Quantitative trait analysis of fruit quality in cucumber: QTL detection, confirmation, and comparison with mating-design variation

A cross within Cucumis sativus var. sativus (GY14 x P1432860) and molecular markers were used to determine the number, magnitudes of effect, and overall variation described for genes conditioning the quantitatively inherited traits of length, diameter, seed-cavity size, color, L/D (length/diameter), and S/D (seed-cavity size/diameter). QTL effects were detected with MAPMAKER/QTL using l00 F3 lines evaluated in a replicated field trial of two harvests over 2 years at one location. Multilocus models were constructed by fixing significant intervals and re-scanning using MAPMAKER/QTL. Marker inclusion in multilocus models was compared to an ANOVA "backward elimination" procedure. Generally the same loci were associated with QTLs among the two methods of model construction. Heritabilities of individual QTLs were confirmed by analysis of related backcrosses (67 BC1P1 lines and 68 BC1P2 lines). The majority of QTLs were confirmed in at least one backcross population. Pairs of backcrosses allowed overall additive variances and heritabilities to be calculated using a North Carolina Design 3 (NC3 design) and estimates were compared to overall variances attributable to markers. Heritability estimates using markers were comparable, but generally lower than additive variances estimated by co-variance relationships in the NC3 design. This suggests that neither the number nor the magnitude of QTL effects were over-estimated. The utility of backcrosses to confirm individual QTLs and the overall variance described by QTLs is recommended to avoid false positives and over-estimation of effects. The number of QTLs, and/or the proportions of phenotypic variation described by markers and the mating design, agreed with previous reports of heritabilities employing similar germplasm.