Genomic insights into the recent evolution of ecotypes: delineating at-risk caribou populations with high gene flow

Abstract

Background

Ecotypes hold unique behavioral and/or life history diversity. Identifying genetic divergence among ecotypes is critical for informing their designations and conservation status. However, detecting these differences can be challenging, particularly for ecotypes that are rooted in behavior, share habitats, and exhibit high gene flow. In Ontario and Quebec (Canada), two ecotypes of woodland caribou (Rangifer tarandus caribou) with overlapping distributions in winter are recognized: eastern migratory and boreal caribou are morphologically similar yet are distinguished by their calving strategy and conservation status. While eastern migratory caribou move between the forest and tundra during spring and summer and are proposed to be Endangered, boreal caribou remain in the forest year-round and are listed as Threatened under the federal Species at Risk Act.

Results

Genome-wide analysis revealed that caribou are more genetically similar within provinces, regardless of whether they are eastern migratory or boreal. While boreal caribou showed a continuous genetic structure between Ontario and Quebec, eastern migratory caribou were genetically distinct between the two provinces. Although there is low genetic differentiation between ecotypes in both provinces, association analysis identified genomic variants that significantly distinguish them. These variants were different in each province and consisted of single nucleotide polymorphisms (SNPs) and small insertions-deletions (indels), with none located in regions exhibiting strong suppressed recombination, such as chromosomal inversions associated with “supergenes”. Instead, genomic variants were primarily found in non-coding regions or those linked to gene regulation. In Ontario, high levels of admixture were detected between ecotypes.

Conclusions

Our findings revealed that ecotypes in Ontario and Quebec are genetically distinguishable. The different polygenic variants found in each province and the absence of “supergenes” suggest a recent and independent evolution of the eastern migratory ecotype in each province. In Ontario, the levels of interbreeding between ecotypes indicate that the area of geographical overlap is larger than what the current distribution reflects. These results add to our current knowledge of caribou genetic structure and movement in the region and can provide insights for future management efforts.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-026-13006-0.

Background

Ecotypes can be defined as phenotypes within a species that have evolved to leverage resources and increase their fitness in specific environments [1]. These forms, therefore, help maintain ecological differences and are important sources of adaptive genetic diversity [1]. Delimitating ecotypes, however, can be challenging, particularly if the ecotypes have a recent origin and/or exhibit high gene flow. During ecotype origin, there is an interaction between phenotypic plasticity and standing genetic variation [2], and the genomic differences can arise and be maintained by several molecular mechanisms from small polygenic regions, large linked regions, and/or epigenetic modifications [3]. In recent years, for example, chromosomal inversions have gained prevalence as a key mechanism for maintaining adaptive variation [4]. Also referred to as supergenes, inversions maintain genetic variation of multiple linked genes by suppressing recombination and have been reported to differentiate ecotypes, such as migratory vs. sedentary cod [5], lagoon vs. marine seahorse [6], southern vs. northern migratory willow warbler [7], and forest vs. prairie mice [8]. This prevalence shows the importance of genomic data to detect fine-scale genetic differentiation between ecotypes, and in the last few years, an increasing number of studies have used genomics to differentiate between ecotypes of species at risk and delineate conservation units [9–13].

Caribou (Rangifer tarandus), also known as reindeer in Europe, is a species with a wide circumpolar distribution that plays an important ecological and cultural role [14]. In Canada, populations have been classified into 12 Designatable Units (DUs) by the Committee on the Status of Endangered Wildlife in Canada (COSEWIC) based on their discreteness and evolutionary significance. This classification considered genetics, morphology, behavior, and distribution [15]. In Ontario and Quebec, two DUs with overlapping distributions can be found: eastern migratory and boreal. These ecotypes present both the North American (NAL) and Beringian–Eurasian (BEL) mitochondrial phylogenetic lineages [16], exhibit little genetic differentiation [17], and show no morphological differences [18]; however, they are distinct by their distribution, calving strategy, and conservation status. While eastern migratory caribou are found in Manitoba, Ontario, Quebec, and Labrador, boreal caribou have a larger distribution that extends from the Northwest Territories to Labrador and inhabits southern latitudes [14]. Regarding their behavior, eastern migratory, which is also referred to as “forest-tundra” ecotype, migrates to the tundra during the calving season to avoid predators and calve in large aggregations, while boreal, known as “forest-dwelling”, exhibits a sedentary behavior remaining in the forest year-round and generally disperses away from each other during calving. Most caribou ecotypes are experiencing population declines [19]. The boreal DU is listed as Threatened under the federal Species at Risk Act, and the eastern migratory DU is currently classified as Endangered under COSEWIC. Although important efforts have been made to understand the genetic relationship of these ecotypes in Ontario and Quebec [20–26], there are still gaps in our understanding of their movement, fine-scale connectivity, adaptive diversity, and interbreeding between these two ecotypes, particularly in their geographical regions of overlap.

The goal of this study is to provide insight into the population genomics of eastern migratory and boreal caribou in Ontario and Quebec. Using a large dataset of 331 whole genome sequences (109 at high coverage and 222 at low coverage) of caribou sampled across their distributions (Fig. 1a & Table S1), we aimed to (i) detect fine-scale genomic structure of eastern migratory and boreal caribou in Ontario and Quebec; (ii) identify any genomic regions associated with their different behavioural strategies; and (iii) estimate levels of admixture between the two ecotypes in Ontario. Results from this study will help improve our understanding and delineation of these ecotypes, which can later facilitate the development of management strategies for their conservation.

Fig. 1.

(a) Map depicting the distribution of eastern migratory (blue) and boreal (green) caribou and sampling locations of high coverage (black points) and low coverage (grey points) samples. PCA results of high coverage data using ~ 2 million SNPs of (b) all samples and (c) a subset of Ontario samples (inside dotted circle), which includes individuals in the ecotype overlapping zone

Methods

Samples and sampling location

This study included caribou samples from Manitoba, Ontario, and Quebec from across the distribution of eastern migratory and boreal caribou (Fig. 1a and Table S1). Samples from Manitoba were eastern migratory individuals from the Pen Island herd. In Ontario, eastern migratory sampling locations included Pen Island, Fort Severn, and Peawanuck. Locations in the overlapping region included Big Trout Lake, Attawapiskat, Missisa, and Ozhiski. Boreal samples were collected in Berens-Sydney, Pickle Lake, Red Lake, Churchill, Brightsand, Nipigon, Auden, Pagwachuan, Hearst, and Cochrane. Quebec eastern migratory samples included 10 individuals from the George River herd, the range of which extends into Labrador. Genomes from eight samples were available in the NCBI-SRA repository (PRJNA634908 and PRJNA984705). From Quebec, we also used data available in the NCBI-SRA repository of 15 eastern migratory and boreal caribou sequenced at 30x (PRJNA843030), however, the geographic coordinates of these samples were not available. Most of the Ontario samples, particularly those located in the region of geographical overlap, were obtained from GPS-collared animals. Thus, their ecotype assignment was based on telemetry data following the approach developed by Pond et al. [18], where eastern migratory individuals spent more than 50% of their calving-season locations in the Hudson Bay Lowlands, and boreal individuals spent more than 50% in the Ontario Shield. For more details about the telemetry assignment, see Fournier et al. Preprint.

DNA extraction and sequencing

For tissue, hair, and blood samples, whole genomic DNA was extracted using Qiagen’s DNeasy blood and tissue kit and following the manufacturer’s protocol. For fecal pellets, DNA was extracted from the mucosal coat using a modified Qiaqen protocol as described in Taylor et al. [27]. DNA was quantified using a Qubit fluorometer (Thermo Fisher Scientific, MA, USA), and its purity was measured with a Nanodrop (Nanodrop Technologies Inc., Wilmington, DE, USA). Only samples with > 20ng/ul and 260/280 score ~ 1.8 were selected for sequencing.

Library preparation and sequencing were performed at The Centre for Applied Genomics (TCAG) at the Hospital for Sick Children (Toronto, Ontario). Library preparation was done following the TruSeq PCR-free DNA Library Preparation Kit protocol, and sequencing was done in an Illumina NovaSeq S4 flowcell 150-bp paired-end.

Data analysis

Scripts used in the analyses can be found at the following URL: https://github.com/LisetteDelgado/Genomics-caribou.

Raw data to genotypes

For both high and low coverage sequencing data, the first step was to check raw sequence read quality using FastQC [28]. Low-quality ends and adapters were removed using Trimmomatic [29], and poly-G tails were removed using fastp [30]. After checking read quality for a second time with FastQC, reads were mapped to the reindeer reference genome (mRanTar1.h1.1) using Bowtie2 [31]. SAM files were converted to BAM files and sorted using samtools [32], and duplicates were removed using picard [33]. After removing duplicates, samtools was used to calculate depth.

Genomes with > 15x were considered high coverage. Using the software GATK [34], individual variant call format (VCF) was called using HaplotypeCaller function, then individual VCFs were merged using CombineGVCFs function, and finally the combined VCF was genotyped using the GenotypeGVCFs function. Insertions-deletions (indels) and low-quality reads (minQ < 20 and minDP 5) were then filtered out using VCFtools [35]. A second VCF, which filtered out missing data > 5%, was also used as input for some of the population genetic analyses described below.

Genomes with <5x were considered low coverage. High-coverage data were down-sampled using the function DownsampleSam from picard, to include them in the low-coverage dataset. On average, low-coverage genomes had a depth of 2.36x (depth ranges: 1.1x to 3x). After removing duplicates, instead of calling for genotypes (VCF), genotype likelihoods (GL) were calculated using the software ANGSD [36]. Briefly, global GLs were calculated to obtain a beagle format using the following parameters: -GL 1, -doGLF 2, -doMaf 1, -minMaf 0.05, -SNP-pval 1e-6. Depth parameters were selected using a conservative approach. The -minDepth was set to assume 1x per sample, and the -maxDepth was the average depth multiplied by the total number of samples. Quality control parameters were set to -minQ 20 -minMapQ 20 -uniqueOnly 1 -remove_bads 1 -only_proper_pairs 1 -trim 0 -C 50.

Population genetics analyses

Principal Component Analysis (PCA) was used to assess genetic structure. For high coverage data, plink [37] was used first to convert VCF files to bed files and then to undertake a PCA with the -pca function. For low coverage data, PCAs were obtained using the beagle file as input and using the software pcangsd [38].

Isolation by distance (IBD) for low coverage data was calculated from genetic distances between populations (F_ST) and geographic distances. With ANGSD, the allele frequencies were first calculated using the -doSAF function. Then, using realSFS program from the ANGSD software, populations’ pairwise F_ST were estimated. Mantel test was calculated using the mantel.rtest function from the ade4 package in R [39].

Admixture K 2 to 9 were tested for high-coverage data using the software Admixture [40]. For low-coverage data, the software NGSadmix [41] was used. Genome-wide association analysis (GWAS) was run using the -assoc function in plink for high-coverage data. A structure plot [42] was also built using 20 microsatellite loci from Fournier et al. (Preprint). Plots from the different analyses were generated in R, except for the Linkage Disequilibrium (LD) heatmaps, which were created using the software LDBlockShow [43].

Mitogenomes were acquired by mapping raw reads to the mitochondria reference genome (from mRanTar1.h1.1 reference genome). After, fasta files were aligned in Mega [44], haplotypes were detected in DnaSP [45], and a median-joining network was calculated and plotted in Network [46]. The mitogenome of a barren-ground caribou (Rangifer tarandus groenlandicus) was used as an outgroup.

Finally, the gene ontology (GO), specifically the genetic enrichment for biological processes, was obtained from Gene Ontology (https://geneontology.org/). The reference organism selected was Bos taurus, and the false discovery rate (FDR) threshold was set to 0.05.

Results

Ecotypes in Ontario and Quebec

The analysis of 109 high-coverage genomes of eastern migratory and boreal caribou from Ontario (including Pen Island herd from Manitoba) and Quebec detected 89,017,035 genetic variants, including indels. After filtering for indels and setting a maximum SNP missing data to 5%, 2,093,176 SNPs remained. PCA revealed that caribou are more genetically similar within each province than by their ecotype (Fig. 1b). Eastern migratory caribou from Ontario and Manitoba cluster closer to boreal caribou from Ontario than to eastern migratory caribou from Quebec. The Pen Island herd in Manitoba showed no genetic differentiation with eastern migratory caribou from Ontario; thus, hereafter, when referring to eastern migratory caribou from Ontario, it includes caribou from the Pen Island herd in Manitoba. Eastern migratory caribou from Ontario and Quebec were distinguished along the PC1 axis (13% of the variation), showing the highest differentiation. Boreal caribou showed a significant isolation by distance (IBD) pattern (Mantel r = 0.839, p = 3e^− 4, Fig. S1) across their distribution from Quebec to Berens (Ontario’s most western population), excluding animals sampled in the region of geographical overlap.

Ecotypes in Quebec clustered separately, although, due to the lack of coordinates and small sample size, further conclusions cannot be made. In Ontario, our dataset included a large sample size and individuals located in the region of geographical overlap. PCA roughly distinguished between the ecotypes in Ontario, with 6.81% of the variation explained by PC1 (Fig. 1c). However, some individuals assigned as boreal using telemetry data clustered closer to eastern migratory caribou. Boreal caribou sampled further away from the region of geographical overlap were not necessarily more genetically distinct. Along the PC2 axis, one individual from the Pen Island herd in Manitoba separated from the rest.

Ancestry

Focusing on populations in Ontario, admixture results from high-coverage genomes showed that K = 2 was the most likely number of ancestral components (Fig. S2). Assuming K = 2, the ancestral component of each individual generally matches the ecotypes assignment based on telemetry or location (Fig. 2). In most cases, eastern migratory individuals have a higher proportion (> 0.99) of K1, and boreal caribou have a higher proportion (> 0.99) of K2. However, for four individuals, ancestry did not match telemetry assignment; caribou with higher proportions of K1 (eastern migratory) were assigned as boreal using telemetry data. In six cases, individuals had an equal proportion of K1 and K2 (50/50). For those six individuals, telemetry data assigned them as boreal. Interestingly, half of the individuals with equal ancestry proportions were sampled outside and south of the region of geographical overlap (Fig. 2). Contrary to whole genome analysis, structure analysis with the same samples but using 20 microsatellite markers could not distinguish between the two ancestral components (Fig. S3).

Fig. 2.

Admixture results assuming K = 2 of high-coverage data of a subset of Ontario caribou. Contribution of eastern migratory ancestry is shown in blue and boreal ancestry in dark green. The asterisks indicate if the individual was found in the overlapping zone. Information regarding mismatches with telemetry findings and 50% contribution of each group is placed at the top

NGSadmix results from low-coverage data, which included a large sample size (i.e., 316 individuals) and boreal individuals from Quebec, revealed three ancestry groups (Fig. 3a, Table S2): (i) eastern migratory located mainly in the northernmost region of Ontario, (ii) boreal located in the central region of Ontario, and (iii) boreal located in the northeastern region of Ontario and in Quebec. Results also showed admixture from eastern migratory ancestry to the boreal caribou south of the current region of overlap. Caribou from Berens (the most western location in Ontario) showed a contribution of almost 50% from the eastern migratory ancestry. To the east, a strong contribution from the boreal Quebec cluster is also observed. Admixture from the X chromosome (i.e., chr 35) exhibited a more pronounced admixture pattern, with most individuals displaying a higher contribution from the third cluster (light green; Fig. 3b). The interpolation map facilitates visualization of the contributions of the three ancestry groups or clusters (Fig. 3c). These results demonstrate that admixture between eastern migratory and boreal caribou occurred in most of their distribution.

Fig. 3.

Admixture results assuming K = 3 of low-coverage data from (a) autosomal chromosomes and (b) chromosome X. (c) Interpolation mapping the admixture results from autosomal chromosomes, areas where clusters (K) have similar contributions are blank. Contribution of Ontario’s eastern migratory ancestry is shown in blue, Ontario’s boreal ancestry in dark green, and Quebec’s boreal ancestry in light green

Analysis of the mitochondrial genome (excluding the d-loop region) of 110 caribou (109 high coverage plus a barren-ground caribou as outgroup) identified 59 haplotypes and revealed the presence of two haplogroups. These haplogroups show no relationship with ecotypes, as eastern migratory and boreal individuals were present in both haplogroups. Individuals from both ecotypes even shared the same mitochondrial haplotype (i.e., haplotype 19) (Fig. S4). The median-joining network also suggests genetic homogenization between provinces, as no clear haplotype distinction between individuals by province can be observed.

Genomic regions of differentiation between ecotypes and provinces

GWAS analysis was performed on the complete genomic dataset that included indels. In Ontario, 61 caribou whose genome-wide ancestry matched telemetry assignments were selected. As observed in Fig. 4a, multiple variants across the genome had significant association with the ecotypes; 7827 variants were above the blue line (P > 10^− 5) and 350 variants above the red line (P > 10^− 8). Around half (51.7%) of the 350 top variants were in non-coding regions, including the top peaks in chromosomes 23 and 27 (Fig. 4a). Variants within coding regions that differentiate ecotypes included Olfactory receptor 7A17 (OR7A17) in chromosome 3, Kinesin-like protein (KIF13B) in chromosome 18, Synergin gamma (SYNRG) in chromosome 21, and Cohesion subunit SA (STAG2) in chromosome X. GO terms using genes in the top 350 variants found no significant enrichment, however, using the genes from the top 7827 variants (i.e., above blue line), significant biological process were found to be overrepresented. These processes included anatomical structure development, cellular response to stimulus, optic nerve morphogenesis, regulation of cardiac muscle contraction, response to potassium ion, among others (Table S3). The variants in the top peaks (i.e., in chr 23 and chr X) alone were not informative markers that can be used to distinguish between ecotypes (Fig. S5a and c), as shown in PCAs built using variants within each peak. Also, the top peaks were not located in regions that show strong linkage blocks (Fig. S5b and d).

Fig. 4.

Manhattan plot showing genome-wide association analysis (GWAS) results comparing high coverage eastern migratory vs. boreal ecotypes in (a) Ontario (41 eastern migratory vs. 20 boreal caribou) and (b) Quebec (15 eastern migratory vs. 10 boreal caribou). The red line indicates the genome-wide significance (1.63 × 10-8) and the blue line indicates the suggestive significance (3.25 × 10-7). Genes’ abbreviations are shown above each peak

For Quebec samples, GWAS comparing the ecotypes showed different peaks than in Ontario (Fig. 4). There were fewer variants significantly differentiating ecotypes in Quebec, only 36 variants above the red line (Fig. 4b). Three peaks were in non-coding regions, although one was located close (i.e., 9,419 bp apart) to an interleukin receptor gene (IL1RAPL1). The second peak in chromosome X was within a histone deacetylase gene (HDAC8), which has the function to repress transcription.

GWAS comparing eastern migratory caribou from Ontario versus Quebec show highly significant variants across the genome (Fig. S6). More than 66 thousand variants were above the red line, some of which were located within 2,735 genes, which suggests adaptive differences between these two eastern migratory populations.

Discussion

Identifying the genetic basis of ecotype differentiation can be key to defining ecotypes and their evolutionary significance, yet detecting genetic variation that distinguishes ecotypes can be challenging, particularly in ecotypes that exhibit little genetic differentiation, live in sympatry, and might be experiencing recent or rapid changes [1]. The genomic analysis of eastern migratory and boreal ecotypes in Ontario and Quebec revealed that caribou are more genetically similar to geographically close populations than to caribou from the same ecotype. Leveraging on a large genetic dataset (i.e., 109 high-coverage genomes and 222 low-coverage genomes) and telemetry information, our results revealed: (i) boreal populations in Ontario and Quebec exhibited an IBD pattern, while eastern migratory populations were genetically distinct in both provinces; (ii) genome-wide resolution was needed to distinguish ancestral components and these components were found to be associated with the ecotypes; (iii) significant genomic variants differentiating ecotypes were not “supergenes” and were different in each province, lastly (iv) higher levels of interbreeding and gene flow between ecotypes than expected were detected in Ontario. Our results suggest a complex, independent, and recent evolutionary history of these ecotypes associated with their movement during the calving season in Ontario and Quebec. We discuss these results in detail below.

Eastern migratory and boreal origin in Ontario and Quebec

Caribou colonized their current ranges in Ontario and Quebec as the Laurentide ice sheet retreated during the late Pleistocene, and are known to have originated from the introgression of two lineages: NAL and BEL [16, 47]. These two lineages are depicted in the network by the two mitochondrial haplogroups, with the NAL haplogroup being more diverse than the BEL haplogroup, which included the barren-ground individual (i.e., outgroup) (Fig. S4). Introgression of barren-ground, which has a migratory behavior, into eastern migratory populations has been previously reported [24, 47], and a previous study suggested that having a BEL haplotype is associated with a higher probability of having a migratory behavior [48]. Our results, however, show that the majority of eastern migratory individuals had the NAL haplotype. The lack of clustering based on provinces at the mitochondrial level suggests a recent evolution and that the differences observed between ecotypes in each province arose from the similar standing genetic variation.

Genomic regions associated with migratory behavior

Whole genome analysis found significant variants within genes associated with ecotypes. These variants were different in Ontario and Quebec (Fig. 4), and were also different from a previous study with central and northern mountain caribou ecotypes, which found 57 SNPs associated with migration behavior [48]. The lack of parallelism and the fact that most variants were in non-coding regions or regions related to gene expression suggest an independent and recent evolution of eastern migratory caribou in Ontario and Quebec. Environmental variations in these geographical regions may also help explain these results (Fig. 4), as it has been shown that the environment plays an important role in caribou [49] and reindeer [50] genetic differentiation.

In both provinces, GWAS showed small peaks in different chromosomes (Fig. 4), suggesting that the changes in calving strategies are not the product of one region with large effects but the result of small polygenic effects. However, a larger sample size can lead to strong statistical power. In Ontario, the number of samples with high coverage from boreal ancestry (southern locations) was limited (i.e., 20 individuals), and the low depth of the low-coverage dataset (i.e., ~2x) was not recommended for GWAS analysis. In Quebec, the number of samples from both ecotypes was also limited, with only 10 boreal caribou. Our GWAS findings of multiple loci with small effect are consistent with the literature; however, we cannot discard a loss of power given the small sample size [51].

In general, ancestry clusters correlated with ecotypes (Figs. 2 and 3); therefore, we can infer that the differences in calving strategy are not purely plastic responses but that there is potentially an early adaptive component. Despite peaks being found in non-coding or regulatory regions, changes in gene expression regulation are more prevalent drivers of local adaptation than changes in coding regions [52]. However, the current high gene flow observed (i.e., microsatellites - Fig. S3), raises the question of how these adaptive variants are maintained. Recently, multiple examples of chromosomal inversions associated with ecotypes can be found in the literature, where genetic architecture helps suppress recombination [4]. However, our top peaks are not located in regions with strong linkage, discarding the presence of inversions (Fig. S5). Although we cannot rule out the presence of small inversions or other structural variants outside the top peaks. Alternative explanations of how adaptive variation is maintained despite gene flow are epigenetic modifications that help reduce recombination, and adaptive variation that follows a source-sink dynamic [53]. Among these mechanisms, epigenetic modifications are the least studied, although interestingly, one of the peaks found was within the histone deacetylase gene, which encodes for an epigenetic enzyme that has a role in repressing gene expression [54]. The latter mechanism suggests that there are “source” populations where gene flow is limited and where adaptive variation is maintained. These “source” populations are potentially where the interpolation map shows the highest contribution of each genetic cluster (Fig. 3c), for example, for boreal in Ontario, would be central southern locations such as Brightsand and Auden (Fig. 1a).

Eastern migratory movement in Ontario

Admixture results showed some discrepancies with telemetry data. In two cases, individuals with a higher proportion of migratory ancestry (60–70%) show boreal or sedentary behavior, and individuals with ~ 50% of eastern migratory ancestry were also assigned as sedentary (Fig. 2). This could suggest that a higher proportion of migratory ancestry (e.g., > 70%) is needed to maintain a migratory behavior. Although telemetry findings in the region suggested that there is no switching between behaviors [18], partial migration has been reported [55]. Therefore, the conflicting results between ancestry and telemetry could be explained by individuals skipping migration for a year. Yet, the telemetry data used in this study is robust, as most of the collared individuals have + 2 years of data. In Alaska-Yukon caribou, differences between genetic and demographic connectivity have also been reported, and episodic contact during the breeding season rather than switching herds, has been hypothesized as the most likely mechanism [56].

Telemetry data in Ontario have suggested that female eastern migratory and boreal distributions do not overlap during the breeding season [18], but our results suggest high interbreeding. Our dataset is skewed to female caribou, with only 14.3% being males; thus, further research with males is recommended to understand the movement and interbreeding between eastern migratory and boreal caribou. Similar interpretation could not be performed for Quebec samples, given the limited sample size and lack of location data, but previous work in that region has shown a significant IBD pattern among boreal caribou [57], and that particularly males disperse more during the rut or breeding season [20].

Conclusions

This study demonstrated the importance of high-resolution genome-wide data in delineating and identifying potentially adaptive genetic differences between ecotypes that share habitats and have a recent evolutionary history. Our results revealed that caribou in Ontario and Quebec are genetically distinct, particularly the eastern migratory ecotype. The non-parallel polygenic variation that distinguishes ecotypes in each province suggests an independent ecotype evolution, and the fact that most variants were in non-coding regions or regions associated with gene regulation suggests a recent ecotype formation or divergence. Results in Ontario detected high levels of interbreeding between ecotypes, revealing a larger range of movement and suggesting that the current region of overlap might need to be reassessed. The differences found between caribou ecotypes in Ontario and Quebec imply that separate conservation and management may be required. Overall, our findings highlight that ecotype formation can occur rapidly, and even though genetic differences arise from the same standing genetic variation, repeated ecotypes can develop from different polygenic variations.

Authors' contributions

MLD assisted with sample selection for sequencing, analyzed and interpreted genomic data, and wrote the original draft of the manuscript. MM and PW contributed to the study design, supervised, and were major contributors to writing the manuscript. CF performed the interpolation analysis and provided support with ecotype behavior metadata. DM, RST, and KS assisted with the study design and sample selection. AR and JJ provided samples from Ontario and analyzed telemetry data of collared caribou. VT and JK provided samples from Manitoba. All authors read and approved the final manuscript.

Funding

NSERC, the Ontario Species at Risk Stewardship Program, and Genome Canada provided funding for this research.

Data availability

Raw data generated and analysed during the current study are available in the Sequence Read Archive (SRA) repository from the National Center for Biotechnology Information (NCBI); low-coverage reads are linked to the BioProject accession number: PRJNA1365123, and high-coverage reads to the BioProject accession number: PRJNA1365846. Bioinformatic scripts used for analyses are available on the following GitHub URL: https://github.com/LisetteDelgado/Genomics-caribou.

Declarations

Ethics approval and consent to participate

Invasive caribou samples from Ontario were collected during collaring. Captures were conducted by the Ontario Ministry of Natural Resources. Procedures were approved by the OMNR Wildlife Animal Care, and the permit numbers are 19 − 440, 20 − 440, 21 − 440, 22 − 440, 23 − 440, and 24 − 440. Invasive caribou samples from Manitoba were collected during caribou collaring conducted by Manitoba’s Department of Natural Resources and Indigenous Futures, under procedures approved by the Manitoba Wildlife Branch.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.