Changes in traditional management practices (grazing and fertilisation) affect vegetation dynamics in the dairy subalpine pastures of the Northern French Alps, with consequences on the floristic composition and the pastoral value of these grasslands. Based on plant functional traits, our research deals with the mechanisms driving the dynamics of vegetation under changing pastoral use. We studied five vegetation types representative of different intensities of pastoral use leading to different levels of fertility. We experimentally demonstrated that the importance of competition, as a structuring factor of plant communities, decreases with decreasing fertility. However, the importance of competition for a given species is affected by the individual response of this species to competition and abiotic stress. Thus, for the same level of resource availability, the importance of competition can be different between two species. Besides, competition tends to be more intense for species with trait values very different from the average trait values of their resident community. We also assessed the functional composition of the five types of vegetation in order to investigate the relationships between plant traits and pastoral use. Decreasing pastoral use selects plants with more conservative strategies, characterized by a smaller plant height, a higher tissue density, a longer life span, a slower growth rate and the production of a more recalcitrant litter than plants exploitative strategies. The replacement of exploitative species by more conservative species is gradual along the gradient. It is known to have consequences on ecosystem functioning by reducing vegetation palatability and reinforcing soil acidification. Quantifying this effect will be the object of future research. Vegetation typologies are commonly used to characterise grasslands in relation to their ecological or agronomic value. We propose here a non-taxonomic criteria, functional composition, as an alternative to floristic criteria for the classification of vegetation. It proved as efficient as the taxonomic criteria to characterise and differentiate the different types of subalpine pastures. The classification was also robust, even when dominant species only were sampled or when differences occurred in the local floristic composition. Finally, we tested a method, adapted from a linear vegetation point transect protocol, to assess the functional composition of grasslands: the trait transect'. Quick and requiring little botanical knowledge, it should be an appropriate tool for monitoring ecosystem changes in response to environmental conditions and management.