Response of common wheat varieties to organic and conventional production systems across Italian locations,and implications for selection

Specific breeding for organic systems may help reduce their yield gap relative to conventional systems by exploiting genotype × system (GS) interaction. Likewise, specific breeding for distinct subregions within a region could capitalize on genotype × location (GL) interaction. Grain yield and test weight of common wheat varieties were evaluated under organic and conventional systems in ten locations spanning from northern to southern Italy, with the objectives of: (i) comparing production systems; (ii) investigating the extent of GS and GL interactions and their relationship with genotypic and environmental characteristics; and (iii) preliminarily comparing, in terms of predicted selection gains, different strategies to cope with GS and GL effects. These effects were investigated in the 2-year Data set 1 including seven genotypes. GS effects were also assessed in the annual Data sets 2 and 3 including 13 and 11 genotypes, respectively. The yield reduction of organic systems relative to conventional ones averaged 28% in Data set 1, 29% in Data set 2 and 14% in Data set 3. Organic systems also tended to a modest test weight reduction. Genetic correlations between systems ranged from high to very high (0.88 ≤ r_g ≤ 0.98) for yield and test weight, owing to nil or limited GS interaction. Broad-sense heritability tended to be higher in conventional systems than organic ones for yield in two data sets (mainly due to lower experimental error) while being similar in the two systems in the other cases. Predicted selection gains suggested nil (yield) or very modest (test weight) advantage of direct selection in organic systems relative to indirect selection in conventional systems, when targeting organic systems. The scope for selection only in conventional systems was reinforced when comparing predicted gains for selection scenarios which target both systems in relation to their foreseeable marketing importance. GL effects for yield and test weight were significant and were modeled by additive main effects and multiplicative interaction analysis. Site classification based on GL effects for yield revealed a larger subregion A including northern and central Italy and a smaller subregion B comprising southern Italy, accordingly with previous, independent studies. Yield selection only in subregion A (with indirect selection gain for subregion B) implied slightly higher predicted gain for A (+4%) and much lower gain for B (−24%) relative to independent, direct selection in each subregion. Selection for specific geoclimatic subregions may have greater importance than selection for specific production systems.