The existence of distinct traits in island versus mainland populations offers opportunities to gain insights into how eco-evolutionary processes operate under natural conditions. We used two island colonization events in the white-winged fairywren (Malurus leucopterus) to investigate the genomic and demographic origin of melanic plumage. This avian species is distributed across most of Australia, and males of the mainland subspecies (M. l. leuconotus) exhibit a blue nuptial plumage in contrast to males of two island subspecies – M. l. leucopterus on Dirk Hartog Island and M. l. edouardion Barrow Island – that exhibit a black nuptial plumage. We used reduced-representation sequencing to explore differentiation and demographic history in this species and found clear patterns of divergence between mainland and island populations, with additional substructuring on the mainland. Divergence between the mainland and Dirk Hartog was approximately 10 times more recent than the split between the mainland and Barrow Island, supporting two independent colonizations. In both cases, estimated gene flow between the mainland and the islands was low, contributing to signals of divergence among subspecies. Our results present demographic reconstructions of mainland-island dynamics and associated plumage variation in white-winged fairywrens, with broader implications regarding our understanding of convergent evolution in insular populations.

We obtained tissue samples from the Western Australian Museum, the South Australian Museum, and the Burke Museum as well as blood collected during fieldwork for 104 M. leucopterus individuals (see Table S1 in main manuscript). Genomic DNA was isolated using a DNeasy Extraction Kit (Qiagen, Valencia, California, USA) according to the manufacturer protocol. Double-digested restriction-site associated DNA markers (ddRADtags) were generated following the protocol of Peterson et al. (2012) with modifications described in Thrasher et al. (2018). All reads were trimmed to 97 bp and then filtered for quality using the FASTX-Toolkit(http://hannonlab.cshl.edu/fastx_toolkit). We removed sequences if a single base had a Phred quality score less than 10 and if more than 95% of bases had a Phred quality score less than 15. Sequences were assembled using a red-backed fairywren (Malurus melanocephalus) reference genome using stacks v 2.41. Sequences were aligned to the M. melanocephalus assembly with bowtie2 version 2.2.2 (Langmead et al. 2009). We required that a SNP be present in a minimum of 80% of the individuals and applied a minor allele frequency filter of 5%. We created a subset of SNPs that included only the first SNP per stack (6,497 SNPs). This data set was used for all analyses, with the exception of the PCA.