Use of edaphic variables to control experimental error: A case study on blocking and use of covariance

Field studies that extend over large areas are commonly blocked in an attempt to minimize experimental error. In an experiment that was designed to determine the effects of exposure to acidic rain and ozone on seedling loblolly pines through three years of growth, principal component and cluster analyses were conducted on the pretreatment soil chemistry of each plot in an attempt to make the plots in each block as homogeneous as possible. After one growing season, a series of edaphic measurements were taken in each plot to quantify soil drainage and aeration. Analyses of covariance were then conducted using the edaphic measures as covariates to determine their effectiveness in controlling experimental error. Even though the chosen blocking pattern was effective in reducing experimental error, the edaphic variable, MOTTLING (depth to mottling clay), was an effective covariate. MOTTLING reduced the experimental error by an additional 25 to 46%. Block sums of squares were also reduced by as much as 94% but still retained significance, suggesting that though there was a strong relationship between the blocking pattern and the MOTTLING pattern, blocking still accounted substantially for variability due to other unknown factors. © 1993 Williams and Wilkins.

Duke Scholars

Author Daniel D. Richter Earth and Climate Sciences