The impact of OTU sequence similarity threshold on diatom‐based bioassessment: A case study of the rivers of Mayotte (France, Indian Ocean)

Abstract

Extensive studies on the taxonomic resolution required for bioassessment purposes have determined that resolution above species level (genus, family) is sufficient for their use as indicators of relevant environmental pressures. The high‐throughput sequencing (HTS) and meta‐barcoding methods now used for bioassessment traditionally employ an arbitrary sequence similarity threshold (SST) around 95% or 97% to cluster sequences into operational taxonomic units, which is considered descriptive of species‐level resolution. In this study, we analyzed the effect of the SST on the resulting diatom‐based ecological quality index, which is based on OTU abundance distribution along a defined environmental gradient, ideally avoiding taxonomic assignments that could result in high rates of unclassified OTUs and biased final values. A total of 90 biofilm samples were collected in 2014 and 2015 from 51 stream sites on Mayotte Island in parallel with measures of relevant physical and chemical parameters. HTS sequencing was performed on the biofilms using the rbcL region as the genetic marker and diatom‐specific primers. Hierarchical clustering was used to group sequences into OTUs using 20 experimental SST levels (80%–99%). An OTU‐based quality index (IdxOTU) was developed based on a weighted average equation using the abundance profiles of the OTUs. The developed IdxOTU revealed significant correlations between the IdxOTU values and the reference pressure gradient, which reached maximal performance using an SST of 90% (well above species level delimitation). We observed an interesting and important trade‐off with the power to discriminate between sampling sites and index stability that will greatly inform future applications of the index. Taken together, the results from this study detail a thoroughly optimized and validated approach to generating robust, reproducible, and complete indexes that will greatly facilitate effective and efficient environmental monitoring.

Assessing insect biodiversity with automatic light traps in Brazil: Pearls and pitfalls of metabarcoding samples in preservative ethanol

“…Our diversity analyses reiterate the impact of sequence similarity levels (cutoffs) on OTU taxonomic composition (Holovachov, Haenel, Bourlat, & Jondelius, 2017;Potter et al, 2017;Tapolczai et al, 2019) since alpha diversity differed when using different cutoffs. The choice of sequence similarity levels greatly depends upon well-curated reference databases, marker of choice, targeted taxa, or a combination of all.…”

Diatom DNA Metabarcoding for Biomonitoring: Strategies to Avoid Major Taxonomical and Bioinformatical Biases Limiting Molecular Indices Capacities

“…In traditional indices, species ecological profiles (optimum and tolerance) are often directly obtained from a large set of data by using simple statistical descriptors of location (e.g., weighted mean) and dispersion (e.g., weighted standard deviation) along a pollution gradient. Interestingly, the morphospecies as a taxonomic unit of the traditional indices can be replaced by molecular taxonomic units, allowing to train a new type of biological indices, the so-called taxonomy-free indices (Apothéloz-Perret-Gentil et al, 2017;Tapolczai et al, 2019). So far, taxonomy-free indices have proved to be an efficient strategy to keep biological information carried by OTUs whose taxonomic assignment is otherwise not possible because of the incomplete reference library.…”

Ecosystems Monitoring Powered by Environmental Genomics: A Review of Current Strategies with An Implementation Roadmap

“…By considering all the OTUs (or ASVs) profiles along a known impact gradient of typical anthropogenic origin, these studies have shown that HTS data represent a virtually unlimited reservoir of new bioindicators. Examples (listed in Table 1) include contamination by pesticide (Thompson et al, 2016;Andùjar et al, 2017), agriculture stressors (Salis et al, 2017) and gradients of eutrophication and urban contamination (Apothéloz et al, 2017;Martínez-Santos et al 2018;Simonin et al, 2019;Tapolczai et al, 2019aTapolczai et al, , 2019b in freshwater systems. In marine environments, it was demonstrated after an oil spill (Bik et al, 2012), in the vicinity of offshore drilling platforms (Lanzén et al, 2016;Laroche et al, 2016Laroche et al, , 2018a and aquaculture sites (Pochon et al, 2015;Dowle et al, 2015;Stoeck et al 2018aStoeck et al , 2018b as well as along eutrophication and urban or industrial contamination gradients in estuaries (Chariton et al, 2010(Chariton et al, , 2015Angly et al, 2015;Lallias et al, 2015;Obi et al, 2016).…”

“…molecular-based index is directly calibrated on the HTS data, which alleviates the qualitative and quantitative biases encountered with the taxonomy-based EG approaches. Such strategy has proven successful for both freshwater benthic diatoms (Apothéloz et al, 2017;Tapolczai et al, 2019aTapolczai et al, , 2019b and for water bacterial and eukaryotic communities in streams and estuarine systems (Chariton et al, 2015;Li et al, 2018). A strategy analogous to the IndVal approach is the use of polynomial quantile regression splines (Andersson, 2008), which has shown great promise for the prediction of organic enrichment in aquaculture sites using eukaryotic and prokaryotic metabarcoding data in parallel .…”

“…A strategy analogous to the IndVal approach is the use of polynomial quantile regression splines (Andersson, 2008), which has shown great promise for the prediction of organic enrichment in aquaculture sites using eukaryotic and prokaryotic metabarcoding data in parallel . For diatoms, the accuracy of the assessment can be largely improved, arguably because the indicator value approach makes use of a larger number of OTUs or ASVs compared to an approach relying solely on their taxonomic assignments (Apothéloz et al, 2017;Tapolczai et al 2019aTapolczai et al , 2019b).…”