Justification for Grazing Intensity Experiments: Analysing and Interpreting Grazing Data

Publisher Website
Google Scholar
Abstract
Grazing trials in which treatments are compared at only 1 grazing intensity greatly outnumber those in which treatments are compared at several grazing intensities. This suggests that, compared to other treatments and the need for replication in grazing trials, researchers consider grazing intensity lower in priority. In this study, a regression modeling approach for analyzing and interpreting data was developed to enhance the value of grazing intensity trials. As an example, results from 5 irrigated bermuda grasses (Cynodon dactylon (L.) Pers) (Callie, Coastal, Brazos and experimental hybrids S-54 and S-16) which were continuously grazed without field replication by Santa Gertrudis steers at 4 grazing intensities were considered. The relationships between average daily gain (ADG) and stocking rate, ADG and herbage present (Mg/ha), and between stocking rate (animals/ha) and herbage present were well described by linear functions for all cultivars, with correlation coefficients (r) mostly above 0.9. Coefficients of determination (R2) for linear regression models derived for ADG vs stocking rate, ADG vs herbage present, and stocking rate vs herbage present were 0.90**, 0.89**, and 0.87**, respectively. Significant cultivar .times. grazing intensity (as measured by stocking rate or herbage present) interactions (P .ltoreq. 0.01) were observed. Furthermore, estimated stocking rates which provided maximum gain/ha ranged from 6.6 to 9.4 animals/ha, and the range in herbage present which provided maximum gain/ha was 0.35 to 1.95 Mg/ha. Callie provided an estimated maximum gain/ha of 881 kg/ha/season, while maximum gain/ha for the other cultivars ranged from 613 to 687 kg/ha/season. Comparison between these 5 cultivars at only 1 grazing intensity would have had very narrow application. The procedure described allowed statistical comparison of cultivars without replication, and inferences about the separate effects of forage quality and quantity on animal performance could be made. Herbage present and cultivar were descriptors of the pasture. Since there was a substantial range of values for herbage present and stocking rate, all important assumptions underlying linear regression were met and designs utilized in analysis of variance were not needed.