Application of Trend Analysis to Horticultural Field Trials1

Abstract

Experimental results from horticultural field trials are obscured by the effect of systematic variation. This variation is directly related to the position of the plot in the field and is referred to as a fertility gradient(s). Trend analysis eliminates the effect of fertility gradients by fitting a polynomial regression equation (response surface model) to the systematic variability in the experimental units. Two cultivar trials of potato (Solanum tuberosum L.) conducted to compare results from trend analysis with that using the standard design analysis indicated that fertility gradients existed in the fields and were of a form that could be adequately fitted by a response surface model. A 3-dimensional plot of the response surface model indicated that the fertility gradients formed a very complex surface which could not be eliminated by experimental design. Of the 3 experimental designs used, the Latin square was the most efficient while the completely random was the least efficient. Trend analysis resulted in a large gain in relative efficiency over the standard analyses of completely random and randomized block designs. It also resulted in a substantial gain over that of a Latin square design. Adjusting the means using a response surface model in trend analysis also improved treatment estimates. Tests of significance using adjusted means were more precise and easier to interpret. Trend analysis proved to be the most efficient way to analyze the data, regardless of the experimental design used.

On the evaluation of the carbon dioxide solubility in polymers using gene expression programming

“…Trend analysis is a well-known applicable technique to visualize the output variation with the change of input variables 52 , 53 . The predictions of the CO 2 solubility correlations are depicted versus temperature and pressure in Fig.…”