Reduced representation sequencing for symbiotic anthozoans: are reference genomes necessary to eliminate endosymbiont contamination and make robust phylogeographic inference?
Anthozoan cnidarians form the backbone of coral reefs. Their success relies on endosymbiosis with photosynthetic dinoflagellates in the family Symbiodiniaceae. Photosymbionts represent a hurdle for researchers using population genomic techniques to study these highly imperiled and ecologically critical species because sequencing datasets harbor unknown mixtures of anthozoan and photosymbiont loci. Here we use range-wide sampling and a double-digest restriction-site associated DNA sequencing (ddRADseq) of the sea anemone Bartholomea annulata to explore how symbiont loci impact the interpretation of phylogeographic patterns and population genetic parameters. We use the genome of the closely related Exaiptasia diaphana (previously Aiptasia pallida) to create an anthozoan-only dataset from a genomic dataset containing both B. annulata and its symbiodiniacean symbionts and then compare this to the raw, holobiont dataset. For each, we investigate spatial patterns of genetic diversity and use coalescent model-based approaches to estimate demographic history and population parameters. The Florida Straits are the only phylogeographic break we recover for B. annulata, with divergence estimated during the last glacial maximum. Because B. annulata hosts multiple members of Symbiodiniaceae, we hypothesize that, under moderate missing data thresholds, de novo clustering algorithms that identify orthologs across datasets will have difficulty identifying shared non-coding loci from the photosymbionts. We infer that, for anthozoans hosting diverse members of Symbiodinaceae, clustering algorithms act as de facto filters of symbiont loci. Thus, while at least some photosymbiont loci remain, these are swamped by orders of magnitude greater numbers of anthozoan loci and thus represent genetic \"noise,\" rather than contributing genetic signal.