Biomonitoring river diatoms: Implications of taxonomic resolution

摘要

Benthic diatoms are routinely used to assess river pollution. Most of the tools based on these organisms exploit the differences of pollution sensitivity between species; as such, species level identification is required. Accurate determination of diatom species requires rigorous training due to the extreme diversity of the group. The level of taxonomic resolution for biomonitoring is still debated. We tested the influence of taxonomic resolution on diatom bioassessment in an ecoregional framework. We used a database of 1967 diatom samples from biomonitoring programs in two French river basins, that reported three kinds of data for each site: (a) taxa abundance, expressed with 6 separate level of taxonomic resolution: species, genus, family, order, class or subdivision level; (b) physical and chemical characterization; (c) ecoregion and river-size class memberships. Mantel tests showed that the influence of taxonomic resolution on assemblage composition description was weak from species to order level. Mantel tests between chemical parameters and diatom assemblages showed that there was an increase in correlation from subdivision to genera resolution. But species and genus resolutions showed equivalent correlations with chemical parameters. Predictive models using diatom data to reconstruct nutrients, organic matter and major-ions content showed an increasing performance from sub-division to species resolution. Nevertheless their performances did not follow the exponential increase of taxa number from sub-division to species: models performances improved only by 12-23% from genus to species depending on the parameter reconstructed whereas number of taxa was multiplied by 10. Finally, we observed that the more precise the taxonomic resolution, the better the correspondence with ecoregion classification. This can be partly explained by diatom endemism and cosmopolitanism which is mostly observed to species level, rarely to genus level and never above.For a quick and robust assessment of river pollution coarse identification is sufficient. Hypotheses to explain such results are that: (1) many species are too rare to describe their ecological requirements with certainty; (2) more environmental descriptors are necessary to explain the presence of some species; (3) the dataset is compromised by identification errors, particularly at the species level. On the other hand, a precise ecoregional bioassessment requires a fine taxonomic resolution; this must be stressed for the European Water Framework Directive which requires an assessment in an ecoregion classification.