133 Prediction of conventional wheat characteristics of hard winter wheats using single kernel parameters.

O. K. CHUNG, J. B. Ohm, and B. W. Seabourn. USDA, ARS, Grain Marketing and Production Research Center, Manhattan, KS 66502.

The single kernel characteristics of 2890 hard winter wheats that were collected from federal nurseries from 1990 to 1997 were obtained from Single Kernel Characterization System (SKCS). Flour yield showed a mean value of 68.8% with a standard deviation of 2.9. Test weight (TW), % large kernels (% LK), 1000 kernel weight (KWT), and near infrared hardness score (NIR-HS) showed mean values of 60 lbs/bu, 28.6 g, 67%, and 66 with standard deviations of 2.3, 4.7, 20.0, and 12, respectively. Single kernel weight obtained from SKCS showed significant correlation coefficients (r) with KWT (r = 0.923), % LK (r = 0.888), TW (r = 0.521) and flour yield (r = 0.384). SKCS hardness index had a significant correlation with NIR-HS (r = 0.557). To develop continuum regression prediction models of TW, % LK, KWT, NIR-HS and flour yield, eight SKCS characteristics and 12 machine parameters were used. Among 2890 wheats, 1200 and 300 wheats were selected as calibration and validation sets, respectively. The prediction model for flour yield showed an R-square of 0.643 and a root mean square error (RMSE) of 1.6 for the calibration set, and an R- square of 0.635 and an RMSE of 1.6 for the validation set. Prediction models of KWT, % LK, NIR-HS, and TW showed R-square values of 0.898, 0.888, 0.625, and 0.472 for the calibration set and 0.893, 0.888, 0.609, and 0.473 for the validation set, respectively.