From environmental DNA sequences to ecological conclusions: How strong is the influence of methodological choices?

Abstract

AbstractAimEnvironmental DNA (eDNA) is increasingly used for analysing and modelling all‐inclusive biodiversity patterns. However, the reliability of eDNA‐based diversity estimates is commonly compromised by arbitrary decisions for curating the data from molecular artefacts. Here, we test the sensitivity of common ecological analyses to these curation steps, and identify the crucial ones to draw sound ecological conclusions.LocationValloire, French Alps.TaxonVascular plants and fungi.MethodsUsing soil eDNA metabarcoding data for plants and fungi from 20 plots sampled along a 1000‐m elevational gradient, we tested how the conclusions from three types of ecological analyses: (a) the spatial partitioning of diversity, (b) the diversity–environment relationship, and (c) the distance–decay relationship, are robust to data curation steps. Since eDNA metabarcoding data also comprise erroneous sequences with low frequencies, diversity estimates were further calculated using abundance‐based Hill numbers, which penalize rare sequences through a scaling parameter, namely the order of diversity q (Richness with q = 0, Shannon diversity with q ~ 1, Simpson diversity with q = 2).ResultsWe showed that results from different ecological analyses had varying degrees of sensitivity to data curation strategies and that the use of Shannon and Simpson diversities led to more reliable results. We demonstrated that molecular operational taxonomic unit clustering, removal of polymerase chain reaction errors and of cross‐sample contaminations had major impacts on ecological analyses.Main conclusionsIn the Era of Big Data, eDNA metabarcoding is going to be one of the major tools to describe, model and predict biodiversity in space and time. However, ignoring crucial data curation steps will impede the robustness of several ecological conclusions. Here, we propose a roadmap of crucial curation steps for different types of ecological analyses.