A Novel Spatially Explicit Integrated Modeling Tool for Assessing P Accumulation in Agricultural Soils

Abstract

An improper fertilization strategy may lead to the accumulation of phosphorus (P) in agricultural soils in the long term, leading to elevated risks of P export by runoff, erosion and leaching to water bodies, thus contributing to their eutrophication. Furthermore, excessive application of nutrients is a waste of limited resources. Although several methods have been published for calculating P balances at different scales, there is a lack of models purposely designed to produce spatially distributed predictions of the long-term evolution of P concentration in soil at regional scale. We devised a modelling tool based on a tiered field- and regional-scale approach, integrating and harmonizing a high number of different datasets related to weather, soil, land use, and farm management. We developed a Land Management Model (LMM) that provides animal manure and mineral fertilizer application rates, based on farm census and land use statistics. Swiss census data comprises georeferenced information for every single farm, such as animal type and number, crop and corresponding area. This information, complemented by fertilisation guidelines and expert knowledge, was distributed in a spatially explicit form to the land use raster. The LMM was coupled with a field scale dynamic model (Environment Policy Impact Calculator, EPIC) which simulates several soil-plant processes including the cycling of P in distinct pools and for different soil layers. EPIC was calibrated at reference sites belonging to the Swiss Soil Monitoring Network (NABO), where soils have been sampled every five years and field management has been recorded annually since 1986. The LMM and EPIC were coupled within a Spatially Explicit Regional Integration Tool (SERIT). Uncertainties related to the spatial distribution of soil properties and crops, as well as amount and P content of animal manure, are taken into account by means of Monte Carlo simulations. The final figures are thus complemented by a measure of their uncertainty, which can be apportioned to different sources. The modelling tool was tested in a study area located in Canton Fribourg, Switzerland, for the period 1995-2009. The soil P temporal evolution was represented by maps. Results indicate a general depletion of both labile and total P fractions from the topsoil, and an accumulation of labile P in the subsoil. The spatial pattern of total P variation in the subsoil is heterogeneous and can be put in relation with the land use, indicating accumulation in arable land and depletion in grassland. Keywords: phosphorus, modelling, spatial, soil, agriculture, uncertainty.