Fig 1.
Effect of sampling effort on estimates of ANOSIM R and PERMANOVA R2 effect sizes and 95% confidence intervals.
Sampling effort represents random subsamples of each comparison group with replacement. ANOSIM R (grey) and PERMANOVA R2 (red) values were calculated with Bray-Curtis dissimilarity and 95% confidence interval based on 1000 subsamples. a) Comparison of FFE communities between bases and tips of P. taeda needles at five-sample intervals from five to 60 samples each. b) Comparison of FFE communities between P. torreyana needles between different geographic locations (Santa Rosa and San Diego, CA) at one-sample intervals from three to 14 samples each.
Fig 2.
Effect of sequencing depths on estimates of ANOSIM R and 95% confidence intervals based on Bray-Curtis dissimilarity indices.
Comparisons were made between FFE communities of a) bases and tips of P. taeda needles at 5, 10, 20 and 60 samples per group and b) between P. torreyana needles between Santa Rosa and San Diego, CA at 3, 6, 9, and 14 samples per group. Sequencing depths were tested at 100, 1000, 5000, 10000, 20000. X-axes are on a log-scale. Dotted lines indicate 95% confidence intervals for different sampling sizes. Solid lines indicate the mean. Trends for different sampling efforts were overlapped in a single plot.
Fig 3.
Effect of sequencing depths on estimates of geographic community dissimilarity and 95% confidence intervals for FFE communities that differ in geography.
FFE communities between nine P. taeda plots across varying distances (1–107 km) were compared. ANOSIM R (grey) and PermANOVA R2 (red) values were calculated with Bray-Curtis dissimilarity and 95% confidence interval based on 1000 subsamples. Sequencing depth tested at 100, 1000, 5000, 10000, 20000. X-axes are on a log-scale. Dotted lines indicate 95% confidence interval. Solid lines indicate the mean.
Fig 4.
β-diversity patterns are revealed with increasing sequencing depths between communities that differ only by random processes.
Non-metric multidimensional scaling of two simulated communities drifted for 20 generations from one community with 5 million individuals and 974 OTUs (taxa) have beta-diversity patterns revealed only after 50,000 sequences per sample, which corresponds to ~1% of the community per sample. Ordinations are based on between-sample dissimilarity calculated with Bray-Curtis (left panel), Jaccard (middle panel). OTU accumulation curves (right panel) demonstrate community sampling with varying sequencing depths. From top to bottom, sequencing depths correspond to a) 100, b) 1000, c) 5000, d) 10000, and e) 50000 sequences. Insets represent the accumulation curves from the previous sequencing depth.