OOS 87-8
Incorporating plant functional diversity into Earth system models: Plant carbon allocation strategies in light- and water-limited ecosystems

Friday, August 14, 2015: 10:30 AM
327, Baltimore Convention Center
Jeremy W. Lichstein, Department of Biology, University of Florida, Gainesville, FL
Tao Zhang, Department of Biology, University of Florida, Gainesville, FL
Ensheng Weng, Princeton University, Princeton, NJ
Caroline E. Farrior, Ecology and Evolutionary Biology, Princeton University, Princeton, NJ
Sergey Malyshev, Princeton University
Elena Shevliakova, Ecology and Evolutionary Biology, Princeton University, Princeton, NJ
Ray Dybzinski, Princeton University, Princeton, NJ
Richard Birdsey, Forest Service, Newtown Square, PA
Stephen W. Pacala, Ecology and Evolutionary Biology, Princeton University, Princeton, NJ

The response of the terrestrial carbon (C) cycle to climate change is a key uncertainty in Earth system models (ESMs). An important component of this uncertainty concerns plant functional diversity, which is typically represented in ESMs by ~10 discrete plant functional types (PFTs). The PFT framework is widely viewed to be inadequate, and several alternatives have been proposed. However, few global vegetation models, and none of the ESMs included in the last IPCC report, include the individual-level competitive mechanisms (e.g., height-structured competition for light) needed to represent plant functional diversity in ESMs in a non-arbitrary manner. We have developed a new land model, designed for ESM coupling, based on a height-structured forest dynamics model, the perfect plasticity approximation (PPA). The new land model, LM3-PPA, allows for an arbitrary number of “species” or PFTs whose spatial-temporal distributions are determined by the outcome of resource competition. We performed modeling experiments across precipitation gradients to evaluate how competitively optimal (fitness-maximizing) plant C allocation (to leaves, wood, and fine-roots) affects the forest C and water cycles relative to biomass- or NPP-maximizing allocation strategies that are implemented in some land models. We also evaluated alternative assumptions about how plant water deficit affects tree mortality rates.


Under moist conditions, where plants rarely experienced water deficit, competitively-optimal and biomass-maximizing trees had similar allocation (and therefore similar C and water cycles). However, under dry condition (frequent water deficit), competitively optimal trees allocated more C to both fine roots and leaves, and less C to wood, compared to biomass-maximizing trees. Increasing the sensitivity of tree mortality to water deficit led to greater differences between competitively-optimal and biomass-maximizing strategies, but did not qualitatively change the results. Increased leaf allocation by competitively optimal trees was explained by their increased fine root density, which led to greater plant water availability, which in turn shifted the benefit:cost ratio of additional leaf layers. The change in the hydrological cycle due to competitive C allocation led to increased evapotranspiration (decreased evaporation more than compensated for by increased transpiration). NPP-maximizing plants tended to have similar leaf allocation but greater fine-root allocation (and thus less wood allocation and C storage) than competitively optimal plants. Our results demonstrate significant C- and water-cycle consequences of incorporating ecological principles (e.g., resource competition) into ESMs. The non-competitive optimization algorithms used in some land models likely contribute to model errors and uncertainty.