Fig 1.
Map of the four collection sites in Willapa Bay, WA, with the distribution of carapace widths (in millimeters) at each site for all European green crabs trapped (gray), those chosen for sequencing (black), and those chosen for sequencing which had identifiable prey DNA in their stomach contents (red). The map inset shows the location of Willapa Bay on the Washington State coast. The main map background is from the U.S. Geological Survey National Map, compiled using terrainr [33].
Table 1.
Collection sites, dates, and sample sizes, with collection-specific information on crab and prey.
The final sample size (“Final N”) is the number of sequenced European green crabs (“N sequenced”) with putative prey items identified in their stomach contents. We also provide the mean crab carapace width (mm; of crabs sequenced) and the overall prey alpha diversity for the given site.
Table 2.
Prey items that occurred in two or more European green crab from any site, to their finest taxonomic classification, and the total number of crab in which that prey taxon was identified (N).
Numbers that are not italicized represent totals for the given taxonomic level, summed across more specific taxonomies (including those not shown here, when n = 1). For a comprehensive summary of prey taxa, see S4 Table.
Fig 2.
Frequency of prey occurrence (FO%), summarized (a) for major prey groups (distribution of the frequencies of occurrence for individual taxa within each group, across all sites), and (b) for taxa which were present in four or more crab across all sites. Prey group identification numbers from (a) are annotated at the top of (b). In (b), total frequency of occurrence is broken down by site (bar color).
Fig 3.
Principal Coordinates Analysis (PCoA) constructed using (a) prey presence/absence and the Jaccard dissimilarity index, and (b) the eDNA index and the Bray-Curtis distance.
Table 3.
β diversity of prey taxa between the sampling sites.
Darker borders outline the β diversities between the two clam bed sites (marked (*); top left) and the two slough sites (marked (+); bottom right).
Table 4.
The significance of diet variability across sites according to PERMANOVAs conducted using (a) the Jaccard dissimilarity index calculated from prey presence/absence and (b) the Bray-Curtis distance calculated from the semi-quantitative eDNA index.
Superscript indicates significance at the α = 0.10 (.), 0.05 (*), and 0.01 (**) levels.
Table 5.
Post-hoc pairwise tests of diet composition between pairs of sites that are of the opposite type (clam bed site v. natural slough site); tests were conducted using (a) a prey presence/absence, and (b) the eDNA index.
Superscript indicates significance at the α = 0.10 (.), 0.05 (*), and 0.01 (**) levels. The Bonferroni correction (for six total pairwise tests for each data set) was applied to obtain the “corrected p-value.” Results from pairwise tests between sites of the same type are reported in S6 Table.
Table 6.
Estimates of amplification efficiencies (α) for all species in the mock communities.
Amplification efficiencies are relative to the Manila clam (R. philippinarum). The α parameter captures the log-ratio of amplification efficiency for a species relative to that of the reference species (here, Manila clam; see Shelton et al. [29] for full description of α). + Not present in sample data, α estimated from separate run of the quantitative model.
Fig 4.
The diet of an “average” green crab which contains the eight calibrated prey species, from each site type (a). Proportions for each prey species are provided in white text in the appropriate block, and represent the mean mixture proportions estimated from the fitted zero-inflated Dirichlet regression model. To put into perspective what subset of all sequencing data is represented by the eight calibrated prey species, we also provide (b) the proportion of sequencing reads remaining after each filtering step, out of the total aligned to NCBI’s BLAST database (“total reads”), and (c) the proportions of all unique prey taxa, and of reads belonging to all prey taxa, captured by the eight calibrated prey species. 95% credible intervals associated with the mean values in (a) can be found in S6 Fig.