COS 40-8
Confronting models with measurements: An optimal allocation model accurately predicts empirical measurements of the rhizosphere marketplace for nitrogen and phosphorus

Tuesday, August 11, 2015: 4:00 PM
303, Baltimore Convention Center
Edward R. Brzostek, Biology, West Virginia University, Morgantown, WV
Joshua B. Fisher, NASA Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA
Richard P. Phillips, Biology, Indiana University, Bloomington, IN

Accurate projections of the future land carbon (C) sink by terrestrial biosphere models depend on how they represent the impacts of nutrients on the partitioning of photosynthate-C between above and belowground pools. While empirical research has highlighted the rhizosphere as a hotspot for the trading of photosynthate for soil nitrogen (N) and phosphorus (P) between roots, mycorrhizal fungi, and free-living microbes, this C allocation pathway is missing in the models. P dynamics are also largely absent or poorly parameterized in models despite the ability of P to limit or co-limit net primary production (NPP) in many ecosystems.  Here, we integrated P dynamics into a cutting-edge plant N model (Fixation & Uptake of Nitrogen: FUN; Fisher et al., 2010, Brzostek et al., 2014) that includes representations of how ecto- and arbuscular mycorrhizal trees differ in the rhizosphere marketplace for nutrients. We confronted and validated FUN against empirical estimates of the C allocated belowground by temperate forests trees to mobilize N and P.


Across 45 temperate forest plots that vary in the distribution of arbuscular mycorrhizal (AM) and ectomycorrhizal (ECM) associated trees centered on the Morgan-Monroe State Forest AmeriFlux site, we found that FUN predicted much of the variability in the empirical estimates of belowground C allocation.  The N-only version captured the empirical patterns of greater root exudation (R2=0.53) and higher carbon costs of N acquisition (R2=0.77, i.e., C expended per unit N gained) that occur as the dominance of ectomycorrhizal trees increases.  However, there was a bias in the N-only model predictions where costs were under- and overestimated in ECM and AM dominated plots, respectively.  This was solved by integrating the interactions between the C costs of N and P uptake, which eliminated this bias and resulted in ~6% of NPP on average being allocated belowground to gain N and P.  Given that terrestrial biosphere and soil C models are now examining how plant C inputs — often assuming that inputs are a constant fraction of NPP — influence soil decomposition, FUN provides an important, robust bridge between terrestrial biosphere models and microbial-soil C models by dynamically predicting the C allocated by plants to the rhizosphere marketplace.