrooke, Quebec. Accessed 2 November 2012. Neal, A.K., G.C. White, R.B. Gill, D.F. Reed, and J.H. Olterman. 1993. Evaluation of mark-resight Model assumptions for estimating mountain sheep numbers. Journal of Wildlife Management 57:436–450. Nicholson, M.C., R.T. Boyer, and J.G. Kie. 1997. Habitat selection and survival of mule deer: tradeoffs associated with migration. Journal of Mammalogy 78:483–504. Nielsen, S.E., M.S. Boyce, G.B. Stenhouse, and R.H.M. Munro. 2002. Modeling grizzly bear habitats in the Yellowhead Ecosystem of Alberta: taking autocorrelation seriously. Ursus 13:45–56. Oehler, M.W., V.C. Bleich, R.T. Bowyer, and M.C. Nicholson. 2005. Mountain sheep and mining: implications for conservation and management. California Fish and Game 91:149–178.

Aurora Wildlife Research Page 78

Elk Valley bighorn sheep project

Otis, D.L., and G.C. White. 1999. Autocorrelation of location estimates and the analysis of radiotracking data. Journal of Wildlife Management 63:1039–1044. Parker, K.L., C.T. Robbins, and T.A. Hanley. 1984. Energy expenditures for locomotion by mule deer and elk. Journal of Wildlife Management 48:474–488. Parker, K.L., P.S. Barboza, and M.P. Gillingham. 2009. Nutrition integrates environmental responses of ungulates. Functional Ecology 23:57–69. Pauley, G.R., J.M. Peek, and P. Zager. 1993. Predicting white-tailed deer habitat use in northern Idaho. Journal of Wildlife Management 57:904–913. Petersburg, M.L., A.W. Alldredge, and W.J. de Vergie. 2000. Emigration and survival of 2-year-old male elk in northwestern Colorado. Wildlife Society Bulletin 28:708–716. Pollock, K.H., S.R. Winterstein, C.M. Bunck, and P.D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7–15. Poole, K.G., R. Serrouya, and K. Stuart-Smith. 2007. Moose calving strategies in interior montane ecosystems. Journal of Mammalogy 88:139–150. Poole, K.G., K. Stuart-Smith, and I.E. Teske. 2009. Wintering strategies by mountain goats in interior mountains. Canadian Journal of Zoology 87:273–283. Portier, C., M. Festa-Bianchet, J.-M. Gaillard, J.T. Jorgenson, and N.G. Yoccoz. 1998. Effects of density and weather on survival of bighorn sheep lambs (Ovis canadensis). Journal of Zoology, London 245:271–278. Puls, R. 1994. Mineral levels in animal health. Diagnostic data. 2nd edition. Sherpa International, Clearbrook, BC, Canada. R Development Core Team (2008). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R- project.org. Resources Inventory Committee. 1999. Standards for Predictive Ecosystem Mapping, Inventory Standard. Version 1.0. November 1999. Terrestrial Ecosystem Mapping Alternatives Task Force, Victoria, BC. Available at http://www.env.gov.bc.ca/fia/pem.htm accessed 4 January 2012. RISC (Resources Information Standards Committee). 2002. Aerial-based inventory methods for selected ungulates: bison, mountain goat, mountain sheep, moose, elk, deer and caribou. Standards for components of British Columbia’s biodiversity No. 32. Version 2.0. Resources Inventory Committee, B.C. Ministry of Sustainable Resource Management, Victoria, British Columbia. Risenhoover, K.L., and J.A. Bailey. 1985. Foraging ecology of mountain sheep: implications for habitat management. Journal of Wildlife Management 49:797–804. Rodgers, A.R., and A.P. Carr. 1998. HRE: The Home Range Extension for ArcView™: User’s manual. Beta Test Version 0.9, July 1998. Centre for Northern Ecosystem Research, Ontario Ministry of Natural Resources, Thunder Bay, Ontario, Canada. Rubin, E.S., W.M. Boyce, C.J. Stermer, and S.G. Torres. 2002. Bighorn sheep habitat use and selection near an urban environment. Biological Conservation 104:251–263. Sappington, J.M., K.M. Longshore, and D.B. Thompson. 2005. Quantifying landscape ruggedness for animal habitat analysis: a case study using bighorn sheep in the Mojave Desert. Journal of Wildlife Management 71:1419–1426

Aurora Wildlife Research Page 79

Elk Valley bighorn sheep project

SAS Institute. 2004. SAS/STAT user’s guide. SAS Institute, Cary, North Carolina, USA. Seaman, D.E., and R.A. Powell. 1996. An evaluation of the accuracy of kernel density estimators for home range analysis. Ecology 77:2075–2085. Shackleton, D.M., C.C. Shank, and B.M. Wikeem. 1999. Natural history of Rocky Mountain and California Bighorn Sheep. Pages 78–138 in R. Valdez and P.R. Krausman, eds. Mountain sheep in North America. Univ. Arizona Press, Tucson, AZ. Schuerholz, G. 1984. Annual ecological cycles of Rocky Mountain bighorn sheep populations with high elevation winter ranges in the East Kootenays. Proceedings of the 4th Biennial Symposium, Northern Wild Sheep and Goat Council 4:192–210. Singer, F.J., E. Williams, M.W. Miller, and L.C. Zeigenfuss. 2000. Population growth, fecundity, and survivorship in recovering populations of bighorn sheep. Restoration Ecology 8(4S):75–84. Smith, T.S., P.J. Hardin, and J.T. Flinders. 1999. Response of bighorn sheep to clear-cut logging and prescribed burning. Wildlife Society Bulletin 27:840–845. Sokal, R.R., and F.J. Rohlf. 1981. Biometry. Second edition. W.H. Freeman and Co., New York. Stelfox, J.G. 1976. Range ecology of Rocky Mountain bighorn sheep. Canadian Wildlife Service Report Series No. 39. Environment Canada, Ottawa, ON. Summit Environmental Consultants Inc. 2012. Teck Coal Limited Elk Valley bighorn sheep habitat study. Project: 2010-8044.000. Sweanor, P.Y., M. Gudorf, and F.J. Singer. 1996. Application of a GIS-based bighorn sheep habitat model in Rocky Mountain Region National Parks. Biennial Symposium of the Northern Wild Sheep and Goat Council 10:118–125. Tabachnick, B.G., and L.S. Fidell. 1996. Using multivariate statistics. HarperCollins, New York, New York, USA. TAESCO. 1985a. Line Creek Biophysical Inventories - Volume 3. Wildlife. Unpublished report prepared for Crows Nest Resources Ltd. [Line Creek Resources] by TAESCO Environmental Consultants Ltd., Duncan, BC. TAESCO. 1985b. Line Creek Biophysical Inventories - Volume 3. Wildlife – Addendum I: Radiotelemetry. Unpublished report prepared for Crows Nest Resources Ltd. [Line Creek Resources] by TAESCO Environmental Consultants Ltd., Duncan, BC. Teske, I., and B. Forbes. 2002. East Kootenay Rocky Mountain sheep inventory: winter 2001 and 2002. Unpublished report, Ministry of Water, Land and Air Protection, Environmental Stewardship Division, Cranbrook, B.C. Tilton, M.E., and E.E. Willard. 1982. Winter habitat selection by mountain sheep. Journal of Wildlife Management 46:359–366. Underwood, E.J., and N.F. Suttle. 1999. The mineral nutrition of livestock, 3rd edition. CABI Publishing, Wallingford, Oxon, U.K. 614 pp. Unsworth, J.W., F.A. Leboan, E.O. Garton, D.J. Leptich, and P. Zager. 1998. Aerial survey: user’s manual. Electronic edition. Idaho Department of Fish and Game, Boise, Idaho. Vore, J.M., and E.M. Schmidt. 2001. Movements of female elk during calving season in northwest Montana. Wildlife Society Bulletin 29:720–725.

Aurora Wildlife Research Page 80

Elk Valley bighorn sheep project

Wakelyn, L.A. 1987. Changing habitat conditions on bighorn sheep ranges in Colorado. Journal of Wildlife Management 51:904–912. White, G.C. 1996. NOREMARK: population estimation from mark-resighting surveys. Wildlife Society Bulletin 24:50–52. White, G.C., and R.A. Garrott. 1990. Analysis of wildlife radio-tracking data. Academic Press, Inc., San Diego, California, USA, 383 pp. White, P.J., T.O. Lemke, D.B. Tyers, and J.A. Fuller. 2008. Initial effects of reintroduced wolves Canis lupus on bighorn sheep Ovis canadensis dynamics in Yellowstone National Park. Wildlife Biology 14:138–146. White, K.S., G.W. Pendleton, D.Crowley, H.J. Griese, K.J. Hundertmark, T. Mcdonough, L. Nichols, M. Robus, C.A. Smith, and J.W. Schoen. 2011. Mountain goat survival in coastal Alaska: Effects of age, sex, and climate. Journal of Wildlife Management 75:1731–1744. Worton, B.J. 1989. Kernel methods for estimating the utilization distribution in home-range studies. Ecology 70:164–168. Wulder, M.A., J.C. White, M. Cranny, R.J. Hall, J.E. Luther, A. Beaudoin, D.G. Goodenough, and J.A. Dechka. 2008. Monitoring Canada’s forests. Part 1: Completion of the EOSD land cover project. Canadian Journal of Remote Sensing 34:549–562. Zeigenfuss, L.C., F.J. Singer, and M.A. Gudorf. 2000. Test of a modified habitat suitability model for bighorn sheep. Restoration Ecology 8(4S):38–46.

Aurora Wildlife Research Page 81

Elk Valley bighorn sheep project

Appendix 1. Capture data for bighorn sheep, Elk Valley, May 2009 – October 2010.

IDNo WH ID Winter Range Date UTMEast UTMNorth Elev (ft) Sex Age Ram Cl GPS No. Frequency L tag R tag colour 01F 2009-1985 Erickson Ridge 26-Feb-09 659278 5507517 4900 F 3 15387 148.100 49 Orange 02F 2009-1978 Erickson Ridge 26-Feb-09 659435 5507141 4700 F 4 15390 148.132 51 Orange 03M 2009-1980 Erickson Ridge 26-Feb-09 662892 5507185 6900 M 3 1 15416 148.501 53 Orange 04M 2009-1986 Sheep Mt 26-Feb-09 659529 5524843 7000 M 2 1 15413 148.492 59 Orange 05F 2009-1988 Sheep Mt 26-Feb-09 659347 5525160 7500 F 5 15396 148.221 06F 2009-1983 Sheep Mt 26-Feb-09 659513 5524832 7000 F 5 15384 148.062 56 Orange 07M 2009-1982 Ewin Ridge 26-Feb-09 660008 5542419 6900 M 4 2 15402 148.332 55 Orange 08M 2009-1984 Ewin Ridge 26-Feb-09 660008 5542419 6900 M 3 1 15412 148.472 57 Orange 09F 2009-1987 Ewin Ridge 26-Feb-09 660480 5543234 8000 F 6 15391 148.142 60 Orange 10F 2009-1990 Ewin Ridge 26-Feb-09 661290 5542035 8400 F 5 15380 148.023 63 Orange 11M 2009-1989 Ewin Ridge 26-Feb-09 659002 5544458 7900 M 3 15415 148.342 62 Orange 12F 2009-1981 Todhunter 26-Feb-09 658594 5552760 7500 F 6 15398 148.262 54 Orange 13F 2009-1977 Todhunter 26-Feb-09 658930 5552818 7500 F 4 15393 148.162 14M 2009-1976 Todhunter 26-Feb-09 657993 5552485 6500 M 2 15408 148.432 15F 2009-1979 Ewin Ridge 26-Feb-09 661239 5541593 8000 F 3 15392 148.151 52 Orange 16M 2009-1992 Todhunter 27-Feb-09 661417 5552966 7000 M 2 1 15400 148.312 65 Orange 17M 2009-1997 Chauncey 27-Feb-09 655386 5558022 7200 M 2 1 15410 148.451 70 Orange 18M 2009-1998 Chauncey 27-Feb-09 655386 5558022 7200 M 2 1 15411 148.460 71 Orange 19F 2009-2000 Brownie 27-Feb-09 658258 5564363 7700 F 6 15389 148.120 73 Orange 20F 2009-2008 Brownie 27-Feb-09 658209 5564304 7600 F 6 15395 148.211 81 Orange 21F 2009-2009 Henretta 27-Feb-09 655914 5569285 7300 F 5 15382 148.042 82 Orange 22F 2009-2001 Henretta 27-Feb-09 655956 5569074 7000 F 3 15383 148.054 74 Orange 23F 2009-1995 Henretta 27-Feb-09 655956 5569074 7000 F 6 15386 148.092 68 Orange 24M 2009-1996 Chauncey 27-Feb-09 657550 5556550 7000 M 2 1 15414 148.273 69 Orange 25F 2009-2002 Greenhills S 27-Feb-09 652941 5550837 7000 F 4 15379 148.011 75 Orange 26F 2009-2003 Greenhills S 27-Feb-09 652941 5550837 7000 F 4 15394 148.182 76 Orange 27M 2009-2011 Greenhills S 27-Feb-09 652874 5551079 7100 M 1 1 15407 148.422 84 Orange 28F 2009-2004 Greenhills S 27-Feb-09 652446 5554826 6900 F 6 15385 148.072 77 Orange 29M 2009-2010 Ewin Ridge 27-Feb-09 660607 5542420 7600 M 3 2 15399 148.304 83 Orange 30M 2009-1999 Ewin Ridge 27-Feb-09 659852 5542909 7600 M 2 1 15401 148.323 72 Orange 31M 2009-2005 Chauncey 27-Feb-09 657160 5557124 7200 M 2 1 15403 148.360 78 Orange 32F 2009-1994 Brownie 27-Feb-09 658473 5564655 8200 F 6 15381 148.031 67 Orange 33F 2009-1993 Henretta 27-Feb-09 655721 5569229 8200 F 6 15397 148.253 99 Orange 34M 2009-2007 Sheep Mt 28-Feb-09 659555 5524887 7100 M 2 1 15404 148.382 80 Orange 35M 2009-2006 Brownie 28-Feb-09 658974 5563895 7200 M 4 2 15417 148.522 79 Orange 36M 2009-2013 Greenhills S 28-Feb-09 656083 5546469 5900 M 3 2 15418 148.553 152 Yellow 37M 2009-2015 Greenhills S 28-Feb-09 656083 5546769 6000 M 5 2 15405 148.401 154 Yellow 38M 2009-2017 Greenhills N 28-Feb-09 647998 5559676 6700 M 3 2 15409 148.442 39M 2009-1991 Greenhills N 28-Feb-09 648189 5560003 7400 M 3 2 15406 148.414 40M 2009-2014 Line Creek 27-May-09 660955 5534656 5200 M 2 1 15412 148.472 153 Yellow 41M 2009-2012 Line Creek 27-May-09 661164 5534899 5400 M 5 3 15413 148.492 151 Yellow 42F 2009-103 Line Creek 25-Aug-09 661450 5534840 5800 F 3 15392 148.151 103 Orange

Aurora Wildlife Research Page 82

Elk Valley bighorn sheep project

IDNo WH ID Winter Range Date UTMEast UTMNorth Elev (ft) Sex Age Ram Cl GPS No. Frequency L tag R tag colour 43F 2009-104 Line Creek 25-Aug-09 661587 5534808 5600 F 4 15388 148.111 104 Orange 44M 2009-101 Line Creek 25-Aug-09 660935 5534646 5200 M 4 2 15413 148.492 45M Greenhills 17-Dec-09 652481 5554202 M 2 1 15409 148.442 46M Chauncey 17-Dec-09 656520 5558007 M 3 2 15406 148.414 47M 2010-105 Line Creek 18-May-10 661140 5525630 5300 M 5 3 15414 148.273 105 Orange 48M 2010-106 Line Creek 18-May-10 659630 5535680 5500 M 3 2 15385 148.072 102 Orange 49F Line Creek 19-Aug-10 658665 5534526 6000 F 3 15389 148.120 106 Orange 50M Line Creek 25-Oct-10 658309 5535183 M 3 2 15413 148.492

IDNo WH ID R Horn length R Horn base L Horn L Horn base Neck (cm) Chest (cm) Total len (cm) Hair Punch Blood Pellets 01F 2009-1985 length15.0 11.5 45 107 147 Y Y Y Y 02F 2009-1978 21.0 13.0 21.0 13.0 50 108 162 Y Y Y Y 03M 2009-1980 41.0 24.0 52 106 152 Y N Y Y 04M 2009-1986 47.5 38.5 46.5 28.0 48 106 162 Y Y Y Y 05F 2009-1988 20.5 12.5 21.2 42 116 155 Y Y Y Y 06F 2009-1983 24.5 25.0 13.8 44 110 180 Y Y Y Y 07M 2009-1982 60.5 36.0 60.5 36.0 52.5 124 184 Y Y Y Y 08M 2009-1984 56.0 32.0 115 150 Y Y Y Y 09F 2009-1987 24.5 24.0 13.5 44 114 163 Y Y Y N 10F 2009-1990 23.5 12.7 22.0 12.5 42 114 173 Y Y Y Y 11M 2009-1989 46.0 28.5 45.5 27.4 45 108 168 Y Y Y Y 12F 2009-1981 25.5 13.5 25.0 14.5 45.5 118 162 Y Y Y Y 13F 2009-1977 23.0 13.5 46.5 116 155 Y N Y N 14M 2009-1976 32.0 22.0 31.0 21.5 44.5 100 157 Y Y Y Y 15F 2009-1979 13.0 48 124 170 N Y Y Y 16M 2009-1992 41.0 27.0 41.0 26.0 49 118 172 Y Y Y Y 17M 2009-1997 50.0 32.0 49 112 162 Y Y Y Y 18M 2009-1998 43.0 26.0 124 152 Y Y Y Y 19F 2009-2000 30.0 13.5 29.5 13.5 47 111 161 Y Y Y Y 20F 2009-2008 22.0 21.5 15.0 48 114 169 Y Y Y Y 21F 2009-2009 25.0 13.0 26.0 13.0 46 116 164 Y Y Y Y 22F 2009-2001 24.0 13.2 24.0 13.2 48 126 166 Y Y Y Y 23F 2009-1995 25.0 13.0 25.5 13.0 43.5 110 159 Y Y Y Y 24M 2009-1996 50.0 30.7 51.0 30.5 50 120 162 Y Y Y Y 25F 2009-2002 25.0 13.0 25.0 13.5 46 114 176 Y Y Y Y 26F 2009-2003 25.0 14.0 25.2 14.5 41 116 172 Y Y Y Y 27M 2009-2011 27.0 20.0 28.0 20.0 50.5 102 162 Y Y Y Y 28F 2009-2004 30.0 13.5 30.5 13.0 49 114 166 Y Y Y Y 29M 2009-2010 51.5 33.0 53.0 34.0 51 120 177 Y Y Y Y 30M 2009-1999 49.0 34.0 51 124 174 Y Y Y Y 31M 2009-2005 46.0 30.0 46.0 30.0 60 116 167 Y Y Y Y 32F 2009-1994 29.0 16.0 28.0 16.0 50 132 165 N Y Y Y 33F 2009-1993 28.0 16.0 27.0 16.0 50 125 162 Y Y Y Y 34M 2009-2007 58.5 35.5 59.0 35.7 51 108 182 Y Y Y Y

Aurora Wildlife Research Page 83

Elk Valley bighorn sheep project

IDNo WH ID R Horn length R Horn base L Horn L Horn base Neck (cm) Chest (cm) Total len (cm) Hair Punch Blood Pellets 35M 2009-2006 81.0 41.5 length80.6 41.8 58 119 195 Y Y Y Y 36M 2009-2013 69.5 39.8 70.5 40.7 53 130 197 Y Y N Y 37M 2009-2015 80.0 38.0 69 136 184 Y N Y N 38M 2009-2017 52.0 35.5 54.5 37.0 116 Y N N N 39M 2009-1991 Y Y N Y 40M 2009-2014 32.5 23.5 43 89 152 Y Y Y Y 41M 2009-2012 76.0 38.0 76.2 38.5 52 119 186 Y Y Y Y 42F 2009-103 14.5 12.0 17.0 12.0 35 92 142 Y Y Y Y 43F 2009-104 21.5 13.5 21.0 13.5 41 106 158 Y Y Y Y 44M 2009-101 58.0 35.0 58.5 35.0 49 109 178 Y N Y Y 45M Y N N Y 46M Y N N N 47M 2010-105 71.0 39.0 62 132 178 Y Y Y Y 48M 2010-106 46.0 30.0 47 99 122 Y N Y Y 49F 11.0 11.0 42 92 147 Y Y Y Y 50M Y N N Y

Aurora Wildlife Research Page 84