Alpine tundra species phenology is mostly driven by climate-related variables rather than by photoperiod

The study of plant phenology has frequently been used to link phenological events to various factors, such as temperature or photoperiod. In the high-alpine environment, proper timing of the phenological cycle has always been crucial to overcome harsh conditions and potential extreme events (i.e. spring frosts) but little is known about the response dynamics of the vegetation, which could shape the alpine landscape in a future of changing climate. Alpine tundra vegetation is composed by an array of species belonging to different phytosociological optima and with various survival strategies, and snowbed communities are a relevant expression of such an extreme-climate adapted flora. We set eight permanent plots with each one in a snowbed located on the Cimalegna plateau in Northwestern Italy and then we selected 10 most recurring species among our plots, all typical of the alpine tundra environment and classified in 3 different pools: snowbed specialists, grassland species and rocky debris species. For 3 years we registered the phenophases of each species during the whole growing season using an adaptation of the BBCH scale. We later focused on the three most biologically relevant phenophases, i.e., flower buds visible, full flowering, and beginning of seed dispersion. Three important season-related variables were chosen to investigate their relationship with the phenological cycle of the studied species: (i) the Day Of Year (DOY), the progressive number of days starting from the 1st of January, used as a proxy of photoperiod, (ii) Days From Snow Melt (DFSM), selected to include the relevance of the snow dynamics, and (iii) Growing Degree Days (GDD), computed as a thermal sum. Our analysis highlighted that phenological development correlated better with DFSM and GDD than with DOY. Indeed, models showed that DOY was always a worse predictor since it failed to overcome interannual variations, while DFSM and marginally GDD were better suited to predict the phenological development of most of the species, despite differences in temperature and snowmelt date among the three years. Even if the response pattern to the three variables was mainly consistent for all the species, the timing of their phenological response was different. Indeed, species such as Salix herbacea and Ranunculus glacialis were always earlier in the achievement of the phenophases, while Agrostis rupestris and Euphrasia minima developed later and the remaining species showed an intermediate behavior. However, we did not detect significant differences among the three functional pools of species.