Since the 1950s, the population of the Brown hare (Lepus europaeus) has dramatically decreased in Switzerland as well as in other European countries. According to the Swiss Ornithological Institute of Sempach, this animal is an adequate indicator of the state of the landscape. Its decrease is probably due to intensification of agricultural practices, extension of urbanism and development of transport networks (Pfister et al., 2002). Today, the Brown hare is on the Swiss Red list of species that must be protected. This thesis presents an analysis of the landscape dynamics of the North-Western part of Switzerland, from the 1950s to the 1990s, through an historical analysis of the state of the landscape as it is perceived by the hare. In particular, we identify the factors of the landscape that are relevant for this animal, determine their trend of evolution, and finally postulate hypotheses for the evolution over the next years and their consequences for the hare populations. For each decade, 936 communes are described, using 40 variables related to agricultural practices, land-use, morphology, urbanisation, climate and geography. This database is called "Landscape". These data, mainly extracted from communal statistics, were found to be adequate to describe the evolution of the composition of the landscape at a large scale. We studied the content, the quality and the organisation of this database "Landscape" with the help of descriptive statistical methods (data distribution, histograms and boxplots at each decade, dynamics of the statistical characteristic values) and using principal components analysis (PCA). The latter reveals two groups of communes: urban (belonging to an agglomeration) and rural. On average, it also shows a very definite evolution of the urban character of both groups, from the 1950s to the 1990s. The three first components of the PCA explain only 45% of the variance. Therefore, we decided not to simplify the database by replacing our variables by these agglomerated variables, as the loss of information would be important. Among the 40 variables, 16 key-variables for the hare were identified. For this, an "Index of abundance of hares" was calculated for a sub-set of 125 communes, defined as the mean number of hares hunted per total communal area between 1980 and 1985. The link between this index and the 40 variables, describing the state of 125 communes in the 1980s, was analysed with the use of correlation, linear multiple regression, the method Pegase (a multivariate hierarchical divisive procedure, based on information theory (Phipps, 1981)), and additional information from the Swiss Ornithological Institute of Sempach. These methods were found to be complementary and their use enables us to take into account several perspectives. It would be interesting to go further by calculating this index for a larger number of communes in order to optimise the use of Pegase. The main trends of evolution of these key-variables, observ
