Agronomy and soils: Effect of sample size on cotton plant mapping analysis and results
Additional Document Info
Although cotton plant mapping has been valuable in understanding growth and development, variation in fruit distribution among plants is a significant mapping challenge. Choosing a sample that is large enough to generate useful information, but small enough to minimize time and resources, can make plant mapping more accessible for evaluating cotton crop growth characteristics throughout the cotton belt. The purpose of this research was to identify the effects of sample size and main-stem node grouping on sample variability. Plants were sampled in 10-m sections from six cotton cultivars at five locations in Georgia in 2009 and one location in 2010. The relative errors associated with sample sizes of one to 50 plants, as well as the statistical power associated with each sample size, were computed. On average, 37 plants per cultivar among five cultivars were required to reach a statistical power of 0.90, with the required number based on the magnitude of difference between cultivars in the fraction of plants having a boll at a given fruiting site. Grouping of main-stem nodes and the use of moving weighted averages decreased the error on a node-by-node basis. The use of these methods resulted in the loss of some node-by-node information that might be of value in particular cases, but the number of plants required to generate the same statistical power and standard deviation was decreased from a mean of 37 to a mean of 19 plants. These techniques should allow the use of smaller plant samples and make plant mapping more accessible. The Cotton Foundation 2011.