Statistical methods for identifying traits associated with high yield potential in durum wheat under drought conditions

The knowledge of key traits contributing to stress tolerance could contribute to increasing the efficiency of the selection process leading to development of drought tolerant genotypes. Over three years (2010-13), 25 durum wheat genotypes were evaluated for different agro-physiological traits under drought conditions. Four statistical procedures including path analysis, stepwise regression, principal component analysis (PCA) and factor analysis (FA) were used to identify the most contributors to grain yield. A mean score index (NISI) based on scaling scores of selection criteria was used for genotypes characterization. The average yield productivity varied between 774 to 2360 kg/ha across years. The statistical procedures confirmed the chlorophyll fluorescence (Fv/Fm), spike length, SPAD reading, plant height, peduncle length and heading date as the most contributors to yield productivity in durum wheat. The methodology of scoring scale provided a simple and easy visualization and identification of resilient, productive and/or contrasting genotypes according to selection criteria. The PCA and FA by justifying the high portion of variability in yield were found to be more efficient for developing proper models for indirect selection.