Schielbelhut et al. (2018) detected allelic shifts in grossly normal P. ochraceus adults and juveniles before and after the SSWS outbreak in California. They found three loci putatively under selection and reported on 100 discriminatory haplotypes between time periods. Interestingly, the 18 SNP loci we detected as most discriminatory in our samples did not occur within 30 kb of those reported by Schielbelhut et al. (2018), and most were between 250 kb and 2 Mb apart. Lack of overlap in these genomic regions is perhaps not surprising given the multitude of differences between the studies, such as year of sampling, the health status of sea stars, and geographic location. For instance, SSWS is known to be associated with spikes in sea water temperature (Bates et al., 2009; Eckert et al., 1999; Eisenlord et al., 2016; Kohl et al., 2016), and daily deviations from annual sea water temperatures in Oregon were more prevalent in 2013/2014 than in California (Miner et al., 2018). Such relevant environmental differences between California and Oregon may hence also cause different selective pressures. Moreover, the reduced-representation nature of RADseq, and the use of different restriction enzymes suggest that a lack of overlap between studies' outlier markers is not indicative of a lack of biological relevance.
The scarcity of SSWS-associated loci isolated is also possibly a result of reduced coverage from the inherent nature of RADseq methods (Lowry et al., 2017a). While RADseq is an efficient and cost-effect method for producing thousands of SNPs along the entire genome, these markers remain sparse. Despite the limitations, many RADseq studies have found loci attributing to adaptive selection when coverage is adequate (Epstein et al., 2016; Lowry et al., 2017a; McKinney, Larson, Seeb, & Seeb, 2017). Marker density aimed at detecting phenotype-genotype associations is recommended to be high and relative to LD in the target species (Lowry et al., 2017a). Based on this metric, our RADseq effort in this study adequately covers the full genome, with on average 2.5 SNPs found per every 15-kb linkage block. Therefore, we argue that our results are not due to low marker density, but perhaps may be improved by genotyping a higher number of individuals.
Our study joins that of Schielbelhut et al. (2018) and Ruiz-Ramos et al. (2020) in assessing genomic variation in the keystone species P. ochraceus and highlighting the importance of understanding the causes and responses to devastating SSWS outbreaks. While current patterns remain obscure, the accumulation of putative functional genomic regions will serve as invaluable resources for continued field and laboratory studies. In addition to physiological and transcriptomic experiments, we suggest the need for a concerted effort to sample large numbers of wasting and apparently unaffected individuals across several geographic regions, and ideally using low-coverage whole-genome sequencing for substantially increased power (Lou, Jacobs, Wilder, & Therkildsen, 2021).