Title: Small Sample Techniques to Evaluate Cotton Variety Trials Author
Submitted to: Journal of Cotton Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: August 13, 2007
Publication Date: March 1, 2008
Citation: Boykin Jr, J.C. 2008. Small Sample Techniques to Evaluate Cotton Variety Trials. Journal of Cotton Science. 12: 16-32. Interpretive Summary: Approximately 20 million bales of cotton are produced in the U.S. annually with about 75% exported. It is important that this cotton is processed efficiently while maintaining the quality demanded by domestic and foreign consumers. New varieties are developed each year with improved fiber properties and increased yield. As these varieties are developed and marketed, it is important to accurately characterize fiber yield and other fiber quality parameters in tests designed for variety selection. Most cotton variety tests are limited to small plots, and sampling methods are limited by the size of these plots. Small samples are either gathered by hand or machine to test in laboratory gins. It is important that the results of these tests agree with actual crop results. This paper presents results of a test which compared small hand-picked samples taken randomly from plots (boll samples) and small samples from the machine picker (grab samples) to whole-plot machine harvested samples. Small samples were ginned on a small laboratory gin, while the machine harvested samples were ginned through a conventional ginning sequence. Lint yield and other fiber quality parameters were determined for each sample method. The differences observed between the sample methods reflect both differences in picking and ginning. There was a good relationship between methods for all properties other than leaf, but care should be used in predicting conventional results with small sample methods. In all cases (except leaf), the overall differences in cultivars were stronger than the inconsistencies between sample methods, so these small sample methods should continue to be a useful tool to predict fiber quality and gin turnout when conventional machinery is not practical or unavailable.
Technical Abstract: Cotton from the Mississippi Regional Cotton Variety Trials was evaluated for differences in gin turnout and High Volume Instrument (HVI) fiber properties. The objective was to determine if cultivar differences after conventional processing were predicted with two small sample techniques. The microgin sample, which represented conventional processing, included entire plots of cotton spindle picked and processed through a typical sequence of gin machinery including drying and cleaning. The grab sample was collected by hand from the picker, and the boll sample was picked by hand from the plant. Grab samples and boll samples were ginned on a smaller 10-saw gin without cleaning machinery. For each cotton property, a statistical model was developed to explain variation due to cultivar, sample method, and other factors; and the interaction between cultivar and sample method was analyzed. The most important finding was that cultivars tended to compare differently, in some cases, depending on the sample method. This was revealed statistically when the interaction between cultivar and sample was significant. The interaction was significant for leaf and reflectance when grab or boll sample data sets were included with the microgin data set. The interaction was significant for length and uniformity when the boll sample data set was included with the microgin data set. The interaction was significant for gin turnout when the grab sample data set was included with the microgin data set. In all other cases, the small sample methods revealed cultivar differences similar to those found in the microgin. For micronaire, strength, and yellowness, relative differences in cultivars determined in the microgin were the same as those found with boll and grab samples.