Population Genetic Structure of Moose (Alces Alces)Of South-Central Alaska

POPULATION GENETIC STRUCTURE OF MOOSE (ALCES ALCES)OF SOUTH-CENTRAL ALASKA

Robert E. Wilson1,2, Thomas J. McDonough3, Perry S. Barboza1, Sandra L. Talbot4, and Sean D. Farley5

1University of Alaska Fairbanks, Institute of Arctic Biology, Fairbanks, Alaska 99775; 3Alaska Department of Fish and Game, 3298 Douglas St., Homer, Alaska 99603; 4U. S. Geological Survey, Alaska Science Center, 4210 University Drive, Anchorage, Alaska 99508; 5Alaska Department of Fish and Game, 333 Raspberry Road, Anchorage, Alaska 99518

ABSTRACT: The location of a population can influence its genetic structure and diversity by impact- ing the degree of isolation and connectivity to other populations. Populations at range margins are often thought to have less genetic variation and increased genetic structure, and a reduction in genetic diversity can have negative impacts on the health of a population. We explored the genetic diversity and connectivity between 3 peripheral populations of moose (Alces alces) with differing potential for connectivity to other areas within interior Alaska. Populations on the Kenai Peninsula and from the Anchorage region were found to be significantly differentiated (FST = 0.071, P < 0.0001) with lower levels of genetic diversity observed within the Kenai population. Bayesian analyses employing assignment methodologies uncovered little evidence of contemporary gene flow between Anchorage and Kenai, suggesting regional isolation. Although gene flow outside the peninsula is restricted, high levels of gene flow were detected within the Kenai that is explained by male-biased dispersal. Furthermore, gene flow estimates differed across time scales on the Kenai Peninsula which may have been influenced by demographic fluctuations correlated, at least in part, with habitat change.

ALCES VOL. 51: 71–86 (2015)

Key words: Alaska, genetic diversity, gene flow, moose, population genetic structure

The pattern of geographical variation in the range margins and highest at the center genetic diversity and divergence is dictated of a species distribution (Yamashita and by the interaction of genetic drift, gene Polis 1995, Schwartz et al. 2003, Eckert et al. flow, and natural selection (Eckert et al. 2008, Howes and Lougheed 2008). Marginal 2008), and these evolutionary processes can populations are more likely to be isolated, be influenced by the location of a population occur in patchy habitats, and may reflect within the species’ geographic range (Briggs recent colonization. Peripheral populations 1996, Wisely et al. 2004, Howes and Lough- are less likely to receive immigrants whereas eed 2008). At the local and regional scales, the core populations typically occupy prime the relative location of a population can habitat and experience greater levels of strongly impact patterns of dispersal and gene flow (Hoffmann and Blows 1994, degree of isolation influenced by both histor- Brown et al. 1995, Wisely et al. 2004, Miller ical and contemporary events (Vucetich and et al. 2010, Schrey et al. 2011). Waite 2003, Eckert et al. 2008), ultimately Evolutionary theory suggests that the determining the level of genetic structure reduction of genetic diversity within periph- and diversity. Genetic diversity is lowest at eral populations can impede adaptation to

2Present address: U. S. Geological Survey, Alaska Science Center, 4210 University Drive, Anchorage, Alaska 99508 71 POPULATION GENETIC STRUCTURE – WILSON ET AL. ALCES VOL. 51, 2015 changing environmental conditions (Brad- moose populations on the Kenai have fluctu- shaw 1991, Hoffmann and Parsons 1991, ated between 5,000–8,000 animals over the Hoffmann and Blows 1994, Blows and Hoff- past several decades (T. J. McDonough, mann 2005). Such adaptation is largely unpublished data), these fluctuations have determined by the availability of additive not been uniform across moose management genetic variation in heritable traits with fit- units on the peninsula. While population size ness consequences. Several studies have in Game Management Unit (GMU) 15C in shown that even small changes in genetic southwest Kenai has increased, numbers in variation can have large effects on popula- GMU 15A (northwest) have declined drasti- tion fitness (Frankham 1995, Reed and cally, ∼40% in the last 20 years as quality Frankham 2003) including juvenile survival forage has diminished since the last major (Coulson et al. 1999, Mainguy et al. 2009, fire in 1969. Relative isolation from neigh- Silva et al. 2009), antler growth (Von Hard- boring regions with a strong history of fluc- enberg et al. 2007), and parasite resistance tuations in population size might lead to (Coltman et al. 1999) within ungulates. reduced genetic variability on the Kenai However, adaptability is also affected by Peninsula which could ultimately be detri- effective dispersal which can have either mental to the long-term health of moose in positive or negative effects on the population this region. depending on the rate of gene flow and the Using microsatellite loci, we compared strength of selection acting on the local levels of genetic variation and gene flow in population (García-Romos and Kirkpatrick 2 areas within the Kenai Peninsula and the 1997, Akerman and Bürger 2014, Bourne Anchorage area. These 3 areas are situated et al. 2014, Frankham 2015). Thus, examin- on the periphery of overall moose distribu- ing conditions (habitat, genetic diversity, tion in Alaska but differ in levels of potential gene flow rates, and life history) under connectivity to the core area of interior which peripheral populations exist can aid Alaska. First, we investigated the connectiv- in understanding the processes that maintain ity between GMUs on the Kenai Peninsula geographical ranges, predicting the conse- that have been affected by a long history of quences of climate change (Parmesan and land alteration and demographic changes. Yohe 2003, Root et al. 2003, Hampe and Second, we predicted that 2 sites within the Petit 2005), and conserving populations at disjunct Kenai Peninsula region, where range margins (Howes and Lougheed 2008). opportunities for genetic exchange may be The Kenai Peninsula is a peripheral more limiting than Anchorage, would exhi- region situated in south-central Alaska that bit relatively lower genetic diversity. was separated from the mainland by a nar- row (16 km wide) isthmus at the end of the METHODS last ice age. Due to its diverse landscape, Sample collection biodiversity in this region is unusually high A total of 163 moose were sampled from at this latitude (Morton et al. 2009), and the 3 populations in south-central Alaska (Fig. 1). moose (Alces alces) is one of the most recog- Ear-plugs and blood were taken from 33 col- nizable and socio-economically important lared female moose in 2008–2010 and 2012 species. Moose populations on the Kenai from the city of Anchorage and adjacent Peninsula are characterized by fluctuations Eagle River (called Anchorage hereafter). in population size, peaking after the occur- In addition, muscle tissue was taken from rence of forest fires that promote optimal for- 32 hunter-killed moose (16 female, 15 age habitat (Oldemeyer et al. 1977). While male, and 1 unknown) during the winter of

72 ALCES VOL. 51, 2015 WILSON ET AL. – POPULATION GENETIC STRUCTURE

with dinucleotide repeat motifs were selected for further analysis that were polymorphic in all populations: BL42, BM888, BM203, BM2830 (Bishop et al. 1994), NVHRT21, NVHRT22 (Røed and Midthjell 1998), RT1, RT5, and RT30 (Wilson et al. 1997). Polymerase chain reaction (PCR) amplifica- tion and electrophoresis followed protocols described in Roffler et al. (2012). Ten per- cent of the samples were amplified and gen- otyped in duplicate for the 9 microsatellite loci for quality control. Fig. 1. Sampling areas for three moose populations in south-central Alaska: An- Analysis of genetic diversity and chorage, Game Management Unit (GMU) population genetic subdivision 15A (northwest Kenai Peninsula), and We calculated allelic richness, inbreed- GMU15C (southwest Kenai Peninsula). ing coefficient (FIS), observed and expected heterozygosities, and tested for deviations – 2011 2012. In spring 2012, blood was taken from Hardy-Weinberg equilibrium (HWE) from radio-collared female moose from and linkage disequilibrium (LD) for each 2 GMU 15A (n = 49; 3,367 km ) and GMU microsatellite locus and population in 2 15C (n = 49; 3,030 km ) on the Kenai Penin- FSTAT ver. 2.9.3 (Goudet 1995). The degree sula, the borders of which are approximately of genetic subdivision among moose popula- 20 km apart. Anchorage samples are archived tions was assessed by calculating overall and at the Molecular Ecology Laboratory, U.S. pairwise FST and RST, adjusting for multiple Geological Survey, Anchorage, Alaska, and comparisons using Bonferroni correction (α Kenai Peninsula samples at the Alaska = 0.05) in Arlequin v3.5.1.3 (Excoffier and Department of Fish and Game, Homer, Lischer 2010). Because the upper possible Alaska. All animal capturing and genetic FST value for a set of microsatellite loci is sampling were conducted under Division of usually <1.0 (Hedrick 2005), we used Wildlife Conservation ACUC approval (# RECODEDATA, version 1.0 (Meirmans – – – 2012 2007, 2013 2021, and 90 05) and 2006) to calculate the uppermost limit of under the University of Alaska Fairbanks FST for our data set. IACUC approval (# 14885 and 182744). We also used the Bayesian-clustering program STUCTURE 2.2.3 (Pritchard et al. Molecular techniques 2000) to determine the level of population Genomic DNA was extracted from blood structure in the autosomal microsatellite and tissue samples using a “salting out” pro- data set. We performed 2 sets of analyses to cedure described by Medrano et al. (1990), look at structure within south central Alaska: with modifications described in Sonsthagen 1) between Anchorage and Kenai Peninsula et al. (2004). Genomic DNA concentrations and 2) within the Kenai Peninsula (GMU were quantified using fluorometry and 15A and 15C). Structure assigns individuals diluted to 50 ng mL–1 working solutions. to populations maximizing Hardy-Weinberg Individuals were initially screened at 17 equilibrium and minimizing linkage disequi- microsatellite loci. Thirteen autosomal loci librium. The analysis was conducted for 1– were found to be polymorphic of which 9 10 populations (K) using an admixture

73 POPULATION GENETIC STRUCTURE – WILSON ET AL. ALCES VOL. 51, 2015 model with 100,000 burn-in iterations and hereafter) whereas MIGRATE gene flow 1,000,000 Markov chain Monte Carlo estimates are averaged over the past n gen- (MCMC) iterations without providing a erations, where n equals the number of priori information on the geographic origin generations the populations have been at of the individuals; the analyses were mutation-drift equilibrium (Beerli and repeated 10 times for each K to ensure con- Felsenstein 1999, 2001). It is generally sistency across runs. We used the ▵K agreed that microsatellite mutation rates method of Evanno et al. (2005) and evalua- are several orders of magnitude higher than tion of the estimate of the posterior probabil- mutation rates of DNA sequences (mito- ity of the data given K, Ln P(D), to chondrial or nuclear; Schlötterer 2000, determine the most likely number of groups Ellegren 2004). Thus, microsatellite markers at the uppermost level of population struc- can reflect recent (within the last 10,000 ture. For the Kenai Peninsula analysis we years) and almost contemporaneous events, used the LOCPRIOR which is able to detect but increases in homoplasy associated with weak signals of population structure in data- microsatellites reduce their ability to capture sets not detectable under standard models older demographic events (Hartl and Clark (Hubisz et al. 2009). We determined if loca- 2007, Hughes 2010). Therefore, MIGRATE tion was informative by the value of r, which analyses are referred to as estimating recent parameterizes the amount of information gene flow. contained by the location of the samples. MIGRATE was run with a full migration Values of r > 1 indicates either there is no model; θ (4Neµ, composite measure of effec- population structure or that structure is inde- tive population size and mutation rate) and pendent of locality. all pairwise migration parameters were estimated individually from the data. Gene Gene flow flow was estimated using maximum likeli- We estimated gene flow between moose hood search parameters; 10 short chains populations using 2 methodologies: (5000 trees used out of 1,500,000 sampled), MIGRATE v3.2.16 (Beerli and Felsenstein 10 long chains (15,000 trees used out 1999, 2001) and BayesAss 3.0 (Wilson and 5,250,000 sampled), and 5 static heated Rannala 2003). These programs differ in chains (1, 1.33, 2.0, 4.0, and 1,000,000; the underlying model used to estimate gene swapping interval = 1). Full models were flow. MIGRATE uses a steady-state two- run 10 times to ensure the convergence of island coalescent model of population differ- parameter estimates. entiation which incorporates parameters BayesAss was initially run with the scaled to the mutation rate (µ), the effective default delta values for allelic frequency population size parameter Θ (4Neµ), and (P), migration rate (m), and inbreeding (F). the migration rate M (m/µ) between popula- Subsequent runs incorporated different delta tions. BayesAss uses an assignment metho- values to ensure that acceptance rate for pro- dology which does not incorporate posed changes was between 20–40% for genealogy or assume that populations are in each parameter to maximize log likelihood Hardy-Weinberg equilibrium (Wilson and values and ensure the most accurate esti- Rannala 2003). Thus, estimates of migration mates (Wilson and Rannala 2003). Final rate can be interpreted differently and at dif- delta values used were ▵P = 0.5 (27% accep- ferent temporal scales. BayesAss reflects tance rate), ▵m = 0.2 (27%), and ▵F = 0.85 gene flow over the last several generations (31%). We performed 10 independent runs (referred to as contemporary gene flow (10 million iterations, 1 million burn-in,

74 ALCES VOL. 51, 2015 WILSON ET AL. – POPULATION GENETIC STRUCTURE and sampling frequency of 1000) and 2 addi- (Cornuet and Luikart 1996). BOTTLENECK tional longer runs (50 million iterations, 5 compares heterozygote deficiency and excess million burn-in) with different random seeds relative to genetic diversity, not to Hardy- to ensure convergence and consistency Weinberg equilibrium expectation (Cornuet across runs. Convergence was also assessed and Luikart 1996). by examining the trace file in program Tracer v1.5 to ensure proper mixing of parameters RESULTS (Rambaut and Drummond 2007). Genetic diversity and population subdivision Population demography Multilocus genotypes were collected To estimate the effective population size from 163 individuals and each individual (Ne) for each GMU on the Kenai Peninsula had a unique genotype. The number of and Anchorage area, we used the approxi- alleles per locus observed ranged from 3.4– mate Bayesian computation method (Beau- 4.7 per population with an overall estimate mont et al. 2002) implemented in the of 5.1 (Tables 1, 2). The observed heterozyg- program ONeSAMP 1.2 (Tallmon et al. osity ranged from 43–55% with an overall 2008). We used a lower prior of 100 for all mean heterozygosity of 49%. The Kenai populations and a maximum prior that Peninsula exhibited a 19% lower allelic rich- reflected the current census size (1,000 for ness (20% in GMU 15A and 25% in GMU Anchorage, 2,000 for GMU 15A, and 3,000 15C) compared to the Anchorage area, and for GMU 15C). Similar values were obtained 3x more private alleles were observed in for larger maximum possible effective popu- the Anchorage region (Table 1). In addition, lation sizes. the observed (H ) and expected (H ) hetero- Lastly, we used BOTTLENECK which o e zygosity was significantly lower (all compares the number of alleles and gene diversity at polymorphic loci under the infi- P-values < 0.0001) in the Kenai Peninsula nite allele model (IAM; Maruyama and (Ho by 18%, He by 16% expected), in Fuerst 1985), stepwise mutation model GMU 15C (15%, 14%), and in GMU 15A (SMM; Ohta and Kimura 1973), and two- (22%, 20%). On average, individuals on the phase model of mutation (TPM; Di Rienzo Kenai showed a greater level of homozygos- et al. 1994; parameters: 79% SSM, variance ity (Kenai: 4.94 loci [SD = 1.46] vs. Ancho- 9; Piry et al. 1999, Garza and Williamson rage: 3.98 loci [SD = 1.51]; t = 4.02, P < 2001). One thousand simulations were per- 0.0001). The inbreeding coefficient (FIS) formed for each population and parameters did not differ significantly from zero in any were changed among 5 runs to evaluate the population (Table 1). All loci and popula- robustness of results. Significance was tions were in HWE and linkage equilibrium. assessed using a Wilcoxon sign-rank test Significant genetic structure was which determines if the average of standar- observed at the 9 microsatellite loci between dized differences between observed and Anchorage and the two GMUs on the Kenai expected heterozygosities is significantly Peninsula (Table 3). No significant differ- different from zero (Cornuet and Luikart ence was found within the Kenai Peninsula. 1996). Significant heterozygote deficiency The upper limit of the FST for our microsa- relative to the number of alleles indicates tellite data set was 0.499. Therefore, the recent population growth, whereas heterozy- overall FST of 0.071 accounted for 14.2% gote excess relative to the number of alleles of the maximum possible level of genetic indicates a recent population bottleneck structure and 19% for the pairwise

75 POPULATION GENETIC STRUCTURE – WILSON ET AL. ALCES VOL. 51, 2015

Table 1. Estimates of genetic diversity of the moose sampled from three locales in south-central Alaska, including: average number of alleles, allelic richness (AR), observed and expected heterozygosities (Ho/He), inbreeding coefficient (FIS), effective population size (Ne) estimated in ONeSAMP and sample size (n) calculated from nine microsatellite loci. Allelic richness is based on smallest sample size of 65 for Anchorage and overall Kenai. Within Kenai Peninsula (GMU 15A and GMU 15C) based on sample size of 49. Kenai Peninsula

Anchorage GMU 15A GMU 15C Overall Kenai No. Alleles 4.67 3.67 3.44 4.00 No. Private Alleles 10 1 2 3

AR 4.59 3.67 3.44 3.78

Ho (SD) / 0.55 (0.02)/ 0.43 (0.02)/ 0.47 (0.02)/ 0.45 (0.02)/ He (SD) 0.56 (0.06) 0.45 (0.07) 0.48 (0.05) 0.47 (0.06)

FIS 0.007 0.056 0.031 0.043

Ne 74.3 (67.6–83.0) 47.9 (43.2–56.8) 36.8 (33.3–43.9) 145.5 (123.3–255.9) n65494998 comparisons between Anchorage and Kenai northern direction of contemporary gene Peninsula GMUs. flow from GMU 15C into 15A (proportion STRUCTURE uncovered genetic parti- of individuals with migrant origin: 27.8% tioning within south central moose popula- in 15A vs. 6.9% in 15C); although the 95% tions, supporting a two-population model confidence intervals do overlap (Fig. 3). (ΔK = 188.3, average Ln P(D) = −2758.7). Asymmetrical recent gene flow as esti- Most individuals from Anchorage were mated by MIGRATE was observed among assigned to one genetic cluster (87.7%), sampled populations. The directionality of whereas individuals from Kenai GMU 15A gene flow was from Kenai Peninsula into and 15C were assigned to a second cluster Anchorage (Fig. 3). The number of migrants with high probability, 93.6 and 92.6%, per generation (Nem) ranged from 2.56 respectively (Fig. 2). Seven Anchorage indi- (GMU 15A; 1.97–3.29) and 2.78 (GMU viduals were assigned to the Anchorage clus- 15C; 2.20–3.48) into Anchorage and 0.99 ter with <60% certainty, conversely, only a (0.74–1.32) and 1.08 (0.78–1.47) into the single Kenai individual was assigned to the Kenai GMU 15A and 15C, respectively. Kenai cluster with <60% certainty. Genetic Within Kenai there was a signal of asymme- partitioning was not observed within Kenai trical gene flow from GMU 15A into 15C Peninsula, as including capture location (3.3 migrants/generation; Fig. 3). (LOCIPRIOR) was not informative (r > 9). Population demography Gene flow The estimated effective size using the Restricted gene flow over the past sev- Bayesian computation method for the eral generations was observed under the Anchorage region was 74.3 (95% CI: 67.6– BayesAss model between Anchorage and 83.0). GMUs 15A and 15C on the Kenai Kenai Peninsula, with 96.8% (93.3–100%) Peninsula had lower estimated effective sizes of the Anchorage population comprised of with non-overlapping confidence intervals with a non-migrant origin (Fig. 3). Within the Anchorage (Table 1). The BOTTLENECK Kenai Peninsula, there was a signal of a analysis showed no evidence of significant

76 ALCES VOL. 51, 2015 WILSON ET AL. – POPULATION GENETIC STRUCTURE

Table 2. Estimates of observed and expected heterozygosity, number of alleles per locus for nine autosomal nuclear microsatellite loci assayed in three moose populations in south-central Alaska. All loci were in Hardy-Weinberg equilibrium. Sample size is in parentheses; Ho = heterozygosity observed, He = heterozygosity expected, and Na = number of alleles. Kenai Peninsula

Anchorage GMU 15A GMU 15C Overall Kenai All populations Locus (65) (49) (49) (98) (165)

NVHRT22 Ho/He 0.69/0.76 0.49/0.54 0.57/0.53 0.53/0.53 0.60/0.68

Na 65 4 6 6

NVHRT21 Ho/He 0.49/0.50 0.55/0.46 0.39/0.45 0.47/0.45 0.48/0.48

Na 53 2 3 5

RT1 Ho/He 0.49/0.46 0.27/0.29 0.39/0.38 0.31/0.34 0.38/0.40

Na 22 2 2 2

RT5 Ho/He 0.54/0.52 0.18/0.21 0.25/0.32 0.21/0.26 0.34/0.40

Na 43 3 3 4

RT30 Ho/He 0.55/0.58 0.69/0.67 0.74/0.72 0.71/0.70 0.65/0.66

Na 54 4 4 5

BM203 Ho/He 0.20/0.20 0.37/0.41 0.51/0.50 0.44/0.46 0.34/0.38

Na 53 4 4 6

BM2830 Ho/He 0.46/0.49 0.37/0.43 0.41/0.41 0.39/0.42 0.42/0.45

Na 22 2 2 2

BM888 Ho/He 0.63/0.65 0.22/0.26 0.20/0.27 0.21/0.27 0.38/0.46

Na 44 3 4 4

BL42 Ho/He 0.91/0.84 0.71/0.81 0.80/0.75 0.76/0.79 0.82/0.83

Na 96 7 8 12

Overall Loci Ho/He 0.55/0.56 0.43/0.45 0.47/0.48 0.45/0.47 0.49/0.53

Na 4.67 3.67 3.44 4.00 5.11

Table 3. Pairwise and overall values of FST and excess) based on the infinite allele model RST calculated from nine microsatellite loci. (IAM) for Kenai GMU 15C. Significant values after Bonferroni correction (P < 0.0001) are marked with an asterisk. DISCUSSION FST RST Climatic and glaciation history has Anchorage played a major role in shaping the evolution- – Kenai GMU 15A 0.094* 0.014 ary history of many taxa in south-central – Kenai GMU 15C 0.092* 0.028 Alaska. It was not until approximately Kenai GMU 15A 7,000 years before present that the Kenai – Kenai GMU 15C 0.001 0.000 Peninsula became distinct and relatively iso- Overall 0.071* 0.016 lated from the mainland by a 16 km wide mountainous isthmus (Pielou 1991, Muhs heterozygosity excess or deficit under the et al. 2001). This isolation has fostered SMM or TPM. However, there was evidence genetically or morphologically distinct popu- of a recent population decline (heterozygote lations for a variety of taxa (e.g., wolverine

77 POPULATION GENETIC STRUCTURE – WILSON ET AL. ALCES VOL. 51, 2015

1.0

0.9

0.8

0.7

0.6

0.5

0.4

Probability of Assignment of Probability 0.3

0.2

0.1

0.0 Anchorage Kenai GMU 15A Kenai GMU 15C

Fig. 2. Structure analysis showing posterior probability of assignment of individuals to each (K = 2) genetic cluster. White bar represents the estimated probability of assignment to cluster one and grey bar is the estimated probability of assignment to cluster two.

2.56 0.01 (1.97-3.29) (0.00-0.04) 0.99 0.01 (0.74-1.32) (0.00-0.02)

0.28 1.18 1.08 2.78 (0.13-0.43) 0.01 (0.89-1.54) (0.78-1.47) 0.02 3.29 (2.20-3.48) (0.00-0.05) (0.00-0.03) (2.63-4.07) 0.07 (0.00-0.25)

Fig. 3. Estimates of (a) recent (number of migrants per generation, Nem) estimated in MIGRATE and (b) contemporary (proportion of individuals with migrant origin, m) calculated in BayesAss for moose populations in south-central Alaska as calculated from nine microsatellite loci with relative magnitude indicated by width of arrow; 95% conﬁdence intervals are in parentheses.

[Gulo gulo], Tomasik and Cook 2005; Amer- populations residing on the Kenai are no ican marten [Ursus americanus], Robinson exception. Using a multi-locus approach, et al. 2007; song sparrow [Melospiza melo- we observed that moose on the Kenai were dia], Patten and Pruett 2009). The moose genetically distinct from those in the

78 ALCES VOL. 51, 2015 WILSON ET AL. – POPULATION GENETIC STRUCTURE mainland Anchorage population and exhib- likelihood of long-distance dispersal between ited significantly lower levels of genetic these two regions. diversity at microsatellite loci. However, connectivity could be mediated through a contact zone north of Loss of genetic diversity between the isthmus located at Portage Valley that is peninsula and mainland used by black bears (Ursus americanus) Populations residing in areas with bar- (Robinson et al. 2007). The isthmus is not riers that limit dispersal (e.g., peninsulas and an absolute/strong barrier as movements of islands) across the landscape are expected to radio-collared moose occur across the isth- have lower genetic variation (Gaines et al. mus; this movement was restricted within 1997). Our results were consistent with intermountain valleys that spanned both Gaines et al. (1997) prediction: moose occu- sides of the isthmus (T. Lohuis, Alaska pying the Kenai Peninsula had significantly Department of Fish and Game, unpublished reduced genetic diversity (∼18%) compared data). Furthermore, STRUCTURE analysis to the nearest mainland population in Ancho- estimated a low probability assignment to a rage. A reduction of genetic variability has genetic cluster for ∼12% of the individuals also been reported for other Alaskan moose in Anchorage, suggestive of genetic populations (Hundertmark 2009, Schmidt exchange that has occurred during or after et al. 2009) as well as other mammals on the population divergence, with higher rate Kenai Peninsula (e.g., Canada lynx [Lynx going into the Anchorage area based on the canadensis], Schwartz et al. 2003). The loss MIGRATE analysis. This northward direc- of genetic variation in peripheral populations tion of gene flow is also found in other may be due to numerous factors such as lim- peninsular populations (Schmidt et al. ited number of connections to other popula- 2009) and may reflect post-colonization tions or smaller population size (Schwartz gene flow rates. Further study of moose in et al. 2003). areas between Anchorage and Kenai Penin- Cook Inlet waters, mountains, and a sula might identify if a contact zone exists highway and railways may represent formid- for moose at the isthmus as seen in other able dispersal barriers for moose between mammals, or if these regions are truly iso- these regions. Although Kenai Peninsula lated as indicated by the contemporary gene and Anchorage are in close geographic flow analysis. proximity (straight line distance over land is ∼105 km), the costs of dispersal over the Relationships within the peninsula rugged terrain and highways or swimming Unlike the potential strong barriers to across the inlet are likely high. In agreement dispersal between the peninsula and main- with limited effective dispersal, we found land populations, there are relatively few restricted contemporary gene flow between natural barriers to movement in the western Kenai Peninsula and mainland Anchorage part of the peninsula, and gene flow esti- populations with confidence intervals sug- mates suggest that there is ongoing genetic gesting there has been limited genetic exchange. The directionality of gene flow exchange over the past several generations. across the western Kenai Peninsula has not Telemetry studies of the sampled females in remained constant over time. Differences in this study showed that individuals remained directionality across time scales may be in the same general area throughout the attributed to the fluctuating nature of moose year (Farley et al. 2012, T. J. McDonough, population dynamics that is correlated at unpublished data), further suggesting a low least in part with habitat change, in particular

79 POPULATION GENETIC STRUCTURE – WILSON ET AL. ALCES VOL. 51, 2015 in GMU 15A where population size fluctu- negatively impact genetic diversity of a ates with major fire events (Oldemeyer et al. population; this may partially explain the 1977, Schwartz and Franzmann 1989, Lor- significantly low genetic diversity on the anger et al. 1991). The habitat in GMU Kenai Peninsula. A reduction in genetic 15A has changed drastically over the last diversity can lower viability and fecundity century after major fires in 1947 and 1969 (Falconer 1981, Ralls et al. 1983, Frankham transformed previously low quality habitat 1995, Crnokrak and Roff 1999), and at the to ideal foraging habitat, which subsequently extreme can lead to inbreeding depression; declined to the current condition (Oldemeyer decreased viability and fecundity occur cur- et al. 1977, Schwartz and Franzmann 1989). rently on the Kenai Peninsula (Franzmann If periodic population increase has been suf- and Schwartz 1985, ADF&G 2013, unpub- ficiently frequent throughout the history of lished data). Whether lower reproductive moose in this area, and dispersal is influ- rates (twinning rates and calf survival) on enced by population density and habitat the Kenai Peninsula are correlated solely quality, we might expect the directionality with genetic variability or are influenced in of gene flow to change over time with more addition, or solely by environmental factors, moose dispersing from areas of high produc- is an area for future investigation. tivity into areas of lower density or less preferred habitat as competition for resources Conservation implications increases. Indeed, contemporary gene flow Although the effects of inbreeding estimated in BayesAss indicates gene flow depression can diminish over time (Lynch from a higher density area (GMU 15C) 1977), a general loss of genetic diversity with better quality habitat into an area char- can be detrimental over evolutionary time acterized by poor habitat conditions and as it may lower the ability of populations to lower density area (15A). respond to environmental stressors such as Moose populations on the Kenai Penin- novel predators, parasites, or climatic condi- sula have also fluctuated in size partially tions (Lacy 1987, Quattro and Vrijenhoek due to human activities (land development 1989, Leberg 1993). Following the recom- and forest fires), with changes in habitat mendations of Frankham et al. (2014), all 3 potentially affecting fertility and survival of populations fall below the minimum effec- young (Klein 1970, Franzmann and Arneson tive population size of 1,000 required to 1973, Schwartz and Franzmann 1989, Testa maintain long-term viability. In addition, and Adams 1989). While moose populations the GMUs on the Kenai Peninsula, when initially respond positively to wildfires considered separately, have an effective through the emergence of optimal habitat, population size lower than both recent populations eventually decline as the habitat (> 100; Frankham et al. 2014) and earlier changes to late succession (non-optimal for- (> 50; Franklin 1980, Soulé 1980) recom- age) vegetation. During the 20 years follow- mendations to avoid inbreeding depression. ing the last major fire in GMU 15A (1969), Indeed, the Kenai Peninsula does have a the population has declined by approxi- higher inbreeding coefficient (although not mately 40%. Current and previous assess- significantly different from zero) and higher ment of calf survival from this area has levels of homozygosity. However, when identified low calf survival (Franzmann et al. considering the Kenai Peninsula as a single 1980, T. J. McDonough, unpublished data). population, the effective population exceeds Such a drastic decline in population size 100 but remains below the threshold for coupled with low productivity can long-term viability.

80 ALCES VOL. 51, 2015 WILSON ET AL. – POPULATION GENETIC STRUCTURE

Neutral loci are commonly used to infer does not imply endorsement by the U.S. evolutionary history of populations and Government. make inferences about overall variation (see Howes and Lougheed 2008), but it is still REFERENCES unclear whether the trends in putatively neu- AKERMAN, A., and R. BÜRGER. 2014. The con- tral loci are reflective of quantitative-trait sequences of gene flow for local adapta- variation found in genes for physiological, tion and differentiation: a two-locus morphological, or life history traits that are deme model. Journal of Mathematical likely important for the adaptive potential Biology 68: 1135–1198. of populations (Merilä and Crnokrak 2001, BEAUMONT, M. A., W. ZHANG, and D. J. Reed and Frankham 2001, Eckert et al. BALDING. 2002. Approximate Bayesian 2008). Therefore, a conclusion that reduced computation in population genetics. – genetic diversity observed at neutral micro- Genetics 162: 2025 2035. BEERLI, P., and J. FELSENSTEIN. 1999. satellite markers reflects reduction of diver- Maximum-likelihood estimation of sity in the genome overall is premature. migration rates and effective population Our results showing significant population numbers in two populations using a coa- structure and limited connectivity to outside lescent approach. Genetics 152: 763–773. populations for the Kenai Peninsula provide ———, and ———. 2001. Maximum like- a working hypothesis for the potential effects lihood estimation of a migration matrix on genetic diversity, which can be tested by and effective population sizes in n subpo- assaying both selectively neutral and func- pulations by using a coalescent approach. tional diversity. Such studies can provide Proceedings of the National Academy of greater resolution on the processes responsi- Sciences USA 98: 4563–4568. ble for the distribution of genetic diversity BISHOP, M. D., S. M. KAPPES,J.W.KEELE,R. among moose populations within south- T. STONE,S.L.F.SUNDEN,G.A.HAWKINS, central Alaska. S. S. TOLDO,R.FRIES,M.D.GROSZ,J. YOO, and C. W. BEATTIE. 1994. A genetic ACKNOWLEDGEMENTS linkage map for cattle. Genetics 136: Funding was provided by Joint Base 619–639. Elmendorf-Richardson, under the guidance BLOWS, M. W., and A. A. HOFFMANN. 2005. A of H. Griese, C. Garner, D. Battle, and R. reassessment of genetic limits to evolu- Graham, and U. S. Geological Survey. tionary change. Ecology 86: 1371–1384. Extensive field assistance was provided by BOURNE, E. C., G. BOCEDI,J.M.J.TRAVIS,R. the military conservation agent program on J. PAKEMAN,R.W.BROOKER, and K. Elmendorf Air base. J. Crouse, J. Selinger, SCHIFFERS. 2014. Between migration and pilots J. DeCreft, T. Levanger, and J. load and evolutionary rescue: dispersal, Fieldman, and the Kenai National Wildlife adaptation and the response of spatially Refuge provided field assistance on the structured populations to environmental Kenai Peninsula. Z. Grauvogel, K. Sage, change. Proceedings of Royal Society B and S. Sonsthagen provided laboratory and 281: 20132795. manuscript advice, while K. Hundertmark BRADSHAW, A. D. 1991. The Croonian Lec- provided advice on marker selection and G. ture, 1991: genostasis and the limits to Roffler and E. Solomon provided help with evolution. Philosophical Transactions of maps. Any use of trade, firm, or product the Royal Society B: Biological Sciences names is for descriptive purposes only and 333: 289–305.

81 POPULATION GENETIC STRUCTURE – WILSON ET AL. ALCES VOL. 51, 2015

BRIGGS, J. C. 1996. Biogeography and punc- Molecular Ecology Resources 10: tuated equilibrium. Biogeographica 72: 564–567. 151–156. FALCONER, D. S. 1981. An Introduction to BROWN, J. H., D. W. MEHLMAN, and G. C. Quantitative Genetics. Longman, London, STEVENS. 1995. Spatial variation in England. abundance. Ecology 76: 2028–2043. FARLEY, S., P. BARBOZA,H.GRIESE, and C. COLTMAN, D. W., J. G. PILKINGTON,J.A. GARDNER. 2012. Characterization of SMITH, and J. M. PEMBERTON. 1999. Para- moose movement patterns, movement site-mediated selection against inbred of black bears in relation to anthropo- Soay sheep in a free-living, island popu- genic food sources, and wolf distribution lation. Evolution 53: 1259–1267. and movement on JBER lands, of Elmen- COULSON, T., S. ALBON,J.SLATE, and J. PEM- dorf AFB and Fort Richardson AP. BERTON. 1999. Microsatellite loci reveal Alaska Department of Fish and Game sex-dependent responses to inbreeding Report, Juneau, Alaska, USA. and outbreeding in red deer calves. Evo- FRANKLIN, I. R. 1980. Evolutionary change in lution 53: 1951–1960. small populations. Pages 135–149 in CORNUET, J. M., and G. LUIKART. 1996. M. E. Soulé and B. A. Wilcox, editors. Description and power analysis of two Conservation Biology: An Evolutionary- tests for detecting recent population bot- Ecological Perspective. Sinauer Associ- tlenecks from allele frequency data. ates, Sunderland, Massachusetts, USA. Genetics 144: 2001–2014. FRANKHAM, R. 1995. Inbreeding and conser- CRNOKRAK, P., and D. A. ROFF. 1999. Inbreed- vation: a threshold effect. Conservation ing depression in the wild. Heredity 83: Biology 9: 792–799. 260–270. ———. 2015. Genetic rescues of small DI RIENZO, A., A. C. PETERSON,J.C.GARZA, inbred populations: meta-analysis reveals A. M. VALDES,M.SLATKIN, and N. B. large and consistent benefits of gene FREIMER. 1994. Mutational processes of flow. Molecular Ecology doi: http:// simple-sequence repeat loci in human 10.1111/mec.13139. populations. Proceedings of the National ———,C.J.A.BRADSHAW, and B. W. BROOK. Academy of Sciences USA 91: 2014. Genetics in conservation manage- 3166–3170. ment: revised recommendations for the ECKERT,C.G.,K.E.SAMIS,andS.C.LOUGHEED. 50/500 rules, red list criteria and pop- 2008. Genetic variation across species’ ulation viability analyses. Biological geographical ranges: the central-marginal Conservation 170: 56–63. hypothesis and beyond. Molecular Ecol- FRANZMANN, A.W., and P. D. ARNESON. 1973. ogy 17: 1170–1188. Moose Research Center Studies. Alaska ELLEGREN, H. 2004. Microsatellites: simple Department of Fish and Game Report, sequences with complex evolution. Nat- Soldotna, Alaska, USA. ure Reviews Genetics 5: 435–445. ———, and C. C. SCHWARTZ. 1985. Moose EVANNO, G., S. REGNAUT, and J. GOUDET. 2005. twinning rates: a possible population Detecting the number of clusters of indi- condition assessment. Journal of Wildlife viduals using the software Structure: a Management 49: 394–396. simulation study. Molecular Ecology ———, ———, and R. O. PETERSON. 1980. 14: 2611–2620. Moose calf mortality in summer on the EXCOFFIER, L., and H. E. L. LISCHER. 2010. Kenai Peninsula, Alaska. Journal of Arlequin suite ver 3.5: A new series of Wildlife Management 44: 764–768. programs to perform population genetics GAINES, M. S., J. E. DIFFENDORFER,R.H. analyses under Linux and Windows. TAMARIN, and T. S. WHITTAM. 1997. The

82 ALCES VOL. 51, 2015 WILSON ET AL. – POPULATION GENETIC STRUCTURE

effects of habitat fragmentation on the HUNDERTMARK, K. 2009. Reduced genetic genetic structure of small mammal popu- diversity in two introduced and isolated lations. Journal of Heredity 88: 294–304. moose populations in Alaska. Alces 45: GARCÍA-RAMOS, G., and M. KIRKPATRICK. 137–142. 1997. Genetic models of adaptation and KLEIN, D. R. 1970. Food selection by North gene flow in peripheral populations. Evo- American deer and their response to lution 51: 21–28. over-utilization of preferred plant spe- GARZA, J. C., and E. G. WILLIAMSON. 2001. cies. Pages 25–46 in A. Watson, editor. Detection of reduction in population Animal Populations in Relation to Their size using data from microsatellite loci. Food Sources. British Ecological Society Molecular Ecology 10: 305–318. Symposium 10. Blackwell, Oxford, GOUDET, J. 1995. FSTAT (version 1.2): a com- England. puter program to calculate F-statistics. LACY, R. C. 1987. Loss of genetic diversity Journal of Heredity 86: 485–486. from managed populations: Interacting HAMPE, A., and R. J. PETIT. 2005. Conserving effects of drift, mutation, immigration, biodiversity under climate change: the selection, and population subdivision. rear edge matters. Ecology Letters 8: Conservation Biology 1: 143–158. 461–467. LEBERG, P. 1993. Strategies for population HARTL,D.L.,andA.G.CLARK. 2007. Principle reintroduction: effects of genetic varia- of Population Genetics, 4th Edition. bility on population growth and size. Sinauer Associates, Sunderland, Massa- Conservation Biology 7: 194–199. chusetts, USA. LORANGER, A. J., T. N. BAILEY, and W. W. HEDRICK, P. W. 2005. A standardized genetic LARNED. 1991. Effects of forest succes- differentiation measure. Evolution 59: sion after fire in moose wintering habitats 1633–1638. on the Kenai Peninsula, Alaska. Alces HOFFMANN, A. A., and M. W. BLOWS. 1994. 27: 100–109. Species borders: ecological and evolu- LYNCH, C. B. 1977. Inbreeding effects upon tionary perspectives. Trends in Ecology animals derived from a wild population and Evolution 9: 223–227. of Mus musculus. Evolution 31: ———, and P. A. PARSONS. 1991. Evolution- 526–537. ary Genetics and Environmental Stress. MAINGUY, J., S. D. CÔTÉ, and D. W. COLTMAN. Oxford University Press, Oxford, 2009. Multilocus heterozygosity, paren- England. tal relatedness and individual fitness HOWES, B. J., and S. C. LOUGHEED. 2008. components in a wild mountain goat, Genetic diversity across the range of tem- Oreamnus americanus population. Mole- perate lizard. Journal of Biogeography cular Ecology 18: 2297–2306. 35: 1269–1278. MARUYAMA, T., and P. A. FUERST. 1985. Popu- HUBISZ, M. J., D. FALUSH,M.STEPHENS, and lation bottlenecks and non-equilibrium J. K. PRITCHARD. 2009. Inferring weak models in population genetics. II. Num- population structure with the assistance ber of alleles in a small population that of sample group information. Molecular was formed by a recent bottleneck. Ecology Resources 9: 1322–1332. Genetics 111: 675–689. HUGHES, A. L. 2010. Reduced microsatellite MEDRANO, J. F., E. AASEN, and L. SHARROW. heterozygosity in island endemics sup- 1990. DNA extraction from nucleated ports the role of long-term effective red blood cells. Biotechniques 8: 43. population size in avian microsatellite MEIRMANS, P. G. 2006. Using the AMOVA diversity. Genetica 138: 1271–1276. framework to estimate a standardized

83 POPULATION GENETIC STRUCTURE – WILSON ET AL. ALCES VOL. 51, 2015

genetic differentiation measure. Evolu- PIELOU, E. C. 1991. After the Ice Age: The tion 60: 2399–2402. Return of Life to Glaciated North MERILÄ, J., and P. CRNOKRAK. 2001. Compar- America. University of Chicago Press, ison of genetic differentiation at marker Chicago, Illinois, USA. loci and quantitative traits. Journal of PIRY, S., G. LUIKART, and J. M. CORNUET. 1999. Evolutionary Biology 14: 892–903. BOTTLENECK: a computer program for MILLER, M. J., E. BERMINGHAM,J.KLICKA, detecting recent reductions in the effec- P. E SCALANTE, and K. WINKER. 2010. Neo- tive population size using allele fre- tropical birds show a humped distribution quency data. Journal of Heredity 90: of within-population genetic diversity 502–503. along a latitudinal transect. Ecological PRITCHARD, J. K., M. STEPHENS, and P. DON- Letters 13: 576–586. NELLY. 2000. Inference of population MORTON, J. M., M. BOWER,E.BERG,D. structure using multilocus genotype MAGNESS, and T. ESKELIN. 2009. Long data. Genetics 155: 945–959. Term Ecological Monitoring Program QUATTRO, J. M., and R. C. VRIJENHOEK. 1989. on the Kenai National Wildlife Refuge, Fitness differences among remnant Alaska: an FIA adjunct inventory. Pages populations of the Sonoran topminnow, 1–17 in W. McWilliams, G. Moisen, Poeciliopsis occidentalis. Science 245: and R. Czapiewski, compilers. Proceed- 976–978. ings of the Forest Inventory and Analysis RALLS, K., J. BALLOU, and R. L. BROWNELL JR. (FIA) Symposium 2008. RMRS- P-56CD. 1983. Genetic diversity in California sea USDA Forest Service, Rocky Mountain otters: Theoretical considerations and Research Station, Fort Collins, Colorado, management implications. Biological USA. Conservation 25: 209–232. MUHS, D., T. A. AGER, and J. E. BEGET. 2001. RAMBAUT, A., and A. J. DRUMMOND. 2007. Vegetation and paleoclimate of the last Tracer v1.4. < http://beast.bio.ed.ac.uk/ interglacial period, central Alaska. Qua- Tracer> (accessed August 2013). ternary Science Reviews 20: 41–61. REED, D. H., and R. FRANKHAM. 2001. How OHTA, T., and M. KIMURA. 1973. A model of closely correlated are molecular and mutation appropriate to estimate the quantitative measures of genetic varia- number of electrophoretically detectable tion? A meta-analysis. Evolution 55: alleles in a finite population. Genetic 1095–1103. Research 22: 201–204. ———,and ———. 2003. Correlation OLDEMEYER, J. L., A. W. FRANZMANN,A.L. between fitness and genetic diversity. BRUNDAGE,P.D.ARNESON, and A. FLYNN. Conservation Biology 17: 230–237. 1977. Browse quality and the Kenai ROBINSON, S. J., L. P. WAITS, and I. D. MARTIN. moose population. Journal of Wildlife 2007. Evaluating population structure of Management 41: 533–542. black bears on the Kenai Peninsula using PARMESAN, C., and G. YOHE. 2003. A globally mitochondrial and nuclear DNA ana- coherent fingerprint of climate change lyses. Journal of Mammalogy 88: impacts across natural systems. Nature 1288–1299. 421: 37–42. RØED, K. H., and L. MIDTHJELL. 1998. Micro- PATTEN, M. A., and C. L. PRUETT. 2009. The satellites in reindeer, Rangifer tarandus, song sparrow, Melospiza melodia,asa and their use in other cervids. Molecular ring species: patterns of geographic var- Ecology 7: 1773–1776. iation, a revision of subspecies, and ROFFLER, G. H., L. G. ADAMS,S.L.TALBOT,G. implications for speciation. Systematics K. SAGE, and B. W. DALE. 2012. Range and Biodiversity 7: 33–62. overlap and individual movements

84 ALCES VOL. 51, 2015 WILSON ET AL. – POPULATION GENETIC STRUCTURE

during breeding season influence genetic SONSTHAGEN, S. A., S. L. TALBOT, and C. M. relationships of caribou herds in south- WHITE. 2004. Gene flow and genetic central Alaska. Journal of Mammalogy characterization of Northern Goshawks 95: 1318–1330. breeding in Utah. Condor 106: 826–836. ROOT, T. L., J. T. PRICE,K.R.HALL,S.H. SOULÉ, M. E. 1980. Thresholds for survival: SCHNEIDER,C.ROSENZWEIG, and J. A. maintaining fitness and evolutionary POUNDS. 2003. Fingerprints of global potential. Pages 151–169 in M. E. Soulé warming on wild animals and plants. and B. A. Wilcox, editors. Conservation Nature 421: 57–60. Biology: An Evolutionary-Ecological SCHLÖTTERER, C. 2000. Evolutionary Perspective. Sinauer Associates, Sunder- dynamics of microsatellite DNA. Chro- land, Massachusetts, USA. – mosoma 109: 365 371. TALLMON, D. A., A. KOYUK,G.H.LUIKART, SCHMIDT, J. I., K. J. HUNDERTMARK,R.T. and M. A. BEAUMONT. 2008. ONeSAMP: BOWYER, and K. G. MCCRACKEN. 2009. a program to estimate effective popula- Population structure and genetic diver- tion size using approximate Bayesian sity of moose in Alaska. Journal of computation. Molecular Ecology – Heredity 100: 170 180. Resources 8: 299–301. SCHREY, A. W., M. GRISPO,M.AWAD,M.B. TESTA, J. W., and G. P. ADAMS. 1989. Body COOK,E.D.MCCOY,H.R.MUSHINSKY, condition and adjustments to reproduc- T. ALBARYRAK,S.BENSCH,T.BURKE, tive effort in female moose (Alces alces). L. K. BUTLER,R.DOR,H.B.FOKIDIS, Journal of Mammalogy 79: 1345–1354. H. JENSEN,T.IMBOMA,M.M.KESSLER- TOMASIK, E., and J. A. COOK. 2005. Mitochon- RIOS,A.MARZAL,I.R.K.STEWART, drial phylogeography and conservation H. WESTERDAHL,D.F.WESTNEAT, genetics of wolverine (Gulo gulo)of P. Z EHTINDJIEV, and L. B. MARTIN.2011. northwestern North America. Journal of Broad-scale latitudinal patterns of genetic Mammalogy 86: 386–396. diversity among native European and VON HARDENBERG, A., B. BASSANO,M.FESTA- introduced house sparrow (Passer domesticus) populations. Molecular BIANCHET,G.LUIKART,P.LANFRANCHI, and Ecology 20: 1133–1143. D. COLTMAN. 2007. Age-dependent genetic effects on a secondary sexual trait SCHWARTZ, C. C., and A. W. FRANZMANN. 1989. Bears, wolves, moose and forest in male alpine ibex, Capra ibex. Molecu- – succession, some management consid- lar Ecology 16: 1969 1980. erations on the Kenai Peninsula, Alaska. VUCETICH, J. A., and T. A. WAITE. 2003. Spa- Alces 25: 1–10. tial patterns of demography and genetic ’ SCHWARTZ, M. K., L. S. MILLS,Y.ORTEGA,L. processes across the species range: null F. RUGGIERO, and F. W. ALLENDORF. hypotheses for landscape conservation 2003. Landscape location effects genetic genetics. Conservation Genetics 4: variation of Canada lynx (Lynx canaden- 639–645. sis). Molecular Ecology 12: 1807–1816. WILSON, G. A., and B. RANNALA. 2003. Baye- SILVA, A. D., J.-M. GAILLARD,N.G.YOCCOZ, sian inference of recent migration rates A. J. M. HEWISON,M.GALAN,T.COULSON, using multilocus genotypes. Genetics D. ALLAINE,L.VIAL,D.DELORME,G.VAN 163: 1177–1191. LAERE,F.KLEIN, and G. LUIKART. 2009. ———,C.STROBECK,L.WU, and J. W. Heterozygosity-fitness correlations COFFIN. 1997. Characterization of micro- revealed by neutral and candidate gene satellite loci in caribou Rangifer taran- markers in roe deer from a long-term dus, and their use in other artiodactyls. study. Evolution 63: 403–417. Molecular Ecology 6: 697–699.

85 POPULATION GENETIC STRUCTURE – WILSON ET AL. ALCES VOL. 51, 2015

WISELY, S. M., S. W. BUSKIRK,G.A.RUSSELL, YAMASHITA, T., and G. A. POLIS. 1995. A test K. B. AUBRY, and W. J. ZIELINKSI. 2004. of the central-marginal model using Genetic diversity and structure of the sand scorpion populations (Paruroctonus fisher (Martes pennanti) in a peninsular mesaensis, Vaejovidae). Journal of Ara- and peripheral metapopulation. Journal chnology 23: 60–64. of Mammaology 85: 640–648.