1. A large-scale grazing management study comparing rotational grazing and continuous grazing with dairy cows at two stocking rates over four complete production seasons is described.2. The four treatments were: (i) controlled grazing, light stocking rate; (ii) controlled grazing, heavy stocking rate; (iii) uncontrolled grazing, light stocking rate; (iv) uncontrolled grazing, heavy stocking rate.Each treatment involved 40 cows for a first 2-year phase and 42 cows for the following 2 years. Each herd had a normal age distribution pattern and seven 2-year-old first lactation heifers (17% of total herd) were introduced each year to maintain this pattern.3. Stocking rate was the more important factor affecting the efficiency of pasture utilization as measured by per acre output of milk and butterfat. In general, high stocking was associated with higher outputs per acre despite lower yields per animal.4. Grazing method was of less importance. In general, controlled rotational grazing was superior to uncontrolled continuous grazing, both per animal and per acre, but the average influence even of these extremes of management was only half that of stocking rate.5. Significant interactions between stocking rate and grazing method existed. Under continuous grazing a point was reached where production per acre declined to the vanishing point with increased stocking rate due to excessive depression of per cow yield: this point was not reached under rotational grazing at the same high stocking levels.6. The results suggest that optimum stocking rate under rotational grazing occurs at a level some 5–10% higher than under continuous grazing. A depression of 10–12% in per cow yield, compared with more lenient grazing, corresponds with optimum stocking level irrespective of the grazing system. This estimate is suggested as a guide line in applying the principles involved.

Animal production in grazing experiments depends both on treatment and on stocking rate. Production at optimum stocking rate is the only valid measure of treatment effect and treatment comparisons should be based on this criterion. Since optimum stocking rates are unknown these comparisons cannot be made directly, instead they must be inferred from the results obtained with the stocking rates actually used. This requires that some form of production stocking rate relationship be postulated. This paper suggests a form for this relationship and considers its implications on experimental design and particularly on the number of stocking rates per treatment.

SUMMARYConniffe, Browne & Walshe (1970) proposed a form of experimental design and statistical analysis for grazing experiments. The original paper did not include any example of the analysis on real data. This note presents such an example and makes some comments on the statistical validity of the method of analysis.

No abstract

No abstract