The sex chromosomes often follow unusual evolutionary trajectories. In particular, the sex-limited Y and W chromosomes frequently exhibit a small but unusual gene content in numerous species, where many genes have undergone massive gene amplification. The reasons for this remain elusive with a number of recent studies implicating meiotic drive, sperm competition, genetic drift and gene conversion in the expansion of gene families. However, our understanding is primarily based on Y chromosome studies, and the W chromosome has been largely overlooked. Here, we conduct a comprehensive investigation into the abundance, variability, and evolution of ampliconic genes on the W both across and within avian species. We find a striking deficit of gene families on the duck W chromosome, as well as conservation in W-linked gene copy number across duck breeds, indicating that gene amplification may not be such a general feature of sex chromosome evolution as Y studies would initially suggest. Furthermore, we show that gene families have expanded independently in the duck and chicken. In particular, using contrasts between modern chicken and duck breeds selected for different female-specific selection regimes and their wild ancestors, we investigate the factors driving the expansion of HINTW, a prominent ampliconic gene family hypothesized to play a role in female reproduction and oogenesis. While we find that HINTW is ampliconic in both species, our results support a role of female-specific selection in driving gene amplification in the chicken but not the duck, challenging the assumption that HINTW is key for female fecundity across the avian phylogeny.

The NanoString nCounter assay was performed according to standard protocol. Briefly, at least 300ng of DNA per sample was fragmented via AluI digestion and then hybridized to the custom CNV CodeSet. Samples included three females and two males from each of the selectively bred chicken and duck breeds, and two female and two male samples from each of the modern ancestor breeds. Samples were distributed randomly over the CNV CodeSets to avoid batch effects. The nCounter Digital Analyzer was used to count and quantify signals of reporter probes. Data analysis was performed using the nSolver Analysis Software.