Determination of wheat kernel morphological variation by digital image analysis: I. Variation in Eastern Canadian milling quality wheats

Twelve kernel morphological features of nine cultivars of milling quality wheats grown in Eastern Canada were determined by digital image analysis. Four cultivars of soft white winter (SWW; cookie/pastry quality) and five cultivars of bread making quality, three hard red spring (HRS) and two hard red winter (HRW) were examined. The HRW cultivars were grown in two different locations. Stepwise discriminant analysis selected seven variables which clustered cultivars by class and the HRS wheats by origin. A four-way linear discriminant model, separating bread and cookie quality wheats, correctly assigned (> 80 %) HRS and HRW classes and totally segregated the SWW wheats. Mis-classification of kernels between the HRS and HRW classes was due to the large variation in morphology within each sample, since no prior screening was applied to the samples. While kernels could be assigned to classes using external morphology, cultivars within the SWW class could not be clearly distinguished.