The study of relationship among traits is essential in assessing the feasibility of joint selection for two or more traits. The path coefficient analysis partitions correlation coefficient into direct and indirect contributions of various traits towards dependent variable thus guiding effective selection. A study was conducted to determine the correlation and path coefficients for sorghum grain yield and 12 quantitative traits among 47 sorghum cultivars collected from southwestern highlands of Uganda. The study was conducted at Kachwekano research farm in Kabale District at an altitude of 2,223 metres above sea level, during sorghum growing season of December 2014 to August 2015. Correlation studies revealed significant trait associations (p<0.001) for grain yield with panicle weight (r=0.938), panicle width (r=0.619), stem girth (r= 0.674), and leaf width (r=0.576). Grain yield revealed a non-significant weak negative correlation with days to 50% flowering (r=-0.011). Regression analysis revealed that panicle weight and plant height had the highest direct effect on grain yield of 1.085 and 0.2097 respectively. Path coefficient analysis revealed that stem girth (0.814), leaf width (0.74901) and panicle width (0.713) had the highest indirect contributions to grain yield. Residual effect was low (R=0.11) with high adjusted R2 value (0.89).

Keywords: correlations, path coefficients, regression, selection, sorghum