Simultaneous selection for multiple traits in sugarcane breeding populations using logistic regression models

Abstract

Simultaneous selection for multiple traits in plant breeding increases the value of cultivars to growers. Visual selection is inefficient while multivariate statistical analysis is too complicated for early stages. The objective of this study was to demonstrate logistic regression models as a statistical tool for simultaneous selection of several traits in sugarcane breeding. Data for stalk number, height, diameter, sucrose and fibre contents, and number of Eldana saccharina Walker (Lepidoptera) bored stalks were collected from BSL13 and SSL13 populations grown in the Midlands region of KwaZulu-Natal, South Africa. Data were analysed using the logistic procedure of SAS. Coefficients for stalk number, height, diameter and number of eldana bored stalks were highly significant (p < 0.0001), which indicated their influence in selecting for high cane yield. Selected genotypes in BSL13 produced 20% more stalks that were 23% taller and 7% thicker than those discarded. For SSL13, selected genotypes produced 22% more stalks that were 21% taller and 10% thicker. The selected genotypes produced 66% higher cane yield for BSL13 and 74% higher cane yield for SSL13. Eldana damage in selected genotypes was 37% lower for SSL13 and 14% lower for BSL13. The logistic regression models can be applied for simultaneous selection in several crops.