eDNA metabarcoding warms up a hotspot of marine biodiversity: revealing underrepresented taxa in visual surveys and historical records from the Gulf of California

Abstract

Environmental DNA (eDNA) metabarcoding is revolutionizing biodiversity monitoring, but comparisons against traditional data rarely include long-term historical inventories. We targeted eukaryotes by amplifying a fragment of the 18S gene from eDNA isolated from seawater samples at 20 sites in the Gulf of California (GC) and contrasted regional taxonomic diversity against 316 simultaneous visual surveys and a historical database with over 5k species. From 61k Amplified Sequence Variants, we identified 850 eukaryotic families, of which half represent new compiled records, including 174 families of planktonic, benthic, and parasitic invertebrates. The 18S eDNA metabarcoding analysis revealed many overseen taxa, highlighting higher taxonomic ranks within micro-invertebrates, microscopic fungi, and other micro-eukaryotes from the supergroups Stramenopiles, Alveolata, and Rhizaria. The database combining all methods has doubled the number of distinct phyla, classes, and orders compared to the historical baseline, indicating biodiversity levels in the GC are much higher than previously assumed. The estimated proportion of historical taxa included in public reference databases was only 18% for species, partially explaining the small portion of 18S eDNA reads that were taxonomically assigned to species level (13%). Each method showed different taxonomic biases, with 18S eDNA metabarcoding detecting few vertebrates, visual surveys targeting only seven metazoan phyla, and the historical records focusing on macroinvertebrates, fish, and algae. Although all methods recovered the main known biogeographic regionalization, the 18S eDNA metabarcoding data did not support the historical pattern of higher diversity in the Central than Northern GC. While combining methods provides a novel view of biodiversity that is much more comprehensive than any individual approach, our study highlights many challenges in synthesizing biodiversity data from traditional and novel sources.