3.4 Beta diversity between sites and replicates
We evaluated inter and intra site-based estimates of β diversity using the Jaccard metric, which weighs all observed taxa equally. At both 1,000 and 10,000 read sampling depths, we found that PCR replicates in both PITS and FITS results cluster by site and that sites from the same geographic area cluster near each other (Figure 4). None of the identified PCR replicate outliers reduced the ability to differentiate sites based on the PCoA. While read sampling depth did not affect dispersion in the PCoA between PCR replicates for either PITS or FITS, increasing read sampling depth changed the position of some sites relative to each other in the ordination (Figure 4). For example, the two Fort Ord sites all had similar PITS taxon composition and could not be distinguished from each other at either read sampling depth. However, the two St.P sites could be differentiated at both read sampling depths for PITS but only at the lower sequencing read depth in FITS. We next calculated β diversity using the Bray-Curtis estimator (Bray and Curtis, 1957), which considers RA, and compared the resulting PCoA plots to those generated using the Jaccard estimator (Figure S1; Figure 4). The plots were similar overall, with slightly closer clustering among replicates at the two St.P sites in the PITS data set. The similarity between results using Jaccard and Bray-Curtis estimates suggests that low abundance taxa may not influence strongly estimates of β diversity.