A Proposed Method for Estimation of Genetic parameters on Forest Trees Without Raising Progeny: Critical Evaluation and Refinement

A method for estimation of genetic parameters of forest trees without raising progeny, the SAKAI and HATAKEYAMA method proposed in 1963, is evaluated. This method assumes that all variation not genetic in origin follows a spatial pattern, and that genetic variation is randomly distributed in space. It is applied here to simulated data and to a progeny trial of maritime pine (Pinus pinaster AIT.). Even though the method has been widely used, it has an inadequate theoretical foundation and seems rarely to give good estimates. With simulated data, the method failed to provide a reasonable estimate of genetic variance when the spatial trends in the data were weak. With data from the progeny trial, the method overestimated the genetic variance. A modification to the SAKAI and HATAKEYAMA method is proposed, using a modern approach to spatial analysis. The modified method has a better theoretical foundation, and gives slightly better estimates of genetic variance when applied to the simulated data. Nevertheless, the estimates only approached the true values when the number of genotypes was large and either the spatial autocorrelation in the error term was strong or the ratio of genetic to error variance was large. The modified method gave a worse estimate of genetic variance than the unmodified SAKAI and HATAKEYAMA method when applied to the progeny trial. It is concluded that the use of the SAKAI and HATAKEYAMA method, either its original form or the modified form, can rarely if ever be recommended. However, if limited information on the genetic relationships between trees is available, either from pedigrees or from molecular-genetic markers, more modern methods of spatial analysis should be of great value in the estimation of genetic parameters.