Filogenômica do gênero Panthera (mammalia, felidae)

Abstract

The evolutionary history of species, especially those derived from rapid and recent radiations, is often embedded in a complex network of past admixtures process. This complex history, associated with the stochastic nature of incomplete lineage sorting (ILS), poses challenges for reconstructing their phylogenetic relationships, even with the use of whole-genome-sequence (wgs) data. Several studies have addressed the evolutionary relationships among the five extant species of Panthera, and recent analyses have revealed extensive genealogical discordance in this group, making it difficult to discern which topology reflects the original sequence of speciation events. So far, phylogenomic studies of Panthera employing wgs data have included a single individual per species, which precludes an assessment of replicability of topological patterns when the species’ representative is changed. Here we employ a Panthera-wide wgs dataset incorporating three jaguar genomes (two of which are novel), along with two representatives each of lions and leopards, to dissect the relationships among these three species. We tested the genome-wide monophyly of each of them, and then investigated their patterns of genealogical discordance. We initially assessed the frequency of each of the three alternative topologies along their genomes, by breaking up the alignment into windows of four different sizes (50 kb, 100 kb, 1 Mb and 5 Mb) and reconstructing the ML phylogeny (tested with nonparametric bootstrapping) for each window. Supported ML phylogenies were then plotted along the genome, and the frequency of each topology was assessed relative to the recombination rate of the respective genomic segment. Using 100-kb windows, we also estimated divergence times between the two sister-species defined by each topology, as well as a ‘normalized age’ in which the depth of the sisterspecies node was divided by that of the preceding (trio) node, to further correct for recombination-rate effects. We also employed 100-kb windows to identify and map introgression events along the genome, assuming different topologies as the ‘species tree’. Genome-wide monophyly for all three tested species was strongly supported, and a clear pattern was observed with respect to their phylogenetic relationships: with all window sizes, the most frequent topology (76% – 95%) united lion+leopard as sisterspecies (topology 1), followed by lion+jaguar (topology 2: 4% – 8%) and leopard+jaguar (topology 3: 1% – 6%). Topology 1 was strongly predominant in regions of high recombination, e.g. terminal chromosomal segments, whereas topologies 2 and 3 were strongly enriched in low-recombination regions, especially in centromeric segments. Absolute divergence times were younger for topologies 2 and 3 (relative to topology 1) on both autosomes and the X chromosome, but this pattern was reversed when we 19 ‘normalized’ the depth of the terminal node using the age of the preceding (trio) node. The introgression analyses indicated pervasive historical admixture among these species, regardless of the assumed species tree. For example, if topology 1 was assumed, 5% of the genome was inferred to derive from lion-jaguar hybridization, while if topology 2 or 3 were assumed, 35% of the genome was detected as introgressed between lion and leopard. Overall, our results indicate that topology 2 (lion+jaguar) most likely reflects the original speciation relationship, given its age, frequency and recombination profile. The jaguar+leopard topology did not differ significantly in age or recombination profile, and thus likely derives from ILS. Remarkably, these results imply that a large proportion of the genome has been overwritten by post-speciation admixture between lion and leopard, leading to a complex mosaic whose phylogenetic resolution requires integration among different approaches.