Abstract
The speed and intensity of global change requires new means of observing biodiversity that are rapid, non-destructive, standardized, widely deployable and in remote ecosystems (deep sea). Conventional inventory methods are based on morphological or acoustic identification of species, which are costly in terms of time and expertise. Beyond these signals, animals also leave traces of DNA in their environment in the form of dermal cells, mucus or feces. The metabarcoding of this environmental DNA (eDNA) consists in collecting this DNA, amplifying and sequencing it to identify the species present using a genetic reference database. However, these reference databases are incomplete, which severely limits the potential of eDNA. The aim of this thesis is to develop an alternative approach based on molecular taxonomic units (MOTUs) to analyze the biodiversity of aquatic macroorganisms, and more particularly that of bony fish. I first performed a global and spatialized synthesis of the taxonomic coverage of the genetic reference database for all bony fishes, which shows an under-representation of species in the tropical zone as well as taxonomic gaps for endangered and non-indigenous species. Only 13% of fish species are sequenced for the most common marker, which excludes any ambition for an exhaustive analysis of biodiversity using only species-level assignments in the short or medium term. Consequently, I have developed a bioinformatics pipeline to generate estimates of diversity using molecular taxonomic units (MOTUs) by fish family. It shows how this MOTU diversity represents an excellent proxy for species diversity at different spatial scales. Then an application of eDNA metabarcoding and the MOTUs approach allowed to estimate the functional diversity, based on species traits, and the phylogenetic diversity, based on the evolutionary history of the species, of tropical fishes in a more exhaustive way than traditional methods (videos, dives). Finally, in a first global analysis of coral reef diversity in eDNA, which brings together 251 samples collected from the Indian Ocean to the Caribbean, the MOTUs approach allows the reconstruction of major trends in fish biogeography but also reveals local spatial heterogeneity hitherto underestimated. While it is now crucial to set up efficient, non-specialist dependent and high temporal frequency monitoring methods to better understand the effects of global changes on biodiversity, this work demonstrates the full potential of eDNA using a MOTUs approach to build robust indicators of several facets of biodiversity at several scales, but also to test theoretical hypotheses underlying the distribution of this biodiversity.