Numerous studies have highlighted plant species shift to further north and higher altitude due to present climate warming. Using indirect factors (such as altitude and latitude) does not allow to affirm the observed effect of climatic change. Moreover, most of the studies used an individual species approach preventing us to quantify and survey the plant communities’ response. We compared here the dynamic equilibrium between French forest plant communities (based on more than 20000 floristic surveys) and mean annual temperatures (MATs) between the period 1965-1987 and the warmer period 1988-2000 (+1°C in average).

We computed a predictive model of MAT from the understory floristic composition of sites (assumed the most reactive plants to climate change). The model combines weighted averaging partial least squares and Breiman’s random forest regressions to take into account species dependence and species-temperature no-linearity relationship. It was calibrated from a dataset of 704 plant species and 2754 floristic surveys before the recent climate warming (1975 to 1985). MATs were then predicted with plant communities for 9589 plots carried out between 1965 and 2000. Predicted temperatures were compared with MATs extracted from a spatialized model (at 1 km resolution).

Results/Conclusions

Our model showed a high goodness of fit (from 1796 validation plots, R²=0.816, RSE=0.85). Floristic predictions exhibited annual response in temperature respect to the climatic variation. Interannual variations in climate reconstruction were lower than temperature trend (mean interannual temperature variations=1.12°C and 1.33°C, respectively). From 1965 to 1987 (4774 observations), we observed low differences between floristic predictions and temperature observations (mean difference=0.06°C, 95% confidence interval (IC_95%)=0.03 to 0.1°C). Since 1988 (4815 observations), mean temperatures predicted from forest plant communities were lower than the temperature observations (mean difference=-0.95°C, IC_95%=-0.98 to -0.92°C). The difference was stronger below 500 m of elevation (mean difference=-1.2, IC_95%= -1.25 to -1.17°C) as compared to higher altitude (mean difference=-0.6, IC_95%=-0.65 to -0.55°C).

Rapid climate warming broke the equilibrium between temperate forest plant communities and climate. In response, species lost and/or turnover occurred to partially recover plant community-temperature equilibrium. Stronger plant communities’ response in highland was probably explained by high species sensitivity to climate change, and short distance migration along the altitudinal gradient to reach their suitable conditions. In contrast, present inertia of lowland forest plant species could lead to community vulnerability face to climate change.