Pedro Vaz Pinto1,2,3,4, Hans Siegismund5, Bettine Jansen van Vuuren6, Nuno Ferrand1,2,3,6, & Raquel Godinho1,2,3
1CIBIO/InBIO – Centro de Investigação em Biodiversidade e Recursos Genéticos, Universidade do Porto, Campus Agrário de Vairão, 4485-661 Vairão, Portugal
2Departamento de Biologia, Faculdade de Ciências, Universidade do Porto, Rua do Campo Alegre 4169-007 Porto, Portugal
3ISCED – Instituto Superior de Ciências da Educação da Huíla, Rua Sarmento Rodrigues, Lubango, Angola
4The Kissama Foundation, Rua Joaquim Capango nº49, 1ºD, Luanda, Angola
5Department of Biology, University of Copenhagen, Ole Maaloes Vej 5, 2200 Copenhagen N, Denmark
6Department of Zoology, Faculty of Sciences, University of Johannesburg, Auckland Park 2006, South Africa
Abstract
Dramatic landscapes and ecological heterogeneity are known to influence the biogeography of Africa, but surprisingly few studies to date have implemented a population genetics approach to investigate intraspecific relationships in African bovids. The sable antelope (Hippotragus niger) is one of the most emblematic antelopes, being an economically important species in Southern Africa, and includes populations of global conservation concern. As a highly specialized species widely distributed across the continent, it is an ideal candidate for a population study, and here we report on results obtained by applying a panel of 57 autosomal microsatellites on a comprehensive dataset of 369 contemporary and 31 historical sable samples. We found sable to be
strongly structured into five geographical clusters delimited by well-defined physical barriers. We also explore patterns of genetic diversity, differentiation and substructuring. The critically endangered giant sable antelope proved to be the most differentiated population, but also showed high levels of genetic depletion, suggesting that genetic drift may have contributed to the differentiation values. We conclude that the five clusters have been maintained with relatively low levels of gene flow, but two contact zones were identified in eastern Zambia and central Mozambique. The results of this study support a five subspecies classification, partially concordant with traditional taxonomy based on morphometrics, but suggesting the existence of a yet undescribed taxon in west Tanzania.
Introduction
African bovids experienced an explosive radiation starting in the Miocene, when most of the extant antelope tribes are thought to have originated in response to a cooling climate and subsequent adaptation to spreading grasslands, savannas and drier habitats (Vrba 1995; Hassanin & Douzery 1999; Fernandéz & Vrba 2005; Hassanin 2012). Based on the fossil record, the tribe Hippotragini shows a marked proliferation of species from the middle to the late Pliocene onwards, but only eight species and three genera survived to historical times, while being restricted to Africa and the Arabian Peninsula (Vrba 1994; Kingdon 2013). Extant hippotragine are large-bodied grazing antelopes in which species belonging to genus Addax and Oryx present some remarkable adaptations to desert or semi-desert environs, while the genus Hippotragus has evolved linked to savannas and woodlands (Kingdon 2013). The climatic fluctuations of the Pleistocene are expected to have influenced the intraspecific variation within the tribe, similarly to what has been reported in other ungulates, as result of speciation in refugia (Flagstad et al. 2001; Lorenzen et al. 2010, 2012; Hewitt 2004), and shaped by the effects of physeogeographic features such as rivers and mountain chains (Cotterill 2003a). First described in South Africa, the sable antelope Hippotragus niger HARRIS 1838 is one of the two existing members of the genus and has specialized to the mesic conditions of the miombo woodlands, a type of broad-leafed deciduous open woodlands dominated by Brachystegia/ Julbernardia spp. trees growing on poor dystrophic soils (Estes 2013). Although sable can marginally occur in different types of savannas in the south of their range, they are one of the larger mammal species more strongly associated with the miombo zone (Estes 2013), and its distribution closely matches the occurrence of this ecotype across eastern and south-central Africa. Generally present in low densities,
sable is typically an ecotone species, utilizing the edges between the woodland and grassland while avoiding very open or waterlogged areas (Estes 2013). As in all remaining hippotragine both male and females grow horns, but sable are unique in exhibiting a marked sexual dimorphism, with bulls being larger in size, much darker in coloration and developing much longer and curved horns (Estes 2013). Sable antelopes are gregarious, forming strongly phylopatric matriarchal herds composed of cows, calves and young, while males disperse before maturing and may temporarily form bachelor groups before eventually establishing their own territories (Estes 2013). Generally sable occupies niches that minimize interspecific competition and it has been suggested that the pronounced sexual dimorphism and social behaviour makes sable the most sedentary in habits and specialized hippotragine species (Estes 2013).
Morphological differences such as body size, coloration patterns, or skull and horn measurements, observed in geographically separated populations, led to the description of four subspecies of sable (Ansell 1972; East 1999; Estes 2013). The typical race or southern sable H. n. niger is characterized by females becoming almost as dark as bulls and is widely distributed across southern Africa to the south of the Zambezi (Groves & Grubb 2011; Estes 2013). A smaller sable with relatively shorter horns from eastern Africa was described as Roosevelt’s sable H. n. roosevelti (Groves & Grubb 2011; Estes 2013). Originally found in Kenya, it was subsequently suggested that Roosevelt’s sable would likely include populations found further south into south-eastern Tanzania and northern Mozambique (Siege & Baldus 1999; Booth 2002). Kirk’s sable H. n. kirkii was first described from south-western Zambia, having different facial mask pattern, males growing longer horns on average and females of brown-reddish colour (Ansell 1972; Groves & Grubb 2011; Estes 2013). Known to Zambia to the north of the Zambezi, the distribution range for the latter subspecies has remained unclear, although Ansell (1972) suggested that it could extend to Malawi, Democratic Republic of Congo (DRC) and western Tanzania. The giant sable H. n. variani is the most geographically distant and isolated population, restricted to a small region in central Angola, and is characterized by a very dark face, and larger skull and horn development in males (Blaine 1922; Ansell 1972; Groves & Grubb 2011; Estes 2013). After comparing a series of sable skulls and skins obtained in museums, Groves (1983) suggested another subspecies H. n. anselli, to be confined to the rift region of Malawi and eastern Zambia, between lake Malawi and the Muchinga escarpment/ Luangwa river, albeit this fifth subspecies has been rarely considered by subsequent authors. More disruptive proposals have also been made, recommending specific status to giant sable (Blaine 1922) or to Roosevelt’s sable (Groves & Grubb, 2011). However, the four subspecies classification remains the most
widely used, even if strictly based on phenotypic traits and lacking well defined boundaries, as critical populations from the DRC, Tanzania, northern Mozambique, Malawi and eastern Zambia, are yet either untackled or problematic.
The first molecular studies to focus on sable antelope made use of one or few mitochondrial fragments and revealed the existence of surprisingly deep maternal lineages within the species (Mathee & Robinson 1999; Pitra 2002). However, these and subsequent results were based on limited sampling efforts and proved difficult to interpret and reconcile with previous taxonomic considerations (Mathee & Robinson 1999; Pitra 2002, 2005; Jansen Van Vuuren et al. 2010; Groves & Grubb 2011). A recent mitogenomic study analysing a much more comprehensive dataset provides some critical insights into the evolutionary history of the species, unveiling a strong association to geomorphological and climatic events, and suggests the possibility of past introgression in Tanzania to clarify previous inconsistencies (Rocha et al. in prep). This study reveals three main maternal sable lineages that subsequently split into nine haplogroups and sub-haplogroups, evolving during the middle to late Pleistocene, and resulted in six geographical groupings, one being the isolated giant sable and the remaining well demarcated by geological features, namely the eastern arc rift system (EARS), the eastern arc mountains (EAM), the Muchinga escarpment and the Zambezi river (Rocha et al. in prep). Five of these mitochondrial groupings would be largely concordant with a five subspecies classification, while leaving a sixth and unresolved grouping in western Tanzania. The co-existence of an additional and much more ancient maternal lineage dating from the early Pleistocene in western Tanzanian animals, has been tentatively explained as resulting from long distance intraspecific outbreeding (Pitra et al. 2002) or from mitochondrial capture following an interspecific hybridization event with an extinct taxon (Rocha et al. in prep).
Once widely distributed across the subcontinent, habitat destruction and consumption use throughout the twentieth century, caused local extinctions and fragmentation of surviving sable populations, leading to the confinement of natural populations mostly isolated in protected areas (East 1999; Estes 2013). Currently, several populations are threatened, including the relic giant sable which regarded as a national icon in Angola and yet listed as critically endangered (IUCN 2008), with a total population estimated at less than 200 animals and managed in situ. Other fragmented sable populations in protected areas have been steadily declining (Ogutu & Owen-Smith 2005; Dunham 2012; Crosmary et al. 2015), and may in the future require drastic conservation measures including restocking from source populations.
Contrasting to decreasing wild populations, the numbers have been recently expanding on private land across southern Africa where the sable ranks as a high value species destined for trophy hunting or breeding purposes, and is one of the most widely represented large ungulates in private game farms (Bothma et al. 2010; Taylor et al. 2016). In South Africa alone, wildlife ranching is a multi-billion rand industry with sable antelope generally regarded as the “glamour” breed, and where game auction prices for breeding bulls have been escalating, and reached a record value of US$ 1.9 million paid for one sable bull in 2015 (Lindsay et al. 2013; Pitman et al. 2016; Taylor et al. 2016). As result of its high economic value, sable are currently bred and managed intensively in many game ranches, and often translocated or introduced outside their historical distribution range (Bothma et al. 2010; Taylor et al. 2016). Some regional variants are more highly prized by the hunting industry for producing better trophies, usually a function of larger horn development in bulls (Crosmary et al. 2013). Although countries may have strict regulations in place to prevent the introduction of exotic subspecies to avoid contamination of the indigenous gene pools, those can prove hard to enforce when dealing with a high value species whose intraspecific taxonomy remains unresolved (Cousins et al. 2010). A better understanding of indigenous sable populations and clarification of their genetic relationships will prove crucial as a conservation tool for a rare species that include endangered wild populations, and to influence government policy making and private ranch owners on the management of such an economically important species.
In this study we present the first population genetic analyses on sable antelope with a comprehensive dataset obtained across the entire species geographic range, combining recent and historical samples. All putative subspecies were included in the sampling effort, as well as geographically intermediate and previously unresolved populations. By using a set of autosomal microsatellites we explore the main contemporary geographic patterns and genetic relationships between sable populations. We predicted sable to display clear patterns of population structure shaped by existing geomorphological barriers and by demographic processes. Specifically we aimed to 1) define the main groups or meta-populations of sable antelope; 2) estimate genetic diversity and relationships among the main groups; 3) explore substructuring at a finer scale within each cluster; 4) test the hypothesis of mitochondrial introgression following hybridization in western Tanzania; 5) provide taxonomic insights. We expect the definition of discrete conservation units will constitute an invaluable tool in designing future policies for management and conservation of the sable antelope.
Materials and Methods
Sample collection
We used a total of 369 modern samples (dating from 1993 to 2015) and 31 historical samples (from 1913 to 1964), from 64 sampling localities covering the whole species native distribution range (Supplementary Table S1). All recognized subspecies from the 11 countries where sable are known to occur, are represented in this dataset.
DNA extraction and amplification
Total genomic DNA was extracted from tissue samples using the QIAGEN DNeasy Blood & Tissue Kit, and following manufacture instructions. Individual multilocus genotypes were scored for a set of 57 species-specific polymorphic microsatellites. The amplification of loci followed the methodology and conditions described in Vaz Pinto et al. (2015), always using negative controls to monitor for possible contaminants. PCR products were separated by size on an ABI3130xl Genetic Analyzer. Allele sizes were scored against the GeneScan 500 LIZ Size Standard, using GENEMAPPER 4.0 (Applied Biosystems) and manually checked twice. The accuracy of genotypes was confirmed through re-amplification and re-analyses of 20% of random selected samples from each locus (Pompanon et al. 2005), resulting in complete concordance among replicates. For historical samples, DNA extractions followed the procedures of Dabney et al. (2013) and were performed in a dedicated room for low quality DNA using negative controls to monitor for contamination. Individual multilocus genotypes were scored for a set of 57 species-specific polymorphic microsatellites following Vaz Pinto et al. (2015). Multiplex reactions to amplify historical samples were split from the original ones to contain few loci per reaction. Amplification of historical samples were replicated four times for each sample and were performed in a dedicated PCR room maintaining conditions to reduce the risk of contamination. PCR products were separated by size in an ABI3130xl genetic analyzer. Allele sizes were scored against the GeneScan 500 LIZ Size Standard, using GENEMAPPER 4.0 (Applied Biosystems) and manually checked twice. The accuracy of genotypes was confirmed through re-amplification and re-analyses of 20% of randomly selected samples from each locus (Pompanon et al. 2005), resulting in complete concordance among replicates. A threshold of 33% of missing data was assumed ending up with a 400 sample final dataset.
Genetic diversity and population structure
Preliminary analyses using the Bayesian clustering software STRUCTURE 2.3.4 (Pritchard et al, 2000; Falush et al. 2003) allowed the identification of main groups, and these were further subdivided into coherent geographical subgroups.
We used the software ARLEQUIN 3.5 (Excoffier & Lischer 2010) to evaluate deviations from Hardy-weinberg equilibrium (HWE) and to test for pairwise linkage disequilibrium (LD) for all loci in the main groups previously identified. Microsatellite diversity was evaluated separately for each main group and geographical subgroup, based on the expected heterozygosity (He), the inbreeding coefficient Fis (Weir & Cockerham 1984), mean number of alleles per locus (Na) and number of private alleles (Npa) for each locus and population, also using ARLEQUIN 3.5, while the allelic richness (AR) was computed with the program FSTAT ver. 2.9.3.2 (Goudet 2001). Significant levels for all tests were adjusted using the sequential method of Bonferroni for multiple comparisons in the same data set (Rice 1989).
We tested sable population structure using the software STRUCTURE with the admixture model and correlated allele frequencies and no prior geographic information. Analyses was performed on 10 independent runs for each K with 1,000,000 MCMC iterations following a burn-in period of 500,000 steps, and number of clusters (K) estimated for 1 to 10. This methodology was firstly used on the whole data set to identify the main populations. Subsequently we applied the same analyses on each of the main groups separately to look at substructuring on a finer scale, with 10 independent runs of 1,000,000 MCMC iterations following a burn-in of 250,000 steps for each K, and number of clusters (K) set from 1 to 7. The software STRUCTURE HARVESTER (Earl 2012) was used to plot the likelihood across the various values of K determined by STRUCTURE, using the mean likelihood L(K) and ΔK (Evanno et al. 2005). Population structure was also tested by model-independent multivariate analyses, carrying out a discriminant analyses of principal components (DAPC) with ADEGENET v1.2.8 in R (Jombart et al. 2010). Unlike STRUCTURE, the software ADEGENET does not make assumptions based on HWE or LD, and is not sensible to relatedness among individuals, being therefore less prone to produce clusters influenced by family groups. We also generated a neighbour-joining network tree with the program POPULATIONS v1.2.32 using Cavalli- Sforza and Edwards chord distance Dc (Cavalli-Sforza & Edwards, 1967).
To assess the levels of genetic differentiation between groups and populations considered, we carried out Fisher’s exact test, analogues of pairwise mean Fst (Weir & Cockerham 1984), and we tested for significance using analyses of molecular variance
among and within groups and populations (AMOVA, Michalakis & Excoffier 1996) with 1,000 permutations on ARLEQUIN 3.5. A Mantel test (Mantel 1967) was performed to evaluate isolation by distance within the main groups, using software GENEPOP 4.2 (Raymond & Rousset 1995; Rousset 2008) with 1,000 permutations for significance.
Results
In total 400 sables were genotyped for 57 autosomal polymorphic microsatellites. The preliminary results for population structure revealed five different geographically- clustered groups separated by well-defined physical boundaries, hereafter named Southern, Eastern, Zambian, West Tanzanian and Angolan (Fig. 1). On a finer scale and using a hierarchical approach, a total of 12 geographically defined subgroups were further identified (Fig. 2).
Fig. 1 – Population structure in pie charts and sample location for sable antelope. The size of each pie is scaled according to the number of individuals sampled at each locality. Pie chart colours depict different genetic clusters as identified by STRUCTURE, and the size of each pie slice shows the average probability of assignment of individuals to various clusters. (a) Representation of clustering analyses of 400 individuals performed by STRUCTURE, for which K=5 had the highest credibility. Vertical bars show the membership of each individual to a given cluster, and different colours correspond to separate clusters. (b) The shaded colour represent sable distribution in Africa assigned to five different geographically coherent populations, and according to the highest assignment probability, calculated with STRUCTURE.
Genetic diversity of sable antelope
Consistent deviations from HWE (P < 0.05 and P < 0.01) after Bonferroni correction, led to the exclusion of seven loci, while all 400 samples in the dataset were still kept below the threshold of 33% of missing data for the final 50 microsatellite markers. Subsequently we found deviations from HWE in nine of 250 tests, but in no case did one marker deviated from HWE on more than one main group (Supplementary Table S2). Significant linkage disequilibrium (P < 0.05) after Bonferroni correction was only observed for one pair of loci (HpN24 – HpN91) for more than one main sable group (results not shown).
Fig.2 – Population structure in pie charts and sample location for sable antelope for the main groups Southern, Eastern, Zambian and Angolan. The West Tanzania group was excluded because it showed no substructure. The size of each pie is scaled according to the number of individuals sampled at each locality. Pie chart colours depict different genetic clusters as identified by STRUCTURE, and the size of each pie slice shows the average probability of assignment of individuals to various clusters. Representation of clustering analyses performed by STRUCTURE, with the K value with the highest credibility for each subgroup. Vertical bars show the membership of each individual to a given cluster, and different colours correspond to separate clusters. The historical samples are signaled with a red diamond symbol.
We observed even values of genetic diversity among Southern, Eastern, Zambian and West Tanzanian sable populations, while the Angolan population consistently exhibited lower diversity for all parameters measured. Statistical T-Tests performed on the five main groups for the genetic diversity parameters were highly significant when involving the Angolan group, and always non-significant for all combinations comparing the
remaining groups (results not shown). The expected heterozygosity (He) ranged from 0.33 (Angolan) to 0.56 (Eastern), while the average number of alleles per locus (Na) ranged from 2.54 (Angolan) to 5.44 (Southern). Number of private alleles (Npa) and allelic richness (AR) were also lowest for the Angolan population (2 and 2.32 respectively) and highest for the Eastern group (26 and 5.05 respectively) (Table 1). The analyses carried out for the 12 subgroups revealed all indexes of genetic diversity to be lowest for the Cangandala population in Angola. The highest He was found on the Mid- Southern population, while Na, Npa and AR scored highest for the West Tanzanian sables for which no subpopulations were observed (Table 1). The results for the inbreeding coefficient Fis were only significant for the subpopulations Mid-Eastern (0.071; P≤0.01), West Zambia (0.038; P≤0.05) and West Tanzanian group (0.031; P≤0.05).
Table 1 - Measures of genetic diversity by main group and subgroups of sable
N He s.d. Fis Na s.d. Npa AR Southern Sable 104 0.500 0.265 - 5.44 3.76 22 4.33 West-Southern 41 0.438 0.260 0.002 3.68 2.08 1 2.78 Mid-Southern 48 0.502 0.257 0.019 4.76 3.11 7 3.25 East-Southern 15 0.465 0.282 0.035 3.58 2.08 2 3.01 Eastern Sable 37 0.562 0.287 - 5.52 3.26 26 5.05 North-Eastern 9 0.453 0.285 0.052 2.80 1.44 2 2.66 Mid-Eastern 20 0.551 0.290 0.071** 4.78 2.71 16 3.67 West-Eastern 8 0.475 0.298 -0.123 3.24 1.66 3 - Zambian Sable 112 0.534 0.254 - 5.52 3.80 14 4.55 Central Zambia 13 0.412 0.258 -0.000 2.66 1.30 1 2.44 South Zambia 38 0.518 0.262 -0.015 4.64 2.68 5 3.38 West Zambia 61 0.527 0.254 0.038* 5.06 3.30 2 3.36
West Tanzanian Sable 89 0.532 0.239 0.031* 5.24 2.83 27 4.33
Angolan Sable 58 0.326 0.260 - 2.54 1.33 2 2.32
Cangandala 9 0.237 0.235 -0.088 1.80 0.76 0 1.70
Luando 49 0.322 0.264 -0.006 2.50 1.34 1 2.03
N, sample size; He, expected heterozygosity; s.d., standard deviation; Fis, inbreeding coefficient with *P ≤ 0.05 and ** P