Model-experiment interaction to improve representation of phosphorus limitation in land models

Abstract:

Carbon-nutrient interactions play important roles in regulating terrestrial carbon cycle responses to atmospheric and climatic change. None of the CMIP5 models has included routines to represent the phosphorus (P) cycle, although P is commonly considered to be the most limiting nutrient in highly productive, lowland tropical forests. Model simulations with the Community Land Model (CLM-CNP) show that inclusion of P coupling leads to a smaller CO₂ fertilization effect and warming-induced CO₂ release from tropical ecosystems, but there are important uncertainties in the P model, and improvements are limited by a dearth of data. Sensitivity analysis identifies the relative importance of P cycle parameters in determining P availability and P limitation, and thereby helps to define the critical measurements to make in field campaigns and manipulative experiments. To improve estimates of P supply, parameters that describe maximum amount of labile P in soil and sorption-desorption processes are necessary for modeling the amount of P available for plant uptake. Biochemical mineralization is poorly constrained in the model and will be improved through field observations that link root traits to mycorrhizal activity, phosphatase activity, and root depth distribution. Model representation of P demand by vegetation, which currently is set by fixed stoichiometry and allometric constants, requires a different set of data. Accurate carbon cycle modeling requires accurate parameterization of the photosynthetic machinery: V_c,max and J_max. Relationships between the photosynthesis parameters and foliar nutrient (N and P) content are being developed, and by including analysis of covariation with other plant traits (e.g., specific leaf area, wood density), we can provide a basis for more dynamic, trait-enabled modeling. With this strong guidance from model sensitivity and uncertainty analysis, field studies are underway in Puerto Rico and Panama to collect model-relevant data on P supply and demand functions. New FACE and soil warming experiments in P-limited ecosystems in subtropical Australia, and tropical Brazil, Puerto Rico, and Panama will provide important benchmarks for the performance of P-enabled models under future conditions.