Interpretation of interactions in sunflower agronomic trials using multiplicative models and climatic information

Abstract

A three-year experiment (2005, 2006, 2007) including three sunflower hybrids and eight sowing dates was carried out in order to study the impact of hybrid × sowing date interaction on oil yield of sunflower hybrids. With an aim of attaching biological significance to the interaction interpretation, information on four climatic factors (minimum, maximum and mean temperature and precipitation) were used. Significant differences were found between hybrids, sowing dates and years regarding their impact on oil yield. The results of 3-way ANOVA showed that all sources of variation were highly significant (main effects and interaction effects). This indicated variations among sunflower hybrids for oil yield and variations in sowing date and hybrid × sowing date effects. The multiplicative interaction between H × R was further separated in two bilinear terms (PC1 and PC2), and both were highly significant. The SREG2 biplot indicated that the hybrid Miro was the best performer at 11 planting dates. During the three-year experiment, Pobednik produced highest oil yields at 10 planting dates and Rimi only at 3. Sowing dates R1-6, R2-6 and R3-6 were best for oil yield, because they had highest PC1 values and near-zero PC2 values were (SREG2). On the basis of percentages in the first significant dimension, three variables (pr3, mx3, mn3) higher than 50% and with high positive values of loading were extracted. The PLS regression tri-plot shows that all variables were distributed in 4 groups with similar (or different) effects on the total interaction. Minimum temperature (mn4) at physiological maturity had the smallest contribution to the H × R interaction for oil yield. Sowing dates R4-5 and R5-5 also had smallest contributions to the H × R interaction, because they were located near zero point (0.0) and because their oil yields were smaller than the average. The results for sunflower oil yield indicated that the relative performance of the hybrids and sowing dates was strongly under the influence of their different reactions to precipitation, maximum and minimum temperatures at the flowering stage.