Modelling of lucerne (Medicago sativa L.) for livestock production in diverse environments

Modelling of lucerne (Medicago sativa L.) for livestock production in diverse environments

Several models exist to predict lucerne (Medicago sativa L.) dry matter production; however, most do not adequately represent the ecophysiology of the species to predict daily growth rates across the range of environments in which it is grown. Since it was developed in the late 1990s, the GRAZPLAN pasture growth model has not been updated to reflect modern genotypes and has not been widely validated across the range of climates and farming systems in which lucerne is grown in modern times. Therefore, the capacity of GRAZPLAN to predict lucerne growth and development was assessed. This was done by re-estimating values for some key parameters based on information in the scientific literature. The improved GRAZPLANmodel was also assessed for its capacity to reflect differences in the growth and physiology of lucerne genotypes with different winter activity. Modifications were made to GRAZPLAN to improve its capacity to reflect changes in phenology due to environmental triggers such as short photoperiods, declining low temperatures, defoliation and water stress. Changes were also made to the parameter governing the effect of vapour pressure on the biomass-transpiration ratio and therefore biomass accumulation. Other developments included the representation of root development and partitioning of canopy structure, notably the ratio leaf : stem dry matter. Data from replicated field experiments across Australia were identified for model validation. These data were broadly representative of the range of climate zones, soil types and farming systems in which lucerne is used for livestock grazing. Validation of predicted lucerne growth rates was comprehensive owing to plentiful data. Across a range of climate zones, soils and farming systems, there was an overall improvement in the capacity to simulate pasture dry matter production, with a reduction in the mean prediction error of 0.33 and the root-meansquare deviation of 9.6 kg/ha.day. Validation of other parts of the model was restricted because information relating to plant roots, soil water, plant morphology and phenology was limited. This study has highlighted the predictive power, versatility and robust nature of GRAZPLAN to predict the growth, development and nutritive value of perennial species such as lucerne.