Massive inter-species introgression overwhelms phylogenomic relationships among jaguar, lion, and leopard.

Abstract

Phylogenomic analyses of closely related species allow important glimpses into their evolutionary history. Although recent studies have demonstrated that interspecies hybridization has occurred in several groups, incorporating this process in phylogenetic reconstruction remains challenging. Specifically, the most predominant topology across the genome is often assumed to reflect the speciation tree, but rampant hybridization might overwhelm the genomes, causing that assumption to be violoated. The notoriously challenging phylogeny of the five extant Panthera species (specifically jaguar [P. onca], lion [P. leo], and leopard [P. pardus]) is an interesting system to address this problem. Here we employed a Panthera-wide whole-genome-sequence dataset incorporating three jaguar genomes and two representatives of lions and leopards to dissect the relationships among these three species. Maximum-likelihood trees reconstructed from non-overlapping genomic fragments of four different sizes strongly supported the monophyly of all three species. The most frequent topology (76-95%) united lion+leopard as sister-species (topology 1), followed by lion+jaguar (topology 2: 4-8%) and leopard+jaguar (topology 3: 0-6%). Topology 1 was dominant across the genome, especially in high-recombination regions. Topologies 2 and 3 were enriched in low-recombination segments, likely reflecting the species tree in the face of hybridization. Divergence times between sister-species of each topology, corrected for local recombination rate effects, indicated that the lion-leopard divergence was significantly younger than the alternatives, likely driven by post-speciation admixture. Introgression analyses detected pervasive hybridization between lions and leopards, regardless of the assumed species tree. This inference was strongly supported by multi-species-coalescence-with-introgression (MSci) analyses, which rejected topology 1 or any model without introgression. Interestingly, topologies 2 and 3 with extensive lion-leopard introgression were unidentifiable, highlighting the complexity of this phylogenetic problem. Our results suggest that the dominant genome-wide tree topology is not the true species tree but rather a consequence of overwhelming post-speciation admixture between lion and leopard.