Back

Quantitative, Multispecies Monitoring at a Continental Scale

Guri, G.; Liu, O. R.; Kelly, R. P.; Shaffer, M. R.; Parsons, K. M.; Ramon-Laca, A.; Nichols, K. M.; Brandao-Dias, P. F. P.; Wells, A.; Shelton, A. O.

2025-10-31 ecology
10.1101/2025.10.30.685180 bioRxiv
Show abstract

Molecular data from environmental samples can reflect the abundance of species DNA, an index immediately relevant to natural-resources management at a broad geographic scale. These data commonly derive from assays designed and targeted for individual species (e.g., using qPCR or ddPCR), or from metabarcoding that use more general PCR primers to amplify many species simultaneously. Multispecies analyses make efficient use of field samples and laboratory time, and speak to inherently multispecies questions of management and ecological interest. However, unlike single-species techniques, metabarcoding alone generally reflects only the environmental DNA (eDNA) proportions of target-species DNA present, not their absolute quantities. Here, we combine qPCR and metabarcoding data derived from the same samples to estimate quantities of eDNA from many fish species along the US West Coast in three dimensions, demonstrating a technique of practical relevance for both management and ecology. We derive spatially explicit maps of eDNA abundances for 12 common species of ecological and management importance and point the way to quantitative surveys of wild species using molecules alone. We find that species distribution maps derived from eDNA largely mirror known species niches and known spatial distributions. Notably, our analyses indentify biodiversity hotspots that align with previously documented regions of ecological significance, such as the Columbia River plume and the Heceta Bank.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.