Extensive admixture among captive sable antelope on South African wildlife farms calls for the integration of genetic data into IUCN assessments

Summary

The IUCN recognizes captive populations as valuable resources for conservation interventions. However, their genetic composition remains insufficiently understood, potentially jeopardizing wild populations. Sable antelope (Hippotragus niger) exemplifies this challenge. Although listed as Least Concern, recent assessments highlight declines in free-living populations, offset by increasing numbers under private ownership. Given the lack of comprehensive genetic data on captive populations, we analyzed genetic diversity and structure in 7,271 individuals from 358 South African wildlife farms using nuclear microsatellite markers. Captive sable represents only two of the five wild ancestries—Zambian and Southern—with widespread admixture and a bias toward Zambian ancestry. The underrepresentation of native Southern sable and the absence of other wild groups underscore the need for caution when using captive individuals in conservation actions. Overall, our results show that population numbers alone are insufficient to guide conservation decisions, reinforcing the importance of incorporating genetic information into species assessments.

Graphical abstract

Highlights

• •Captive sable in South Africa represents only a subset of wild ancestries
• •Native Southern sable is underrepresented in South African private wildlife farms
• •Widespread admixed ancestry challenges the conservation value of captive individuals
• •Genetic metrics are pivotal alongside population size in conservation assessments

Wildlife genetics; Nature conservation; Ecology; Zoology; Genetics

Introduction

Historic and current population declines shape demographic stochasticity and lead to changes in genetic diversity that may challenge population persistence.¹ Large mammals have been particularly affected by these declines. Human activities, including habitat destruction and overexploitation, are detrimental to species that require large areas and that are integrated into complex ecosystems.² Furthermore, such species usually have longer generation lengths and hence require longer times for population recovery.³

In response to the increasing extinction risks associated with population declines, the International Union for Conservation of Nature (IUCN) has developed guidelines on the use of ex situ management, which include captive breeding as a potential conservation tool under specific and well-defined circumstances.^4,5 Successful breeding programs require careful attention to the maintenance of genetic diversity^6,7,8 as prolonged captivity can lead to genetic changes, which ultimately affect fitness traits such as reproduction and survival.⁹ Without careful genetic management, adverse outcomes including inbreeding and outbreeding depression, founder effects, loss of genetic diversity, or adaptation to captivity may arise, diminishing the value of captive populations for conservation.^9,10 Hence, reintroduction efforts using captive-bred individuals must be based on a thorough understanding of their genetic history, both in wild and captive populations.^8,11 In addition, the introduction of hybridized or backcrossed intraspecific lineages into wild populations may lead to the erosion of locally adapted genetic variation.¹²

Captive breeding and management of large mammals have also been implemented for non-conservation purposes, such as wildlife farming, i.e., breeding non-domesticated animals for commercial use, including sales, hunting, and wildlife watching and nature tourism.¹³ Although primarily profit-driven, private wildlife farms may contribute to species conservation by increasing the population size of the farmed species.^14,15 Moreover, wildlife farms also preserve natural habitats, which can potentially lead to positive outcomes for the conservation of wild species in general.^16,17,18 However, in the absence of data on the genetic provenance of captive populations, concerns arise about their long-term effectiveness in conservation and, ultimately, whether the use of these individuals in conservation efforts risks disrupting natural populations.^19,20 Comparative studies assessing genetic diversity between wild populations in national parks or reserves and their farmed counterparts remain scarce. Those available essentially report no significant differences between captive and wild populations, as documented for African buffalo, red deer, or red-legged partridge.^21,22,23

Private wildlife farming is a key economic activity in southern Africa,^24,25 particularly in South Africa, where approximately 9,000 private wildlife farms span ∼205,000 km² or 17% of the country’s land area.²⁶ The private wildlife industry generates around 7.5 billion USD in annual revenues and employs a large number of people.^27,28 National monitoring and regulation remained limited until the late 1980s or early 1990s, when national and provincial legislative frameworks were introduced in South Africa, including the National Environmental Management Act (Act 107 of 1998) and the National Environmental Management: Biodiversity Act (Act 10 of 2004). The absence of such frameworks in earlier periods allowed transboundary translocations of wildlife, often across large geographical distances (see, e.g., Alpers et al.).²⁹ Although the current legislation is aimed at promoting sustainable management of private wildlife farming, there may be unintended consequences. For example, the mandatory fencing of wildlife farms in South Africa disrupts natural dispersal routes and increases the risk of genetic isolation and inbreeding.^17,30,31

Sable antelope (Hippotragus niger) is one of Africa’s most iconic large mammals. The species occurs in southern and eastern Africa and is strongly associated with miombo woodlands.^32,33 Comprehensive evolutionary studies of the species across its range provided support for the recognition of five genetically well-differentiated groups corresponding to distinct geographic populations: Angolan, Southern, Zambian, Eastern, and West Tanzanian.^34,35 At the global level, sable antelope is classified as least concern on the IUCN Red List of Threatened Species,³⁶ despite having faced population declines across its formerly extensive range. These declines have been especially pronounced in populations at the southernmost edge of the species’ distribution.^37,38,39,40 Accordingly, within South Africa, the Southern sable antelope was regionally classified as vulnerable in 2016,⁴¹ a status retained in the most recent regional reassessment.⁴² The global IUCN assessment also provides relevant context for the private wildlife-farming system, as it reports contrasting population trends across land-use contexts, noting that “population trends are more or less stable in protected areas (∼50%), increasing on private land (∼25%), and decreasing elsewhere (∼25%)” and further that “any further decrease in the free-living population may be compensated by the continued growth of its numbers on private farms and conservancies.”³⁶ In South Africa, this private-land component largely corresponds to an expanding private wildlife industry, primarily driven by the high economic value of sable antelope for trophy hunting and breeding.^24,43 Although it is recognized that several translocations of sable antelope took place into South Africa before the implementation of national legislation—at least from other southern African countries—there is limited evidence regarding the specific wild origins of animals currently present on wildlife farms. In addition, sable breeding programs often rely on the enhancement of certain phenotypic traits, including crossing different subspecies to select specific characters, which can be detrimental to biodiversity conservation⁴⁴; in sable antelope, horn length has been one such characteristic under selection.⁴⁵ Therefore, IUCN’s recognition of rising population trends on private land as an indicator of overall species stability does not address the essential issue of whether these privately managed populations maintain the genetic diversity present in the distinct wild populations found throughout Africa.

In this study, we investigate the origins and genetic background of captive sable antelope on private wildlife farms in South Africa, and how trading within South Africa impacts the genetic diversity of these privately managed populations. Specifically, we assess the presence and proportion of the different wild genetic groups on private wildlife farms, and compare their genetic signatures, including genetic diversity and levels of admixture, to establish whether these captive populations accurately reflect the diversity of their wild counterparts. For this, we amalgamated one of the most comprehensive datasets available for a wild species in captivity, comprising nearly 8,000 samples collected across South African private wildlife farms and genotyped using microsatellite markers. We sought to investigate whether captive populations are indeed representative of their wild counterparts, thereby providing an assessment of their potential role in future conservation efforts. More broadly, our work offers novel insights into the growing body of knowledge on captive populations and their representativeness as proxies for wild populations.

Results

The initial filtering of multilocus genotypes amplified across the 34 sable-specific microsatellite loci yielded a final dataset comprising 7,271 captive sable individuals (Figure 1), after discarding 143 samples that exceeded the missing data threshold of 15% and 574 samples identified as duplicates.

Figure 1.

Dataset organization steps

The final three datasets consist of (A) Captive Zambian, (B) Captive Southern, and (C) Selected Wildlife Farms. See also Figures S1 and S2; Table S2.

Each step shows the number of individuals, with the outer arc indicating the proportion of individuals at each step. The flowchart depicts the stepwise process of filtering and organizing the initial dataset of 7,271 captive individuals into three final subsets. Of the 7,988 samples that were genotyped, 7,271 individuals remained after removing 143 samples due to high missing data and 574 samples identified as duplicates. Starting with the full set, individuals with an admixture assignment (Qi) < 0.9 were excluded, reducing the sample size by 5,458 individuals. The remaining 1,634 captive Zambian individuals and 179 captive Southern individuals underwent pairwise relatedness analysis, leading to the removal of highly related individuals (pairwise relatedness ≥ 0.4). This resulted in 453 individuals in the Captive Zambian dataset and 109 individuals in the Captive Southern dataset. A further filtering step identified five private wildlife farms with ≥ 20 individuals of Zambian ancestry, forming the Selected Wildlife Farms dataset with 175 individuals.

Captive ancestry and relatedness

The initial clustering analysis revealed that captive sable on South African private wildlife farms primarily consisted of individuals from only two wild ancestry groups, namely, Zambian and Southern, using the five wild populations as reference (Figure 2A). In the second clustering step, where only the two wild ancestries found among the captive individuals were used as references (Figure S1), 1,634 non-admixed individuals were assigned to Zambian and 179 to Southern ancestries across 116 and 49 wildlife farms, respectively. Twenty-six wildlife farms housed both Zambian and Southern non-admixed sables. This analysis further indicated that 75% (n = 5,458) of the captive individuals had admixed ancestry between Zambian and Southern groups (Qi < 0.9, Figure S1), and were present on 98% of the 358 wildlife farms sampled. The proportions of Zambian and Southern ancestry for admixed individuals varied widely across farms (0–100%). On average, the proportion of Zambian-Southern admixed individuals on farms was significantly higher (87.2% ± 22.8%) than those assigned uniquely to Zambian (mean: 10.8% ± 21.6%) or Southern ancestry (2.0% ± 9.1%) (p < 0.05 for all comparisons; Figure 2B). Our results indicated that some farms included animals with high assignments to wild ancestry groups, whilst the majority of the farms included mostly admixed animals. Admixed individuals were removed from further analysis.

Figure 2.

Ancestry and population structure of captive sable antelope on 358 private wildlife farms in South Africa

(A) Bayesian clustering analysis results for 7,271 captive sable antelope individuals using STRUCTURE for K = 5, with wild groups used as reference. Each individual is represented by a vertical bar indicating its membership in each genetic cluster.

(B) Boxplot showing the percentage of Zambian, Southern and admixed captive individuals. Each dot represents the proportion of each group present at a given farm.

(C) Principal component analysis (PCA) of Captive Zambian and Captive Southern datasets alongside individuals representing Zambian and Southern wild ancestries. See also Figure S1; Tables S3–S5.

Simulations identified TrioML as the most suitable relatedness estimator for datasets of non-admixed individuals of Zambian and Southern ancestry (Table S1). Mean relatedness for both groups was close to zero (r = 0.07, Figure S2), although a substantial proportion of individuals showed high pairwise relatedness (r ≥ 0.4; 69%). Filtering highly related individuals out of the dataset led to a Captive Zambian dataset of 453 individuals from 84 farms, and a Captive Southern dataset of 109 individuals from 44 farms (Figure 1). This filtering step also resulted in only five private wildlife farms, each holding at least 20 sable individuals with the same wild ancestry (all of Zambian origin), resulting in a total of 175 individuals that comprise our Selected Wildlife Farms dataset (see Table S2 for detailed information on these farms). Notably, no farm met the 20-individual threshold for Southern ancestry, preventing farm-specific analysis for this group (Figure 1). It is important to note that our dataset represents a subset of individuals from each private farm rather than the whole group of individuals from a specific farm.

Population differentiation and migration

The PCA confirmed strong genetic overlap between captive and wild individuals within both groups (Zambian and Southern), reinforcing the genetic similarity between captive and wild sable antelope (Figure 2C). Population differentiation analyses showed low Fst values between wild and captive individuals for both Zambian (Fst = 0.025, p = 0) and Southern ancestry (Fst = 0.022, p = 0) (Table S3). In support, AMOVA results revealed that most of the genetic variation occurs within individual samples (92% in Zambian and 89% in Southern), followed by variation among individuals within populations (5% in Zambian and 9% in Southern), while the lowest variation was observed between wild and captive groups (3% in Zambian and 2% in Southern) (Table S4). In contrast, higher differentiation was found between the Captive Zambian and Captive Southern datasets (Fst = 0.174, p = 0), consistent with PCA results, which showed distinct clustering between these groups (Figure 2C). Although most variation still occurs within individual samples (80%), the second highest component of variation was found between the two captive groups (17%, Table S5). Lastly, differentiation among farms showed low Fst values (0.02–0.06; Table S6), with AMOVA indicating residual variation between farms (2%) (Table S7).

In the Selected Wildlife Farms dataset, Bayesian clustering identified a maximum ΔK at K = 3. However, the mean log likelihood probabilities showed minimal variation between K = 1 and K = 3, indicating weak population structure among farms (Figure S3B). In accordance, individual assignments for the different K values did not separate individuals originating from different farms (Figure S4A). This result is corroborated by the PCA, which shows a general overlap of individuals from the five farms (Figure S4B).

Pairwise gene flow was detected among the five wildlife farms comprising the Selected Wildlife Farms dataset. However, values were generally low, ranging from 0.006 to 0.188, with two farms primarily acting as sources and the others primarily functioning as sinks (Figure 4A; Table S8). Within-farm gene flow ranged from 68% to 82% (Table S8), indicating that external contributions from other wildlife farms accounted for 18%–32% of the genetic composition of individual farms. This suggests a considerable level of individual exchange across farms.

Figure 4.

Differentiation, genetic diversity and migration patterns among private wildlife farms in the Selected Wildlife Farms dataset

(A) Schematic representation of gene flow among the five farms, with recent migration rates estimated using BAYESASS. Arrows indicate the direction of migration, with thickness proportional to the migration rate per generation. Circle size represents within-farm migration and dashed lines indicate farms with higher migration rates (Table S8).

(B) Graphical representation of allelic richness (AR) and expected heterozygosity (He) for each of the five farms and their subsets, and the Captive Zambian dataset included for comparison.

(C) Boxplot of individual inbreeding coefficients for each of the five farms and their subsets, calculated using COANCESTRY. Asterisks (∗) indicate groups with significant pairwise differences in inbreeding coefficients (p < 0.05). For details on specific pairwise comparisons, refer to Table S12. Captive Zambian dataset included for comparison. See also Figures S3 and S4; Tables S2, S6, S7, and S11.

Genetic diversity

Genetic diversity metrics were not significantly different between the Captive Zambian and Captive Southern groups, and their respective wild counterparts (p > 0.05; Figures 3A and S5; Tables S9 and S10). Any observed discrepancies, such as those in allele counts and private alleles, disappeared when comparing with the randomized subsets (Tables S9 and S10). For the Selected Wildlife Farms dataset, genetic diversity did not differ significantly from the overall captive dataset (p > 0.05) (Figure 4B; Table S11). However, allelic richness (AR) was slightly lower within each farm compared with the entire Captive Zambian dataset (Figure 4B; Tables S9 and S11).

Figure 3.

Genetic diversity of Zambian and Southern ancestries

(A) Graphical representation of allelic richness (AR) and expected heterozygosity (He) in the wild and captive Zambian and Southern sable groups.

(B) Boxplot of individual inbreeding coefficients calculated using COANCESTRY for the wild and captive Zambian and Southern sable groups. See also Figure S5; Tables S9, S10, S13, and S14.

Both captive groups exhibited significantly lower individual inbreeding coefficients than their respective wild groups, regardless of sample size differences (Figures 3B and S5). In the Selected Wildlife Farms dataset, individual inbreeding coefficients varied across the five farms (Figure 4C), with four farms showing significant differences (p < 0.05) compared with subsets (Table S12), suggesting that these results were not driven by sample size.

Estimated effective population sizes (Ne) were 275 (95% CI = 230–329) for Captive Zambian and 80 (95% CI = 57–113) for Captive Southern (Table S13). These values highlight a marked difference between the two groups, with the Captive Zambian maintaining a considerably higher effective population size.

Discussion

Since the 1970s, captive sable antelope numbers have grown considerably, primarily on private wildlife farms in South Africa, and to a lesser extent, in Namibia. This increase has led to captive sable being recognized as a stronghold for the species, in contrast with the declines documented in several wild populations across its formerly widespread range, particularly within the Southern group.^{36,37,38,39,40} Although the genetic footprint of wild sable populations is documented,^34,35 a similar comprehensive population genetic assessment of captive sable has been lacking thus far. To our knowledge, this is the first study to examine the genetic ancestry, admixture patterns, and genetic diversity parameters in captive sable from southern Africa, with direct comparisons with wild counterparts.

Our ancestry analyses revealed that the captive population resident on private wildlife farms in South Africa is limited to two of the five wild sable ancestries present across the species’ range. Specifically, Zambian and Southern wild groups were identified, whereas none of the other groups were present. Several factors may explain why captive sable from other origins are absent. The Angolan giant sable (H. n. variani) is renowned for its remarkable phenotype, including long and massive horns and black facial markings, which has earned its recognition as the most iconic of all sable antelope.^32,46 However, this subspecies is listed as Critically Endangered, with an estimated population of fewer than 310 individuals.^36,47 Moreover, strict regulations by the Angolan Government prohibit the export of animals from Angola.⁴⁸ Fortunately, the in situ conservation plan implemented in recent years for the Angolan giant sable has been successful.^47,48 The absence of wild Eastern ancestry among captive sable may be attributed to their smaller body size and shorter horns compared with the other southern African ancestries,^32,46 making their phenotype commercially less valued. However, consistent with reports of rare historical translocations from Malawi and Mozambique into South Africa, we detected five samples showing apparent Eastern ancestry, but these were excluded from the final dataset due to low genotype quality. Lastly, West Tanzanian ancestry is also not represented among the captive sable population. This population occurs in a geographically isolated region distant from South Africa and has only recently been recognized as a distinct population group.³⁵

Sable of Zambian ancestry dominate on private wildlife farms in South Africa, with 22% of the individuals analyzed representing this wild genetic group, while sable of Southern ancestry comprises only 2% of the analyzed captive individuals. The substantial underrepresentation of Southern sable is somewhat unexpected, considering that this is the wild group that occurs locally.³² During the rise of private wildlife farming, largely in the 1970s and 1980s, animals were relatively easily moved from neighboring southern African countries into South Africa due to the absence of policies restricting the use of non-native species.³¹ During this time, iconic wildlife species such as sable antelope,⁴⁹ roan antelope,²⁹ and nyala,⁵⁰ amongst others, were imported. Sable individuals with Zambian ancestry have consistently been the focus of South African wildlife farmers, perhaps due to their, on average, larger horn size and darker facial markings compared with Southern sable.^32,46 Although some phenotypic differences between lineages have been described, information on their functional or adaptive significance remains limited and would require dedicated genomic and ecological studies. The current predominance of Zambian ancestry in our captive dataset thus reflects historical management preferences. This pattern is particularly unfortunate given that recent population declines have been mainly reported within the Southern lineage.^37,38,39,40 Perhaps not surprisingly, the large population of captive Zambian sable (1,634 individuals in our dataset) shows a small effective population size (Ne = 275), well below the ideal 500 individuals proposed to retain genetic diversity.⁵¹ The small effective population size likely reflects the small number of family groups originally translocated into South Africa, combined with subsequent selective breeding practices.⁵²

Our study revealed that most of the captive sable exhibit admixed ancestry between Zambian and Southern wild populations. Efforts to assign a wild origin to captive sable individuals have been in demand since the early 2000s. Initially, genetic testing was based on mitochondrial DNA alone, which fails to capture the complexity of genetically admixed backgrounds due to its maternal inheritance. Although a number of prominent sable antelope breeders incorporated genetic data into their management, the majority of sable owners employed practices focused on economically valuable traits such as horn size.^31,52 Since 2015, new genetic tools have become available for sable,⁵³ allowing for nuclear DNA testing and a more informed management of the captive population.

Both Zambian and Southern captive sable exhibit moderate genetic diversity across all tested parameters, with no significant differences from their wild counterparts, suggesting an effective preservation of genetic diversity in the captive stock, notwithstanding the relatively small effective population sizes. This, along with the low differentiation observed between captive and wild groups, reflects translocation practices that captured a substantial proportion of the genetic diversity present in wild populations. Although detailed historical records are limited, the observed levels of genetic diversity suggest that translocations from wild populations into captivity may have occurred across multiple events or over an extended period, thereby mitigating the strong founder effects typically associated with captive populations.^9,54 The maintenance of genetic diversity in captive sable may be influenced by the frequent exchange of individuals between farms, which occurs through approximately 40 auctions per year, managed by registered auctioneering companies, in addition to numerous private sales. Auctions work as a dilution factor for the effects of genetic isolation and prevent the formation of distinct captive lineages. Interestingly, sable introduced into the United States (US) during the first half of the twentieth century, currently held on both private ranches and in zoological gardens, display genetic diversity differences between the two captive populations.⁵⁵ However, unlike South African private wildlife farms and US ranches, the US zoological gardens operate as closed populations that do not incorporate new sable individuals, thus potentially favoring genetic drift and amplifying differences between sable housed in zoological gardens and those on wildlife ranches. We acknowledge that the use of a quite different molecular panel by Gooley et al.⁵⁵ (5,000 SNPs) and our study (34 microsatellites) may limit the comparison.

Private wildlife farms operate as semi-closed populations, with individual exchanges primarily influenced by market demands. In our comprehensive analysis of more than 7,200 individuals from 358 farms, only five farms contained more than 20 individuals of the same wild ancestry that were not closely related, all from Zambian ancestry. Our clustering analysis revealed no genetic structure between the farms, and a moderate to high percentage of exchanged individuals per generation, which mirrors a metapopulation management approach. This management practice aligns with the recognized understanding that introducing a single migrant per generation can enhance genetic diversity and reduce inbreeding^56,57,58,59 and has been previously suggested for captive sable in the US.^55,60

From a management perspective, our results highlight the importance of incorporating genetic information into breeding decisions on private wildlife farms. In programs aimed at preserving distinct lineages, limiting admixture between genetic groups would help preserve their genetic distinctiveness, reduce the risk of disrupting locally adapted gene combinations, and maintain evolutionary potential. Breeding programs should also maintain genetic diversity by minimizing inbreeding and avoiding repeated use of a limited number of high-value individuals. When conservation actions are considered, the ancestry and genetic diversity of candidate individuals should be assessed in relation to the intended recipient population and alongside demographic, ecological, epidemiological, and logistical considerations.^61,62 Our results can therefore inform the identification of genetically appropriate candidates but should not, on their own, be used to establish an individual’s suitability for a specific conservation action. Accordingly, admixed individuals may remain suitable for commercial breeding programs but should not be considered for conservation actions aimed at preserving the genetic composition of differentiated wild populations. These findings also reinforce the need for stronger coordination between the public and private sectors, as previously highlighted by Parrini et al.⁴¹ and for the integration of genetic information into breeding, translocation, and reinforcement decisions.

The large number of wildlife populations held in captivity offers a remarkable opportunity for future conservation efforts, even when they are not maintained primarily for conservation purposes. Sable antelope is a notable example, as a numerically large captive population is maintained primarily to support the commercial wildlife industry. However, our results indicate that captive sable antelope on South African wildlife farms should not be considered a conservation reservoir as a whole. Only two of the five wild genetic groups are represented, and fewer than 25% of the analyzed individuals show no evidence of admixture. This pattern, likely driven by management practices lacking sufficient genetic guidance, may hinder effective conservation initiatives that seek to preserve the integrity of genetic ancestry (e.g., Cook et al.).⁶³ Nevertheless, the non-admixed captive groups retain levels of genetic diversity comparable with those of their wild counterparts and may therefore be potential candidates for future rewilding actions, subject to the genetic and non-genetic assessments outlined above.

Currently, the IUCN Red List assessment does not incorporate metrics of genetic diversity, which have been widely recognized as pivotal components of species viability.^{64,65,66,67,68,69,70} While IUCN assessments report overall population trends, numerical data alone provide an incomplete picture of a species’ genetic health and long-term viability. This limitation is further compounded by the current lack of clearly defined operational taxonomic units within the species. Preserving the integrity of wild populations when relying on captive populations requires a robust assessment of the latter’s genetic background. Our results provide a concrete example of the need to integrate genetic data into conservation assessments, showing that, in our case study, only a small subset of individuals could be validated for conservation actions when aiming to match the genetic background of the original population. Despite the ongoing debate regarding the crossing of distinct lineages to potentially reduce extinction risks,^{71,72,73,74,75} we argue that this approach should be regarded carefully when wild populations of highly differentiated genetic groups continue to thrive in their natural habitats, and their genetic integrity can still be maintained, as in the case of the sable antelope. As a practical first step, incorporating genetic metrics into conservation frameworks could involve leveraging existing datasets to assess genetic diversity, inbreeding, and population structure, and to define management units. In addition, routine genetic screening of individuals used in conservation actions, together with the development of shared genetic databases, could support more informed decision-making. While the implementation of such approaches requires consideration of logistical and financial constraints, the use of already available data provides an immediately applicable starting point. Our study demonstrates that a large captive population should not be blindly considered as a stronghold of a species’ genetic diversity or as an undoubted source for successful conservation outcomes. Conservation policies advocating the potential utilization of captive populations would benefit from a comprehensive understanding of their genetic background to avoid inadvertently compromising the conservation of wild species.

Limitations of the study

In this study, we provide a comprehensive assessment of the genetic composition of captive sable antelope on wildlife farms, while acknowledging some limitations. First, the captive dataset is geographically restricted to South Africa. Second, sample sizes are uneven across farms, reflecting inherent differences in herd size and management within the private wildlife industry. Future work integrating genomic data and captive populations from other locations will be important to better assess their suitability for conservation actions and to identify alternative sources of individuals.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Mariana Meneses-Ribeiro (mariana.ribeiro@cibio.up.pt).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •The genetic data used in this study were generated within the context of a commercial service offered to private wildlife farmers. Public deposition of the dataset is restricted due to contractual obligations and its commercially sensitive nature. Nevertheless, access to the data may be considered on a case-by-case basis upon request, subject to the approval of the relevant data owners and stakeholders.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

The authors acknowledge the research support of the Portuguese Foundation for Science and Technology (FCT) under the project UIDP/50027/2020. M.M.-R. and R.G. were supported by the FCT (PhD grant 2021.05120.BD and research contract 2022.07926.CEECIND, respectively). We further acknowledge Susana Lopes for her skilled laboratory and technical assistance, and the CIBIO lab staff for continued support.

Author contributions

R.G., M.M.-R., B.J.v.V., and N.F. conceived and conceptualized the study. R.G. coordinated the project. B.J.v.V. and S.D. established and maintained collaborations with private wildlife farmers and led the collection of captive samples and associated contextual information. P.V.P. provided the wild sable antelope genotypes and background data for the reference wild dataset. M.M.-R. performed the laboratory work and conducted data analyses with guidance from R.G., B.J.v.V., K.-P.K., and N.F. M.M.-R. and R.G. wrote the manuscript with input from all authors. All authors revised and approved the final version of the manuscript.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Biological samples
7,988 captive sable antelope genotypes	This study	N/A
Critical commercial assays
DNeasy Blood & Tissue Kit	QUIAGEN	Cat# 69504
Deposited data
Wild sable antelope genotypes	Vaz Pinto35	N/A
Oligonucleotides
34 microsatellite primer pairs (See Table S14)	Vaz Pinto et al.53	https://doi.org/10.1007/s10344-014-0880-6
Software and algorithms
GeneMapper v4.0	Applied Biosystems	Cat# 4440915
GenAlEx v6.5	Peakall and Smouse76,77	https://biology.anu.edu.au/research/software/genalex
STRUCTURE v2.3.4	Pritchard et al.78; Falush et al.79,80; Hubisz et al.81	https://web.stanford.edu/group/pritchardlab/structure.html
COANCESTRY v1.0.1.7	Wang et al.82	https://doi.org/10.1111/j.1755-0998.2010.02885.x
ADEGENET	Jombart et al.83	https://cran.r-project.org/package=adegenet
FSTAT v2.9.3	Goudet84,85	https://www2.unil.ch/popgen/softwares/fstat.htm
ARLEQUIN v3.5	Excoffier and Lischer86	https://cmpg.unibe.ch/software/arlequin35/
STRUCTURE HARVESTER v0.6.94	Earl and vonHoldt87	https://github.com/dentearl/structureHarvester
BAYESASS v3.0.5.6	Wilson and Rannala88	https://github.com/brannala/BA3
HYBRIDLAB v1.0	Nielsen et al.89	https://doi.org/10.1111/j.1471-8286.2006.01433.x

Experimental model and study participant details

For this study, microsatellite genotype data were generated for captive sable antelope (Hippotragus niger) maintained on private wildlife farms across South Africa. Information on sex and age was not consistently available across the dataset and these variables were therefore not included in the analyses. Animals were privately owned and managed by the respective wildlife farms. In addition, previously generated microsatellite genotype data from wild sable antelope representing the wild genetic groups were used as reference data.

Method details

Sampling and molecular analysis

A total of 7,988 samples (hair, blood, or cartilage) of captive sable were collected from 358 private wildlife farms across South Africa between 2015 and 2023. Additional details on sample types are provided in Method S1. Genomic DNA was extracted using the QIAGEN DNeasy Blood and Tissue Kit, following the manufacturer’s protocol. A set of 34 species-specific microsatellite markers (Table S14) was selected based on reported levels of polymorphism and the absence of null alleles or genotyping errors.⁵³ To minimise errors, each genotype was amplified in duplicate following the cycling conditions outlined in Vaz Pinto et al.⁵³ Amplicons were separated on an ABI 3130xl Genetic Analyser (Applied Biosystems), with allele sizes scored against the GeneScan 500 LIZ Size Standard using the GeneMapper v4.0 software (Applied Biosystems) and manually verified twice, independently. Genotypes were compared using GenAlEx v6.5 software^76,77 to detect and remove potential duplicate samples (those matching in all loci or with mismatches up to 2 loci). Samples with > 15% missing data were excluded from the dataset.

Quantification and statistical analysis

Captive ancestry and relatedness

The ancestry of captive individuals was assessed by clustering analysis in STRUCTURE v2.3.4^78,79,80,81 using two sequential and nested steps. First, we tested the ancestry of the captive individuals against a wild reference dataset comprising the five wild sable groups³⁵ to aid the estimation of allelic frequencies. This reference dataset was generated in the same laboratory, ensuring it is calibrated for allele binning. Wild reference individuals were assigned to known populations, while captive individuals were left unassigned, allowing STRUCTURE to infer their ancestry. Secondly, we refined the analysis by repeating the clustering, this time using only the wild reference groups identified in the first analysis as present in the ancestry of captive sable. This strategy minimises artefacts and stochastic errors, which could lead to misleading ancestry assignments in captivity. Clustering analyses were performed using the admixture model with correlated allele frequencies (Usepopinfo activated for reference samples) and 10 independent replicates of the Markov Chain Monte Carlo (MCMC), each with 10⁶ iterations including a burn-in period of 500,000 steps. The number of genetic clusters (K) was set to 5 or 2, respectively, for the first and second steps, matching the number of wild groups in the analyses. To define a threshold that accurately allows identifying individuals admixed between distinct wild origins, we generated simulated genotypes for the two wild ancestries found in captive sable (see Method S2 for details). Simulated genotypes were predominantly assigned to their group with Qi ≥ 0.9, which was then used to classify individuals as non-admixed among captive sable. Admixed individuals were excluded in subsequent analyses to avoid artefacts.

Pairwise relatedness between captive individuals was estimated using COANCESTRY v1.0.1.7⁸² for all pairs of captive individuals from the same ancestral group and applying 1,000 bootstrap replicates, after determining the optimal relatedness estimator through simulations (Method S3). Individuals with a pairwise relatedness ≥ 0.4 were excluded from all downstream analyses.

The dataset was then reorganized into three subsets: (A) captive individuals representing Zambian wild ancestry (Qi ≥ 0.90), hereafter named “Captive Zambian”; (B) captive individuals representing Southern wild ancestry (Qi ≥ 0.90), hereafter named “Captive Southern”; and (C) all private farms with ≥ 20 individuals from the same wild ancestry (Zambian or Southern), hereafter named “Selected Wildlife Farms” (see Figure 1).

Population differentiation and migration

To further investigate genetic clustering, we used the datasets defined above to conduct two principal component analyses (PCA) using the ADEGENET package⁸³ in R 3.5.0⁹⁰: one combining the Captive Zambian and Captive Southern datasets with both wild ancestries, and one with the Selected Wildlife Farms dataset only. Unlike STRUCTURE, PCA is based solely on genotype similarity and does not rely on biological assumptions.

Genetic differentiation among captive datasets and their respective wild reference groups, as well as between private wildlife farms, was calculated using pairwise Fst.^84,85 Hierarchical genetic structure was examined using an Analysis of Molecular Variance (AMOVA) based on allele frequencies.⁹¹ In particular, we were interested in partitioning total variance components into those derived from within and among Captive Zambian and Captive Southern datasets, as well as their respective wild ancestries, and within and among private farms included in the Selected Wildlife Farms dataset. Fst and AMOVA estimates were performed using 1,000 permutations with ARLEQUIN v3.5.⁸⁶

To further assess population structure, we tested whether the Selected Wildlife Farms dataset represents a single genetic unit or multiple distinct clusters corresponding to different farms, using STRUCTURE. This analysis was performed without reference wild individuals, and the number of genetic clusters investigated ranged from K=1 to K=6, using the admixture ancestry model with the same running parameters. The number of clusters was determined using log-likelihood values and delta statistics⁹² in STRUCTURE HARVESTER v0.6.94.⁸⁷

Additionally, we used a Bayesian method implemented in the BAYESASS v3.0.5.6 software⁸⁸ to estimate recent rates of exchange for non-admixed individuals between private farms included in the Selected Wildlife Farms dataset. We estimated posterior probabilities of the rate of recent migration using 20,000,000 MCMC iterations, with the first 200,000 iterations discarded as burn-in.⁸⁸

Genetic diversity

Genetic diversity was estimated for Captive Zambian and Captive Southern datasets and compared to their wild counterparts to evaluate the effects of captivity. To control for sample size differences, we randomly sampled 10 subsets of captive individuals matching wild dataset sizes. Genetic diversity parameters were also calculated for the Selected Wildlife Farms dataset to examine potential differences when captive individuals from different farms are amalgamated into a unique group in comparison with considering each farm an independent group. For this comparison, we randomly sampled from the Captive Zambian dataset to match the smallest sample size among the Selected Wildlife Farms dataset.

We used FSTAT v2.9.3^84,85 to estimate allelic diversity (AD), allelic richness (AR), and population inbreeding coefficients (Fis) following Weir and Cockerham’s method.⁹³ Individual inbreeding coefficients were calculated using COANCESTRY. GenAlEx was used to calculate observed (Ho) and expected heterozygosity (He), number of alleles (Na), number of private alleles (NpA) and total number of alleles (Atot). Significant differences in AD, AR, He, Na and individual inbreeding coefficient between subsets were tested using Wilcoxon-Mann-Whitney tests in R 3.5.0.⁹⁰ Additionally, effective population size was estimated for each captive group using COANCESTRY.