Bergh, B. O.
A study was made of F1, F2, F3 and backcross progenies derived from the hybridization of two parental lines. One of these lines had small, two-loculed, prolate-shaped fruits while the other had much larger, many-loculed, oblate-shaped fruits. In addition, the two parents introduced marker genes linked with major known fruit-size loci. It was found that in this cross there was segregation at two such loci. Linkage studies indicated that the two loci were probably o and f.
The effects of segregation at what is assumed to be o and f appeared to be qualitatively very similar; it was concluded that in both cases the genotypes determining larger fruits probably also determined a larger number of fruit locules and a more oblate shape. Hence, in these progenies fruit size, locule number, and shape index (equatorial diameter/polar diameter) all increased together.
Locule number was found to be the most satisfactory character for indicating segregation at o and at f. Therefore it alone was measured as the index character for more detailed investigation. A logarithmic transformation of the locule number data was shown to be desirable in order to reduce metrical bias. Segregation at f was found to account for about 50 per cent of the locule number variation among these progenies, while segregation at o accounted for about 30 per cent of such variation. The remaining variation is presumably due to minor genes. Evidence for the segregation of such minor genes was noted.
There were no indications of epistasis involving the two major segregating loci, but evidence was found for epistatic interactions involving the minor genes.
Dominance at f was considerable for few locules, while at o there was almost no dominance from the arithmetic data and a little dominance for many locules following the logarithmic transformation. Potence as calculated from the non-segregating generations (the two parents and the F1) agreed closely with potence calculated only from the dominance relationships at o and f. Average dominance of the segregating genes was calculated by two methods. The first of these had been introduced by Fisher, Immer and Tedin (1932). From the data of the present progenies, this method resulted in highly variable and statistically unreliable values. The second method was suggested by Comstock and Robinson (1952). Applied to the data of these progenies this method appeared to be an efficient way of estimating average dominance. The values obtained agreed closely with analogous values calculated from the dominance ratios estimated earlier for o and for f.