Modelling rangeland productivity in response to degradation in a semi-arid climate

Abstract

Modelling rangeland is essential for capturing changes at the large temporal and spatial scales at which these systems respond to climate and institutional changes and increasing population pressure, but rangeland models applicable to data sparse regions are rarely available. We developed and evaluated a novel rangeland model aimed at simulating rangeland at different stages of degradation using limited parameterisation and measurements.The developed model Linrange is a biophysical simulation model of the aboveground part of a mixed grass sward, combined with sub-models for evapotranspiration, soil water dynamics, and root development. Main processes of the biomass model are growth through a source/sink limited mechanism, reserve storage and remobilisation, basal area dynamics, winter dormancy. The grass sward is simulated based on average species characteristics of the dominating grass community.We show that a model based on simplified biophysical processes and a single set of parameters for a mixed sward can satisfactorily simulate mixed-species rangeland vegetation. The model also could reproduce year-to-year phytomass dynamics, including for exceptionally wet and dry years. Without calibrating specifically for it, the model was able to reproduce observed water-use efficiency values, indicating a good representation of the relationship between the main limiting factor, water, and productivity. By recalibrating the model using only five parameters associated with degradation, the accuracy of simulated degraded rangeland states was close to that of undegraded rangeland. We therefore consider the Linrange model a good tool for research on rangeland dynamics and degradation resulting from management and climate. We also point to directions of further model improvement, particularly regarding the modelling of parameter changes with degradation states.