SUMMARY In this thesis, the populations of brown trout (Salmo trutta) inhabiting the region of Valencia were studied and compared, as well as the characteristics of the habitats and the relations habitat-abundance, from different perspectives and with different statistical techniques. The study was done in the remaining populations of brown trout in the region, in rivers and streams with Mediterranean conditions and relevant habitat variability. The study was carried out in the 4 rivers where native brown populations exist, Ebron and Vallanca (Valencia province), Palancia and Villahermosa (Castellon province). The surveys took place during the summer, from the year 2003 to 2006. The habitat characterization followed a field protocol adapted from the Basinwide Visual Estimation Technique, BVET, of common application in some areas of the United States of America by the U.S. Forest Service. In this case, 4 basic types of mesohabitats or hydromorphological units were identified, i.e. pool, glide, riffle and rapid. Every mesohabitat in the study sites (300 m long, around the units of electro-fishing) was identified and characterized with these variables: length, mean width, mean depth, maximum depth, and depth of the tail crest in the pools. An index of cover designed in Spain was estimated, as well as the woody debris (number and size by classes). The percentages of substrate types were visually estimated, based on a simplified version of the classes of the American Geophysical Union. Other variables were derived in the laboratory, such as the volume of water in the mesohabitat and relative distance between riffles. Also mean velocity was estimated for the analyses of habitat by fuzzy logic. The total number of mesohabitat units sampled (physical properties) in the 4 years was of 964. The trout populations were sampled by electro-fishing by mesohabitats. The field method was the removal without replacement, with 3 passes, and the assessment was based on the method of Carle and Strub. The total number of trouts captured in the 4 years was of 1188. The habitat characterization allowed the analyses with robust statistical techniques. The classification of mesohabitats in the four classes was evaluated, based on the variability within and among classes. The variables more relevant for the habitat variability were identified. The results indicated that the rivers Villahermosa and Ebron were more stable and homogeneous than Vallanca and Palancia. The riffles and glides were more homogeneous than pools and rapids. The length, weight, growth, density, biomass and production of the populations were evaluated and compared with other trout rivers where relevant monitoring were published. The calculations were based on robust statistical methods. The growth in length was large, and there was a notable variability in density and biomass among rivers. The annual production was low and the turnover ratio was very low, in comparison with other trout rivers in Europe and North America. Finally, the length-weight relation (Ricker equation) and the growth curves following Von Bertalanffy model were calculated. A density-dependent relation was found between the fork length and small trout density in two rivers, Vallanca and Palancia. In order to relate the habitat characteristics and the fish abundance, and to detect density-independent relations, a correspondence analysis was performed with Spearman ranks, including all the habitat variables and fish density. The results indicated that all the variables (except fine substrate) were relevant for fish density in some of the rivers (except the river Palancia), and the critical variables varied with the rivers. The results also indicated the high variability among study sites. The relations habitat-abundance were also analyzed separately for inmatures (< 2 years) and adults (?2 years), and separately for slow (pool/glide) and fast water (riffle/rapid) mesohabitats. With a multivariate approach, models of Generalized Additive Models (GAM) were calculated, and several relations were demonstrated. The density of inmatures trout was negatively related with the mesohabitat volume, both in slow and fast water habitats. The density of adult brown trout, in slow habitats, was positively related with maximum depth and the distance between riffles; in fast water habitats, the density increased with the cover index. Finally, also fuzzy logic models of fish density were applied with mesohabitat variables, in comparison with random forests. There were no important differences in performance between the 2 approaches, both demonstrating the relevance of mean width, cover index and mean water velocity (at mesohabitat scale) to interpret the trout density. The fuzzy models combined the ecological relevance of the results with the model transparency, and they are easier to interpret for managers, therefore they mean a good tool to help in decision making about river restoration and management, for the conservation of the Mediterranean brown trout in the Iberian Peninsula.