|Mcmullen, Michael - IOWA STATE UNIV|
Submitted to: Crop Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: July 30, 2007
Publication Date: January 10, 2008
Citation: Doehlert, D.C., Jannink, J., Mcmullen, M.S. 2008. Size distributions of different orders of kernels within the oat spikelet. Crop Science. 48:298-340 Interpretive Summary: Oat kernel size is of importance to the oat milling industry because different sized kernels require different amounts of energy to dehull. We have analyzed distributions of oat kernel sizes using digital image analysis and found that oat kernel sizes seem to be distributed among two distinctive groups, which we call a bimodal distribution. In the field, oats develop within a structure called a spikelet. Spikelets generally contain two kernels. The primary kernel of this spikelet is always larger than the secondary kernel. In this study, we dissected oat kernel spikelets and demonstrated that the source of the bimodal distribution of oat kernels is the two kernel oat spikelet. We have developed a computer program to analyze the bimodal size distributions and have tested applications of this program for the selection of more uniformly sized kernels. We conclude that varieties with large kernels also have the most kernel size variation.
Technical Abstract: Oat kernel size uniformity is of interest to the oat milling industry because of the importance of kernel size in the dehulling process. Previous studies have indicated that oat kernel size distributions fit a bimodal better than a normal distribution. Here we have demonstrated by spikelet dissection and digital image analysis that the source of the bimodal distribution is the two-kernel spikelet of the oat panicle. Primary kernels from these spikelets form the larger kernel size subpopulation and the secondary kernels make up a subpopulation of smaller sized kernels. Kernels from single and triple kernel spikelets cause departures from bimodal distributions. Calculation of size at the modes by statistical analysis of digital images provides good, but not exact, estimations of primary and secondary kernel sizes. Differences in size between these two kernel types, kernel area variance, and coefficient of variation all provide useful information for the evaluation of variation in kernel size, although the bimodal information does provide a more accurate description of the actual kernel distributions. Low test weight may generate inconsistencies in kernel size evaluations according to kernel mass and to linear dimensions. Selection for oat with uniformly large kernels may prove difficult because large kernels were associated with the greatest amount of variation.