Relationship between kernel size features and test weight in Triticum durum

Test weight or bulk density in durum wheat is a physical quality characteristic considered by semolina millers. High test weight values are desirable because they positively influence market grade and price. This study reports data on kernel size features, determined on a sample size of only 25 kernels, replicated three-times, of 16 commercial durum wheat cultivars grown in two locations in southern Italy, to ascertain whether some kernel traits could be related to test weight. For each year and cultivar, the analysis of variance for all of characteristics showed that sample size effect was not significant, enabling the use such a small sample for further investigations. The kernel trait with the highest variation for year was kernel width. The absolute variation of 1994 with respect to 1993 growing season was 30.9%. While kernel weight or volume did not correlate with test weight, a negative association with kernel length (r = -0.61, P = 0.05) and perimeter (r = -0.57, P = 0.05) was found. The kernel shape factors, rectangular aspect ratio (RAR) and circularity shape factor (CSF), showed a positive correlation with test weight (r = 0.51, P = 0.05 and r = 0.59, P = 0.05, respectively). The shape factors were negatively correlated both with kernel length and perimeter and positively with kernel width. A predictive model for test weight (TW = 38.7 + 5.1.ETW, where the estimated test weight [ETW] was computed as individual kernel weight/estimated kernel volume [EKV] ratio), was highly correlated with actual test weight values (r = 0.82, P = 0.001). The effectiveness of the linear model was confirmed when a set of 10 advanced lines of durum wheat were considered, although a slightly lower correlation (r = 0.73, P = 0.05) between actual and predicted test weight values was found. Because the extent of predictability of this approach might be more effective in early-generation lines, the application of the findings in a durum wheat breeding program would be advisable.