The use of physiological traits for indirect selection may help to increase wheat yield potential. Fruiting efficiency (FE, grain number per unit of spike dry weight at anthesis −SDWa-) may be an alternative to increase grain number per unit area (GN), but the possible cross-over genotype (G)×environment (E) interaction and trade-off with SDWa and grain weight (GW) may limit its usefulness. The present paper aimed to determine (i) the degree of G×E interaction of FE at plot, main stem and tiller levels, (ii) the relationship between GN and FE, and (iii) the possible trade-off between FE and (a) SDWa, and (b) GW. Three FE contrasting genotypes (B11 and B19 – high FE, and B2002 – low FE) were grown under different Nitrogen (N) availability at sowing [(i) low – Ni: ca. 50kgNha−1; (ii) middle −130kgNha−1 and (iii) high −200kgNha−1). Two other treatments were applied within the 200N: (iv) late sowing, i.e. delaying sowing 30days in relation to optimum date (200N+LS) and (v) shading the canopy during stem elongation (200N+SH). The FE showed G×E interaction (year×genotype and genotype×treatment) but it only explained 13.4% of its variation. The environment (year) had similar impact than the genotype, explaining both ca. 38% of FE variation (19% each). The FE of tiller spikes was higher than that of main stem spikes, but the difference among cultivars was maintained. Thus, canopies of the same genotype with different proportion of main stem and tiller spikes may present different FE value at plot level. Nevertheless, a high correlation (r=0.82, p<0.0001) and a close 1:1 relationship was observed between FE at plot level and FE in main stems, for all treatments and cultivars, suggesting that sampling the main stem spikes represents the FE at plot level in dense canopies. A G×E cross-over interaction was observed when the sowing was delayed one month (200N+LS) from the optimum date because B19 reached anthesis later than B2002, exploring higher heat stress index (°Ch>27°C) around anthesis, which reduced FE. The correlation of GN with FE was high and significant (r=0.74, p<0.00001) for all treatments and for a wide range of SDWa values (from 90 to 188gm−2). The B19, selected as a high FE genotype, showed higher GN than B2002 (selected as a low FE genotype) under all treatments, while B11 (also of high FE) showed similar or higher GN than B2002, depending on treatment. Thus, despite the G×E interaction, the genotypes with higher FE had more GN under a wide range of environments. Nevertheless, the SDWa and GW were negatively correlated with FE (for SDWa r=−0.73 p<0.005, without considering 200N+SH; and for GW r=−0.48 p<0.01 for all treatments or r=−0.75 p<0.08 for genotypes only). Then, special attention should be given to SDWa and GW when selecting for higher FE in a breeding program.
Read full abstract