The use of repeated measurements analysis for the evaluation of seasonal variation in the dry matter yield and nutritive value of perennial ryegrass (Lolium perenne L.) cultivars

Abstract

Summary. Seasonal variation in herbage yield, and in the concentrations of water-soluble carbohydrates, crude protein, in vitro dry matter digestibility, neutral detergent fibre, acid detergent fibre and lignin in the herbage was measured for 5 perennial ryegrass (Lolium perenne L.) cultivars grown in 3 contrasting environments. Two separate trials were sown in one of the environments. The yield and nutritive value data were subjected to antedependence analyses to detect the presence of serial correlations between data from different harvest dates. Significant (P<0.05) antedependence was found in 2, 3, 3 and 6 of the 7 variables in the 4 experiments respectively. It was shown that serial correlations between harvest dates had the potential to cause the overestimation of the significance of cultivar effects in ANOVA of data from individual harvest dates, and in harvest and cultivar × harvest effects in combined ANOVA across all harvest dates unless statistical procedures appropriate for use with repeated measurements data were adopted.