Spike fertility index (SF) has been proposed as a promising trait to be used as a selection criterion in wheat breeding programs aimed at increasing grain yield, but no actual evidence of its successful application has been reported. In this study, 146 recombinant inbred lines derived from a cross between 'Baguette 10' and 'Klein Chajá', Argentinean spring bread wheat cultivars with contrasting SF, were evaluated during three crop seasons (2013, 2014 and 2015) at Balcarce, Argentina. Grain yield, grain number/m 2 , grain weight, and SF were measured at maturity. Changes in grain yield (i.e., responses to selection) after application of different selection strategies, including different selection criteria and selection intensities, were determined. Significant correlations were observed between grain number and grain yield, SF and grain yield, and SF and grain weight. Analysis of SF variance components showed a significant genotype 9 environment interaction, but it represented only 9% of the total variation, whereas 51% of the variation was genetic, resulting in a high narrow-sense heritability (0.84). The use of SF as a selection criterion, either solely or in combination with selection for high yield, increased yield, resulting in higher and more stable yields than if selecting for high yield alone. Our findings support the use of spike fertility index as a selection criterion for increasing genetic progress and stability of yield in bread wheat breeding programs.

In wheat, grain number is considered as the product of spike dry weight (SDW) and the number of grains per unit of SDW, that is an indicator of spike fertility (SF). The aim of this study was to determine the heritability of SF and the effect of early selection for high SF on its relationship with other spike yield components. Two field experiments were conducted in the south‐eastern Pampas (Argentina) with 400 F2 and F2:3 families obtained from two crosses between varieties with contrasting SF (PIG/SSN and B10/KCJ). Heritability estimates in PIG/SSN and B10/KCJ were, respectively, 0.60 and 0.51 by variance component analysis, 0.43 and 0.43 by F2 : F3 parent–offspring regression and 0.30 and 0.28 by realized heritability analysis. The existence of transgressive segregation (i.e. the occurrence of families with SF values that were more extreme than those of the parents) was observed. The top 25% F3 families with the highest SF had 12% more grains per spike, despite a 13% and 5% decrease in SDW per spike and weight per grain, respectively, than the remaining families. These results give support to the application of early selection for high SF.