Sampling and genomic data acquisition
Holdfast epibiota were collected from intertidal Durvillaeadistributed along the uplifted rocky shores of the Akatore region as well as its northern and southern flanking non-uplifted coasts (Fig. 1). This spatial sampling of invertebrates was closely aligned to previous sampling regimes for their Durvillaea host species to facilitate precise genomic comparisons of both macroalgae and epifauna. Accordingly, we carried out a fine-scale sampling (sampling every 2-6 km to exclude non-habitable sandy beaches) from within an approximately 100 km of coastline (Figure 1; Table S1). As our studied invertebrates can move between D. poha and D. antarctica (but not D. willana , which has a more dense holdfast free of burrowing invertebrates) we randomly collected samples from both hosts at different sites. To reduce the effect of including closely related individuals in our genetic analyses we analyzed samples from multiple holdfasts at each site (Nikula et al. 2010). Details of laboratory procedures and single nucleotide polymorphism (SNP) acquisition are described in the Supplementary Methods and Table S2. The final high quality SNP dataset consisted of 13991 SNPs for O. neglectus , 4809 SNPs for L. segnis and 8162 SNPs for P. karaka . In addition to the holdfast invertebrate datasets, we retrieved the previously generated VCF files for all three Durvillaeaspecies from Parvizi et al. 2020 (available at https://datadryad.org/stash/dataset/doi:10.5061/dryad.pg4f4qrkm) and reanalyzed them to have a consistent analytical approach when comparing the population genomic structure and demographic history of hosts and epifauna.