Citrullus lanatus (Thunb.) Matsum. & Nakai - Cucurbit Breeding ...

The possible gain from selection per cycle was predicted as h n
2
! " 2
( P)
multiplied by the selection
differential in standard deviation units k for selection intensities of 5%, 10%, or 20% (Hallauer and Miranda,
1988). The statistical analysis was performed using the SAS-STAT statistical package (SAS Institute, Cary,
North Carolina).
Results and Discussion
In our study, resistance to gummy stem blight in watermelon was not inherited as a single gene, as
previously described by Norton in PI 189225 (Table 1). The expected segregation ratios for the inheritance of
the db gene were not observed in the F 2 and backcross generations, when PI 189225 was crossed with the
susceptible 'NH Midget'. Similar results were obtained in greenhouse and field tests for the other three families
tested, involving PI 482283 and PI 526233 as resistant parents.
The lack of fit to the single gene hypothesis suggests that gummy stem blight resistance in watermelon
could be inherited as a quantitative trait locus (QTL). Most likely, multiple QTLs could be involved in the
complete expression of resistance. Nevertheless, the distribution of our F 2 data was strongly skewed towards
susceptibility (Fig. 1) and far from the expected bell-shaped (normal) distribution for quantitative traits. This
distribution pattern would suggest the presence either of a single gene or a QTL with high environmental
variation, or of QTLs regulating the expression level of a major gene.
A similar distribution was recorded in all four families, with the exception of the field test of the
family PI 526233 × 'Allsweet'. Higher variability in the field than in the greenhouse tests and low correlation
among tests is commonly found when screening for resistance to gummy stem blight (Gusmini and Wehner,
2002) and may be caused by differences in microclimate in the field.
In our analysis, the variances of the six generations tested were generally consistent across families.
Larger differences in variance estimates among families and within generation were found in the field test,
compared to the greenhouse test (Table 2). Genetic variance was larger than environmental variance in three of
the four crosses (Table 3). The larger environmental variance in the cross PI 482283 × 'Calhoun Gray' was
determined solely by the field test. A large genetic component was found also for this cross in the greenhouse
108