Prediction of weather damage of mungbean seed in tropical Australia. I. Relation between seed quality, weather, and reproductive development

Abstract

Assessment of the potential for mungbean cropping in the Australian monsoon tropics required a model that could predict pre-harvest seed quality from long-term climatic data. Empirical relations between seed quality and pre-harvest weather were developed from field-grown mungbean using 22 sowings over 3 seasons. Seed quality reflected visual symptoms of weather damage expressed as the percentage of undamaged seed. A minimum exposure to rainfall was required before seed quality was reduced. After this minimum was exceeded, the effect of additional rainfall was cumulative and the percentage of unweathered seed decreased proportionally until a maximum was reached whereby all susceptible seed was weather damaged. The percentage of unweathered seed was best predicted as a function of the cumulative duration of rainfall events. Exposure to at least 300 min of rainfall was required before seed quality was downgraded. Exposure to 4000 min of rainfall was required to reach the maximum threshold. The linear decline in the percentage of unweathered seed was accurately predicted with independent data (r2 = 0.84) by a function that combined the cumulative duration of rainfall and the standard deviation of evaporation. This function reflected the weathering process, that is, cumulative exposure to moisture and the extent of drying of the atmosphere between rainfall events. Alternatively, where pluviograph data were unavailable, combining the sum of rainfall events (>0.5mm) with the standard deviation of evaporation and mean daily solar radiation was also highly correlated with the proportion of unweathered seed; accurate predictions were made using independent data during crop ripening (r2 = 0.93) and after ripening (r2 = 0.72). Weather damage was sensitive to the timing of reproductive development relative to rainfall; adjusting climate variables for cohort-specific exposure removed the confounding effects caused by the daily ripening of pods. Time to flowering was accurately predicted, 2–3 days from observed, using mean daily photoperiod and temperature. As expected, rate of progress from flowering to the first ripe pod and crop maturity was dependent on photoperiod, temperature, and moisture availability. The proportion of pods ripe on any day was highly (P < 0.01) correlated with the proportion of the pod-ripening phase completed.