Modelling biological N fixation and grass-legume dynamics with process-based biogeochemical models of varying complexity

Pastures comprising grass-legume mixtures along with the substitution of biological nitrogen fixation (BNF) for fertilisers offer the opportunity to reduce nitrous oxide emissions directly from soils without negative impacts on productivity. Models offer the potential to test how legumes can be used to meet environmental and production goals but many models used to simulate greenhouse gas (GHG) emissions from grasslands have either a poor representation of grass-legume mixtures and BNF, or poor validation of these features. Our objective is to examine how such systems are currently represented in two process-based biogeochemical models, APSIM and DailyDayCent (DDC), when compared against an experimental dataset with different grass-legume mixtures at three fertiliser rates. Furthermore, we propose a novel approach of loosely coupling DDC to APSIM, to increase DDC’s capability when representing a range of grass-legume proportions. We found that both models could capture key aspects of grass-legume dynamics and BNF to reasonably simulate the interactions as legume proportions in the pasture changed, particularly in mixtures with a high clover fraction. Although there is a need for further exploration, our work suggests that single species models should not be used for grass-legume mixtures beyond about 30% legume content unless used based on a similar approach adopted here.