The use of repeated-measurements analysis to describe the overwintering of white clover cultivars

An experiment was set up to investigate the morphological and physiological changes occurring in two white clover (Trifolium repens) (clover) varieties, AberHerald and Huia, growing in binary mixture with two perennial ryegrass (Lolium perenne) (grass) varieties, Preference and Ba 10761. Measurements were made on four occasions during winter/spring 1993–94. The statistical procedure of antedependence analysis was used to check for the existence of serial correlations among data from the different sampling dates. Evidence of antedependence was found in only three of the seventeen variables measured, and an order of antedependence for each of these was estimated. The antedependence structure providing the best fit to the data was then used in an analysis of covariance on the data set for that variable. The rest of the seventeen variables could be safely analysed by a split-plot analysis of variance for each sampling date separately. This result validated the sampling protocol used in the experiment.