Systematic comparison of the relative accuracy of vegetation surveys and soil DNA metabarcoding : Assessing plant biodiversity at different spatial scales

Abstract: Analysis of soil-derived DNA has been shown to minimize problems seen during traditional vegetation surveys, e.g. by matching the eDNA to a reference database for taxonomic identification rather than relying solely on taxonomic expertise. However, it has been debated to what extent and how accurately eDNA acts as a proxy for biodiversity. The reliability on eDNA and the awareness on influencing factors is also important for palaeoenvironmental reconstructions where above-ground vegetation cannot be used. This study aimed to investigate how well modern soil-derived eDNA reflects the contemporary vascular vegetation communities in a subarctic environment, and if the efficiency of the taxonomic identification differed between spatial scales. Near-surface soil samples at altitudinal gradients along numerous transects were collected in combination with vegetation surveys. The eDNA was amplified through metabarcoding using the P6 loop region of the chloroplast trnL intron, followed by a high-throughput sequencing. No difference in the number of identified taxa between eDNA and vegetation survey was seen at landscape scale. In contrast, the number of identified taxa was consistently higher in the vegetation survey at smaller spatial scales. The efficiency of identified taxa per scale remained stable for the vegetation survey, whereas for eDNA, a decreasing trend was seen. This study highlights the variations on taxa identification between both methods and which factors might cause it. Combining the methods allows for a more precise modern biodiversity estimation, as well as to minimize wrongful conclusions. This allows for a more accurate palaeoenvironmental reconstructions, which in turn can improve future species conservation decisions.