Map showing the location of the 16 lion populations included in the analysis. In the legend, the composition of the datasets and the number of included microsatellite loci is indicated. Lion range data from IUCN (2014). Region definitions from IUCN SSC Cat Specialist Group (2006a; b). doi:10.1371/journal.pone.0137975.g001 

Contexts in source publication

Context 1

... on mtDNA, the current taxonomic division is challenged [2 – 4]. However, mtDNA is a single, non recombining locus in the maternal lineage and does not permit the detection of admixture events and sorting at multiple loci as may occur in autosomal markers. Therefore, the observed pattern in mtDNA data may not adequately depict the underlying genetic complexity. Autosomal data are needed to corrob- orate the topology based on mtDNA, since conflicting patterns between phylogenies based on mtDNA and phylogenies based on autosomal markers have been described in several other species [24 – 29]. Most commonly a monophyletic pattern is detected in the mtDNA, but is not supported, or is contradicted, by phylogenies based on autosomal loci. This is often explained by incomplete lineage sorting, as coalescence time in mtDNA is four times shorter than in autosomal markers. Since lineage sorting during the process of coalescence has a random nature, this could also lead to an ‘ incorrect ’ gene tree by mtDNA markers if populations ’ divergences were closely spaced in time. Female philopatry is another strong contributing factor in mtDNA trees. As gene flow in lions is biased towards the male sex [30,31], gene trees based on autosomal markers may show less discrete groups. This argument has been used by Antunes et al . (2008) to explain incongruent patterns in their lion data based on mtDNA and autosomal markers. Taxonomic revisions have potentially far-reaching ramifications with regard to management (e.g., CITES, USFWS, IUCN) and, therefore, should be approached cautiously. Ideally, proposed revisions should be supported by a combination of biogeographic, mtDNA and autosomal DNA, and morphological data. In this study, we analyzed 20 microsatellite loci for lions from thirteen wild populations, one of which is located in West Africa (Benin) and four in Central Africa (Chad, DRC and two from Cameroon). Furthermore, we included microsatellite data from another West African population in Senegal and from two distinct zoo populations of Ethiopian lions representing the region where the two major genetic lineages (i.e., West/Central Africa and East/Southern Africa) may connect. To compare the phylogenetic clusters derived from the microsatellite data and to check for congruence with previously published patterns, we included data from 1,454 base pairs (bp) of the mitochondrial DNA for each sampling location. Using this approach, we are aiming to contribute to the ongoing discussion about lion taxonomy by answering four questions: 1) Do autosomal data support previously described phylogenetic groupings in the lion in general and the distinct position of the West/Central African lion in particular?, 2) Can an effect of sex-biased gene flow be detected?, 3) How genetically distinct are the sampled populations, at both the continental and regional scales, and do levels of genetic diversity vary amongst regional subdivisions, with a special focus on West/Central Africa? and 4) Are there signs for reduced genetic diversity in particular lion populations with an emphasis on West/Central Africa? Our study is the first to include multiple lion populations from West/Central Africa, using both autosomal and mtDNA markers in a phylogenetic con- text covering the entire current geographic range of the lion. We processed a total of 48 samples from eight populations, including one population from West Africa (Benin), four populations from Central Africa (two from Cameroon, one from Chad and one from DRC), two populations from East Africa (Ethiopia2 (captive) and Kenya) and one population from Southern Africa (Zambia). Except for Ethiopia2, all samples origi- nated from free-ranging lions, with no known history of anthropogenic introductions of lions from other populations. Samples were collected in full compliance with specific permits (CITES and permits related to national legislation in the countries of origin). Details on permits, sample storage, DNA extraction, polymerase chain reaction (PCR) amplification, fragment analysis and quality control are given in S1 File. See S1 Table and S2 Table for used loci and primer information. All microsatellite allele length data are given in S2 File. Generated microsatellite data were supplemented by published data for the same 20 loci from another six populations [32], together summarized as Dataset 1. Dataset 2 [12] consists of all 15 samples from Ethiopia1 (captive) with ten analyzed loci, of which six are overlapping with our dataset. For two samples from Ethiopia1, all 20 microsatellites were analyzed and added to Dataset 1. Dataset 3 (Panthera/AMNH) contains microsatellite data from 12 loci for seven lions from Senegal, which could not be resized to Dataset 1 and were therefore only included for calculation of diversity indices and bottleneck statistics (for details on permits and the processing of Senegal samples, see S3 File). An overview of datasets used in each analysis is provided in Fig 1 and Table 1. STRUCTURE 2.3.3 [33] was used for assessing population structure in Dataset 1 with unknown loci scored as missing data. Simulations were run assuming the admixture model with correlated allele frequencies. Ten runs were performed for K = 1 to K = 8, using 10,000,000 permutations and a burn-in period of 1,000,000. To check the assignment of Ethio- pia1 to any of the clusters identified by STRUCTURE, we included the two Ethiopian samples for all 20 microsatellites. Structure Harvester [34] was used to determine the most likely number of clusters, following the Δ K method [35]. CLUMPP was used to combine replicate runs and avoid label switching [36]. Clustering of individuals was further assessed by performing Principal Component Analysis (PCA) in GenAlEx 6.501 [37]. A neighbour-joining tree was created based on D A distance in POPTREE2 using 1,000 bootstraps [38]. For each sampling location, a mitochondrial region of 1,454 bp that encompassed cyto- chrome B (cytB), tRNAThr, tRNAPro and part of the control region was included for a number of individuals (Table 1). Details on polymerase chain reaction (PCR) amplification and sequencing are given in S1 File. Sequences were deposited in GenBank and supplemented by sequences previously published by Bertola et al . (2011) (see S4 File for sequence data and S3 Table for accession numbers). Variable sites and nucleotide diversity were calculated using ARLEQUIN 3.5 [39]. For phylogenetic analysis, a haplotype network was created using the median-joining algorithm in Network 4.6.1.1 (www.fluxus-engineering.com). A repeat region of cytosines of variable length was excluded due to unknown homology (positions 1382 – 1393) and all remaining characters were included with equal weighting. For AMOVA of Dataset 1, individuals for which all 20 loci were analyzed were included as either 1) without an indicated substructure (as all 1 group), 2) following IUCN classification (Africa; Asia), 3) following a North/South division as was indicated from the haplotype network, or 4) using the four groups identified by STRUCTURE (West/Central Africa; East Africa; Southern Africa; India). Isolation By Distance (IBD) was assessed by correlating geographic to genetic distances and using a Mantel ’ s permutation test with 999 permutations, as imple- mented in GenAlEx 6.501 [37]. In addition, AMOVA and IBD analysis were performed on a regional level, using the regions as indicated above (Africa; North; South; West/Central Africa; East Africa; Southern Africa). Pairwise F ST and Nei ’ s genetic distances were computed with GenAlEx 6.501 [37] for microsatellite data and with ARLEEQUIN 3.5 for mtDNA data [39]. The average number of alleles per locus (Na) was calculated using ARLEQUIN 3.5 [39]. Pri- vate allelic richness (A p ) was calculated with HP-Rare 1.1 [40] including statistical rarefaction to compensate for different sample sizes. GenAlEx 6.501 [37] was used to calculate observed (Ho) and unbiased expected heterozygosity (uHe) [41]. To obtain insights into the risk of emergent inbreeding, F IS per population was calculated in FSTAT [42] and the occurrence of recent bottlenecks was evaluated by using the program Bottleneck [43,44]. The Bottleneck test is based on the theory that during a bottleneck the allele numbers are reduced faster than the heterozygosity, leading to an excess of heterozygosity compared to the expected heterozygosity under the mutation-drift equilibrium. The program was run for 10,000 iterations, using the stepwise mutation model (SMM). Significant (P < 0.05) results from the Wilcoxon signed-rank test were scored, as this test proved to be the most powerful and robust when used with few ( < 20) polymorphic loci ...

Context 2

... and describing patterns of mitochondrial (mtDNA) and nuclear genetic variation is a crucial component to fully understanding the evolutionary history of a species. High quality phylogeographic data that represent the underlying genetic complexity are important for taxonomy and contribute to designing effective conservation strategies. This is of particular importance for species such as the lion ( Panthera leo ) that occupy large geographic ranges within which disjunct populations may not allow for natural dispersal and gene flow. Increas- ing habitat fragmentation and variable anthropogenic factors have created a growing need to manage lions at the population level [1]. In addition, several recent publications have sparked the discussion whether the current taxonomic nomenclature for the lion is justified [2 – 4]. Two subspecies of lion are officially recognized by the IUCN, based on genetic data [5,6]: the African lion ( Panthera leo leo ), ranging throughout sub-Saharan Africa with the exception of dense rain forest, and the Asiatic lion ( Panthera leo persica ), which exists as a single population in the Gir forest, India. Although all African lion populations are considered as belonging to the African subspecies ( P . l . leo ), distinct subgroups have been recognized based on mor- phology [7,8] and genetics [2 – 5,9 – 12]. Analyses of morphometric data has led to the distinction of at least three extant clades ( “ subspecies ” ) on the African continent [7]. Lions from the northern part of their range further showed a relatively close relationship to the Asiatic subspecies [7,8]. This pattern was confirmed by phylogenetic analysis of mitochondrial haplotypes only, based on which lions in West/Central Africa were described as a genetically distinct group with a relatively close genetic relationship to the Asiatic subspecies [2 – 4] (region definitions from [13,14], see Fig 1). The genetic dichotomy that separates the West/Central African lion populations from East and Southern African populations has also been found in other large mammal species and is often reflected in their taxonomy including African buffalo ( Syncerus caffer ) [15,16], roan antelope ( Hippotragus equinus ) [17], hartebeest ( Alcelaphus busela- phus ) [18,19], giraffe ( Giraffa camelopardalis ) [20,21] and cheetah ( Acinonyx jubatus ) [22,23]. Due to the genetic differentiation within the African lion and the nested position of the Asiatic lion subspecies within the West/Central Africa clade based on mtDNA, the current taxonomic division is challenged [2 – 4]. However, mtDNA is a single, non recombining locus in the maternal lineage and does not permit the detection of admixture events and sorting at multiple loci as may occur in autosomal markers. Therefore, the observed pattern in mtDNA data may not adequately depict the underlying genetic complexity. Autosomal data are needed to corrob- orate the topology based on mtDNA, since conflicting patterns between phylogenies based on mtDNA and phylogenies based on autosomal markers have been described in several other species [24 – 29]. Most commonly a monophyletic pattern is detected in the mtDNA, but is not supported, or is contradicted, by phylogenies based on autosomal loci. This is often explained by incomplete lineage sorting, as coalescence time in mtDNA is four times shorter than in autosomal markers. Since lineage sorting during the process of coalescence has a random nature, this could also lead to an ‘ incorrect ’ gene tree by mtDNA markers if populations ’ divergences were closely spaced in time. Female philopatry is another strong contributing factor in mtDNA trees. As gene flow in lions is biased towards the male sex [30,31], gene trees based on autosomal markers may show less discrete groups. This argument has been used by Antunes et al . (2008) to explain incongruent patterns in their lion data based on mtDNA and autosomal markers. Taxonomic revisions have potentially far-reaching ramifications with regard to management (e.g., CITES, USFWS, IUCN) and, therefore, should be approached cautiously. Ideally, proposed revisions should be supported by a combination of biogeographic, mtDNA and autosomal DNA, and morphological data. In this study, we analyzed 20 microsatellite loci for lions from thirteen wild populations, one of which is located in West Africa (Benin) and four in Central Africa (Chad, DRC and two from Cameroon). Furthermore, we included microsatellite data from another West African population in Senegal and from two distinct zoo populations of Ethiopian lions representing the region where the two major genetic lineages (i.e., West/Central Africa and East/Southern Africa) may connect. To compare the phylogenetic clusters derived from the microsatellite data and to check for congruence with previously published patterns, we included data from 1,454 base pairs (bp) of the mitochondrial DNA for each sampling location. Using this approach, we are aiming to contribute to the ongoing discussion about lion taxonomy by answering four questions: 1) Do autosomal data support previously described phylogenetic groupings in the lion in general and the distinct position of the West/Central African lion in particular?, 2) Can an effect of sex-biased gene flow be detected?, 3) How genetically distinct are the sampled populations, at both the continental and regional scales, and do levels of genetic diversity vary amongst regional subdivisions, with a special focus on West/Central Africa? and 4) Are there signs for reduced genetic diversity in particular lion populations with an emphasis on West/Central Africa? Our study is the first to include multiple lion populations from West/Central Africa, using both autosomal and mtDNA markers in a phylogenetic con- text covering the entire current geographic range of the lion. We processed a total of 48 samples from eight populations, including one population from West Africa (Benin), four populations from Central Africa (two from Cameroon, one from Chad and one from DRC), two populations from East Africa (Ethiopia2 (captive) and Kenya) and one population from Southern Africa (Zambia). Except for Ethiopia2, all samples origi- nated from free-ranging lions, with no known history of anthropogenic introductions of lions from other populations. Samples were collected in full compliance with specific permits (CITES and permits related to national legislation in the countries of origin). Details on permits, sample storage, DNA extraction, polymerase chain reaction (PCR) amplification, fragment analysis and quality control are given in S1 File. See S1 Table and S2 Table for used loci and primer information. All microsatellite allele length data are given in S2 File. Generated microsatellite data were supplemented by published data for the same 20 loci from another six populations [32], together summarized as ...