Migration, movements, and survival in a partially migratory elk (Cervus canadensis) population

Migration provides an adaptive strategy to improve fitness by allowing individuals to exploit gradients of resources and changes in predation risk. In recent decades, the extent and prevalence of migration has declined in numerous ungulate species including many populations of elk (Cervus canadensis). Resident elk are often more closely associated with human activity, are more readily involved in agricultural conflicts and may contribute to overgrazing on some ranges. We evaluated migratory trends, survival rates, and causes of mortality in a partially migratory elk population in southeastern British Columbia, Canada, and compared these parameters with data from a 1982–1996 study from the same area. Analysis of 201 animal‐years (n = 78 collared cow elk) between 2016 and 2022 showed a ratio of 52% migratory to 48% resident elk, similar to what was found during the 1982–1996 study (55% migratory to 45% resident; n = 40 cows). Among the migrants, 55% were standard migrants (traveling moderate to longer distances and changing little in elevation), 35% elevational, and 10% mixed/atypical. We detected a 14% yearly switching rate between migratory and resident strategies. We recorded 30 mortalities: 47% from human causes, predominantly elk‐vehicle/train collisions (33% of mortalities), apparent starvation/old age (17%), and predation (17%). Notably, while mortality from natural causes was similar between strategies, human‐caused mortality was nearly twice as high in resident elk. Signs of nutritional stress and lower pregnancy rates indicated potential forage limitations. Migrants had higher average survival rates (0.90) compared to residents (0.83), a shift from the 1982–1996 study that recorded higher resident survival rates (0.98), the same migrant survival rate (0.90), and fewer elk‐vehicle/train collisions. Cow elk in our study made fewer and shorter movements into upper mountain tributaries and greater use of mine properties than observed during the 1982–1996 study. Wildlife managers should consider opportunities to enhance elk forage within traditional high‐elevation summer ranges and mitigations to reduce elk‐vehicle/train collisions. Further research is needed to quantify reproductive success, monitor calf survival, and determine relative forage quality among summering areas between migratory strategies, and determine winter ranges of long‐distant migrants.

study.Wildlife managers should consider opportunities to enhance elk forage within traditional high-elevation summer ranges and mitigations to reduce elk-vehicle/train collisions.Further research is needed to quantify reproductive success, monitor calf survival, and determine relative forage quality among summering areas between migratory strategies, and determine winter ranges of long-distant migrants.
Migration is complex in many populations of ungulates, and it is rare that all members of a population follow the same strategy.Many populations are partially migratory (Barker, Mitchell, Proffitt, & Devoe, 2019;Berg et al., 2019;Chapman et al., 2011), composed of longdistance migrants, short-distance migrants, and residents who largely forego migration.Partial migration may be a result of adjusting density-dependent fitness when resources are limited (Lundberg, 1988) and may indicate population-level demographic balancing (Hebblewhite & Merrill, 2011), but the benefits to migration are not always clear (Peterson et al., 2022).Flexibility in migratory behavior may be more common than previously thought (Berg et al., 2019).
In recent decades, the extent and prevalence of migration have declined in many populations of ungulates (Berger, 2004;Bolger et al., 2008;Hebblewhite et al., 2006;Kauffman et al., 2018).Overhunting, anthropogenic barriers in the landscape (e.g., fences and highways), and habitat loss have contributed to the decline of ungulate migrations in many areas (Bolger et al., 2008).Human-caused mortality differentially applied to residents or migrants may alter population growth in favor of one migratory strategy over another (Monteith et al., 2014).Semi-permeable barriers arising from industrial development and other anthropogenic sources can result in detours around established routes, shifts in timing and rate of migration, and decreased foraging during pauses in migration (Lendrum et al., 2013;Sawyer et al., 2013;Seidler et al., 2014).Decreases in native forage quality at high elevations, such as through drought, forest maturation, or over-grazing, may lower fat reserves and pregnancy rates of females and/or increase predation rates, resulting in lower recruitment for migrants (Cole et al., 2015;Hebblewhite & Merrill, 2011;Proffitt et al., 2016).Other factors may contribute in more subtle ways.For example, climate change can cause a shift in the timing, quantity, and duration of the spring peak in forage nutritional quality and spatial variation in plant phenology (Mysterud, 2013).
Migration usually helps ungulates escape predation by moving beyond predator foraging range and reducing vulnerability to regulation (Berg et al., 2023;Fryxell et al., 1988;Fryxell & Sinclair, 1988).However, predator densities and associated risks during migration can increase through recolonizing predator populations.These factors coupled with reduced predator numbers in areas of high human use (Berger, 2007) and rich forage from irrigated crops in lower-elevation habitats associated with humans (Middleton, Kauffman, Mcwhirter, Cook, et al., 2013) can reduce the productivity of migrants compared with residents.In addition, migration may in part be density-dependent as a result of intraspecific competitive release, where migrants may switch to a resident strategy at low abundance and residents may switch to a migrant strategy at high abundance, resulting in increased proportion of migrants with higher density on shared range (Berg et al., 2019;Eggeman et al., 2016;Mysterud et al., 2011).Switching migratory behavior may also increase resiliency to changing environmental and anthropogenic conditions (Zuckerman et al., 2023).
Elk in western North America are an iconic species of high-mountain wilderness and also occur in many urban and suburban settings.Seasonal migrations between distinct summer and winter ranges are common in many elk populations (Irwin, 2002), including those in British Columbia, where migration distances vary from a few kilometers up to 100 km (Phillips & Szkorupa, 2011;Poole & Mowat, 2005).Changes in migration strategies have been observed in many elk populations that have reduced the proportion of the population that migrates long distances to high-elevation summer ranges and the amount of time spent on these summer ranges (Cole et al., 2015;Hebblewhite et al., 2006;Middleton, Kauffman, Mcwhirter, Cook, et al., 2013;Mulligan, 2021).Resident elk are often more closely associated with human activity, tend to more readily be involved in agricultural conflicts (Hegel et al., 2009) and may contribute to overgrazing on some ranges (Phillips & Szkorupa, 2011;Smyth, 2014).Several hypotheses have been suggested for changes in elk migratory behavior, including declining habitat quality on migrant ranges, increased access to hay and planted crops, predator avoidance (primarily wolf [Canis lupus] but also grizzly bear [Ursus arctos]), and winter range enhancement (e.g., shrub-mowing, tree removal, fertilizing, and grass seeding) (Cole et al., 2015;Hebblewhite et al., 2006;Middleton, Kauffman, McWhirter, Jimenez, et al., 2013;Mulligan, 2021).
Elk migration strategies have shifted in numerous areas throughout western North America.The proportion of migratory elk declined significantly between the late 1970s and early 2000s in an East Slopes Alberta population (Hebblewhite et al., 2006), but subsequently increased in the mid-2010s with a shift in migration patterns from the western high elevation range in Banff National Park, to eastern range subjected to more disturbance from oil and gas and forestry (Berg et al., 2021).Migrant cows had higher pregnancy rates and winter calf weights but lower survival of migrant cows and calves compared with resident elk, largely a result of elevated grizzly bear predation during June migration (Hebblewhite & Merrill, 2007), with resident elk forgoing high-quality food to reduce predation risk, mainly from wolves, by selecting areas close to human activity and roads (Hebblewhite & Merrill, 2009;Nelson et al., 2012).Interestingly, in this same population, calves born to resident mothers in the mid-2010s had 45% lower survivorship (Berg et al., 2023).Studies conducted on elk populations in British Columbia's Rocky Mountain Trench document a reduction in the proportion of elk that were migratory from 80% in the late 1980s-early 1990s, to about 50% in the late 2000s and 40% in the mid to late 2010s (Mulligan, 2021) and that migratory elk now spend 1 month longer on winter range (Jamieson & Hebert, 1993;Phillips & Szkorupa, 2011).In the nearby Elk Valley of the East Kootenay, British Columbia, approximately 45% of elk collared from a 1982-1996 study were resident, residing year-round on the west slopes of Natal Ridge near the Elkview coal mine (Gibson & Sheets, 1997).It is unknown if the low proportion of migratory elk in the Elk Valley suggests that migration is declining in this area.
Over the last half century, major changes to the East Kootenay ecosystem have unfolded, including forest encroachment and flooding of low-elevation habitat (largely affecting the Trench), forestry, climate change, expansion of agriculture and urban development, and changes to predator-prey dynamics including expanding white-tailed deer (Odocoileus virginianus) and general increases in predator populations (Hatter et al., 2018;Mowat, 2007;Mowat et al., 2022).Additional influences within the Elk Valley study area are four active open pit coal mines, which over the past 30 years have resulted in increasing areas of restricted access and no hunting (primarily for mine safety reasons) as well as small but increasingly abundant areas of reclamation that provide high quantity and quality forage (Poole, 2013;Smyth, 2014).Rare, Red-and Blue-listed high elevation grasslands occur primarily on the east side of the Elk Valley, providing high-quality forage at high elevations (BC Conservation Data Centre, 2023; Poole et al., 2016;Smyth, 2014).
The purpose of this study was to identify migratory patterns of elk in the Elk Valley, and to contrast mortality causes and survival estimates between migratory and resident elk.Over a 6-year period, we determined migratory status, seasonal ranges and migration corridors, timing and extent of movements, causes of mortality, and survival rates.We also summarized the composition of elk observed during spring carryover counts and Teck Coal Limited (Teck Coal) winter aerial wildlife surveys, using these data along with adult survival from collared individuals to estimate population growth.We provide comparison of migratory trends, survival rates, and causes of mortality with data from the 1982-1996 Elk Valley study, which monitored cow elk using conventional VHF collars and visual markings (Gibson & Sheets, 1997).While study area focus, telemetry methodologies (VHF collars rendering results arguably more conservative) and analytical techniques differed, the 1982-1996 study offers excellent comparisons of elk ecology decades apart.

| Study area
The Elk Valley study area lies within the Rocky Mountains of southeastern British Columbia.The study was conducted from January 2016 to January 2022 in a 2400 km 2 area defined by collar locations of the Elk Valley elk population (Figure 1).Main disturbances include forestry and coal mining.Coal mining has occurred in the valley since the early 1900s, and the current configuration of mines was constructed during the 1970s and early 1980s. 1 During the mid-1980s, over 800 elk wintered on and adjacent to Natal Ridge (where a smaller mine footprint was located-$3070 ha disturbance footprint in 1997 and $4500 ha in 2022 [L.Amos, Teck Coal Ltd., pers.comm.]), and the valley population was loosely estimated at over 4000 elk (B. Warkentin, BCMOE, pers. comm., 1984in Gibson & Sheets, 1997), compared to an estimated 2300, 2800, and 1700 individuals in 2003, 2013, and 2021, respectively (Beswick & Fontana, 2003;Chow, 2022;Szkorupa et al., 2013).Elk abundance declined during our period of study.Teck Coal aerial winter wildlife surveys conducted on mine properties with consistent effort, timing, and personnel indicated a decline since the recent peak in 2015 to just over half as many individuals observed in 2018, with a partial rebound in numbers in 2022 (L.Amos, Teck Coal Ltd., unpublished data).Within the Elk Valley, elk, along with white-tailed deer, are likely one of the most abundant ungulates; other ungulate species, in approximate order of relative abundance, include mule deer, moose (Alces alces), bighorn sheep, and mountain goat (Oreamnos americanus).Potential predators include wolves, grizzly bears, black bears (U. americanus), cougars (Felis concolour), wolverines (Gulo gulo), and coyotes (Canis latrans).The 1982-1996 study area was located on the ridge (and current mine) east of Sparwood (Figure 1).

| Capture and collaring
We captured 78 cow elk from late January to early April 2016 (n = 42), 2017 (n = 29), and 2018 (n = 7) with capture effort spread within winter ranges distributed throughout the study area.Three elk were censured from further analysis as a result of suspected capture-related mortality.We captured elk using two corral traps moved to five areas (29% of captures), and helicopter netgunning (71%; Krausman et al., 1985).Capture and handling protocols were conducted under BC Animal Care permit CB16-220413, with handling time kept to a minimum (<10 min) following live animal capture guidelines (RIC, 1998).We determined the pregnancy of captured elk from analysis of blood progesterone levels (Prairie Diagnostic Services Ltd., Saskatoon, SK).
We fitted 77 adult cow elk (≥2 years old) and 1 yearling with GPS radiocollars (Vectronic Aerospace Gmbh, Berlin, Germany, primarily Globalstar Survey collars), programmed to attempt a fix every 13 hours.The collars collected 97,052 collar locations, which we analyzed using a Geographic Information System (GIS; QGIS 3.4, Open Source Geospatial Foundation, www.qgis.com)and programs written in R (Version 3.5.1,R Development Core Team, 2018).

| Migration characterization
We initially examined each animal-year through the program Migration Mapper version 2.3 (Merkle et al., 2022; https://migrationinitiative.org/content/migrationmapper) and classified migratory behavior based on net squared displacement (Börger & Fryxell, 2012) to classify residents (with largely overlapping seasonal ranges) and migrants (with spatially distinct seasonal ranges) and migratory timing in spring and fall.If an elk left the winter or summer range but returned one or multiple times, the last date the animal was in the former range was set as the start of the migration.We did additional post-processing that included comparing seasonal range overlap and migratory distance (horizontal and vertical).Migratory elk needed to have less than 15% of their summer locations within their winter seasonal range (95% Brownian bridge utilization distribution), and had to reside in these distinct ranges for >30 days (Eggeman et al., 2016).We assumed elk captured between early March and early April were within winter range for those individuals and we saw no evidence to the contrary.As subcategories of migrants, we classified standard migrants, traveling moderate to longer distances and changing little in elevation, and elevational migrants, which were individuals that migrated <12 km Euclidean distance (Cagnacci et al., 2015;Cole et al., 2015) between seasonal range centroids (95% Brownian bridge utilization distribution) but whose mean elevation changed >250 m between winter and summer ranges (Barker, Mitchell, Proffitt, & Devoe, 2019;Hebblewhite et al., 2008).As above we ensured these distinct elevational ranges were inhabited for at least 30 days (Spitz et al., 2018).
We also classified mixed migrants and atypical migrants, which displayed behaviors that were not easy to classify as standard or elevational migrants.These animals generally migrated but did not return to the same range in the fall (mixed migrants; Bunnefeld et al., 2011), or displayed other strategies that were migrant-like but with an unclear or atypical migration pattern (atypical migrants).We omitted these atypical migrants from statistics on migration, but pooled them with migrant animals for survival analysis (both were alive until collar failure).We also identified elk that dispersed or were nomadic, but excluded these for consideration of migrant-to-resident ratio.We did not classify animalyears that had data for an incomplete migration season.We calculated the migratory strategy switching rate as the proportion of individuals with ≥2 years of monitoring that switched migratory strategies in consecutive years.We used a generalized linear mixed model to test for differences in migratory strategies among years, with individual elk as a random term.
We defined winter range for migrants as the time between the end of their fall migration, and the start of their spring migration.For residents, we defined winter based on the dates other migrants migrated.For each year, we defined a resident winter season as the time between the 66th percentile of fall arrival dates, and the 33rd percentile of spring leaving dates.

| Mortality and survival
We generally investigated elk mortalities within 2-4 days (median 2 days; mean 3.6 days [SE 0.60]; range 0-12 days; n = 27).We investigated three additional mortality sites 13 days and 3-4 months after death because of collar malfunction or mine safety concerns.Where possible, we collected a femur for bone marrow examination (following Neiland, 1970) and relevant tissue samples.From evidence at site (e.g., mortality site patterns, predator tracks, and scat) or via necropsy or tissue examination, we ascribed death to predation (wolf, grizzly, and cougar), human harvest, collision with vehicle or train, other human-caused accident, disease, starvation, or unknown natural causes.
Following Schuyler et al. (2019), we excluded from mortality and survival analysis elk that died within 2 weeks of capture.Elk were right censored from survival analysis once their radio collars went offline.We calculated survival rates annually and between resident and migratory elk using Cox proportional hazard models (Andersen & Gill, 1982).We also examined survival within three periods to estimate seasonal survival and cause-specific mortality rates: summer (1 June to 30 September), early winter (1 October to 31 January), and late winter (1 February to 31 May).These seasons were selected to capture summer migratory periods and expected differences in winter survival (Eggeman et al., 2016;Hebblewhite et al., 2006Hebblewhite et al., , 2018)).We used 1000 bootstrapped datasets to assess the variability of our results and to calculate measures of uncertainty.We started the survival year on February 1, the default start for a biological year in Migration Mapper.We summarized sources of cause-specific mortality and calculated cause-specific mortality rates (combined human and natural causes).Differences among survival rates were bootstrapped to provide 95% confidence intervals.
We calculated population growth rates using the Hatter and Bergerud (1991) method, which uses annual adult female survival rates (S; as calculated above), and calf recruitment rates (R; see Composition, below) to estimate population growth (lambda): λ = S/(1 À R).We estimated annual growth rates with migrants and residents pooled using spring carryover counts (2016-2019 and 2021-2022) and Teck Coal mine aerial surveys (2016-2018 and 2022), and a combined-year geometric mean growth rate broken out by residents and migrants (assuming they had the same calf recruitment).

| Composition
Carryover (recruitment) surveys were conducted during April and early May 2016-2022 (excluding 2020) to measure post-winter calf elk recruitment.Methodology followed Stent and Phillips (2013) and consisted of set roadside survey routes.Elk Valley mine aerial survey data up to 2018 and 2022 were provided by Teck Coal; no aerial surveys were conducted during 2019-2021.We also obtained ratio data (corrected for sightability) from complete inventories of the entire Elk Valley in 2021 (Chow, 2022).
During winter 2016, when sample size was roughly similar between capture methods, 69% of corral-trapped elk (n = 16) and 65% of helicopter-captured elk (n = 20) were migratory, suggesting little difference in selection of migratory behavior between capture methods.
We monitored 55 elk for ≥2 migratory years, totaling 130 pairs of animal-years.Twelve elk switched strategies a total of 18 times, a switching rate averaging 14% per year.There was tendency for resident elk to switch to a migratory strategy (12 switched from resident to migrant and 6 switched from migrant to resident), and most elk who switched migratory strategies made relatively subtle changes in seasonal movements within a similar area.Nearly half of the switching events were attributed to four elk.

| Changes in distance and elevation
Distinct patterns of distance and elevational migration were evident in Elk Valley cow elk (Figure 2, Table 2).Over one third (36%) of migratory elk traveled >20 km between seasonal ranges.Migrants accessed seasonal ranges by traveling distances up to 52 km and changing elevation up to 971 m.Migrants had on average little overlap in seasonal ranges; 88% had <5% overlap (Figure 2, Table 2).
Different patterns of migration between standard and elevational migrants were evident (Figure 2).Some elk traveled moderate to longer distances and changed little in elevation, while others moved short distances into higher terrain in immediately adjacent ranges to attain large changes in elevation.Elevational migrants tended to have slightly greater seasonal range overlap than standard migrants.

| Seasonal ranges and migration routes
Winter range was primarily restricted to the main valleys and overlapped extensively between resident and migrant elk, while migrant elk summer range expanded further into tributary valleys (Figure S1).The main standard migration routes followed main valleys, with elevational migrants making shorter, less linear movements as they moved up in elevation (Figure 3).

| Mortality and survival
We detected 30 mortalities of collared individuals between February 2016 and January 2022: 12 migrants, 14 residents, and 4 died before migratory status could be determined.Humans caused 47% of mortalities, predominantly vehicle and train collisions (Figure 4a).Four of the seven elk-vehicle collisions and one of the three train collisions occurred on or adjacent to Teck Coal property.Among natural causes, predation was verified in five (17%) of detected mortalities.Apparent starvation/old age caused four deaths.Percent fat content in bone marrow (n = 12 elk) ranged from a mean of 73% (range 51%-91%) for the five elk killed by predation, 62%-89% for three elk which died of unknown natural causes, 24% for an elk which died of an infection, 23% for an elk killed by a train collision, and 6% and 8% for two elk which died from starvation/ malnutrition.
Annual mortality rates from natural causes were similar between migratory strategies (difference [diff] mig- res = 0.015 [95% confidence interval, CI: À0.05 to 0.08]) but there was weak evidence for higher mortality rates from human causes for residents compared with migrants (diff mig-res = À0.031[95% CI: À0.092 to 0.030]) (Figure 4b), in large part due to higher mortality from train collisions and other human sources of mortality (Figure S2).For all elk the annual mortality rate from vehicle and train collisions combined averaged 0.05.
Survival rates varied among years (Figure S3) and averaged across all years 0.827 (5th to 95th percentile = 0.756-0.893)for residents and 0.895 (0.844-0.941) for F I G U R E 3 Migration paths of standard migrant and elevational migrant cow elk, Elk Valley, 2016-2022, overlain on migration paths of cow elk during the 1982-1996 (Natal Ridge) study (Gibson & Sheets, 1997).migrants (diff migrant-resident = 0.067 [À0.015-0.151])(Figure 4c).Survival rates were slightly higher during summer than early or late winter (diff summer-winterearly = 0.010 [À0.0235 to 0.0435], diff summer-winterlate = 0.021 [À0.014 to 0.056]).Migrants generally had higher survival compared to residents in summer and early winter, and similar survival in late winter (Figure 4d).3).Calf ratios obtained from spring carryover count (road) surveys in April and May were generally lower than ratios obtained from aerial surveys conducted in February.

| Composition and growth rates
Estimated annual population growth rates varied among years using recruitment derived from carryover counts ( 2016

| DISCUSSION
Migratory behavior was variable among collared cow elk in the Elk Valley from 2016 to 2022, but at the broadest scale was approximately equal between migrants and residents.Migration distance and elevation changes were on average less than found in elk populations in the West  Phillips & Szkorupa, 2011).Wyoming elk migrated an average of 52-69 km with maximum distances of 93-103 km (Kauffman et al., 2018).Resident Elk Valley elk remained in lower elevation habitats year-round and had lower survival compared with migrants, largely due to higher human-caused mortality rates for residents.
Comparison with the 1982-1996 data (Gibson & Sheets, 1997) suggested little change in the proportion of residents and migrants within the population between the 1980s-mid-1990s (45% resident) and the late 2010searly 2020s (48%), or in the pattern of standard and elevational migration.The current study was conducted largely during a period of overall population decline, which should have favored more residents in the population (Berg et al., 2019;Eggeman et al., 2016).The relative stability in proportions of migration strategies is in contrast to several studies that have detected reductions in the proportion of migrants in elk populations and less time used by migrants on summer ranges (Hebblewhite et al., 2006;Jamieson & Hebert, 1993;Phillips & Szkorupa, 2011).In the Kootenay Trench studies, the proportion of migratory elk declined by half from 80% during the late 1980s and early 1990s to about 40% during the mid-2010s during a period when elk numbers increased, then declined back to 1990s levels (Mulligan, 2021).
Cow elk in the current study made fewer and shorter movements into tributaries of the upper Elk Valley and greater use of mine properties than observed during the 1982-1996 study (Figure 3).In the 1982-1996 study, 8 out of 40 cows migrated 42-84 km, compared with only 1 out of 9 cows in the current study captured in the same area which migrated 41 km, with the remaining elk migrating <10 km or being resident.These changes are analogous to changes in migratory tactics from long-distant migrants to elevational or short-distant migrants, which can be seen as elk responding dynamically to changes in natural and anthropogenic environments (Zuckerman et al., 2023), in this case, the $45% increase in mine footprint over the past 30 years, which may reduce predation risk, and the increase in reclamation that provide high quantity and quality forage (Poole, 2013;Smyth, 2014).Local observations from hunters and guide outfitters confirmed that elk still inhabit the upper Elk Valley, albeit at much lower densities than in the past.
Declines in proportions of long-distant migrants may be related to changes in forage quality (Middleton, Kauffman, Mcwhirter, Cook, et al., 2013) that lower fat reserves, pregnancy rates, and recruitment (Cole et al., 2015;Proffitt et al., 2016).Winter bone marrow fat indices for several individuals were low and starvation accounted for 13% of detected mortalities, indicative of nutritional stress (Cook et al., 2001) and suggesting forage quality and/or availability may be an issue within the Elk Valley.Forage quality and availability in the Elk Valley have likely reduced over time from year-round elk grazing pressure on winter ranges (Phillips & Szkorupa, 2011), cattle and horses on open ranges (Forest Practices Board, 2008), and wildfire suppression (Forest Practices Board, 2016;Ross & Wikeem, 2002).Wildlife exclusion fencing has reduced the amount of unfenced, irrigated agriculture land by about 40% in the Rocky Mountain Trench over the past 20 years (Mulligan, 2021), and a similar trend has likely occurred in the Elk Valley.However, reclamation activities on mine operations starting in the 1980s likely intensified during the 1990s and 2000s, which may have enhanced forage quality and quantity on mine footprints (Smyth, 2014), influencing changes in patterns of migration (Poole, 2013).Agricultural lands in the Elk Valley bottomlands may provide high forage quality for much of the year, but low winter marrow fat indices suggest these sources are not available or adequate during some winters.Individuals may be more likely to migrate away from irrigated agricultural areas if better forage is available elsewhere (Barker, Mitchell, & Proffitt, 2019) or if they experience high density in their winter range (Eggeman et al., 2016).The high-elevation grasslands may provide high-quality sources of forage for migrants, but forage assessments that consider per capita elk availability would be needed.Poor body condition can also be caused by animals entering the winter in poor condition, suggesting summer and fall forage may be limiting (Cook et al., 2013), and/or having inadequate forage during winter.
Declines in the proportion of long-distant migrants may also be related to higher predation rates for migrants (Hebblewhite et al., 2018; yet lower predation on migrant summer ranges: Middleton, Kauffman, Mcwhirter, Cook, et al., 2013).Verified predation rates on adult cows during the current study were not high; however, we did not examine predation rates on calves which are generally much higher (Barber-Meyer et al., 2008;Middleton, Kauffman, Mcwhirter, Cook, et al., 2013) and may be skewed towards resident elk (Berg et al., 2023).
Within the current study, migration strategies used by individual elk switched at a mean rate of 14% per year, with relatively subtle changes in seasonal movements and a subtle shift towards increased migration.Snow depths and winter severity likely influenced the extent (the probability of an elk migrating increases following a severe winter) and timing (spring migration is delayed with increased severity and snow depth) of migration within the Elk Valley among years (cf Eggeman et al., 2016;Kauffman et al., 2018;Lendrum et al., 2013;Zuckerman et al., 2023).In the current study, elk occasionally but not consistently appeared to switch migration strategies according to winter severity.Migratory strategy switching is likely related to elk abundance (more switching from resident to migrant at higher elk abundance) and age (as they aged migrants tended to stop switching, whereas residents showed some tendency to increase switching) (Eggeman et al., 2016).Increases in density also tend to increase migration distances (Eggeman et al., 2016;Mysterud et al., 2011), which may have contributed to the longer migratory distances observed in the 1982-1996 study (Gibson & Sheets, 1997).
Human-related causes were responsible for nearly half of the mortalities detected, having a greater impact on residents than migrants.This contrasted with the 1982-1996 data where natural causes were the major source of mortality for residents, and hunting (and poaching) for migratory cows (Gibson & Sheets, 1997).In the East Kootenay Trench studies, human-caused mortalities (65% of all known mortalities; n = 26) were nearly double non-human-caused mortalities (35%); in contrast to our findings, much of the human-caused mortality in the Trench was due to hunting which resulted from the then more liberal harvest regulations (Phillips & Szkorupa, 2011).During cow harvests, hunters may target large, healthy, prime-aged females with the greatest reproductive value, which will likely have a disproportionate effect on population growth (Sergeyev et al., 2021).
Vehicle and train collisions were the single highest cause of elk mortality in the current study.Mortalities from vehicle and train collisions (average annual mortality rate of 0.05) appeared to be much higher than what occurred during the 1982-1996 study (5% of all known elk mortalities; Gibson & Sheets, 1997).Upwards of two thirds of wildlife-vehicle collisions are likely unreported, often due to animals dying off the highway right of way (Lee et al., 2021)-neither of the two elk-vehicle collision mortalities on numbered highways were reported in provincial databases.Wildlife-train collisions within the study area are also likely underreported and may represent a significant source of elk mortality.Only two of the four collared elk known to be struck by a train were reported to the Report All Poachers and Polluters (RAPP) line (E.Chow, Ministry of Forests, unpubl.data).
While some of the unknown natural mortalities may have been from predation, predation was not identified as a major mortality source of adult cows.Although wolves recovered in southeastern British Columbia during the 1900s and 2000s (Mowat et al., 2022) and Elk Valley grizzly bear numbers have increased since 2012 (Lamb et al., 2019), we saw little evidence of high wolf or other carnivore predation rates.This finding contrasts with many areas where predator mortalities contribute a greater proportion of overall mortality (Barber-Meyer et al., 2008;Berg et al., 2023;Hebblewhite et al., 2018).However, predation rates on calves may be high especially from bears (Barber-Meyer et al., 2008;White et al., 2010) for migrants (Hebblewhite et al., 2018) and residents (Berg et al., 2023), and may have increased over the past several decades.
Annual survival rates varied over years, were lower for residents compared with migrants, and were lower in late winter compared with early winter.The mean survival rate of residents during the 1982-1996 study was 0.98 (Gibson & Sheets, 1997), significantly higher than rates for residents in the current study (0.83), while mean annual survival rates for migrants were nearly identical between studies (0.90).Hebblewhite et al. (2018) found only slightly higher survival by residents (0.85) compared with migrants (0.81) in eastern Alberta.We found annual median survival rates within the mid-point of ranges of populations in the western United States ($0.78-0.92),where a suite of carnivores is present but where harvest is generally the largest mortality factor (Brodie et al., 2013;Raithel et al., 2007).
Calf ratios averaged about 27 or 32 calves:100 cows during the study (mean of road versus aerial surveys, respectively; Table 3).Adult cow survival rates below 0.85 coupled with 25-30 calves:100 cows ratios usually indicate a declining population (IDFG, 2014: Table 2).Our simple modeling suggests that during the study the overall population was stable to declining, but there was annual variation and differences between migrants and residents.Survival rates observed during the East Kootenay Trench studies in 1986-1993 and 2007-2010 were 0.92 and 0.81, respectively (Szkorupa & Mowat, 2010).Elk Valley cow pregnancy rates varied widely among years, possibly related to annual differences in cow elk health and fat reserves, or reduced breeding success (reduced pregnancy rates and/or delayed conception dates) as a result of continual harassment during the rut (Davidson et al., 2012;Squibb et al., 1986).Pregnancy rates also varied annually from 0.66 to 0.98 over 4 years in the Ya Ha Tinda elk population in eastern Alberta (Hebblewhite & Merrill, 2011).
Differential mortality among migratory strategies may permit the coexistence of resident and migratory components to a population, but the advantages of each will be sensitive to changes in reproductive success and survival (Monteith et al., 2014).Poor recruitment and survival resulting from a particular migratory tactic, given natal and adult fidelity to a particular seasonal range, will inherently reduce the proportion of individuals in the population employing the more costly tactic.Quantifying the reproductive success of each tactic would be a useful measure in future studies.

| CONCLUSIONS
Either gradual loss (greater mortality or switching behavior) of migratory individuals or reduced productivity of migratory cows could alter the proportion of migratory individuals within a population (Cole et al., 2015;Hebblewhite et al., 2006;Middleton, Kauffman, McWhirter, Jimenez, et al., 2013).However, the proportion of migrant elk within the partially migratory Elk Valley population appears to be relatively stable between two studies conducted $30 years apart, and current mortality data do not suggest excessive mortality of migratory cows.Notably, the distance and location of migratory movements changed over time, becoming shorter in length and more focused on mine properties.It would be instructive to determine the relative forage quality among summering areas for resident and migratory elk, including animals associated with mine properties during both winter and summer.Determining the winter ranges used by elk currently summering in the upper Elk Valley may also provide insights into why migration distances are eroding.
Calf mortality may be an important factor influencing population growth rates (Barber-Meyer et al., 2008;Berg et al., 2023;Middleton, Kauffman, Mcwhirter, Cook, et al., 2013).Calf survival monitoring would add an important age-class to current survival rates and, if collaring of neonates were involved (Barber-Meyer et al., 2008), cause-specific mortality between migration strategies.Surveys targeting collared elk and the presence of calves could give an indication of the timing of calf mortality between strategies.Camera collars on female elk could also help identify timing, and potentially cause, of calf mortality (Kaczensky et al., 2019).
Mitigation of highway and railway mortalities should continue to be a management priority.A companion spatial analysis identified several areas of high importance for both highway and railway crossings (Poole & Lamb, 2022).Identified important areas of elk crossings can feed into mitigation planning to reduce elk-vehicle collisions and enhance connectivity in the broader Elk Valley (Lee et al., 2019).Teck Coal may wish to further mitigate high-risk areas for elk-vehicle collisions in the high-importance elk crossing areas within the first few kilometers south of their northern mine gatehouse.In addition, the majority of the high-importance areas for rail crossings within the study area occurred on the north track between Sparwood and the northern coal mine, tracks dedicated to transport of coal.Teck Coal may wish to collaborate with the rail company to explore methods to reduce the risk of elk-train collisions along this section of track.Mortality from elk-train collisions may be an important and underreported mortality source, and clear reporting of these data is lacking.

F
I G U R E 1 Collared cow elk locations from the Sparwood elk project, 2016-2022.Not shown are the movements of 1 elk north and 1 elk east into Alberta.The 1982-1996 study area was Natal Ridge east of Sparwood.

F
I G U R E 4 (a) Frequency (n = 30) of adult cow elk mortalities by mortality cause, Elk Valley, February 2016 to January 2022.(b) Annual mortality rates for human and natural causes for resident and migrant elk.(c) Average annual survival rates for resident and migrant elk.(d) Average seasonal survival rates for resident and migrant elk.Boxes bound the 25th and 75th percentiles, solid line within the box indicates the median, and the whiskers extend to 1.5 times the interquartile range of the observations.T A B L E 3 Elk observations and calf: adult ratios from spring carryover counts (Road) and Teck Coal or government aerial surveys (Aerial), Elk Valley, February-May 2016-2019 and 2021= 38 km, range 16-63 km distance; x = 1000 m elevation change; Poole & Mowat, 2005) and East Kootenay Rocky Mountain Trench (up to 100 km;