Including the phosphorus cycle into the LPJ-GUESS dynamic global vegetation model (v4.1, r10994) – global patterns and temporal trends of N and P primary production limitation

Abstract. Phosphorus (P) is a critical macronutrient for plant growth, often limiting plant production in areas where plant demand is higher than soil supply. In contrast to nitrogen (N), P cannot be sourced from the atmosphere; therefore, where it is rare, it becomes a strong constraint on primary production. Due to this, most dynamic global vegetation models (DGVMs) are incorporating a prognostic P cycle in addition to N, improving their ability to correctly predict stocks and fluxes of carbon and how climate change may impact N and/or P limitations to soil processes and plant productivity. We included the P cycle into an individual-based DGVM, Lund–Potsdam–Jena General Ecosystem Simulator (LPJ-GUESS, v4.1, r10994), in order to improve model performance with regard to observations of vegetation and soil N and P stocks and fluxes in comparison to the N-only (LPJ-GUESS-CN) model version. The new model version (LPJ-GUESS-CNP v1.0) includes soil organic P dynamics, P limitation of organic matter decomposition, P deposition, temperature- and humidity-dependent P weathering, plant P demand and uptake, and P limitations to photosynthesis. Using the CNP version of LPJ-GUESS, we also estimated global spatial patterns of nutrient limitation to plant growth as well as the temporal change in plant N and P limitation during the 20th and early 21st century, evaluating the causes for these temporal shifts. We show that including the P cycle significantly reduces simulated global vegetation and soil C and N stocks and fluxes, in particular in tropical regions. The CNP model simulation improves the fit to global biomass observations in relation to the CN simulation. The CNP model predicts predominant P limitation of plant growth in the tropics, and N limitation in the temperate, boreal, and high-altitude tropical regions. The CNP model also correctly predicted the global magnitude (∼50 PgP) and the spatial pattern of total organic P stocks. P-limited regions cover less land surface area (46 %) than N-limited ones but are responsible for 57 % of the global gross primary productivity (GPP) and 68 % of vegetation biomass, while N-limited regions store a larger portion of total carbon stocks (55.9 %). Finally, the model shows that globally, primary production limitation to N availability decreased and limitation to P increased from 1901 to 2018, with N being more responsive to temperature and P than CO2 changes. We conclude that including the P cycle in models like LPJ-GUESS is crucial for understanding global-scale spatial and temporal patterns in nutrient limitation and improving the simulated carbon stocks and fluxes.