A METHOD TO ADJUST SIMULTANEOUSLY FOR SPATIAL MICROSITE AND COMPETITION EFFECTS

A patchy microsite mosaic gives rise to positive spatial correlations among first-order neighbors in field trials. Competition for a resource in short supply tends to enhance size differences among first-order neighbors. Simulation studies demonstrated that positive spatial autocorrelations will seriously inflate family variance components and, thus, the individual narrow sense heritability in sib trials with multiunit plots in randomized complete block designs. Moderate levels of competition between first-order neighbors had the effect of lowering this inflation. Results from simulated single-tree plot designs were never biased. In the absence of strong competition, this design type is clearly preferable. An effective method for adjusting results obtained from multiunit plot designs is presented. Adjusted values of family variances and heritabilities were within 6% of the true values in two of three simulations. Adjustments are based on an iterative nearest neighbor (NN) regression on family values and a NN adjustment of ANOVA residuals. An application of the NN adjustments to height and diameter data in a jack pine (Pinus banksiana Lamb.) sib test lowered heritabilities of height to the expected level and indicated similar levels of genetic control of the two traits.