During the last decades, many efforts have been invested to predict hydrochemical changes under climate change. These studies generally couple outputs from Global Circulation Models (GCM) or, more recently from Regional Climate Models (RMC) and several downscaling techniques (statistical and dynamics) which transpose regional predictions to a local scale, so that the outputs can be used by hydrochemical models to predict future changes in hydrochemistry. This study will be pioneering as it will be one of the first to link global change to the ecological response through a modelling framework. This is achievable since the project will use the component modelling techniques developed in Euro-limpacs; what will be new is linking of these component models. The project has been already initiated from the collaboration between Dr. Andrew Wade, from the Aquatic Environments Research Centre in Reading , and Prof. Sovan Lek from the CNRS in Toulouse ( France ), within the framework of the EUROLIMPACS European project. As such I aim to maximise the value for money by working for a PhD utilising my research efforts on the project. I have already calibrated the HBV ( Arheimer and Wittgren, 1994 ) and INCA ( Whitehead et al. 1998 ) hydrological and hydrochemical models to simulate flow and nitrate dynamics in the Garonne basin and these models have been set up for the Lambourn. My future objectives are: i) to apply several downscaling techniques to predict nutrients and discharge fluxes under future change scenarios; ii) to calibrate and validate several pressure/impact models to predict freshwater fish biodiversity according to environmental factors; iii) to make the link between the down-scaled scenarios and the coupled models to predict future changes on freshwater fish biodiversity ; iv) to analyse the sensitivity and uncertainty in our results to make emerge some coherent propositions in term of future management of freshwaters. Table 1 gives a listing of the different steps expected during the next two years, with the third year mainly assigned for writing the thesis.

Conservation biology has been developed for decades. Formerly, the species was listed in the international and national red lists based on the long-term field observation. Recently, quantitative methods are used widely in conservation biology, such as population viability analysis, conservation hotspots, and conservation priorities and so on. Objective evaluation of population decline and species extinction risk is conducted for sound decision making.

There are 163 fish species in the upper Yangtze like a tremendous biological diversity storeroom. 44 endemic species of these were threatened by anthropic activities, such as overfishing, pollution, and damming. Some species can be pushed on the verge of extinction. However, more hydroelectric projects, which modify physical environment and regulate flow, are built and planed, such as Three Gorges dam. The conservation strategies instituted for preserving the endemic fish resource will be of urgent need, such as restocking. However, the funding can not satisfy the conservation need. Therefore, these strategies for those species which have high risk should be of priority. It is necessary to identify those most endangered fish. One way is to model probability of extinction of each species by population viability analysis. However, we are far from predicting extinction, except of a few well-studied species (Brook et al, 2000) . Alternative approach is logistic regression model which had been well used by some studies (Polishchuk, 2002) . In this study, logistic regression model will be used to estimate the conservation priories of these endemic fish in order to direct the allocation of the funding.

Stream fish life-history traits are currently used as bioindicators of perturbations in the aquatic environment. This PhD project aims at performing a holistic approach in the assessment of the ecological status of streams and rivers in the Garonne river basin. Perturbations such as climate change and contaminants (metals and pesticides) will be evaluated and linked to alterations in stream fish assemblages and their life-history traits.

Data on physical-chemical parameters, hydromorphology and fish assemblages, collected during the last decade at various sites along the Garonne basin, will be used to perform multivariate analysis and statistical models. Further field sampling will be carried out along a gradient of chemical perturbations, focusing on target fish species. Traits such as growth, reproduction, morphology and health will be registered and further analysis performed. To test the out-coming hypothesis of the effects of toxicants on fish biological responses, laboratory experiments will be carried out with juveniles of autochthon species. Sub-lethal levels of key-pollutants will be checked for their effect on the behaviour, performance and general well-being of the exposed individuals.

All obtained data should provide a base on which to build robust assessment tools of the current levels of biodiversity and ecological status of streams in South-West Europe. Practical methods for performing such assessments will be established.

Abstract :

In a context of river biodiversity loss, this work aims at understanding the influence of hydrological connectivity on fish distribution in the Loire catchment at different spatial (the whole catchment and the floodplain) and biological (the assemblage and the European eel's population) scales. The results show that hydrological connectivity strongly drives fish distribution patterns at both spatial scales. It determines the habitat quality and the composition and structure of fish assemblages. Low connectivity levels create lentic habitats that favour exotic species and disfavour threatened native species. Connectivity also determines fish opportunities to move between various elements of the hydrosystem as it has been shown for the European eel.

Résumé :

L'altération de la connectivité hydrologique est considérée comme un des principaux facteurs responsables de l'érosion de la biodiversité dulçaquicole, et notamment de l'ichtyodiversité. L'objectif de cette thèse était d'analyser le rôle de la connectivité hydrologique sur la distribution de l'ichtyofaune du bassin versant de la Loire à différentes échelles spatiales – celle du bassin versant et celle de la plaine inondable de la Loire aval – et biologiques – celle de l'assemblage et celle de la population d'anguille européenne ( Anguilla anguilla ). Nous avons pour cela utilisé deux sources de données : celles concernant l'échelle du bassin versant sont issues du Réseau Hydrobiologique et Piscicole du Conseil Supérieur de la Pêche, et celles concernant la plaine inondable sont issues de nos propres campagnes d'échantillonnage par pêche électrique.

A l'échelle du bassin versant de la Loire, l'étude des assemblages de poissons par la méthode des cartes auto-organisatrices (SOM) révèle une organisation le long du gradient longitudinal en cohérence avec la zonation de Huet. Cependant, les assemblages typiques des secteurs lentiques (zone à "brème") sont présents en amont de leur distribution théorique suggérant une évolution des secteurs intermédiaires lotiques vers des secteurs lentiques (zones à "barbeau" et à "ombre"). A l'échelle de la plaine inondable, le gradient de connectivité hydrologique entre le fleuve et les annexes hydrauliques dans la dimension latérale génère une forte hétérogénéité environnementale qui permet la coexistence d'espèces aux stratégies écologiques variées. Nous avons ainsi pu décrire une zonation latérale des assemblages de poissons. L'analyse des guildes de préférence de courant montre que l'hydrologie est une variable déterminante qui structure les assemblages le long des gradients longitudinal et latéral selon des patrons assez comparables et en cohérence avec le River Continuum Concept et le Flood Pulse Concept . En outre, les espèces exotiques rencontrées sont pour la plupart des espèces colonisant les habitats lentiques. A l'opposé, les espèces vulnérables étaient surtout liées aux milieux lotiques qui sont les plus fragiles.

La population d'anguille européenne de Loire est apparue très structurée dans le gradient longitudinal. Nos résultats confirment, à l'aide d'un modèle logistique basé sur l'analyse des données de présence-absence de différentes classes de tailles, l'existence de patrons de distribution taille-dépendants. Les anguilles de moins de 150 mm sont localisées surtout dans les secteurs aval et le front de colonisation progresse vers l'amont puis se stabilise pour les anguilles de plus de 450 mm . Nos résultats suggèrent que l'altération de la connectivité longitudinale par des obstacles entraîne un ralentissement de la colonisation des jeunes anguilles et que le niveau de recrutement fluvial est actuellement trop faible pour stimuler la colonisation des zones amont. Par ailleurs, dans le gradient latéral de la plaine inondable, la distribution des anguilles est homogène quelque soit le degré de connectivité entre les annexes et le cours principal, sauf pour les plus petites anguilles (moins de 150 mm ) qui sont plus abondantes dans les milieux connectés. Ces anguilles, récemment arrivées en eaux douces et entreprenant leur migration au printemps, n'ont probablement pas pu bénéficier des connexions lors des crues hivernales pour coloniser les annexes les plus isolées.

L'ensemble de nos résultats montrent la nécessité de maintenir les patrons naturels de connectivité hydrologique pour, d'une part, garantir une diversité d'habitats suffisante et, d'autre part, permettre aux poissons de se déplacer librement entre éléments de l'hydrosystème, que ce soit à l'échelle du bassin versant dans son ensemble ou bien à celle de la plaine inondable.

In this study, we applied the SOM for ecological quality assessment using benthic macroinvertebrates in aquatic ecosystem. First, Chapter I, SOM was utilized to extract information from complex data of environmental variables and benthic macroinvertebrate communities residing in different micro-habitats. Although the sampling was carried out in a limited area, the patterns of environmental variables revealed spatial heterogeneity. The clustering of benthic macroinvertebrate communities in the trained SOM was efficient in showing temporal variation and evaluating water quality according to the conditions of different micro-habitats. Consequently, local spatial heterogeneity is important in revealing dynamics of community abundance and biotic indices, especially regarding restoration processes in polluted streams. Chapter II, the samples were grouped into three main clusters corresponding to distinct EPTC assemblages in the tributary streams of the Garonne River catchment, southern France . Lower richness and diversity of macroinvertebrates were observed in the areas affected by agricultural land use, being associated with high Total Dissolved Solids (TDS), Nitrate (NO3) and Chemical Oxygen Demand (COD). Tolerant EPTC species were identified as controlling parameters for the changes in the assemblages collected at the agriculture-impacted sites. These findings confirmed that agricultural land use has disturbing effects on instream conditions of the riparian forest corridor that naturally serves as a buffer zone. In chapter III, the trained SOM showed the diverse scope of communities and were efficient in tracking temporal changes in community states. Biological water quality indices were correspondingly different in different clusters by the trained SOM. The SOM could be an alternative map for covering diverse scope of community states in water quality monitoring for the large-scale, long-term data, thus would broaden the scope of ecological informatics implemented to ecosystem assessment. In chapter IV, the SOM was efficient in patterning and visualizing community characteristics and accordingly classified the samples corresponding to the geographical regions and to the degree of disturbances. The patterned groups accordingly revealed the impact of pollution: the species abundance and richness accordingly reflected geographical conditions and anthropogenic disturbances, while some tolerant species were selectively collected in the disturbed areas. The communities clustered by the trained SOM were also characterized with the species-abundant patterns. The SOM was efficient in linking various information on community composition, geographic distribution, and ‘species–abundance' patterns in an integrative manner. The SOM could serve as an efficient ecological map for specifying ecoregions and for providing comprehensive view on ecological status of the communities sampled on a large scale

Keywords: Yangtze River; Gezhouba Dam; Acipenser sinensis; Artificial propagation program; Microsatellite; Genetic diversity; Artificial neural network.

Artificial or natural perturbations, space and time changing, affect continuously aquatic communities. These communities are then good indicators of environment quality and are used to control ecosystem quality ("biomonitoring"). For the assessment of river quality, species composition associated to environmental characteristics should be taking into account. However, the complexity and especially the non-linearity of the relations between the communities and their environment lead to propose advanced modelling techniques.
In the interface between modelling and ecology, this thesis aims to assess the European aquatic communities' sensitivity to the environment perturbations using complex techniques, based on a combination of classical and modern modelling techniques, as artificial neural networks.
Explorative, predictive and explanatory capacities of artificial neural networks were used to study the structure and the diversity of three aquatic communities selected as part as European project PAEQANN (diatoms, macroinvertebrates and fish). Communities' ordination and classification were done on self-organising map, non-supervised learning algorithm, to define the whole of communities. With environmental features, defined communities have been predicted using supervised learning algorithm (backpropagation). Then, the development of sensitivity analysis leads to explain ecosystems functioning determining the role of environmental features in the communities' organisation.
However, due to the communities' size and the complexity of the relationships with their environment, the prediction of their large number of species is impossible. Progressive process was proposed, using the reduction of the community to simple descriptors as the species richness, or more complex as guilds until types assemblages; or also using combinations of several methods, neuronal and classical statistics.
The developed methodologies aim to understand specific actions in the restoration and conservation of aquatic biocenose integrity, as part as the water framework directive. These significant results confirm aquatic communities modelling difficulties and the necessity of powerful modelling tools. This thesis opens interesting perspectives in modern technique emerging and their combination in ecology.

Key-words: Communities ecology; Aquatic ecosystem; Water quality; Modelling; Artificial neural networks; Fish; Macroinvertebrates; Diatoms; Water Framework Directive.

In lakes, fish populations live in distinct, contiguous or identical habitats. The relationships between fish and their environment as well as fish population interactions are complex and difficult to identify in a natural environment. However, these trends constitute the basis of the community structure.

The study of fish spatial distribution, habitat and community structure of lake Pareloup required the development and use of advanced modelling tools such as generalized additive models and artificial neural networks.

We first made a global scale (i.e. whole lake scale) study of the fish spatial distribution using hydroacoustic techniques. Results showed a pronounced affinity of the fish for the shallow littoral areas. Moreover, gillnet catches performed in the littoral zone revealed a heterogeneous spatio-temporal distribution of fish in this area.

On a smaller spatial scale, electrofishing point abundance sampling allowed us to precisely define the microhabitat of young of the year perch (Perca fluviatilis L.) and roach (Rutilus rutilus L.) and showed the influence of some environmental variables on these populations. On the basis of these results, we extended our study to the whole fish community. Even though the spatial occupancy of the fish populations was highly influenced by the physical environment, interpopulation interactions (i.e. predation and competition) played an important role in the community structure. Moreover, young of the year fish habitat use and community structure was shown to vary with time and was explained using artificial neural networks.

Finally, the anthropic impact of hydropeaking operations on two cyprinids was investigated. We first studied the effect on roach growth of a major disturbance (i.e. total draining of the lake) on a long time scale (i.e. 10 years). Then, we studied the impact of the autumn water level decrease (i.e. annual disturbance) on young of the year rudd (Scardinius erythrophtahlmus L.) habitat and survival. These results showed that draining rejuvenated the lake ecosystem, whereas high mortality occurred among young of the year rudd with the autumn water level decrease, due to the disappearance of their natural habitat.