Determinants of barley grain yield in drought-prone Mediterranean environments

The determinants of barley grain yield in drought-prone Mediterranean environments have been studied in the Nure x Tremois (NT) population. A large set of yield and other morpho-physiological data were recorded in 118 doubled-haploid lines of the population, in multi-environment field trials (18 site-year combination). Agrometeorological variables have been recorded and calculated at each site too. Four main periods of barley development were considered, vegetative, reproductive early and late grain filling phases, to dissect the effect on yield traits of the growth phases. Relationships between agrometeorological variables, grain yield (GY) and its main components (GN and GW) were also investigated by correlation. Results firstly gave a clear indication of the involvement of water consumption in determining GY and GW (r²=0.616, P=0.007 and r²=0.703, P=0.005, respectively) calculated from sowing to the early grain filling period, while GN showed its highest correlation with the total photothermal quotient (PQ) calculated for the same period (r²=0.646, P=0.013). With the only exception of total PQ calculated during the vegetative period, all significant correlations with GY were associated to water-dependent agrometeorological parameters. As a second result, the NT segregating population allowed us to weight the amount of interaction due to genotypes over environments or to environments in relation to genotypes by a GGE analysis; 47.67% of G+GE sum of squares was explained by the first two principal components. Then, the introduction of genomic information at major barley genes regulating the length of growth cycle allowed us to explain patterns of adaptation of different groups of NT lines according to the variants (alleles) harbored at venalization (Vrn-H1) in combination with earliness (Eam6) genes. The superiority of the lines carrying the Nure allele at Eam6 was confirmed by factorial ANOVA testing the four possible haplotypes obtained combining alternative alleles at Eam6 and Vrn-H1. Maximum yield potential and differentials among the NT genotypes was finally explored through Finlay-Wilkinson model to interpret grain yield of NT genotypes together with yield adaptability (Ya), as the regression coefficient b_i; Ya ranged from 0.71 for NT77 to 1.20 for NT19. Lines simply harboring the Nure variants at the two genes behaved as highest yielding (3.04 t ha^-1), and showed the highest yield adaptability (b_i=1.05). The present study constitutes a starting point towards the introduction of genomic variables in agronomic models for barley grain yield in Mediterranean environments.