Nakhavali, Mahdi André, Mercado, Lina M., Hartley, Iain P., Sitch, Stephen, Cunha, Fernanda V., di Ponzio, Raffaello, Lugli, Laynara F., Quesada, Carlos A., Andersen, Kelly M., Chadburn, Sarah E., Wiltshire, Andy J., Clark, Douglas B., Ribeiro, Gyovanni, Siebert, Lara, Moraes, Anna C. M., Schmeisk Rosa, Jéssica, Assis, Rafael, and Camargo, José L.
Most land surface models (LSMs), i.e. the land components of Earth system models (ESMs), include representation of nitrogen (N) limitation on ecosystem productivity. However, only a few of these models have incorporated phosphorus (P) cycling. In tropical ecosystems, this is likely to be important as N tends to be abundant, whereas the availability of rock-derived elements, such as P, can be very low. Thus, without a representation of P cycling, tropical forest response in areas such as Amazonia to rising atmospheric CO 2 conditions remain highly uncertain. In this study, we introduced P dynamics and its interactions with the N and carbon (C) cycles into the Joint UK Land Environment Simulator (JULES). The new model (JULES-CNP) includes the representation of P stocks in vegetation and soil pools, as well as key processes controlling fluxes between these pools. We develop and evaluate JULES-CNP using in situ data collected at a low-fertility site in the central Amazon, with a soil P content representative of 60 % of soils across the Amazon basin, to parameterize, calibrate, and evaluate JULES-CNP. Novel soil and plant P pool observations are used for parameterization and calibration, and the model is evaluated against C fluxes and stocks and those soil P pools not used for parameterization or calibration. We then evaluate the model at additional P-limited test sites across the Amazon and in Panama and Hawaii, showing a significant improvement over the C- and CN-only versions of the model. The model is then applied under elevated CO 2 (600 ppm) at our study site in the central Amazon to quantify the impact of P limitation on CO 2 fertilization. We compare our results against the current state-of-the-art CNP models using the same methodology that was used in the AmazonFACE model intercomparison study. The model is able to reproduce the observed plant and soil P pools and fluxes used for evaluation under ambient CO 2. We estimate P to limit net primary productivity (NPP) by 24 % under current CO 2 and by 46 % under elevated CO 2. Under elevated CO 2 , biomass in simulations accounting for CNP increase by 10 % relative to contemporary CO 2 conditions, although it is 5 % lower compared to CN- and C-only simulations. Our results highlight the potential for high P limitation and therefore lower CO 2 fertilization capacity in the Amazon rainforest with low-fertility soils. [ABSTRACT FROM AUTHOR]