Background/Question/Methods  A mechanistic-based representation of tree growth and mortality is essential for predicting the effects of environmental perturbations on forest ecosystems. Thus, we are developing an approach for predicting forest dynamics based on species-specific, mechanistic information. The framework includes an individual-based model (IBM) of tree growth and mortality that incorporates species-specific traits related to physiological, anatomical, morphological, and allometric properties. We used the IBM, rigorously informed by trait data obtained from the literature, to address the questions: (1) How variable (within and across species) are key functional traits? And, (2) what are the implications of trait variability for tree growth and mortality, and forest diversity and productivity? We focus on two important traits out of the 15+ traits (parameters) in the IBM: specific leaf area (SLA) and wood density (WD). To address (1), we conducted a Bayesian meta-analysis of over 1500 records each of SLA and WD to obtain species-specific trait estimates and to evaluate within and between species trait variability. To address (2), we input the variable trait estimates into the IBM and simulated tree growth and mortality for a subset of US species under different light regimes to evaluate the implications of trait variability.

Results/Conclusions  The meta-analyses revealed that SLA and WD are highly variable within a species due to site effects (SLA), light (SLA), precipitation (WD), tree age (SLA, WD), and size (WD). SLA exhibited high plasticity such that it is 3.2x higher for shaded seedlings compared to adults in the sun; WD was only 1.1x higher for trees in low (300 mm) versus high (2000 mm) precipitation sites. A taxonomic signal emerged such that standardized SLA and WD differed by a factor of 5.4 and 7.6, respectively, across 305 species. We input the variable SLA and WD estimates into the IBM such that each tree's traits change over time with changing light, tree age, and size (variable trait (VT) model). We also implemented the “standard approach” and assumed constant trait (CT) values for each tree based on nominal species values. The VT and CT approaches produced dramatically different results for growth rate and lifespan under different light regimes; variable SLA allowed increased growth and survival compared to the CT model. This highlights the importance of incorporating trait variability into plant growth and mortality models. We are extending the IBM to many species to evaluate the role of functional traits on forest dynamics.