Impact of genotype multiply environment interaction on response to selection in sugarcane

The impact of genotype x environment interaction on the response to 3 methods of selection in sugarcane was assessed over 3 locations and 3 cropyears. The 3 selection methods considered were 2 previously used procedures, mass (individual) selection using a visual appraisal of clonal performance and family selection using selection rate, and a new procedure, family selection using a net merit grade (NMG) derived from weighing family plots. Within the 2 forms of family selection, light mass selection based on a visual appraisal of clonal performance was used to reject the poorest clones. The gains from selection, cdculated as the deviation in performance of the select groups from the performance of a group of randomly chosen clones, were assessed for each of these 3 methods. All 3 selection methods gave similar gains over the random group when averaged over all locations and crop-years. Family selection based on NMG is recommended for routine use as it is the least labour-intensive and most cost-effective. Genotype x environment interaction, on both a family and clonal basis, was large enough to affect the gains made from selection. These effects provided some support for location-specific selection. Furthermore, selections based on the results from the central experiment station produced relatively poor gains when assessed over all locations and crop-years. This indicates that the experiment station may be a poor location for centralised selection. The use of at least 1 other location may provide a buffer against infrequent environmental factors rendering worthless family or clonal discrimination at any particular location.