Confronting models with data: carbon-phosphorus interaction under elevated CO2 in a mature forest ecosystem (EucFACE)

The importance of phosphorus (P) in plant function and ecosystem biogeochemistry has led to the addition of a P cycle to a range of vegetation models, but the predictions of these P-enabled models have rarely been evaluated with ecosystem-scale data. Here, we confronted eight state-of-the-art, P-enabled models with data from EucFACE, a P-limited Eucalyptus forest subject to long-term Free-Air CO₂ Enrichment. We evaluated the capability of the models to capture the observed elevated CO₂ responses in this ecosystem. We show that the inclusion of phosphorus-cycle is necessary to more realistically simulate ecosystem function and biogeochemistry, but this enhanced capacity did not directly translate into improved prediction accuracy. Specifically, models diverged in capturing the observed CO₂ responses, with simulation accuracy depending upon model assumptions about plant physiology, allocation, plant-soil interactions and soil nutrient processes. Confronting models with experimental responses observed at EucFACE represents a valuable opportunity to improve our understanding of the carbon-phosphorus interaction under rising CO₂, and is an important step towards more accurate predictions of the future land carbon sink under climate change.