Populations growing at the warm margins of the species' range are more prone to experience higher water stress compared to populations inhabiting the core of their distribution. Thus, assessing tree vulnerability to drought is crucial to improve prediction of forest mortality and species range limits. We quantified the abundance of two oak species (Quercus robur and Quercus ilex) along a water stress gradient in a coastal forest located at the southern edge of the distribution of Q. robur. We assessed their ecophysiological responses to drought during a wet and a dry year and determined their vulnerability to drought under field conditions. The abundance of Q. ilex was high all along the water stress gradient, whereas the abundance of Q. robur dramatically declined with decreasing water availability. During dry years, the level of native embolism was significantly higher for Q. robur than for Q. ilex due to species differences in vulnerability to xylem cavitation. Q. robur had a narrower hydraulic safety margin than Q. ilex and operated very close to the species threshold of hydraulic failure, making it highly vulnerable to drought-induced mortality. In the current context of increasing drought frequency and severity, survival of Q. robur populations will be threatened at warm range margins.

A major unknown in the context of current climate change is the extent to which populations of slowly migrating species, such as trees, will track shifting climates. Niche modelling generally predicts substantial northward shifts of suitable habitats. There is therefore an urgent need for field-based forest observations to corroborate these extensive model simulations. We used forest inventory data providing presence/absence information from just over a century (1880–2010) for a Mediterranean species (Quercus ilex) in forests located at the northern edge of its distribution. The main goals of the study were (i) to investigate whether this species has actually spread into new areas during the Anthropocene period and (ii) to provide a direct estimation of tree migration rate. We show that Q. ilex has colonised substantial new areas over the last century. However, the maximum rate of colonisation by this species (22 to 57 m/year) was much slower than predicted by the models and necessary to follow changes in habitat suitability since 1880. Our results suggest that the rates of tree dispersion and establishment may also be too low to track shifts in bioclimatic envelopes in the future. The inclusion of contemporary, rather than historical, migration rates into models should improve our understanding of the response of species to climate change.

No abstract