Effects of Field Design Factors and Spatial Variability on Forage Yield Estimation.

Field-based agronomic and genetic research projects are subject to numerous externalities that impact efficiency of field trials. Modification of experimental designs factors (i.e. block size, plot size, block shape, and number of replicates) and postdictive spatial analysis can improve the precision and efficiency of cultivar-mean estimation. The objective of this experiment was to conduct a retrospective analysis of long-term perennial ryegrass (Lolium perenne L.), Italian ryegrass (Lolium multiflorum), and hybrid ryegrass (Lolium boucheanum) forage yield trials to (i) quantify the effect of experimental design factors on trial precision, (ii) predict the number of replicates, locations, and harvest years required to detect expected cultivar mean differences, and (iii) quantify the effect of spatial variability on cultivar-mean estimation. A total of 142 yield trials sown between 2001 and 2011 across five locations by the Department of Agriculture, Food and Marine (DAFM), Ireland, were considered for this analysis. Blocking was ineffective when long and narrow blocks were used. Results from total sum of squares of residuals indicated a decreasing trend in the contribution of row and column sum of squares (SS) with increasing block size from 6 (71% of total SS) to 33 cultivars/block (51% of total SS). Coefficient of heterogeneity (b) was sensitive to increasing block size and number of replicates, but the responses were different between square and rectangular blocks. For two harvest years, at least six replicates were required to attain power of 0.8 when the detection difference between cultivars was 10% of trial mean. Two-dimensional first order autoregressive (AR1) spatial models were significantly better than the randomized complete block design for 76% of the field trials.   Results from this retrospective analysis indicate that decreasing plot size in conjunction with two-dimensional incomplete block designs can buffer the effects of spatial variability in the DAFM trials.