Abstract
The continuing pressure on natural resources and the expansion of urban development necessitates up-to-date knowledge about the distribution and status of different kinds of nature. Land-cover maps are created by mapping nature and serve as an important tool for sound decision-making in spatial planning and conservation of nature. Nature in Norway (NiN) is today implemented as the official system for mapping of Norwegian nature. NiN and other systems for land-cover mapping are sensitive to several issues pertaining to inaccuracy and low repeatability. As these issues stem from subjectivity, the possibility of adopting molecular tools in the mapping of nature types could be a desirable development, increasing objectivity. Metabarcoding is an increasingly popular tool for ecological research based on the use of high throughput sequencing of short, informative DNA fragments and DNA-based identification of species. This study aims to assess the feasibility of using eDNA metabarcoding to determine NiN types, circumventing issues concerning human error. To do this, soil samples were obtained from sites previously mapped using the NiN system using a robust consensus assessment process. Vegetation analysis was also performed at each site. DNA from the soil samples were amplified using the non-coding plastid trnL (UAA) intron P6 loop and the internal transcribed spacer 2 (ITS2) region of the nuclear ribosomal DNA. The resulting species lists were combined and subsequently used to assign each sample to a NiN type using Sørensen dissimilarity values. To verify the success of the method created for assignment of soil eDNA metabarcoding to nature types, the results obtained were compared to the “true” types from the previous study. Sørensen dissimilarity assigned the correct NiN minor type 22.6% of the times, while 64.5% of the samples were assigned to ecologically similar minor types. This is lower than the 65.0% correct assignments reported for human observers at the same type level in a previous study. Each sample had low standard deviation for assignment to types overall. Comparing results from metabarcoding and vegetation showed some discrepancies in species composition and samples displayed relatively high pairwise dissimilarity values. Investigation into the two important ecological variables used in the mapping of NiN types showed that one was more clearly present in the data acquired from metabarcoding than the other. The divergence between the types assigned to soil samples and the “true” types can possibly be explained by the fact that the studies were not performed simultaneously, introducing 7 uncertainty. The use of previously mapped sites limited the choice of nature type to study and the span of minor types within the chosen nature type, affecting the ability to make the correct assignment of type. The use of presence/absence data also affected the assignments as disproportionate weight is given to taxa, both disproportionately large and small. This study has shown that the assignment of NiN types from soil eDNA metabarcoding is possible, although not yet satisfactory in regard to type delimitation. For this method to be implemented in trials, there should be a goal of getting the correct assignment percentage up to an acceptable level, while reducing the cost of land-cover mapping.