Determination of wheat kernel morphological variation by digital image analysis: II. Variation in cultivars of soft white winter wheats

Abstract

Within the Eastern Canadian soft white winter whets (CEWW), variation in whole kernel morphology is difficult to detect by visual inspection although functional tests show that differences exist between cultivars. To evaluate the morphological variation with the soft white winter wheat cultivars, whole kernel, crease and bran features were measured quantitatively using a digital image analysis system. Whole kernel features alone were inadequate for correct cultivar classification while crease or bran features improved cultivar classification. The inclusion in the discriminant model of variables derived from combinations of whole kernel, crease and bran features improved correct kernel classification to better than 80 %. From a six-cultivar discriminant model, the classification of Augusta and Fredrick (from four test sites) demonstrated that environmental influences greatly affected kernel morphology and subsequent classification. However, in a two-way model, Augusta and Fredrick kernels were correctly identified irrespective of environmental influence.