Does modelling of spatial heterogeneity matter?

Summary

The presence of grazing animals contributes to a much greater spatial heterogeneity in the cycling of C and N within grasslands, compared to arable land. This is primarily due to the deposition of excreta in patches rather than evenly over the field. However, an additional source of heterogeneity is introduced because grazing animals do not evenly utilise the grassland. Herbage contaminated by their faeces will tend to be rejected, as will poor quality (low digestibility) vegetation. There is therefore good reason to expect that C and N cycling will be highly spatially variable in grazed grassland. Hitherto, grassland models have assumed C and N dynamics to be uniform within a field. In this paper, modelling is used to investigate the potential effect of spatial heterogeneity on the estimated C sequestration and N emissions in grazed grassland.
The FASSET farm model (www.fasset.dk) was modified to describe the dynamics of spatial heterogeneity in grassland. For each day that the grassland is grazed, the area of sward receiving urine is calculated from the stocking density, frequency of urination and an assumed standard area per urine. This calculation is repeated for faeces. The grassland is initially assumed to be spatially homogenous i.e. is comprised of a single ‘basal’ patch, with an area equal to that of the field. Each day, three new patches are created to represent the newly formed excretal areas, each containing independent versions of the sub-models of the soil/plant system. Two of the new patches are created with a size equal to the area of the basal patch that covered by the sum of all excretions of either dung or urine from each day. The third is an ‘overlap’ patch, representing the situations where excreta is deposited on areas that have previously received dung or urine. The area attributed to the basal patch and older dung or urine patches is reduced accordingly. Herbage offtake and the amount of C and N in excreta is calculated using a cattle model based on the Danish factorial system for energy and protein partitioning but with the addition of diet selection.
Three scenarios of increasing complexity were examined; 1) no heterogeneity, 2) heterogeneity but no diet selection and 3) heterogeneity plus diet selection. The model was run for 5 years.
When the soils were initiated with a C content typical for a Danish soil, there was considerable annual C sequestration. The simulated C sequestration was only marginally affected by inclusion of spatial heterogeneity, as there is a fairly linear relationship between C return to the soil and total sequestration. However, the simulated N2O emissions and nitrate leaching were higher in the scenarios including spatial heterogeneity, as the relationships between the N inputs and these emissions are highly non-linear.