The physiological and genetic bases of water-use efficiency in winter wheat
Winter wheat (Triticum aestivum L.) is the most extensive arable crop in the UK grown on about 2M ha p.a. There is a need to identify traits to ameliorate yield losses to drought which are on average about 15% per year. These losses will be exacerbated with predicted climate change. The overall objectives of the present study were to investigate the physiological and genetic bases of water-use efficiency (ratio of above-ground dry matter production to evapotranspiration; WUE) in winter wheat grown in UK conditions and to quantify relationships between WUE and yield performance under drought. The present study used a doubled haploid (DH) population of 33 lines derived from a cross between Beaver and Soissons, known from previous work to contrast for WUE. Two glasshouse experiments (2002/3 and 2003/4) and two field experiments (one at ADAS Gleadthorpe, Nottinghamshire in 2002/3 and the other at Sutton Bonington, University of Nottingham in 2004/5) were conducted. In the glasshouse experiments, two irrigation treatments (with and without irrigation) were applied to four genotypes (two parents and two DH lines), and in the field two irrigation treatments (rainfed and fully irrigated) were applied to the two parents and the 33 DH lines. A range of physiological traits was measured, including developmental stages, carbon isotope discrimination (Δ13C), leaf gas-exchange variables, green areas and biomass at sequential samplings, and these traits were related to grain yield. Transpiration efficiency (ratio of above-ground dry matter production to transpiration; TE) was assessed using the established inverse relationship between TE and Δ13c. In the glasshouse, WUE measured as the regression slope of dry matter on water use, did not differ amongst genotypes in 2003, but did in 2004. Soissons showed higher WUE than other genotypes under irrigation, and also higher WUE than Beaver under drought. For measurements of TE according to Δ13. Soissons and line 134G showed lower Δ13C values (higher TE) than line 134E and Beaver (P<0.05) in 2004 under both irrigation and drought. Soissons and line 134G showed consistently higher TE on account of lower stomatal conductance (gs ) and sub-stomatal C02 concentration (C) values. The early developing Soissons and line 134G exhibited greater flag-leaf green area persistence under drought than the late developing Beaver. Beaver tended to use more water than Soissons under both irrigation and drought, but reductions in water use under drought were similar amongst genotypes. Lower seasonal water use for Soissons than Beaver was associated with a smaller root system. There was a tendency for dry matter of Beaver to be more depressed under drought than Soissons in both the years. Overall, it was not possible to detect significant differences in biomass responses to drought amongst the genotypes, but there were consistent genetic differences in WUE and TE observed under both irrigated and droughted conditions. In the field experiments, the onset of drought coincided broadly with anthesis. The average grain yield losses under drought were 0.5 t ha-1 in 2003 and L6 t ha-1 in 2005. Averaging across site/seasons, Δ13C correlated positively with grain yield amongst the 35 genotypes under irrigation (r--0.35; P<0.05) and under drought (r--0.54; P<0.01), indicating a negative trade off between TE and yield. A 13C decreased under drought and a higher TE was associated with a reduction in average flag-leaf gs, measured from flag leaf emergence to anthesis + 4weeks. Stomatal conductance was measured for eight of the 33 DH lines including the parents, and there was a trend for lower Δ13C (higher TE) to be associated with lower gs, The genetic differences in gs, were generally associated with corresponding decreases in Ci and net photosynthetic rate (A). Therefore results suggested that the negative relationship between TE, as indicated by Δ13C and yield was associated with corresponding reductions in seasonal water use. There was a nonsignificant irrigation x genotype interaction at Gleadthorpe in 2003 and Sutton Bonington in 2005 for Δ13C indicating that this trait was of high heritability. There was an irrigation x genotype interaction for grain yield (P<0.05). A small number of genotypes showed higher yield associated with low Δ13C and these outlier lines could potentially be identified for breaking the negative linkage between yield and delta. In summary, WUE was negatively correlated with yield under drought in this population; and season-long water use appeared to be the most important component affecting yield levels under drought. It is suggested that selecting genotypes indirectly for high Δ13C (low WUE) may be a strategy to improve grain yield under drought. In the quantitative genetic analysis, the putative QTLs identified for target physiological traits were generally different at Gleadthorpe in 2003 and Sutton Bonington in 2005. The most confident putative QTLs for Δ13C were mapped on chromosomes 3B (LOD=2.32) and 2D (LOD=1.43). The identification of QTLs as potential candidate genes on these chromosomes may be associated directly with WUE in the Beaver x Soissons DH mapping population. The Δ13C QTL on chromosome 3B was detected commonly in both the irrigation environments and the direction of allelic effects was consistent with the parental differences in Δ13C. This QTL may therefore represent a novel gene for optimising WUE. It is suggested that breeders could optimise TE by selection according to a marker for this gene involving further fine-mapping to identify a marker tightly linked to the gene. Such a marker would also provide a target for gene discovery in future work. The results suggest that water use is the most important component of Passioura's yield model for yield improvement under UK conditions. Nevertheless, WUE and harvest index and their responses under drought will also likely play a role in yield improvement through breeding in the UK targeted at drought-prone environments in future years.