Assessing intraspecific trait variation and covariation across the temperate-boreal ecotone of the northeastern United States

Background/Question/Methods

Trait-based analyses provide powerful new tools for developing a generalizable, mechanistic understanding of how forest communities are responding to ongoing environmental changes. Tree functional traits vary in response to climatic and other environmental changes in space and time, yet trait-based approaches are often represented with a single mean value, under the assumption that trait variation within species is negligible. Mean trait values are particularly problematic when changes in traits of dominant tree species shift along environmental gradients, causing species trait crossover and thus changes in species competitive hierarchy that drives community structure and function. Consequently, a better understanding of intraspecific trait variability along climatic and other environmental gradients is necessary. We collected data on nine leaf and stem traits for 13 tree species along light and temperature gradients in montane forests spanning the ecotone between the temperate deciduous and high elevation coniferous forests of New England and northern New York. We assessed 1) how trait variation is partitioned within and across taxa using a variance partitioning analysis, and 2) how traits vary across the two studied environmental gradients, using linear mixed effects models and a non-metric multidimensional scaling analysis.

Results/Conclusions

We found that levels of intraspecific trait variability differed considerably between traits but with marginal evidence of species crossover, while interspecific trait variability remained dominant and was largely driven by differences between angiosperms and gymnosperms. Traits were also significantly influenced by their light environment, regional climate, and developmental stage. Interestingly, and in contrast to previous studies, we found that traits covariance in our study resembled global level trends, suggesting that trait dimensions such as the Leaf Economic Spectrum persist at local scales. Our results suggest that ecological strategies of tree species assessed in this study may be quantified with relatively few traits due to the persistence of trait dimensions at fine spatial scales, and that representing species with a single trait value may be sufficient for trait-based analyses of tree communities.