Spatial Genetic Structure of Rocky Mountain Bighorn Sheep Ovis Canadensis Canadensis at the Northern Limit of Their Native Range

Canadian Journal of Zoology

Spatial genetic structure of Rocky Mountain bighorn sheep Ovis canadensis canadensis at the northern limit of their native range

Journal: Canadian Journal of Zoology

Manuscript ID cjz-2019-0183.R2

Manuscript Type: Article

Date Submitted by the 10-Dec-2019 Author:

Complete List of Authors: Deakin, Samuel; University of Alberta, Department of Biological Sciences Gorrell, Jamieson; Vancouver Island University, Department of Biological Sciences Kneteman,Draft Jeffery; Government of Alberta, Alberta Environment and Parks Hik, David; University of Alberta, Department of Biological Sciences; Simon Fraser University, Department of Biological Sciences Jobin, Richard; Government of Alberta, Fish and Wildlife Enforcement Branch Coltman, D.; University of Alberta, Department of Biological Sciences

Is your manuscript invited for consideration in a Special Not applicable (regular submission) Issue?:

POPULATION GENETICS < Discipline, ALPINE < Habitat, TERRESTRIAL < Keyword: Habitat, UNGULATES < Taxon

https://mc06.manuscriptcentral.com/cjz-pubs Page 1 of 56 Canadian Journal of Zoology

1 Spatial genetic structure of Rocky Mountain bighorn sheep Ovis canadensis

2 canadensis at the northern limit of their native range

4 Samuel Deakin, Department of Biological Sciences, University of Alberta

5 Jamieson C. Gorrell, Department of Biological Sciences, Vancouver Island University

6 Jeffery Kneteman, Alberta Environment and Parks, Government of Alberta

7 David S. Hik, Department of Biological Sciences, University of Alberta; and Department

8 of Biological Sciences, Simon Fraser University

9 Richard M. Jobin, Fish and Wildlife Enforcement Branch, Government of Alberta

10 David W. Coltman, Department of Biological Sciences, University of Alberta

11 Draft

1 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 2 of 56

24 Abstract:

25 The Canadian Rocky Mountains are one of the few places on Earth where the spatial

26 genetic structure of wide-ranging species have been relatively unaffected by

27 anthropogenic disturbance. We characterised the spatial genetic structure of Rocky

28 Mountain bighorn sheep (Ovis canadensis canadensis (Shaw, 1804)) in the northern

29 portion of their range. Using microsatellites from 1495 individuals and mitochondrial DNA

30 sequences from 188 individuals, we examined both broad and fine scale spatial genetic

31 structure, assessed sex-biased gene flow within the northern portion of the species range,

32 and identified geographic patterns of genetic diversity. We found that broad-scale spatial

33 genetic structure was consistent with barriers to movement created by major river valleys.

34 The fine-scale spatial genetic structureDraft was characterized by a strong pattern of isolation-

35 by-distance, and analysis of neighborhood size using spatial autocorrelation indicated

36 gene flow frequently occurred over distances of up to 100 km. However, analysis of sex

37 specific spatial autocorrelation and analysis of mitochondrial haplotype distributions failed

38 to detect any evidence of sex-biased gene flow. Finally, our analyses reveal decreasing

39 genetic diversity with increasing latitude, consistent with patterns of post-glacial

40 recolonization of the Rocky Mountains.

42 Keywords: Spatial genetic structure, post-glacial recolonization, ungulates, terrestrial,

43 alpine, Rocky Mountain bighorn sheep, Ovis canadensis canadensis

2 https://mc06.manuscriptcentral.com/cjz-pubs Page 3 of 56 Canadian Journal of Zoology

47 Introduction

49 In the absence of anthropogenic habitat fragmentation, the natural population genetic

50 structure of species can be defined by natural geographic and ecological barriers (Slatkin

51 1985; 1987), or the dispersal capabilities of the species (Wright 1942; Kimura and Weiss

52 1964). When natural fragmentation is present, population spatial genetic structure

53 correlates with natural barriers to gene flow (Slatkin 1985; 1987). When a habitat lacks

54 natural barriers, spatial genetic structure may exhibit clinal differentiation if the population

55 range exceeds the maximal dispersal distance of the species (Kimura and Weiss 1964),

56 referred to as isolation-by-distance (Wright 1942). In the modern world anthropogenic

57 habitat fragmentation is common andDraft increasing, and the extent of pristine habitats where

58 natural phenomena can be studied is increasingly limited.

60 The Canadian Rocky Mountains are a vast, relatively undisturbed, mountainous habitat

61 which are part of an important international conservation initiative (Chester 2015). Areas

62 of high anthropogenic activity in the northern Rocky Mountains are avoided by certain

63 species (Proctor and Paetkau 2004; Rogala et al. 2011); however, these areas are minor

64 when considering the vastness of the region and the intensity of human activity in other

65 regions. Prior studies in this region identified broad-scale spatial genetic structure of

66 several species is correlated with natural landscape features which impede gene flow,

67 such as high elevation (Adams and Burg 2015) and major valleys (Serroya et al. 2012),

68 or subtle ecological factors such as landcover preference (Keyghobadi et al. 1999;

69 Gayathiri Samarasekera et al. 2012; Cullingham et al. 2016). Finer scale spatial genetic

3 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 4 of 56

70 structure in the northern Rocky Mountains is primarily influenced by isolation-by-distance

71 (Forbes and Hogg 1999; Keyghobadi et al. 1999; 2005; Gayathiri Samarasekera et al.

72 2012; Weckworth et al. 2013; Cullingham et al. 2016). Therefore, this region offers a

73 range minimally affected by anthropogenic activities, where the natural spatial genetic

74 structure of a species can be observed.

76 Following the Wisconsin glaciation, many species of flora and fauna recolonized

77 northwestern North American including the Rocky Mountains. The two main refugial

78 sources for this post-glacial recolonization were the Beringian refugium in northeastern

79 Siberia and northwestern North America, and the Southern refugium, comprised by most

80 of the modern-day mainland UnitedDraft States (Hewitt 2004; Shafer et al. 2010). Genetic

81 signatures of recolonization, primarily resulting from the founder effect (Frankham 1997),

82 are well documented for many contemporary populations of flora (Mitton et al. 2000;

83 Jørgensen et al. 2002; Anderson et al. 2006; Godbout et al. 2008) and fauna (Stone and

84 Cook 2000; Flagstad and Røed 2003; McDevitt et al. 2009;Latch et al. 2009). By

85 examining patterns of genetic diversity across the landscape, we can make inference

86 about a species recolonization patterns in the northern Rocky Mountains (Shafer et al.

87 2010).

89 The range of Rocky Mountain bighorn sheep (Ovis canadensis canadenis (Shaw, 1804))

90 extends along the Rocky Mountains from Arizona to Alberta and British Columbia, to

91 approximately 55o N. For the southern portion of the species range, studies into the

92 movement (DeCesare and Pletscher 2006) and genetic structure (Miller et al. 2012;

4 https://mc06.manuscriptcentral.com/cjz-pubs Page 5 of 56 Canadian Journal of Zoology

93 Driscoll et al. 2015) of sheep populations are numerous. In the northern portion of their

94 range studies of movement patterns (Festa-Bianchet 1986a; Poole et al. 2016) and

95 population genetic structure (Luikart and Allendorf 1996; Forbes and Hogg 1999) are

96 more limited. Unlike the southern areas, the northern ranges are relatively unfragmented,

97 unimpacted by translocations, and unaffected by population declines or disease-related

98 die-offs. Notably, respiratory disease has never been reported north of the Bow River

99 (Fig. 1) (Government of Alberta 2015). Thus, the northern Rocky Mountains are well

100 suited to study the natural spatial genetic structure of Rocky Mountain bighorn sheep in

101 a relatively intact landscape. Additionally, bighorn sheep in the northern Rocky Mountains

102 are at the northern limit of the species range, and therefore should show signatures of the

103 range edge effect and recent post-glacialDraft recolonization.

104

105 Broad-scale spatial genetic structure of Rocky Mountain bighorn sheep is likely influenced

106 by natural landscape features. Studies of desert bighorn sheep (Ovis canadensis nelsoni

107 (Merriam, 1897)) indicate most fenced highways are barriers to gene flow (Epps et al.

108 2005; 2018). However, most of highways in the northern Rocky Mountains lack fences.

109 Instead, it is likely that natural features of the heterogenous landscape are barriers to the

110 gene flow of these alpine ungulates. Alpine ungulates prefer to occupy high elevations;

111 thus, their movement and distribution are primarily restricted by the availability of this

112 habitat. Due to this habitat preference, the population genetic structure of thinhorn sheep

113 (Ovis dalli (Nelson, 1884)) and mountain goats (Oreamnos americanus (Blainville, 1816))

114 have been found to correlate with areas of high elevation (Worley et al. 2004; Shafer et

115 al. 2011; Sim et al. 2018). Similarly, Rocky Mountain bighorn sheep show preferences for

5 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 6 of 56

116 higher elevation alpine meadows (Fiesta-Bianchet 1988; Loehr 2006), situated close to

117 steep cliffs which can be used to escape predators (Fiesta-Bianchet 1988; Loehr 2006).

118 Therefore, we expect broad-scale Rocky Mountain bighorn sheep spatial genetic

119 structure to correlate with high elevations, due to reduced gene flow across lower

120 elevations.

121

122 At a finer scale, the adaptations of mountain sheep to alpine environments enable their

123 movement across the rugged terrain of mountain ranges to be unimpeded. Argali sheep

124 (Ovis ammon polii (Blyth, 1840)) exhibit low levels of population genetic differentiation,

125 indicating high levels of gene flow persist, despite inhabiting mountainous terrain (Luikart

126 et al. 2011). For Rocky Mountain bighornDraft sheep, male seasonal movements of 33-48 km

127 have been reported in Alberta and Montana (Festa-Bianchet 1986a; DeCesare and

128 Pletscher 2006), and analysis of microsatellite data (Forbes and Hogg 1999) indicates

129 moderate to strong patterns of isolation-by-distance across the entire species range. In

130 the mountains, Rocky Mountain bighorn sheep are likely able to move relatively freely,

131 and therefore their finer scale spatial genetic structure is likely to exhibit a pattern of

132 isolation-by-distance.

133

134 Dispersal is well documented across many taxa, with resource competition, kin

135 competition, and inbreeding avoidance commonly theorised as evolutionary drivers

136 (Pusey 1987; Lawson Handley and Perrin 2007). Most mammalian species exhibit male

137 biased dispersal (Pusey 1987), however previous studies of mountain sheep found no

138 evidence of sex-biased dispersal (Roffler et al. 2014). Bighorn sheep social structure is

6 https://mc06.manuscriptcentral.com/cjz-pubs Page 7 of 56 Canadian Journal of Zoology

139 highly segregated between the two sexes. Females and sexually immature males coexist

140 until males become sexually mature, at which point they leave the herd to associate with

141 exclusively male bachelor groups (Festa Bianchet 1991). Females exhibit high levels of

142 natal philopatry and rarely emigrate from their home ranges (Festa-Bianchet 1986a;

143 Rubin et al. 1998; Boyce et al. 1999; DeCesare and Pletscher 2006). However, instead

144 of exhibiting permanent dispersal (Festa Bianchet 1991), males undertake seasonal

145 breeding migrations (Hogg 2000; Poole et al. 2016), visiting other populations during the

146 rut to maximise their reproductive success and returning to their home range following

147 the rut (Hogg 2000). Due to the rarity of female dispersal and the occurrence of male

148 breeding migrations, we expect that vast majority of gene flow across the landscape to

149 be male mediated. Draft

150

151 Two species of mountain sheep recolonized northwestern North America following the

152 Wisconsin glaciation (Loehr et al. 2006). Thinhorn sheep recolonized from the Beringian

153 refugium and other minor northern refugia (Loehr et al. 2006; Sim et al 2016). It is

154 hypothesised that bighorn sheep recolonized northwestern North America from the

155 Southern refugium (Cowan 1940; Geist 1999). Luikart and Allendorf (1996) reported

156 bighorn sheep from the northern portion of the range exhibit a reduction in genetic

157 diversity compared to the southern portion, consistent with the hypothesis that the

158 Southern refugium sourced the recolonization of northwestern North America. Given the

159 linear distribution of bighorn sheep in the northern Rocky Mountains, the strength of the

160 founder effect should increase with geographic distance from the Southern refugium,

161 when examined on a finer scale. We therefore hypothesized that current populations of

7 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 8 of 56

162 bighorn sheep in the northern Rocky Mountains will exhibit declining genetic diversity with

163 increasing latitude.

164

165 We used microsatellite markers and mitochondrial DNA sequence data to characterize

166 the population structure of the Rocky Mountain bighorn sheep in the northern portion of

167 their range. Our main objectives were to: I) identify potential barriers to gene flow which

168 cause board-scale spatial genetic structure, II) examine fine-scale spatial genetic

169 structure and the extent of contemporary gene flow between sampling locations, III) test

170 for sex differences in population genetic structure and gene flow across the landscape of

171 the northern Rocky Mountains, and IV) characterize fine-scale spatial patterns of genetic

172 diversity across the landscape, thatDraft may have arisen as a result of post glacial

173 recolonization.

174

175 Methods

176

177 Sampling

178 Biological material used for DNA analysis was collected over multiple years, using various

179 techniques, by multiple personnel (Fig. 1 and Table 1). The majority of our samples were

180 collected as faeces during the winters from 2013-2015 by Government of Alberta staff.

181 Additionally, blood, skin, and hair samples were collected from sheep at the Luscar-Greg

182 mine site, Ram Mountain, Sheep River, Castle-Yarrow, Narraway, and Radium by

183 Government of Alberta or British Columbia staff, or university researchers under protocols

184 approved by the University of Alberta Animal Use and Care Committee (certificate

8 https://mc06.manuscriptcentral.com/cjz-pubs Page 9 of 56 Canadian Journal of Zoology

185 number 610901) and the University of Calgary Animal Care Committee (protocol number

186 BI11R-14), following the guidelines set out by the Canadian Council on Animal Care.

187 Finally, we obtained genetic data from hunter submitted horn cores analysed by the

188 Alberta Fish and Wildlife Forensics Unit (AFWFU). GPS coordinates were recorded for

189 all sampling locations, with exception to hunter submitted horn cores which were grouped

190 by wildlife management unit. All sampling locations, with the exception of Ram Mountain,

191 were entirely native populations. For Ram Mountain we used a pedigree (Poirier et al.

192 2019) to select only individuals of native ancestry for genotyping.

193

194 DNA isolation

195 We extracted DNA from biological materialDraft using protocols optimized for each material

196 type. Skin and hair samples were digested and extracted using the DNeasy Blood and

197 Tissue kit (Qiagen, Venlo), following the manufacturer’s recommended procedure. Blood

198 was treated with ammonium-chloride potassium (ACK) (Brown et al. 2016) prior to

199 undergoing the same procedure as skin and hair samples. For faecal samples, three

200 pellets from each sample were soaked in 1X phosphate-buffered saline (PBS) for 20

201 minutes before being swabbed around the exterior with a cotton-tip applicator, which in

202 turn was washed with Aquastool (MoBiTec, Goettingen), the material was digested and

203 extracted following the Aquastool protocol.

204

205 Microsatellite genotyping

206 We amplified extracted DNA across 13 nuclear microsatellite loci and one sex-

207 determining amelogenin marker (Table 2). Loci were amplified either individually or in one

9 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 10 of 56

208 of three multiplexed groups (for further details regarding master mixes, thermocycler

209 profiles and co-loading see Supplementary Information S1). We ran amplified products

210 on an Applied Biosystems 3730 capillary sequencer and alleles were scored based on

211 fragment size using GENEMAPPER 4.0 software (Applied Biosystems, Foster City).

212 Faecal samples were amplified and scored in triplicate to ensure data quality and

213 minimize allelic drop-out. A subset of skin samples were also genotyped by the AFWFU

214 to ensure bin calibration prior to merging datasets.

215

216 We extracted and genotyped 1,252 faecal samples in triplicate, the software GIMLET 1.3

217 (Valiere 2002) was used to produce consensus genotypes across the three replicates for

218 each faecal sample. We accepted anyDraft allele that was present in at least two out of the

219 three replicates but then only retained samples that achieved >60% genotyping success

220 across the 13 loci, resulting in 957 samples (76% success rate). We used CERVUS 3.0

221 (Kalinowski et al. 2007) to identify identical genotypes as a result of repeatedly sampling

222 the same individuals. Identical samples were pooled into consensus genotypes until only

223 unique samples remained. In five instances where identical genotypes were found in two

224 different sampling locations, each genotype was removed from one sampling location at

225 random. When there was only one mismatch between samples in the same location, we

226 scrutinized the electropherograms produced by GENEMAPPER 4.0 (Applied Biosystems,

227 Foster City) a second time. Our final list contained 613 unique faecal genotypes (49% of

228 the original starting samples). We also genotyped 309 sheep from blood, skin, and hair

229 samples and pooled 573 horn cores from the AFWFU for a total sample size of 1,495

230 unique genotypes (922 samples with GPS coordinates).

10 https://mc06.manuscriptcentral.com/cjz-pubs Page 11 of 56 Canadian Journal of Zoology

231

232 Mitochondrial D-loop sequencing

233 We selected 202 individuals from 18 georeferenced sampling locations spread across the

234 length of the sampling area to be sequenced (Fig. 1 and Table 1). We amplified a portion

235 of the D-loop region using two different PCR conditions depending on whether the

236 template DNA was sourced from faeces or tissue and then performed clean-up using

237 ExoSap (further details in Supplementary Information S2). For each individual two PCR

238 amplifications were performed, which resulted in two technical replicates. We sequenced

239 the D-loop portion using Applied Biosystems 3730 capillary sequencer. Both replicates

240 were sequenced in both directions, resulting in two forward sequences and two reverse

241 sequences for each individual. TheDraft four sequences were aligned in Geneious 11.1.4

242 (https://www.geneious.com, Kearse et al. 2012) to produce a consensus sequence. If two

243 or more sequences were of poor quality and failed to align when producing a consensus

244 sequence, the individual was subject to resequencing. We used BLASTn to align the

245 haplotypes discovered against other bighorn sheep accessions in the NCBI database.

246

247 Summary statistics

248 We used Microsatellite Analyser 4.05 (Dieringer and Schlotterer 2003) to calculate the

249 global expected heterozygosity, observed heterozygosity, and number of alleles, which

퐻푒푥 ― 퐻표푏푠 250 we used to calculate Wright’s inbreeding coefficient 푓 = . We repeated these 퐻푒푥

251 calculations for each sampling location, with the addition of rarefied allelic richness to

252 account for uneven sample size between sampling locations. We also used Microsatellite

253 Analyser 4.05 to calculate Nei’s (1972) standard genetic distance corrected for small

11 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 12 of 56

254 sample sizes and FST (Weir and Cockerham 1984) among sampling locations.

255 Significance of the FST distances were tested with 10,000 permutations.

256

257 Broad-scale spatial genetic structure

258 To investigate subtle population structure that might not be apparent when using pairwise

259 distance estimates we performed a principal coordinate analysis in Adegenet 2.1.1

260 (Jombart et al. 2018). Additionally, we used the Bayesian approach developed in

261 STRUCTURE 2.3 (Pritchard et al. 2000). The admixture analysis included correlated

262 allele frequencies and was run for 500,000 MCMC iterations after a burn-in period of

263 50,000 and was repeated across 10 independent runs for values of K1-K8. This analysis

264 was done for the entire 1,495 individualDraft dataset and no prior sampling location information

265 was included. Following the initial runs on the entire dataset, individuals occurring

266 between the Bow and Athabasca Rivers as well as individuals north of the Athabasca

267 River were subjected to separate follow-up analyses, each with 10 independent runs of

268 K1-K4 to investigate substructure as recommended by Janes et al. (2017). The entire

269 admixture analysis was also repeated using prior location data. We used CLUMPAK

270 (Kopelman et al. 2015) to combine the 10 independent runs for each value of K and to

271 produce admixture plots. We produced likelihood plots using the Evanno et al. (2005)

272 method in the R package pophelper 2.2.7 (Francis 2017).

273

274 To further test if population genetic structure was influenced by the presence of major

275 river valleys, we used a hierarchical analysis of molecular variance (AMOVA) in GenAlEx

276 6.5 (Peakall and Smouse 2006; 2012). The 1,495 individuals were grouped by the 49

12 https://mc06.manuscriptcentral.com/cjz-pubs Page 13 of 56 Canadian Journal of Zoology

277 sampling locations, which were in turn grouped a priori into five regions separated by the

278 Bow River, the North Saskatchewan River, the Athabasca River, and the Smoky River

279 (Fig. 1). Pairwise genetic distance was measured with FST and significance was tested

280 using 9,999 permutations.

281

282 To examine the effect each river valley had on genetic distance individually, we performed

283 a series of partial Mantel tests using subsets of georeferenced samples from locations

284 separated by major rivers. To test the Athabasca River, we created a subset of individuals

285 from sampling locations between the North Saskatchewan and Smoky Rivers. For the

286 Bow River, we created a subset of individuals from sampling locations south of the North

287 Saskatchewan River. We used individualsDraft from sampling locations between the Bow and

288 Athabasca Rivers, to test the North Saskatchewan River. Finally, for the Smoky River we

289 used a subset of individuals from sampling locations north of the Athabasca River. For

290 each of these subsets the R package Adegenet was used to calculate Nei’s (1972)

291 standard genetic distances and Euclidean geographic distances matrices. Partial Mantel

292 tests were performed between the matrix of Nei’s distances and a binary matrix denoting

293 whether pair were on the same (0) or opposite (1) sides of the river, while conditioning on

294 Euclidean geographic distance using the software ZT 1.1 (Bonnet and Van de Peer 2002).

295 Each partial Mantel test was performed for 1,000,000 permutations.

296

297 Fine-scale spatial genetic structure

298 To examine the effect of Euclidean geographic distances on genetic distances, we tested

299 for the common effect of isolation-by-distance. To avoid problems caused by small

13 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 14 of 56

300 sample sizes, some sampling locations were either pooled or removed. Any sampling

301 locations with fewer than 10 individuals were pooled with the closest sampling location

302 within 15 km, unless they were separated by one of the major rivers (Fig. 1). Sampling

303 locations that were pooled were assigned the GPS coordinates of the sampling location

304 with the most samples prior to pooling. Any sampling locations with less than 10

305 individuals which were further than 15 km or separated by a major river from another

306 sampling location were removed from the dataset. This resulted in a dataset of 918

307 individuals from 33 georeferenced sampling locations. Nei’s (1972) standard genetic

308 distances and Euclidean geographic distances were calculated among sampling

309 locations with the R package Adegenet. We then performed a Mantel test in Adegenet to

310 compare Nei’s distance to EuclideanDraft distance among all pairs of locations and plotted this

311 relationship with ggplot 3.1.0 (Wickham 2016). To address the potential issues

312 hierarchical population structure can cause for analysis of isolation-by-distance

313 (Meirmans 2012), we performed a partial Mantel test between Nei’s genetic distances

314 and Euclidean distances, while conditioning on a binary matrix denoting whether pair

315 belonged to the same (0) or different (1) genetic clusters identified by admixture analysis,

316 using the software ZT 1.1 (Bonnet and Van de Peer 2002). Mantel and partial Mantel

317 tests were performed for 1,000,000 permutations.

318

319 We performed a spatial autocorrelation to investigate spatial genetic structure and infer

320 gene flow across the landscape. We calculated five relatedness coefficients between the

321 922 georeferenced individuals using SpaGeDi 1.5 (Hardy and Vekemans 2002). These

322 were Moran’s I (1950), Queller and Goodnight’s estimate of relatedness (1989), Lynch

14 https://mc06.manuscriptcentral.com/cjz-pubs Page 15 of 56 Canadian Journal of Zoology

323 and Ritland’s estimate of relatedness (1999), Wang’s relatedness estimator (2002), and

324 the measure of similarity developed by Li et al. (1993). All of the metrics showed highly

325 similar patterns; therefore, we used the Queller and Goodnight (1989) estimator of

326 pairwise relatedness for subsequent analysis. We divided the Queller and Goodnight

327 (1989) estimates of pairwise relatedness into 17 distance classes based upon the

328 Euclidean geographic distance between the pair of individuals tested. The 17 distances

329 classes ranged from 0-300 km split by 20 km intervals, with one class consisting of all

330 individuals >300 km apart. The mean values for each class were then subjected to a two-

331 tailed test, to determine if the mean relatedness of the distance class exhibited a

332 significantly positive or negative difference from zero.

333 Draft

334

335 Geographic patterns associated with mitochondrial variation, and spatial patterns of sex-

336 specific genetic variance

337 To assess differences in population genetic structure between the sexes and to make

338 inferences about sex-biased gene flow, we examined the distribution and differentiation

339 of maternally inherited haplotypes. We ran a hierarchical AMOVA in Arlequin 3.5.2.1

340 (Excoffier and Lischer 2010) for 99,999 permutations, using all sequenced individuals

341 from 18 sampling locations which were in turn grouped a priori into five regions separated

342 by the Bow River, the North Saskatchewan River, the Athabasca River, and the Smoky

343 River (Fig. 1). A haplotype network was produced in the R 3.4.4 package Adegenet. We

344 calculated Nei’s standard genetic distance and FST in the R package Adegenet and

345 Arlequin respectively. Using nuclear microsatellite data and mitochondrial FST values we

15 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 16 of 56

346 performed a calculation used by Roffler et al. (2014) to quantify sex-biased dispersal. If

347 gene flow rates are equal between the sexes mitochondrial FST values should be two to

348 four times higher than nuclear FST values (Birky et al. 1983; Larsson et al. 2009; Allendorf

349 et al. 2013). Additionally, we performed the previously described spatial autocorrelation

350 analysis for each sex separately.

351

352 Geographic patterns of genetic diversity

353 To investigate the effect of latitude on genetic diversity, we calculated the expected

354 heterozygosity and rarefied allelic richness of georeferenced sampling locations. We used

355 the same modified dataset used to investigate the effect of isolation-by-distance. Allelic

356 richness was rarefied to the numberDraft of samples in the smallest of the pooled sampling

357 locations. The relationships between expected heterozygosity and latitude, and between

358 rarefied allelic richness and latitude were tested using a Pearson’s correlation in R 3.4.4

359 (R core team 2013) and plotted with ggplot.

360

361

362 Results

363

364 Genotyping success and quality control

365 We genotyped 1,495 individuals from the 49 sampling locations across the northern

366 Rocky Mountains, of which 587 were female and 908 were male (Table 1). The average

367 genotyping success across all samples and loci was 92.88%. Mitochondrial DNA was

368 successfully sequenced in 188 of 202 individuals and aligned at a ~220 bp segment of

16 https://mc06.manuscriptcentral.com/cjz-pubs Page 17 of 56 Canadian Journal of Zoology

369 the D-loop mitochondrial region, comprising six unique haplotypes. When aligned to the

370 NCBI records for bighorn sheep our haplotypes were a 99% match with the bighorn sheep

371 mitochondrial genome (accession number: MH094035.1). The six haplotypes were

372 uploaded to the NCBI GenBank, with the accession numbers MK660218, MK660219,

373 MK660220, MK660221, MK660222, and MK660223.

374

375 Summary statistics

376 The total number alleles per locus ranged from four (BMC1222) to 15 (BM4505), with

377 BMC1222 also having the lowest observed and expected heterozygosity. We found all

378 loci deviated from Hardy-Weinberg equilibrium when samples were pooled as a single

379 population (Table 2). Across samplingDraft locations, the average number of alleles per locus

380 ranged from 2.92 to 7.85, and when rarefied to the sample size of the location with the

381 fewest individuals, allelic richness ranged from 2.45 to 3.37. Observed heterozygosity

382 ranged from 0.47 to 0.78, while expected heterozygosity ranged from 0.51 to 0.73 across

383 sampling locations. Wright’s inbreeding coefficient within each sampling location ranged

384 from -0.186 to 0.094 (full results in Table 1). The average pairwise FST distances and

385 Nei’s standard genetic distances between all 49 sampling locations were 0.09 and 0.187

386 respectively (table of genetic distances available in Table S1).

387

388 Broad-scale spatial genetic structure

389 Two clusters of individuals were apparent in a plot of the first two principal components

390 (Fig. S1). One cluster contained all samples south of the Bow River including Fernie, Elko

391 and Radium (FR), Castle Yarrow (CY), South of Bow (SO), Sheep River (SR), and

17 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 18 of 56

392 Radium (RD), while the other cluster contained all sampling locations north of the Bow

393 River with geographically close sampling locations grouped together. Principal

394 coordinates one and two explained 5.42% and 3.35% of the variance respectively. The

395 Bayesian clustering analysis indicated between four and six genetic clusters in the entire

396 dataset (Figs. 2, S2, and S3); K=4 was indicated as the most probable K in the likelihood

397 plots (Fig. S4). North of the Athabasca River, two genetic clusters were apparent (Figs.

398 2, S2, and S3) and K=2 was most supported by the likelihood plots (Fig. S4). For the

399 region between the Bow and Athabasca Rivers, two genetic clusters were apparent (Figs.

400 2, S2, and S3), and K=2 was the most supported by the likelihood plots (Fig. S4). From

401 the Bayesian clustering analysis, we concluded there are five genetic clusters in the entire

402 dataset defined by major rivers. TheDraft clusters (as seen in Fig. 2) correspond to regions

403 south of the Bow River (dark orange, K4 plot), between the Bow and the North

404 Saskatchewan Rivers (light green, K2* plot), between the North Saskatchewan and the

405 Athabasca Rivers (dark blue, K2* plot),between the Athabasca and the Smoky Rivers

406 (dark green, K2* plot), and north of the Smoky River (red, K2* plot). The geographic

407 distribution of genetic clusters are shown in Figure 1.

408

409 Our AMOVA of 1,495 individuals was significant (p < 0.01) at all hierarchical levels with

410 major river valleys accounting for 5% of the total genetic variance. Differences among the

411 49 sampling locations explained 8%, while another 8% was among individuals within

412 locations, and the remaining 79% was explained by differences within individuals.

413 Sampling locations separated by the Athabasca, North Saskatchewan, and Smoky Rivers

414 were observed to be more genetically distant than expected for their geographic

18 https://mc06.manuscriptcentral.com/cjz-pubs Page 19 of 56 Canadian Journal of Zoology

415 proximity, and sampling locations not separated by these rivers were less genetically

416 distant than expected (Fig. 3). Sampling locations either side of the Bow River appeared

417 to only show a trend for isolation by geographic distance (Fig. 3). Results of partial Mantel

418 tests for each river confirm the Bow River (r= 0.183, p= 0.234) did not significantly

419 influence genetic distance when controlling for Euclidean geographic distance. However,

420 the North Saskatchewan (r= 0.551, p< 0.001), Athabasca (r= 0.823, p< 0.001), and

421 Smoky (r= 0.792, p< 0.001) had a significant influence on genetic distance when

422 controlling for Euclidean geographic distance.

423

424 Fine-scale spatial genetic structure

425 A strong pattern of isolation-by-distanceDraft was apparent between Nei’s standard genetic

426 distance and Euclidean geographic distance among the 33 georeferenced sampling

427 locations (Fig. 4). This pattern was statistically significant (r= 0.84, p< 0.001), with

428 geographic distance explaining 84% of the variance in genetic distance. When controlling

429 for genetic hierarchical population genetic structure, isolation-by-distance was still

430 statistically significant (r= 0.82, p< 0.001), with geographic distance explaining 82% of the

431 variance in genetic distance. We also observed a negative relationship between pairwise

432 relatedness and increasing geographic distance across the study area. Positive average

433 relatedness values were observed out to 100 km (commonly referred to as neighborhood

434 size) and negative relatedness values observed past 200 km (Fig. 5) a trend which was

435 consistent across all relatedness estimators.

436

19 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 20 of 56

437 Geographic patterns associated with mitochondrial variation and spatial patterns of sex-

438 specific genetic variance

439 We identified six haplotypes in the mitochondrial D-loop region. The majority of individuals

440 (151) shared the M haplotype, which is found throughout the sampling area. We observed

441 no pattern in the distribution of the 201C haplotype, however haplotypes 131A, 144G,

442 and 186T span across different rivers while haplotype 239C was restricted to a single

443 sampling location (Fig. 6). Grouping locations a priori into regions by major rivers

444 accounted for 5.98% of the total genetic variance, while 34.49% of variation occurred

445 among sampling locations within regions and 59.53% of variation occurred within

446 sampling locations (p< 0.001). The haplotype network (Fig. 6), indicated the 131A, 144G,

447 186T, 201C, and 239C haplotypesDraft all differed from the M haplotype by only one

448 nucleotide each. Our mitochondrial FST values (Table S2) had a mean of 0.224, which

449 was 2.46 times higher than our mean microsatellite FST value of 0.091. Both our sex-

450 specific spatial autocorrelation and comparison of mitochondrial and microsatellite FST

451 values indicated the absence of sex-biased gene flow. When we performed spatial

452 autocorrelation upon each sex independently, positive average relatedness values were

453 observed out to 100 km for both males and females, negative average relatedness was

454 observed for males past 200 km and past 300 km for females (Fig. 5). Both our

455 comparison of mitochondrial and microsatellite FST values and sex-specific spatial

456 autocorrelation indicated the absence of sex-biased gene flow.

457

458 Spatial patterns of genetic diversity

20 https://mc06.manuscriptcentral.com/cjz-pubs Page 21 of 56 Canadian Journal of Zoology

459 We observed a reduction in genetic diversity with increasing latitude, indicated by the

460 negative correlations between the expected heterozygosity of the 33 georeferenced

461 sampling locations and increasing latitude (Pearson r= -0.645, p< 0.001, df= 31, Fig. 7),

462 and rarefied allelic richness (Pearson r= -0.579, p< 0.001, df= 31, Fig. 7) of the 33

463 georeferenced sampling locations and increasing latitude.

464

465 Discussion

466

467 Our analyses of bighorn sheep population genetic structure provide clear inferences

468 about patterns of gene flow and historical recolonization in the northern Rocky Mountains

469 region. We observed spatial geneticDraft structure consistent with the hypothesis that major

470 river valleys function as a barrier to gene flow. Additionally, we observed a significant

471 isolation-by-distance pattern within our study area, with Euclidean geographic distances

472 accounting for the majority of genetic distance between sampling locations, suggesting

473 geographically limited gene flow. Observed patterns of individual pairwise relatedness

474 indicate Rocky Mountain bighorn sheep in this region exhibit positive average relatedness

475 (neighbourhood size) up to ~100 km. From our analyses of mitochondrial markers and

476 sex-specific pairwise relatedness, we did not observe any indication of differing

477 population genetic structure between the two sexes, thus suggesting the absence of sex-

478 biased gene flow. Finally, the distribution of genetic diversity we observed indicates the

479 post-glacial recolonization of the region proceeded from the Southern refugium

480 northwards in a stepping-stone pattern.

481

21 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 22 of 56

482 We found all loci exhibited excess homozygosity when samples were pooled as a single

483 population and a mean inbreeding coefficient of 0.094, possibly resulting from the

484 Wahlund effect (Wahlund 1928). When the dataset was separated by sampling location,

485 no substantial deviation from Hardy Weinberg equilibrium was detected and the average

486 inbreeding coefficient was reduced to -0.012, thus indicating the initial deviations from

487 Hardy Weinberg equilibrium were due to population substructure. Sampling locations

488 exhibited genetic differentiation, with an average FST of 0.092 was found between all

489 sampling locations.

490

491 Broad-scale spatial genetic structure

492 We observed spatial genetic structureDraft defined by major river valleys, which functioned as

493 barriers to gene flow. The AMOVA indicated that 5% of genetic variance within the dataset

494 was among the five regions separated by four major rivers (the Bow River, the North

495 Saskatchewan River, the Athabasca River, and the Smoky River). From our principal

496 component analysis, we observed that sampling locations south of the Bow River were

497 genetically distinct from sampling locations north of the Bow River, and sampling locations

498 north of the Bow River grouped together with other sampling locations based upon

499 geographic proximity (Fig. 2). Bayesian clustering analysis performed on the whole

500 dataset indicated the presence of five genetic clusters (Figs. 1, 2, S2, S3, and S4). The

501 five clusters were defined by major rivers, a single genetic cluster was identified for each

502 of the following regions: south of the Bow River, between the Bow and North

503 Saskatchewan Rivers, between the Athabasca and Smoky Rivers, and north of the

22 https://mc06.manuscriptcentral.com/cjz-pubs Page 23 of 56 Canadian Journal of Zoology

504 Smoky River. This overall pattern indicated population genetic structure correlated with

505 the inter-river regions, and these major rivers functioned as barriers to gene flow.

506

507 When further examined by region specific isolation-by-distance plots and partial Mantel

508 tests, we observed the North Saskatchewan, Athabasca, and Smoky Rivers to have an

509 effect on genetic distance (Fig. 3). The Bow River did not appear to have an effect on

510 genetic distance despite individuals from south of the Bow River appearing highly

511 dissimilar from individuals from the main range (Fig. 3). Gene flow between sampling

512 locations either side of the Bow River is already rare due to the geographic distances

513 between sampling locations, thus we did not detect any effect of isolation caused by the

514 Bow River. Instead we observed geneticDraft distances between sampling locations which are

515 consistent with a pattern of isolation-by-distance (Fig. 3). The presence of major rivers

516 appears to be a barrier to the gene flow of bighorn sheep populations, when populations

517 are located within close geographic proximity to major rivers.

518

519 Although our analyses identify major rivers as barriers to gene flow, we cannot infer

520 whether river valleys are a physical barrier to gene flow, or represent lowland, heavily

521 vegetated habitats far from escape terrain, which are avoided by

522 bighorn sheep. The genetic structure of bighorn sheep is consequently correlated with

523 areas of high elevation similar to thinhorn sheep and mountain goats (Worely et al. 2004;

524 Shafer et al. 2011; Sim et al. 2016). We cannot infer whether it is solely rivers, the lowland

525 habitats rivers represent, or a combination of both which fragments the landscape for

526 bighorn sheep. We conclude that the primary barriers to the gene flow of Rocky Mountain

23 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 24 of 56

527 bighorn sheep are river valleys, as a result of the natural landscape features associated

528 with them.

529

530 Fine-scale spatial genetic structure

531 Across the ~680 km mostly linear sampling area we observed a strong pattern of isolation-

532 by-distance, characterised by a high correlation between Euclidean geographic distance

533 and genetic distance (Fig. 4). Euclidean distance accounted for 82% of genetic distance

534 when controlling for genetic hierarchical spatial genetic structure. Thus, beneath the

535 hierarchical population genetic structure caused by major rivers, geographic distance is

536 a major factor determining the genetic variation of Rocky Mountain bighorn sheep in the

537 northern Rocky Mountains. When comparedDraft to other large mammals from the region, the

538 isolation-by-distance pattern observed in our data has a stronger correlation than reported

539 in brown bears (Forbes and Hogg 1999), caribou (Weckworth et al. 2013) and grey wolves

540 (Forbes and Hogg 1999; Cullingham et al. 2016). It is unsurprising bighorn sheep from

541 this region show such a strong pattern of isolation-by-distance. Bighorn sheep are habitat

542 specialists; thus, their dispersal is more limited than more generalist species. Our data

543 indicate a stronger correlation between Euclidean geographic and genetic distance than

544 previously observed in Rocky Mountain bighorn sheep (Forbes and Hogg 1999), however

545 this is likely a result of our more intense sampling regime concentrated in a smaller area.

546

547 The strong pattern of isolation-by-distance is indicative of a stepping-stone model of

548 population structure; with gene flow occurring frequently among geographically close sub-

549 populations, however as sub-populations become separated by more distance gene flow

24 https://mc06.manuscriptcentral.com/cjz-pubs Page 25 of 56 Canadian Journal of Zoology

550 declines (Kimura and Weiss 1964). Previous studies document male breeding migrations,

551 in which males leave their home range to reproduce with other nearby populations (Hogg

552 2000). This behaviour fits the stepping-stone model (Kimura and Weiss 1964); breeding

553 migrations result in localized gene flow, thus genetic distance between sub-populations

554 increases with geographic separation. This clinal genetic differentiation due to geographic

555 separation is typical for any species occupying a region of greater size than its maximal

556 dispersal capability (Kimura and Weiss 1964).

557

558 We found bighorn sheep exhibit significant spatial genetic structure across the northern

559 Rocky Mountains. The results of our spatial autocorrelation analysis indicated average

560 positive relatedness up to 100 km betweenDraft pairs of individuals (Fig. 5), commonly known

561 as the neighbourhood size. The neighbourhood size indicates the geographical extent to

562 which individuals are more related than expected, thus inferring the range at which gene

563 flow occurs frequently. Although individual movement cannot be quantified from gene

564 flow, gene flow cannot occur in the absence of movement. Our finding corroborates the

565 telemetry data from DeCesare and Pletscher (2006) and observational data from Festa-

566 Bianchet (1986a), indicating that bighorn sheep are capable of movements over

567 significant distances.

568

569 Sex-biased gene flow

570 To assess the presence or absence of sex-biased gene flow, we examined mitochondrial

571 DNA sequence data from a subset of individuals and performed sex-specific spatial

572 autocorrelation across all individuals. A lack of variation in the mitochondrial d-loop region

25 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 26 of 56

573 we sequenced restricted the extent to which we could assess female population genetic

574 structure. We identified six haplotypes, the most common haplotype (M haplotype)

575 occurred in ~80% of individuals and was distributed across the entire region. Most other

576 haplotypes appeared to be geographically localized (Fig. 6). Similar to the pattern

577 observed in nuclear markers, the AMOVA found the presence of major rivers to account

578 for 5.98% of genetic variance across our sampling range, thus indicating the gene flow of

579 maternally inherited markers was equally restricted by the presence of major rivers. Our

580 comparisons of mitochondrial marker and nuclear marker FST distances indicated no sex-

581 biased gene flow. Mitochondrial FST distance values had a mean 2.46 times higher than

582 the mean of the microsatellite FST distance values. In the absence of sex-biased gene

583 flow, mitochondrial marker distancesDraft are expected to be two to four times greater than

584 the microsatellite marker FST distances. Therefore, indicating the absence of sex-biased

585 gene flow, similar to the situation described for thinhorn sheep (Roffler et al. 2014). We

586 also did not observe sex differences in spatial autocorrelation based on nuclear markers.

587 Thus, our analyses of mitochondrial and nuclear markers exhibited no evidence of sex-

588 biased gene flow.

589

590 We conclude that each sex contributes to gene flow equally, however previous literature

591 on female philopatry (Festa-Bianchet 1986b), male movement (Festa-Bianchet 1986a),

592 and male migrations (Hogg 2000) suggest male-biased gene flow is biologically probable.

593 Our lack of evidence for male-biased dispersal possibly resulted from a combination of

594 sampling methods and male behaviour. Field observations have shown that males

595 perform seasonal breeding migrations but return to their home range after the rut (Hogg

26 https://mc06.manuscriptcentral.com/cjz-pubs Page 27 of 56 Canadian Journal of Zoology

596 2000). We obtained the majority of our samples during the winter, after males would have

597 already returned from their breeding migrations. Therefore, males may have been

598 sampled in close proximity to their location of birth, inhibiting our ability to detect any

599 difference between male and female neighbourhood size. Although our data did not

600 support the hypothesis of male-biased gene flow, we cannot rule out the possibility that

601 the sex-specific pattern of spatial genetic structure may change at different times

602 throughout the year.

603

604 Post-glacial recolonization

605 We observed significant evidence that bighorn sheep recolonized the northern Rocky

606 Mountains from the Southern refugiumDraft following the cessation of the Wisconsin

607 glaciation. The reduction in the expected heterozygosity and rarefied allelic richness of

608 sampling locations with increasing latitude (Fig. 7), and the strong pattern of isolation-by-

609 distance observed, indicates the recolonization of the northern Rocky Mountains

610 northwards in a stepping-stone (Kimura and Weiss 1964) fashion from the Southern

611 refugium. Our finding aligns with previous theories (Cowan 1940; Geist 1999), genetic

612 studies of bighorn sheep across the species range (Luikart and Allendorf 1996), and

613 studies of other taxa which recolonized this region from the Southern refugium (Rueness

614 et al. 2003; Runck and Cook 2005; Godbout et al. 2008; Latch et al. 2009; McDevitt et al.

615 2009; Gayathiri et al. 2012). Based upon both our findings and the prior literature we

616 confidently conclude that the Southern refugium sourced the recolonization of the

617 northern Rocky Mountains by bighorn sheep.

618

27 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 28 of 56

619 Conclusion

620 In summary, in the northern Rocky Mountains, Rocky Mountain bighorn sheep population

621 genetic structure correlates with inter-river regions, and isolation-by-distance explains

622 genetic distance within and across the broad-scale population genetic structure.

623 Additionally, spatial genetic structure indicates gene flow up to 100 km, with no evidence

624 of sex-biased dispersal (but this is likely related to limitations of our sampling methods).

625 Finally, patterns of genetic diversity indicate the post-glacial recolonization of the northern

626 Rocky Mountains by bighorn sheep was sourced from the Southern refugium. Our results

627 also have wider implications beyond just mountain sheep. Our study characterizes the

628 population genetic structure of a habitat specialist across a vast area relatively

629 unfragmented by anthropogenic disturbance.Draft Second, our findings highlight the aversion

630 that habitat specialists exhibit toward non-optimal habitats, and how the avoidance of non-

631 optimal habitat can function to fragment populations and restrict gene flow across the

632 landscape. Third, our inability to detect differential population genetic structure between

633 the sexes highlights the difficulty of studying sex-bias gene flow in species which do not

634 exhibit typical permanent dispersal. Finally, patterns of bighorn sheep post-glacial

635 recolonization in the Rocky Mountains add further support to the growing body of

636 knowledge suggesting that many vertebrates moved northwards to recolonize

637 northwestern North America from Southern refugia.

638

639 Acknowledgments

640 We thank Anh Dao for her assistance with laboratory work, and Rhiannon Peery for

641 bioinformatical assistance. We also thank Corey Davis, Joshua Miller, and the entirety

28 https://mc06.manuscriptcentral.com/cjz-pubs Page 29 of 56 Canadian Journal of Zoology

642 of Coltman lab for their helpful discussion and comments throughout this project. We

643 also thank Kathreen E. Ruckstuhl (NSERC RGPIN-2018-03913), Peter Neuhaus, and

644 Nigel Caulkett, and their team for sharing tissue samples from Sheep River bighorns.

645 We thank Alberta Fish and Wildlife especially Anne Hubbs, Mark Bradley of Jasper

646 National Park, and Banff National Park for providing Rocky Mountain bighorn sheep

647 species range data. Finally, we thank our funding sources: the Alberta Conservation

648 Association, Alberta Fish and Game Association, Government of Alberta (Environment

649 and Parks), Canadian Wildlife Federation, Natural Sciences and Engineering Research

650 Council of Canada (NSERC RGPIN-2018-04354 to DWC), and the Wild Sheep

651 Foundation.

652 Draft

653

654

655

656

657

658

659

660

661

662

663

664

29 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 30 of 56

665 References

666 Adams, R.V., and Burg, T.M. 2015. Influence of ecological and geological features on

667 rangewide patterns of genetic structure in a widespread passerine. Heredity,

668 114(2): 143-154.

669 Beatty, G.E., and Provan, J. 2010. Refugial persistence and postglacial recolonization

670 of North America by the cold‐tolerant herbaceous plant (Orthilia secunda). Mol.

671 Ecol. 19(22): 5009-5021.

672 Birky, C.W., Maruyama, T., and Fuerst, P. 1983. An approach to population and

673 evolutionary genetic theory for genes in mitochondria and chloroplasts, and some 674 results. Genetics, 103(3): 513-527.Draft

675 Bishop, M.D., Kappes, S.M., Keele, J.W., Stone, R.T., Sunden, S.L.F., Hawkins, G.A.,

676 et al. 1994. A Genetic Linkage Map for Cattle. Genetics, 136(2): 619-639.

677 Bleich, V.C. 2015. Mountain Sheep in California: Perspectives on the Past, and

678 Prospects for the Future. Proceedings of the Northern Wild Sheep and Goat

679 Council, 15: 1-13.

680 Bonnet, E., and Van de Peer, Y. 2002. ZT: a software tool for simple and partial Mantel

681 tests. J. Stat. Softw. 7(10): 1.

682 Boyce, W.M., Ramey, R.R., Rodwell, T.C., Rubin, E.S., and Singer, R.S. 1999.

683 Population subdivision among desert bighorn sheep (Ovis canadensis) ewes

684 revealed by mitochondrial DNA analysis. Mol. Ecol. 8(1): 99-106.

30 https://mc06.manuscriptcentral.com/cjz-pubs Page 31 of 56 Canadian Journal of Zoology

685 Brown, W.E., Hu, J.C., and Athanasiou, K.A. 2016. Ammonium–Chloride–Potassium

686 lysing buffer treatment of fully differentiated cells increases cell purity and resulting

687 neotissue functional properties. Tissue Eng. Part C Methods, 22(9): 895-903.

688 Buchanan, F.C., and Crawford, A.M. 1993. Ovine microsatellites at the OarFCB11,

689 OarFCBl28, OarFCB193, OarFCB266 and OarFCB304 loci. Anim. Genet. 24(2):

690 145.

691 Buchanan, F.C., and Crawford, A.M. 1992. Ovine dinucleotide repeat polymorphism at

692 the MAF209 locus. Anim. Genet. 23(2): 183.

693 Buchanan, F.C., Swarbrick, P.A., and Crawford, A.M. 1992. Ovine dinucleotide repeat 694 polymorphism at the MAF65 locus.Draft Anim. Genet. 23(1): 85.

695 Burns, J.A. 2010. Mammalian faunal dynamics in Late Pleistocene Alberta,

696 Canada. Quat. Int. 217(1-2): 37-42.

697 Chester, C.C. 2015. Yellowstone to Yukon: Transborder conservation across a vast

698 international landscape. Environ. Sci. Policy, 49:75-84.

699 Cowan, I.M. 1940. Distribution and variation in the native sheep of North America. Am.

700 Midl. Nat. 24(3): 505-580.

701 Cullingham, C.I., Thiessen, C.D., Derocher, A.E., Paquet, P.C., Miller, J.M., Hamilton,

702 J.A., and Coltman, D.W. 2016. Population structure and dispersal of wolves in the

703 Canadian Rocky Mountains. J. Mammal. 97(3): 839-851.

704 DeCesare, N.J., and Pletscher, D.H. 2006. Movements, connectivity, and resource

705 selection of Rocky Mountain bighorn sheep. J. Mammal. 87(3): 531-538.

31 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 32 of 56

706 Dieringer, D., and Schlötterer, C. 2003. Microsatellite analyser (MSA): a platform

707 independent analysis tool for large microsatellite datasets. Mol. Ecol. Notes, 3(1):

708 167-169.

709 Driscoll, C.C., Driscoll, J.G., Hazekamp, C., Mitton, J.B., and Wehausen, J.D. 2015. A

710 tale of two markers: Population genetics of Colorado Rocky Mountain bighorn

711 sheep estimated from microsatellite and mitochondrial data. J. Wildl. Manage.

712 79(5): 819-831.

713 Dueck, G.S. 1998. Genetic relations and phylogeography of woodland and

714 barrenground caribou. M.Sc. thesis, Department of bioscience, The University of 715 Alberta, Edmonton, AB. Draft 716 Earl, D.A., and vonHoldt, B.M. 2012.STRUCTURE HARVESTER: a website and

717 program for visualizing STRUCTURE output and implementing the Evanno

718 method. Conserv. Genet. Resour. 4(2): 359-361.

719 Ede, A.J., Pierson, C.A., and Crawford, A.M. 1995. Ovine microsatellites at the OarCP9,

720 OarCP16, OarCP20, OarCP21, OarCP23 and OarCP26 loci. Anim. Genet. 26(2):

721 129-130.

722 Epps, C.W., Crowhurst, R.S., and Nickerson, B.S. 2018. Assessing changes in

723 functional connectivity in a desert bighorn sheep metapopulation after two

724 generations. Mol. Ecol. 27(10): 2334-2346.

725 Epps, C.W., Palsbøll, P.J., Wehausen, J.D., Roderick, G.K., Ramey, R.R., and

726 McCullough, D.R. 2005. Highways block gene flow and cause a rapid decline in

727 genetic diversity of desert bighorn sheep. Ecol. Lett. 8(10): 1029-1038.

32 https://mc06.manuscriptcentral.com/cjz-pubs Page 33 of 56 Canadian Journal of Zoology

728 Evanno, G., Regnaut, S., and Goudet, J. 2005. Detecting the number of clusters of

729 individuals using the software structure: a simulation study. Mol. Ecol.14(8): 2611-

730 2620.

731 Excoffier, L., and Lischer, H.E.L. 2010. Arlequin suite ver 3.5: a new series of programs

732 to perform population genetics analyses under Linux and Windows. Mol. Ecol.

733 Resour. 10(3): 564-567.

734 Festa-Bianchet, M. 1986a. Site fidelity and seasonal range use by bighorn rams. Can. J.

735 Zool. 64(10): 2126-2132.

736 Festa-Bianchet, M. 1986b. Seasonal Dispersion of Overlapping Mountain Sheep Ewe 737 Groups. J. Wildl. Manage. 50(2):Draft 325-330.

738 Festa-Bianchet, M. 1991. The social system of bighorn sheep: grouping patterns,

739 kinship and female dominance rank. Anim. Behav. 42(1): 71-82.

740 Festa-Bianchet, M. 1988. Seasonal Range Selection in Bighorn Sheep: Conflicts

741 between Forage Quality, Forage Quantity, and Predator Avoidance. Oecologia,

742 75(4): 580-586.

743 Flagstad, O., and Røed, K.H. 2003. Refugial Origins of Reindeer (Rangifer tarandus L.)

744 Inferred from Mitochondrial DNA Sequences. Evolution, 57(3): 658-670.

745 Forbes, S.H., and Hogg, J.T. 1999. Assessing population structure at high levels of

746 differentiation: microsatellite comparisons of bighorn sheep and large

747 carnivores. Anim. Conserv. 2(3): 223-233.

33 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 34 of 56

748 Francis, R.M. 2017. POPHELPER: an R package and web app to analyse and visualize

749 population structure. Mol. Ecol. Resour. 17(1): 27-32.

750 Frankham, R. 1997. Do island populations have less genetic variation than mainland

751 populations?. Heredity, 3(78): 311-327.

752 Gayathiri Samarasekera, G. D. N., Bartell, N.V., Lindgren, B.S., Cooke, J.E.K., Davis,

753 C.S., James, P.M.A., et al. 2012. Spatial genetic structure of the mountain pine

754 beetle (Dendroctonus ponderosae) outbreak in western Canada: historical patterns

755 and contemporary dispersal. Mol. Ecol. 21(12): 2931-2948.

756 Geist, V. 1971. Mountain sheep a study in behaviour and evolution, 1st edn, The 757 University of Chicago Press, London.Draft

758 Georges, M., and Massey, J. 1992. Polymorphic DNA markers in Bovidae. World

759 Intellectual Property Org., Geneva. WO Publ.

760 Godbout, J., Fazekas, A., Newton, C.H., Yeh, F.C., and Bousquet, J. 2008. Glacial

761 vicariance in the Pacific Northwest: evidence from a lodgepole pine mitochondrial

762 DNA minisatellite for multiple genetically distinct and widely separated refugia. Mol.

763 Ecol. 17(10): 2463-2475.

764 Government of Alberta. 2015. Management plan for bighorn sheep in Alberta. Available

765 at: https://open.alberta.ca/dataset/8106954c-67d8-45b9-a5ca-

766 90c0bf218c8f/resource/b1485228-428b-47ed-aaaa-5cede694e8a4/download/2015-

767 bighornsheepmgmtplan-draft-jun25-2015a.pdf

34 https://mc06.manuscriptcentral.com/cjz-pubs Page 35 of 56 Canadian Journal of Zoology

768 Gutiérrez-Espeleta, G., Kalinowski, S., Boyce, W.M., and Hedrick, P. 2000. Genetic

769 variation and population structure in desert bighorn sheep: implications for

770 conservation. Conserv. Genet. 1(1): 3-15.

771 Hardy, O.J., and Vekemans, X. 2002. SPAGeDi: a versatile computer program to

772 analyse spatial genetic structure at the individual and population levels. Mol. Ecol.

773 Notes, 2(3): 618-620.

774 Hewitt, G.M. 2004. Genetic consequences of climatic oscillations in the Quaternary.

775 Philos. Trans. R. Soc. Lond. B Biol. Sci. No. 359(1442): 183-195.

776 Hogg, J.T. 2000. Mating system and conservation at large spatial scales. In Vertebrate 777 mating systems. Edited by M. Apollonio,Draft M. Festa-Bianchet and D. Mainardi. World 778 scientific, Singapore. pp. 214-252.

779 Janes, J.K., Miller, J.M., Dupuis, J.R., Malenfant, R., Gorrell, J.C., Cullingham, C.I., and

780 Andrew, R.L., 2017. The K= 2 conundrum. Mol. Ecol. 26(14): 3594-3602.

781 Jombart, T. 2008. adegenet: a R package for the multivariate analysis of genetic

782 markers. Bioinformatics, 24(11): 1403-1405.

783 Jørgensen, S., Hamrick, J.L., and Wells, P.V. 2002. Regional patterns of genetic

784 diversity in Pinus flexilis (Pinaceae) reveal complex species history. Am. J. Bot.

785 89(5): 793-800.

786 Kalinowski, S.T., Taper, M.L., and Marshall, T.C. 2007. Revising how the computer

787 program cervus accommodates genotyping error increases success in paternity

788 assignment. Mol. Ecol. 16(5):1099-1106.

35 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 36 of 56

789 Kearse, M., Moir, R., Wilson, A., Stones-Havas, S., Cheung, M., Sturrock, et al.

790 2012. Geneious Basic: an integrated and extendable desktop software platform for

791 the organization and analysis of sequence data. Bioinformatics, 28(12): 1647-1649.

792 Keyghobadi, N., Roland, J., and Strobeck, C. 1999. Influence of landscape on the

793 population genetic structure of the alpine butterfly Parnassius smintheus

794 (Papilionidae). Mol. Ecol. 8(9): 1481-1495.

795 Keyghobadi, N., Roland, J., and Strobeck, C. 2005. Genetic differentiation and gene

796 flow among populations of the alpine butterfly, Parnassius smintheus, vary with

797 landscape connectivity. Mol. Ecol. 14(7): 1897-1909.

798 Kimura, M., and Weiss, G.H. 1964. TheDraft stepping stone model of population structure 799 and the decrease of genetic correlation with distance. Genetics, 49(4): 561-576.

800 Kopelman, N.M., Mayzel, J., Jakobsson, M., Rosenburg, N.A., and Mayors, I. 2015.

801 CLUMPAK: a program for identifying clustering modes and packaging population

802 structure inferences across K. Mol. Ecol. Resour. 15(5): 1179-1191.

803 Larsson, L., Charlier, J., Laikre, L., and Ryman, N. 2009. Statistical power for detecting

804 genetic divergence—organelle versus nuclear markers. Conserv. Genet. 10(5):

805 1255-1264.

806 Latch, E.K., Heffelfinger, J.R., Fike, J.A., and Rhodes, O.E.J. 2009. Species-wide

807 phylogeography of North American mule deer (Odocoileus hemionus): cryptic

808 glacial refugia and postglacial recolonization. Mol. Ecol. 18(8): 1730-1745.

36 https://mc06.manuscriptcentral.com/cjz-pubs Page 37 of 56 Canadian Journal of Zoology

809 Lawson Handley, L.J., and Perrin, N. 2007. Advances in our understanding of

810 mammalian sex‐biased dispersal. Mol. Ecol. 16(8): 1559-1578.

811 Leonard, J.A., Vilà, C., and Wayne, R.K. 2005. Legacy lost: genetic variability and

812 population size of extirpated US grey wolves (Canis lupus). Mol. Ecol. 14(1): 9-17.

813 Li, C.C., Weeks, D.E., and Chakravarti, A. 1993. Similarity of DNA fingerprints due to

814 chance and relatedness. Hum. Hered. 43(1): 45-52.

815 Loehr, J., Worley, K., Grapputo, A., Carey, J., Veitch, A., and Coltman, D.W. 2006.

816 Evidence for cryptic glacial refugia from North American mountain sheep

817 mitochondrial DNA. J. Evol. Biol. 19(2): 419-430.

818 Luikart, G., and Allendorf, F.W. 1996.Draft Mitochondrial-DNA variation and genetic-

819 population structure in Rocky Mountain bighorn sheep (Ovis canadensis

820 canadensis). J. Mammal. 77(1): 109-123.

821 Luikart, G., Amish, S., Winnie, J., Beja-Pereira, A., Godinho, R., Allendorf, F., and

822 Harris, R. 2011. High connectivity among argali sheep from Afghanistan and

823 adjacent countries: Inferences from neutral and candidate gene

824 microsatellites. Conserv. Genet. 12(4): 921-931.

825 Lynch, M., and Ritland, K. 1999. Estimation of pairwise relatedness with molecular

826 markers. Genetics, 152(4): 1753-1766.

827 McDevitt, A.D., Mariani, S., Hebblewhite, M., Decesare, N.J., Morgantini, L., Seip, D.,

828 Weckworth, B.V., and Musiani, M. 2009. Survival in the Rockies of an endangered

37 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 38 of 56

829 hybrid swarm from diverged caribou (Rangifer tarandus) lineages. Mol. Ecol. 18(4):

830 665-679.

831 Meirmans, P.G. 2012. The trouble with isolation by distance. Mol. Ecol. 21(12): 2839-

832 2846.

833 Miller, J.M., Poissant, J., Hogg, J.T., and Coltman, D.W. 2012. Genomic consequences

834 of genetic rescue in an insular population of bighorn sheep (Ovis canadensis). Mol.

835 Ecol. 21(7): 1583-1596.

836 Mitton, J.B., Kreiser, B.R., and Latta, R.G. 2000. Glacial refugia of limber pine (Pinus

837 flexilis James) inferred from the population structure of mitochondrial DNA. Mol. 838 Ecol. 9(1): 91-97. Draft

839 Moran, P.A.P. 1950. Notes on continuous stochastic phenomena. Biometrika, 37(1-2):

840 17-23.

841 Nei, M. 1972. Genetic distances between populations. Am. Nat. 106: 283–292.

842 Peakall, R., and Smouse, P.E. 2006. genalex 6: genetic analysis in Excel. Population

843 genetic software for teaching and research. Mol. Ecol. Notes, 6(1): 288-295.

844 Penty, J.M., Henry, J.M., Ede, A.J., and Crawford, A.M. 1993. Ovine Microsatellites at

845 the OarAE16, OarAE54, OarAE57, OarAE119, and OarAE129 loci. Anim. Genet.

846 24(3):219.

847 Poirier, M.A., Coltman, D.W., Pelletier, F., Jorgenson, J. and Festa‐Bianchet, M. 2019.

848 Genetic decline, restoration and rescue of an isolated ungulate population. Evol.

849 Appl. 12(7): 1318-1328.

38 https://mc06.manuscriptcentral.com/cjz-pubs Page 39 of 56 Canadian Journal of Zoology

850 Poole, K.G., Serrouya, R., Teske, I.E., and Podrasky, K. 2016. Rocky Mountain bighorn

851 sheep (Ovis canadensis canadensis) winter habitat selection and seasonal

852 movements in an area of active coal mining. Can. J. Zool. 94(11): 733-745.

853 Pritchard, J.K., Stephens, M., and Donnelly, P. 2000. Inference of population structure

854 using multilocus genotype data. Genetics, 155(2): 945-959.

855 Proctor, M., and Paetkau, D. 2004. A genetic-based spatial analysis of grizzly bears in

856 Alberta. Alberta Sustainable Resource Development: Fish and Wildlife Division: 1-

857 27 pp.

858 Pusey, A.E. 1987. Sex-biased dispersal and inbreeding avoidance in birds and 859 mammals. Trends Ecol. Evol. 2(10):Draft 295-299.

860

861 Queller, D.C., and Goodnight, K.F. 1989. Estimating relatedness using genetic markers.

862 Evolution, 43(2): 258-275.

863 Rioux-Paquette, E., Festa-Bianchet, M., and Coltman, D.W. 2010. No inbreeding

864 avoidance in an isolated population of bighorn sheep. Anim. Behav. 80(5): 865-871.

865 Roffler, G., Talbot, S., Luikart, G., Sage, G., Pilgrim, K., Adams, L., and Schwartz, M.

866 2014. Lack of sex-biased dispersal promotes fine-scale genetic structure in alpine

867 ungulates. Conserv. Genet. 15(4): 837-851.

868 Rogala, J.K., Hebblewhite, M., Whittington, J., White, C.A., Coleshill, J., and Musiani, M.

869 2011. Human activity differentially redistributes large mammals in the Canadian

870 Rockies National Parks. Ecol. Soc. 16(3): 16.

39 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 40 of 56

871 Rubin, E.S., Boyce, W.M., Jorgensen, M.C., Torres, S.G., Hayes, C.L., O'Brien, C.S.,

872 and Jessup, D.A. 1998. Distribution and abundance of bighorn sheep in the

873 peninsular ranges, California. Wildl. Soc. Bull. 26(3): 539-551.

874 Ruckstuhl, K.E. 1998. Foraging behaviour and sexual segregation in bighorn

875 sheep. Anim. Behav. 56(1): 99-106.

876 Rueness, E.K., Stenseth, N.C., O'Donoghue, M., Boutin, S., Ellegren, H., and Jakobsen,

877 K.S. 2003. Ecological and genetic spatial structuring in the Canadian lynx. Nature,

878 425(6953): 69-72.

879 Runck, A.M., and Cook, J.A. 2005. Postglacial expansion of the southern red‐backed

880 vole (Clethrionomys gapperi) in DraftNorth America. Mol. Ecol. 14(5): 1445-1456.

881 Serrouya, R., Paetkau, D., McLellan, B.N., Boutin, S., Campbell, M. and Jenkins, D.A.

882 2012. Population size and major valleys explain microsatellite variation better than

883 taxonomic units for caribou in western Canada. Mol. Ecol. 21(11): 2588-2601.

884 Shafer, A.B.A., Cullingham, C.I., Côté, S.D., and Coltman, D.C. 2010. Of glaciers and

885 refugia: a decade of study sheds new light on the phylogeography of northwestern

886 North America. Mol. Ecol. 19(21): 4589-4621.

887 Shafer, A.B.A., Côté, S.D., and Coltman, D.W. 2011. Hot spots of genetic diversity

888 descended from multiple Pleistocene refugia in an alpine ungulate. Evolution, 65(1):

889 125-138.

890 Shannon, N.H., Hudson, R.J., Brink, V.C., and Kitts, W.D. 1975. Determinants of spatial

891 distribution of Rocky Mountain bighorn sheep. J. Wildl. Manage. 2(39): 387-401.

40 https://mc06.manuscriptcentral.com/cjz-pubs Page 41 of 56 Canadian Journal of Zoology

892 Sharma, M., Singh, A., Sharma, N., and Rawat, S. 2012. Embryo sexing in cattle. J.

893 Mamm. Ova Res. 29(3): 119-123.

894 Sim, Z., Davis, C.S., Jex, B., Hegel, T., and Coltman, D.W. 2018. Management

895 implications of highly resolved hierarchical population genetic structure in thinhorn

896 sheep. Conserv. Genet. 20(2): 1-17.

897 Sim, Z., Hall, J.C., Jex, B., Hegel, T.M., and Coltman, D.W. 2016. Genome‐wide set of

898 SNPs reveals evidence for two glacial refugia and admixture from postglacial

899 recolonization in an alpine ungulate. Mol. Ecol. 25(15): 3696-3705.

900 Slatkin, M. 1985. Gene flow in natural populations. Annu. Rev. Ecol. Syst. 16(1):393-

901 430. Draft

902 Slatkin, M. 1987. Gene flow and the geographic structure of natural

903 populations. Science, 236(4803): 787-792.

904 Stone, K.D., and Cook, J.A. 2000. Phylogeography of black bears (Ursus americanus)

905 of the Pacific Northwest. Can. J. Zool. 78(7): 1218-1223.

906 Swarbrick, P.A., Buchanan, F.C., and Crawford, A.M. 1991a. Ovine dinucleotide repeat

907 polymorphism at the MAF36 locus. Anim Genet. 22(4): 377-378.

908 Swarbrick, P.A., Buchanan, F.C., and Crawford, A.M. 1991b. Ovine dinucleotide repeat

909 polymorphism at the MAF64 locus. Anim Genet. 22(4): 375-376.

910 Valière, N. 2002. gimlet: a computer program for analysing genetic individual

911 identification data. Mol. Ecol. Notes, 2(3): 377-379.

41 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 42 of 56

912 Wahlund, S. 1928. Zusammensetzung von populationen und korrelationserscheinungen

913 vom standpunkt der vererbungslehre aus betrachtet. Hereditas, 11(1): 65-106.

914 Wang, J. 2011. coancestry: a program for simulating, estimating and analysing

915 relatedness and inbreeding coefficients. Mol. Ecol. Resour. 11(1): 141-145.

916 Wang, J. 2002. An estimator for pairwise relatedness using molecular

917 markers. Genetics, 160(3): 1203-1215.

918 Weckworth, B.V., Musiani, M., Decesare, N.J., McDevitt, A.D., Hebblewhite, M., and

919 Mariani, S. 2013. Preferred habitat and effective population size drive landscape

920 genetic patterns in an endangered species. Proc. R. Soc. Lond. B Biol. Sci. 921 280(1769): 1-9. Draft

922 Weir, B.S., and Clark Cockerham, C. 1984. Estimating F-statistics for the analysis of

923 population structure. Evolution, 38(6): 1358-1370.

924 Wilson, G.A., Strobeck, C., Wu, L., and Coffin, J.W. 1997. Characterization of

925 microsatellite loci in caribou (Rangifer tarandus), and their use in other

926 artiodactyls. Mol. Ecol. 6(7): 697-699.

927 Worely, K., Strobeck, C., Arthur, S., Carey, J., Schwantje, H., Veitch, A., and Coltman,

928 D.W. 2004. Population genetic structure of North American thinhorn sheep (Ovis

929 dalli). Mol. Ecol. 13(9): 2545-2556.

930 Wright, S. 1942. Isolation by distance. Genetics, 28(2): 114-137.

931

932

42 https://mc06.manuscriptcentral.com/cjz-pubs Page 43 of 56 Canadian Journal of Zoology

933 Figure captions

934 Fig. 1. Sampling locations and their admixture of each of the five genetic clusters

935 identified, with major river valleys highlighted in blue. See Table 1 for sampling location

936 names and other details. Insert a shows the geographic location of our study area.

937 Insert b shows the species range of Rocky Mountain bighorn sheep (Ovis canadensis

938 canadensis) in Canada (blue crosshatch) and the distribution of our sampling locations.

939 *Due to the resolution of this figure the location of EK (Etna Knoll) and MP (Morro Peak)

940 may appear ambiguous, we confirm both sampling locations are located south of the

941 Athabasca River. [Figure was created using ArcMap version 10.6.1. Base map of

943 Canada 2017]. Draft

944

945 Fig. 2. Admixture plot for K4 for all individuals sampled, and separate plots of K2 for the

946 region between the Bow and Athabasca river and the region north of the Athabasca river

947 (K2* each region analysed separately). The clusters represent south of the Bow River

948 (dark orange), between the Bow and the North Saskatchewan Rivers (light green),

949 between the North Saskatchewan and the Athabasca Rivers (dark blue), between the

950 Athabasca and the Smoky Rivers (dark green), and north of the Smoky River (red). Black

951 bars represent the major rivers.

952

953 Fig. 3. Regression of Nei’s standard genetic distance (1972) on Euclidean geographic

954 distance for sampling locations situated on the same and different sides of major rivers,

955 represented by filled and unfilled circles respectively. The effect of distance on genetic

43 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 44 of 56

956 distance in each region is as follows: Athabasca River, r= 0.736, p<0.001. Bow River, r=

957 0.938, p<0.001. North Saskatchewan River, r= 0.487, p<0.001. Smoky River, r= 0.281,

958 p=0.041. The effect of each river on genetic distance when controlling for geographic

959 distance is as follows: Athabasca River, r= 0.823, p<0.001. Bow River, r=0.183, p=0.234.

960 North Saskatchewan River, r= 0.551, p<0.001. Smoky River, r= 0.792, p<0.001.

961

962 Fig. 4. Regression of Nei’s (1972) standard genetic distance on Euclidean geographic

963 distance between 33 sampling locations of Rocky Mountain bighorn sheep (Ovis

964 canadensis canadensis) in the northern portion of the species range (r= 0.844, p<0.001).

965

966 Fig. 5. Average pairwise Queller andDraft Goodnight estimate of relatedness (1989) by

967 distance class. The lower and upper limit of the box are the 25th and 75th percentile

968 respectively, the middle bar represents the median, and the whiskers extend to the

969 furthest data point which is within 1.5 times the inter-quartile range. An asterisk below +

970 or – indicates a distance class mean pairwise relatedness significantly more or less than

971 zero. Mean relatedness of the total population shown as dashed line with a value of -

972 0.0014.

973

974 Fig. 6. The haplotype network and the proportion of the haplotypes 131A (blue), M

975 (red), 144G (yellow), 201C (pink), 186T (green), and 239C (cyan), at each sampling

976 location. Plain red circles indicate populations with no variation from the M haplotype.

977 Major rivers highlighted in blue. [Figure was created using ArcMap version 10.6.1. Base

44 https://mc06.manuscriptcentral.com/cjz-pubs Page 45 of 56 Canadian Journal of Zoology

978 map of provincial, state, and North American boundaries, and shaded relief from ESRI

980

981

982 Fig. 7. Correlations between the expected heterozygosity of the 33 georeferenced

983 sampling locations and latitude (Pearson r=-0.645, p<0.001, df=31), and between the

984 rarefied allelic richness of the 33 georeferenced sampling locations and latitude (Pearson

985 r=-0.579, p<0.001, df=31).

Draft

45 https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 46 of 56

Tables

Table 1. Description of the 49 sampling locations, including site abbreviation (Abrv), latitude (Lat), longitude (Long), number of females typed at microsatellite loci (♀), number of males typed at microsatellite loci (♂), total number of individuals typed at microsatellite loci (Total), number of individuals with mtDNA sequence (mtDNA seq), number of alleles (A), allelic richness corrected for small sample size (Ar), observed heterozygosity (Hobs), expected heterozygosity (Hexp), and Wright’s inbreeding coefficient (f). An asterisk (*) indicates samples from this location were submitted by hunters.

mtDNA Sampling location Abrv Lat Long ♂ ♀ Total seq A Ar Hobs Hexp f Castle Yarrow CY 49.27 -114.20 0 27 27 7 5.62 2.94 0.69 0.67 -0.033 Radium RD 50.64 -116.05 2 2 4 2.92 2.57 0.56 0.56 0.000 Fernie, Elko, Radium* FR 107 5 112 7.85 3.37 0.66 0.72 0.091 South of Bow* SO 72Draft10 82 6.77 3.25 0.70 0.72 0.028 Sheep River SR 50.64 -114.66 18 12 30 10 5.38 3.30 0.72 0.71 -0.011 Forbidden Creek Trail FC 51.82 -115.93 9 13 22 10 4.62 2.95 0.68 0.67 -0.022 Malloch Creek- Banff Park MB 51.88 -115.95 21 2 23 5.23 2.97 0.73 0.67 -0.101 Lost Guide LO 51.94 -116.00 0 6 6 3.85 3.23 0.68 0.75 0.094 South Ram River Headquarters RH 51.95 -116.16 7 11 18 5.00 2.99 0.66 0.66 0.009 Ram Range RG 52.07 -116.15 10 14 24 9 5.00 3.10 0.70 0.69 -0.009 Cline River CR 52.20 -116.62 6 6 12 10 4.85 3.02 0.74 0.68 -0.091 Abraham Lake AL 52.22 -116.26 9 5 14 7.38 3.33 0.70 0.73 0.048 Ram River* RR 154 0 154 5.30 3.06 0.67 0.66 -0.003 Ram Mountain RM 52.36 -115.79 32 34 66 12 4.77 3.20 0.74 0.72 -0.037 Job Creek JC 52.37 -116.72 6 5 11 10 4.62 3.21 0.78 0.72 -0.084 Bighorn River BR 52.42 -116.51 9 0 9 4.62 3.07 0.71 0.69 -0.035 Opabin B OB 52.43 -116.58 8 4 12 5.15 3.12 0.69 0.67 -0.025 Aztec Lake AZ 52.46 -116.99 0 12 12 10 5.00 3.16 0.73 0.70 -0.039

https://mc06.manuscriptcentral.com/cjz-pubs Page 47 of 56 Canadian Journal of Zoology

Opabin A OA 52.54 -116.72 3 5 8 3.91 3.07 0.61 0.64 0.046 Blackstone BS 52.55 -116.53 4 9 13 3.77 2.76 0.61 0.64 0.046 Isaac Creek IC 52.55 -117.01 8 11 19 5.00 3.05 0.70 0.68 -0.022 Cairn River CA 52.69 -117.12 8 4 12 10 4.85 3.12 0.70 0.70 -0.004 Russel Creek RC 52.82 -117.15 2 26 28 5.31 3.05 0.75 0.67 -0.119 Grave Flats GF 52.90 -117.01 8 0 8 4.23 3.02 0.64 0.66 0.030 Drummond DR 52.95 -117.38 8 0 8 4.23 3.06 0.71 0.67 -0.052 Cadomin Mountain CM 52.97 -117.20 6 10 16 16 4.92 2.96 0.68 0.65 -0.050 Cadomin* CD 69 5 74 6.54 3.15 0.66 0.69 0.032 Luscar-Gregg Mine LG 53.06 -117.42 31 143 174 20 6.77 3.19 0.69 0.69 -0.006 Mystery Lake ML 53.13 -117.67 19 22 41 10 5.77 3.04 0.66 0.68 0.019 Makwa Ridge MR 53.05 -117.79 10Draft15 25 5.23 3.03 0.66 0.68 0.036 Morro Peak MP 53.04 -118.08 4 3 7 3.46 2.78 0.65 0.64 -0.026 Etna Knoll EK 53.07 -118.02 6 3 9 4.08 3.11 0.69 0.65 -0.071 Disaster Point DP 53.18 -117.96 3 6 9 4.38 3.13 0.66 0.66 0.008 Folding Mountain FM 53.21 -117.69 16 5 21 5.08 2.93 0.67 0.64 -0.036 Windy Point WP 53.21 -117.99 2 6 8 3.46 2.79 0.60 0.64 0.057 Brule Tunnel BT 53.23 -117.87 1 20 21 10 4.54 2.78 0.64 0.62 -0.031 Stornoway Mountain ST 53.29 -118.39 7 9 16 10 4.00 2.54 0.54 0.56 0.026 Munn Creek MC 53.52 -118.32 8 11 19 8 4.77 2.93 0.65 0.63 -0.032 Monaghan MO 53.61 -119.05 3 11 14 3.85 2.77 0.68 0.59 -0.138 Rocky Pass RP 53.63 -118.83 4 17 21 4.15 2.75 0.59 0.60 0.019 North Berland NB 53.67 -118.76 13 8 21 12 4.31 2.84 0.65 0.63 -0.032 Wilmore* WM 72 3 75 5.15 2.85 0.61 0.62 0.019 Mount Mawdsley MM 53.80 -119.12 4 10 14 3.77 2.72 0.66 0.60 -0.097 Stearn A SA 53.91 -119.29 11 26 37 10 4.69 2.80 0.58 0.61 0.048 Horn Creek HC 53.97 -119.53 4 9 13 3.77 2.61 0.58 0.57 -0.023

https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 48 of 56

Grande Cache Coal Mine GC 54.03 -119.20 13 22 35 4.23 2.56 0.50 0.53 0.062 Smoky River* SM 76 0 76 4.62 2.52 0.51 0.53 0.047 Gorman Mountain GM 54.18 -120.01 1 6 7 3.08 2.45 0.47 0.51 0.073 Narraway/Mount Torrens NW 54.27 -119.91 2 6 8 10 3.62 2.67 0.67 0.57 -0.186

Draft

https://mc06.manuscriptcentral.com/cjz-pubs Page 49 of 56 Canadian Journal of Zoology

Table 2. Descriptive statistics for 13 microsatellite loci and one sex-determining marker used to quantify genetic diversity in Rocky Mountain bighorn sheep (Ovis canadensis canadensis), including: locus name, number of genotypes, genotyping success rate, number of alleles (k), allele size range (bp), observed heterozygosity of entire dataset(Hobs), expected heterozygosity of entire dataset (Hexp), and Wright’s inbreeding coefficient (f).

Percentage Size range Locus Source Genotypes success (%) k (bp) Hobs Hexp f BM1225 Bishop et al., 1994 1407 94.11 11 237-261 0.653 0.735 0.112

BM4505 Bishop et al., 1994 1371 91.71 15 263-301 0.661 0.716 0.077

BMC1222 Bishop et al., 1994 1437 96.12 4 286-292 0.317 0.340 0.068

FCB266 Buchanan & Crawford, 1993 1434 95.92 6 88-102 0.715 0.770 0.071 MAF209 Buchanan & Crawford, 1992 1475 Draft98.66 9 107-123 0.717 0.751 0.045 MAF36 Swarbrick, Buchanan & Crawford, 1991a 1477 98.79 7 90-108 0.629 0.693 0.092 MAF64 Swarbrick, Buchanan & Crawford, 1991b 1374 91.91 8 109-123 0.726 0.788 0.079 MAF65 Buchanan, Swarbick & Crawford, 1992 1349 90.23 12 101-135 0.681 0.794 0.142 OarAE16 Plenty et al. 1993 1453 97.19 11 82-106 0.775 0.854 0.093 OarCP26 Ede, Pierson & Crawford, 1995 1408 94.18 12 131-159 0.632 0.724 0.127 Rt9 Wilson et al., 1997 1159 77.53 11 119-147 0.734 0.797 0.079 TGLA122 Georges and Massey, 1992 1194 79.87 10 129-146 0.624 0.718 0.131 TGLA387 Georges and Massey, 1992 1406 94.05 13 132-155 0.701 0.783 0.104 SheepAML Gokulakrishnan et al., 2013 1495 100 2 121-168 - - -

https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 50 of 56

Draft

https://mc06.manuscriptcentral.com/cjz-pubs Page 51 of 56 CanadianDraft Journal of Zoology https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 52 of 56

Draft

https://mc06.manuscriptcentral.com/cjz-pubs Page 53 of 56 Canadian Journal of Zoology

Draft

https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 54 of 56

Draft

https://mc06.manuscriptcentral.com/cjz-pubs Page 55 of 56 Canadian Journal of Zoology

Draft

https://mc06.manuscriptcentral.com/cjz-pubs Canadian Journal of Zoology Page 56 of 56

Draft

https://mc06.manuscriptcentral.com/cjz-pubs