Back

Holistic monitoring of freshwater and terrestrial vertebrates by camera trapping and environmental DNA

Holm, A. M. R.; Knudsen, S. W.; Maansson, M.; Pedersen, D. E.; Nordfoss, P. H.; Johansson, D. K.; Gramsbergen, M.; Havmoller, R. W.; Sigsgaard, E. E.; Thomsen, P. F.; Olsen, M. T.; Moller, P. R.

2022-11-24 zoology
10.1101/2022.11.23.517571 bioRxiv
Show abstract

The anthropogenic impact on the worlds ecosystems is severe and the need for non-invasive, cost-effective tools for monitoring and understanding those impacts are therefore urgent. Here we combine two such methods in a comprehensive multi-year study; camera trapping (CT) and analysis of environmental DNA (eDNA), in river marginal zones of a temperate, wetland Nature Park in Denmark. CT was performed from 2015 to 2019 for a total of 8,778 camera trap days and yielded 24,376 animal observations. The CT observations covered 87 taxa, of which 78 were identified to species level, and 73 were wild native species. For eDNA metabarcoding, a total of 114 freshwater samples were collected from eight sites in all four seasons from 2017 to 2018. The eDNA results yielded a total detection of 80 taxa, of which 74 were identified to species level, and 65 were wild native species. While the number of taxa detected with the two methods were comparable, the species overlap was only 20 %. In combination, CT and eDNA monitoring thus yielded a total of 115 wild species (20 fishes, four amphibians, one snake, 23 mammals and 67 birds), representing half of the species found via conventional surveys over the last ca. 20 years (83% of fishes, 68 % of mammals, 67 % of amphibians, 41 % of birds and 20 % of reptiles). Our study demonstrates that a holistic approach combining two non-invasive methods, CT and eDNA metabarcoding, has great potential as a cost-effective biomonitoring tool for vertebrates.

Matching journals

The top 3 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.