Man against machine: Do fungal fruitbodies and eDNA give similar biodiversity assessments across broad environmental gradients?

The majority of Earth's biodiversity is unknown. This is particularly true for the vast part of soil biodiversity, which rarely can be observed directly. Metabarcoding of DNA extracted from the environment (eDNA) has become state-of-the-art in assessing soil biodiversity. For assessing fungal communities, eDNA metabarcoding is seen as an attractive alternative to classical surveying based on fruitbodies. However, it is unknown whether eDNA metabarcoding provides a representative sample of fungal diversity and census of threatened species. Therefore conservation planning and assessment are still based on fruitbody inventories. Based on a dataset of large ecological width, representing both soil eDNA metabarcoding and expert inventories of fungal fruitbodies in Denmark, we document for the first time the validity of eDNA sampling and metabarcoding as a practical inventory method and a measure of conservation value for fungi. Fruitbody data identified fewer species in total and per site, and had larger variance in site richness. Focusing on macrofungi – the class Agaricomycetes, and in turn the order Agaricales – metrics of total richness and compositional similarity converged between the methods. eDNA was suboptimal for recording the non-soil dwelling fungi such as lichens and polypores. ß-Diversity was similar between methods, but more variation in community composition could be explained by environmental predictors in the eDNA data. The fruitbody survey was slightly better at finding red listed species. We find a better correspondence between biodiversity indices derived from fungal fruitbodies and DNA-based approaches than indicated in earlier studies. We argue that (historical) fungal community data based on fruitbody forays – with careful selection of taxonomic groups – may be interpreted together with modern DNA-based approaches to obtain a richer picture of the full mycobiota of the site, and for addressing historical changes. We estimated the costs of the two inventory approaches to be approximately similar for practical applications.