COS 70-8
Decomposing leaf mass into photosynthetic and structural components explains divergent patterns of trait variation within vs. among plant communities

Wednesday, August 12, 2015: 10:30 AM
339, Baltimore Convention Center
Masatoshi Katabuchi, Department of Biology, University of Florida, Gainesville, FL
Kaoru Kitajima, Graduate School of Agriculture, Kyoto University, Kyoto, Japan
Joseph S. Wright, Smithsonian Tropical Research Institute, Panama
Jeanne L. D. Osnas, Department of Forestry and Natural Resources, Purdue University, West Lafayette, IN
Jeremy W. Lichstein, Department of Biology, University of Florida, Gainesville, FL

Leaf mass per area (LMA) is thought to be a key trait underlying the leaf economic spectrum. Among communities, the relationship between LMA and leaf lifespan (LL) is strong, whereas relationships between LMA and area-normalized rates of photosynthesis (Amax/area) and respiration (Rdark/area) are weak. Within communities, however, LMA shows strong correlations with Amax/area and Rdark/area, whereas LMA shows weak correlations with LL. A simple hypothesis that may explain these seemingly contradictory patterns is that (1) photosynthetic leaf mass mainly affects Amax and Rdark, whereas structural leaf mass mainly affects LL; but (2) these relationships are obscured by analyzing LMA alone. Thus, statistically decomposing LMA into photosynthetic LMA (LMAp) and structural LMA (LMAs) should provide improved mechanistic understanding of the leaf economic spectrum, and provide insights that can guide the representation of trait variation in global carbon-cycle models. We used hierarchical Bayesian methods to decompose LMA into LMAp and LMAs using leaf trait data from tropical forest sites in Panama and the global GLOPNET database.


Models that decomposed LMA into LMAp and LMAs predicted LL, Amax and Rdark better than a standard LMA-based model for both the Panama and GLOPNET datasets. Evergreen species had greater LMAs than deciduous species, and canopy leaves had greater LMAp than understory leaves. LMAs accounted for most of the LMA variation in GLOPNET, whereas LMAp accounted for most of the LMA variation in the Panama data. These results are consistent with the strong correlation between LMA and LL in GLOPNET, and the strong correlation between LMA and both Amax and Rdark in Panama. LMAp was positively correlated with nitrogen and phosphorus per unit leaf area. Our analysis demonstrates that decomposing LMA into two axes of trait variation (i.e., photosynthetic and structural tissue mass) can explain disparate leaf trait relationships that cannot be explained by LMA alone. A limitation of our model is that LMAp and LMAs have not been directly measured, and in reality some leaf tissue mass likely contributes to multiple functions. Nevertheless, our model explains a variety of patterns that are not explained by the one-dimensional leaf economic spectrum, and may provide a useful framework for representing leaf functional variation in carbon-cycle models.