Persistent genetic connectivity in caribou may buffer against inbreeding effects

TL;DR

Caribou in Alaska show genetic unity despite spatial separation, which helps prevent inbreeding even after a population decline.

Contribution

The study reveals that spatial substructuring in caribou does not lead to genetic differentiation or inbreeding.

Findings

Genetic analysis shows no strong differentiation between east, central, and west caribou subgroups.

Low inbreeding levels are maintained due to the herd's previously large population size.

Caribou movement maintains genetic cohesion across a large geographic range.

Abstract

The Mulchatna Caribou Herd (MCH) in southwest Alaska has undergone significant demographic fluctuations, with a 94% decline over the past three decades, reducing the population from an estimated 200,000 to 13,000 individuals. This decline and concurrent range contraction, coupled with radio-telemetry and global positioning system (GPS)-collar studies, revealed indications of herd substructure. Females showed fidelity to one of two spatially distinct calving aggregations (designated as east and west), an attribute typically used to define individual herds in Alaska, and to three breeding areas within the greater MCH range (designated east, central, and west). To assess the genetic consequences of the population decline and apparent spatial substructuring, we analyzed genotyping-by-sequencing data from 121 adult female caribou. We found no strong genetic differentiation between east, central, and west subgroups, suggesting that the herd currently functions as a single genetic population despite observed spatial structuring during breeding and calving. In addition, we found no significant levels of inbreeding, likely due to the previously large population size. Overall, our results support the idea that the movement of the MCH is best characterized by a spatially structured but genetically cohesive population, in which connectivity between migratory and nonmigratory groups occurs over a large geographic range.

Figures

Point density of GPS locations from 87 adult female Mulchatna caribou (MCH) during the breeding season (September 25–October 10) from 2021 to 2023 in southwest Alaska.Kernel density estimation (500 m bandwidth, 30 m resolution) was used to identify breeding clusters, with higher densities represented in red (high), yellow (medium), and blue (low). Females were grouped into three primary breeding areas: west, central, and east, separated by major topographic features such as the Nuyakuk and Nushagak River and Wood River Mountains. Most of the females remained in the same breeding area across the years, though 12% switched breeding grounds at least once. The inset map shows the study area’s location in southwest Alaska.

Genetic structure of MCH.(A) Population structure of the Mulchatna caribou herd (MCH) inferred using Structure. Although K = 2 is shown to visualize potential admixture, extensive shared ancestry across all individuals indicates that the herd is best interpreted as a single genetic population. (B) Bar plot illustrating Admixture analysis results for the MCH at K = 2. The analysis indicates widespread shared ancestry among individuals; slightly stronger ancestry assignment relative to Structure reflects known methodological differences between the algorithms rather than biologically meaningful subdivision. Together, these results suggest significant genetic mixing among groups, irrespective of observed spatial movements. (C) Principal coordinates analysis (PCoA) of the MCH demonstrates high similarity among individuals and close population clustering, further supporting weak population structure.

Observed heterozygosity and inbreeding coefficients for MCH.(A) Observed heterozygosity and (B) inbreeding coefficient (FIS) for the west, central, and east MCH sample groups. Error bars represent 95% confidence intervals around these means. The observed heterozygosity values are similar across all groups. The FIS values indicate low levels of inbreeding within each group. The overlapping confidence intervals across MCH groups suggest consistent genetic diversity and minimal differentiation.

Full text

Introduction

Genetic diversity is a key component of population resilience and adaptability, providing the genetic materials needed to respond to environmental changes, disease pressures, and other survival challenges (Crandall et al., 2000; Moritz, 2002; DeWoody et al., 2021). Caribou (Rangifer tarandus) are a widely distributed species adapted to often harsh environmental conditions (Bernes et al., 2015; Pedersen et al., 2021). However, recent synchronous declines in abundance have been reported in various caribou populations across much of their range (Gunn et al., 2009; Vors & Boyce, 2009; Gunn, 2016; Plante et al., 2020). For example, the Mulchatna caribou herd in southwest Alaska seems to experience routine fluctuations in abundance (Skoog, 1968; Valkenburg et al., 2003). Historical records indicate that caribou numbers in the Bristol Bay region peaked in the 1860s before undergoing a 60-year decline, followed by a significant recovery and range expansion in the late 20th century (Van Lanen, Neufeld & Mcdevitt, 2018; Barten & Watine, 2020). In 1974, the first effort to accurately count the Mulchatna Caribou Herd (MCH) resulted in a minimum count of 13,079 animals (Taylor, 1981). By the mid-1990s, the herd had grown considerably before experiencing another substantial decline of approximately 94% in the late 1990s, dropping from an estimated 200,000 individuals to around 13,000 (Van Lanen, Neufeld & Mcdevitt, 2018; Barten & Watine, 2020). During this decline, the Mulchatna’s annual range contracted, and signs of population substructuring emerged as evident by divergent migrations, seasonal ranges, and distinct calving grounds by two primary groups of animals referred to as ‘east’ and ‘west’ subgroups displaying different demographic parameters (Demma, 2019; Barten & Watine, 2020; Demma & Sattler, 2022). Theory suggests that this repeated population size rise and fall may have removed genetic variants from the population, resulting in low levels of variability in the current population (DeWoody et al., 2021).

Defining a population can be a challenge in ecology, but typically populations are spatially discrete with occasional movement among populations; however, this model may not best represent how caribou populations function. Instead, Hinkes et al. (2005) suggested that as caribou herd sizes increase, previously isolated herds begin to overlap, and larger migratory herds may subsume smaller non-migratory ones, thereby promoting gene flow between groups (Roffler et al., 2012). Conversely, when herd sizes are smaller, isolation between herds increases (Hinkes et al., 2005). These dynamics support gene flow over long periods, facilitating retention of genetic variants even in smaller herds where drift may have been expected to result in the loss of these variants (Boulet et al., 2007; Roffler et al., 2012).

Previous genetic studies using microsatellite and mitochondrial markers have shown that the Mulchatna caribou herd is genetically homogeneous within the herd but clearly differentiated from other Alaskan caribou herds at broader spatial scales (Colson, Mager & Hundertmark, 2014; Mager et al., 2014). Colson, Mager & Hundertmark (2014) and Mager et al. (2014) suggest that natural landscape features and migratory behavior primarily drive large-scale genetic structure in Alaskan caribou. However, because these earlier analyses relied on a limited number of genetic markers and were based on samples collected before the recent population decline, when the herd was larger and exhibited greater spatial mixing, the extent to which fine-scale genetic structure exists within the contemporary Mulchatna herd remained unresolved. The use of single nucleotide polymorphism (SNP) data in the present study provides substantially higher resolution to evaluate population genetic structure and connectivity within this large, wide-ranging migratory system.

Rapid and severe demographic reductions, like those seen in the MCH, often lead to loss of genetic variation (England et al., 2003; Taylor et al., 2024; Stringer et al., 2024). This raises concerns about the MCH’s long-term viability, particularly because reduced genetic diversity can limit adaptability (Lande & Shannon, 1996; Bozzuto et al., 2019; Kardos et al., 2021; Taylor et al., 2024). Small population sizes also increase the risk of inbreeding, which occurs when closely related individuals mate (Reed et al., 2003). This leads to increased homozygosity and can result in inbreeding depression, a reduction in fitness characterized by lower survival and reproductive rates (Charlesworth & Charlesworth, 1987; Ralls, Ballou & Templeton, 1988; Charlesworth & Willis, 2009). Effective population size, which is typically a small fraction of census population size, determines the rate at which genetic diversity is lost through genetic drift and the rate at which inbreeding accumulates across generations (Wright, 1931; Lande, 1988; Reed et al., 2003). These factors can compound demographic challenges and counteract recovery efforts for declining populations.

Caribou play an important ecological role and represent a critical resource for the people in Alaska, supporting indigenous subsistence, recreational hunting, and wildlife viewing (Wolfe & Walker, 1987; Fall, 2016; Parlee, Sandlos & Natcher, 2018; Mallory & Boyce, 2018). Therefore, understanding the causes and potential consequences of caribou population declines is essential for establishing effective conservation strategies. In this study, we consider the potential genetic effects of the large fluctuations in population abundance in the MCH. To do so, we evaluated the contemporary structure of the MCH and quantified the divergence and standing genetic diversity within the herd. We then assessed how the current genetic structure of the MCH might impact long-term population stability to inform management strategies.

Methods

Study area

The MCH inhabits southwest Alaska, within a region bounded by the southern portion of the Alaska Range, the lower Kuskokwim River, and Bristol Bay (Fig. 1). The region is characterized by a mix of graminoid and shrub tundra, areas of boreal forest, streams and rivers lined with willow (Salix spp.) and alder (Alnus spp.), expanses of low-lying wetlands, while higher elevations feature lichen heath dominated substrates (Viereck et al., 1992; Clark et al., 2010; Van Lanen, Neufeld & Mcdevitt, 2018; Macander et al., 2022). To support the downstream analysis of genetic diversity and evaluate contemporary opportunities for gene flow in the MCH, we identified areas of caribou breeding using global positioning system (GPS) locations from 87 adult females between September 25 to October 10, 2021, through 2023. This period was selected as the breeding season because most calves in the MCH are born between May 13 and 28th (Demma, 2019), and the mean caribou gestation is approximately 230 days (McEwan & Whitehead, 1972; Bergerud, 1975; Ropstad, 2000). To identify areas of concentrated breeding, we used kernel density estimation in ArcGIS Pro 10.1 (ESRI), with a 500-meter bandwidth, 30-meter resolution, and used natural breaks in point densities to identify areas with low, medium, and high point density. These outputs show concentrated locations east of the Nushagak River (i.e., east breeding group), west of the Nushagak River and east of the Wood River Mountains (i.e., central breeding group), and west of the Wood River Mountains (i.e., west breeding group) during the breeding period (Fig. 1). The breeding concentrations are dynamic, commonly comprising many groups of various sizes. In the MCH, the main breeding concentrations in the east and west are typically separated by over 240 kilometers, although smaller concentrations (i.e., central breeding group) and other isolated groups are distributed in between (Fig. 1). Over the three years of GPS data, 12% of adult females switched breeding grounds at least once.

Point density of GPS locations from 87 adult female Mulchatna caribou (MCH) during the breeding season (September 25–October 10) from 2021 to 2023 in southwest Alaska.Kernel density estimation (500 m bandwidth, 30 m resolution) was used to identify breeding clusters, with higher densities represented in red (high), yellow (medium), and blue (low). Females were grouped into three primary breeding areas: west, central, and east, separated by major topographic features such as the Nuyakuk and Nushagak River and Wood River Mountains. Most of the females remained in the same breeding area across the years, though 12% switched breeding grounds at least once. The inset map shows the study area’s location in southwest Alaska.

DNA sequencing and analysis

This research was conducted in full compliance with institutional, national, and international guidelines on the ethical treatment of wildlife. The Division of Wildlife Conservation Institutional Animal Care and Use Committee approved all procedures (IACUC Protocol Nos. 0102-2020-61 and 0102-2021-51; approved 6 October 2020 and 2 September 2021, respectively). Research activities adhered to the guidelines of the American Society of Mammalogists for the use of wild mammals in research (Sikes, 2016), as well as to the US Animal Welfare Act (7 U.S.C. §§2131–2159) and associated regulations, and Alaska Department of Fish and Game policies, and permitting requirements (USDA, 1990). In accordance with the 3Rs principle (Replacement, Reduction, Refinement), we minimized animal handling, limited sample sizes to those necessary to achieve scientific objectives and used non-lethal sampling techniques. All measures were taken to reduce stress and disturbance, and taxon-specific guidelines were followed to ensure the humane treatment of caribou during capture, handling, and sampling (including ADF&G caribou capture and handling protocols). Between 2020 and 2022, a total of 121 whole blood samples (n = 48 east, n = 26 central, and n = 47 west) were collected from adult female Mulchatna caribou into EDTA-coated blood tubes and stored at −80 °C. Blood samples were sent to the University of Minnesota Genome Core for DNA extraction, genotyping-by-sequencing library preparation, and sequencing. Extractions were performed using the salting out method, a protocol optimized for whole blood samples (Aljanabi & Martinez, 1997). Genotyping-by-sequencing libraries were prepared at the University of Minnesota Genomics Core using the in-line GBS method based on the Poland protocol (Elshire et al., 2011). This protocol employed a PstI + MspI double digest for DNA fragmentation and barcoding. The in-line GBS method allowed for 376-plex multiplexing of samples after ligation. All 121 samples were sequenced on the Illumina platform using single-end reads (101 bp).

Sequencing data were assessed for quality and then used to assign genotypes to the sequenced individuals. First, raw fastq files were filtered for quality control using Fastp version 0.23.4 (Chen et al., 2018), using an existing caribou reference genome assembly (GenBank GCA_949782905.1). We aligned reads to the reference genome using the Burrows-Wheeler Aligner (BWA-MEM2 version 2.2.1) mem algorithm (Li & Durbin, 2010). Following alignment, the resulting bam files were sorted and indexed using Samtools version 1.19.2 (Li et al., 2009). Variant calling was conducted jointly across all samples using Freebayes version 1.3.7 (Garrison & Marth, 2012) with the default setting options. Using bcftools version 1.19, the VCF file generated by Freebayes was filtered to remove low-quality variants (QUAL <30), sites with low minor allele frequency (MAF <0.01), monomorphic sites (AC = 0), and missing data (F_MISSING >0). The variants were also pruned to retain only one SNP from each pair that was in linkage disequilibrium (LD). We filtered LD with 50,000 bp windows and a filtering of 0.5 r^2^ threshold (Cavedon et al., 2022b).

Using our genotype data, we investigated herd structure using three different programs. We first used the Bayesian clustering program Structure version 2.3.4 (Pritchard, Stephens & Donnelly, 2000) by converting the filter VCF file using PGDSpider version 2.1.1.5 (Lischer & Excoffier, 2012). We allowed Structure to reach approximate stationarity with a burn-in of 100,000 iterations followed by 250,000 sampling iterations for values of K = 1 − 4 (Schweizer et al., 2016; Cavedon et al., 2022a), where each run was performed 10 times. We determined the most parsimonious number of genetic clusters (K) suggested by Structure using Structure Harvester (Earl & Von Holdt, 2012). To ensure that we also effectively evaluated the occurrence of a single population, we also ran Admixture version 1.3 with a cross-validation flag for clusters of K = 1 − 4 (Alexander, Novembre & Lange, 2009) with 1,000 bootstraps and ensured population structure using a principal coordinate analysis (PCoA) using the package Adegenet (version 2.1.7) in R version 3.4.4 (R Core Team, 2023).

We used the R package Adegenet (Jombart, 2008) to calculate the pairwise FST among our sample groups defined geographically by breeding locations (west, central, and east). The FST values were computed based on the method of Weir & Cockerham (1984), which provided an estimate of population differentiation by partitioning genetic variance among and within these groups. Bootstrapping with 1,000 replicates was performed to derive 95% confidence intervals for FST estimates. For the Mulchatna groups, we estimated observed heterozygosity, expected heterozygosity, and the inbreeding coefficient FIS (as an index of inbreeding arising from non-random mating) using bootstrapped estimates (1,000 iterations) with 95% confidence intervals implemented in the R package dartRverse version 1.0.6 (Kardos et al., 2016; Gruber et al., 2018). Deviations from Hardy–Weinberg equilibrium were evaluated using a global chi-square test across all loci implemented in the R package pegas version 1.3 (Paradis, 2010). This test was applied to assess whether genotype frequencies conformed to expectations under random mating. Observed heterozygosity was additionally examined as a complementary proxy for genome-wide homozygosity, which is expected to correlate with runs of homozygosity (Kardos et al., 2016). To ensure that population structure and demographic inferences were based primarily on neutral genetic variation, we screened the SNP dataset for loci potentially under selection using pcadapt version 4.4.1 (Luu, Bazin & Blum, 2017), following the distinction between adaptive and neutral genetic diversity outlined by Holderegger, Kamm & Gugerli (2006).

Results

We successfully sequenced 121 females and generated between 0.9 million and 11 million reads per individual, with an average of 4.35 million reads per sample. Across all samples, about 560.7 million reads were produced, with an average quality score >30. Our analysis pipeline output initially included 3,683,544 sites across the 121 samples, with 22,901 SNPs after quality filtering. The dataset included 48 females from the east group, 26 from the central group, and 47 from the west group.

Our Structure Harvester, Admixture, and PCoA results suggest that the MCH exists as a single genetic population. Using Structure Harvester, the best-supported Structure model was K = 1; however, K = 2 was visualized to illustrate potential admixture among groups. Visual inspection of individual ancestry coefficients at K = 2 showed extensive admixture among all individuals, indicating that K = 1 provides the most biologically meaningful interpretation (Fig. 2A). Admixture analyses similarly supported a single population, with low genetic differentiation and widespread shared ancestry among all groups (Fig. 2B). Although Admixture displayed slightly stronger ancestry assignment than Structure, this difference is expected because the two methods use different statistical frameworks and optimization procedures to estimate ancestry proportions, rather than reflecting true population subdivision. Importantly, despite these visual differences, both methods converge on the same biological conclusion that the Mulchatna caribou herd currently forms a single genetic population. Principal coordinates analysis (PCoA) further supports this interpretation, with individuals from all groups forming a tightly clustered distribution and exhibiting very low genetic divergence (Fig. 2C).

Genetic structure of MCH.(A) Population structure of the Mulchatna caribou herd (MCH) inferred using Structure. Although K = 2 is shown to visualize potential admixture, extensive shared ancestry across all individuals indicates that the herd is best interpreted as a single genetic population. (B) Bar plot illustrating Admixture analysis results for the MCH at K = 2. The analysis indicates widespread shared ancestry among individuals; slightly stronger ancestry assignment relative to Structure reflects known methodological differences between the algorithms rather than biologically meaningful subdivision. Together, these results suggest significant genetic mixing among groups, irrespective of observed spatial movements. (C) Principal coordinates analysis (PCoA) of the MCH demonstrates high similarity among individuals and close population clustering, further supporting weak population structure.

We found low levels of genetic differentiation between groups: FST between west and east groups (FST = 0.0000, 95% CI [0.0000–0.0000]), west and central groups (FST = 0.0008, 95% CI [0.0006–0.0009]), and east and central groups (FST = 0.0004, 95% CI [0.0002–0.0005]) indicated no significant genetic differentiation. Using a global chi-square test across all loci to test deviations from Hardy–Weinberg equilibrium, our results showed no significant deviation from random mating (χ^2^ = 5.95, df = 1.09, p = 0.371), indicating that allele frequencies largely conform to Hardy–Weinberg expectations. All mean inbreeding coefficients were low, indicating little inbreeding attributable to non-random mating (west = 0.058, 95% CI [0.039–0.087]; central = 0.098, 95% CI [0.071–0.14]; east = 0.064, 95% CI [0.045–0.095]). Observed heterozygosity values were similar across all groups (west = 0.18, 95% CI [0.17–0.18]; central = 0.17, 95% CI [0.16–0.18]; east = 0.18, 95% CI [0.17–0.19]), as indicated by overlapping CIs (Fig. 3). Expected heterozygosity values were likewise similar across all groups (west = 0.18, 95% CI [0.17–0.18]; central = 0.18, 95% CI [0.18–0.19]; east = 0.18, 95% CI [0.18–0.19]), with overlapping CIs. We used pcadapt to evaluate whether any SNPs deviate strongly from neutral expectations. We performed this analysis on the full dataset and found that <1% of our SNPs showed evidence of being outliers (N = 221 loci). We interpreted this very small number of outliers as indicating that the overwhelming majority of loci behave as neutral markers and that our estimates of population structure, heterozygosity, and inbreeding are unlikely to be biased by selection. Because severe recent bottlenecks and strong selection can generate widespread departures from neutrality across the genome, the lack of a large outlier signal further supports the interpretation that our dataset primarily reflects neutral demographic processes within the Mulchatna herd.

Observed heterozygosity and inbreeding coefficients for MCH.(A) Observed heterozygosity and (B) inbreeding coefficient (FIS) for the west, central, and east MCH sample groups. Error bars represent 95% confidence intervals around these means. The observed heterozygosity values are similar across all groups. The FIS values indicate low levels of inbreeding within each group. The overlapping confidence intervals across MCH groups suggest consistent genetic diversity and minimal differentiation.

Discussion

Female philopatry to calving areas is characteristic of migratory caribou (Gunn & Miller, 1986; Valkenburg & Davis, 1986), and a behavioral pattern observed in the Mulchatna herd. If traditional definitions were applied today, the Mulchatna would be considered two herds (Skoog, 1968). However, our results support the idea that the east and west subgroups of the MCH may function as a spatially structured but genetically cohesive population, similar to adjacent caribou populations that exhibit high connectivity rather than operating as discrete populations (Hinkes et al., 2005; Roffler et al., 2012). Although year-round radio-telemetry monitoring and GPS collar data documented consistently different seasonal range use between the east and west subgroups (Barten & Watine, 2020; Demma & Sattler, 2022), we did not detect strong genetic differentiation (Fig. 1). Our results are consistent with previous genetic studies showing that the Mulchatna caribou herd is genetically homogeneous within the herd but distinct from neighboring herds at broader spatial scales (Colson, Mager & Hundertmark, 2014; Mager et al., 2014). Using genome-wide SNPs, we extend these earlier findings by confirming that even with higher genomic resolution, the MCH currently exhibits weak internal genetic structure and high connectivity among breeding groups.

The apparent disconnect between spatial and genetic structure is consistent with broader caribou population dynamics, where individuals alter their movement behaviors in response to environmental and demographic pressures (Hinkes et al., 2005). For the MCH, historical fluctuations in abundance have shaped these patterns, with greater spatial overlap among adjacent herds during periods of high abundance and more restricted overlap during declines (Van Lanen, Neufeld & Mcdevitt, 2018). In fact, records from the late 1980′s indicate that the Kilbuck herd, a distinct non-migratory herd resident to the southern Kuskokwim and Kilbuck mountains, experienced increased overlap with Mulchatna caribou as the Mulchanta’s numbers swelled and range expanded (Hinkes, 1989). Similar density-dependent movement responses have been observed in other caribou populations, where high-density conditions promote longer-distance dispersal, while lower densities reinforce site fidelity and subgroup cohesion (Skoog, 1968; Hemming, 1975; Hinkes et al., 2005; Solmundson et al., 2023). These findings highlight the importance of considering behavioral plasticity and ecological context when interpreting genetic structure in migratory caribou.

Observed and expected heterozygosity were highly similar across populations, and inbreeding coefficients were low and did not suggest concerning levels of inbreeding in the Mulchatna caribou herd (Fig. 3). Although the absolute values of heterozygosity were modest, this was expected for reduced-representation SNP datasets, which typically yield lower estimates of expected heterozygosity than microsatellite or whole-genome–based markers (Andrews et al., 2016; Lemopoulos et al., 2019). Because FIS reflects only the component of inbreeding arising from non-random mating, this measure does not account for inbreeding resulting from historical reductions in population size (Keller, 2002). To complement FIS and account for broader genomic patterns of inbreeding, we therefore relied on genome-wide observed heterozygosity and neutrality tests as indirect indicators of elevated homozygosity and historical inbreeding. Observed heterozygosity and outlier analyses showed no evidence that elevated homozygosity or loci under selection disproportionately influenced our results (Holderegger, Kamm & Gugerli, 2006). These findings indicate no strong genomic signal of elevated inbreeding within the Mulchatna caribou herd.

Previous studies have shown that caribou populations across their range often exhibit moderate to high levels of genetic diversity, shaped by their characteristic cycles of population expansion and contraction (Courtois, 2003; Yannic et al., 2014; Dedato et al., 2022; Taylor et al., 2024; Mager et al., 2024; Fournier et al., 2024). These cyclical fluctuations in abundance are a defining feature of migratory caribou herds (Gunn, 2003; Joly et al., 2011) and can influence patterns of genetic diversity and inbreeding over time. Importantly, mixing with nearby herds during periods of high abundance may help mitigate the genetic consequences of population declines, including inbreeding (Klütsch et al., 2016). While many declining populations experience inbreeding depression following isolation—manifested as reduced reproductive success and adaptability (Charlesworth & Charlesworth, 1987; Charlesworth & Willis, 2009; Hedrick & Garcia-Dorado, 2016; Poirier et al., 2019; Jong et al., 2020; Kardos et al., 2023; Clement et al., 2024), the historical cycles of expansion and contraction in caribou populations may offer some protection through occasional gene flow and the purging of deleterious alleles (Lacy, 1997; Reed et al., 2003; Hedrick & Garcia-Dorado, 2016; Kardos et al., 2021; Taylor et al., 2024). Within this context, continued connectivity within the Mulchatna caribou herd and across the broader region may provide an important buffer against the negative genetic effects of prolonged low abundance, as observed in other ungulate systems such as tule elk (Cervus canadensis nannodes; Sacks, Davis & Batter, 2024). Given the historical fluctuations in MCH population size and the continued exchange of females between breeding groups (e.g., 12% of females switched breeding grounds at least once in 3 years) even during the current period of low abundance, concern over inbreeding in this herd remains minor, provided that key seasonal habitats remain intact to support movement and gene flow between subgroups.

Although we did not detect multiple genetic populations, movement among groups may be uneven. For example, the central portion of the herd may already be functioning as a sedentary group, as has been seen in other herds (Hinkes et al., 2005). Even with increased movement frequency, genetic differentiation may be accumulating in this group, reinforcing the idea that movement dynamics within the MCH are not uniform and that some subgroups may be more vulnerable to isolation and drift than others. If the central group is largely resident, it may experience elevated genetic drift and associated concerns compared to more migratory segments of the herd. Similar patterns have been observed in other caribou populations, such as the George River Herd, where declining numbers have led to increased genetic differentiation between migratory and sedentary segments (Boulet et al., 2007). However, movement patterns and thus genetic divergence are likely more complex. Sedentary groups may still play a key role in maintaining gene flow among larger aggregations, particularly within a spatially structured yet genetically connected population, which may also apply to the MCH (Hinkes et al., 2005; Roffler et al., 2012).

Our findings also provide insight into how connectivity affects long-term genetic diversity. Across wildlife populations, connectivity has played a crucial role in maintaining genetic diversity and buffering against population decline (Lowe & Allendorf, 2010). While gene flow has limited major differentiation in the MCH so far, population genetics theory and empirical results suggest that reductions in movement could shift this balance (i.e., increased drift effects due to isolation), leading to reduced genetic variation (DeWoody et al., 2021). This is particularly concerning given that genetic diversity enhances a population’s capacity to adapt to environmental stressors such as disease outbreaks and climate variability (Hohenlohe, Funk & Rajora, 2021; Kardos et al., 2021). Maintaining connectivity is therefore important for the long-term stability of the MCH, given elevated brucellosis antibody prevalence during this study (Sattler, 2021), a disease that reduces population growth potential through abortions and reduced pregnancy rates (Cotterill et al., 2018; Yang et al., 2019), and ongoing climate-driven environmental change in Alaska (Macander et al., 2022). Many wildlife studies emphasize the role of connectivity in preventing inbreeding and supporting adaptive potential (Solmundson et al., 2023; Theissinger et al., 2023; Lamka & Willoughby, 2024). This is especially relevant for disease resilience, as higher genetic diversity increases the likelihood that individuals carry traits conferring resistance to pathogens (Allendorf, Luikart & Aitken, 2012; Kardos et al., 2021).

While our findings highlight the importance of connectivity in maintaining genetic health, one limitation of our study is the lack of information on male movement behavior and its role in mediating gene flow within the MCH. For example, in the adjacent Nelchina and Mentasta caribou herds in south-central Alaska, despite having discrete calving ranges, male movements between breeding groups during years of fall range overlap facilitated opportunities for gene flow (Roffler et al., 2012). During two of three breeding seasons, radiocollared mature Mentasta males moved 60–85 km to areas occupied exclusively by Nelchina females, facilitating gene flow despite spatial separation during calving. Given the movement distances documented for Mentasta bulls, it is conceivable that most, if not all, rutting groups within the MCH range could be connected via male-mediated gene flow, while spatial segregation is evident during calving.

Conclusions

Integrating genetic data into conservation planning is essential for preserving biodiversity and ensuring the long-term resilience of wildlife populations. Our findings suggest that the MCH currently functions as a single genetic population. Maintaining connectivity will be critical for safeguarding genetic diversity in the MCH over the long term. Regular genetic monitoring can serve as an early warning system for emerging genetic isolation, enabling wildlife managers to implement timely and targeted conservation actions. By incorporating genetic assessments into long-term management strategies, conservation efforts can more effectively support the stability and adaptive potential of caribou and other species.