Environmental DNA reveals that rivers are conveyer belts of biodiversity information

Abstract

DNA sampled from the environment (eDNA) is a useful way to uncover biodiversity patterns. By combining a conceptual model and empirical data, we test whether eDNA transported in river networks can be used as an integrative way to assess eukaryotic biodiversity for broad spatial scales and across the land-water interface. Using an eDNA metabarcode approach, we detect 296 families of eukaryotes, spanning 19 phyla across the catchment of a river. We show for a subset of these families that eDNA samples overcome spatial autocorrelation biases associated with the classical community assessments by integrating biodiversity information over space. In addition, we demonstrate that many terrestrial species are detected; thus suggesting eDNA in river water also incorporates biodiversity information across terrestrial and aquatic biomes. Environmental DNA transported in river networks offers a novel and spatially integrated way to assess the total biodiversity for whole landscapes and will transform biodiversity data acquisition in ecology.

Figure 1. Study area and location of sampling sites where environmental DNA samples and classical sampling methods were carried out.

The direction of flow for the river Glatt is northwest (blue arrow). The main stem of the river originates from the outflow of lake Greifensee. Scale bar, 2 km. Coloured regions represent the catchment upstream of each sampling point. Letters are used to indicate the position in the river network starting from the outflow ‘a' to ‘f' and the two sampled tributaries ‘ab' and ‘cd'. Sources for GIS data were from Swisstopo (DHM25, Gewässernetz Vector 25) and reprinted with permission.

Figure 2. Total eukaryotic diversity detected from the river Glatt using environmental DNA metabarcoding.

(a) The number of families per phylum (N=296) sampled and confirmed as known to be present in Switzerland or known from all four neighbouring countries (Austria, France, Germany and Italy). The inset further breaks the most abundant phylum (Arthropoda) into the number of families sampled by class (N=196). (b) Grey bars depict the number of sequences receiving a taxonomic identification at the family level for each phylum.

Figure 3. Percent terrestrial or freshwater species for the subset of each phylum detected in eDNA.

Species were confirmed as known to be present in Switzerland or known from all four neighbouring countries (Austria, France, Germany and Italy; N=255). Number in brackets indicates the number of species confirmed for each phylum.

Figure 4. Difference of benthic macroinvertebrate family richness and community dissimilarity estimated between the environmental DNA and kicknet sampling.

(a) α-diversity measured at each site and log₁₀ transformed taxon–area relationship for eDNA and kicknet samples. Slopes of line and the y intercept are significantly higher for eDNA compared with kicknet (P<0.0001, P=0.0002, respectively), indicating that eDNA samples a greater amount of diversity over a greater area compared with kicknets. (b) Correlation of community dissimilarity with along-stream geographic distances between sample sites. Solid line for kicknet β-diversity indicates a significant positive relationship with stream distance (P=0.005). Dashed line for eDNA β-diversity indicates no significant relationship with distance (P=0.44).

Figure 5. Conceptual model of environmental DNA dynamics in a hypothetical river network.

While classical sampling only detects a fraction of real local diversity due in part to spatial autocorrelation, eDNA sampling allows an estimate of catchment-level diversity, including both aquatic and terrestrial taxa, and integrates this information across space due to downstream transport of eDNA.