- One of the most economically important genes in squash (Cucurbita spp.) is the B gene which conditions precocious depletion of chlorophyll in preanthesis fruit. The B genes are found naturally occurring in at least three of the five domesticated species of Cucurbita, including C. pepo, C. maxima and, most recently C. moschata. Additionally, there are reports of a fourth B gene in C. argyrosperma but no inheritance or allelism tests have been reported. In addition to improving fruit appearance, the B gene can dramatically increase flesh carotenoids such as β carotene, a precursor to vitamin A. The B gene in C. pepo (Bpepo) is well studied and has been introduced into commercial cultivars such as ‘Golden Zucchini’. The B of C. maxima (Bmax) is found in commercial cultivars, including the economically important processing squash, ‘Golden Delicious’. More recently, B was discovered in C. moschata (bmos) and has been crossed into a tromboncino background (C. moschata) at OSU. The three different sources of B have distinct differences in their expression depending on background and environment. To overcome fertility barriers and allow allelism tests between C. pepo and C. maxima, two sources of the B gene (Bpepo and Bmax) were transferred into C. moschata, which crosses readily with the other species. The tests for allelism that followed, revealed that Bpepo and Bmax are separate unlinked genes. A preliminary inheritance study found that the B of C. moschata differed from the incompletely dominant B of C. pepo and C. maxima in being recessive. Thus, it was believed to be non-allelic and therefore designated bmos. This was unexpected and warranted further investigation. The objectives of the present study were to evaluate the mode of inheritance, allelic relationships, and phenotypic characteristics among the three sources of B; Bpepo (B1), Bmax (B2) and bmos (b3), in new backgrounds of C. moschata. Parental (P), filial (F), backcross (BC), and test cross (TC) generations were evaluated for precocious pigmentation in fruit and other above ground vegetative organs. In agreement with the previous report, in crosses involving the dark green tromboncino parent, bmos fit a one gene 3:1 (green:yellow/bicolor) phenotypic ratio in the F2 generation (P = 0.55) and a 1:1 (green:yellow/bicolor ratio in the BC1 to the yellow/bicolor parent (P = 0.84). However, contrary to previous reports, the data did not follow Mendelian inheritance in wide crosses involving ‘Ayote’ (C. moschata) or ‘Green Striped Cushaw’ (C. argyrosperma) parents. A pattern of weak infrequent heterozygous expression was observed when crossing to the dark green tromboncino and strong frequent heterozygous expression when crossing to Ayote and Green Striped Cushaw. Thus, bmos was inherited as a single recessive gene in some backgrounds but not in others. Contrary to previous reports, results from inheritance tests of Bmax did not fit a one gene 3:1 (yellow:green) phenotypic ratio in F2 generations (P < 0.001) due to an excess of green segregants. However, they did fit a one gene 1:1 (yellow:green) ratio in the BC1 to the green parent (P = 0.44). Additionally, results from an inheritance test of Bpepo did not fit a 3:1 (yellow:green) ratio in the F2 generations (P < 0.001) due to an excess of green segregants. In allelism tests between bmos and Bpepo, we observed a one gene 1:0 (yellow:green) ratio in the F2 and BC1 to the bmos parent indicating that Bpepo and bmos are allelic while Bpepo is the top dominant. There was some question as to the allelic nature of the two Bpepo parents as they were obtained from different sources and the parent AC1-B was observed to have stronger expression than the other parent CN3-B. Additionally, a putative allele of B, Bw (weak B) has been reported in the literature and may have been crossed into C. moschata. We observed a one gene 1:0 (bicolor/yellow:green) ratio in the F2 and BC1 to the CN3-B parent indicating that the B of AC1-B and the B of CN3-B are allelic. Additionally, our results suggest that CN3-B may carry a more weakly expressed allele, but more tests are still needed to verify this. In allelism tests between bmos and Bmax the results in two out of three F2 generations fit a two gene 13:3 yellow:green ratio (P = 0.35). Additionally, the BC1 to the bmos parent fit a two gene 3:1 yellow:green ratio (P = 0.38) indicating that Bmax is non-allelic and epistatic to bmos. Interestingly, the F2 that deviated from the 13:3 model also had an excess of green plants resulting from a significant deficiency in the Bmax class. Finally, allelism tests indicate that Bmax is non-allelic and epistatic to Bpepo although a pattern of excess green segregants was again observed. In the F2 of test 1, two out of three crosses fit a two gene 12:3:1 ratio of -/- B2/- : B1/- B2+/B2+ : B1+/B1+ B2+/B2+ (P = 0.25-0.81). The F2 of test 1 that deviated from the model (P < 0.01), also had an excess of green plants resulting from a significant deficiency in the Bmax class. By contrast, the TC to the green parent did not fit a two gene 2:1:1 model of -/- B2/B2+ : B1/B1+ B2+/B2+ : B1+/B1+ B2+/B2+ (P < 0.01) due to a deficiency in the green plants from an excess in the Bpepo class. However, the BC1 to the Bpepo parent did fit a two gene 2:1:1 ratio of -/- B2/B2+: B1/B1 B2+/B2+: B1/B1+ B2+/B2+ (P = 0.78). The F2 of test 2, did not fit a two gene 12:3:1 ratio (P < 0.01) due to an excess in green plants resulting from a deficiency in the Bpepo class. The TC to the green parent also did not fit a two gene model, again, due to an excess of green segregants. However, test 2 used a different Bpepo parent that was found to have weaker more variable expression when crossed into new genetic backgrounds. Accordingly, the BC1 to the Bpepo parent produced a significant number of green segregants that were presumed to be non-expressing Bpepo heterozygotes. When these were placed into the B1/B1+ B2+/B2+ class the BC1 fit a 2:1:1 two gene ratio of (52 -/- B2/B2+ : 21 B1/B1 B2+/B2+ : 28 B1/B1+ B2+/B2+, χ2 = 1.06, P = 0.59). Finally, pooled data from the F2 of test 1 and 2 deviated significantly from a two gene 12:3:1 model (P < 0.001) due to an excess of green segregants. However, pooled data did fit the two gene 2:1:1 model (P = 0.49). Contrary to test 1, it was Bpepo and not Bmax that was deficient in most generations in test 2. The combination of allelism tests indicate that bmos and Bpepo are non-allelic and hypostatic to Bmax. Additionally, results indicate that bmos is a recessive allele of Bpepo and part of a multiallelic series. The order of dominance is (B > B+ > bmos). Although the results vary significantly in the allelism test between Bmax and Bpepo, when taken together, and examined in more detail, the best interpretation is in support of the two gene hypothesis. The results of this study raise questions about patterns of deviation from expected ratios and inheritance in certain genetic backgrounds. The deviations from two gene models did not indicate allelism but were generally the result of incomplete penetrance of either Bmax or Bpepo. The failure of Bmax or Bpepo to express always resulted in an excess of green plants. This tendency to deviate from expected ratios observed with the B genes was unexpected and should be investigated further. The results from this study help to clarify the relationships between B genes and alleles while adding to our understanding of their phenotypic expression in new genetic backgrounds. This knowledge should accelerate progress in the improvement of C. moschata in breeding programs.