Population genomics for symbiotic anthozoans: can reduced representation approaches be used for taxa without reference genomes?

Population genetic studies of symbiotic anthozoans have been historically challenging because their endosymbioses with dinoflagellates have impeded marker development. Genomic approaches like reduced representation sequencing alleviate marker development issues but produce anonymous loci, and without a reference genome, it is unknown which organism is contributing to the observed patterns. Alternative methods such as bait-capture sequencing targeting Ultra-Conserved Elements are now possible but costly. Thus, RADseq remains attractive, but how useful are these methods for symbiotic anthozoan taxa without a reference genome to separate anthozoan from algal sequences? We explore this through a case-study using a double-digest RADseq dataset for the sea anemone Bartholomea annulata. We assembled a holobiont dataset (3854 loci) for 101 individuals, then used a reference genome to create an aposymbiotic dataset (1402 loci). For both datasets, we investigated population structure and used coalescent simulations to estimate demography and population parameters. We demonstrate complete overlap in the spatial patterns of genetic diversity, demographic histories, and population parameter estimates for holobiont and aposymbiotic datasets. We hypothesize that the unique combination of anthozoan biology, diversity of the endosymbionts, and the manner in which assembly programs identify orthologous loci alleviates the need for reference genomes in some circumstances. We explore this hypothesis by assembling an additional 21 datasets using the assembly programs pyRAD and Stacks. We conclude that RADseq methods are more tractable for symbiotic anthozoans without reference genomes than previously realized.