Matching habitat typology and ecological assemblages can be useful in environmental management. We examined whether a priori defined riverine sections correspond with distinct fish assemblage types along the >2000 km long course of the Danube River, Europe.We also tested whether different sampling methods (i.e. day and night inshore electric fishing and offshore benthic trawling) provide consistent typological results. Analysis of assemblage similarities, indicator species analysis, non-metric multidimensional scaling (NMDS) and kmeans analyses indicated that fish assemblages of the a priori defined Upper-, Middle and Lower-Danubian sections differed slightly, within class variability was high. Although indicator species analysis showed that the Upper-Danube belongs to the barbel (Barbus barbus) zone and the Middle-and Lower Danube belong to the bream (Abramis spp) zone, indicator values of the character species were generally low. The NMDS analyses suggested a weak gradient in assemblage structure along the course of the river with relatively high variability between neighbouring sites. K-means analyses revealed that many sampling sites were in a different class than the a priori defined sections, and classifications at other group numbers did not lead to better classification outcome. Overall, the results do not suggest clearly distinguishable assemblage types with distinct boundaries in the potamal section of a great river. Nevertheless, the division of the potamon to smaller sections may explain some variability in fish assemblage structure, and could be used for bioassessment purposes. The study also shows the importance of multihabitat and multigear surveys in the typological assessment of great rivers.

Environmental DNA (eDNA) metabarcoding is an effective method for studying fish communities but allows only an estimation of relative species abundance (density/ biomass). Here, we combine metabarcoding with an estimation of the total abundance of eDNA amplified by our universal marker (teleo) using a quantitative (q)PCR approach to infer the absolute abundance of fish species. We carried out a 2850-km eDNA survey within the Danube catchment using a spatial integrative sampling protocol coupled with traditional electrofishing for fish biomass and density estimation.Total fish eDNA concentrations and total fish abundance were highly correlated. The correlation between eDNA concentrations per taxon and absolute specific abundance was of comparable strength when all sites were pooled and remained significant when the sites were considered separately. Furthermore, a nonlinear mixed model showed that species richness was underestimated when the amount of teleo-DNA extracted from a sample was below a threshold of 0.65 × 10 6 copies of eDNA. This result, combined with the decrease in teleo-DNA concentration by several orders of magnitude with river size, highlights the need to increase sampling effort in large rivers. Our results provide a comprehensive description of longitudinal changes in fish communities | 397 PONT et al.

No abstract

eDNA metabarcoding is an effective method for studying fish communities but allows only an estimation of relative species abundance (density / biomass). Here, we combine metabarcoding with an estimation of the total abundance of eDNA amplified by our universal marker (teleo) using a qPCR approach to infer the absolute abundance of fish species. We carried out a 2850 km eDNA survey within the Danube catchment using a spatial integrative sampling protocol coupled with traditional electrofishing for fish biomass and density estimation. Total fish eDNA concentrations and total fish abundance were highly correlated. The correlation between eDNA concentrations per taxon and absolute specific abundance was of comparable strength when all sites were pooled and remained significant when the sites were considered separately. Furthermore, a non-linear mixed model showed that species richness was underestimated when the amount of teleo-DNA extracted from a sample was below a threshold of 0.65.106 copies of eDNA. This result, combined with the decrease in teleo-DNA concentration by several orders of magnitude with river size, highlights the need to increase sampling effort in large rivers. Our results show a comprehensive description of longitudinal changes in fish communities and underline our combined metabarcoding/qPCR approach for biomonitoring and bioassessment surveys when a rough estimate of absolute species abundance is sufficient.

In complement to the JDS4 traditional fish survey (TFS, mainly by electrofishing), a fish eDNA metabarcoding-based survey has been implemented within the framework of the monitoring program organized by DNAqua-Net and in collaboration with the INTERREG "MEASURES" program. Water samples were collected at 29 sites from the source to the mouth of the Danube River, and at 18 tributaries. Two water samples (mean volume 29 L per sample) were collected at each site and the water filtered in situ. Twelve PCR replicates were performed per sample using teleo primers. In total, 80 taxa were detected, of which 19 corresponded mainly to farmed fish or food fish due to eDNA release in waste waters, most often downstream of large cities. Of the remaining 61 taxa, 50 taxa were identified at the species level, six taxa at the genus level, and five taxa at a higher taxonomic level. Concerning the Danube river itself, 69 and 57 species/markers were detected along the Danube River by TFS and eDNA surveys, respectively. 50 of these taxa were detected by both methods. Nine species were captured by TFS alone, mainly due to gaps in reference librairies and too colosely related barcodes. Eight species were only detected by eDNA. Except for the Salvelinus group, these were all benthic species, which are difficult to catch by electrofishing in large rivers (Acipenser ruthenus, Acipenser stellatus, Benthophilus sp., Romanogobio uranoscopus, Sabanejewia balcanica, Umbra krameri). The species richness tended to increase from upstream to downstream and the relative number of DNA sequences per taxa showed a clear succession of species along the river. Barbatula barbatula, Cottus gobio, Hucho hucho, Lampletra planeri, Phoxinus phoxinus and Thymallus thymallus were restricted to the Upper Danube whereas A. ruthenus, Neogobius fluviatilis, S. balcanica and Scardinius erythrophtalmus were detected from Vienna to the Danube river mouth. Abramis brama, Alburnus alburnus, Cyprinus carpio, Silurus glanis, Sander spp, Zingel streber were detected all along the river course; Alosa spp. and Syngnathus abaster downstream from the Iron Gate; A. stellatus and U. krameri only on the most downstream site (Danube delta). The calculation of a fish index, based on the common metrics used to intercalibrate national fish assessment methods on a European scale, classified most of the sites as being in moderate ecological status. Comparison of the indicative ecological status calculated using the same assessment method based on TSF and eDNA data at the common Danube sites showed a similar classification for six of the 13 sites and a difference of one class for the remaining seven sites.