Range-wide phylogeographic analyses of leopards (Panthera pardus) reveal African mito-nuclear discordance and previously unrecognized diversity

Range-wide phylogeographic analyses of leopards (Panthera pardus) reveal African mito-nuclear discordance and previously unrecognized diversity | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 16 March 2026 V1 Latest version Share on Range-wide phylogeographic analyses of leopards (Panthera pardus) reveal African mito-nuclear discordance and previously unrecognized diversity Authors : Danielle Del Castillo 0009-0004-8646-921X , Corey Anco , Seth Cunningham , Alexis Neffinger , Faruk Mamugy , Arame Ndiaye 0000-0001-9403-9130 , Hana Raza 0000-0003-1482-1718 , Kamta Tchoffo Roméo Omer , Evon Hekkala , and Laura D. Bertola [email protected] Authors Info & Affiliations https://doi.org/10.22541/au.177366294.41173847/v1 230 views 75 downloads Contents Abstract Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Leopards, as the most widely distributed felids, remain phylogeographically understudied due to sampling gaps across its vast range distribution. In this study, we assembled the most comprehensive genomic dataset to date, spanning all the subspecies (n=198 from archival tissues, field collections and published data). Mitochondrial genomes resolve seven Asian subspecies plus three African haplogroups (West, Central and South), while nuclear genomes reveal admixtures with weak structure (FST <0.05) highlighting 30-75% of mito-nuclear discordance likely driven by ILS. Moreover, we identify in Africa a Congo Basin diversity hotspot (Par II) and three sub-Saharan lineages (Par I, II and III). These data can be further used as a genomic reference baseline, enabling forensics to trace origin of illegal trade of leopard body parts seizures (e.g. for the west African haplogroup). This is critical as leopards face 70% historical range loss and escalating CITES Appendix I seizures. Thus, resolving marker-type discrepancies is essential for effective conservation action plans particularly when facing ongoing species decline. Introduction: The leopard ( Panthera pardus ) is the most widely distributed species of the Felidae family , currently ranging across Africa, the Middle East, into far East Asia (1–3). Insights into the spatial distribution of genetic lineages can aid conservation efforts, lend a backbone for effective planning of protected areas and migration corridors, identify appropriate populations for wildlife translocations, and trace the origins of illegally harvested parts for trade. Although several studies have characterized leopard genetic diversity and distribution on a more local scale (4–15), few studies have tackled this on a range-wide scale. Attempts to classify leopard diversity on a broader geographic scale (16–19) are often limited in the number of localities, which creates sampling gaps and can lead to spurious results when delineating genetic lineages. Studies with more comprehensive sampling have relied on a small number of SNPs (20), microsatellites or mitochondrial DNA (17,21,22), which provide only a limited view of genome-wide diversity, and, in the case of mitochondrial DNA, could even misrepresent population structure (23). Both aspects, limited sampling and reliance on small and non-representative datasets, can therefore confound classification of conservation units. To date, a comprehensive analysis of the leopards’ overall diversity and spatial distribution of genomic patterns that combines denser sampling across their entire geographic range is missing. The first broad scale genetic studies, based on mitochondrial and microsatellite loci, suggested genetic population subdivisions across eight Asian subspecies and one contiguous African subspecies, P. p. pardus (16,17). More recently, studies including whole genome data and broader sampling efforts have corroborated these initial works (18,19). Asian lineages show strong patterns of differentiation and relatively low levels of diversity, whereas their African counterparts exhibit nearly the opposite genomic composition, with weak population structure and high diversity (18,19). This weak population structure on the African continent is in strong contrast with patterns from other co-distributed taxa, which are thought to be driven by historical expansions and contractions of forest and desert biomes, thereby temporarily reducing suitable habitat for some species, leading to differentiation between refugial populations (19,24). The presumed continent-scale connectivity in African leopards was attributed to their high adaptability to diverse habitats, their versatility as predators, and the lack of migration barriers given their ecology. This may have been further influenced by the continuously large effective population size (sustained through Pleistocene climate and habitat fluctuations), also reflected by high levels of diversity relative to other members of the genus Panthera (19). In contrast with the weak population structure observed in nuclear genomic data within Africa, analysis of the mitochondrial locus NADH dehydrogenase 5 (NADH5), supported five geographically distinct haplogroups across sub-Saharan Africa, where F ST calculations between haplogroups rivaled and, in some cases, exceeded those used for the subspecies classifications in Asia (21). These results were used to delineate three distinct African clades (21) and challenge the view that leopards in Africa should be treated as one continent-wide panmictic unit. A similar study on the NADH5 locus found divergence of two lineages, one spanning the majority of the continent, the other restricted to the southernmost portion of Africa (22). These data suggest that signatures of population structure do not align across marker types, and that mitochondrial-nuclear (mito-nuclear) discordance may be common, at least in parts of the leopards’ range. Mitochondrial DNA markers have been widely used for investigating phylogeographic patterns due to their ease of use, even in low quality samples, however it can yield misleading results. Principally, in the absence of recombination, the entire mitogenome can only be treated as one locus, which can lead to misleading inferences due to stochastic lineage sorting, introgression, or selective sweeps (23,25–27). Specifically, patterns of admixture are less straightforward to detect, in species with female philopatry, population structure is overestimated, and due to faster coalescence times, mitochondrial DNA has the tendency to oversplit and overestimate divergence times. For this reason, genomic data are needed to corroborate inferences based on mitochondrial data only. Further, mito-nuclear discordance is known to result in misidentified lineages when only mitochondrial data are considered, which potentially impacts conservation actions, as was highlighted for the African elephant ( Loxodonta africana ) (28). In addition to the identification of spatial patterns of mito-nuclear discordance, it is equally important to investigate the drivers of discordance (29,30). There are several explanations, biologically relevant to leopards, for why mito-nuclear discordance occurs. The simplest stems from incomplete lineage sorting (ILS), whereby lineage sorting in the nuclear genome has not caught up to differentiation established within the mitochondrial genome over the same period of time. Because their effective population size for mitochondrial loci is four times smaller than for nuclear genomes (due to the nature of haploid mitogenomes versus diploid nuclear genomes), mitogenomes will sort at a rate four times faster than nuclear genomes (29,31,32). Demographic and behavioral asymmetries may also account for mito-nuclear discordance. For example, sex-biased dispersal may contribute to dispersal of the nuclear genome in situations where there is no corresponding movement of mitochondrial DNA (or vice versa) (32). Toews & Brelsford, 2012 (29) suggest that mitochondrial structure, in the absence of nuclear, can be attributed to male-biased gene flow, the primary means of dispersal in leopards (3). We can disentangle ILS from male-biased dispersal (although these are non-mutually exclusive), by observing patterns of the geographic distribution of nuclear partitioning, as compared to the range extent of the mitochondrial haplotype distribution of the same individuals (28,29). ILS does not generate predictable distributions of mitochondrial haplotypes versus nuclear admixture components across space. Therefore, if ILS is driving the discordance, we will observe a continuous and random distribution of nuclear population groups across or outside of the extent of the mitochondrial distribution (29). However, if male-biased dispersal is the primary driver of the discordance, we expect patterns of overlap within nuclear genomic population partitions across the extent of mitochondrial distribution (29). Here, we improve on existing knowledge by obtaining new samples from leopard range states, mostly by leveraging natural history collections, and building upon the work of Paijmans et al., 2021 (18), Pečnerová et al., 2021 (19), and Tensen et al., 2024 (15) to generate the most comprehensive geographic study on leopards to date. Intersection and re-analyses of these datasets gives a much more complete picture of population genetic relationships between leopard populations throughout their entire range. We further explore how mitochondrial genomes (mitogenomes) compare with genome-wide nuclear patterns of diversity. As mitochondrial markers are still an important tool for wildlife and forensic studies, in particular in the Global South, having an in-depth comparison will allow mitochondrial data to be further used and results to be correctly interpreted. Indeed, despite its limitations, mitochondrial DNA remains invaluable in various areas of focus such as wildlife forensics for species identification or tracing the origin of confiscated leopard body parts (skins, claws, bones…) in illegal wildlife trade. As such, genomic resources covering as much of the leopard’s range as possible, and an in-depth comparison between marker types will pave the way for the implementation of a robust validated reference database. Based on existing studies showing different patterns across marker types, suggesting that mito-nuclear discordance exists within their range, this in-depth comparison is of particular relevance for leopards, as discordant marker signatures can lead to incorrect inferences regarding population structure and assignment, with disastrous management decisions. Sampling and DNA extraction Archival museum specimens were collected per approved destructive sampling proposals submitted to the American Museum of Natural History (AMNH), New York, NY, USA; the Field Museum (FMNH), Chicago, IL, USA; and Le Muséum National d’Histoire Naturelle (MNHN), Paris, France. Bone and tissue fragments (< 0.5 cm in length) were excised with sterile tools from one hundred sixteen (n = 116) archival P. pardus samples housed in the department of Mammalogy within the division of Vertebrate Zoology collections at the AMNH; the Gantz Family Collection Center Mammal Collections, housed at FMNH; and the Mammals Collection housed at MNHN. Sample origin and locality data were recorded from the specimen tag and archived field notes for those samples of which the data was available (Table A1). Through field collection, four (n = 4) fecal samples (with recorded locality data), one each from Nigeria, Cameroon, Sierra Leone, and Cote D’Ivoire were collected in compliance with local and international legislation. Additionally, raw genomic data were downloaded from Paijmans et al., 2021 (18) (n = 19; ex situ individuals were excluded), Pečnerová et al., 2021 (19) (n = 50), and Tensen et al., 2024 (15) (n = 9) to supplement the above data set (Table A1). The total data set comprises one hundred ninety eight (n = 198) individuals across the entire geographic distribution of leopards (Figure 1, Table A1). To maintain the integrity of the samples, all specimens and DNA extractions were handled and processed in PCR (polymerase chain reaction)-free rooms with protocols in place (sterile bench procedures, UV sterilized rooms) for protection against contamination of the low-quality, low-quantity, and/or degraded DNA common in archival samples. Archival samples were submerged in up to 500µL of 1x phosphate buffered saline (PBS) and incubated at room temperature for 48 hours prior to isolation of DNA. Samples were then digested in a solution of 180µL of Qiagen ATL Buffer (Qiagen; Hilden, Germany) and 20µL of proteinase K at 55°C for 48 hours. An additional 20µL of proteinase K was added after the first 24 hours of incubation. Following digestion, DNA was extracted and purification of DNA was executed using the Qiagen MinElute PCR Purification Kit (Qiagen; Hilden, Germany) with the following modifications to the manufacturer’s protocol: DNA extract was incubated in 5mL of PB binding buffer for five minutes and filtered through MinElute spin column tubes, then washed twice with 750µL of PE buffer, followed by elution with two subsequent 30µL volumes of EB buffer (pre-heated to 55°C). Fecal DNA extractions followed the QiaAmp Fast DNA Stool Mini Kit (Qiagen; Hilden, Germany) protocol. DNA extracts were quantified using a Qubit fluorometer (Life Technologies, Inc.). Library Preparation Illumina libraries were prepared using NEBNext Ultra II Library Prep with Sample Purification Beads kit (New England Biolabs; Ipswich, MA, USA) following manufacturer’s protocols, with the following modifications: adaptors were diluted 1:20 with distilled H 2 O. Indexing PCR was performed using 2µL NEBNext Multiplex Unique Dual Index Oligos for Illumina (New England Biolabs; Ipswich, MA, USA), and 25µL NEB mastermix (from kit) per 23µL library sample under the following conditions: 98°C for 30 seconds; 12 cycles of {98°C for 10 seconds, 65°C for 30 seconds, 72°C for 30 seconds}; 72°C for 5 minutes; 10°C hold. Library enrichment PCR was performed in quadruplicate per sample using 0.5µL Phusion High-Fidelity DNA polymerase (New England Biolabs; Ipswich, MA, USA), 10µL buffer, 1µL dNTPs, 1.5µL dimethyl sulfoxide (DMSO), 1µL bovine serum albumin (BSA), 30µL distilled H2O, and 0.5µL each of primers IS5 and IS6 (33) per 5µL of sample under the following conditions: 95°C for 30 seconds; 12 cycles of {95°C for 10 seconds, 65°C for 30 seconds, 68°C for 30 seconds}; 10°C hold. Column purification was performed after library enrichment using Qiagen MinElute PCR purification kit (Qiagen; Hilden, Germany): library-prepped DNA extract was washed in 750µL of PB binding buffer through MinElute spin column tubes, then washed twice with 750µL of PE buffer, followed by elution with two subsequent 30µL volumes of EB buffer (pre-heated to 55°C). DNA concentrations of library preps were determined using a Qubit fluorometer (Life Technologies, Inc.) and Bioanalyzer 2100 (Agilent Technologies, Inc; Santa Clara, CA, USA). Shotgun Sequencing and Target Capture Library-prepped samples were processed in two ways : 1) shotgun sequencing, and 2) with target-bait hybridization and bead purification for the mitogenome with a commercially available leopard-specific mitochondrial baits set (mito-baits; Daicel Arbor Biosciences; Ann Arbor, MI, USA), following protocols from the myBaits Hybridization Capture Kit: Custom DNA-Seq or RNA-Seq Manual v5.03 (Daicel Arbor Biosciences; Ann Arbor, MI, USA). Samples were then pooled together (up to 33 samples per pool) into equimolar concentrations and pools were sequenced for 150bp paired-end read lengths on an Illumina HiSeq4000 (Illumina, Inc.; San Diego, CA, USA) at GENEWIZ (Azenta Life Sciences; South Plainfield, NJ, USA) facilities. Illumina raw, demultiplexed reads were assessed for quality using FastQC software (34), followed by screening for exogenous DNA using FastQScreen (35). Sequenced data, plus raw genome-wide data downloaded from Pečnerová et al., 2021 (19) and Paijmans et al., 2021 (18) were then adaptor-trimmed and filtered for quality using Skewer software (36). Filtering parameters included trimming bases with low quality (minimum quality > 30), removing degenerative reads (containing more than 15% ambiguous bases (“N”)), and discarding reads shorter than 40 bp after trimming. Filtered reads were mapped to the leopard genome (37), and the leopard mitogenome (38) using the Burrows-Wheeler Aligner software, BWA v0.7.17 (39), and to a nuclear mitochondrial insertions (numts) sequence which is known to occur as a tandem repeat in the Panthera genome, using the competitive mapping approach employed by Curry et al., 2021 (40). Samtools v1.5 (41) was used to sort and index reads, while retaining only those reads that mapped to the nuclear or mitochondrial leopard genomes. Duplicate reads from PCR error were filtered out using Picard v2.17.8 (Broad Institute, GitHub Repository) “MarkDuplicates” command. Following read mapping to the mitochondrial genome, whole mitochondrial genome sequences were generated for each sample using ANGSD (42) “doFasta” command, and samples with more than 50% missing or ambiguous bases (“N”) were removed from the data set. Following mapping of the nuclear genome, samples with less than 5 million reads were eliminated from the data set. We used ANGSD (42), commonly used for analyzing low coverage data, such as samples sourced from ancient, archival, or other degraded sources, to estimate genotype likelihoods for each individual and to identify variable sites (single nucleotide polymorphisms - SNPs) across individuals. Additional quality filtering was incorporated into this step (base quality >30, mapping quality >30). Intersecting and downsampling with published data To build upon previously generated whole-genome datasets (18,19), we intersected the variable positions from these published data with our shotgun data, exploring different levels of missing data (as obtained from the “-minInd” prompt) using principal component analyses (PCAs) integrated in ANGSD (42) using the “-doCov” prompt. As we consistently observed a batch effect with the data derived from Pečnerová et al. (2021) (19) (Figures A1-A2), we opted to include these data for mitochondrial analyses only. Only samples with less than 50% missingness in all variant sites were retained for downstream analyses. In total, four sets of filtered data were generated: one mitogenomic set (n = 147) and one nuclear (n = 82) for the entire leopard distribution; and one mitogenomic set (n = 121) and one nuclear (n = 57) for the African distribution (Table A2). We found intersection of our data with previously published data sources challenging, as we recovered minimal universal overlap of genomic sequence data across research groups (Table A4), thus interpreting relatively subtle structure from a smaller number of shared sites. For this reason, we took a subset of 30 of the highest quality in-house extracted sequences (containing the most reads per sample across geographic occurrences), and following the same procedures as described above, retained 16 individuals passing all filters (Table A2). This allowed us to verify the patterns of population structure based on the full dataset, but with higher genomic representation (more sites). Population Assignment and Analyses Mitochondrial genome To assess patterns of mitochondrial phylogeography, consensus sequences of the full mitochondrial genomes of each sample, plus an Amur leopard ( P. pardus orientalis ) and an African lion ( P. leo ) as outgroups (Genbank accessions KP202265.1 and NC028302.1, respectively), were aligned by the “muscle” algorithm using the R statistical software package “msa” (43). Aligned sequences were reviewed and filtered for missingness and ambiguities: samples with more than 15% of bases assigned as “N” or “-“ were removed from the alignment. The filtered alignment was used as input for generating haplotype networks using PopArt v1.7 (44), employing the median joining network algorithm with epsilon=0. Outliers, identified by particularly long branch lengths, were further examined employing the Basic Local Alignment Search Tool for nucleotides (nucleotide BLAST) available through the National Institute of Health’s (NIH) National Center for Biotechnology Information (NCBI) (45). Samples in which parts of the mitogenome were identified as a different species in the top hits were removed from the analysis. In the final haplotype network, populations were delineated based on visual assessment of clade arrangement. Hudson’s F ST (46,47) was calculated between populations using the R package “KRIS” (48). Additionally, Nei’s F ST (49) was calculated between populations using the R package “PopGenome” (50). To corroborate population delineation, an analysis of molecular variance (AMOVA) was performed using the R package “ade4” (51). Finally, a suite of summary statistics was calculated using the R package “PopGenome” (50), including Tajima’s D (52), Watterson’s theta (53,54), and pairwise nucleotide diversity. Sample F1444, an infant skull with associated tissue, was attributed as a leopard collected in Somalia in 1896, however, the data consistently aligned to P. leo in the mitochondrial analyses. Since lion cubs display prominent body spotting during early life stages, which can closely resemble leopard cub pelage, we suspect that this may have led to misidentification at the time of collection. Nuclear genome Admixture and population structure of the nuclear genome was inferred using the Bayesian clustering algorithm employed in NGSadmix (55) over 10 replicates for both SNP datasets (total distribution and African distribution). The best K was calculated by assessing Delta K across the replicates with the highest log likelihoods (Ln(K)) within the R statistical software base packages (56). Individuals were assigned to a clade based on their ancestry proportion. To evaluate differentiation between clades, delineated using Admixture proportions, Hudson’s F ST (46,47) was calculated using winsfs (57). PCAs were generated in R base software (56) from covariance matrices produced in ANGSD using the -doCov prompt (42). Winsfs (57) was also used to generate a suite of population genetics summary statistics from the site frequency spectrum (sfs) for each delineated clade, including Tajima’s D (52), heterozygosity (58), and Watterson’s theta (θ) (53,54). Effective population size (Ne) was estimated from dividing θ/μ, where μ = 1.0E10 -8 , the mutation rate employed by Pečnerová et al., 2021 (19). Mito-nuclear discordance To assess mito-nuclear discordance (29), genetic distance based trees were generated and compared using FastME v2.0 software (59) and ANGSD ngsDist (42) for the mitochondrial and nuclear pairwise genetic distances, respectively, followed by tree generation using the R packages “ape” (60) and “phytools” (61). The African lion ( Panthera leo ) was used as an outgroup (Genbank accession NC_028302.1 and SRR10764410 for the mitochondrial and nuclear outgroups, respectively). Range extents for inferred populations based on mitochondrial delineations were explored using minimum convex polygons around occurrence points for each population using the R package “adeHabitatHR” (62). Nuclear derived delineations for each occurrence were mapped onto the geographic range extent of mitochondrial assigned populations in sub-Saharan Africa, as described in Ishida et al., 2021 (28), using QGIS v3.34.2. For each individual, the mitochondrial haplotype assignment was plotted against the percentage of nuclear ancestry corresponding to each of K lineages across those same individuals. Results: Sequencing reads were generated with an average of 8,541,216 reads per sample, providing an average of 1.712x depth of coverage. An average of 653,985 reads per sample were recovered for the mitochondrial genome at an average depth of 883x. Mitochondrial Population Delineation Global Distribution An alignment for the total distribution of leopards (n = 147 samples, Table A2) consisted of 16,712 sites, 2,076 of which are polymorphic. The haplotype network (Figure 2) revealed haplogroups in Asia corroborating the geographic distribution of classified Asian subspecies (17,63). Populations on the African continent are diverged from the Asian subspecies, with the exception of P.p.nimr , which groups on the same lineage as populations in West Africa. Within Africa, we identify three major haplogroups: a wide-spread continental haplogroup (“Central”), a haplogroup occurring in the Gulf of Guinea (“West”), and one primarily covering Southern Africa (“South”) (Figure 2, Figure 3). F ST estimates highlight differentiation between African and Asian leopards (Nei’s F ST = 0.67652; Hudson’s F ST = 0.24154), as well as between all Asian subspecies and the three identified African haplogroups (Central/South: Nei’s F ST = 0.69588; Hudson’s F ST = 0.14241; Central/West: Nei’s F ST = 0.58206; Hudson’s F ST = 0.13591; and South/West: Nei’s F ST = 0.48519; Hudson’s F ST = 0.13736) (Table 1). Although Hudson’s F ST estimates have a lower range than the Nei’s estimates, patterns of differentiation between haplogroups remain proportionally consistent. An AMOVA comparing the pairwise Euclidean genetic distances between these lineages confirmed that there is greater Euclidean genetic distance between these groups than expected by chance (p=0.01, CI=95%). Nucleotide diversity (π) was 0.0115 across the total distribution and ranged from 0.00032 (Sri Lankan subspecies, P.p.kotiya ) to 0.001991 (Indian subspecies, P. p. fusca ) in Asia, and 0.002656 (Central African haplogroup) to 0.009217 (West African haplogroup) in Africa. Watterson’s theta (θ) was 330.962 for the total distribution and ranged from 5.333 (Sri Lankan subspecies, P.p.kotiya ) to 36.96 (Indian subspecies, P. p. fusca ) in Asia, and 105.622 (West African haplogroup) to 141.818 (Central African haplogroup) in Africa. Tajima’s D was -1.322 for the total distribution and ranged from -2.026 (Central African haplogroup) to 4.587 (West African haplogroup) (Table 1). For the two subspecies represented by a single sample, P. p. nimr and P. p. melas , diversity metrics could not be calculated. African Distribution A mitochondrial alignment of n = 121 samples (Table A2) was generated for the African distribution of leopards, composed of 16,721 sites, where 1,948 of these sites were polymorphic. A haplotype network showed that the previously mentioned widespread “Central” haplogroup consists of four closely related and spatially overlapping clades (East1, East2, Sudano-Guinean (SG), and Northeast (NEast)) (Figure 3). F ST estimation for all six minor lineages are in Table A3, ranging from 0.35145 to 0.52055 for Nei’s F ST ; and -0.0044 to 0.03611 for Hudson’s F ST compared across the four “Central” haplogroups, East1, East2, SG, and NEast. While the “West” haplogroup ranged from 0.48518 to 0.64131 for Nei’s F ST ; and 0.11518 to 0.16235 for Hudson’s F ST , compared to all other groups. The “South” haplogroup ranged from 0.48518 to 0.75611 for Nei’s F ST ; and 0.07823 to 0.20332 for Hudson’s F ST , compared to all other groups. An AMOVA corroborates these groups (p=0.01, CI=95%), with significant Euclidean genetic distances between these haplogroups. Nuclear Population Assessment For the intersection with previously published genomes, we have prioritized retaining the broadest geographic representation possible (64). Due to the nature of degraded archival and fecal samples, the number of shared sites across datasets are limited. With 70% of samples sharing reads across sites (Figure A3, Table A4), we retain a total of 1,168 shared genomic sites in 82 individuals (Table A4). For the African distribution, we proceeded using a minimum of 55% of samples sharing reads per site (Figure A6, Table A4), resulting in a total of 3,418 sites in 57 individuals (Table A4). PCAs assessing sample clustering for all intersection filtering strategies are found in Figures A1-A6). DeltaK estimates indicated K=10 for the range-wide leopard distribution, and K = 5 for the African continent, although we proceed with K=7 (the second best DeltaK), as it maintains consistency with the seven African ancestral groups identified in the range-wide dataset (out of K=10) (Figure A7). Estimated admixture proportions show gradual shifts across the total distribution of leopards, and within the African continent, and differ widely across samples (Figure 4). In Asia, three clusters emerge along a geographic gradient from west to east, roughly separating the Near East, the Indian subcontinent, and the Far East (Figure 4A). Individuals were assigned to “lineages” by the majority ancestral proportion (percentage of each respective K) represented within each individual’s admixture estimate (as opposed to their geographic occurrence or proximity). Admixture plots for K=3 thru K=10 for the total distribution, and K=3 thru K=7 for the African distribution are shown in Figures A8-A9. To make comparable groups for mito-nuclear discordance comparisons across the African distribution (because there are three mitochondrial haplogroups), we loosely consolidate K=7 ancestral lineages into three nuclear clusters, based on F ST estimates, where groups with similar allelic variation between them were combined (as represented by color blocks in Table 2): 1) PAR-I (shades of pink), 2) PAR-II (shades of green), and 3) PAR-III (blue), using the previously denoted “PAR” nomenclature to maintain consistency with previous designations (17,21) (Table 3). Nuclear clusters PAR-I thru PAR-III reflect and correspond to mitochondrial haplogroups Central, South and West, respectively, for use in downstream mito-nuclear discordance comparisons. African and Asian lineages are separated by 0.086462 allelic differentiation (Table 2). While there is geographic overlap between African clusters, allele frequency estimates indicate they are genetically differentiated (Table 2; Table A5). Heterozygosity estimates are relatively high, at nearly 12% for the total distribution of leopards, and around 8% for the African distribution (Table 2). Population level heterozygosity estimates range from 0.011 for the PAR-III lineage to 0.089 for the PAR-IIA lineage in Africa, and from 0.038 ( P. p. tuliana ) to 0.059 ( P. p. fusca ) in Asia (Table 2). The highest heterozygosity from the PAR-IIA lineage is spatially constrained in southern Cameroon (Figure 4), indicating a diversity hotspot. Effective population sizes (Ne) for Asia = 104,610.44, and for Africa, Ne = 122,362.00, while population level estimates range from 31,824.765 ( P. p. tuliana ) to 52,285.935 ( P. p. fusca ) in Asia, and from 11,943.311 (PAR-III) to 68,277.444 (PAR-IIA) in Africa (Table 2). Tajima’s D is negative across all lineages (Table 2). In the PCA plot of the total distribution dataset, we see African and Asian samples cluster discretely along PC1 (Figure 4E). Within the Asian samples, assigned admixture groups are distinguishable across PC1. For the African samples, we see large overlap of samples assigned to different ancestry groups by NGSadmix. Plots for PC3-PC8 shown in Figures A10-A11. Admixture estimates for our subsetted data (n= 16,903 genomic sites) mirror those seen in the total and African distribution datasets (Figure A12), confirming that the patterns we observed in our datasets with low genomic representation hold through the addition of more genomic sites, thus validating our inferences here. Unfortunately, none of the samples here represent ancestral contributions from the PAR-I or PAR-III groups (a byproduct of choosing high-sequence-yield individuals to include in the subset a priori ), however the PAR-II distributions here (Figure A12) are mostly consistent with individual assignments in the full data sets (Figure 4), where individuals with homogeneous ancestry in the full dataset retained their ancestry, and those with admixed ancestry containing signatures from PAR-II in the full dataset converged on homogeneous ancestry from PAR-II. While using a dataset with a larger number of SNPs, due to the low number of samples, we expect admixture to inflate individual ancestry proportions. Admixture plots for the subsetted data set for K=3 through K=10 are shown in Figure A13. A PCA plot of these data further confirms these patterns and shows the clustering of individuals from Africa versus individuals from Asia across PC1, and those from the PAR-II group across PC2 (Figure A12.B). Although whole genome data and mitochondrial genomes show mostly comparable patterns and are largely in line with the classic subspecies delineations in Asia (18,65), we observe a discordant pattern in Africa. Genetic distance based trees for the mitochondrial and nuclear genomes confirm a pattern of mito-nuclear discordance (Figure A14), although bootstrap support is low, possibly due to weak population structure and limited number of SNPs. Most PAR-I individuals are assigned to the Central mitochondrial haplogroup (concordant). Those assigned to PAR-II are mostly concordant to the mitochondrial South haplogroup assignments, although many individuals are assigned as well to the Central mitochondrial haplogroup (discordant). PAR-III shows only one out of two individuals concordant to their mitochondrial assignments, while the mitochondrial tree has an additional individual in the mitochondrial West haplogroup that is discordant to the nuclear assignment. Interestingly, two samples, A52041 and A52044, inhabiting Cameroon’s Congo rainforest belt, indicated by a star (*) in Figure A14, belong to the South mitochondrial haplogroup, although in this distance tree are more closely related to samples assigned to the West mitochondrial haplogroup. These samples are responsible for extending the South mitochondrial range northwards into the Congo River Basin (Figure 3, Figure 5B, Figure A14). We mapped the nuclear admixture proportions for each assigned nuclear population over the range extent of each assigned mitochondrial population (Figure 5). In addition, both mitochondrial and nuclear clade assignment were jointly visualized with mitochondrial assignment on the x-axis, against the ancestry proportion of each individual’s nuclear genome on the y-axis (Figure 5D-F). In this representation, samples clustering in the bottom left corner and upper right corner indicate congruence between mitochondrial and nuclear assignment. We see PAR-I samples fall within 100% overlap of the mitochondrial Central range extent (Figure 5A). 40.9% of these samples (18 out of 44 total samples) are discordant across mitochondrial and nuclear assignments (Figure 5D). Individuals assigned as PAR-II exceed the boundaries of the South mitochondrial range in approximately 15% of occurrences (Figure 5B). 48.5% of these (16 out of 33 total samples) have discordant mitochondrial and nuclear assignments (Figure 5E). Occurrences of PAR-III assigned individuals fall less than 50% within the West mitochondrial range (Figure 5C). Out of three samples assigned as mitochondrial West, and two samples assigned to PAR-III, only one sample, NI-017, overlapped for both (Figure 5F). As the rest of the samples were assigned to neither mitochondrial West or PAR-III lineages, we calculate for this cluster that 75% (3 out of 4 total samples) have discordant assignments (Figure 5F). Discussion: Here, we analyzed range-wide data from all leopard subspecies to investigate patterns of genomic variation across the leopards’ full distribution. This study covers regions formerly not represented (for example, the Congo River Basin) and adds comprehensive support for regions previously undersampled (for example, West Africa) (Figure 1). This was partially achieved by leveraging archival museum and natural history collections (66), which represent the majority of new data for this study. Although repurposing existing datasets can come with challenges, such as limited meta data and batch effects, we build on existing genomic resources for leopards to make the analyses as comprehensive as possible. Mitochondrial genome All seven recognized Asian subspecies (67) split out in our mitochondrial network according to their taxonomic divisions and geographical ranges (Figure 2, Table 1), as well as three mitochondrial haplogroups in Africa (Figures 2-3, Table 2). Within the African continent, the Central haplogroup occupies the major portion of sub-Saharan Africa (Figures 2-3, Figure 5A), and consists of four largely geographically overlapping sublineages stretching throughout the Sahel and into southern Africa. While most individuals assigned to the South haplogroup are found inhabiting the southern portion of the continent south of the Cunene and Zambezi Rivers, some samples extend the range of this haplogroup northwards into the Congo River Basin (Figures 2-3, Figure 5B). The West haplogroup is geographically constrained to regions stretching from the Gulf of Guinea towards the northern borders of the Sahel (Figures 2-3, Figure 5C). Importantly, these three haplogroups are in line with the patterns observed by Anco et al., 2017 (22) and reaffirms these population assignments. Morris et al., 2024 (21) inferred two mitochondrial lineages, respective to this study’s Central and South haplogroups, and while both lineages are widely distributed, Morris et al.’s PAR-II has a stronger geographic presence in the south of the continent, similar to what is shown here. Nucleotide diversity appears highest on the African continent, compared to the Asian subspecies, and is primarily driven by the West haplogroup. Elevated mitochondrial diversity in the West haplogroup may reflect deeper coalescence time, historically larger effective population size, demographic stability, or past admixture, although distinguishing among these processes would require explicit demographic modeling. Nuclear genome These data support the partitioning of Asian leopard populations into three major lineages (Figure 4A): Near East, Central Asia, and Far East, previously identified by Paijmans et al. 2021 (18). Due to limited sample size, we were unable to fully recapitulate individual subspecies, although observed population clustering aligns with an east to west gradient. African lineages show high diversity, coupled with weak population structure (Figure 4A-B), as was previously shown by Pečnerová et al. , 2021 (19). Paijmans et al., 2021 (18) tentatively suggests three lineages in Africa: a Northwestern population identified by the same Moroccan individual used in our study, a distinct Western population, and a geographically broad third population encompassing the remainder of their data, which widely represents coarse sampling from South Africa, the African interior, and East Africa as far north as Eritrea. With a more in-depth, population-level sampling, Pečnerová et al., 2021 (19) identified four lineages: a Western, a Southern, and two Eastern populations. Both studies are limited by the included sampling locations, which leaves extensive gaps in the leopard distribution, consequently potentially inflating the reported subdivision and missing distinct lineages. However, all studies seem to agree on a widespread central group and a distinct western group, whereas the southern group was not explicitly recovered by Paijmans et al., 2021, likely due to limitations in sampling. Also common to all of these are weak signals of differentiation between groups, and high heterozygosity across groups (driven primarily by ancestry from PAR-II) (Tables 2-3). We recognize that marker density can affect lineage resolution (23,25,27,68–70), and this may influence our results here, given our small number of SNPs after intersection filtering across datasets (n=1,168 and n=3,418 for the range-wide and African distributions, respectively) (Table A4). However, our subsetted data with increased SNP representation (n=16,903) show that our inferences here with limited universal shared sites are robust enough to capture patterns resolved with broader genomic representation (Figures A12-13). Biogeographical Implications Our results recovered some unexpected relationships, which combined, reveal interesting patterns of dispersal. For instance, in the nuclear genome, PAR-II is more closely related to Asian lineages than to either PAR-I or PAR-III African lineages, per F ST estimates (Table 2). And Arabian sample F77747 ( P. p. nimr ) revealed a closer relationship to West African leopards in both mitochondrial and genomic signatures than to its Asian cousins (Figure 2, Figure 4), which is also supported with previous published data (12,17,71). This may be reflective of connectivity across the Sahelian corridor (72), Horn of Africa, and the Arabian Peninsula, and begs additional research on possible gene flow across the Red Sea Corridor and colonization of the Arabian Peninsula. Of course, our Middle Eastern data are limited to only one individual, whereas the whole of Saudi Arabia, Oman and the northern Peninsula are not represented. Broader sampling across the entire Arabian Peninsula would be needed to fully resolve dispersal pathways. Additionally we find the westernmost sample, M1867, located in Senegal, nested within the Northeastern mitochondrial haplogroup (Figures 2-3). We also observe that the Moroccan sample SMNH582373 showed a closer mitochondrial relationship to Asian haplogroups than to African clades (Figure 2), while in the haplotype network excluding the Asian samples, it placed as basal to the Northeast haplogroup (Figure 3). These combined lines of evidence could alternatively point to longitudinal dispersal across the Sahelian corridor, stretching from the Guinean coast of Africa to the Red Sea, and encompassing arid to semi-arid habitats across north-south gradients of ecological diversity, moisture, and temperature (68). Because this pattern is based only on mitochondrial data, alternative explanations such as incomplete lineage sorting or single-locus phylogenetic limitations cannot be ruled out. However, additional North and West African sampling is essential before strong biogeographic conclusions can be drawn. Pleistocene expansions and contractions of Saharan refugia, as was shown for the Horned viper ( Cerastes cerastes ) (73), could also explain these patterns, but is less likely, given that occurrences of PAR-III individuals are distributed largely across the southern extent of the Sahelian corridor (Figure 4, Figure 5C). Future demographic modeling to test alternative dispersal scenarios is needed to confirm these potential demographic histories. The Congo River Basin has also been shown here as important for leopards, hosting genomic signatures of all mitochondrial haplogroups (Figures 2-3), ancestry from all nuclear lineages (Figure 4), and harboring the highest diversity (Tables 1-2, Table A5), indicating high levels of connectivity through the Congo rainforest belt. Rainforests therefore may act as both a source and archive of genomic diversity for leopards. The Congo Basin itself is well known for its exceptional biodiversity and high levels of endemism (74). Temporal connectivity has been suggested for savannah species, such as African buffalo ( Syncerus caffer ), for which populations in Cameroon were shown to have a close relationship with populations from Angola (75). A potential western rainforest refugia connection was also originally proposed to link lion populations, however this could not be corroborated, given uncertain origin and breeding history of analyzed specimens (24,76). Considering that leopards are doing far better in dense forests than lions and other savannah species (77–79), the Central African rain forest is thought to be less of a barrier for their dispersal (19). High admixture proportions and wide-spread clades identified in this study support this claim. One working hypothesis for the combined evidence described above suggests a possible Northeastern African origin for leopards, followed by expansion to Asia across the Arabian peninsula, and dispersal to the western and southern portions of the African continent through a rainforest connectivity hub in the Congo River Basin. This is supported by fossil evidence which points to an Eastern African origin of leopards (80,81). However, given that Arabia forms a critical biogeographic bridge between Africa and Asia, our uneven sampling there could influence these conclusions. Formal phylogeographic modeling with robust geographic sampling throughout the Middle East would be crucial to investigate this pattern. Looking towards Asia, Javan sample M1874 ( P. p. melas ), while being the farthest geographically from African resident leopards, has fewer mitogenomic polymorphic differences from African clades, compared to all other Asian clades (excluding P. p. nimr , discussed above), reaffirming previous research that have identified Javan leopards as mitochondrially distinct from their Asian counterparts (12,16,17,71). Paijmans et al., 2018 have recovered Javan leopards as basal to all Asian subspecies (64). They propose that recent divergence times (shallow coalescence) of mainland Asian leopards, resulting from Pleistocene population bottlenecks, but excluding Java, reduced genetic diversity in mainland Asian lineages, but not Javan leopards (64). This may likely explain why they are closely related to African leopards, which also were not influenced by bottlenecks to the same extent that mainland Asian leopards experienced. High genetic diversity in widespread taxa such as leopards, in the absence of geographic structure (this study, 18,19), reflects large effective population size (this study, (4) during the Pleistocene, enabled by ecological versatility rather than refugial isolation. Leopards are not necessarily restricted to the same differentiating forces that refugia pose to other taxa, as genomic signatures indicate widespread connectivity across different regions, and the ability to inhabit or traverse environments considered unsuitable for other co-distributed taxa (19). This may help explain why leopards do not illustrate strong phylogeographic breaks observed in other taxa across sub-Saharan Africa (24,82), despite subjection to the same climatic fluctuations and refugial expansions/contractions other taxa experience there. Mito-nuclear discordance Mitochondrial signals of phylogeographic breaks can exist even in continuously distributed species with little or no barriers to nuclear gene flow (83,84), which is confirmed for leopards by our results here. All three lineages described here (PAR-I, PAR-II and PAR-III) showed some level of discordance (Figure 5D-E, Figure A12), although it was strongest in PAR-III (Figure 5F), whereas for PAR-I and PAR-II, the majority of samples were concordant. Given the random and continuous distribution of nuclear ancestral contributions across the mitochondrial range extents, we conclude that ILS is the major driving factor behind the discordance. Since discordance is not a major contributor to PAR-I and PAR-II, we conclude that the high levels of admixture are the result of high gene flow perpetuated by male-biased dispersal, their primary means of dispersal (3), more so than ILS. For PAR-III, on the other hand, the majority of samples had discordant mito-nuclear assignments (Figure 5F), indicating that ILS is strongly influencing this lineage. PAR-II and PAR-III (to a stronger degree) have narrower representation of nuclear ancestry across their distribution, as compared to PAR-I (Figure 4, Figure 5D-F), and these are also much further dispersed across the continent relative to their mitochondrial ranges (Figure 5). For PAR-II, this is to a much stronger degree, observing PAR-II nuclear representation across nearly the entire continent (Figures 4-5). Coupled with the strong mitochondrial separation of these groups (Figures 2-3, Table 1), these patterns suggest that these are lineages for which the nuclear genome has not fully sorted because gene flow from male-biased dispersal is competing with lineage sorting in the nuclear genome. Whether these lineages are relics from Pleistocene refugia, or newly emerging lineage diversification is not clear. Further fine-scale demographic analyses would clarify the origin of each of these groups and their corresponding genomic relationships and structuring. Considering the mitochondrial haplotype networks and nuclear structure analyses together, we find a widely distributed PAR-I lineage, driven primarily by male-biased dispersal, characterized by weak population structure (Figure 4, Tables 2-3). This is the only group that reported a negative Tajima’s D in the mitochondrial genome (Table 2). We also recognize the PAR-II lineage characterized by mitochondrial differentiation (Figures 2-3, Table 1) in the presence of weak nuclear structure (Figure 4, Tables 2-3), which we also attribute to male-biased dispersal (Figure 5). This group harbors a diversity hotspot for leopards in Cameroon (Figure 4, Table 2, Table A5). This group is also most closely related to the Asian lineages (excluding Arabia and Java). Lastly, we recognize a genomically distinct PAR-III lineage, only detected in a relatively small proportion of samples, characterized by mito-nuclear discordance primarily driven by ILS (Figure 5). This group has relatively low heterozygosity and a much smaller effective population size (Tables 2-3), although it has the highest mitochondrial diversity (Table 1). Arabian leopards are most closely related to this group. Given the weak nuclear differentiation across leopards, and widespread signatures of admixture, we are careful not to designate these groups as discrete populations or subspecies, rather as independent lineages that are mixing across geographic space. The difficulty in delineating geographically discrete populations could indicate ecological or environmental isolation (IBE) rather than genomic isolation by distance (85,86). Leopards occupy a diverse range of habitats, including semi-arid environments, tropical and temperate rainforests, mountains, deserts, grassland plains, shrublands, and even suburbs (3,87–89), making IBE a potential route of future investigation into leopard structure. Conservation Implications The International Union for the Conservation of Nature (IUCN) lists the African leopard as vulnerable (67). As of 2024, new subspecies categorizations by the IUCN recognize seven subspecies designations across Asia ( P. p. nimr, P. p. tuliana (previously P. p. saxicolor ), P. p. fusca, P. p. kotiya, P. p. delacouri, P. p. orientalis, P. p. melas ), and one contiguous African population, P. p. pardus , with five of these classified subspecies listed as endangered or critically endangered (64,67). Currently, leopards face increasing threats, including loss of habitat and prey populations, and persecution for perceived and realized threats to livestock, illegal harvesting parts for local and global trade, and poorly managed trophy hunting (88,90). This has reduced the leopards’ historical distribution by up to 75% (88) and more range losses are projected to occur in the future (91). Furthermore, only 17% of the leopard’s current global range has been designated as protected areas (88). Yet its potential as a conservation umbrella species (92), or “umbrella capacity,” is substantial given its wide distribution in diverse habitats, and its large spatial and ecological overlap with other species (93,94). In fact, the leopard outranks all other felids in this umbrella capacity (93). Currently, the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES), in partnership with the International Union for the Conservation of Nature (IUCN), the Convention on the Conservation of Migratory Species of Wild Animals (CMS), and the African Carnivore Initiative (ACI) have published the Roadmap for the Conservation of the Leopard in Africa (RCLA v1.0) (95), which highlights regions of concern for leopard decline, and places importance on developing and implementing leopard conservation plans on the continental, regional, and national levels. Our research fills gaps identified in Objective 2 of the RCLA, by deepening our understanding of genomic diversity and hotspots for leopards from regions with previously little knowledge, particularly from West and Central Africa, regions of priority interest designated in the RCLA. For instance, PAR-III, occurring in areas already under severe threat of decline (88,95), is challenged with low heterozygosity and relatively small effective population size, which can be detrimental for its sustained survival. And tropical rainforest belts in Central Africa doubly serve as corridors and habitat, with the capacity to hold high levels of diversity, suggesting that prioritizing the establishment/designation and maintenance of protected areas (PAs) within rainforested regions could be a crucial step towards maintaining leopard diversity. While individual Asian subspecies have been given individualized conservation status by the IUCN (66), African leopards ( P. p. pardus ) have been listed as vulnerable, and are given the same conservation priority across the entire continent with no recent regional assessments. While the lineages we identified in this study are overlapping, we show that the African continent harbours extensive diversity in differentiated lineages. Doing regional assessments, both for the IUCN Red List and Green Status, could help to understand the status of leopard populations more locally. How to deal with identifying conservation units, especially when geographic delineation is challenging, has been widely discussed (23,96–99), and, as addressed above, is not the goal of our study. Rather, our aim was to highlight cryptic diversity by greatly expanding sampling localities, including into previously undersampled regions, and address mito-nuclear discordance as a potential risk for conservation strategies based on mitochondrial data only. However, the geographically and comprehensive sampling framework generated in this study, will provide a genomic reference baseline for further assignment of leopard samples across their extant range and this is undoubtedly an important input in forensics particularly when facing illegal trade of leopard body parts.Molecular tools are increasingly being used to inform conservation decisions or for forensic applications in the context of illegal wildlife trade. A recent work, partially repurposing existing data, has generated a leopard specific SNP panel for across the leopard’s global range and produced a baseline dataset of 900 individuals (20). This panel was optimized and validated for use with low quality, non-invasively collected samples. Although application of this panel still depends on available laboratory capacity, the nature of the dataset allows analysis with relatively limited computational resources, making it more accessible across range states. New data generated through this panel can be added to the baseline, allowing even more detailed inferences. As many range countries currently are still building capacity to execute genome-wide analyses for wildlife species, mitochondrial data are still widely used. And as such, the range-wide mito haplogroups (Fig 2-3), even if overlapping, provide an important step for Wildlife Forensic applications to be able, once results are validated, to trace Leopard seizures through origin assignments along the trade routes. For leopards, a species with wide-spread mito-nuclear discordance, a thorough comparison as shown here, will hopefully help to make correct inference from the data, even if at present this is restricted to mitochondrial data. It illustrates how genomic resources, such as the ones generated in this study, and an in depth understanding of spatial patterns from various marker types can contribute to the development of tools for informing conservation action. Literature Cited: 1. Myers N. The Leopard Panthera pardus in Africa. [Internet]. Morges, Switzerland: International Union for Conservation of Nature and Natural Resources; 1976 [cited 2026 Feb 16]. (IUCN MONOGRAPH No. 5). Available from: https://www.jstor.org/stable/2402577?origin=crossref doi:10.2307/2402577 2. Marciszak A, Krajcarz MT, Krajcarz M, Stefaniak K. The first record of leopard Panthera pardus Linnaeus, 1758 from the Pleistocene of Poland. Acta Zool Cracoviensia - Ser Vertebr. 2011 Jul 29;54(1):39–46. doi:10.3409/azc.54a_1-2.39-46 3. Stein AB, Hayssen V. Panthera pardus (Carnivora: Felidae). Mamm Species. 2013;45(900):30–48. 4. Spong G, Johansson M, Björklund M. High genetic variation in leopards indicates large and long‐term stable effective population size. Mol Ecol. 2000 Nov;9(11):1773–82. doi:10.1046/j.1365-294x.2000.01067.x 5. Uphyrkina O. Conservation Genetics of the Far Eastern Leopard (Panthera pardus orientalis). J Hered. 2002 Sep 1;93(5):303–11. doi:10.1093/jhered/93.5.303 6. Perez I, Geffen E, Mokady O. Critically Endangered Arabian leopards Panthera pardus nimr in Israel: estimating population parameters using molecular scatology. Oryx. 2006 Jul;40(3):295–301. doi:10.1017/S0030605306000846 7. Rozhnov VV, Lukarevskiy VS, Sorokin PA. Application of molecular genetic characteristics for reintroduction of the leopard (Panthera pardus L., 1758) in the Caucasus. Dokl Biol Sci. 2011 Apr;437(1):97–102. doi:10.1134/S0012496611020062 8. Rozhnov VV, Sorokin PA, Lukarevskiy VS, Naidenko SV, Hernandes-Blanko JA, Lukarevskiy SV. Individual identification of Amur leopards (Panthera pardus orientalis) using molecular-genetic methods and the population size estimation. Biol Bull. 2013 Apr;40(2):124–9. doi:10.1134/S106235901302012X 9. Dutta T, Sharma S, Maldonado JE, Wood TC, Panwar HS, Seidensticker J. Fine‐scale population genetic structure in a wide‐ranging carnivore, the leopard ( P anthera pardus fusca) in central I ndia. Austin J, editor. Divers Distrib. 2013 Jul;19(7):760–71. doi:10.1111/ddi.12024 10. Sugimoto T, Aramilev VV, Kerley LL, Nagata J, Miquelle DG, McCullough DR. Noninvasive genetic analyses for estimating population size and genetic diversity of the remaining Far Eastern leopard (Panthera pardus orientalis) population. Conserv Genet. 2014 Jun;15(3):521–32. doi:10.1007/s10592-013-0558-8 11. McManus JS, Dalton DL, Kotzé A, Smuts B, Dickman A, Marshal JP, et al. Gene flow and population structure of a solitary top carnivore in a human‐dominated landscape. Ecol Evol. 2015 Jan;5(2):335–44. doi:10.1002/ece3.1322 12. Wilting A, Patel R, Pfestorf H, Kern C, Sultan K, Ario A, et al. Evolutionary history and conservation significance of the Javan leopard Panthera pardus melas . J Zool. 2016 Aug;299(4):239–50. doi:10.1111/jzo.12348 13. Asad M, Martoni F, Ross JG, Waseem M, Abbas F i, Paterson AM. Assessing subspecies status of leopards ( Panthera pardus ) of northern Pakistan using mitochondrial DNA. PeerJ. 2019 Jul 16;7:e7243. doi:10.7717/peerj.7243 14. Zhou C, Liu Y, Zhang R, Zheng X, Zhao G, Li F, et al. Chromosome-level Genome Assembly of the High-altitude Leopard ( Panthera pardus ) Sheds Light on Its Environmental Adaptation. Fraser B, editor. Genome Biol Evol. 2022 Sep 6;14(9):evac128. doi:10.1093/gbe/evac128 15. Tensen L, Emami-Khoyi A, Khan A, Camacho G, Swanepoel L, Fischer K. Mitogenomic Characterization of South African Leopards and the Effect of Past Climatic Events. Daniels S, editor. J Zool Syst Evol Res. 2024 May 4;2024:1–14. doi:10.1155/2024/2174469 16. Miththapala S, Seidensticker J, O’Brien SJ. Phylogeographic Subspecies Recognition in Leopards (Panthera pardus): Molecular Genetic Variation. Conserv Biol. 1996;10(4):1115–32. 17. Uphyrkina O, Johnson WE, Quigley H, Miquelle D, Marker L, Bush M, et al. Phylogenetics, genome diversity and origin of modern leopard, Panthera pardus . Mol Ecol. 2001 Nov;10(11):2617–33. doi:10.1046/j.0962-1083.2001.01350.x 18. Paijmans JLA, Barlow A, Becker MS, Cahill JA, Fickel J, Förster DWG, et al. African and Asian leopards are highly differentiated at the genomic level. Curr Biol. 2021 May;31(9):1872-1882.e5. doi:10.1016/j.cub.2021.03.084 19. Pečnerová P, Garcia-Erill G, Liu X, Nursyifa C, Waples RK, Santander CG, et al. High genetic diversity and low differentiation reflect the ecological versatility of the African leopard. Curr Biol. 2021 May;31(9):1862-1871.e5. doi:10.1016/j.cub.2021.01.064 20. Mamugy F, Bertola LD, De Vry AM, Dussex N, Shiffthaler B, Paijmans JLA, et al. SNP panel for non-invasive genotyping of leopard ( Panthera pardus ). Genetics; 2024 doi:10.1101/2024.11.01.621452 21. Anco C, Kolokotronis SO, Henschel P, Cunningham SW, Amato G, Hekkala E. Historical mitochondrial diversity in African leopards ( Panthera pardus ) revealed by archival museum specimens. Mitochondrial DNA Part A. 2017 Apr 3;29(3):455–73. doi:10.1080/24701394.2017.1307973 22. Morris DR, McWhorter TJ, Boardman WSJ, Simpson G, Wentzel J, Coetzee J, et al. Unravelling the maternal evolutionary history of the African leopard ( Panthera pardus pardus ). PeerJ. 2024 Apr 11;12:e17018. doi:10.7717/peerj.17018 23. Allendorf FW. Genetics and the conservation of natural populations: allozymes to genomes. Mol Ecol. 2017 Jan;26(2):420–30. doi:10.1111/mec.13948 24. Bertola LD, Jongbloed H, Van Der Gaag KJ, De Knijff P, Yamaguchi N, Hooghiemstra H, et al. Phylogeographic Patterns in Africa and High Resolution Delineation of Genetic Clades in the Lion (Panthera leo). Sci Rep. 2016 Aug 4;6(1):30807. doi:10.1038/srep30807 25. Hey J, Nielsen R. Multilocus Methods for Estimating Population Sizes, Migration Rates and Divergence Time, With Applications to the Divergence of Drosophila pseudoobscura and D. persimilis. Genetics. 2004 Jun 1;167(2):747–60. doi:10.1534/genetics.103.024182 26. Maddison WP, Knowles LL. Inferring Phylogeny Despite Incomplete Lineage Sorting. Collins T, editor. Syst Biol. 2006 Feb 1;55(1):21–30. doi:10.1080/10635150500354928 27. Edwards S, Bensch S. Looking forwards or looking backwards in avian phylogeography? A comment on Zink and Barrowclough 2008. Mol Ecol. 2009 Jul;18(14):2930–3. doi:10.1111/j.1365-294X.2009.04270.x 28. Ishida Y, Oleksyk TK, Georgiadis NJ, David VA, Zhao K, Stephens RM, et al. Reconciling Apparent Conflicts between Mitochondrial and Nuclear Phylogenies in African Elephants. Murphy WJ, editor. PLoS ONE. 2011 Jun 8;6(6):e20642. doi:10.1371/journal.pone.0020642 29. Toews DPL, Brelsford A. The biogeography of mitochondrial and nuclear discordance in animals. Mol Ecol. 2012 Aug;21(16):3907–30. doi:10.1111/j.1365-294X.2012.05664.x 30. Keyghobadi N. The genetic implications of habitat fragmentation for animals. Can J Zool. 2007 Oct;85(10):1049–64. doi:10.1139/Z07-095 31. Funk DJ, Omland KE. Species-Level Paraphyly and Polyphyly: Frequency, Causes, and Consequences, with Insights from Animal Mitochondrial DNA. Annu Rev Ecol Evol Syst. 2003 Nov;34(1):397–423. doi:10.1146/annurev.ecolsys.34.011802.132421 32. Zink RM, Barrowclough GF. Mitochondrial DNA under siege in avian phylogeography. Mol Ecol. 2008 May;17(9):2107–21. doi:10.1111/j.1365-294X.2008.03737.x 33. Meyer M, Kircher M. Illumina sequencing library preparation for highly multiplexed target capture and sequencing. Cold Spring Harb Protoc. 2010;5:1–11. 34. Andrews S. FastQC: A Quality Control Tool for High Throughput Sequence Data. Available Online http://www.bioinformatics.babraham.ac.uk/projects/fastqc. 2010. 35. Wingett SW, Andrews S. FastQ Screen: A tool for multi-genome mapping and quality control. F1000 Res. 2018;7:1338. 36. Jiang H, Lei R, Ding SW, Zhu S. Skewer: a fast and accurate adapter trimmer for next-generation sequencing paired-end reads. BMC Bioinformatics. 2014;15:182. 37. Kim S, Cho YS, Kim HM, Chung O, Kim H, Jho S, et al. Comparison of carnivore, omnivore, and herbivore mammalian genomes with a new leopard assembly. Genome Biol. 2016 Dec;17(1):211. doi:10.1186/s13059-016-1071-4 38. Li G, Davis BW, Eizirik E, Murphy WJ. Phylogenomic evidence for ancient hybridization in the genomes of living cats (Felidae). Genome Res. 2016 Jan;26(1):1–11. doi:10.1101/gr.186668.114 39. Li H, Durbin R. Fast and accurate short read alignment with Burrows–Wheeler transform. Bioinformatics. 2009 Jul 15;25(14):1754–60. doi:10.1093/bioinformatics/btp324 40. Curry CJ, Davis BW, Bertola LD, White PA, Murphy WJ, Derr JN. Spatiotemporal Genetic Diversity of Lions Reveals the Influence of Habitat Fragmentation across Africa. Heye E, editor. Mol Biol Evol. 2021 Jan 4;38(1):48–57. doi:10.1093/molbev/msaa174 41. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009 Aug 15;25(16):2078–9. doi:10.1093/bioinformatics/btp352 42. Korneliussen TS, Albrechtsen A, Nielsen R. ANGSD: Analysis of Next Generation Sequencing Data. BMC Bioinformatics. 2014 Dec;15(1):356. doi:10.1186/s12859-014-0356-4 43. Bodenhofer U, Bonatesta E, Horejš-Kainrath C, Hochreiter S. msa: an R package for multiple sequence alignment. Bioinformatics. 2015 Dec 15;31(24):3997–9. doi:10.1093/bioinformatics/btv494 44. Leigh JW, Bryant D. popart : full‐feature software for haplotype network construction. Nakagawa S, editor. Methods Ecol Evol. 2015 Sep;6(9):1110–6. doi:10.1111/2041-210X.12410 45. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol. 1990 Oct;215(3):403–10. doi:10.1016/S0022-2836(05)80360-2 46. Hudson RR, Slatkin M, Maddison WP. Estimation of levels of gene flow from DNA sequence data. Genetics. 1992 Oct 1;132(2):583–9. doi:10.1093/genetics/132.2.583 47. Bhatia G, Patterson N, Sankararaman S, Price AL. Estimating and interpreting F ST : The impact of rare variants. Genome Res. 2013 Sep;23(9):1514–21. doi:10.1101/gr.154831.113 48. Chaichoompu, K, Abegaz, F, Tongsima, S, Shaw, PJ, Sakuntabhai, A, Pereira, L, et al. KRIS: keen and reliable interface subroutines for bioinformatic analysis. R Package Version 11. 2018. 49. Nei M. Analysis of Gene Diversity in Subdivided Populations. Proc Natl Acad Sci. 1973;70(12):3321–3. 50. Pfeifer B, Wittelsbürger U, Ramos-Onsins SE, Lercher MJ. PopGenome: An Efficient Swiss Army Knife for Population Genomic Analyses in R. Mol Biol Evol. 2014 Jul;31(7):1929–36. doi:10.1093/molbev/msu136 51. Dray S, Dufour AB. The ade4 Package: Implementing the Duality Diagram for Ecologists. J Stat Softw. 2007;22(4). doi:10.18637/jss.v022.i04 52. Tajima F. Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics. 1989 Nov 1;123(3):585–95. doi:10.1093/genetics/123.3.585 53. Watterson GA. On the number of segregating sites in genetical models without recombination. Theor Popul Biol. 1975 Apr;7(2):256–76. doi:10.1016/0040-5809(75)90020-9 54. Dung SK, López A, Barragan EL, Reyes RJ, Thu R, Castellanos E, et al. Illuminating Women’s Hidden Contribution to Historical Theoretical Population Genetics. Genetics. 2019 Feb 1;211(2):363–6. doi:10.1534/genetics.118.301277 55. Skotte L, Korneliussen TS, Albrechtsen A. Estimating Individual Admixture Proportions from Next Generation Sequencing Data. Genetics. 2013 Nov 1;195(3):693–702. doi:10.1534/genetics.113.154138 56. R Core Team. R: A Language and Environment for Statistical Computing. [Internet]. Vienna, Austria: R Foundation for Statistical Computing; 2025. Available from: https://www.R-project.org/ 57. Rasmussen MS, Garcia-Erill G, Korneliussen, TS, Wiuf C, Albrechtsen A. Estimation of site frequency spectra from low-coverage sequencing data using stochastic EM reduces overfitting, runtime, and memory usage. Genetics. 2022;222(4). doi:10.1093/genetics/iyac148 58. Nei M, Li WH. Mathematical model for studying genetic variation in terms of restriction endonucleases. Proc Natl Acad Sci. 1979 Oct;76(10):5269–73. doi:10.1073/pnas.76.10.5269 59. Lefort V, Desper R, Gascuel O. FastME 2.0: A Comprehensive, Accurate, and Fast Distance-Based Phylogeny Inference Program: Table 1. Mol Biol Evol. 2015 Oct;32(10):2798–800. doi:10.1093/molbev/msv150 60. Paradis E, Schliep K. ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. Schwartz R, editor. Bioinformatics. 2019 Feb 1;35(3):526–8. doi:10.1093/bioinformatics/bty633 61. Revell LJ. phytools 2.0: an updated R ecosystem for phylogenetic comparative methods (and other things). PeerJ. 2024 Jan 5;12:e16505. doi:10.7717/peerj.16505 62. Calenge C. adehabitatHR: Home Range Estimation. R package version 0.4.22; 2024. Available from: https://github.com/clementcalenge/adehabitathr 63. Stein AB, Gerngross P, Al Hikmani H, Balme G, Bertola LD, Drouilly M, et al. Panthera pardus. The IUCN Red List of Threatened Species 2024. e.T15954A254576956. 2024;e.T15954A254576956. doi:https://dx.doi.org/10.2305/IUCN.UK.2024-1.RLTS.T15954A254576956.en 64. Wiens JJ. Missing data and the design of phylogenetic analyses. J Biomed Inform. 2006 Feb;39(1):34–42. doi:10.1016/j.jbi.2005.04.001 65. Paijmans JLA, Barlow A, Förster DW, Henneberger K, Meyer M, Nickel B, et al. Historical biogeography of the leopard (Panthera pardus) and its extinct Eurasian populations. BMC Evol Biol. 2018 Dec;18(1):156. doi:10.1186/s12862-018-1268-0 66. Holmes MW, Hammond TT, Wogan GOU, Walsh RE, LaBarbera K, Wommack EA, et al. Natural history collections as windows on evolutionary processes. Mol Ecol. 2016 Feb;25(4):864–81. doi:10.1111/mec.13529 67. Stein, A. B., Gerngross P, Al Hikmani H, Balme, G., Bertola, L., Drouilly M, et al. Panthera pardus (amended version of 2025 assessment). The IUCN Red List of Threatened Species 2025: e.T15954A286153337. 2025. Available from: https://www.iucnredlist.org/species/15954/286153337 doi:10.2305/IUCN.UK.2025-2.RLTS.T15954A286153337.en 68. Luikart G, England PR, Tallmon D, Jordan S, Taberlet P. The power and promise of population genomics: from genotyping to genome typing. Nat Rev Genet. 2003 Dec;4(12):981–94. doi:10.1038/nrg1226 69. Dupuis JR, Roe AD, Sperling FAH. Multi‐locus species delimitation in closely related animals and fungi: one marker is not enough. Mol Ecol. 2012 Sep;21(18):4422–36. doi:10.1111/j.1365-294X.2012.05642.x 70. Pálešová K. Methodological Advances in the Analysis of Genetic Population Structure: Implications for Biodiversity Conservation. Acta Fytotech Zootech. 2025 Mar 31;28(1):28–39. doi:10.15414/afz.2025.28.01.28-39 71. Luo S, Zhang Y, Johnson WE, Miao L, Martelli P, Antunes A, et al. Sympatric Asian felid phylogeography reveals a major Indochinese–Sundaic divergence. Mol Ecol. 2014 Apr;23(8):2072–92. doi:10.1111/mec.12716 72. Le Houerou HN. The Rangelands of the Sahel. J Range Manag. 1980 Jan;33(1):41. doi:10.2307/3898226 73. Liz AV, Mochales-Riaño G, Velo-Antón G, García-Cardenete L, Brito JC, Carranza S, et al. Historical biogeography of the Saharan horned viper enlightens past dynamics of hyperarid desert habitats. J Arid Environ. 2024 Jun;222:105171. doi:10.1016/j.jaridenv.2024.105171 74. White LJT, Inogwabini BI, Sonke B, Koutika LS, Abernethy KA, Braun JJ, et al. Congo Basin Resilience and Sustainability: From the Past to the Future. New York: Science Panel for the Congo Basin & Sustainable Development Solutions Network (SDSN); 2025. p. 45. (Congo Basin Assessment Report). Report No. 75. Van Hooft WF, Groen AF, Prins HHT. Phylogeography of the African buffalo based on mitochondrial and Y‐chromosomal loci: Pleistocene origin and population expansion of the Cape buffalo subspecies. Mol Ecol. 2002 Feb;11(2):267–79. doi:10.1046/j.1365-294X.2002.01429.x 76. Bertola LD, Van Hooft WF, Vrieling K, Uit De Weerd DR, York DS, Bauer H, et al. Genetic diversity, evolutionary history and implications for conservation of the lion (Panthera leo) in West and Central Africa: Phylogeography of the lion (Panthera leo). J Biogeogr. 2011 Jul;38(7):1356–67. doi:10.1111/j.1365-2699.2011.02500.x 77. Henschel P. The Status and Conservation of Leopards and Other Large Carnivores in the Congo Basin, and the Potential Role of Reintroduction. In: Hayward MW, Somers MJ, editors. Reintroduction of Top‐Order Predators. 1st ed. Wiley; 2009. p. 206–37. doi:10.1002/9781444312034.ch10 78. Mavinga, FB. A camera trap assessment of factors influencing leopard (Panthera pardus) habitat use in the Nouabalé-Ndoki National Park, Republic of Congo. [South Africa]: University of Cape Town; 2018. 79. Corredor-Ospina N, Kreyer M, Rossi G, Hohmann G, Fruth B. First report of a leopard (Panthera pardus)–bonobo (Pan paniscus) encounter at the LuiKotale study site, Democratic Republic of the Congo. Primates. 2021 Jul;62(4):555–62. doi:10.1007/s10329-021-00897-8 80. Turner A, Antón M, editors. The big cats and their fossil relatives: an illustrated guide to their evolution and natural history. New York: Columbia Univ. Press; 1997. 234 p. 81. Werdelin L, Yamaguchi N, Johnson WE, O’Brien SJ. Phylogeny and evolution of cats (Felidae). Biol Conserv Wild Felids. 2010;59–82. 82. Lorenzen ED, Heller R, Siegismund HR. Comparative phylogeography of African savannah ungulates 1 . Mol Ecol. 2012 Aug;21(15):3656–70. doi:10.1111/j.1365-294X.2012.05650.x 83. Irwin DE. Phylogeographic breaks without geographic barriers to gene flow. Evolution. 2002 Dec;56(12):2383–94. doi:10.1111/j.0014-3820.2002.tb00164.x 84. Nosil P. Speciation with gene flow could be common. Mol Ecol. 2008 May;17(9):2103–6. doi:10.1111/j.1365-294X.2008.03715.x 85. Wang IJ. Examining the full effects of landscape heterogeneity on spatial genetic variation: a multiple matrix regression approach for quantifying geographic and ecological isolation: special section. Evolution. 2013 Dec;67(12):3403–11. doi:10.1111/evo.12134 86. Wang IJ, Bradburd GS. Isolation by environment. Mol Ecol. 2014 Dec;23(23):5649–62. doi:10.1111/mec.12938 87. Nowell K, Jackson P. Wild cats: status survey and conservation action plan. Gland (Switzerland): IUCN Species survival commission; 1996. (IUCN/SSC Action plans for the conservation of biological diversity). Report No. 88. Jacobson AP, Gerngross P, Lemeris Jr. JR, Schoonover RF, Anco C, Breitenmoser-Würsten C, et al. Leopard ( Panthera pardus ) status, distribution, and the research efforts across its range. PeerJ. 2016 May 4;4:e1974. doi:10.7717/peerj.1974 89. Sunquist M, Sunquist F. Wild Cats of the World. University of Chicago Press; 2002. 90. Sandom CJ, Williams J, Burnham D, Dickman AJ, Hinks AE, Macdonald EA, et al. Deconstructed cat communities: Quantifying the threat to felids from prey defaunation. Leung B, editor. Divers Distrib. 2017 Jun;23(6):667–79. doi:10.1111/ddi.12558 91. Mitchell C, Bolam J, Bertola LD, Naude VN, Gonçalves da Silva L, Razgour O. Leopard subspecies conservation under climate and land‐use change. Ecol Evol. 2024 May;14(5):e11391. doi:10.1002/ece3.11391 92. Caro TM. Conservation by Proxy: Indicator, Umbrella, Keystone, Flagship, and Other Surrogate Species. Washington DC: Island Press; 2010. 1 p. 93. Dickman AJ, Hinks AE, Macdonald EA, Burnham D, Macdonald DW. Priorities for global felid conservation: Global Felid Conservation. Conserv Biol. 2015 Jun;29(3):854–64. doi:10.1111/cobi.12494 94. Macdonald EA, Hinks A, Weiss DJ, Dickman A, Burnham D, Sandom CJ, et al. Identifying ambassador species for conservation marketing. Glob Ecol Conserv. 2017 Oct;12:204–14. doi:10.1016/j.gecco.2017.11.006 95. IUCN SSC Cat Specialist Group. Roadmap for the conservation of leopards in Africa. Muri/Bern, Switzerland; 2019. 96. Moritz C. Defining ‘Evolutionarily Significant Units’ for conservation. Trends Ecol Evol. 1994 Oct;9(10):373–5. doi:10.1016/0169-5347(94)90057-4 97. Palsboll P, Berube M, Allendorf F. Identification of management units using population genetic data. Trends Ecol Evol. 2007 Jan;22(1):11–6. doi:10.1016/j.tree.2006.09.003 98. Bertola LD, Nicholson SK, Bauer H, Masembe C, Chege M, Asfaw T, et al. Subspecies introgression: insights and implications for lion management and policy. Cat News. 2025;84. 99. Geue J, Bertola L, Paulette B, Brüniche-Olsen A, Da Silva J, DeWoody JA, et al. Practical genetic diversity protection: an accessible framework for IUCN subpopulation and Evolutionarily Significant Unit identification [Internet]. Biodiversity; 2025 [cited 2026 Feb 26]. Available from: https://ecoevorxiv.org/repository/view/9129/ doi:10.32942/X2RK9Q Table 1. Mitochondrial F ST and Summary statistics across the total distribution of leopards. F ST measurements for ten haplogroups (color coded by clade assignment from Figure 2) across the entire leopard distribution on top, F ST Africa versus Asia below. Bottom of diagonal represents Nei’s F ST (Nei, 1973). Top diagonal represents Hudson’s F ST (Hudson et al., 1992; Bhatia et al., 2013) designated in light gold. Total Dist = global estimates across all samples, π = nucleotide diversity, θ = Watterson’s theta, n=sample size, NA=no data. F ST Haplotype Diversity π θ Tajima’s D ASIA AFRICA ORIENTALIS DELACOURI TULIANA NIMR FUSCA KOTIYA MELAS CENTRAL WEST SOUTH ORIENTALIS (n=5) -0.1443 0.14171 0.47384 0.0325 -0.1669 0.09341 0.31424 0.38353 0.32938 1 0.002135 39.36 1.632979 DELACOURI (n=3) 0.20984 0.11625 0.43457 -0.0409 -0.1196 0.04615 0.29694 0.36478 0.32475 1 0.001719 28.66667 NA TULIANA (n=8) 0.6181 0.66935 0.29389 0.12393 0.03498 -0.1002 0.22967 0.25502 0.24217 1 0.001328 28.53994 0.236559 NIMR (n=1) 0.94664 0.957 0.96695 0.40963 0.34 -0.2 0.11389 0.04312 0.1944 NA NA NA NA FUSCA (n=5) 0.53031 0.58608 0.60971 0.95149 -0.068 0.00247 0.29914 0.35365 0.3404 1 0.001991 36.96 1.776951 KOTIYA (n=3) 0.72735 0.77954 0.80304 0.99216 0.7245 -0.0069 0.24677 0.30516 0.27383 1 0.00032 5.33333 NA MELAS (n=1) 0.89329 0.914 0.93123 1 0.89939 0.98406 0.00373 -0.0308 0.01763 NA NA NA NA CENTRAL (n=96) 0.87688 0.88788 0.89717 0.90824 0.88078 0.92421 0.87384 0.13591 0.14241 0.996641 0.002656 141.2457 -2.02602 WEST (n=4) 0.72354 0.73317 0.74283 0.66301 0.72973 0.77045 0.68446 0.58206 0.13736 1 0.009217 141.8182 4.587346 SOUTH (n=21) 0.7928 0.80471 0.81135 0.78506 0.79851 0.83875 0.78245 0.69588 0.48519 0.985714 0.006359 105.6219 0.95152 Total Dist. (n=147) 0.995006 0.0115 330.9616 -1.32229 F ST ASIA AFRICA ASIA (n=26) 0.24154 0.997436 0.006965 238.5965 -1.29642 AFRICA (n=121) 0.67652 1 0.005443 166.1444 -0.60291 Table 2. Nuclear F ST and Summary statistics across the total distribution of leopards. F ST measurements for K=10 populations across the entire leopard distribution on top, color coded by ancestry assignment (Figure 4). F ST Africa versus Asia below. Bottom diagonal represents Hudson’s F ST (Hudson et al., 1992; Bhatia et al., 2013). Ne = effective population size, θ = Watterson’s theta, n=sample size. F ST Ne Hetero-zygosity θ Tajima’s D ASIA AFRICA TULIANA FUSCA ORIENTALIS PAR-II A PAR-II B PAR-II C PAR-II D PAR-I A PAR-I B PAR-III TULIANA (n=8) 31824.765 0.0380634 0.01273 -2.2964 FUSCA (n=6) 0.051458 52285.935 0.0588973 0.020914 -2.10915 ORIENTALIS (n=8) 0.111692 0.034188 35712.185 0.0427608 0.014285 -2.28956 PAR-II A (n=16) 0.168236 0.112388 0.169637 68277.444 0.0894316 0.027311 -2.56095 PAR-II B (n=8) 0.158562 0.084457 0.15919 0.052851 40792.641 0.0475953 0.016317 -2.20817 PAR-II C (n=10) 0.166423 0.09221 0.169109 0.046809 0.015877 40457.862 0.0496218 0.016183 -2.3284 PAR-II D (n=6) 0.26432 0.165883 0.259407 0.139473 0.107413 0.108709 19399.557 0.019667 0.00776 -1.75248 PAR-I A (n=7) 0.434838 0.335257 0.420907 0.341303 0.348834 0.361858 0.440658 15407.930 0.0144831 0.006163 -1.40608 PAR-I B (n=6) 0.416583 0.308204 0.384085 0.335216 0.333331 0.351497 0.397013 0.511042 20575.833 0.0205671 0.00823 -1.70073 PAR-III (n=4) 0.381453 0.281046 0.372709 0.275722 0.276533 0.286474 0.381624 0.541853 0.500467 11943.311 0.0109381 0.004777 -1.34851 Total Dist. 2374188.3 0.11953 0.0718 -2.84605 F ST ASIA AFRICA ASIA (n=22) 104610.44 0.1478372 0.041844 -2.75652 AFRICA (n=57) 0.086462 122362.00 0.0828314 0.048945 -2.83095 Table 3. Nuclear F ST and Summary statistics across the African distribution of leopards. Hudson’s F ST (Hudson et al., 1992; Bhatia et al., 2013) measurements for African distribution of leopards for three groups as derived from the consolidation of K=7 admixture lineages, plus summary statistics. Ne = effective population size; θ = Watterson’s theta, n=sample size. PAR-I PAR-II PAR-III PAR-I (n=13) 0.0511094 0.0695369 -2.853541 173842.2 PAR-II (n=40) 0.148329 0.0606371 0.0717955 -2.893528 179488.638 PAR-III (n=4) 0.48233 0.432981 0.0015389 0.00274993 0.0026777 6874.81601 Total Africa 0.1478372 0.04184417 -2.756521 104610.442 Appendix A: Supplementary Tables Table A1. Leopard published, archival and fecal specimens origin and collection date (n = sample size, AMNH = American Museum of Natural History, FMNH = Field Museum of Natural History, MNHN = Le Museum National d’Histoire Naturelle, N/A = specimen tag and/or field notes do not contain the information) West Africa Cote D’Ivoire Ghana Liberia Nigeria Senegal Sierra Leone Togo 1 6 1 1 1 1 1 CCNP011 7246, 7465, 7466, 7547, 7548 - 7549 A213648 NI-017 M1867 SLGR019 M1909-263 Comoe National Park Jang Oku Chierese Kananto Grupe Monrovia Gashaka-Gumpti National Park Larchet Gola Rainforest National Park Togodo 2019 1998 N/A 2009 1842 2019 1909 D. Del Castillo Pečnerová et al., 2021 AMNH P. Henschel MNHN D. Del Castillo MNHN Central Africa Cameroon Chad Democratic Republic of the Congo Equatorial Guinea Gabon 19 1 23 1 2 A170290 – A170291, A170292, A170293, A170294 – A170295, A170296, A170297 – A170298, A170299, A170300, A170306, A170307, A170309, A54334 – A54335, A87236, CDRB015, MFN_MAM_056545 A164151 A208770, A189390, A52005, A52006, A52008, A52009, A52010, A52013, A52017, A52024, A52026, A52028, A52035, A52036, A52038, A52039, A52041-A52042, A52043-A52044, A52045, A52046, A52047, F27502 SMNH581240 A120285, MFN_MAM_056389 Akom Ekowon Mvie Akom Efulan Akom Messambe Efulen Evuma Okon Nr Metet Efulen Dja National Park Bafreng Fort Archambault District Kivu District, Bunyakiri Ubangi District Faradje Niapu Akenge Garamba Haute Zaire, Poko Haute Zaire, Gamangui Medje Vankerckhovenville (Nzoro) Bafuka Virunga Mts, Pass between Mt Mikeno & Mt Karisimbi N/A Oytem (Oyem) Akoafim 1934-35 1935 1934 1936 1935 1934 1934 1934 N/A 1936 N/A 1923 1932 2019 1908 1952 N/A 1937 1911 1912-13 1913 1913 1912 1913 1910 1910 1911 N/A 1964 N/A N/A AMNH D. Del Castillo Paijmans et al., 2021 AMNH AMNH FMNH Paijmans et al., 2021 AMNH Paijmans et al., 2021 East Africa Eritrea Ethiopia Kenya Malawi Somalia Sudan Tanzania 1 8 8 2 1 3 22 SMNH595313 A54019, A54020, F1446, F27006, F32943, F31530, F147633, MFN_MAM_040560 A179130, A34745, A34746, A34747, A88608, F22364, F84800, ZMUC3490 A161733, NMSZ_Z_2004.205 F1444 A238222, F85215, F30779, A42216, A81301, A81302, A85161, A88393, F27279, F127842, MFN_MAM_056356, 3241, 3244, 3243, 4343, 4346, 4443, 4352, 4354 5180 – 5181, 5525, 5519, 5521, 8647 N/A Harrar, Jimabero Harar N/A Sidamo, Boram border Shoa, Addis Ababa N/A Addis Ababa Ngare Ndare Cherangangi Hills Elgeyo Forest N/A Eastern province, Machakos district, Athi river N/A N/A Chibotela N/A Woqooyi Galbeed, Tuyo Plain Nimule National Park Easter; Wad Medani Torit district, Rahad R, Abid N/A Rungwe Serengeti Plains Bamboo Forest Mwanza region, Katungulu Mara region, Bariadi district, Serengeti Plains, Seronera Konde, Pemba Island Maswa, Mbono Maswa, Kimali Ugalla west Selous Ugalla Maswa Selous la1 1911 1921 1896 1926 1926 1929 N/A N/A 1958 1912 1913 1933 1905 N/A 1945 1946 1960 1896 1962 1948 1928 N/A 1929 N/A 1933 1927 1928 N/A 1995 1995 1993-95 1994 1996-97 1996-97 1998 Paijmans et al., 2021 AMNH FMNH Paijmans et al., 2021 AMNH FMNH Paijmans et al., 2021 AMNH Paijmans et al., 2021 FMNH AMNH FMNH AMNH FMNH Paijmans et al., 2021 Pečnerová et al., 2021 South Africa Angola Botswana Namibia South Africa Zambia Zimbabwe 4 4 15 11 18 1 A80611, A80612, A80613, F83654 A169460, F34590, F34591, F35257 A165112, 7934 – 7942, 7943 – 7944, 7946, 7949 A81845, Los, Lyd, Mak, Mal, Mid, Pie, Red, Sab, Sku, SMNH581311 A89842, ZMUC4446, 2469, 2523, 6342, 6344, 6346, 6348, 6349, 6351, 6352 – 6353, 6354, 6355, 6356, 6357, 6358, 6359 F89917 Bie, Chitau Huila Ngamiland, Bushman Pits Chobe, Kabulabula Ngamiland, Mababe Flats Ghanzi Kaokoland Otjiworongo Okahondja Outjo Winchoek Transvaal Loskop Lydenberg Makubu Doornhoek Malelane Middelburg Piet Retief Red Sabi Sands Skukuza Durban N/A N/A Kafue N Luangwa N Luangwa E Nsumbu N Mumbwa C Mfuwe E Zambezi C Kasempa NW Lunga, Luswishi NW Livingstone S Mpulungu Port N Lochnivar S Kabwe C Chiawa C N/A 1925 1954 1950 1930 1930 N/A 1953 1997-98 1998 1996 1997 N/A 2024 1840 1939 1960 1995 1995 N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A AMNH FMNH AMNH FMNH AMNH Pečnerová et al., 2021 AMNH Tensen et al., 2024 Pečnerová et al., 2021 AMNH Paijmans et al., 2021 Pečnerová et al., 2021 FMNH North Africa Morocco 1 SMNH582373 N/A 1933 Paijmans et al., 2021 Middle East Afghanistan Iran Iraq Palestine Turkey Yemen 1 4 1 1 2 2 MFN_MAM_083486 A88720, F57956, F97876, F97877 F98279 MFN_MAM_056095 A69446, M1849-19 F77747, F77749 N/A Bujnurd, Gouladah Mt Feinale, Isfahan Kolahe-Ghazi Sistan and Baluchistan, Damin Bushehr, Ahram Bahgdad N/A Zincurli Ninfi Sea of San’a San’a, Al’Asr 1984 1938 N/A 1962 1963 1952-53 N/A N/A N/A 1951 1951 Paijmans et al., 2021 AMNH FMNH FMNH Paijmans et al., 2021 AMNH MNHN FMNH Asia China India Indonesia Korea Laos Malaysia Sri Lanka Thailand Vietnam 3 7 2 1 1 1 3 2 3 A60106, A265943, F33469 A113745, A54462, F34872, F27443, F91259, M1856, ZMUC29 M1855, M1874 SMNH605501 F31792 Malay F99534, F99535, MFN_MAM_047501 A55553, MFN_MAM_013705 F34366, F46832, MFN_MAM_050746 Shaanxi, Wanhsien Tibet Szechwan, Rombatza Madras, Mavatuva Attinan, Mysore Maharashtra, Chanda district, Allapalli Uttar Pradesh, Kheri Assam, Raimona, Dt Goalpara Cote De Malabar N/A Sumatra Java N/A Phongsaly N/A Meda, Maha, Nuwara Ceylon N/A N/A Cochin, China, Saigon Annam, Darlac province, Ban Me Thuot N/A 1925 1948 1929 1936 1923 1926 1927 N/A N/A 1839 1821 N/A N/A 1929 N/A 1955 N/A 1927 1908 1930 1937 1962 AMNH FMNH AMNH FMNH MNHN Paijmans et al., 2021 MNHN Paijmans et al., 2021 FMNH AMNH FMNH Paijmans et al., 2021 AMNH Paijmans et al., 2021 FMNH Paijmans et al., 2021 Unknown Turkey/Gabon Ethiopia 2 1 2 A69443, A69445 F145984 A119215, A248492 “Gaboon”/Turkey? N/A N/A N/A N/A N/A AMNH FMNH AMNH Table A2. Final sample pools for analysis. List of samples used in each final dataset post-filtering strategies, with corresponding geographic occurrences in latitude and longitude. AMNH = American Museum of Natural History, FMNH = Field Museum of Natural History, MNHN = Le Museum National d’Histoire Naturelle, N/A = specimen tag and/or field notes do not contain the information. * indicates highest sequencing yield individuals included in the subsetted verification of admixture patterns. Total Dist. African Dist. Total Dist. African Dist. Latitude Longitude Source 2469 2469 -14.5 26.2 Pečnerová et al., 2021 2523 2523 -11.9 32.3 Pečnerová et al., 2021 3241 3241 -3.2 34.6 Pečnerová et al., 2021 3243 3243 -3.2 34.6 Pečnerová et al., 2021 3244 3244 -3.2 34.6 Pečnerová et al., 2021 4343 4343 -5.8 31.9 Pečnerová et al., 2021 4346 4346 -5.8 31.9 Pečnerová et al., 2021 4352 4352 -7.8 38.2 Pečnerová et al., 2021 4354 4354 -7.8 38.2 Pečnerová et al., 2021 4443 4443 -5.8 31.9 Pečnerová et al., 2021 5180 5180 -5.8 31.9 Pečnerová et al., 2021 5181 5181 -5.8 31.9 Pečnerová et al., 2021 5519 5519 -3.2 34.6 Pečnerová et al., 2021 5521 5521 -3.2 34.6 Pečnerová et al., 2021 5525 5525 -5.8 31.9 Pečnerová et al., 2021 6342 6342 -12.1111 32.10556 Pečnerová et al., 2021 6344 6344 -11.2131 30.09389 Pečnerová et al., 2021 6346 6346 -15 27 Pečnerová et al., 2021 6348 6348 -13.0667 31.81667 Pečnerová et al., 2021 6349 6349 -15.3 29.6 Pečnerová et al., 2021 6351 6351 -15.3 29.6 Pečnerová et al., 2021 6353 6353 -13.4583 25.83389 Pečnerová et al., 2021 6354 6354 -13.3 26.6 Pečnerová et al., 2021 6355 6355 -8.8 31.1 Pečnerová et al., 2021 6356 6356 -8.76667 31.1 Pečnerová et al., 2021 6358 6358 -14.4 28.5 Pečnerová et al., 2021 6359 6359 -15.1667 29.76667 Pečnerová et al., 2021 7246 7246 10.2 -2.5 Pečnerová et al., 2021 7547 7547 9.2 -2 Pečnerová et al., 2021 7548 7548 9.2 -2.2 Pečnerová et al., 2021 7549 7549 9.2 -2.2 Pečnerová et al., 2021 7934 7934 -20.4 16.7 Pečnerová et al., 2021 7935 7935 -20.4 16.7 Pečnerová et al., 2021 7936 7936 -20.4 16.7 Pečnerová et al., 2021 7937 7937 -20.4 16.7 Pečnerová et al., 2021 7938 7938 -20.4 16.7 Pečnerová et al., 2021 7939 7939 -20.4 16.7 Pečnerová et al., 2021 7940 7940 -20.4 16.7 Pečnerová et al., 2021 7941 7941 -20.4 16.7 Pečnerová et al., 2021 7942 7942 -20.4 16.7 Pečnerová et al., 2021 7943 7943 -22 16.9 Pečnerová et al., 2021 7944 7944 -22 16.9 Pečnerová et al., 2021 7946 7946 -20.1 16.2 Pečnerová et al., 2021 7949 7949 -22.6 17.1 Pečnerová et al., 2021 8540 8540 2.2 31.8 Pečnerová et al., 2021 8647 8647 -7.8 38.2 Pečnerová et al., 2021 A113745 A113745 31.419 76.5938 AMNH A119215 NA NA AMNH A161733 A161733 A161733 A161733 -13.0286 33.9612 AMNH A164151* A164151* A164151* A164151* 9.1429 18.3923 AMNH A165112 A165112 -18 13 AMNH A169460 A169460 A169460 A169460 -20.5176 26.59448 AMNH A170290 A170290 2.933333 11.05 AMNH A170293 A170293 2.933333 11.23333 AMNH A170294 A170294 A170294 A170294 2.45 10.25 AMNH A170295 A170295 A170295 A170295 2.383333 10.03333 AMNH A170296* A170296* A170296* A170296* 2.4069 12.6924 AMNH A170297 A170297 A170297 A170297 3.283333 12.46667 AMNH A170299 A170299 A170299 A170299 2.8 11.63333 AMNH A170300 A170300 A170300 A170300 3.8528 15.1198 AMNH A170306 A170306 A170306 A170306 3.1 12.43333 AMNH A170309 A170309 3.766667 12.06667 AMNH A189390 A189390 A189390 A189390 3.332191 20.30103 AMNH A208770 A208770 A208770 A208770 -2.07232 28.56424 AMNH A213648* A213648* A213648* A213648* 6.3345 -10.6443 AMNH A238222 A238222 3.6133 32.1361 AMNH A248492 A248492 A248492 NA NA AMNH A265943 31.0137 93.1163 AMNH A34745 A34745 -1.4972 36.6726 AMNH A34746 A34746 A34746 -1.6135 36.7174 AMNH A34747 A34747 A34747 A34747 1.073 35.2871 AMNH A42216 A42216 A42216 A42216 -6 35 AMNH A52005 A52005 3.735 29.70972 AMNH A52006 A52006 A52006 A52006 3.735 29.70972 AMNH A52009 A52009 3.735 29.70972 AMNH A52013 A52013 3.735 29.70972 AMNH A52024 A52024 A52024 A52024 3.735 29.70972 AMNH A52026 A52026 A52026 A52026 2.423611 26.44222 AMNH A52028 A52028 A52028 A52028 2.423736 26.44232 AMNH A52038 A52038 A52038 A52038 2.93922 26.82375 AMNH A52039 A52039 A52039 A52039 4.195592 29.48093 AMNH A52041 A52041 A52041 A52041 -4.76532 16.53735 AMNH A52042 A52042 A52042 A52042 -4.76532 16.53735 AMNH A52043 A52043 A52043 A52043 2.17 27.33 AMNH A52044 A52044 A52044 A52044 2.17 27.33 AMNH A52046* A52046* A52046* A52046* 3.332667 29.49336 AMNH A52047 A52047 A52047 A52047 3.33819 28.5389 AMNH A54019 A54019 9.0104 34.7012 AMNH A54020 A54020 A54020 9.0104 34.7012 AMNH A54334 A54334 3.433333 11.75 AMNH A54462* A54462* 12.29266 76.63854 AMNH A60106* A60106* 35.25 109.25 AMNH A69446 A69446 38.0155 35.44235 AMNH A81302* A81302* A81302* A81302* -9.13333 33.66667 AMNH A81845* A81845* A81845* A81845* -25.8667 29.16667 AMNH A85161 A85161 -2.83333 35 AMNH A87236 A87236 A87236 A87236 2.783333 10.53333 AMNH A88393 A88393 A88393 A88393 -8.83333 33.33333 AMNH A88720 A88720 37.4797 57.2698 AMNH A89842 A89842 -14.3333 28.5 AMNH CCNP011 CCNP011 CCNP011 CCNP011 8.833098 -3.82586 D Del Castillo field collection CDRB015 CDRB015 CDRB015 CDRB015 3.175714 12.80592 D Del Castillo field collection F127842* F127842* F127842* F127842* -2.26667 34.78333 FMNH F1444 F1444 9.153122 44.83064 FMNH F145984 F145984 F145984 F145984 9 39.5 FMNH F147633* F147633* F147633* F147633* 9 39.5 FMNH F27006 F27006 9 39.5 FMNH F27279* F27279* F27279* F27279* -2.51667 32.66667 FMNH F27443 28.57992 77.92998 FMNH F27502 F27502 -1.48639 29.43528 FMNH F30779 F30779 12.71667 30.65 FMNH F31530 F31530 9.024325 38.74923 FMNH F31792 F31792 21.75 102.3333 FMNH F34366* F34366* 10.77658 106.7009 FMNH F34590 F34590 F34590 F34590 -17.8333 24.96667 FMNH F34591 F34591 F34591 F34591 -19.2222 23.96833 FMNH F34872* F34872* 19.43172 80.06377 FMNH F46832 12.6594 108.0389 FMNH F57956 F57956 32.39097 51.82404 FMNH F77747 F77747 15.2725 44.28389 FMNH F84800 F84800 1 38 FMNH F85215* F85215* F85215* F85215* 14.39227 33.52804 FMNH F89917* F89917* F89917* F89917* -20.3 30.95 FMNH F91259 F91259 26.17669 90.62634 FMNH F97876 F97876 27.43333 60.78333 FMNH F97877 F97877 28.69833 51.57083 FMNH F98279* F98279* 33.34058 44.40088 FMNH F99534 6.184 81.0828 FMNH F99535 6.184 81.0828 FMNH Los Los -25.456 29.381 Tensen et al., 2024 Lyd Lyd -24.717 30.411 Tensen et al., 2024 M1849-19 38.44749 27.15214 MNHN M1867 M1867 14.5 -14.25 MNHN M1874 -7.49167 110.0044 MNHN M1909-263 M1909-263 M1909-263 M1909-263 6.323056 1.215833 MNHN Mak Mak -24.347 30.275 Tensen et al., 2024 Mal Mal -25.002 31.582 Tensen et al., 2024 MFN_MAM_013705 MFN_MAM_013705 14.28333 100.5 Paijmans et al., 2021 MFN_MAM_040560 MFN_MAM_040560 9.024325 38.74923 Paijmans et al., 2021 MFN_MAM_047501 MFN_MAM_047501 7.75 80.75 Paijmans et al., 2021 MFN_MAM_050746 MFN_MAM_050746 12.8 108.65 Paijmans et al., 2021 MFN_MAM_056095 MFN_MAM_056095 32.58333 35 Paijmans et al., 2021 MFN_MAM_056356 MFN_MAM_056356 -5.91667 35.91667 Paijmans et al., 2021 MFN_MAM_056389 MFN_MAM_056389 -0.71931 11.7713 Paijmans et al., 2021 MFN_MAM_056545 MFN_MAM_056545 MFN_MAM_056545 5.97857 10.17867 Paijmans et al., 2021 MFN_MAM_083486 MFN_MAM_083486 33.40189 66.26564 Paijmans et al., 2021 Mid Mid -25.565 25.565 Tensen et al., 2024 NI-017 NI-017 NI-017 NI-017 7.316667 11.01667 P Henschel field collection NMS_Z_2004 NMS_Z_2004 NMS_Z_2004.205 -13.9669 33.78725 Paijmans et al., 2021 Pie Pie -26.894 30.838 Tensen et al., 2024 Red Red -24.545 30.815 Tensen et al., 2024 Sab Sab -25.447 31.561 Tensen et al., 2024 Sku Sku -25.565 25.565 Tensen et al., 2024 SMNH581240 SMNH581240 1.5925 10.82361 Paijmans et al., 2021 SMNH581311 SMNH581311 -29.8579 31.0292 Paijmans et al., 2021 SMNH582373 SMNH582373 31.63359 -8.00916 Paijmans et al., 2021 SMNH595313 SMNH595313 15.33805 38.93184 Paijmans et al., 2021 SMNH605501 SMNH605501 37.58333 127 Paijmans et al., 2021 ZMUC29 ZMUC29 21.87431 82.00145 Paijmans et al., 2021 ZMUC3490 ZMUC3490 0.113209 38.12935 Paijmans et al., 2021 ZMUC4446 ZMUC4446 -15.4167 29 Paijmans et al., 2021 Table A3. Mitochondrial F ST estimates for African distribution of leopards . F ST measurements for six clades identified in African haplotype network (Figure 3). Column label colors represent groups as illustrated in Figure 3. Gold cells highlight F ST values > 50% allelic differentiation. Bottom of diagonal represents Nei’s F ST (Nei, 1973). Top of diagonal represents Hudson’s F ST (Hudson et al., 1992; Bhatia et al., 2013) SG 0.01568 0.03611 -0.0186 0.13907 0.20332 EAST1 0.35145 -0.0173 -0.0044 0.12473 0.11157 EAST2 0.48413 0.47571 0.00313 0.11518 0.07823 NEAST 0.44344 0.43328 0.52055 0.16235 0.17149 WEST 0.62184 0.61576 0.64131 0.5907 0.15585 SOUTH 0.74234 0.73979 0.75611 0.70664 0.48518 Table A4. Downsampling data sets for intersection . Filtering Strategy across each data set was examined for a minimum threshold percentage of individuals that share called sites across all called genotypes (using -minInd flag in ANGSD genotype calling). Number of sites indicates the total sites retained by each filtering strategy. n 1 = the number of individuals included in the analysis after filtering out samples with less than 5 million mapped reads. n 2 = the final number of individuals retained after individuals with less than 50% informative calls were removed from each filtered strategy data set. * = not measured due to “batch effects”, as observed in PCA plots for each data set (Supplementary Figures A1-A6). Filtering strategies with less than 1,000 sites not considered. In red, the data sets chosen for final admixture and population statistics analyses. In-House extractions + Paijmans et al., 2021 + Pečnerová et al., 2021 Total Distribution No filter minInd 50% minInd 60% minInd 70% minInd 75% minInd 80% minInd 90% minInd 95% minInd 100% 3,942,493 39,131 6,467 1,706 769 361 160 73 9 196 125 * * * In-House extractions + Paijmans et al., 2021 + Pečnerová et al., 2021 African Distribution No filter minInd 10% minInd 20% minInd 30% minInd 40% minInd 50% minInd 75% minInd 90% minInd 100% 3,957,005 3,890,407 3,795,692 3,648,325 3,346,423 429,411 2,026 192 11 152 122 * * * * * * In-House extractions + Paijmans et al., 2021 Total Distribution No filter minInd 10% minInd 20% minInd 30% minInd 40% minInd 50% minInd 60% minInd 70% minInd 75% minInd 80% minInd 85% minInd 90% minInd 95% minInd 100% 2,113,425 2,029,561 1,231,736 110,608 15,020 7,228 3,231 1,168 638 345 168 104 70 13 117 117 117 117 117 117 117 117 68 68 72 74 75 79 81 82 In-House extractions + Paijmans et al., 2021 African Distribution No filter minInd 10% minInd 20% minInd 30% minInd 40% minInd 50% minInd 60% minInd 70% minInd 75% minInd 80% minInd 85% minInd 90% minInd 95% minInd 100% 1,980,923 1,865,970 1,014,269 85,896 16,037 7,683 4,088 1,709 982 518 219 110 74 15 83 43 * * * * * * * In-House extractions Total Distribution No filter minInd 10% minInd 20% minInd 30% minInd 40% minInd 50% minInd 55% minInd 60% minInd 65% minInd 70% minInd 75% minInd 90% minInd 100% 1,649,438 624,574 41,343 10,610 5,190 3,882 2,818 1,780 1,122 728 197 115 10 145 145 145 145 145 145 145 145 145 64 74 77 78 88 88 88 98 98 In-House extractions African Distribution No filter minInd 10% minInd 20% minInd 30% minInd 40% minInd 50% minInd 55% minInd 60% minInd 65% minInd 70% minInd 75% minInd 80% minInd 90% minInd 100% 1,732,592 817,835 97,392 16,685 8,484 4,940 3,418 2,453 1,598 996 684 325 165 29 113 113 113 113 113 113 113 113 113 32 40 47 49 52 55 57 57 57 Table A5. Nuclear F ST estimates for African distribution of leopards . F ST measurements for K=7 admixture populations. Label colors represent groups as illustrated in Figure 4. Bottom of diagonal represents Hudson’s F ST (Hudson et al., 1992; Bhatia et al., 2013). Ne = effective population size; θ = Watterson’s theta. PAR-II A PAR-II B 0.08041 PAR-II C 0.057319 0.052156 PAR-II D 0.088504 0.095556 0.06465 PAR-I A 0.33419 0.391805 0.355578 0.415497 PAR-I B 0.224029 0.26254 0.240866 0.295252 0.447946 PAR-III 0.411332 0.482709 0.434241 0.50691 0.619261 0.53260 Ne 822782.79 334838.45 409028.81 468466.66 233820.80 441390.89 90936.72 Hetero- zygosity 0.0773844 0.0393471 0.0286337 0.0294701 0.0139817 0.035941 0.001539 θ 0.024881 0.0101255 0.0123690 0.0141664 0.0070707 0.013347 0.0027499 Tajima’s D -0.019867 -0.012029 -0.009989 -0.014671 -0.009423 -0.017832 0.0026777 Figure Legends: Figure 1 . Sample Collection . Distribution of samples included in this study (n = 120 present study; n = 19 from Paijmans et al., 2021; n = 50 from Pečnerová et al., 2021; n = 9 from Tensen et al., 2024) across the full extent of Panthera pardus geographic range. Basemap of leopard range downloaded from IUCN Red List https://www.iucnredlist.org/species/15954/274970607 Figure 2. Mitochondrial population structure across the total distribution of leopards. On top, a haplotype network, circles represent unique haplotypes, where the size of each represents the number of individuals, branch lengths represent the number of nucleotide differences between haplotypes, and haplogroups (genetic clades) are identified by colors as labeled. * = P. p. orientalis reference. Below, a map of the geographic distribution of the corresponding samples, color coded by haplogroup. Size of the circle indicates the number of individuals at each occurrence (range = 1 to 9 individuals). Pie charts represent the percentage of different haplogroups present at each occurrence, colors as indicated in the haplotype network. Black circle in Somalia represents sample F1444, which matched with the P. leo outgroup and may represent a mislabeled specimen. Figure 3. Mitochondrial population structure across the African distribution of leopards. On the left, a haplotype network, circles represent unique haplotypes, where the size of each represents the number of individuals belonging to the haplotype, branch lengths represent the number of nucleotide differences between haplotypes, and haplogroups (genetic clades) are identified by colors as labeled. * indicates Moroccan and Senegalese samples that fall within haplogroup clades outside of their geographic proximity. On the right, a map of the geographic distribution of the corresponding samples, color coded by haplogroup assignment. Size of the circle indicates the number of individuals at each occurrence (range = 1 to 9 individuals, larger circles indicate more individuals). Pie charts represent the percentage of different haplogroups present at each occurrence, colors as indicated in the haplotype network. Black circle in Somalia represents sample F1444, which matched with the P. leo outgroup and may represent a mislabeled specimen. Figure 4. Nuclear admixture proportions across leopard geographic extent . (A) Admixture proportions represented in histogram for the Total distribution dataset (n = 1,168 nuclear genomic sites across 82 individuals), labeled by sample ID, organized in order across geographic distribution (labeled by country and geographic region across Africa and by country and subspecies classification across Asia, in colored horizontal bars corresponding to mitochondrial haplogroups). Colors indicate ancestry proportions for K = 10 groups. (B) Map of the sample admixture estimates across their geographic distribution, where each circle represents one sample, and the pie charts indicate the proportions of mixed ancestry. (C) Admixture proportions represented in histogram for the African distribution dataset (n = 3,418 nuclear genomic sites across 57 individuals), labeled by sample ID, organized in order across geographic distribution (labeled by country and geographic region, in horizontal bars). Colors indicate ancestry proportions for K = 7 groups. (D) Map of the sample admixture estimates across their geographic distribution, where each circle represents one sample, and the pie charts indicate the proportions of mixed ancestry. Red boxes in (A), (C), (B) and (D) designate a leopard diversity hotspot, as indicated by high heterozygosity. (E-F) Principal component analyses of leopards across their geographic extent and in Africa . PC plots for the total distribution dataset (E) and the African distribution dataset (F), color coded by sample population assignment corresponding to the admixture plots. PC contributions reported as percentages on each axis. Figure 5. Investigating overlap of mitochondrial and nuclear population assignments in African leopards. (A-C) Maps of the distribution of individuals from nuclear genomic clusters PAR-I (A), PAR-II (B), and PAR-III (C), represented as pie charts of ancestral admixture contributions for each occurrence, overlaid onto the range extents of the mitochondrial Central haplogroup (A), South haplogroup (B), and West haplogroup (C); (D-F) Assignment of individuals to mitochondrial haplogroup (0% or 100%) on x-axis, plotted against the nuclear admixture proportions from PAR-I (D), PAR-II (E), and PAR-III (F) for those same individuals on the y-axis. Colored boxes represent samples with mitochondrial haplogroup assignments (pink, green and blue for Central, South and West haplogroups respectively). Grey boxes represent samples not assigned to the corresponding mitochondrial haplogroups. Purple outlines in (D-F) highlight mito-nuclear discordant assignments. Ne = effective population size and π = heterozygosity for each of the three nuclear clusters PAR-I thru PAR-III (Table 3). Appendix A: Supplementary Figures Figure A1. PCA clustering patterns across serial dilution of shared genomic sites. PCA’s for sample data across in-house extracted samples plus those downloaded from Pečnerová et al., 2021 and Paijmans et al., 2021 for the total distribution of leopards. “minInd” indicates the threshold percentage of samples that need to have an informative call at each site for the site to be included in the dataset. As the number of required shared sites across samples increases, the number of called sites decreases. Pečnerová samples, in purple, largely do not share overlapping sites with the other data sets, and were therefore removed from the nuclear analyses. Figure A2. PCA clustering patterns across serial dilution of shared genomic sites. PCA’s for sample data across in-house extracted samples plus those downloaded from Pečnerová et al., 2021 and Paijmans et al., 2021 for the African continental distribution of leopards. “minInd” indicates threshold percentage of samples that need to have an informative call at each site for the site to be included in the dataset. As the number of required shared sites across samples increases, the number of called sites decreases. Pečnerová samples, in purple, largely do not share overlapping sites with the other data sets, and were therefore removed from the analyses. Figure A3. PCA clustering patterns across serial dilution of shared genomic sites. PCA’s for sample data across in-house extracted samples plus those downloaded from Paijmans et al., 2021 for the total distribution of leopards. “minInd” indicates threshold percentage of samples that need to have an informative call at each site for the site to be included in the dataset. As the number of required shared sites across samples increases, the number of called sites decreases. Figure A4. PCA clustering patterns across serial dilution of shared genomic sites. PCA’s for sample data across in-house extracted samples plus those downloaded from Paijmans et al., 2021 for the African continental distribution of leopards. “minInd” indicates threshold percentage of samples that need to have an informative call at each site for the site to be included in the dataset. As the number of required shared sites across samples increases, the number of called sites decreases. Figure A5. PCA clustering patterns across serial dilution of shared genomic sites. PCA’s for sample data across in-house extracted samples for the total distribution of leopards. “minInd” indicates threshold percentage of samples that need to have an informative call at each site for the site to be included in the dataset. As the number of required shared sites across samples increases, the number of called sites decreases. Figure A6. PCA clustering patterns across serial dilution of shared genomic sites. PCA’s for sample data across in-house extracted samples for the African continental distribution of leopards. “minInd” indicates threshold percentage of samples that need to have an informative call at each site for the site to be included in the dataset. As the number of required shared sites across samples increases, the number of called sites decreases. Figure A7. Calculating Best K . Log Likelihoods (BestLn(K)) were recorded across ten replicates for each K from 2:8 for the African distribution and 2:15 for the total distribution of leopards. The highest likelihoods from each group of replicates are plotted in the top three panels. DeltaK was measured for each of these and plotted as shown in the bottom panels. Figure A8. Admixture plots for K=3 thru K=10. Admixture proportions represented in histograms for the Total distribution dataset (n=1168 nuclear genomic sites), labeled by sample ID, organized in order across geographic distribution (labeled by country and geographic region across Africa and by country and subspecies classification across Asia, in horizontal bars). Figure A9. Admixture plots for K=3 thru K=7. Admixture proportions represented in histogram for the African distribution dataset (n=3418 nuclear genomic sites), labeled by sample ID, organized in order across geographic distribution (labeled by country and geographic region, in horizontal bars). Figure A10. Principal component analyses. PC plots for the total distribution dataset for PCs 3-8, color coded by sample population assignment corresponding to the admixture plots (Individuals assigned to population groups based on the majority proportion of admixture contributions within each individual). PC contributions reported as percentages in parentheses on each axis. (A) PC3 on x-axis, PC4 on y-axis (B) PC5 on x-axis, PC6 on y-axis (C) PC7 on x-axis, PC8 on y-axis. Figure A11. Principal component analyses. PC plots for the African distribution dataset for PCs 3-8, color coded by sample population assignment corresponding to the admixture plots (Individuals assigned to population groups based on the majority proportion of admixture contributions within each individual). PC contributions reported as percentages in parentheses on each axis. (A) PC3 on x-axis, PC4 on y-axis (B) PC5 on x-axis, PC6 on y-axis (C) PC7 on x-axis, PC8 on y-axis. Figure A12. Nuclear population structure based on a subset of high quality samples only. (A) Admixture proportions represented in histogram across top for the Total distribution subsetted dataset (n = 16,903 nuclear genomic sites), labeled by sample ID, organized in order across geographic distribution (labeled in horizontal bars by country and geographic region across Africa and by country and subspecies classification across Asia). Colors indicate ancestry proportions for K = 10 groups. (B) PCA plot for the total distribution subsetted dataset. PC contributions reported in percentages next to axis labels. Figure A13. Admixture plots for K=3 thru K=10. Admixture proportions represented in histograms for the high-quality, subsetted distribution dataset (n=16,903 nuclear genomic sites across 16 individuals), labeled by sample ID, organized in order across geographic distribution (labeled by country and geographic region across Africa and by country and subspecies classification across Asia, in horizontal bars). Figure A14. Genetic distance trees based on mitochondrial and nuclear SNPs of African leopards. Comparison of branch position for each individual across distance trees, mitochondrial on the left, nuclear on the right. * = samples A52041, A52044, see description in text. Bootstrap support indicated at each node. Branches colored by population assignments as observed in mitochondrial haplotype networks, and nuclear admixture plots, respectively. Data Accessibility: Raw sequence data are available with accession numbers XXX-XXX. Competing Interests Statement: The authors declare no competing interests Author Contributions : Danielle Del Castillo: Conceptualization (lead); Methodology (lead); Data Curation (equal); Investigation; Formal Analysis; Visualization; Writing - original draft (lead); Writing - review and editing (equal). Corey Anco: Data Curation (equal); Writing - original draft (supporting); Writing - review and editing (equal). Seth W. Cunningham : Data Curation (equal); Writing - original draft (supporting); Writing - review and editing (equal). Alexis Neffinger: Data Curation (equal). Faruk Mamugy : Writing - review and editing (equal). Arame Ndiaye: Writing - review and editing (equal). Hana Raza: Writing - review and editing (equal). Kamta Tchoffo Roméo Omer: Writing - review and editing (equal). Evon Hekkala: Conceptualization (supporting); Methodology (supporting); Data Curation (equal); Resources. Writing - original draft (supporting); Writing - review and editing (equal). Laura Bertola: Conceptualization (supporting); Methodology (supporting); Data Curation (equal); Writing - original draft (supporting); Writing - review and editing (equal). Acknowledgements: We would like to thank Fordham University Graduate School of Arts and Sciences, Department of Biological Sciences for academic, laboratory and professional support, as well as Fordham University’s Advanced Research Computing, Education Technologies and Research Computing department, a division of the Office of Information Technology for providing their assistance and access to research computing resources (https://www.fordham.edu/advanced-research-computing) and for which Andrew Angelopoulos and Upendar Thaduri were very helpful. We thank the Henry Luce Foundation for their generous Clare Booth Luce Fellowship award and professional development funds, as well as the Freedman Fellowship, which supported our laboratory benchwork. We thank The American Museum of Natural History for access to archival collections and sampling. Additionally, AMNH provided use of their ancient biomolecular clean room laboratory facility for handling and processing of sample tissues, and their HPC server for bioinformatics and computational resources, both operated by the Institute for Comparative Genomics and the Information Technology Department at AMNH, and supported by the Simons Foundation and generous donor support to the Museum. We are thankful as well for bioinformatics support from Sajesh Singh and Dean Bobo at AMNH. We also thank The Field Museum of Natural History and Le Muséum National d’Histoire Naturelle for access to archival collections and sampling. We acknowledge Philip Henschel of Panthera (Panthera.org) for donated scat samples. Our field-collection work was supported by Dr. Adam Jakab and Mr. Bella Barry of the Guinean Park Foundation; Dr. K. E. Linsenmair, Dr. N’Golo Kone, Dr. Erik Frank, Dr. Juan LaPuente, Mr. David Kouamé Kouassi, Mr. Koffi Kouadio, and local guides Oumar and Max of the Comoé National Park Research Station - University of Würzburg; Dr. Kevin Njabo, Dr. Eric Nana, Ms. Ruth Nsang, and Mr. Chi Elvis of the Congo Basin Institute (CBI) - University of California, Los Angeles (UCLA) and the International Institute of Tropical Agriculture (ITTA); Mr. Alain Djao Lamwe of the Cameroon Ministry of Forestry and Wildlife (MINFOF) and local guides Jean Jaques Avoto and Biango; and Dr. Benjamin Barca and local guide Sheriff of the Gola Rainforest Conservation Center (GRCC). Lastly, thanks to the park conservators and eco-guards for their field support, as well as government Ministries for granting research and export permits and providing access to the protected areas. Information & Authors Information Version history V1 Version 1 16 March 2026 Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords molecular evolution molecular genetics sequencing terrestrial vertebrate Authors Affiliations Danielle Del Castillo 0009-0004-8646-921X Fordham University View all articles by this author Corey Anco Buffalo Bill Center of the West View all articles by this author Seth Cunningham American Museum of Natural History View all articles by this author Alexis Neffinger Fordham University View all articles by this author Faruk Mamugy Sveriges lantbruksuniversitet Fakulteten for Skogsvetenskap View all articles by this author Arame Ndiaye 0000-0001-9403-9130 TRACE Wildlife Forensics Network View all articles by this author Hana Raza 0000-0003-1482-1718 Leopards Beyond Borders View all articles by this author Kamta Tchoffo Roméo Omer Panthera View all articles by this author Evon Hekkala Fordham University View all articles by this author Laura D. Bertola [email protected] Tata Institute of Fundamental Research National Centre for Biological Sciences View all articles by this author Metrics & Citations Metrics Article Usage 230 views 75 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Danielle Del Castillo, Corey Anco, Seth Cunningham, et al. Range-wide phylogeographic analyses of leopards (Panthera pardus) reveal African mito-nuclear discordance and previously unrecognized diversity. Authorea . 16 March 2026. DOI: https://doi.org/10.22541/au.177366294.41173847/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.177366294.41173847/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'9fe05c005b19c13d',t:'MTc3OTE2NDk1Mw=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();