Dynamics of sex chromosome evolution in a rapid radiation of cichlid fishes

Abstract

Sex is a fundamental trait determined by environmental and/or genetic factors, including sex chromosomes. Sex chromosomes are studied in species scattered across the tree of life, yet little is known about tempo and mode of sex chromosome evolution among closely related species. Here, we examine sex chromosome evolution in the adaptive radiation of cichlid fishes in Lake Tanganyika. Through the analysis of male and female genomes from 244 cichlid taxa (189 described species with 5 represented with two local variants/populations; 50 undescribed species) and of 396 multitissue transcriptomes from 66 taxa, we identify signatures of sex chromosomes in 79 taxa, involving 12 linkage groups. We find that Tanganyikan cichlids have the highest rates of sex chromosome turnover and heterogamety transitions known to date. We show that sex chromosome recruitment is not at random. Moreover convergently emerged sex chromosomes in cichlids support the “limited options” hypothesis of sex chromosome evolution.

Fig. 1.. Sex chromosome evolution in the adaptive radiation of cichlid fishes in LT.

Sex chromosome state and ancestral state reconstruction in LT cichlids (permissive dataset) are placed on a time-calibrated species tree (17), with background color shading indicating the 13 cichlid tribes of the LT radiation. The inner circle at tips shows identified sex-linked LGs. Colors refer to the LGs of the reference genome [23 LGs in the used reference genome represent the 22 chromosomes of O. niloticus, the most common chromosome number for African cichlids (21)]; two- or multi-colored symbols at tips indicate sex chromosomal signals that were detected on two or more reference LGs, suggesting chromosomal rearrangements between O. niloticus and LT cichlids. Symbols at the outer circle indicate the heterogametic SD system identified in each species (square, XY; triangle, ZW; no symbol, no system has been identified). Empty circles at tips indicate that within the dataset used here, no sex chromosome could be identified, nor has, to the best of our knowledge, genetic SD been investigated in these species elsewhere. Pie charts at nodes represent the probability for an LG being a sex chromosome at this time derived from ancestral state reconstructions.

Fig. 2.. Nonrandom sex chromosome distribution in LT cichlids.

(A) Use of different LGs as sex chromosomes. Bars represent the number of times a reference genome LG has been detected as sex-linked at the species level with colors referring to tribe (permissive dataset). (B) The occurrence of SD systems. Bars represent how often an XY or ZW heterogametic SD system was identified at the species level (permissive dataset) and with colors referring to tribe. (C) Association between species richness and sex chromosome turnover. The number of sex chromosome turnovers leading to the tips of each tribe (permissive dataset) is associated with the number of species investigated in each tribe (pGLS, P = 0.0043, coefficient = 0.039). Dots are colored according to tribes; the line represents the linear model fitted to the data. (D) Boxplots showing the expected number of sex chromosome recruitments if recruitment was at random (10,000 permutations). Boxplot centerlines represent the median, box limits the upper and lower quartiles, and whiskers the 1.5× interquartile range. Outliers are not shown. Ten reference LGs were never implicated in a turnover event in LT cichlids. Under random recruitment in the simulations, this pattern occurred only in 2.01% of all simulations. Yellow dots indicate the number of observed sex chromosome recruitments per reference genome LG derived from ancestral state reconstructions (permissive dataset), gray background shading represents chromosome length in megabases derived from the reference genome, and numbers below each boxplot indicate the number of previously described sex-determining genes on these reference genome LGs.

Fig. 3.. Convergent evolution of LG19 as XY sex chromosome in two Tanganyikan cichlid tribes.

The phylogenetic tree of X and Y haplotype sequences does not group the P. paradoxus Y haplotype with the Tropheus Y haplotypes but supports the species tree, suggesting convergent evolution. The Y haplotype of the non-LT riverine haplochromine O. indermauri groups with the Y haplotypes of the Tropheus species, supporting monophyly of this sex chromosomal system. The scale bar indicates the number of substitutions per site; values at nodes represent bootstrap support (% of 1000 bootstraps, if no value is shown the node support was 100%).

Fig. 4.. Sex chromosome differentiation in LT cichlids.

(A) Size distribution of sex-differentiated regions. The size of these regions corresponds to the proportion of the reference genome LG with windows that have more sex-specific SNPs than two times the mean across all windows. (B) Per-species proportion of the chromosome(s) showing sex differentiation and corresponding estimated ages of the sex chromosomal system based on ancestral state reconstructions on a time-calibrated species tree. The degree of differentiation is not associated with the estimated age of origin (pGLS, P = 0.9049, coefficient = 0.0011).

Fig. 5.. Sex chromosome evolution in African cichlids.

Phylogenetic relationships in African cichlids are based on previous studies (17, 19). Sex chromosome occurrences are denoted with reference to the 22 chromosomes of the genome of the Nile tilapia (O. niloticus, tribe Oreochromini). Note that the naming of the chromosomes relates to previous naming of LGs and is missing “21” because LG21 became part of LG16 in the course of establishing chromosome length genome assemblies. “B” refers to B chromosomes, i.e., supernumerary chromosomes found in some cichlid species. Nile tilapia strains exist with an XY heterogametic system on LG1 and an XY heterogametic system with a Y-specific amh SD gene on LG23, respectively (52)). Cichlid lineages of LT are indicated with black branches, cichlids from other lakes or rivers with gray branches. Sex chromosome information is derived from this study and previously published summaries (19, 20) that are also based on the 22 O. niloticus chromosomes as the most common African cichlid karyotype (21).