Il est désormais avéré que les pratiques intensives de production agricole mises en œuvre depuis les années 1960 en Europe, sont responsables de la dégradation des écosystèmes et notamment de la qualité des eaux de surface et souterraines. Afin de respecter les objectifs de résultat définis par la Directive Cadre sur l’Eau à l’échéance 2015, les gestionnaires de la ressource ont désormais ciblés prioritairement la protection des zones d’alimentation des captages. Cela nécessite une évaluation environnementale à une échelle adaptée aux mesures mises en œuvre. Ces éléments expliquent les objectifs de cette thèse, à savoir la contribution au développement de méthodes d’évaluation environnementale des activités agricoles à l’échelle des territoires, en complément de celles déjà existantes. Nous avons ainsi utilisé (i) des méthodes d’analyse et de modélisation multicritère (ELECTRE) combinées avec un système d’information géographique au niveau de petits bassins versants élémentaires de 2 à 5 km2 (sur le bassin de l’Oir, en Basse Normandie, Nord-ouest de la France et celui d’Auradé sur les Coteaux de Gascogne, Sud-ouest de la France) (ii) une approche par modélisation spatiale (dénommée PIXAL) des risques agro-environnementaux en utilisant la télédétection et un SIG, au niveau d'un grand espace à enjeu environnemental de plusieurs milliers de km2, (Coteaux de Gascogne). Dans les deux approches, nous avons montré que la combinaison de critères de vulnérabilité des eaux et de pression agricole choisis de façon pertinente, permettent d’évaluer les risques agro-environnementaux à différents niveaux d’organisation spatiale. L'évaluation à la parcelle agricole (niveau de la prise de décision par les acteurs de terrain) procure une représentation au niveau du bassin versant élémentaire. L'évaluation au pixel de l’image satellitale (Landsat 5-TM) permet une représentation à différents niveaux de bassins versants emboîtés par agrégation. En perspective, un couplage entre un modèle d'analyse multicritère avec un modèle spatial, pourrait permettre d’améliorer la performance de l’évaluation aux différents niveaux de l’action publique ou de terrain et d’apporter des éléments de réponse supplémentaires à la problématique du changement d’échelle spatiale. / There is evidence that intensive farming practices applied in Europe since the 1960s are responsible for the degradation of certain ecosystems, and a reduction in the quality of surface and ground water. Excessive use of nitrogen-based fertilisers and pesticides have led to high concentrations of these substances being found in rivers, thus compromising raw drinking water sources. Soil erosion – caused by simplified crop rotations, feeding livestock on corn instead of grass, and the removal of elements such as embankments, trees, hedges, etc. – has been conducive to the transfer of these particles into surface water. The resulting turbidity has led to the clogging up of spawning grounds, with many negative effects on their biological quality. Complementing existing approaches, this thesis will focus on the development of methods for carrying out territorial environmental analysis of farming activities. For our study, we used (i) a multicriteria analysis and modelling (ELECTRE) combined with a Geographic Information System (GIS) on a selection of small elementary watersheds (ranging in size from 2 km² to 5 km²). These were located in the Oir basin, in Lower Normandy Region, in northwest France, and the Auradé basin, in the hills of Gascony region, southwest. We also employed (ii) a spatial modelling approach (PIXAL) to represent agri-environmental risks. Applied to an environmentally-sensitive area (hills of Gascony region) of several thousand square kilometres, the PIXAL method involved evaluating agri-environmental risks through a combination of remote sensing and GIS. In both cases, we showed that choosing the most relevant criteria allows agri-environmental risks to be assessed at different organisational scales. Analysis at farm plot level (where decisions are made by stakeholders in the field) gives a representation of the elementary watershed. Pixel-based evaluation, using satellite imaging (Landsat TM5), provides a nested multi-aggregation view of different watersheds. Placing this study in perspective, coupling ELECTRE-style multicriteria analysis with a spatial model could be an effective way to improve analysis at different policy-making and territorial scales, as well as providing additional support where assessment needs to switch between those varying scales.

The objectives of the study were to analyse rural youth poverty alleviation in the Zambezi Region, problems, and possibilities in order to determine how rural youth agricultural and non-agricultural livelihood activities have improved their livelihood and alleviated poverty. The Sustainable Livelihoods Framework has helped the researcher to understand and explain rural youth poverty alleviation, problems, and possibilities in the study area. The study employed a mixed-method approach, with more emphasis on qualitative research. The study had a total size sample of 223, respondents, including focus group discussions. The findings indicate that no livelihood analysis was carried out for rural youth in agriculture and non-agricultural activities in the Zambezi Region. For this reason, agricultural and non-agricultural programmes and projects have not addressed rural youth poverty alleviation. The findings also show that rural youth in agriculture and non-agricultural activities are exposed to numerous factors affecting their livelihood, as most of them cannot efficiently affect the vulnerability context. Furthermore, rural youth depends on their families and friends for livelihood support since they lack access to most livelihood assets. Despite livelihood improvement, rural youth faces problems that impede poverty alleviation in both agriculture and non-agricultural livelihood activities; lack of participation in policy deliberations, access to markets, Information Communication Technologies, training in both agriculture and non-agricultural activities, limited access to financial services and limited access to land. The study demonstrates that successful agricultural and non-agricultural programmes need to take rural youth sincerely to influence rural youth poverty alleviation. It is hoped that this study will contribute to the limited literature on rural youth poverty alleviation and also provide a rural youth critique in agriculture and non-agricultural activities in Namibia. / Development Studies / D. Phil. (Development Studies)