Phylogenomic analyses of all species of swordtail fishes (genus Xiphophorus) show that hybridization preceded speciation

Abstract

Hybridization has been recognized to play important roles in evolution, however studies of the genetic consequence are still lagging behind in vertebrates due to the lack of appropriate experimental systems. Fish of the genus Xiphophorus are proposed to have evolved with multiple ancient and ongoing hybridization events. They have served as an informative research model in evolutionary biology and in biomedical research on human disease for more than a century. Here, we provide the complete genomic resource including annotations for all described 26 Xiphophorus species and three undescribed taxa and resolve all uncertain phylogenetic relationships. We investigate the molecular evolution of genes related to cancers such as melanoma and for the genetic control of puberty timing, focusing on genes that are predicted to be involved in pre-and postzygotic isolation and thus affect hybridization. We discovered dramatic size-variation of some gene families. These persisted despite reticulate evolution, rapid speciation and short divergence time. Finally, we clarify the hybridization history in the entire genus settling disputed hybridization history of two Southern swordtails. Our comparative genomic analyses revealed hybridization ancestries that are manifested in the mosaic fused genomes and show that hybridization often preceded speciation.

Fig. 1. Phylogenetic trees constructed using a maximum-likelihood method for the mitochondrial genome (~15 kb, left) and nuclear genome sequences (~342 Mb, right).

Numbers on the nodes represent bootstrap support values; nodes without numbers means that they are 100% supported.

Fig. 2. Phylogenetic trees showing incongruent topology from coalescent- and concatenated-based methods.

A Alignment of 10,000 trees constructed from recombination-free genetic loci showing the phylogenetic discordance across the genome. B A coalescent-based phylogeny constructed with weightedASTRAL. C A concatenated-based phylogenetic tree constructed by using the maximum-likelihood method.

Fig. 3. Extensive reticulate hybridization during evolution of the genus Xiphophorus.

A Heatplot showing the estimated admixture fraction between two species on the x- and y-axis. Species are listed on the x- and y-axis in the species tree. The values in the matrix (f_b) were calculated using Dsuite with the f-branch statistic, referring to excess allele sharing between the branch identified on the expanded tree on the y-axis (relative to its sister branch) and the species identified on the x-axis, indicating the hybrid origin of these alleles. B Reticulate evolution of the X. clemenciae, X. mixei and X. monticolus clade estimated using SNaQ with h = 1. C Reticulate evolution of lineage X. xiphidium estimated using SNaQ with h = 1.

Fig. 4. Dot plots revealing genome regions derived from hybridization for species with or without hybrid history.

Trees on the right are in the format of ((P1, P2), P3), where the admixture signal between P2 and P3 was investigated. Values (f_dM) on the y-axis were calculated using Dsuite with modified f_d statistics developed by Malinsky et al., referring to the fraction of excess alleles in sliding-windows sharing between P2 and P3 than between P1 and P3, indicating hybridization origin of the genome region. Statistics were averaged over windows of 100 informative SNPs, moving forward by 50 SNPs. Row A refers to a genome without hybridization resources.