Objective:

To determine the size and shape of the experimental unit for yield trials of Brachiaria, hybrid CIAT 3608, in the Santa Cruz canton, Guanacaste, Costa Rica, through the application of the multiple regression method.

Materials and Methods:

A uniformity trial was planted and the obtained data were used to apply the multiple regression method. According to it, the residual variability was measured as variation coefficient, corresponding to the different simulated sizes and shapes of experimental unit, and modeling was done depending on length and width. Two multiple regression models were adjusted: one of them included the term of interaction among the regressors, and the other did not.

Results:

The model without the interaction term was the one with the best fit. It had higher R2 and all the estimated coefficients were significant (p < 0,05). Then, the partial derivatives were calculated with regards to the length and width of the regression equation, estimated by the model without interaction, and they were equal to -1. The system of resulting equations was solved and the combination of width and length of the experimental unit, which minimizes the variation coefficient, was obtained. For this work, this combination turned out to be 4,95 m of length and the width is 6,03 m.

Conclusions:

In the framework of the conditions under which this work was conducted, to obtain accurate results with the species Brachiaria hybrid CIAT 3608, one experimental unit of 5 X 6 m of length and width, respectively, is sufficient.

        · abstract in Spanish     · text in Spanish     · English ( pdf ) | Spanish ( pdf )