Non-lethal rapid biodiversity assessment

International audience

English. For several animal taxa, non-lethal techniques that do not rely on collecting individuals are routinely used to assess biodiversity (e.g. point counts in birds). Identification often relies on the ability of the observer, are subjected to errors, but populations are not impacted. Thus, multiple counting sessions (MCS) that allow using robust analyses (e.g. unbiased Chao richness estimate) are available. However, for most species (e.g. arthropods), trap systems must be set up. Killed individuals are collected and later accurately identified in the laboratory, but unbiased MCS become unavailable. Environmental DNA bar-coding provides an alternative, yet it requires important technical support and is not designed for MCS. Lethal rapid biodiversity assessments (RBA), derived from classical trap surveys and based on less accurate identifications (morphospecies are used), have been successfully developed to relax technical constraints. In this study, we combined non-lethal and RBA approaches to address logistical, analytical and ethical issues. We tested five versions of a protocol to visually survey the macro-fauna of hedgerows. A large number of individuals were directly identified in the field, mostly arthropods but also vertebrates. Identification error varied with taxonomic level and lineage, but remained low at the morphospecies level. Importantly, estimates tended to reach asymptotes, suggesting that local richness was appropriately appraised. Like any technique, non-lethal RBA (NL-RBA) present both advantages and weaknesses, and may improve the toolbox to survey biodiversity.