Phylogeographic studies of continental clades, especially when combined with palaeoclimate modelling, provide powerful insight into how environment drives speciation across climatic contexts. Australia, a continent characterized by disparate modern biomes and dynamic climate change, is a model system for reconstructing the impact of past and present environments on diversification. Here we use genomic-scale data (1310 exons and whole mitogenomes from n = 111 samples) to investigate Pleistocene diversification, cryptic diversity, and secondary contact in the Australian delicate mice (Hydromyini: Pseudomys), a recent radiation spanning almost all Australian environments. Across northern Australia, we find no evidence for introgression between cryptic lineages within Pseudomys delicatulus sensu lato, with palaeoclimate models supporting contraction and expansion of suitable habitat since the last glacial maximum. Despite multiple contact zones, we also find little evidence of introgression at a continental scale, with the exception of a potential hybrid zone in the mesic biome. In the arid zone, combined insights from genetic data and palaeomodels support a recent expansion in the arid specialist P. hermannsburgensis, and contraction in the semi-arid P. bolami. In the face of repeated secondary contact, differences in sperm morphology and chromosomal rearrangements are potential mechanisms that maintain species boundaries in these recently diverged species. Additionally, we describe the western delicate mouse as a new species and recommend taxonomic reinstatement of the eastern delicate mouse. Overall, we show that speciation in an evolutionarily young and widespread clade has been driven by environmental change, and potentially maintained by divergence in reproductive morphology and chromosome rearrangements.

Molecular Data Collection

For frozen and ethanol preserved tissue samples, we extracted total genomic DNA using a standard Qiagen DNeasy Blood and Tissue Kit. For historical museum specimens, we used a phenol-chloroform extraction protocol optimised for historical DNA (Roycroft, MacDonald, et al., 2021; Roycroft, Moritz, et al., 2022). We prepared DNA libraries following a modified (Roycroft, Moussalli, et al., 2020) Meyer and Kircher (2010) protocol. Genomic libraries from historical specimens were treated with PreCR Repair Mix (New England BioLabs) to repair degraded DNA, following the manufacturer’s instructions. Samples were uniquely indexed with dual barcodes, and up to 96 sample libraries were pooled and hybridised using the custom sequence capture probe design for murine rodents described in Roycroft, Moussalli, et al., (2020). This probe set targets 1.27Mb of genomic DNA (1417 exons, SeqCap EZ Developer Library; Roche NimbleGen). Post-capture libraries were sequenced on either an Illumina HiSeq 2500 or Novaseq SP platform (100 bp paired-end reads) at the Australian National University’s Biomolecular Resource Facility.

Bioinformatic Processing

Raw sequence data were processed using Exon Capture Pipeline for Phylogenetics (ECPP, https://github.com/Victaphanta/ECPP), following the settings described in Roycroft, Moussalli, et al., (2020), using the high-quality de novo mapping approach described in Roycroft, Moritz, et al., (2022). Targets were aligned using MAFFT (Katoh & Standley, 2013) and processed with BMGE (Criscuolo & Gribaldo, 2010) to remove poorly represented regions. We excluded loci with average heterozygosity >3% across samples (putative paralogs) and filtered the final dataset to include only loci that were >90% sample complete. The final filtered exon dataset consisted of 1310 exons, making up a combined nucleotide alignment of 1.15 Mb. 

See Methods section of associated manuscript for full details and citations.