Résumé
A marked increase in air temperature has been observed in France since the end of the 1980s for fruit trees and vine regardless of cropping areas. As phenology is mainly influenced by temperature, changes in vegetative and flowering phenology timing will have agronomic consequences on these perennial productions. It is thus necessary to characterize the present and future trends of bud phenology in the context of global warming. Firstly, long-term series of phenological data extracted from a national data base devoted to fruit trees and vine (called PhénoClim®) have revealed changes in the time-course variations in flowering and vegetative phenology. Surprisingly, quite similar trends toward advances of flowering dates in various fruit trees and bud break dates in vine have been observed in different French cropping areas. In apple and pear trees, a statistical study of phenological series using multiple change-point models clearly support the idea of an abrupt change in the mean dates of flowering stages since the end of the 1980s, the most likely period for this discontinuity being between 1988 and 1989. Similar results were obtained considering phenological series of bud break dates in vine. The concomitancy with the global warming recorded in France since the end of 1980s leads to conclude that these advances in bud phenology could be actual impacts of this warming. Secondly, regarding the analysis of the phenological evolution in terms of temperature effects, modelling works based on the same conception of successive chilling and heat effects were developed to analyze the past and future impacts on bud phenology. For apple tree, a user-oriented software (called Pollenoscope) was exploited to understand the recent advance of mean flowering dates. The results suggest that two opposite trends simultaneously occurred in France since the end of the 1980s: a slower rate of completion of chilling requirements and a higher rate of completion of heat requirements. Nevertheless, the observed advance of mean flowering dates could result of more pronounced increases of temperatures from January to April, corresponding to the period of heat requirements completion for buds in France. In vine, a new model (called Brin) was developed introducing a chilling sub-model contrary to many traditional vine models. The optimal parameters were defined using the PhénoClim® database and the software STICS. The prediction of bud break date was improved by the model Brin. Moreover, a simulation over a future period (2070-2099) was realized using the Brin phenological model and the Arpege climatic model on the basis of A2 and B2 scenarios of IPCC. An increase of the mean advance of bud break could be observed for vine in France, although the dormancy break by chilling would have a marked tendency to be later, as previously suggested for apple tree in a recent past.