MSc

The present study whose objective is to evaluate the problems concerned with field experimentation on cashew was undertaken using the yield records of 625 uniformly treated cashew trees collected from cashew research station, Madakathara for a period of five years starting from 1976. A single row of trees was discarded on either side of the experimental field to eliminate border effect. Thus the experimental material consisted of 576 trees raised from same parental stock in a 24 x 24 compact block arrangement. The period 1976 to 1980 for which there was no change in treatment was considered to be the pre-experimental period and the year following the application of treatment was considered to be the experimental period. Observations on certain yield contributing characters pertaining to the year 1978 to 1979 such as height, spread and trunk girth were also gathered.
Plots of different sizes and shapes were formed by combining yields of adjoining trees in various possible ways, a single tree being considered as the ultimate unit. The plots were grouped into blocks of different sizes and cv was worked out for plots of different dimensions when plots were arranged in blocks and not arranged.
The trees were found to be highly heterogenous even though they were raised from the same parental stock. Single tree plots were found to be the most efficient when viewed from the point of view of maximum relative percentage information and consequently could be recommended for conducting field experiments on cashew. Two tree plots could also be recommended due to certain practical considerations shape of plot did not seem to have a consistent effect on variability.
Fair field Smith’s equation gave a good fit to the data and the parameter ‘b’ was found to be high in both the cases when trees are arranged in blocks and when they are not arranged. It was observed that 2 plot blocks were the most efficient for conducting field experiments on cashew. The efficiency of blocking decreased with an increase in plot size.
The result showed that experimental error could be considerably reduced by the use of incomplete block designs. As an effective method of controlling tree to tree variation the method of ranking based on their past performance can be adopted. Efficiency of local control was considerably increased by ranking of trees as compared to mere grouping of trees with respect to their geographical contiguity. Pre-experimental year’s yield was found to be the most important yield contributing character. Trunk girth of tree was also found to have significant influence on the inherent yielding ability of the trees.
A selection index evolved was found to be strongly correlated with experimental yield. The correlation coefficient observed for this variate was higher than that for other yield contributing characters. Maximum efficiency in analysis of covariance was noted by using selection index as covariate. Efficiency of covariance analysis decreased with an increase in plot size. The optimum pre-experimental period for calibration and covariance analysis was found to be two year’s immediately prior to the start of the experiment.