Genomic evidence for homoploid hybrid speciation in a marine mammal apex predator

Abstract

Hybridization is widespread and constitutes an important source of genetic variability and evolution. In animals, its role in generating novel and independent lineages (hybrid speciation) has been strongly debated, with only a few cases supported by genomic data. The South American fur seal (SAfs) Arctocephalus australis is a marine apex predator of Pacific and Atlantic waters, with a disjunct set of populations in Peru and Northern Chile [Peruvian fur seal (Pfs)] with controversial taxonomic status. We demonstrate, using complete genome and reduced representation sequencing, that the Pfs is a genetically distinct species with an admixed genome that originated from hybridization between the SAfs and the Galapagos fur seal (Arctocephalus galapagoensis) ~400,000 years ago. Our results strongly support the origin of Pfs by homoploid hybrid speciation over alternative introgression scenarios. This study highlights the role of hybridization in promoting species-level biodiversity in large vertebrates.

Fig. 1.. Sampling setup and population structuring of the focal taxa.

(A) Distribution of fur seal species and localities assessed in this study. From the left to the right: New Zealand fur seal (NZfs), A. forsteri, green; Galapagos fur seal (Gfs), A. galapagoensis, pink; Peruvian fur seal (Pfs) A. australis, hatched blue and pink lines; and South American fur seal (SAfs), A. australis, light blue. (B) Principal components analysis (PCA) of superimposed whole-genome resequencing (stars) and double digest restriction site–associated DNA sequencing (ddRAD-seq) (circles) data. (C) PCA of ddRAD-seq data with only the three focal taxa. WGS, whole-genome shotgun. (D) Admixture plots based on ddRAD-seq data with K = 3 and 4. Samples from Isla Foca are indicated in yellow in (B) and (C) and are specified in (D).

Fig. 2.. Species trees and divergence time among fur seals species.

(A) Species tree estimated with ASTRAL-III based on 7810 independent gene trees. Numbers are node support. (B) StarBEAST2 species tree with point estimates of divergence times based on 300 independent loci of 50 kb. Node numbers are the divergence times (million years ago). Blue bars represent the 95% confidence interval. The bar at the bottom of the phylogeny is a scale bar for the confidence interval. All nodes with support = 1. NZfs, A. forsteri; Gfs, A. galapagoensis; Pfs, A. australis; SAfs, A. australis; and the outgroup, the Afs, A. gazella.

Fig. 3.. Topology weighting across the genome for the origin of the Pfs.

(A) The main three alternative topologies for the position of the Pfs and (B) their relative weightings across the whole genomes. (C) Smoothed weights across the five largest scaffolds (delimited by the plus signals).

Fig. 4.. Peruvian fur seals are a hybrid species.

(A) f-branch results. The values in the matrix refer to excess allele sharing (f_b statistics) between the tree branches in the y axis and the species/populations on the x axis. Dotted lines in the phylogeny represent ancestral lineage, and nondotted lines represent extant lineages. The asterisks indicate significant results (P < 0.01). Gray cells are inestimable relationships. AR (Argentina), FK (Falklands), and CH (Chile). (B) TreeMix maximum likelihood phylogeny with the standard error (s.e.) and the direction of introgression (arrows) and (C) admixture graph supported the Pfs as a mixed genome between the Gfs and the SAfs. Dotted lines are the direction of introgression, and numbers are the percentage of introgression with their respective confidence intervals in gray. SAfs from AR and FK are combined as an Atlantic population (B). An introgression between the NZfs and the SAfs was also detected. NZfs is the basal taxon.

Fig. 5.. Test of demographic scenarios for the origin of the Pfs.

(A) The HHS (scenario 1) and the three alternative introgression scenarios tested. Ts and Tg are the duration (in generations) of the intermediate subpopulation/s in the non–homoploid hybrid speciation (HHS) scenarios. (B) Model checking showing that the model fits the data well. (C) Posterior probability (point and 95% confidence interval) of each scenario (scenario choice) computed with logistic regression. (D) Posterior distributions of parameters Ts and Tg in scenarios 2 to 4 showing that they approached the minimum possible value of 0 (x axis).

Fig. 6.. Inferred demographic parameters.

Divergence times in million years (left) and ancestry percentages (above the branches) as estimated with BP&P. Below the numbers are the confidence intervals.