Prediction of hybrid performance in grain sorghum using RFLP markers

Abstract

Heterosis is an important component of hybrid yield performance. Identifying high yielding hybrids is expensive and involves testing large numbers of hybrid combinations in multi-environment trials. Molecular marker diversity has been proposed as a more efficient method of selecting superior combinations. The aim of this study was to investigate the value of molecular marker-based distance information to identify high yielding grain sorghum hybrids in Australia. Data from 48 trials were used to produce hybrid performance-estimates for four traits (yield, height, maturity and stay green) for 162 hybrid combinations derived from 70 inbred parent lines. Each line was screened with 113 mapped RFLP markers. The Rogers distances between the parents of each hybrid were calculated from the marker information on a genome basis and individually for each of the ten linkage groups of sorghum. Some of the inbred parents were related so the hybrids were classified into 75 groups with each group containing individual hybrids that showed similar patterns of Rogers distances across linkage groups. Correlations between hybrid-group performance and hybrid-group Rogers distances were calculated. A significant correlation was observed between whole genome-based Rogers distance and yield (r = 0.42). This association is too weak to be of value for identifying superior hybrid combinations. One reason for the generally poor association between parental genetic diversity and yield may be that important QTLs influencing heterosis are located in particular chromosome regions and not distributed evenly over the genome. Variation in the sign and magnitude of correlations between Rogers distance and hybrid-group performance for particular linkage groups observed in this study support this hypothesis. The concept of using diversity on individual linkage groups to predict performance was explored. Using data from just two linkage groups 38% of the variation in hybrid performance for grain yield could be explained. A model combining phenotypic trait data and parental diversity on particular linkage groups explained 71% of the variation in grain yield and has potential for use in the selection of heterotic hybrids.