Submitted to: Journal of Animal Science
Publication Type: Abstract Only
Publication Acceptance Date: 3/24/2004
Publication Date: 7/26/2004
Citation: Seman, D.H., Stuedemann, J.A., Douglass, L.W. 2004. Number of experimental units required to detect differences in grazing time. Journal Of Animal Science. Interpretive Summary:
Technical Abstract: The objective of this study was to determine the number of cattle necessary to detect differences in time spent grazing by steers on E+ tall fescue. In each of two years, 12 yearling Angus steers grazed tall fescue pastures, three pastures were highly infested (64%)(Neotyphodium coenophialum) KY-31 fescue, while three had low infestation rates (32%). One steer on each pasture received an experimental drug to overcome effects of endophyte infection and one received a placebo on Monday, Wednesday, and Friday during the three-week (1988) and four-week (1989) study. All 12 steers were fitted with vibracorders to record continuous grazing behavior during the studies. Statistical analyses of time spent grazing/h was conducted by the Mixed Procedure of SAS with day as a repeated measure and pasture considered a random effect. Mixed Procedure calculated eight standard errors, one for each endophyte, drug and year combination and resulted in a range of values when estimating the number of experimental units. It would take from 14 to 115 steers/treatment to detect a 5% difference in minutes spent grazing/h á=.05. Up to 30 steers/treatment would be needed to detect a 10% difference, 7 to 15 to detect a 15% difference, 2 to 9 to detect a 20% difference, 4 to 6 for a 25% difference, and 2 to 5 to detect a 30% difference. These analyses point out the importance of calculating an accurate standard error with which to estimate experimental units. Box-Jenkins ARIMA time series analysis showed that minutes spent grazing/hour differed (P<.05) between days when steers were gathered and given drug and days when steers were left alone in 1988 and during week 1 in 1989. Grazing behavior at any given time of the day was correlated with the previous one or two hours. A significant 24-h seasonal component showed that grazing behavior repeated (P<.01)at the same time on subsequent days. Results suggest that it may not be necessary to monitor grazing behavior continuously day after day to obtain a representative sample. Data on the days when normal grazing behavior is disrupted, e.g., by gathering steers, may not be representative of time spent grazing throughout a given collection period.