SYMP 5-4 - How should trait based ecology deal with intraspecific trait variability?

Tuesday, August 7, 2012: 9:15 AM
Portland Blrm 252, Oregon Convention Center
Justin Wright, Biology, Duke University, Durham, NC
Background/Question/Methods

One of the primary goals of trait based ecology is to predict how community composition and ecosystem function will be change in response to perturbations such as climate change, eutrophication, or invasion and extinction events based on information about the distribution of traits within the species pool.  To date, most attempts to do so have used approaches that assume that single species can be adequately characterized by a single value of a given trait.  However, it is well known that intraspecific variability can be quite large for many traits – indeed, evolution by natural selection demands significant intraspecific variation in traits.  Determining when and how to incorporate this variability is a major challenge for trait based ecology.  Here we review evidence for the degree to which intraspecific variability exists for ecologically relevant traits and discuss three potential approaches for addressing intraspecific trait variability within the context of trait based ecology.

Results/Conclusions

Recently published papers have shown that as much as 48% of the variability in ecologically relevant traits such as leaf N, specific leaf area (SLA), and specific root length (SRL) along broad gradients is present within species, suggesting that ignoring intraspecific variability might have serious implications for trait based models.  One possible strategy is to assume that such variability is essentially ecologically irrelevant noise and to continue approaches that assume a single value of a trait for each species.  This approach is least likely to be problematic when addressing questions at the regional scale and above where species differences in trait values are likely to be much greater than intraspecific variability.  A second approach assumes that traits vary predictably along environmental gradients, so that models simply need to incorporate terms that include environmental variables.  We show that despite significant species by environment interactions (indicating species-specific trait responses to changes in N availability and water table depth), traits such as SLA, Leaf N, and photosynthetic rates (Amass) are predictable from one environment to another.  However, multivariate trait responses are significantly less predictable.  This suggests that a third approach may be necessary for some ecological questions, namely an approach that actively measures and incorporates intraspecific trait variability into models.  We demonstrate an example of how this might be accomplished, but also point out the challenges, both in developing the database necessary for such an approach and in developing the model structure to fully embrace intraspecific trait variability.