Path Analysis to Control The Multicollinearity among Yield Components In Maize (Zea mays L.) | ||
| Egyptian Journal of Agronomy test test test test test test long long test toooooooooooo longgggg | ||
| Article 6, Volume 34, Issue 2, December 2012, Pages 213-226 PDF (457.24 K) | ||
| Document Type: Original Article | ||
| DOI: 10.21608/agro.2012.101 | ||
| Abstract | ||
| A FIELD experiment was carried out at Mallawy Agricultural Research Station, Elmenia Governorate during 2010 and 2011 seasons to evaluate the performace of sixteen maize hybirds (11 single crosses and 5 three way crosses). Also, to use a modified model of path analysis to control the multicollinearity among yield components when studying the relationship between grain yield and its related characters.The experimental design used was randomized complete block with four replications. As a multiple linear regression model, the independence among the explanatory variables is an important assumption to do the path analysis characterized by validity and goodness of fit. But, this assumption is rarely satisfied because there are strong associations among yield components, which called the multicollinearity problems. Therefore, the present investigation introduced and evaluated a modified path analysis model that could deal and correct the negative effects of multicollinearity problem. Results appeared that the tested descriptive statistics for all studied characters were located at the statistically acceptable range. Highly significant and positive associations were observed between grain yield (ton ha-1) and each of days to 50 % tasseling, and silking, ear height, number of ears plot-1, ear length and number of kernels row-1. The Highest values of Variance Inflation Factor (VIF above 10) were recorded for most studied characters (7 out of 11 traits) using the coventional path analysis model indicating the presence of multicollinearity. Consequently , the path coefficients were flactuated recording very low values (close to be zero) and other inflated values (above 1). Also, there were some unexpected signs for some path coefficients such as the negative sign of the direct effect for number of kernels row-1. The previous negative effects were statistically enough to reject the normal model of path analysis. More accurate results were obtained using the modified model of path analysis because it can overcome the adverse effects of multicollinearity dilemma. The proposed model revealed that the traits of number of ears plot-1 and ear length exerted the greatest influence directly or indirectly upon grain yield indicating their importance as selection criteria in improvement of maize breeding programs. | ||
| Keywords | ||
| Maize; Multicollinearity; path analysis | ||
|
Statistics Article View: 789 PDF Download: 832 |
||
