Evaluation of parental-type classification for the inbred-backcross method of estimating gene number

An attempt was made to estimate gene numbers for four quantitative fruit traits in tomato using the inbred-backcross method. This led to a critical review of the method as used in publications to date. Estimation of gene number via the inbred-backcross method depends on the classification of lines into parental and nonparental types. Examination of prior publications shows that in several papers the classification of lines was done using tolerance intervals or ellipses rather than confidence regions. This would result in an underestimation of the gene number. No genes could be detected for any of the tomato fruit traits assayed except first color date using previous methods of line classification. For reasons discussed, we suggest use of confidence rectangles to classify lines as being parental and nonparental. Using this method, 14, 20, 25, and 30 genes were obtained, respectively, for fruit firmness, first color date, soluble solids, and fruit weight. For reasons discussed, these could be underestimations of actual numbers of genes affecting these traits. This study has implications concerning the precision of gene number estimates by other methods. Estimates of gene number similar to those previously published for the inbred-backcross method have been obtained using other methods, suggesting that the alternative methods may also be inaccurate