The development of modem soybean (Glycine max L. Merrill) has produced a highly productive crop. However, the crop is not well adapted to environments in which water stress is common. In Australia, the cropping of soybean is limited by lack of adaptation to stress-prone environments and in particular, by the rapid 'firing' of leaves once soil water has been depleted.
This thesis examines the case for breeding more drought-tolerant soybean varieties, using trait-based selection. Experiments were conducted to establish the range of genotypic variation in four traits with putative value in enhancing leaf survival during drought: leaf water potential, osmotic adjustment, leaf epidermal conductance and critical relative water content.
Variation for these traits was explored in a wide range of soybean genotypes from different regions of the world, and also in several wild Glycine species that are adapted to drought-prone environments in Australia. Experiments were conducted to explore the interrelations among these traits and to establish the relation between genotypic variation in these traits and variation in leaf survival under water-stressed conditions. To assist in the further assessment of the effects of these traits, in isolation and in combination, a mathematical model was developed to describe the interactions among the traits and to explore new combinations and predict their putative effects on leaf survival. The effect of combinations of traits was predicted with a level of accuracy comparable with the magnitude of the errors of measurement in the original trait determinations due to the combination of many trait estimates into the model parameterisation.
The heritability of epidermal conductance, osmotic adjustment and leaf relative water content was examined and all traits were found to be moderately heritable.
The model was used to explore the likely changes in leaf area over time as available soil water was depleted, given the range of heritable variation in the several leaf traits in soybean and its wild relatives. Substantial re-ranking of the putative leaf area survival occurred with different levels of expression of traits, due to the effect of profligate or conservative use of soil water early in the simulated water stress.
Comparison of the combinations of traits in current varieties with model predictions provided circumstantial evidence that, in eastern Australia, past selection may have driven the combination of traits conditioning leaf survival into response strategies appropriate to maximise productivity in the different environments under which selection was undertaken. However, most environments have been relatively waterstress free and desirable genotypic variation in leaf traits appeared not to occur in varieties in modem plant breeding programs.
The main conclusion of this thesis is that sufficient additional desirable genotypic variation exists in soybean for levels of leaf traits that affect leaf survival during water stress to enable more drought tolerant varieties to be bred. With this additional variation, it is postulated that traits could be combined into a stress response strategy with sufficiently enhanced leaf survival to permit dryland cropping to be substantially expanded in tropical/subtropical Australia. Whether the breeding investment to achieve the outcome would be justified economically would need to be assessed through a detailed benefit: cost analysis.