Predicting phosphorus loss from structured soils through macropores

Macropore transport is an important process of phosphorus (P) loss from tile-drained agricultural land to surface waters where P inputs may cause accelerated eutrophication. Many laboratory experiments or plot studies have shown that P loss by macropore transport increases with increasing concentrations of mobilizable P in the topsoil. However, operational models that quantify the risk of P losses by macropore transport based on typically available information on soil properties, including P status and soil hydrological properties, are currently lacking.

This study has collated and analyzed comprehensive existing data from standardized column-leaching experiments with 193 topsoils from different locations in Denmark. In addition to general physical and chemical soil properties including soil P pools, water, and P transport were measured on the large undisturbed soil columns. This data has been used to investigate relationships between P loss and soil properties under varying degrees of macropore transport. Specifically, we have used two statistical methods to analyze relationships between variables and to explore predictive models – multiple linear mixed models (MLMM) and structural equation modeling (SEM). The latter technique allows for testing complex causal relationships among observed and latent variables.

Our SEM approach has so far yielded rather poor model fits, and the model structures for estimating the loss of dissolved and particulate P from the columns were characterized by low significance. This was partly due to missing data. In contrast, different MLMM fitted the measured dissolved and particulate P losses satisfactorily. Water-extractable P and saturated hydraulic conductivity were the most important variables for estimating dissolved P losses, while colloid mobilization in soils and tritium leaching breakthrough time explained particulate P losses to a large degree.

Our initial statistical analyses show that P loss in dissolved and particulate form from large columns under macropore runoff scenarios can be reasonably explained by soil properties that are typically mapped in Denmark. This approach could bridge empirical and mechanistic modeling and facilitate mapping the risk of P loss by macropore transport.