1. Natural experiments, in the form of disturbance from spates, were used to study the resistance and resilience of interstitial communities. Investigations were conducted in a by‐passed section of the Rhône River characterized by an artificial hydrology with frequent spates separated by regular minimum discharge of 30 m3 s–1. 2. Three areas of a bar were studied, upwellings at the head of the bar (stations 1 and 2), and downwelling at the tail of the bar (station 3). In the head of the bar the substratum was characterized by stable cobbles, while mobile gravels dominated in the tail of the bar. At each station, samples were derived from four depths (0.5, 1.0, 1.5 and 2.0 m below the surface of the substratum). Fifteen spates occurred during the study period whose peak discharge ranged from 50 to 1640 m3 s–1. Temporal variations of the fauna were studied by comparing the spate effect observed 1 day (resistance), 7 days (resilience) and 17 days after the spate. Within‐class correspondence analysis was used to compare the temporal variability of the fauna within each class {station/depth}. 3. The fauna differed markedly between the three stations, and the relative density of stygobionts (i.e. hypogean fauna) decreased from 55% at station 1 to 4% at station 3. The spatio‐temporal variability increased dramatically from station 1 to station 3. 4. The results suggest that the hyporheic zone acts as a patchy refugium: the stations were more or less active refugial zones, depending on hydrology (upwelling or downwelling), substratum stability and spate amplitude. 5. The downwelling station was the main refugium area for benthic taxa. Important migrations of benthic groups (e.g. Gammarus, Cladocera) or hyporheic taxa (e.g. Cyclopoida and Harpacticoida) were observed deep into the sediment (2 m). Vertical movements of stygobionts (Niphargus, Niphargopsis) were also observed at high amplitude spates. These movements were very important (great numbers of individuals migrated) at low and medium magnitude spates, but were unimportant at high discharge, when the threshold of sediment instability was exceeded. In this case the substratum became mobile and induced drift of benthic organisms. 6. Conversely, in the upwelling stable stations, accumulation was less important (lower number of species and lower densities) but more constant with increasing discharge, suggesting that substratum stability is also a key factor. 7. Generally recovery was rapid at all stations (within 7 days) but no relationships were found between resilience (rate of recovery) and the amplitude of spates.

Summary Copepoda, Ostracoda and ‘Cladocera’ are important meiobenthic Crustacea which can be both numerically abundant and species rich in running waters. Harpacticoids and ostracods are well adapted to benthic life because they are typical crawlers, walkers, and burrowers. Many cladocerans are substratum dwellers, but most benthic species among these can also swim. Cyclopoids which are generally good swimmers are nevertheless often bottom frequenters and actively colonise sediment interstices (the hyporheic zone). For each of the three major taxa, morphological characteristics are presented, specimen collection and preparation are described and references to available taxonomical keys are provided. Biological characteristics are extremely diverse among and within the three taxa, resulting in a great variety of strategies in meiobenthic crustaceans. Characteristics of reproduction, sexual dimorphism, cyclomorphosis and population parameters (i.e. clutch size, lifespan, growth, moulting) are provided for some of the most common species. Important differences between the three main taxa were found at the species level. Ecological requirements such as hydraulic microhabitats and geomorphologic features of the streambed are the major determinants of species diversity and abundance for benthic microcrustacea of lotic habitats. Many studies on the ecology of these communities are limited by a lack of knowledge of the life history characterisitics of lotic (especially interstitial) crustacean populations.

1. This paper develops a framework of spatial and temporal variability for a habitat typology of the Upper Rhone River (France) and its alluvial floodplain that is based on about 17 years of data collection and analysis. The aim was to provide a scale of spatial-temporal variability for river habitat templet predictions on trends in species traits and spedes richness. 2. In developing this framework, eight physical-chemical variables were available and could be considered for twenty-two habitat types: seventeen superfidal (surface) and five interstitial (0.5 m below the substrate surface). These habitat types were selected in two areas (Jons and Bregnier-Cordon) after geomorphological considerations and because of differences in their biological characteristics. 3. The data sets used were processed by a 'fuzzy coding' method using, for each variable, the frequency distribution (by modalities = categories) of all measurements and monthly means over an annual scale. Two tables were produced; the first corresponded to an expression of the total variability, and the second represented an evaluation of the temporal variability. 4. Each of these tables was analysed by correspondence analysis, which provided factorial scores that were used to calculate, by habitat type and by variable, a total variability and a temporal variability in terms of cumulated variability of factorial scores for the eight physical-chemical variables. The rationale in describing variability from these two tables is that total variability equals temporal variability plus spatial variability. The spatial variability was then detennined by the difference between total and temporal variability. From this procedure, a positioning of the twenty-two habitat types on the spatial and temporal variability axes was obtained. 5. The estimate of spatial variability did not consider any error term that may have occurred in the above model; it was then tested by an independent assessment of the spatial variability using thirteen variables in nine major habitat types. A high correlation between the two ways of assessing spatial variability (r = 0.85, P< 0.004) underscored the reliability of the spatial variability that was calculated previously. 6. The river habitat templet obtained for the Upper Rhone and its alluvial floodplain appears to be appropriate to test the predictions on patterns of spedes traits and spedes richness in the framework of spatial and temporal variability. 311 312 B. Cellot et al.