Analysis of phosphorus stock variation in soil and biomass during an Eucalyptus rotation: a step towards modelling the phosphorus cycle

Forests play a fundamental role in supporting global biodiversity, supplying key resources such as timber, paper, and energy, and acting as one of the largest terrestrial carbon sinks. Forest productivity, however, is constrained by several environmental factors, including the availability of carbon dioxide (CO2) and essential nutrients such as nitrogen (N), phosphorus (P), and potassium (K). Over the past decades, an increasing number of terrestrial ecosystem models (TEMs) have incorporated representations of nutrient cycles, most frequently considering N (Zaehle et al. 2009, Vuichard et al. 2019) more rarely P (Goll et al. 2012, 2017, Jiang et al. 2024) and K (CASTANEA model, see Cornut et al. 2022a, b).

Our study contributes to that effort by focusing on to the quantification and modelling of phosphorus (P) cycles, based on data and model simulations from Eucalyptus plantations in Brazil. As a starting point, we studied the fluxes of P between the soil (i.e., soil organic and inorganic P stocks) and the trees (i.e., aboveground biomass and P stocks). To do so, we used data from a P fertilization experiment conducted at the Itatinga experimental station (University of Sao Paulo) with various forms of P fertilizers. Using allometric relations and concentration measurements, we quantified the mass of phosphorus in each compartment of the trees (leaves, branches, trunk wood, bark and roots) during the entire rotation and compared it to the variation of P stock in the soil, measured in different chemical forms.

Results showed that, compared to control conditions (no fertilizer added), phosphorus fertilization increased the tree biomass production, the amount of P accumulated in plant tissues, as well as increasing the soil P stocks. However, the magnitude of these effects depended on the type of fertilizer used. Complexed humic phosphate, designed to enhance phosphorus bioavailability, produced the highest tree biomass and phosphorus mineralomass. In contrast, rock phosphate was most effective at increasing total soil phosphorus stocks. This outcome aligns with previous findings, as rock phosphate is less readily absorbed by plants than more soluble forms. Accounting for the spatial heterogeneity in soil P concentrations proved essential when computing the ecosystem P. In 5 out of 6 treatments we observed apparent P losses (i.e. unclosed P balance), which may reflect underestimation of deep root biomass and P pools in deeper soil layers. By contrast, in rock phosphate treatment, apparent P gains probably stemmed from overestimations of soil P related to uncertainties in the estimation of soil P spatial variability.

Given limitations in the data, we are currently considering incorporating a simplified representation of soil P in the CASTANEA model, representing only a small number of phosphorus compartments (organic vs. mineral form and availability to trees).