Multivariate statistical analysis of yield-determining factors

Abstract

This study is aimed at developing a multivariate statistical method to identify yield-determining factors and to evaluate their relative magnitude. Principal component analysis was used to summarize the rice growing environment in a “hydromorphic” field into a number of factors which may affect the yield performance. Four major factors were recognized; i) Soil moisture factor, ii) Soil texture factor, iii) Fertilizer factor, and iv) Insect damage factor.

Using the scores of the indicated factors as independent variables, stepwise regression analysis was performed to derive a yield prediction function. The coefficient of determination (R2) reached 0.924 and the magnitude of the factor effect on yield followed the order of Soil moisture≫Insect damage>Fertilizer. Soil texture was not significant as a yield-determining factor.

Although the current report is only a case study, the method developed should be useful for predicting yield potentials of plant varieties adapting to different soil environments.

Key Words:

• hydromorphic soils
• principal component analysis
• regression analysis
• yield-determining factor