Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Identification of Novel Single Nucleotide Polymorphisms (SNPs) in Deer (Odocoileus spp.) Using the BovineSNP50 BeadChip

  • Gwilym D. Haynes,

    Affiliation Department of Biological Sciences, Behavioral and Molecular Ecology Research Group, University of Wisconsin – Milwaukee, Milwaukee, Wisconsin, United States of America

  • Emily K. Latch

    Affiliation Department of Biological Sciences, Behavioral and Molecular Ecology Research Group, University of Wisconsin – Milwaukee, Milwaukee, Wisconsin, United States of America

Identification of Novel Single Nucleotide Polymorphisms (SNPs) in Deer (Odocoileus spp.) Using the BovineSNP50 BeadChip

  • Gwilym D. Haynes, 
  • Emily K. Latch


Single nucleotide polymorphisms (SNPs) are growing in popularity as a genetic marker for investigating evolutionary processes. A panel of SNPs is often developed by comparing large quantities of DNA sequence data across multiple individuals to identify polymorphic sites. For non-model species, this is particularly difficult, as performing the necessary large-scale genomic sequencing often exceeds the resources available for the project. In this study, we trial the Bovine SNP50 BeadChip developed in cattle (Bos taurus) for identifying polymorphic SNPs in cervids Odocoileus hemionus (mule deer and black-tailed deer) and O. virginianus (white-tailed deer) in the Pacific Northwest. We found that 38.7% of loci could be genotyped, of which 5% (n = 1068) were polymorphic. Of these 1068 polymorphic SNPs, a mixture of putatively neutral loci (n = 878) and loci under selection (n = 190) were identified with the FST-outlier method. A range of population genetic analyses were implemented using these SNPs and a panel of 10 microsatellite loci. The three types of deer could readily be distinguished with both the SNP and microsatellite datasets. This study demonstrates that commercially developed SNP chips are a viable means of SNP discovery for non-model organisms, even when used between very distantly related species (the Bovidae and Cervidae families diverged some 25.1−30.1 million years before present).


Single nucleotide polymorphisms (SNPs) are increasingly becoming the marker of choice for investigating contemporary and evolutionary genetic processes (e.g. [1], ). SNPs have many advantages over more traditionally used allozymes, microsatellite loci and chain-termination (Sanger) sequencing of select loci. These include availability in high numbers, presence in coding and non-coding regions, low-scoring error rates, relative ease of calibration between different studies and conformation to simple models of mutation. Furthermore, SNPs can be genotyped using high-throughput protocols that allow thousands of loci to be scored simultaneously, even from low quality DNA samples [2], [5], [6]. In species with fully sequenced genomes (i.e., ‘model’ organisms), panels of SNP markers that cover the entire genome can be devised to allow marker-trait association studies of high statistical power and accuracy (e.g. [7], [8], [9]). SNPs are also useful in researching the genetics of non-model organisms, and can be used in place of or in tandem with microsatellite markers to investigate kinship [10], individual identification [11], parentage inference [12] and population structure [13]. In addition, a SNP panel including both selectively neutral loci and loci under selection could be beneficial in studies of non-model organisms, as neutral loci can be used to make inferences about long-term demographic processes (e.g., migration) whereas loci under selection can be used to differentiate recently diverged lineages or identify genomic regions involved in local adaptation, reproductive isolation or speciation [4], [14], [15].

Developing a panel of SNP markers can be a challenge when working with non-model organisms. While next-generation sequencing technologies have greatly reduced the cost of DNA sequencing [16], performing such sequencing on enough individuals to identify SNPs (with minimal bias) is still outside the resources of many projects. One means of SNP discovery that does not require extensive sequencing is to use commercially available SNP chips developed for a related, well-studied model species. SNP chips are microarrays specifically customized for genotyping known SNP loci, and allow thousands of such loci to be scored simultaneously for two alleles. Recently, SNP chips from agricultural species have been used to identify SNPs in closely related, non-model species. For example, Miller et al. [17] identified 868 SNPs in bighorn (Ovis canadensis) and thinhorn sheep (Ovis dalli) using the OvineSNP50 BeadChip developed for domestic sheep (Ovis aries). Similarly, Pertoldi et al. [18] used the BovineSNP50 BeadChip developed for cattle (Bos taurus) to genotype 2 209 polymorphic loci in European (Bison bonasus) and American bison (B. bison bison and B. bison athabascae). These studies confirm that cross-species application of commercial SNP chips can be a successful strategy for SNP discovery in non-model organisms. This strategy, however, has only been applied to SNP development in non-model species closely related to the focal species. Domestic sheep diverged from bighorn and thinhorn sheep approximately 3.1 million years ago (MYA) [19], while cattle and bison diverged 1.2−2.1 MYA [20]. The use of commercial SNP chips in non-model organisms therefore warrants further investigation regarding their utility in more divergent lineages.

In this study, the potential utility of commercial SNP chip technology for identification of SNPs in non-model organisms is tested between two lineages that diverged approximately 25.1−30.1 MYA, deer (family Cervidae) and cattle (family Bovidae) [21]. The Illumina BovineSNP50 BeadChip developed for commercial SNP genotyping of B. taurus is used to genotype DNA samples from a diverse species complex of deer indigenous to North America: mule deer and black-tailed deer (Odocoileus hemionus ssp.), and white-tailed deer (O. virginianus) [22]. A suite of novel SNPs is characterized, and putatively neutral and selected loci are identified. A range of population genetic analyses are implemented using these SNPs and a panel of 10 microsatellite loci to assess whether the newly identified SNPs behave in a predictable fashion.


Of the 54 609 SNPs on the chip, 21 131 (38.7%) were scored successfully in at least 90% of individuals, and 1068 of these loci were polymorphic. Minor allele frequency (MAF) is widely used to describe the genetic variability of two-allele SNPs, and refers to frequency of the least common SNP allele. MAF for each locus overall and within each deer lineage is detailed in Table S1. MAF varied across loci and between lineages. The majority of minor alleles were at low frequencies of 0.1 or less, and some loci that were polymorphic overall were monomorphic within a single lineage (Table 1; Table S). To minimize ascertainment bias, all polymorphic SNPs were included in downstream analyses, regardless of the level of genetic variability. The microsatellites were successfully genotyped for 98.6% of alleles, with 4−13 alleles detected at each locus.

The analysis in lositan identified 878 SNP loci as neutral, 116 as being under positive selection and 74 under balancing selection after adjustment for multiple testing (Table S1). Departures from HWE were non-significant in all analyses (Table 2). The standard deviation was high for all genetic diversity measures (Table 2), likely because of the small sample sizes analyzed. Expected heterozygosity (HE) and observed heterozygosity (HO) were generally lower for SNPs than for microsatellites, though this difference between marker types was only significant in mule deer (Table 2). FIS differed markedly between species and datasets but was also generally lower for SNPs than for microsatellites (Table 2). The overall P(ID) (Table 2) was extremely low for both the 1068 polymorphic SNPs (3.4×10−162) and the 878 neutral SNPs (3.0×10−123), attesting to the high discriminatory power of these markers. Although P(ID) was an order of magnitude higher for microsatellites (3.6×10−12; Table 2 ) than for SNPs, this value still indicates a very high discriminatory power for the microsatellites as it is well above the P(ID) value of at least 10-3–10-4 recommended for wildlife forensic applications [23].

Table 2. Hardy-Weinberg Equilibrium (HWE) p-values, expected heterozygosity (HE), observed heterozygosity (HO), and FIS for mule deer (MD), black-tailed deer (BTD) and white tailed deer (WTD) with associated p values.

All three deer lineages were distinguished from each other in all analyses and for all datasets. Fisher’s exact test in genepop returned significant (p-value <0.001) departures from panmixia in all pairwise comparisons. The analyses in structure all returned K = 3. For the microsatellites, the highest ΔK value was at K = 3, with the mule deer, black-tailed deer and white-tailed deer each partitioned into distinct clusters (Table 3). Both SNP datasets initially returned a highest ΔK value at K = 2, where mule deer and black-tailed deer were clustered together to the exclusion of white-tailed deer. As structure only identifies the upper most level of population structure [24], the analyses were rerun without the white-tailed deer to determine if additional substructure could be identified within the cluster containing mule deer and black-tailed deer. The highest ΔK in both these subsequent analyses was two (Table 4), with the mule deer and black-tailed deer partitioned into discrete genetic clusters. Finally, FCA readily separated each of the three lineages into distinct clusters. These clusters were completely discrete for the microsatellites (Figure 1A), while the SNPs placed the mule deer and black-tailed deer into partially overlapping but still discernible clusters (Figure 1B, 1C).

Figure 1. Factorial component analysis (FCA) of mule deer (MD), black-tailed deer (BTD) and white-tailed deer (WTD) estimated using (a) microsatellites, (b) all 1068 polymorphic SNPs, and (c) the 878 SNPs identified as selectively neutral.

Table 3. Analysis in STRUCTURE for all 28 deer using 10 microsatellites, all 1068 polymorphic SNPs and the 878 putatively neutral SNPs.

Table 4. Analysis in STRUCTURE using only mule deer and black-tailed deer for all 1068 polymorphic SNPs and 878 putatively neutral SNPs.

All datasets and all measures of genetic distance clearly identified mule deer and black-tailed deer as more closely related to one another than either was to white-tailed deer (Figure 1). This pattern is consistent with previous studies of morphological characters [25], nuclear DNA [26], [27] and the Y-chromosome [28]; although it should be noted that mitochondrial DNA studies have revealed a different pattern, with mule deer and white-tailed deer being most closely related [26], . FST was higher for SNPs than for microsatellites in two of the three comparisons (Figure 1), likely because FST has a tendency to be reduced by high levels of polymorphism [32][35]. D and Dm were far higher for microsatellites than for SNPs (Figure 1). D is an explicit measure of allele frequency differences between sample groups that makes no correction for high numbers of alleles. High mutation rates (and therefore large numbers of alleles) typical of microsatellites therefore lead to higher values of D relative to loci with low mutation rates and low numbers of alleles, such as SNPs [36]. Dm is similarly elevated increased by high levels of heterozygosity [37], and is likely elevated here by the higher HO values detected for microsatellites in mule deer and black-tailed deer than for SNPs (Table 2).


Of the 54 609 loci on the BovineSNP50 BeadChip, 21 131 (38.7%) SNPs were successfully genotyped in at least 90% of individuals, and 1068 (2.0% of the total; 5.1% of genotyped loci) were polymorphic in deer. In comparison, Pertoldi et al. [18] successfully genotyped a far greater proportion of loci (96.7–98.7%) and detected 4% of loci as polymorphic using the same SNP chip in bison; and Miller et al. [17] successfully genotyped over 90% of loci in closely related species of sheep using the OvineSNP50 BeadChip, yet found only 1.7% of sites to be polymorphic (868 out of a total of 49 034 loci). The lower rate of genotyping success in this study when compared with Pertoldi et al. [18] and Miller et al. [17] is expected, given the 25.1−30.1 million year divergence between Bovidae (B. taurus) and Cervidae (O. hemionus and O. virginianus) [21]. The level of polymorphism, however, is unexpectedly high and could result from historically high population sizes of mule deer, black-tailed deer and white-tailed deer in North America [24]. In contrast, the bison species analyzed by Pertoldi et al. [18] have undergone several severe population bottlenecks, while the wild sheep species investigated by Miller et al. [17] live in relatively small, isolated populations. The identification of 1068 novel, polymorphic SNPs in this study demonstrates that commercial SNP chip technology is a viable and potentially underutilized means of discovering SNP loci in non-model species, even when used between highly divergent lineages.

Both neutral loci and loci potentially under selection were detected in this study, including 878 neutrally evolving, 116 under the influence of positive selection, and 74 influenced by balancing selection (Table S1). A suite of loci that includes both neutral and selected loci will be useful for a variety of applications. Most population genetic analyses, for example, assume that the genetic markers employed are selectively neutral. Loci under positive selection, however, can be essential in distinguishing between recently diverged species and populations that are otherwise difficult to distinguish using neutral makers [14], [38]. Characterizing genomic regions under balancing selection could identify advantageous genes and alleles that move between populations, such as loci involved in disease resistance (e.g., [39]). Thus, a necessary first step in any genetic study is to accurately characterize suites of loci that match study objectives and ensure the application of appropriate analytical models and correct interpretation of results.

Population genetic inferences made with the SNPs identified here were consistent with current taxonomic nomenclature and with previous studies of nuclear [27] and Y-chromosome [28] DNA and morphological characters [25] that identified mule and black-tailed deer as closely related and white-tailed deer as a more divergent evolutionary lineage. All measures of genetic distance (FST, D and Dm) reported lower differentiation between mule deer and black-tailed deer than between white-tailed deer and either O. hemionus lineage (Figure 2). Consistent with the analyses of microsatellites performed here, the three lineages were clearly delineated using exact tests, assignment tests, and FCA using the dataset of all 1068 polymorphic SNPs or the 878 neutral SNPs. Extremely low P(ID) values both overall and within individual lineages suggests that these SNPs would be very useful for fine-scale population genetic analyses requiring unambiguous individual identification. In this study, we used only ‘pure’ representatives of each lineage (as identified by previous genetic analyses; [40]). Further characterization of these SNPs would be necessary to determine their power and accuracy for delineating lineages in areas of sympatry where individuals may be of mixed ancestry.

Figure 2. Genetic distance measures estimated between mule deer, black-tailed deer and white-tailed deer using 10 microsatellites (white), all 1068 polymorphic SNPs (dark grey) and 878 putatively neutral loci (pale grey).

(a) FST (with standard deviation), (b) Jost’s D (with standard error) and (c) Nei’s minimum distance, Dm.

The level of within-population inbreeding (FIS) differed markedly between datasets (Table 2) and warrants further explanation here. The FIS statistic ranges from −1 to 1, with negative values indicating an excess of heterozygosity and positive values indicating excess homozygosity relative to expectations under HWE. For each lineage, deer were sampled from disparate locations, and as such are expected to belong to different populations and to therefore return positive FIS values consistent with homozygote excess (Wahlund effect). In accordance with these expectations, positive FIS values were returned for all lineages for microsatellites (although FIS was not significantly different from zero in white-tailed deer) and for SNPs in black-tailed deer and white-tailed deer. In contrast, statistically significant negative FIS values were returned in mule deer when all 1068 SNPs or the 878 neutral SNPs were analyzed (Table 2). The unexpected heterozygote excess in the SNP data in the mule deer lineage could be caused by a high proportion of low-frequency alleles in mule deer which would in turn lead to an artificially high HO. Of the 429 loci that were polymorphic in mule deer, 54% (n = 232) had a minor allele frequency (MAF) less than 0.1 (Table 1). This was higher than the proportion of similarly low-frequency alleles found in black-tailed deer (46%; 200 of 434 polymorphic loci within the black-tailed deer lineage) and white-tailed deer, where the MAF could not be less than 0.125 on account of only 4 individuals being analyzed (if at a given locus only one of the four individuals is heterozygous, the MAF of that locus will be 0.125) (Table 1). Multilocus genotypes from additional individuals would be necessary to more fully evaluate potential mechanisms for the observed heterozygote excess in mule deer.

Any process of SNPs discovery carries some risk of ascertainment bias, where the overall pattern of genetic diversity is not accurately represented by the sampled SNPs. In general, small screening panel size, overly stringent SNP identification algorithms, and bias toward polymorphic loci in SNP selection can lead to inaccurate inferences of genetic diversity, population genetic structure, and phylogenetic relationships [5], . The small sample size of deer initially screened for SNPs in the present study will almost certainly have led to some polymorphic sites not being detected, in particular those sites harboring rare alleles. In addition, the screening of SNPs identified in B. taurus for use in O. hemionus and O. virginianus is likely biased in favor of conserved genomic regions that still retain polymorphisms ancestral to the divergence between Cervidae and Bovidae. Such loci may not be representative of the evolutionary changes that have since occurred within the Cervidae family. The selection of SNPs for the Bovine SNP50 BeadChip that are distributed in a roughly even fashion across the B. taurus genome, however, should minimize the effects of this bias. Downstream applications can avoid compounding ascertainment bias by randomly selecting a panel of SNPs for analysis, rather than using only SNPs that exceed a minimum, predefined level of polymorphism [5].

One of the most attractive incentives for using model species to identify SNPs in non-model species is the availability of annotations that link SNP variation to DNA sequences and ultimately to biological processes. Although no deer genomes have yet been fully sequenced and annotated, the genomic location of each SNP identified in this study can be mapped on various versions of the B. taurus genome (e.g., the Btau 4.2 assembly, compiled by the Bovine HapMap Consortium, or the UMD3.1 assembly, compiled by the Center for Bioinformatics and Computational Biology at the University of Maryland). The position of each SNP on both Btau4.0 and UMD3.1 is provided in Table S1. However, the level of divergence between our model and non-model species (25–30 MYA) may not permit accurate chromosomal locations to be determined for all identified SNPs. Multiple chromosome rearrangements have occurred in the Bovidae and Cervidae lineages since their divergence, which is especially evident in a change in karyotype from 2n = 70 in cervids O. virginianus and O. hemionus to 2n = 60 in the bovid B. taurus [44]. In spite of these large-scale rearrangements, alignment of deer DNA sequences to the B. taurus genome has been successful for next-generation sequences generated from O. virginianus [45], presumably owing to regional synteny. Still, caution is warranted when interpreting results obtained from alignments between such divergent lineages.

The SNPs characterized in this study would likely be useful in a variety of applications for an array of cervid species, given the high cross-species amplification success we observed. Neutral SNPs can be readily applied to more traditional population genetic analyses, such as characterizing population structure, quantifying genetic diversity and inferring migration rates. Loci under natural selection could be used to investigate genetic mechanisms underpinning natural selection and adaptation, or to differentiate recently diverged populations, species and ecotypes that are otherwise difficult to distinguish using neutral loci [46]. Such investigations are relevant not only for evolutionary research but also for conservation and management of mule deer, black-tailed deer and white-tailed deer. In addition to being important game species, the U.S. Fish and Wildlife Service lists the Cedros Island mule deer (O. h. cerrosensis), Florida Key white-tailed deer (O. v. calvium) and Columbian white white-tailed deer in western Oregon (O. v. leucurus) as ‘Endangered’ [47]. White-tailed deer are also threatened in Venezuela by overhunting and habitat loss [48]. Thorough delimitation of subpopulation boundaries, identification of locally adapted populations and characterization of genetic diversity patterns will therefore be highly useful in informing regional conservation and management strategies. These commercial SNP chips could even be applied to other cervids of conservation or management concern; for example, those listed as threatened on the IUCN Red List [49] (hog dear, Axis spp, revised to genus Hyelaphus in [50]; Père David’s deer, Elaphurus davidianus; Patagonian huemul, Hippocamelus bisulcus).

This study demonstrates the potential utility of commercially available SNP chip technology for identifying SNP loci in non-model organisms. As polymorphic SNPs were identified between lineages that diverged up to 30.1 MYA, SNP chips developed for model organisms can likely identify SNPs in a far wider range of organisms than previously realized. The porcine, ovine, equine and bovine SNP chips, for example, could be used to collectively to develop a panel of SNPs for wide range of highly divergent ungulates; while SNP chips developed for dogs (Canis lupus familiaris) could likely identify polymorphic SNPs in a wide range of Carnivora species that would otherwise require extensive DNA sequencing. The cross-species utilization of SNP chips is therefore an exciting avenue of future research.

Materials and Methods

Ethics Statement

Samples were collected by Department of Natural Resources staff in Washington and Oregon from hunter-harvested animals between 2003 and 2009. Ethics approval was not required or sought for this research, as the samples were hunter-harvested and thus not collected specifically for this study, and no additional observational or field data were collected.

Study Organism

Mule deer and black-tailed deer are both classified as O. hemionus. Morphological [51], [52] and genetic studies [26], [28], [53], [54], however, strongly support the separation of this species into two highly distinct lineages that diverged in allopatry during the last glacial maximum. Black-tailed deer include subspecies O. h. columbianus and sitkensis and are found throughout the Pacific Northwest, west of the Cascade Mountains and north to Alaska along the Pacific Coast. Mule deer include subspecies O. h. hemionus, fulginatus, californicus, inyoensis, eremicus, crooki, peninsulae, sheldoni, and cerrosensis, and are found east of the Cascade Mountains and throughout western and central North America, Canada, and Mexico. White-tailed deer (O. virginianus) are more widespread than O. hemionus, being found throughout northern South America, Central America, Mexico, central and eastern North America and in a number of isolated populations in western North America. White-tailed deer can be subdivided into as many as 38 subspecies [45], [55], [56]. All three types of deer within this species complex show extensive local adaptation and population structuring [53], [57], [58], yet all have a conserved karyotype of 2n = 70 chromosomes [44] and are capable of extensive hybridization and introgression in regions of sympatry [26], [28], [40], . Notably, all three lineages overlap within our study area in western Oregon, making this region a natural experiment for testing specific hypotheses about such evolutionary processes as hybridization, local adaptation, and reproductive isolation. However, for the purposes of this study, only ‘pure’ samples from each lineage were used (as identified in previous genetic analyses; [40]).

Sample Collection and DNA Genotyping

To evaluate the feasibility of cross-species SNP chip genotyping as a means of SNP discovery, tissue samples were collected from twelve mule deer, twelve black-tailed deer and four white-tailed deer in Washington and Oregon, USA (Figure 3) between 2003 and 2009. Previous genetic analyses identified these deer as ‘pure’ representatives of their respective lineages, i.e., no evidence inter-lineage ancestry [40]. Genomic DNA for each of the 28 deer sampled was genotyped at a commercial lab (Genetic Visions, Inc.) using an Illumina BovineSNP50 Genotyping BeadChip. In addition, 10 selectively neutral microsatellite loci (BM848, Odh_C, Odh_E, Odh_K, C273, Odh_G, Odh_P, Odh_O, RT24, and T40) were PCR-amplified and genotyped according to Latch et al. [67] in all individuals so that statistical inferences made with SNPs could be compared with microsatellites.

Figure 3. Map of sampling locations for mule deer (MD), black-tailed deer (BTD) and white-tailed deer (WTD).

Identification of Neutral Loci and Loci Under Selection

Genetic analyses of wild populations depend on accurately characterizing whether the genetic loci used are under selection. Theoretical models in population genetics typically assume that the markers employed are selectively neutral; including loci under selection can bias inferences about migration rates, genetic diversity, population genetic structure, and phylogenetic relationships. Loci should therefore be screened for signatures of selection prior to population genetic analyses, in order to ensure that appropriate analytical models are used and results are interpreted correctly [15], [43]. Genomic studies, in contrast, are primarily concerned with identifying genes or genomic regions involved in evolutionary processes and can hence benefit from specifically targeting genomic regions suspected to be under selection (e.g. [68]). To identify SNPs potentially under selection in the present study, the FST-outlier method [69] was implemented in the Bayesian program lositan [70]. lositan simulates the expected distribution of Wright’s inbreeding coefficient FST vs expected heterozygosity (HE) for a given set of genetic markers under the island model of migration [71]. Loci under positive selection are expected to show greater levels of interpopulation differentiation than neutral loci (i.e., higher FST/HE ratio), whereas loci under balancing selection are expected to show lower levels (i.e., lower FST/HE ratio) of differentiation [72]. lositan was run for 10 000 000 simulations, under the “neutral” mean FST and forced mean FST settings, with a two-tailed significance level of 0.05. The mule deer, black-tailed deer, and white tailed deer were designated as different ‘populations’ in the analysis. P-values were adjusted for multiple testing using the B-Y method of false discovery rate correction [73] in the R-project package multtest [74].

Statistical Analyses

The statistical properties of the newly identified SNPs were compared with the 10 neutral microsatellite loci to verify that the SNPs were behaving in a predictable fashion. A range of common statistical analyses were implemented using all 1068 polymorphic SNPs identified here (see Results), the 878 SNPs identified as selectively neutral (see Results) and the 10 microsatellite loci to characterize population genetic structure in mule deer, black-tailed deer, and white-tailed deer. Departures of genotype frequencies from expectations under Hardy-Weinberg equilibrium (HWE) were tested using Fisher’s exact test in genepop 4.1 [75]. Heterozygosities were estimated in arlequin [76], and FIS for each deer lineage was calculated in genetix [77]. The unbiased theoretical expected probability of identity P(ID) was calculated for each suite of loci over all deer and within the mule deer and black-tailed deer lineages [23], [78].

To determine if either suite of SNPs (1068 polymorphic loci or 878 neutral loci) could be used to distinguish between mule deer, black-tailed deer and white-tailed deer, significant differences in allele frequencies were assessed using Fisher’s exact test in genepop 4.1 [75]. Assignment tests were also performed in structure 2.3.3 [79], [80] under the Allele Frequencies Correlated Model and Admixture Model. The newly developed Sampling Locations as Priors Model was also used, as this model incorporates pre-defined sample group information (in this case, each individual was identified a priori as mule deer, black-tailed deer or white-tailed deer) to allow population structure to be detected at lower levels of divergence and with less data than earlier versions of structure [81]. Assignment tests were run for K = 1−6, with 50 000 burn-in steps and 500 000 iterations for each value of K. Tests were performed three times for each value of K, and the ΔK statistic of Evanno et al. [24] was used to determine the most likely value of K for each data set. Factorial correspondence analysis (FCA) was implemented in genetix 4.05.2 [77] in order to represent genetic relationships among individual deer graphically.

Three measures of genetic distance were calculated to further confirm that the newly identified SNPs exhibit patterns of genetic variation and structure in accordance with theoretical expectations. Weir and Cockerham’s [82] measure of FST was calculated in genetix [77], and the standard deviation was estimated using 10 000 permutations. The more recently developed Jost’s D [83] was estimated in genodive [84] and the standard error calculated against a background of 10 000 permutations. Nei’s minimum genetic distance, Dm [37], was estimated in populations 1.2.31 [85]. FST is one of the most commonly used measures of genetic differentiation and is used in lositan to detect loci under selection, despite being strongly affected by high levels of polymorphism [32][35]. D provides an unbiased quantification of differences in allele frequencies between populations without being affected by levels of genetic diversity and heterozygosity the way FST and its analogues are [83]. Dm performs well in recently diverged lineages and when mutation rate is low [37], and is therefore well suited for SNP data (low mutation rates and numbers of alleles [86]) and the study system (recently diverged lineages [53]).

Supporting Information

Table S1.

Genome location, outlier-analysis in LOSITAN and Minor Allele Frequency (MAF) data of the 1068 polymorphic SNPs identified in O. hemionus and O. virginianus. Only SNPs that were genotyped in at least 90% of individuals were included in the analysis. The chromosomal position of each SNP on the Bos taurus genome assemblies UDM3.0 and BTAU4.0 is included. N/A values in the UMD3.0 assembly indicate that SNPs that are not mapped to this genome assembly. Zero values on the BTAU4.0 assembly are indicative of SNPs that could not be mapped to this assembly.



We would like to thank C. Michael Cowan from Genetic Visions Inc. (Middleton, Wisconsin) for performing SNP genotyping and for assistance with data formatting. We also thank Tiago Antao for providing assistance in using lositan, and Jim Heffelfinger for assistance with sample collection. Colleagues in the Latch lab, Elizabeth Kierepka, Ona Alminas and Rachael Toldness, provided helpful comments on an earlier version of the manuscript.

Author Contributions

Conceived and designed the experiments: EKL GDH. Performed the experiments: EKL GDH. Analyzed the data: EKL GDH. Contributed reagents/materials/analysis tools: EKL GDH. Wrote the paper: EKL GDH.


  1. 1. Seeb JE, Carvalho G, Hauser L, Naish K, Roberts S, et al. (2011) Single-nucleotide polymorphism (SNP) discovery and application of SNP genotyping in nonmodel organisms. Molecular Ecology Resources 11: Suppl. 11–8.
  2. 2. Garvin MR, Saitoh K, Gharrett AJ (2010) Application of single nucleotide polymorphisms to non-model species: a technical review. Molecular Ecology Resources 10: 915–934.
  3. 3. Slate J, Santure AW, Feulner PGD, Brown EA, Ball AD, et al. (2010) Genome mapping in intensively studied wild vertebrate populations. Trends in Genetics 26: 275–284.
  4. 4. Allendorf FW, Hohenlohe PA, Luikart G (2010) Genomics and the future of conservation genetics. Nature Reviews Genetics 11: 697–709.
  5. 5. Helyar SJ, Hemmer-Hansen J, Bekkevold D, Taylor MI, Ogden R, et al. (2011) Application of SNPs for population genetics of nonmodel organisms: new opportunities and challenges. Molecular Ecology Resources 11: 123–136.
  6. 6. Morin PA, McCarthy M (2007) Highly accurate SNP genotyping from historical and low-quality samples. Molecular Ecology Notes 7: 937–946.
  7. 7. Brooks SA, Gabreski N, Miller D, Brisbin A, Brown HE, et al. (2010) Whole-genome SNP association in the horse: identification of a deletion in myosin Va responsible for lavender foal syndrome. PLoS Genetics 6: e1000909.
  8. 8. Kolbehdari D, Wang Z, Grant JR, Murdoch B, Prasad A, et al. (2008) A whole-genome scan to map quantitative trait loci for conformation and functional traits in Canadian Holstein bulls. Journal of Dairy Science 91: 2844–2856.
  9. 9. Davoli R, Fontanesi L, Cagnazzo M, Scotti E, Buttazzoni L, et al. (2003) Identification of SNPs, mapping and analysis of allele frequencies in two candidate genes for meat production traits: the porcine myosin heavy chain 2B (MYH4) and the skeletal muscle myosin regulatory light chain 2 (HUMMLC2B). Animal Genetics 34: 221–225.
  10. 10. Krawczak M (1999) Informativity assessment for biallelic single nucleotide polymorphisms. Electrophoresis 20: 1676–1681.
  11. 11. Chakraborty R, Stivers DN, Su B, Zhong Y, Budowle B (1999) The utility of short tandem repeat loci beyond human identification: Implications for development of new DNA typing systems. Electrophoresis 20: 1682–1696.
  12. 12. Anderson EC, Garza JC (2006) The power of single-nucleotide polymorphisms for large-scale parentage inference. Genetics 172: 2567–2582.
  13. 13. Morin PA, Martien KK, Taylor BL (2009) Assessing statistical power of SNPs for population structure and conservation studies. Molecular Ecology Resources 9: 66–73.
  14. 14. Via S, West J (2008) The genetic mosaic suggests a new role for hitchhiking in ecological speciation. Molecular Ecology 17: 4334–4345.
  15. 15. Luikart G, England PR, Tallmon D, Jordan S, Taberlet P (2003) The power and promise of population genomics: from genotyping to genome typing. Nature Reviews Genetics 4: 981–994.
  16. 16. Glenn TC (2011) Field guide to next-generation DNA sequencers. Molecular Ecology Resources 11: 759–769.
  17. 17. Miller J, Poissant J, Kijas J, Coltman D (2011) A genome-wide set of SNPs detects population substructure and long range disequilibrium in wild sheep. Molecular Ecology Resources 11: 314–322.
  18. 18. Pertoldi C, Wójcik JM, Tokarska M, Kawałko A, Kristensen TN, et al. (2010) Genome variability in European and American bison detected using BovineSNP50 BeadChip. Conservation Genetics 11: 627–634.
  19. 19. Bunch TD, Wu C, Zhang YP, Wang S (2006) Phylogenetic analysis of snow sheep (Ovis nivicola) and closely related taxa. Journal of Heredity 97: 21–30.
  20. 20. MacEachern S, McEwan J, Goddard M (2009) Phylogenetic reconstruction and the identification of ancient polymorphism in the Bovini tribe (Bovidae, Bovinae). BMC Genomics 10: 177.
  21. 21. Hassanin A, Douzery EJP (2003) Molecular and morphological phylogenies of Ruminantia and the alternative position of the Moschidae. Systematic Biology 52: 206–228.
  22. 22. Gilbert C, Ropiquet A, Hassanin A (2006) Mitochondrial and nuclear phylogenies of Cervidae (Mammalia, Ruminantia): Systematics, morphology, and biogeography. Molecular Phylogenetics and Evolution 40: 101–117.
  23. 23. Waits LP, Luikart G, Taberlet P (2001) Estimating the probability of identity among genotypes in natural populations: cautions and guidelines. Molecular Ecology 10: 249–256.
  24. 24. Evanno G, Regnaut S, Goudet J (2005) Detecting the number of clusters of individuals using the sortware STRUCTURE: a simulation study. Molecular Ecology 14: 2611–2620.
  25. 25. Mackie RJ (1981) Interspecific relationships. In: Wallmo OC, editor. Mule and black-tailed deer of North America. Lincon, NE: University of Nebraska Press. pp. 487–507.
  26. 26. Cronin MA (1991) Mitochondrial and nuclear genetic relationships of deer (Odocoileus spp.) in western North America. Canadian Journal of Zoology 69: 1270–1279.
  27. 27. Gavin TA, May B (1988) Taxonomic status and genetic purity of Columbian white-tailed deer. Journal of Wildlife Management 52: 1–10.
  28. 28. Cathey JC, Bickham JW, Patton JC (1998) Introgressive hybridization and nonconcordant evolutionary history of maternal and paternal lineages in North American deer. Evolution 52: 1224–1229.
  29. 29. Carr SM, Ballinger SW, Derr JN, Blankenship LH, Bickham JW (1986) Mitochondrial DNA analysis of hybridization between sympatric white-tailed deer and mule deer in west Texas. Proceedings of the National Academy of Science of the United States of America 48: 9576–9580.
  30. 30. Cronin MA (1991) Mitochondrial-DNA phylogeny of deer (Cervidae). Journal of Mammalogy 72: 553–566.
  31. 31. Cronin MA, Vyse ER, Cameron DG (1988) Genetic relationships between mule deer and white-tailed deer in Montana. Journal of Wildlife Management 52: 320–328.
  32. 32. Wright S (1978) Evolution and the Genetics of Populations, Vol. IV. Variability Within and Among Natural Populations. Chicago: University of Chicago Press.
  33. 33. Charlesworth B (1998) Measures of divergence between populations and the effect of forces that reduce variability. Molecular Biology and Evolution 15: 538–543.
  34. 34. Nagylaki T (1998) Fixation indicies in subdivided populations. Genetics 148: 1325–1332.
  35. 35. Hedrick P (1999) Highly variable loci and their interpretation in evolution and conservation. Evolution 53: 313–318.
  36. 36. Whitlock MC (2011) G’ST and D do not replace FST. Molecular Ecology 20: 1083–1091.
  37. 37. Nei M (1987) Molecular Evolutionary Genetics. New York: Columbia University Press.
  38. 38. Nosil P, Schluter D (2011) The genes underlying the process of speciation. Trends in Ecology & Evolution 26: 160–167.
  39. 39. Xu TJ, Sun YN, Wang RX (2011) Allelic polymorphism, gene duplication and balancing selection of the MHC class II DAB gene of Cynoglossus semilaevis (Cynoglossidae). Genetics and Molecular Research 10: 53–64.
  40. 40. Latch EK, Kierepka EM, Heffelfinger JR, Rhodes OE (2011) Hybrid swarm between divergent lineages of mule deer (Odocoileus hemionus). Molecular Ecology 20: 5265–5279.
  41. 41. Clark AG, Hubisz MJ, Bustamante CD, Williamson SH, Nielsen R (2005) Ascertainment bias in studies of human genome-wide polymorphism. Genome Research 15: 1496–1502.
  42. 42. Albrechtsen A, Nielsen FC, Nielsen R (2010) Ascertainment biases in SNP Chips Affect Measures of Population Divergence. Molecular Biology and Evolution 27: 2534–2547.
  43. 43. Smith CT, Antonovich A, Templin WD, Elfstrom CM (2007) Impacts of marker class bias relative to locus-specific variability on population inferences in Chinook salmon: a comparison of single-nucleotide polymorphisms with short tandem repeats and allozymes. Transactions of the American Fisheries Society 136: 1674–1687.
  44. 44. Gallagher DS, Derr JN, Womack JE (1994) Chromosome conservation among the advanced pecorans and determination of the primitive bovid karyotype. Journal of Heredity 85: 204–210.
  45. 45. Seabury CM, Bhattarai EK, Taylor JF, Viswanathan GG, Cooper SM, et al. (2011) Genome-wide polymorphism and comparative analyses in the white-tailed deer (Odocoileus virginianus): a model for conservation genomics. Plos One 6: e15811.
  46. 46. Via S (2009) Natural selection in action during speciation. Proceedings of the National Academy of Sciences of the United States of America 106: 9939–9946.
  47. 47. U.S. Fish and Wildlife Service (2011) Endangered Species Database (, searched using the term ‘Odocoileus’ on December 6th 2011.
  48. 48. Moscarella RA, Aguilera M, Escalante AA (2003) Phylogeography, population structure, and implications for conservation of white-tailed deer (Odocoileus virginianus) in Venezuela. Journal of Mammalogy 84: 1300–1315.
  49. 49. IUCN (2011) The IUCN Red List of Threatened Species. Version 2011.2. Downloaded on 10 November 2011.
  50. 50. Groves C (2006) The genus Cervus in eastern Eurasia. European Journal of Wildlife Research 52: 14–22.
  51. 51. Taylor WP (1956) The Deer of North America. Harrisburg, PA: Stackpole Books.
  52. 52. Wallmo O (1981) Mule and black-tailed deer of North America. In: Lincoln , editor. NE. University of Nebraska Press.
  53. 53. Latch EK, Heffelfinger JR, Fike JA, Rhodes OE (2009) Species-wide phylogeography of North American mule deer (Odocoileus hemionus): cryptic glacial refugia and postglacial recolonization. Molecular Ecology 18: 1730–1745.
  54. 54. Polziehn RO, Strobeck C (1998) Phylogeny of wapiti, red deer, sika deer, and other North American cervids as determined from mitochondrial DNA. Molecular Phylogenetics and Evolution 10: 249–258.
  55. 55. Wilson DE, Reeder DM (2005) Mammal Species of the World: A Taxonomic and Geographic Reference, third edition. Baltimore: John Hopkins University Press.
  56. 56. Baker RH (1984) White-tailed Deer: Ecology and Management; Halls LK, editor. Harrisburg: Stockpole Books.
  57. 57. DeYoung RW, Demarais S, Honeycutt RL, Rooney AP, Gonzales RA, et al. (2003) Genetic consequences of white-tailed deer (Odocoileus virginianus) restoration in Mississippi. Molecular Ecology 12: 3237–3252.
  58. 58. DeYoung RW, Demarais S, Honeycutt RL, Gonzales RA, Gee KL, et al. (2003) Evaluation of a DNA Microsatellite Panel Useful for Genetic Exclusion Studies in White-Tailed Deer. Wildlife Society Bulletin 31: 220–232.
  59. 59. Bradley RD, Bryant FC, Bradley LC, Haynie ML, Baker RJ, et al. (2003) Implications of hybridization between white-tailed deer and mule deer. The Southwestern Naturalist 48: 654–660.
  60. 60. Jackson HHT (1921) A hybrid deer of the F2 generation. Journal of Mammalogy 2: 140–143.
  61. 61. McClymont RA, Fenton M, Thompson JR (1982) Identification of cervid tissues and hybridization of serum albumin. Journal of Wildlife Management 46: 540–544.
  62. 62. Kay CE, Boe E (1992) Hybrids of white-tailed and mule deer in western Wyoming. Great Basin Naturalist 52: 290–292.
  63. 63. Carr SM, Hughes GA (1993) Direction of introgressive hybridization between species of North American deer (Odocoileus) as inferred from mitochondrial cytochrome-b sequences. Journal of Mammalogy 74: 331–342.
  64. 64. Hornbeck GE, Mahoney JM (2000) Introgressive hybridization of mule deer and white-tailed deer in southwestern Alberta. Wildlife Society Bulletin 28: 1012–1015.
  65. 65. Derr JN (1991) Genetic interactions between white-tailed and mule deer in the southwestern United States. Journal of Wildlife Management 55: 228–237.
  66. 66. Hughes GA, Carr SM (1993) Reciprocal hybridization between white-tailed deer (Odocoileus virginianus) and mule deer (O. hemionus) in western Canada: evidence from serum-albumin and mtDNA sequences. Canadian Journal of Zoology 71: 524–530.
  67. 67. Latch EK, Amann RP, Jacobson JP, Rhodes OE (2008) Competing hypotheses for the etiology of cryptorchidism in Sitka black-tailed deer: an evaluation of evolutionary alternatives. Animal Conservation 11: 234–246.
  68. 68. Bimova BV, Macholan M, Baird SJE, Munclinger P, Dufkova P, et al. (2011) Reinforcement selection acting on the European house mouse hybrid zone. Molecular Ecology 20: 2403–2424.
  69. 69. Beaumont MA, Nichols RA (1996) Evaluating loci for use in the genetic analysis of population structure. Proceedings of the Royal Society of London Series B Biological Sciences 263: 1619–1626.
  70. 70. Antao T, Lopes A, Lopes RJ, Beja-Pereira , Luikart G (2008) LOSITAN: A workbench to detect molecular adaptation based on an Fst-outlier method. BMC Bioinformatics 9: 323–327.
  71. 71. Wright S (1931) Evolution in Mendelian populations. Genetics 16: 97–159.
  72. 72. Nielsen R (2005) Molecular signatures of natural selection. Annual Review of Genetics 39: 197–218.
  73. 73. Benjamini Y, Yekutieli D (2001) The control of the false discovery rate in multiple hypothesis testing under dependency. Annals of Statistics 29: 1165–1188.
  74. 74. Pollard KS, Dudoit S, van der Laan LJ (2008) Multiple testing procedures: R multtest package and applications to genomics UC Berkeley Division of Biostatistics Working Paper Series. 1.21.1 ed.
  75. 75. Raymond M, Rousset F (1995) GENEPOP (version 1.2): population genetics software for exact tests and eumenicism. Journal of Heredity 86: 248–249.
  76. 76. Excoffier L, Laval G, Schneider S (2005) Arlequin ver. 3.0: An integrated software package for population genetics data analysis. Evolutionary Bioinformatics Online 1: 47–50.
  77. 77. Belkhir K, Borsam P, Chikhi L, Raufaste N, Bonhomme F (2004) Genetix, Logiciel Sous WindowsRM Pour la Génétique Des Populations. Laboratoire Génome, Population, Interactions CNRS URM 5000, Université de Montpellier II, Montpellier, France.
  78. 78. Paetkau D, Strobeck C (1994) Microsatellite analysis of genetic variation in black bear populations. Molecular Ecology 3: 489–495.
  79. 79. Pritchard JK, Stefens M, Donelly P (2000) Inference of population structure using multilocus genotype data. Genetics 155: 945–959.
  80. 80. Falush D, Stephens M, Pritchard JK (2003) Inference of Population Structure Using Multilocus Genotype Data: Linked Loci and Correlated Allele Frequencies Genetics 164: 1567–1587.
  81. 81. Hubisz MJ, Falush D, Stephens M, Pritchard JK (2009) Inferring weak population structure with the assistance of sample group information. Molecular Ecology Resources 9:
  82. 82. Weir BS, Cockerham CC (1984) Estimating F-statistics for the analysis of population structure. Evolution 38: 1358–1370.
  83. 83. Jost L (2008) GST and its relatives do not measure differentiation. Molecular Ecology 17: 4015–4026.
  84. 84. Meirmans PG, Van Tienderen PH (2004) GENOTYPE and GENODIVE: two programs for the analysis of genetic diversity of sexual organisms. Molecular Ecology Notes 4: 792–794.
  85. 85. Langella O (1999) Populations 1.2.30: a population genetic software. Roscoff, France. UPR9034: Available at Centre National de la Recherche Scientifique, Evolution, Génomes et Spéciation.
  86. 86. Brumfield RT, Beerli P, Nickerson DA, Edwards SV (2003) The utility of single nucleotide polymorphisms in inferences of population history. Trends in Ecology & Evolution 18: 249–256.