Consideration of Cost and Accuracy for Trial Design of Advanced Apple Selections

Field evaluation of advanced apple selections is expensive and, with limited resources, the design of these trials is a trade-off between maximizing accuracy of identifying truly elite candidates and minimizing cost.  Here we investigate the effect of alternative trial designs on the cost and accuracy of fruit quality assessment using instrumental and sensory traits in the Washington State University Apple Breeding Program as a model.  Critical percentage difference, response to selection, and correlated response were used as measures of accuracy. The number of locations, years and harvests per year were decreased from the current design of three to either two or one. For most traits, the loss in accuracy from a reduced design was less than 5%. Reduction in the number of harvests resulted in the smallest loss in accuracy for all traits, but only a negligible decrease in program cost. Reducing both number of harvests and locations to two resulted in a greater loss in accuracy, but still relatively small, and would allow the program to evaluate twelve additional candidates per year for a similar total program cost. Overall, the total cost of the program could be reduced if the design is reduced, with minimal loss in accuracy and the additional capability to trial more selections. The methods employed in this analysis offer a framework for other tree fruit breeding programs to investigate their own trial design accuracy and efficiency questions.