A method for the determination of optimum plot size in experiments with rubber trees (Hevea)
1987
Rossetti, A. G. | Gomes, F. P.
On the account of scarcity of information about plot size and the great variability of genetic material, large plots are usual in experiments with rubber trees (Hevea). This paper studies data from four experiments carried out in the Amazon Basin (Brazil) by a method that uses the intraclass coefficient of correlation among trees within plots, takes into account the guard rows, and defines as optimum plot size the number of test trees per plot, which minimises the estimate of variance of a treatment mean, for a fixed total number of trees per treatment. The first experiment (in the State of Para) provided data on yield (averages of three years). The second one (in the State of Amazonas) furnished data on yield too. The other two experiments (in the State of Amazonas) furnished data on yield by microtapping (HMM), number of latex vessel rings, diameter of latex vessels, and stem girth at 1.30 m from the graft. The study leads to the following recommendations: for plots with a complete guard row: two test rows, with six to twelve trees, or three test rows, with nine to eighteen trees, or else four test rows, with sixteen to twenty trees. For plots with one half guard row: two test rows, with four to ten trees, or three test rows, with six to fifteen trees, or else four rows with sixteen to twenty test trees. When small plots are used, it is necessary to increase the number of replications. But this can be obtained even with a considerable reduction of area, and without loss of precision.
显示更多 [+] 显示较少 [-]