Using logistic regression models to determine optimum combination of cane yield components among sugarcane breeding populations

Abstract

Plant breeding aims to produce cultivars that possess optimum trait combinations for yield and other traits. Path coefficient analysis in sugarcane showed number, height and diameter of stalks control cane yield. The objectives of this study were to determine optimum combinations of number, height and diameter of stalks for high sugarcane yield in agro-ecological regions and evaluate the increase in yield when optimum combinations were applied during selection. Data for number, height, diameter of stalks and cane yield measured from seven breeding populations, each representing an agro-ecological region in South Africa, were analysed using logistic regression. The optimum trait combinations for low yield agro-ecological environments was higher number of stalks that were taller, whereas in high yield environments, thicker stalks were favourable. Harvest ages 18–24 months suited genotypes with more stalks than 12 months. The genotypes selected using optimum trait combinations produced 43% to 89% higher cane yield than those discarded. The high-yielding genotypes produced more stalks that were taller and thicker than the low-yielding genotypes. The highest yield advantage of 72% to 89% were attained in regions with lowest yield potential. The benefits from using trait combinations were highest in breeding programmes located in harsh agro-ecological regions.