Aim Studying clearly delineated populations in marine lakes, islands of sea, we investigate the interplay of habitat size, immigration, and priority effects in shaping marine population genetic structure. Location Marine lakes and coastal locations in Indonesia, Palau, Papua New-Guinea and Australia. Taxon Mussels (Mytillidae, Brachidontes spp.) Methods Populations were sampled from four coastal locations and 22 marine lakes of similar age (~8,000 years), yet differing in size (0.04 - 4.7 km2) and degree of connection to the adjacent sea. While some lakes are highly connected, allowing potential influx of larvae from the sea, others have very limited water exchange. We assessed the phylogeographic structure and demographic history using mitochondrial and nuclear DNA sequence data, and combined this with geometric morphometrics. The effects of lake characteristics on population genetic diversity and structure were tested using linear regression and Mantel tests. Results Each lake contained one of six distinct genetic lineages, which were characterized by deep phylogenetic splits and significant morphometric differences. These lineages likely represent separate species. The lineages showed similar demographic patterns, with lakes containing founder populations that rapidly expanded and diverged. Genetic diversity within lake populations was significantly correlated with lake area, but not with physical connection to the adjacent sea. Within lineages that occurred in multiple lakes there was strong population structure (average ΦST 0.65), which did not conform to an isolation-by-distance pattern or to the degree of dispersal potential. Main Conclusions Marine lakes across a gradient of physical isolation show strong population structure and evidence for in situ divergence. We hypothesize that the observed genetic structure is the result of priority effects. In addition, reduction of habitat size appears to reduce genetic diversity, even at very small spatial scales. Our findings are relevant in the context of ongoing alterations to coastal hydrodynamics, which lead to habitat reduction and influence migration among populations at fine spatial scales.

To identify morphospecies, geometric morphometric analyses were processed for a total of 419 digital images of 13 mussel populations in Indonesia (lineage A–D, F). The shells of lineage E were not available for analysis. The analyses were performed following the procedure described in Becking et al. (2016). Briefly, mussels were positioned in a standardized way and photographed. Shell outlines were drawn via curves starting from the beak of the mussel using tpsDig (Rohlf, 2010). Curves were converted to semi-landmarks and standardized for size and orientation in tpsRelw (Rohlf, 2010). We obtained centroid size and relative warp scores for each specimen. Furthermore, repeatability was confirmed for the first three relative warp axes; thus, subsequent analyses were performed only on the first three axes. To study differences among groups, we performed a nonparametric permutational multivariate analysis of variance (PERMANOVA, 1,000 permutations) implemented in past 2.11 based on Euclidean distances (Hammer, Harper, & Ryan, 2001). We corrected for multiple testing by calculating Bonferroni-corrected P values.