APPLICATION OF MULTIVARIATE STATISTICAL ANALYSIS FOR BREEDING STRATEGIES OF SPRING SAFFLOWER (Carthamus tinctorius L.)

Abstract

This study aimed to assess oil yield components and their interrelationships of spring safflower lines and varieties by using different statistical techniques to increase the oil yield in safflower breeding program. Field experiments were conducted at the Transitional Zone Agricultural Research Institute in Eskisehir, Turkey during 2014, 2015 and 2016. Correlation, simple linear regression, stepwise multiple regression, path, principal component and cluster analyze were used to investigate the relationships between oil yield and some components in spring safflower. The results revealed that characters affecting oil yield, which is important to determine selection criteria in plant breeding, vary according to statistical methods. Therefore, to obtain reliable result, it is essential to use multivariate statistical methods for scanning significant characters in studied material. According to the numbers of common characters determined in different statistical analyzes; oil content, seed weight, seed yield and number of head per plant would be important selection criteria for improved oil yield in the breeding material studied. The lines and varieties may be used in hybridization program and their hybrids may yield more transgressive sergeants for these characters for oil yield improvement.