Increasing yield potential in wheat: complementing conventional breeding by application of novel physiological and germplasm strategies

• Australian National University, Australia
• University of Queensland, Australia
• CSIRO Plant Industry, Australia

Wheat accounts for more than 25% of the world's cereal output, and is the main source of calories for more than 1.5 billion people. More than half the 220 million ha sown to wheat is in developing countries; and at least 70% of the developing world's wheat area is planted to varieties originating ultimately with CIMMYT (the International Maize and Wheat Improvement Centre).

CIMMYT has played a major role in increasing the productivity of wheat around the world for more than 30 years. However, in spite of its achievements, projections indicate that global ability to supply wheat will lag behind demand in the future. Over the next 20 years, demand is expected to grow at 1.3% per year worldwide, and by 1.8% in developing countries, yet the area sown to wheat is expected to grow by only 0.14% annually in the same period. In many countries, farmers' yields are close to the maximum potential yields obtained on experimental stations in the same area, and so it is unlikely that on-farm improvements can be a source of further gains.

There is therefore an urgent need to develop new and more efficient wheat-breeding methods to complement existing techniques, as well as to identify new traits to drive faster yield gains by exploiting the true biological yield potential of wheat.

Evidence suggests that some physiological traits - in particular, those relating to stomatal aperture - have considerable potential for improving the efficiency of yield gains. These traits, known as SATs (stomatal aperture-related traits) need to be studied and their usefulness evaluated. That was the main thrust of this project. A second activity was to elucidate the physiological basis for the relationship between SATs and yield potential.

The main aim was to capitalise on promising new techniques, based on physiological criteria, which could be used for selecting high-yielding wheat varieties for breeding.

The work was divided into three sub-projects. The first focused on evaluating genetic gains associated with using stomatal aperture-related traits (SATs) as physiological selection criteria, and this took place at CIMMYT (in Mexico) and in Australia.

The second sub-project identified underlying physiological and genetic mechanisms of the SATs. The aim was to provide a clear understanding of the link between the observed SATs and yield in CIMMYT material, so that scientists could more easily predict ways of further increasing yield using physiological reasoning. Part of this research examined whether high-yield lines have a less conservative response to incipient environmental stresses (such as higher evaporation rates or drying topsoil), and also investigated photosynthetic rates and responses to heat stress.

The third sub-project assessed the potential of genetic sources of variation in other physiological traits, such as high biomass production, longer rapid spike growth phase, or large kernel size. The team evaluated new sources of germplasm and elite lines and produced homozygous sister lines from contrasting parents to establish genetic links between traits and potential yield gains.

In evaluating genetic gains associated with use of stomatal aperture-related traits (SATs) as early generation selection criteria, the scientists measured yield and SATs on large plots and small observation plots, respectively, on five populations of random inbred lines (RILs) - from elite/elite breeder crosses - over three crop cycles in a high-yield irrigated environment in NW Mexico. Visual estimates of yield followed by actual yield were also measured on small plots. SATs evaluated were as follows:
leaf conductance or porosity (POR), measured once during both booting and grain-filling stages by measuring six to eight flag leaves per plot on sunny days in two crop cycles
canopy temperature (CT) measured between three and five times on sunny, relatively windless days during both the booting and grain-filling stages in all three cycles
carbon isotope discrimination (CID) on leaves at approximately the first node stage as well as harvested grain in one cycle.

On average, genetic correlations between SATs measured on small plots and yield on large plots varied from 0.5 for CID to 0.7 for CT. Correlated phenotypic response of yield in response to selection for the best 25% and worst 25% genotypes, based on their SATs, showed significant variation. Approximate genetic gains associated with the 25% selection intensity were: 50 g m-2 for yield itself, 40 g m-2 for CID, and 30 g m-2 for CT, POR, and visual estimates. Since CT, POR and visual estimates are the cheapest and easiest to measure, multiple regression analysis was conducted to establish if traits were additive.
The result of analyses on all five crosses combined over years suggested that while between 13 and 56% of yield could be explained with a single trait, 26-63% of yield could be explained by combining them. The conclusion is that a combination of visual as well as simple-to-use SAT evaluations is desirable. This conclusion was backed by measurements of CT in parallel populations of breeder-selected lines, in which the combination of CT with visual selection identified not only a larger proportion of high-yielding lines but also the most productive lines.
Other experiments led to the following conclusions
(a) genetic gains in response to selection for SATs are equally possible with both advanced and early-generation breeding materials;
(b) CT measured at higher vapour pressure deficit (VPD) was associated with a better correlation with yield;
(c) SATs measurements are relatively robust to different stages of development;
(d) CT showed a larger correlation with yield than CT-depression;
(e) economic assessment of the use of SATs within CIMMYT's wheat breeding program lent strong support for their potential value in reducing costs, for example by discarding physiologically substandard lines prior to extensive yield testing.
The team sought to identify the underlying physiological mechanisms associated with SATs by testing the following hypotheses:
(a) Is the association of SATs with yield a function of sink strength (studied using genetic and physiological models)?
(b) Do high-yield potential lines have a less conservative stomatal response to incipient environmental stresses?
(c) Are good vascular systems the key to higher rates of transpiration?
(d) Do high-yielding lines have intrinsically higher photosynthetic potential, permitting favourable growth of the spike before anthesis, and of the grain after anthesis?
(e) Are the cooler canopies of high-yield potential lines less susceptible to the effects of heat stress?
The team concluded that the relationship between SATs and yield is probably driven by multiple factors (and may depend on specific germplasm). These factors include: (i) sink strength, (ii) a less conservative response to incipient soil water deficit that may be associated with a superior vascular system; (iii) superior photosynthetic capacity.
In undertaking an assessment of the value of traits related to source/sink balance, the scientists developed three populations of random inbred lines (RILs) to test association of source/sink (SS) traits with yield and biomass in high-yield environments. The SS traits formed three groups relating to (i) phenological pattern, (ii) assimilation capacity up until anthesis, (iii) partitioning of assimilates to reproductive structures at anthesis.
The largest genetic gains in performance traits were associated with the second group. However, traits from the other groups were also identified as being genetically linked to improvement in performance parameters. The trait most consistently associated with performance traits was biomass at anthesis (BMA).
The identification of a number of SS traits associated with yield and biomass, which both PCA and multiple regression suggest as being at least partially independent of one another, support the idea that cumulative gene action could be achieved by adopting a physiological trait-based breeding approach where traits from different groups are combined in a single background.