Experimental design factors affecting error variation in orchardgrass

Several experimental design factors over which the researcher has control may affect the precision of field experiments. The objective of this experiment was to identify the effects of changing plot size, incomplete block size, and incomplete block shape on precision of forage yield and quality estimates in orchardgrass (Dactylis glomerata L.). Orchardgrass seed was planted uniformly into 288 plots of various sizes. Forage yield, in vitro dry matter digestibility (IVDMD), and neutral-detergent fiber (NDF) were determined over a 2-yr period. Error variances (V-w) were computed for variously sized incomplete blocks (in one or three strips and from 1 to 12 plots strip(-1)) within each plot size and expressed as log(V-w). Values of log(V-w) for IVDMD and NDF concentration were unaffected by experimental design factors. Values of log(V-w) for forage yield increased with increasing number of plots per strip, but these effects varied among plot sizes and for one- vs. three-strip blocks. Three-strip blocks had a mean log(V-w) 71% greater than one-strip blocks, an effect that could be offset only by doubling the number of replicates in a field test. To obtain maximum precision in Eases where a large number of replicates is impossible or impractical, orchardgrass held experiments should employ incomplete block designs, particularly when limited seed stocks or land requires a plot size near 1.4 m(2). Incomplete block designs with this plot size should not contain multiple strips within incomplete blocks.