Understanding the ecological status of aquatic ecosystems and the impact of anthropogenic contamination requires correlating exposure to toxicants with impact on biological communities. Several tools exist for assessing the ecotoxicity of substances, but there is still a need for new tools that are ecologically relevant and easy to use. We have developed a protocol based on the substrate-induced respiration of a river biofilm community, using the MicroRes (TM) technique, in a pollution-induced community tolerance approach. The results show that MicroRes (TM) can be used in bioassays to assess the toxicity toward biofilm communities of a wide range of metals (Cu, Zn, Cd, Ag, Ni, Fe, Co, Al and As). Moreover, a community-level physiological profile based on the mineralization of different carbon substrates was established. Finally, the utility of MicroRes (TM) was confirmed in an in-situ study showing gradient of tolerance to copper correlated to a contamination gradient of this metal in a small river. A modified MicroRes (TM) technique as a tool for measuring induced tolerance to heavy metals of a microbial biofilm community.

We consider an optimal consumption and pollution problem that has two important features. Environmental damages due to economic activities may be irreversible and the level at which the degradation becomes irreversible is unknown. Particular attention is paid to the situation where agents are relatively impatient and/or do not care a lot about the environment and/or Nature regenerates at low rate. We show that the optimal policy of the uncertain problem drives the economy in the long run toward a steady state while, when ignoring irreversibility, the economy follows a balanced growth path accompanied by a perpetual decrease in environmental quality and consumption, both asymptotically converging toward zero. Therefore, accounting for the risk of irreversibility induces more conservative decisions regarding consumption and polluting emissions. In general, however, we cannot rule out situations where the economy will optimally follow an irreversible path and consequently, will also be left, in the long run, with an irreversibly degraded environment.

This text focuses on the identification, efficiencies, classification and management of landscape features having a potential buffer function regarding diffuse phosphorus, because of their specific structure (vegetation-soil) and of their location at the interface between sources (farm infrastructures, emitting fields…) and surface water bodies. These buffers are very diverse and correspond to natural landscape features (wetlands, riparian areas…) as well as manmade structures (constructed buffer strips or intermediate cases such as field margins, hedgerows). Their role and efficiency depends on the local factors controlling the retention processes (internal organisation and properties of the buffer), on the position within the watershed, and on the landscape context which reciprocally determines the overall buffer capacity of a watershed. On that basis, we recognize the diversity of the buffers in structure and functioning and thus in the way they attenuate the signal, their limitations (sustainability, side effects) and their hierarchic organisation at the watershed scale.

Dans les bassins versants méditerranéens où les zones cultivées en vigne occupent une part importante de l'espace, les pratiques culturales sont un facteur important d'apparition de risques de pollution diffuse des eaux par les pesticides. Au cours de la thèse, une démarche de modélisation couplée pression-impact a été développée. La question de recherche a porté sur la prise en compte dans le modèle construit de la dynamique des variables d'état du milieu modifiées par les actions culturales. Deux variables ont été identifiées : l'infiltrabilité de la surface du sol et la concentration en matières actives au sol. A l'échelle du bassin versant, pour réaliser l'évaluation des impacts, on s'est appuyé sur l'utilisation du modèle hydrologique distribué MHYDAS qui permet de simuler les flux d'eau et de polluants selon différents scenarii de pratiques et climatiques. Pour représenter la répartition spatio-temporelle des opérations culturales sur chacune des parcelles du bassin versant, le modèle décisionnel DHIVINE a été construit. Il permet de simuler les itinéraires techniques des viticulteurs. Pour chaque exploitation du bassin versant, le modèle est basé sur une représentation de la conduite technique annuelle du vignoble sous forme de plans d'activités. Ces plans mettent en oeuvre des indicateurs d'états de l'agrosystème et prennent en compte les caractéristiques des ressources productives de l'exploitation ainsi que les arbitrages à réaliser entre opérations culturales et parcelles concurrentes. Des modèles biophysiques ont été développés pour simuler les indicateurs de la décision ainsi que l'impact des différentes pratiques sur les états de surface du sol qui conditionnent l'infiltrabilité du sol en surface. Une typologie fonctionnelle des états de surface basée sur leurs propriétés d'infiltrabilité du sol a été utilisée pour développer une démarche visant une modélisation continue dans le temps de l'infiltration. Elle intègre une modélisation parallèle de la dynamique des états de surface du sol nu et du développement de la couverture herbacée par les adventices. Un couplage des modèles a été réalisé sous la plate-forme OpenFLUID. La démarche adoptée ici permet de représenter les dynamiques différentes des variables d'état du milieu en fonction de différentes pratiques culturales et semble adaptée pour l'évaluation des impacts en terme de pollution de l'eau par les pesticides à l'échelle de bassins versants viticoles méditerranéens. | In Mediterranean catchments in which vineyards account for a major share of the area, cropping practices are an important factor with respect to increasing risks of diffuse pesticide pollution of water. This study was about developping an approach for pressure-impact modelling. The research question considered dynamics of agrosystem status variables modified by cropping actions. Two variables were identified : soil surface infiltrability and active material concentration on the soil. On a catchment scale, to assess impacts, we used the distributed hydrological model MHYDAS to simulate water and pollutant flows according to different cropping practice and climatic scenarios. To represent the spatiotemporal distribution of cropping operations on each plot, the decision model DHIVINE was built. It simulates vinegrowers's crop management sequences. For each farm in the catchment, the model is based on a representation of the vineyard annual technical management strategy in the form of activity plans. These plans involve agrosystem status indicators and take into account the characteristics of productive resources on the farm and arbitration between cultivation operations and competing plots. Biophysical models were developped to simulate decision indicators and impacts of cropping practices on soil surface characteristics which are drivers of soil surface infiltrability. A functional typology of soil surface states based on their soil infiltrabilty properties was used to develop an approach for temporally continuous modelling of infiltration. It was based on a parallel modelling of dynamics (i) of soil surface characteristics of bare soil and (ii) of weed cover development. Model coupling was realized with OpenFLUID platform. The approach developped herein allows simulating dynamics of agrosystem status variables as a function of cropping practices. It seems adapted for assessing hydrological impacts at the catchment scale in Mediterranean environments.