Computational protocol: Evaluating a multigene environmental DNA approach for biodiversity assessment

Similar protocols

Protocol publication

[…] Deconvolution, trimming, and quality-based filtering of the NGS data from the 16S, 18S, trnL, ITS, COI, and COI-spun eDNA datasets resulted in 65,786-768,208 high quality reads per marker. Error-correction of the sequence reads was performed using Acacia []. Operational taxonomic units (OTUs) by eDNA marker were determined using the UPARSE [] pipeline with a 97 % sequence similarity clustering threshold (in all cases except Fig. where we vary the sequence similarity threshold). Additionally, an alternative set of OTUs for each amplicon dataset was constructed in which all of the single-read OTUs were removed (see Additional file for a full set of parallel analyses to match those described below, none of the major conclusions are affected by this alternative data treatment). OTUs were assigned to phyla using BLAST+ and MEGAN 5 [] (Figs. and ). Fig. 2Fig. 3Fig. 4Diversity statistics were calculated for both eDNA marker datasets (Table ) and those collected using conventional methods (Table ) with the R package vegetarian []. Alpha, beta, and gamma diversities all decreased steeply as the similarity threshold for OTU clustering decreased from 100 to 97 %. The diversities were generally less sensitive to changes in the similarity threshold between 90–97 % (Fig. ). Beta diversities were less sensitive to the choice of OTU similarity threshold than the alpha and gamma diversity estimates.Rarefaction curve analysis for each of the eDNA markers indicates different sampling properties for the different diversity statistics (Fig. ). Measures of alpha and gamma diversities were highly dependent on the number of sequences, with most gene regions not asymptoting to a maximum. On the other hand, beta diversities trended towards a stable measure after a few thousand sequence reads for all the eDNA markers examined. Beta diversities within and among plots varied for the different markers (Fig. ). Beta diversities were low within plots for 16S, but were highly variable between pairs of plots. Fig. 5Fig. 6The 18S marker showed intermediate levels of beta diversities, both within and between pairs of plots, whereas the remaining four eDNA markers had high beta diversities within and especially between pairs of plots. A regression analysis of pairwise beta diversity against the elevational difference between plots (Fig. ) shows that among the conventional methods, trees, seedlings and invertebrates have the strongest positive correlation. This decrease in compositional similarity with increasing elevational separation is analogous to the well-established distance-decay relationship [, ]. Among the eDNA markers, the COI and 18S markers showed the strongest positive correlation between pairwise beta diversity and elevational difference (COI: r=0.49, p<0.001; 18S: r=0.48, p<0.001). All of the correlations were significant using PERMANOVA [] except 16S and trnL (Table ). Fig. 7 […]

Pipeline specifications

Software tools Acacia, UPARSE, MEGAN
Application 16S rRNA-seq analysis