Eutrophication is one of the most common causes of water quality impairment of inland and marine waters. Its best-known manifestations are toxic cyanobacteria blooms in lakes and waterways and proliferations of green macro algae in coastal areas. The term eutrophication is used by both the scientific community and public policy-makers, and therefore has a myriad of definitions. The introduction by the public authorities of regulations to limit eutrophication is a source of tension and debate on the activities identified as contributing or having contributed decisively to these phenomena. Debates on the identification of the driving factors and risk levels of eutrophication, seeking to guide public policies, have led the ministries in charge of the environment and agriculture to ask for a joint scientific appraisal to be conducted on the subject. Four French research institutes were mandated to produce a critical scientific analysis on the latest knowledge of the causes, mechanisms, consequences and predictability of eutrophication phenomena. This paper provides the methodology and the main findings of this two years exercise involving 40 scientific experts.

1. Relationships between fish and their habitat over whole geographic regions, which are evident from studies of many streams and species, can improve understanding of lotic communities and provide reliable management tools. Nevertheless, most habitat preference studies have been based on single sites, and confined to small streams and to game species. 2. Regional habitat preference models, based on local velocity, depth and roughness, were developed for twenty‐four species and their size classes commonly found in large European streams. Fish surveys were conducted in six large streams in southern France over an 8‐year period. To limit the influences of habitat variables other than those studied, we estimated fish preferences within each survey and averaged this information across surveys. Preferences were fitted with confidence intervals and their sensitivity to field uncertainty was evaluated. 3. Most species and size classes had significant preferences for local habitat conditions which were consistent across the region. Habitat preferences predominant in the region overall were not always observed at any one site, but habitat conditions preferred on average in the region were never actually avoided locally. These results support the use of regional preference models for fish and the development of similar models for other lotic groups whose sensitivity to local habitat conditions has been reported elsewhere.

No abstract

1. One current approach to the prediction of community characteristics is to use models of key local‐scale processes (e.g. niche dimensions) affecting individuals and to estimate the effects of these attributes over larger scales. We tested this approach, focusing on how the hydraulic habitat structures fluvial fish communities. 2. We used a recent statistical habitat model to predict fish community characteristics in eleven reaches in the Rhône river basin in France. Predictions were made ‘blindly’ since most reaches were not used to calibrate the model. The model reflects species preferences for local hydraulics. We made predictions of the fish community from the local hydraulic conditions found in the reaches under low flow conditions. The overall abundance and the relative abundance (both as indices) of fish species, specific size classes and species traits (i.e. reproductive, trophic, morphological and others) were predicted. We summarized our predictions of the relative abundance of species as two ‘community structure indices’ using Principal Component Analysis. 3. Our predictions from low‐flow hydraulics were compared with long‐term observations of fish communities. The relative abundance of species actually observed depended largely on zoogeographic factors within the Rhône basin which could not be predicted by the model. The model predicted 13% of the variance in the indices of relative abundance at the species level and 23% of this variance at the trait level for all zoogeographic regions combined. However, when focused on reaches within a geographic region, the model explained up to 47% of the same variance. Therefore, geographic regions act as ‘filters’ on the relative abundance of species, but hydraulics do affect fish communities within a given geographical context. 4. For the synthetic ‘community structure indices’, we obtained good predictions from hydraulics independently of the geographical context (variance explained up to 95%). These indices were linked to simple key hydraulic characteristics of river reaches (Froude and/or Reynolds number). The indices enabled interpretations of the links between hydraulics, geomorphology, discharge and community patterns. These links were consistent with existing knowledge of species and their traits. 5. In addition to the above validations, the habitat model partly explained the observed effects of impoundment on fish communities. 6. The present results show that stream hydraulics strongly impact fish community structure. Consequently, our findings confirm that community characteristics can be predicted using models of the local‐scale habitat requirements of the species forming the community.

The quantification of management impacts on fish populations requires a reliable estimation of the local hydraulic habitat variability. In the case of complex natural flows, deterministic hydraulic models are expensive and not adapted to the description of local velocities. Statistical descriptions of the velocity distribution as a function of easily obtained input variables are therefore an attractive alternative. Existing velocity data on several French stream segments with intermediate-to large-scale roughness were analyzed statistically to define a shape parameter of the point velocity frequency distributions. Dimensional analysis was used to model this shape parameter, and thereby the velocity distributions, as a function of simple average descriptors of stream reaches (discharge, mean roughness, mean depth, and mean width). Such predictive models of the physical habitat availability for fish in the probability domain can provide stream managers with cost-effective decision tools. low-flow conditions, because of the violation of the common assumptions of uniformity and the unpredictability of turbu-2367 2368 LAMOUROUX ET AL.: PREDICTING VELOCITY FREQUENCY DISTRIBUTIONS LAMOUROUX ET AL.: PREDICTING VELOCITY FREQUENCY DISTRIBUTIONS 2369