Can functional traits explain variation in plant life history strategies?

Background/Question/Methods

Recent studies have focused attention on relationships between plant functional traits and demographic vital rates (e.g. survival or growth). One of the motivations for linking traits with demographic performance is to understand the physiological basis of whole plant life history evolution. Because functional traits mediate species' interactions with the environment, they should reflect the selective pressures that species have experienced. However, correlations between functional traits and raw demographic vital rates are often weak and do not directly address questions about life history evolution. To more directly assess selective pressures, we correlated functional traits obtained from the TRY database with vital rate elasticities derived from matrix projection models contained in the COMPADRE II database. We tested the following three predictions: 1) Species with low specific leaf area (or long leaf life spans) should have high elasticity to survival and low elasticity to growth; 2) Wood density should also be positively correlated with survival elasticities; 3) Seed mass should be positively correlated with survival and negatively correlated with recruitment elasticities.

Results/Conclusions

We generated trait-elasticity records for up to 184 plant species. For all four functional traits, generalized linear regressions and quantile regressions on elasticities supported our hypotheses, though the relationships were stronger for seed mass and wood density than for specific leaf area and leaf life span. Relationships between functional traits and raw survival rates were also significant, whereas raw growth and recruitment rates were seldom related to traits. These significant, if noisy, trends emerged despite potentially large measurement errors. Our results provide strong evidence that functional traits help drive, or at least covary with, plant life history evolution. However, the high degree of unexplained variation in the trait-demography relationships may limit the utility of trait-based approaches for addressing applied problems.