Thursday, August 5, 2010 - 4:00 PM

OOS 44-8: Using historic databases to assess forest plant species range elevational shifts

Jonathan Lenoir1, Jean-Claude Gégout2, and Jens-Christian Svenning1. (1) Ĺrhus University, (2) AgroParisTech, UMR1092 AgroParisTech-INRA

Background/Question/Methods

Have species shifted their ranges in pace with rapidly changing climate between a relatively cooler period (1905-1986) and a relatively warmer period (1986-2005)? Here, we aimed to assess changes in plant species elevational ranges in temperate forests, a currently understudied ecosystem. We focused on shifts in the mid-range elevational optimum, as the mid-range optimum can be estimated with greater certainty than range boundaries. Our study spanned the entire forested elevational gradient in six Western European mountain ranges. From two large floristic databases (~28,000 surveys), we extracted two well-balanced datasets, including 3,991 surveys each. One dataset represented the period (1905-1985) before mean annual temperature started to consistently increase in the mid-1980s, while the other one represent the post-1985 warming period (1986-2005). We used both generalized linear models (GLM) and generalized additive models (GAM) to estimate changes in species mid-range elevation optimum between the two periods. By using a bootstrap procedure, we additionally simulated random shifts to estimate the proportion of species expected to shift their mid-range elevation optimum by chance alone (random population fluctuations, data idiosyncrasies, and observer errors).

Results/Conclusions

There was an upslope shift in plant species mid-range elevation optimum averaging ~65 m for 171 forest plant species between 1905-1985 and 1986-2005. Although clearly indicative of a climate change-driven impact, this elevation shift was only half that expected from the observed temperature increase (0.7-0.9 °C, ~120-150 m). We also found that ~30% of the studied species shifted their mid-range elevation downslope, twice as many as expected by purely stochastic processes. Overall, we conclude that both the smaller-than-expected shift and the occurrence of non-random downslope range shifts should be taken more explicitly into consideration when making predictions of the effects of future climate change scenarios on species distributions. Finally, we propose a conceptual and testable model that explains the observed range shifts of species as resulting from the effects that climate and land-use change, separately or in concert, might have on species interactions and distributions.