Consistent gains in grain yield in dry environments have been made by empirical breeding although there is disturbing evidence that these gains may have slowed. There are few examples where an understanding of the physiology and the genetics of putative important drought-related traits has led to improved yields. Success will first depend on identifying the most important traits in the target regions. It will then depend on accurate and fast phenotyping, which, in turn, will lead to: (1) trait-based selection being immediately transferable into breeding operations and (2) being able to identify the underlying genes or the important genomic regions (quantitative trait loci), perhaps leading to efficient marker-based selection (MBS). Genetic complexity, extent of genotype × environment (G × E) interaction and sampling cost per line will determine value of phenotyping over MBS methods. Here, we review traits of importance in dry environments and review whether molecular or phenotypic selection methods are likely to be the most effective in crop improvement programs and where the main bottlenecks to selection are. We also consider whether selection for these traits should be made in dry environments or environments where there is no soil water limitation. The development of lines/populations for trait validation studies and for varietal development is also described. We firstly conclude that despite the spectacular improvements in molecular technologies, fast and accurate phenotyping remains the major bottleneck to enhancing yield gains in water-limited environments. Secondly, for most traits of importance in dry environments, selection is generally conducted most effectively in favourable moisture environments.

The high temperature responses of photosynthesis and respiration in wheat are an underexamined, yet potential avenue to improving heat tolerance and avoiding yield losses in a warming climate.

The improvement in grain yield of wheat throughout Australia through both breeding and management has been impressive. Averaged across all farms, there has been an approximate doubling of yield per unit area since ~1940. This has occurred across a broad range of environments with different rainfall patterns. Interestingly, the gain in the driest years (9 kg ha–1 year–1 or 0.81% year–1) has been proportionally greater than in the most favourable years (13.2 kg ha–1 year–1 or 0.61% per year) when expressed as yield relative to 2012. These data from all farms suggest that further yield progress is likely, and evidence is presented that improved management practices alone could double this rate of progress. The yield increases achieved have been without any known compromise in grain quality or disease resistance. As expected, improvements have come from both changed management and from better genetics, as well as from the synergy between them. Yield improvements due to changed management have been dramatic and are easiest to quantify, whereas those from breeding have been important but more subtle. The management practices responsible have largely been driven by advances in mechanisation that enable direct seeding, more timely and flexible sowing and nutrient management, and improved weed and pest control, many of which have been facilitated by improved crop sequences with grain legumes and oilseeds that improve water- and nutrient-use efficiency. Most of the yield improvements from breeding in Australia have come from conventional breeding approaches where selection is almost solely for grain yield (together with grain quality and disease resistance). Improvements have primarily been through increased harvest index (HI), although aboveground biomass has also been important. We discuss future opportunities to further increase Australian rainfed wheat yields. An important one is earlier planting, which increases resource capture. This will require knowledge of the genes regulating phenological development so that flowering still occurs at the optimum time; appropriate modifications to sowing arrangements and nutrient management will also be required. To improve yield potential, we propose a focus on physiological traits that increase biomass and HI and suggest that there may be more scope to improve biomass than HI. In addition, there are likely to be important opportunities to combine novel management practices with new breeding traits to capture the synergy possible from variety × management interactions. Finally, we comment on research aimed at adapting agriculture to climate change.

Using a field to lab approach, mature deep-rooting traits in wheat were correlated to root phenotypes measured on young plants from controlled conditions. Mature deep-rooting root traits of 20 wheat genotypes at maturity were established via coring in three field trials across 2 years. Field traits were correlated to phenotypes expressed by the 20 genotypes after growth in four commonly used lab screens: (i) soil tubes for root emergence, elongation, length, and branching at four ages to 34 days after sowing (DAS); (ii) paper pouches 7 DAS and (iii) agar chambers for primary root (PR) number and angles at 8 DAS; and (iv) soil baskets for PR and nodal root (NR) number and angle at 42 DAS. Correlations between lab and field root traits (r2=0.45–0.73) were highly inconsistent, with many traits uncorrelated and no one lab phenotype correlating similarly across three field experiments. Phenotypes most positively associated with deep field roots were: longest PR and NR axiles from the soil tube screen at 20 DAS; and narrow PR angle and wide NR angle from soil baskets at 42 DAS. Paper and agar PR angles were positively and significantly correlated to each other, but only wide outer PRs in the paper screen correlated positively to shallower field root traits. NR phenotypes in soil baskets were not predicted by PR phenotypes in any screen, suggesting independent developmental controls and value in measuring both root types in lab screens. Strong temporal and edaphic effects on mature root traits, and a lack of understanding of root trait changes during plant development, are major challenges in creating controlled-environment root screens for mature root traits in the field.