Previous eDNA-based studies have revealed high microbial and metazoan diversity in perennial rivers28,29. However, although the contribution of dry riverbeds to ecosystem functioning and services of NPRs is increasingly recognized11–13, the biodiversity of dry riverbeds remains poorly characterized relative to perennial rivers. We applied genomic tools to samples collected during a global-scale, coordinated distributed experiment, and our approach is the first to reveal patterns and drivers in the biodiversity of dry riverbeds, a knowledge that is becoming increasingly critical in the context of global change as NPRs become more prevalent. Strong linkages occurred among the beta-diversity of Bacteria, Fungi, Algae, Protozoa and Nematoda, which, assuming adequate representation of the major abiotic variables, suggest that biotic interactions might be more important to the assembly of dry riverbed communities than environmental filters or spatial variables.
Community composition of dry riverbeds revealed by eDNA metabarcoding
Our analyses indicate that specific ecological adaptations to dry conditions and low-resource environments partially shape bacterial community assembly in dry riverbeds. Increased resource availability in the form of DOC and SRP promoted copiotrophic bacteria, whereas prolonged dry periods induced a shift towards oligotrophs. Taxa with an oligotrophic strategy likely outcompeted copiotrophs in dry conditions owing to their greater resistance to osmotic stress and higher investment in extracellular enzyme machinery, allowing them to cope with reduced substrate diffusion17,30,31. The selection of oligotrophic taxa under dry conditions might explain the observed increases in the relative abundance of Actinobacteria and Alphaproteobacteria and the decreases in the proportion of Betaproteobacteria and Bacteroidetes in response to sediment desiccation32,33. This suggests that mechanisms underlying the assembly of bacterial communities exposed to drying operate through adaptations to reduced water and resource availability, potentially leading to a progressive shift in bacterial communities toward terrestrial soil communities in NPRs during dry phases.
Such mechanisms could also apply to fungal communities, as sediment drying can promote an increase in the abundance of Basidiomycota33, which are expected to be oligotrophs and more resistant to dry conditions than copiotrophic Ascomycota30. Freshwater fungal communities are generally dominated by Ascomycota and Chytridiomycota, while Basidiomycota are less common in aquatic than in terrestrial habitats15,34. However, we found no indication that dry period duration affected the ratio of Ascomycota:Basidiomycota in NPRs. Instead, the relative abundance of Basidiomycota increased with increasing sediment C:N and land cover of forest within the catchment, suggesting that the inputs of complex organic matter in the form of leaf litter promoted fungal taxa with the ability to degrade recalcitrant polymers such as lignin35. The selection of distinct ecological strategies under dry conditions might have implications for organic matter decomposition and nutrient cycling31.
Our results provide new insights into biodiversity patterns of a broad range of taxa that have been largely disregarded in NPRs to date. In particular, Ciliophora and Cercozoa dominated the protozoan community as shown in soils36, perennial rivers sediments37 and both pond and stream sediments experiencing recurring drying events38. Protozoan communities play a key role in structuring microbial food webs through their consumption of bacteria, fungi, algae or detrital matter39. During dry phases, the relative contribution of Protozoa to organic matter and nutrient cycling might increase due to their better ability to cope with adverse environmental conditions than larger taxa15,40.
Most arthropods identified were Collembola, Ostracoda, and Arachnida (mainly mites), indicating that we captured the diversity of benthic or interstitial meiofauna, whereas eDNA of macroinvertebrates was poorly detected by our metabarcoding approach. Sediments can provide a refuge for some aquatic invertebrate resting forms during dry phases41, which would allow their detection by eDNA analyses. Yet, essentially terrestrial taxa such as Collembola or Arachnida strongly contributed to the arthropod community detected in our study. This observation supports previous reports that colonization of dry riverbeds by terrestrial arthropods from adjacent riparian areas contributes to the assembly of communities during dry phases20,42.
Resource availability and climate as drivers of OTU richness in dry riverbeds
Both indirect (latitude, climate) and direct drivers (relating to energy, nutrient and water availability) were important predictors of OTU richness in NPR sediments. However, considering the main predictors of OTU richness, most taxa were under the control of climatic variables, while bacteria were primarily influenced by resource availability. A latitudinal biodiversity gradient was detected for most taxa, with OTU richness peaking at mid-latitudes (absolute latitude between 30° and 45°) and decreasing at lower or higher latitudes. This finding aligns with previous studies reporting a hump-shaped relationship of fungal taxon richness with latitude in perennial streams43 but contrasts with conventional macroecological patterns. This pattern could reflect the indirect influence of other climatic variables that were negatively correlated with latitude, such as mean annual temperature or potential evapotranspiration44, for which a peak in OTU richness was also observed at intermediate values for most taxa. Temperature and potential evapotranspiration being positively correlated, this pattern could also derive from increased osmotic stress or reduced substrate diffusion under dry conditions45. Yet, lower temperatures might offset the benefit of increased water availability in colder regions44. Indirect effects on resource availability could also explain such patterns. Specifically, elevated temperature and associated high carbon mineralization rates at lower latitudes46 might also limit organic carbon accumulation in NPR sediments11,13, which, combined with increased evapotranspiration, would reduce water and energy availability for heterotrophic organisms.
The increase in OTU richness of heterotrophic organisms with the availability of organic resources aligns with the positive influence of organic matter inputs on microbial diversity in marine sediments47 and increases in soil biodiversity along resource/productivity gradients48. Conversion of carbon to microbial biomass allows its transfer to higher trophic levels (Nematoda, Protozoa, Arthropoda), although these metazoans also act as decomposers that consume detrital organic matter. The saturating relationship of OTU richness with carbon resources as opposed to a unimodal relationship suggests that competitive exclusion might not strongly affect the local diversity of heterotrophic communities in dry riverbeds, as reported from soils48. Nutrient limitation, particularly by phosphorus, appeared also to influence the biodiversity of dry riverbeds, as indicated by a sharp increase in OTU richness as sediment phosphorus content increased. This suggests that the biodiversity of NPR might be constrained by resource availability, at least in the most oligotrophic conditions. Furthermore, our results indicate that heterotrophic microbial communities might be more limited by carbon or phosphorus than by nitrogen availability49. However, while algal richness is typically constrained by nutrient limitation in freshwaters50, nutrient availability had less of an influence on algal and plant OTU richness in dry riverbeds than climatic variables. Algal richness rapidly decreased with increasing dry period duration, probably reflecting the reduced number of algal taxa able to persist in dry riverbeds under extended periods of water limitation51. Our results suggest that ongoing and predicted increases in temperature and aridity might negatively impact sediment biodiversity in dry riverbeds, especially under conditions of low resource availability.
The analysis of rarefaction curves indicated that the sequencing depth was sufficient to recover the diversity of eukaryotic organisms in our samples (Sup. Figure 10) but that part of the bacterial diversity in our samples remained uncovered (Sup. Figure 11). Increasing sequencing depth could have identified additional and potentially rare bacterial OTUs but the number of reads selected for rarefaction was sufficient to identify trends and relative variation in alpha-diversity among samples52 without altering our estimates of beta-diversity53,54.
Tight linkages between beta-diversity across multiple taxa
The weak correlations we observed in dry riverbeds between potential climatic, environmental, and spatial drivers of beta-diversity of different taxonomic groups suggest that interactions among taxa might have more influence than abiotic conditions or dispersal constraints. Partial correlation networks revealed a strong correlation between the beta-diversity of Bacteria and Fungi, which, assuming adequate representation of the major abiotic variables, could reflect biotic interactions of these two groups of microbes. Such interactions might include competition for and facilitation of organic carbon and nutrient acquisition. For example, Bacteria benefit from the release of labile molecules during the decomposition of complex polymers by diverse extracellular fungal enzymes55. The observed link between the beta-diversity of Bacteria and Algae may reflect bacterial consumption of algal exudates56. Strong associations of the beta-diversity of Protozoa and both Fungi and Algae also occur in perennial rivers57 and could reflect predation by protozoans on fungi and algae58. Similar relationships were observed between the beta-diversity of Nematoda and Bacteria, and to a lesser extent Algae and Fungi, in line with the different feeding habits of Nematoda59. Arthropoda were the only group whose beta-diversity did not closely correlate with that of other taxa, in spite of potential of arthropods to control the structure of benthic microbial decomposers and primary producers60.
Potential limitations
The results reported herein should be considered within the limitations associated with our sampling design and metabarcoding approach. Our global assessment was biased toward temperate and dry climates, a frequent limitation to global biodiversity studies61,62. Hence, further research is needed to increase the global coverage of NPRs, particularly in polar, boreal, tropical and continental climate zones. We also acknowledge the potential biases associated with metabarcoding targeting eDNA from sediment samples, which encompasses intracellular DNA from living, dormant and dead organism cells/tissues as well as extracellular DNA actively excreted by living microorganisms or passively released from lysed cells27. The use of eDNA could thus inflate biodiversity estimates and obscure relationships between habitat characteristics and taxa distribution. However, even if relic DNA makes a significant contribution to the total DNA pool, it might not bias eDNA-based biodiversity estimates63, as it generally originates from cell lysis or active release from the local community and thus also reflects its composition. Coupling eDNA with eRNA approaches might enable quantification of the relative contribution of active and non-active organisms to biodiversity estimates64, although eRNA approaches are limited by methodological constraints25.
As NPRs become increasingly prevalent due to ongoing global change, predicted increases in temperature and aridity could negatively affect sediment biodiversity in dry riverbeds, especially under conditions of low resource availability. The implications of these changes in biodiversity on associated ecosystem functioning and services remain unknown.