A physiological approach to disease yield loss relationships in barley (Hordeum vulgare L.)

Methods for estimation of crop losses caused by disease are based largely on empirical models. Recently there has been an increased awareness that a more mechanistic approach with a greater understanding of the physiological nature of host/pathogen systems would provide more accurate and biologically meaningful estimates of crop loss. Two autumn sown barley experiments were conducted in 1983 and 1984 in Canterbury, New Zealand, to study the physiological basis of the relationship between yield loss and disease. Disease primarily leaf rust, caused by Puccinia hordei and scald, caused by Rhynchosporium secalis was measured based on physiological considerations as per cent green leaf area (%GLA). Disease reduced grain yield significantly largely by decreasing grain weight and grain number in both experiments. Observed yield variation was described by empirical single and multiple point models developed from linear regressions of disease and yield or yield loss. These models were improved significantly by the inclusion of an estimator of healthy crop yield, such as sowing time or crop growth duration. This improvement demonstrated that models based on disease measurements only were not appropriate for predicting the yield of a crop sown at different times during one season or in different seasons. Thus a physiological approach to disease yield loss relationships is necessary to explain yield variation caused by different production environments. Physiological studies of the effect of disease on yield were related primarily to the process of grain filling. The relative contribution to grain filling of assimilates from stem stored carbohydrate reserves and current photosynthesis is dependent on the production environment. It was hypothesised that disease may affect yield potential and the realization of that potential during grain filling, by the effect on storage and utilization of stored carbohydrate reserves. This was investigated by measuring the weight of stem stored reserves from the product of stem weight and per cent carbohydrate at approximately two week intervals from stem elongation to maturity. The maximum weight of stem stored carbohydrates occurred at the onset of grain filling and decreased thereafter until maturity in all treatments. The stem carbohydrate weight reduction between the onset of grain filling and maturity was 1.9 t ha⁻¹ in the nil disease and 0.9 t ha⁻¹ in the disease treatment in 1983. The following season stem carbohydrate weight reductions were 2.8, 2.0 and 2.6 t ha⁻¹ for the nil disease (ND), disease throughout the season (FD) and disease during grain filling (LD) treatments. These results demonstrated that disease affected the weight of stem carbohydrates potentially available for grain filling, depending on the target yield severity and time of epidemic onset. Radioactive labelling techniques were used to test the hypothesis that crops with similar yield potential at the onset of grain filling but with a reduced supply of current photosynthates, caused by disease during grain filling, may compensate by remobilization of stem reserves. The weight of stored carbohydrate reserves which contributed to grain yield was estimated by radioactive techniques with several conservative assumptions, as 1.3 t ha⁻¹ in the LD and 1.0 t ha⁻¹ in the ND treatment. These estimates demonstrate that demand for stored assimilates and the weight remobilized was greater when disease was present during grain filling than when absent. If excess carbohydrate reserves are available to be remobilized to the grain disease thresholds will occur below which there is no significant yield loss, as a result of movement of stored assimilates to the grain in response to the greater demand. This can be used as the basis for a predictive management system and was discussed in relation to a winter wheat crop simulation model (ARCWHEAT I). These investigations indicate that the basis on which the model partitions stored carbohydrate reserves to grain filling would need to be altered to incorporate the effects of disease. The model apportions a fixed amount (30%) of the weight of assimilates at anthesis to grain filling. It was suggested that the proportion could be measured directly at the onset of grain filling. The potential of a regression of accumulated %GLA and weight of stem carbohydrate at the onset of grain filling to determine the weight of assimilates remobilized was also discussed. The onset of grain filling was suggested as more appropriate than anthesis for the allocation of stem reserves for grain filling. The findings of this research can be applied to partitioning submodels of future barley simulation models to provide predictive models for disease management purposes.