PREDICTION OF SITE INDEX AND APPLE ROOTSTOCK PERFORMANCE FROM ENVIRONMENTAL VARIABLES

Previously, we developed stability analysis models for nine rootstocks tested over 19 apple producing states and demonstrated significant ro otstock-site interactions for cumulative yield (CY) and trunk cross-se ctional-area (TCSA). A key input in these models is site index (SI), e stimated from the mean over rootstocks within site. Our goal was to ex tend the usefulness of these models by developing further models to es timate SI for untested sites from climate variables. Prediction of SI from mean daily maximum temperature (T) and total moisture (M) (sum of precipitation and irrigation) was evaluated over five periods based o n previous work and on approximate phenological phases of the apple tr ee over the geographic area included in this study: A = December-Janua ry (dormant), B = February-April (prebloom), C = May-June (set), D = J uly-September (growth to harvest), and E = October-November (postharve st to leaf senescence), resulting in ten climate-time variables for mo del development. Complete records of T and M were available from 11 an d 9 states, respectively, SITCSA Was not significantly correlated with any T or M variable and therefore may have been influenced more by so il factors or deviations in orchard management practices than by clima te. SICY was well correlated with T-C, T-D and M(C), but T-C and M(C) were strongly collinear. Further model development concentrated on pre diction of SICY from T variables over data from ten states (with Calif ornia removed as an outlier). All one-, two-, and three-variable multi ple regression models were evaluated. We concluded that SICY was best predicted from a linear relationship with T-C.