Discussion
Recent microcosm experiments and model simulations have unravelled how lateral (Harvey et al. 2020) and longitudinal (Jacquet et al. 2022a) exchanges of resources can shape network-scale biodiversity and how drying patterns may influence the dynamic of resource processing (Catalàn et al. 2022) and communities (Jacquet et al. 2022b) within river networks. However, comprehensive field-based evidence is missing. Our network-scale field study confirms the paramount effect of drying in governing resource stocks, community composition, ecosystem functioning and their relationships in river meta-ecosystems. As expected in H1, flow intermittence had a positive effect on instream leaf quantity, but a negative effect on shredder abundance and hence on organic matter decomposition rates. Partially agreeing with H2, invertebrate communities and decomposition rates changed throughout the river network in response to upstream connectivity, but not always linearly. Interestingly, invertebrate-driven decomposition peaked at intermediate levels of upstream fragmentation, revealing the potential positive effects of upstream drying on the functioning of downstream ecosystems. Instream and riparian responses were generally weakly related but the relationship between riparian and instream decomposition rates changed with flow permanence (FK; H3) and network location (CK; H4), suggesting that the links between riparian and instream ecosystem functioning can change depending on the disturbance regime and network-scale connectivity. Although invertebrate richness did not directly relate to decomposition, community composition and shredder abundance did, and the strength of these relationships was lower among intermittent and headwater sites (H5), suggesting that flow intermittence and low connectivity may induce mismatches between community composition and ecosystem functioning.
Although the negative effects of flow intermittence on decomposition can be linked to changes in leaf resource nutritional quality (del Campo et al. 2021a), most field studies attribute these negative effects to losses of primary consumers such as shredders (Datry et al. 2011, Schlief and Mutz 2011, Abril et al. 2015, Northington and Webster 2017). We found that increasing flow intermittence had a seasonal positive effect on instream resource quality and quantity, but the negative effects on invertebrate richness and shredder abundances drove the associated decreases in decomposition. Our results echo those from Catalàn et al. (2022), who demonstrated through simulations that coarse particulate organic matter (e.g. leaves) tend to accumulate and remain unprocessed – thus conserving a higher reactivity – on dry riverbeds. Such accumulation can result from decreases in transport caused by low or null water discharges and the absence or low abundance of efficient decomposers such as aquatic shredders (Northington and Webster 2017). In our study, decomposition was associated to invertebrate community variability among perennial but not intermittent sites, suggesting that flow intermittence may create mismatches between community structure and ecosystem functions, here decomposition. Such mismatches may result from the variable levels and paths of recovery communities are on in these frequently disturbed environments, preventing communities from reaching equilibrium and thus use available resources optimally (Fukami et al. 2010, Brose and Hillebrand 2016, Datry et al. 2016). Conversely, in hydrologically more stable, perennial sites, communities are more likely to be at equilibrium with their environment and their composition may better reflect resource availability and uses. As commonly found in the literature, microbially driven decomposition was little affected by flow intermittence, likely as a result of the microorganisms’ capacities to (1) sustain activity during dry phases, especially if moisture is preserved and (2) to recover their activity within days to weeks of flow resumption (Foulquier et al. 2015, Arroita et al. 2018, Truchy et al. 2020).
River dendritic structure, i.e. multiple small streams branching into larger rivers, promotes gradual increases in biodiversity (Finn et al. 2011, Altermatt 2013) and metabolic activity (Diamond et al. 2021) as influx of organisms and resources (e.g. nutrients and dissolved organic matter) increase with water discharge from up to downstream. Because of this structure, communities in headwaters typically respond more strongly to disturbance than in the mainstem where mass dispersal can override the negative effect of disturbance (Tornwall et al. 2017). However, responses to natural drying can contrast with this pattern due to the fragmenting effect of drying, reducing connectivity and dispersal even in the mainstem (Crabot et al. 2020, Gauthier et al. 2020). We found that invertebrate richness responded more strongly to drying in the mainstem than in headwaters. This bigger loss of species with increasing intermittence in the mainstem sites may thus result from reduced mass effects in this fragmented network and from the higher richness typical of low-order perennial rivers, where communities have more to lose when facing a disturbance than already species-poor headwaters. We also found that invertebrate-driven decomposition increased with the distance to the source and was related to shredder abundance in headwaters only, agreeing with the theory that ecosystem functioning and BEF relationships should peak at intermediate levels of network connectivity (Leibold et al. 2017). Stronger relationship between decomposition and community multidimensional variability in headwaters than in mainstems may also reflect the high dependence of smaller, shaded streams on leaf resources and the stronger community specialization towards the use of leaf resource in the former. Contrastingly, microorganism-driven decomposition and leaf litter quality increased linearly with distance to the source, suggesting that microorganisms activity may increase from upstream to downstream, at least during the summer season, when microbial activity is likely boosted by higher temperature (Friberg et al. 2009). This may in turn increase leaf litter quality through microbial nitrogen immobilization from the water column.
The effects of drying on communities and ecosystem functions are relatively well documented at local reach scales (Datry et al. 2011, Foulquier et al. 2015), whereas network-scale responses owing to fragmentation remain overlooked, although evidence is emerging for fish (Jaeger et al. 2014) and invertebrate communities (Gauthier et al. 2020, Sarremejane et al. 2021). The proportion of perennial reaches upstream had a positive effect on leaf stocks, shredder abundance and invertebrate-driven decomposition, suggesting that connectivity to upstream perennial habitats is key in determining downstream leaf transfer (Catalàn et al. 2022), communities (Pineda-morante et al. 2022, Fournier et al. 2023) and ecosystem functions (Harvey et al. 2017). Interestingly, decomposition was higher when upstream connectivity was intermediate, suggesting that intermediate extents of flow intermittence upstream may promote higher functional rate in downstream habitats. In rivers, where the flux of resources and organisms is directional (i.e. dictated by the unidirectional flow of water), downstream ecosystem functions may thus depend on the disturbance regime in – and connectivity to – upstream habitats. An intermediate amount of upstream intermittence could therefore promote downstream decomposition rates by providing 1) pulses of high quality leaf resources and 2) influx of diverse species best adapted to resource use, from a variety of upstream habitats, following rewetting events (Northington and Webster 2017, del Campo et al. 2021a, Catalàn et al. 2022).
Decomposition is usually faster in aquatic than in riparian habitats due to moisture limitation hindering microbial activity and the presence of less efficient and abundant detritivore invertebrate communities in the latter (Hutchens and Wallace 2002, Tiegs et al. 2019, Lohse et al. 2020, Simões et al. 2021). Accordingly, we found lower microorganism- and invertebrate-driven decomposition, shredder abundance and invertebrate richness in network-wide riparian habitats. We found weak evidence of relationships between instream and riparian invertebrate communities and changes along flow permanence gradients, suggesting that community overlap between these habitats was limited. Although long dry phases (i.e. > year) may lead to invertebrate community overlap between riparian and instream habitats due to convergences in environmental characteristics (Steward et al. 2022), the dry phases the Albarine experienced in 2021 were likely too short (< 2 months) to promote such community homogenization. However, the relationship between instream and riparian invertebrate-driven decomposition, and more weakly so instream and riparian invertebrate richness, changed with the distance to the source. The negative relationships observed in headwaters may suggest a compensation phenomenon where high invertebrate community richness and decomposing activity in riparian habitat may promote low richness and decomposition in adjacent instream habitats, and vice versa (Abelho and Descals 2019). This may occur if riparian decomposers use high quality compounds, leaving instream communities with lower quality resource entering through runoff, though we did not observe a negative relationship between leaf litter quality and quantity across both habitats. Conversely, microbial-led decomposition instream and in riparian habitats were positively related and this relationship was stronger among intermittent sites, suggesting that the microbial decomposer communities in both habitats may be related and more strongly so as flow intermittence increases. Instream and riparian microbial communities are likely to interact as suggested by compositional overlap between riparian soil and instream bacterial communities (Ruiz-González et al. 2015) and between senescent leaf and instream leaf-litter fungal communities (Koivusaari et al. 2019). Site-specific homogeneity in environmental characteristics such as humidity may also explain co-variability between riparian and instream microbial decomposition, particularly in headwaters where tree canopy may help preserving humidity in both the riparian and instream area.