Data from: Metabarcoding of soil environmental DNA replicates plant community variation but not specificity

While metabarcoding of plant DNA from their environment is an exciting
method that can supplement inventorying of live plant species, the
accuracy and specificity has yet to be fully assessed over complex
continuous landscapes. In this work, we evaluate plant community profiles
produced via metabarcoding of soil by comparing them to a morphological
survey. We assessed plant communities by metabarcoding of soil DNA in 130
sites along ecological gradients (nutrients, succession, moisture) in
Denmark using chloroplast trnL region (10-143 bp) primer set and
compared the resulting communities to communities produced with a longer
nuclear ITS2 region (~216 bp) and a morphological survey. We found that
the community variation observed within the morphological survey was well
represented by molecular surveys, with significant correlation with both
community composition and richness using both primer sets. While the
majority of the ITS2 sequences could be assigned to species (over 80%), we
had less success with the trnL sequences (70%), which was only
possible after restricting the reference database to local species. We
conclude that the community profiles produced by metabarcoding can be
highly effective in performing large-scale macroecological studies.
However, the discovery rates and taxonomic assignments produced via
metabarcoding remained inferior to morphological surveys, but manual
curation of databases improves the specificity of
assignments made by the trnL primers, and improves
the accuracy of the assignments made with the ITS2
primers. Finally, we suggest that a greater percentage of named diversity
would be recovered by increasing soil sampling with the use of additional
universal primer sets.