Broad sense heritability and genotype × environment interaction for carbon isotope discrimination in field-grown wheat

Carbon isotope discrimination (-) has been proposed as a possible selection criterion for greater water use efficiency in breeding programs for water-limited environments because it provides an integrative assessment of genotypic variation in leaf transpiration efficiency. Considerable genotypic variation for - has been demonstrated in wheat, but environmental factors may cause even larger changes in the value of - measured in plant dry matter, which could compromise the effective use of - in breeding programs. In this study we assess broad-sense heritability of - and the significance of genotype x environment interaction for - in field-grown wheat. Another objective was to identify the most effective growth stage or plant part to characterize genotypic variation in -. Experiments were done using several large sets of genotypes (between 8 and 40, usually c. 20) grown in a range of field environments spanning the southern Australian wheat-belt. Carbon isotope discrimination was determined on unreplicated grain samples from seven Interstate Wheat Variety trials grown in 1983 and 1984 and on several plant parts taken from replicated experiments conducted at four locations in south-west New South Wales from 1985 to 1988. From these replicated experiments broad-sense heritabilities for - were calculated on a genotype mean basis h2-M) and on a single-plot basis (h2-P). In dry matter sampled from several environments, site-mean - ranged from 21.0 x 10-3 to 18.9 x 10-3 for early-formed dry matter and from 16.4 x 10-3 to 13.4 x 10-3 for grain. When followed in a single environment, the value of - fell from c. 20 x 10-3 in early-formed leaves to 15.4 x 10-3 in the grain. Variation among genotypes in - of different plant parts was always significant, and was typically c. 2 x 10-3 . Among Australian wheats, low values of - (implying greater transpiration efficiency) were strongly associated with the WW15 genetic background. Estimates of broad-sense heritability for - averaged over 95%, on a genotype mean basis, in experiments where common genotypes were grown in numerous environments. In individual trials, heritability was lowest for plant material sampled near anthesis (average value for h2-M, 83% and for h2-p, 62%) and greatest for dry matter laid down before or during early stem elongation (average value for h2-M, 95% and for h2-P 88%). Even though heritability for grain - was also relatively high (average value for h2-M, 92% and for h2-P, 79%), genotypic differences in grain - are difficult to interpret because of the likelihood of some changes in genotype ranking for - resulting from differences among genotypes in the degree of water stress encountered during grain filling. As well, the contribution of remobilized carbon to grain - may vary between environments and genotypes. We conclude that, for wheat, assessment of genotypic variation in - should be most effective under well-watered conditions using dry matter laid down early in plant development.