Corrolations have been observed previously between locule number and fruit weight. However, a phenotypic correlation can be quite misleading in interpreting genetic phenomena. Therefore, consideration was given to genetic correlations obtained from the genetic components of variance and covariance. Such a correlation could be contributed to from two causal sources, viz., linkage and/or pleiotropy.
The genetic correlation between these two traits was measured in two populations. One was a segregating (backcross) population, the other a non-segregating population of the P1-F1 type of experimental set-up (Griffing, Genetics 35: 303-321. 1950). The latter involved six strains and all possible hybrids. (The strains involved were Red Currant, Gold Ball, Devon, Sterling Castle, Matchless, and Pan American.) In the non-segregating population, the genetic correlation observed would not be expected to involve any contribution due to linkage) if we can assume that the six strains involved are a random sample of all possible genotypes for these two traits. The magnitude of this correlation would be due solely to pleiotropic gene action. In the segregating material, on the other hand, both pleiotropy and linkage would contribute to the magnitude of the genetic correlation. Thus, a comparison of the two correlations should permit an evaluation of the relative magnitudes of linkage and pleiotropy as contributory causes.
The genetic correlations observed were (on the basis of logarithmic data): segregating population: rG = 0.69 +/- 0.07, non-segregating population: rG = 0.69. Thus, we may conclude that the genetic correlation of approximately 0.7 is the result of pleiotropy; that is, the two traits have genes in common.
It should be noted, parenthetically, that a difficulty exists in precisely defining pleiotropy as a distinct entity from exceedingly close linkage (Mather, 1949.)
A knowledge of the source and magnitude of such a correlation is of value in practical breeding work, where the two traits concerned are of interest. We know beforehand, e.g., that attempts to recombine the traits will not be very successful in that recombination due to crossing-over of linked genes will not be expected to occur. Conversely, if the traits are both desired, selection for one of them will suffice, as it will automatically tend to select for the other, thus saving expenditure of effort.