Development of a variety selection decision support system: genotype × environment analyses and system validation.

The objective of this study was to develop a basic variety selection decision support system (DSS) for the South African sugar industry. The approach adopted considered legal aspects, cutting cycle, and structured genotype-by-environment (GxE) analyses. Data extracted from SASRI’s variety trial database were categorised into different regions, harvest ages (12, 18, 24 months) and harvest seasons (early, mid, late). Restricted maximum likelihood (REML) analyses were conducted within regions to determine varietal adaptability to different harvest ages and seasons. Varietal adaptability to different yield potential conditions was determined using the site regression technique, and was interpreted from the slope of the regression curves. Highly significant variety x harvest age and variety x season interactions allowed for the appropriate categorisation of varieties. The analysed data were used to create simple ‘yes/no’ matrices, which were housed within a relational database. A web interface linked to the database allows users to specify characteristics of their production environment. The VChoice™ system then selects appropriate varieties that conform to specified criteria and eliminates non-compliers in a stepwise approach. The system was subsequently validated against expert extension specialist opinion, where the number of choices made by experts was compared to corresponding selections made by the DSS. The results showed acceptable performance (84% success rate) by the DSS. The system is currently implemented as a webbased application and is planned to form the framework of a more comprehensive system that will include environmental factors and actual yield data to assist with variety selection.