In this study, we sequenced the first complete mitogenomes of African leopards. They originated from South Africa, where, according to our data, two unique clades occur. The study aimed to reveal how South African leopards fit into continental patterns of genetic differentiation and examine how the climatic history of South Africa may have shaped phylogeographic patterns. Our results show a deep divergence between the two clades, which was dated back to approximately 800.000 years ago. We hypothesize that these clades originated during the climatically unstable Mid-Pleistocene, as seen in other savanna-dwelling mammals [24–26, 51, 53]. Thus, the genetic clades seen in Mpumalanga provinces are not the result from a regional, historical dispersal barrier, per [23], but instead remnants of past climatic events that took place on a larger, continental scale [11]. As such, leopards reflect the unique climatic history of South Africa, which has resulted in eminent and endemic genetic diversity. Genomic nuclear data from our study region are needed to confirm our phylogenetic hypotheses.
This is the first genomic study on African leopards that confirms the existence of two deeply diverged mitochondrial clades. The Southern African clade (SA) is restricted to South Africa and Mozambique, whereas the central African clade (CA) extends far into West-Central and East Africa [11]. The mitogenomes were reconstructed using Bayesian and Maximum Likelihood analysis, showing independent evolutionary histories for each lineage. Thus, the admixture in Mpumalanga illustrates that the separately evolved clades have since come into secondary contact. The two clades of African leopard have previously been found by Uphyrkina et al. [17] and Ropiquet et al. [14], using mitochondrial sequences. Interestingly, these diverged lineages were not resolved using neutral nuclear data. Nevertheless, a few neutral microsatellite markers are not representative of genome-wide variation, which is subject to natural selection [54]. One study that applied a whole genome approach to study leopard phylogeographic structure has not yet included South African leopards [12], whereas another genomic study only included one historic sample from Durban, which did illustrate a particularly high heterozygosity [2].
Based on the calculated divergence time, the SA clade likely evolved due to climatic shifts during the Mid-Pleistocene, which lasted from 1.6 to 0.52 mya. This supports previous studies on genetic diversification worldwide [55, 56]. In sub-Saharan Africa, the Pleistocene was marked by cool and dry cycles alternating with warm and wet cycles, which drove repeated expansions and contractions of savanna grasslands [57]. The Pleistocene refuge theory states that maintaining habitat patches in South Africa enabled species persistence through repeated ice ages and promoted the divergence between populations (Arctander et al., 1999). Southwest Botswana and northwest South Africa underwent a period of intense aridification from about 2.5 to 1.7 mya, during which sand dunes were repositioned across the Makgadikgadi Palaeo-lake [58]. We expect that this caused the historical isolation of the Kalahari-Namibian-Cape region, which has been well-documented in other African taxa [25, 26, 59]. Although leopards are extremely versatile habitat generalists with large ranging habits [2] (Paijmans et al., 2018), their prey have a more patchy distribution due to specific habitat requirements, such as savanna woodlands and the availability of permanent water sources [53]. Several studies have documented similar spatial genetic patterns among African ungulates as a result of Pleistocene climatic shifts (see [60–63].
The coalescence time between African and Asian leopard mitochondrial lineages was ~ 0.9 mya, similar to previous mitochondrial studies [17, 19]. When comparing South African leopard mitogenomes with Asian conspecifics, it became evident that African leopards exhibit a level of genetic differentiation comparable to, or even more extensive than, Asian subspecies, which has previously been suggested [11]. Because a subspecies is traditionally defined as a “phenotypically distinct and discrete geographic population that differs in heritable genetic variation” [64], we consider the diverged lineages not to represent subspecies. Leopards within the Mpumalanga province (whether part of the SA or CA clade) are not phenotypically distinct. They are admixed across a sizeable hybrid zone that extends well into Mozambique and most likely across the Northern Cape [26]. Nevertheless, the delimitation of a subspecies is still widely debated, and it is now acknowledged that subspecies can live in sympatry with interspecific hybridization [65]. Our data mainly highlights that leopards did not maintain low population differentiation and lack of migration barriers across Africa during the Pleistocene, as suggested by Paijmans et al. [19] and Pečnerová et al. [12], but contain deeply diverged lineages. The divergence between the SA and CA was calculated to be 2%, which is higher than what was measured for the Asian subspecies (1.1%). Therefore, it is justified to refer to the South African leopard as a unique genetic entity, sometimes referred to as an evolutionary significant unit [66].
That we did not see clear a subspecies distinction in Asian leopards using mitogenomes can mean one or two, not mutually exclusive, things. First, the NCBI Genbank database does not always provide a sample location or subspecies delineation, which means some samples may have been mislabelled. Secondly, the two subspecies occurring in northern China are not as distinct as we considered them to be. The latter would be supported by the minor differentiation among any of the Asian samples, regardless of their origin, and by the clustering of subspecies that were clearly labelled as belonging to either the North-Chinese leopard P. p. japonensis or Amur leopard P. p. orientalis. The first study on phylogeographic structure in leopards [67] already assigned these to the same East Asian group, and proposed that only leopards in Central Asia are distinct (P. p. saxicolor and P. p. sindica). They mentioned that further sampling is required for subspecies recognition in this region. The genetic study that followed [17] has suggested subspecies status of North-Chinese leopards from combined NADH-5 (611 bp) and control region (116 bp) sequences. However, the latter study also concluded that they could not be significantly differentiated using microsatellites. Our mitogenomic data seems to confirm this. More exhaustive mitogenome sampling would be required for conclusive insights.
That leopards in Northern China and the Amur region show lower divergence than previously assumed might be beneficial to their conservation. Even though conservation management would normally allow for the protection of historical differences among unique lineages, as well as adaptive differences that may be present across the species range [68], there has been an increasing acknowledgment that hybridization could also benefit species when their conservation status is dire [69–71]. For instance, introgression from other subspecies can be beneficial when it increases the genetic variability of populations that suffer from a severe bottleneck or inbreeding depression, which is the case for the Endangered Amur leopard and Near-Threatened North-Chinese leopard [72]. For the Amur leopard, it is known that only two haplotypes occur, which only differ by a point mutation [16]. For future reintroduction, admixture between the presumed subspecies could be considered to allow for increased genetic diversity and novel adaptive variation [73], which might be the only tool left to revive these rare leopards.
When looking at other historical samples, one unexpected result in our phylogenetic analysis is that SA leopards cluster with Burundi in Central-East Africa and Algeria in Northern Africa. With regards to Algeria, no ecological scenario will be able to explain this observation, which leads us to look into other directions, such as sequence errors, possible DNA contamination, and the authenticity of the specimens. For instance, leopard skins have been traded for several decennia, which might have resulted in discrepancies in museum collections [74]. The sample originated from the Museum of Denmark and was collected in 1850. When mapped to the reference genome, 1514 base pairs were missing, and 131 base pairs were unambiguous, which is indicative of some level of DNA degradation. Although basic steps were taken to avoid sequencing artifacts, such as 2X serial captures, we cannot verify whether contamination has taken place, which is common in museum collections [48]. As for Burundi, the authors concluded that this sample's mitochondrial diversity and basal position suggest that Central-East Africa is the potential origin of leopards [2]. It is, therefore, highly interesting that this sample clusters with one of our samples (Mpu1) from the SA clade. Considering the presence of leopards for at least 2 million years in East Africa, it is possible that the SA lineage was once part of this population and split afterward, with the loss of intermediate lineages [2]. Further research is required, preferably using modern samples from areas surrounding Burundi. More importantly, we still consider our results strong enough to support the existence of the Southern African clade.
Our data on African leopards illustrates that unique evolutionary histories can be maintained over large geographic timescales in the face of gene flow. Speciation has long been assumed to require geographic barriers to gene flow, but there has been growing recognition that adaptive radiation can occur under sympatric conditions as well [75]. We expect that hybridization of the African leopard clades has been occurring since the Holocene, which is likely to have caused significant homogenization in the nuclear genome [76]. Nevertheless, every genome contains some ‘barrier loci’ that seem to resist the homogenizing effects of gene flow [77], due to which we still expect to find remnants of the CA and SA division when a whole genome approach is applied. This would be particularly interesting for the Cape leopard, which is morphologically distinct from the rest of South Africa [22]. Although phylogenetic clades may have originated in response to past climatic events and temporal fluctuations in population size, leopards can cross hundreds of kilometres through unsuitable habitat and human-modified landscapes [78], and dispersal is currently considered to occur widely across southern Africa [79]. Therefore, despite their distinct evolutionary histories, we consider the clades to be part of the same, interconnected metapopulation, which do not require separate conservation approaches [79].
In conclusion, our results show the unique history of South African leopards, and support the hypothesis of a southern glacial refuge during the Pleistocene, after which they spread northeast during the Holocene and came into secondary contact with the Central African leopard clade. Although the mitochondrial clades are deeply diverged and date back to approximately 800.000 years ago, a whole genome approach is advised to confirm this hypothesis.