Of the American Society for Horticultural Science

Broccoli (Brassica oleracea L. var. italica Plenck) and caulifl ower (B. oleracea var. botrytis DC) are closely related botanical varieties. The underlying genetic bases of their phenotypic differences from each other are not well understood. A molecular genetic marker enabling B. oleracea germplasm curators and breeders to predict phenotype from seeds or seedlings would be a valuable tool. Mutant alleles at fl ower developmental pathway loci BoAP1-a, BoCAL-a, and glucosinolate biosynthetic pathway locus BoGSL-ELONG have been reported to be associated with a caulifl ower phenotype. We surveyed mutant alleles at these three loci in a genetically diverse sample of broccoli and caulifl ower accessions from the U.S. Department of Agriculture, Agricultural Research Service (USDA-ARS) Plant Genetic Resources Unit (PGRU) and the University of Warwick, Genetic Resources Unit of Warwick HRI (HRI). Phenotypic and genotypic data were collected for multiple plants per accession during two fi eld seasons. Simple genetic models assuming dominance or codominance of alleles were analyzed. Goodness-of-fi t tests rejected the null model that the mutant genotype was associated with a caulifl ower phenotype. A correlation analysis showed that BoAP1-a and BoCAL-a alleles or loci were signifi cantly correlated with phenotype but the fraction of variation explained was low, 4.4% to 6.3%. Adding BoGSL-ELONG to the analysis improved predictive power using the linear regression procedure, Maximum R-square Improvement (max R2). In the best three-variable model, only 24.8% of observed phenotypic variation was explained. Because tested genetic models did not hold robustly for the surveyed accessions, it is likely that there are multiple genetic mechanisms that infl uence whether the phenotype is broccoli or caulifl ower. Our results in commercial cultivars indicate that other genetic mechanisms are more important in determining the horticultural type than are BoAP1-a and BoCAL-a. Received for publication 16 Mar. 2006. Accepted for publication 21 June 2006. We thank Susan Sheffer and Paul Kisly for their excellent technical assistance, Drs. Heidi Schwaninger and Lori Hinze for helpful reviews of the manuscript, and Dr. David Astley for seed from the collection at Warwick HRI. 1To whom reprint requests should be addressed; e-mail: [email protected] The cole crops (Brassica oleracea) are characterized by diverse morphological forms that likely resulted from selection for various edible parts. Examples include, leaves that form a head [cabbage (B. oleracea var. capitata L.)], non-heading leafy types {kale and collard greens (B. oleracea var. acephala DC), chinese kale [B. oleracea var. alboglabra (L.H. Bailey) Musil]}, an enlarged stem [kohlrabi (B. oleracea var. gongylodes L.)], axillary buds {brussels sprouts [B. oleracea var. gemmifera (DC) Schultz]}, and immature infl orescences [broccoli (B. oleracea var. italica), caulifl ower (B. oleracea var. botrytis)]. Broccoli and caulifl ower are similar in that precociously large infl orescences (fl ower buds, pedicels, and peduncles in the case of broccoli) constitute the edible part of the plant. This has generated much confusion regarding their distinction in both scientifi c and popular literature (Kalia and Sharma, 2004). A robust defi nition was proposed by Gray (1982) based on the relative ontogeny of broccoli vs. caulifl ower at marketable maturity. Broccoli heads are a mass of fully differentiated fl ower buds, while caulifl ower crowns (known as curds) consist of proliferated fl oral meristems, about 90% of which abort prior to developing into buds (Gray, 1982). Arguing that the curding character of caulifl ower is under major gene control, two lines of evidence were presented by Crisp (1982). First, reproductive fi tness of caulifl ower is low compared to broccoli. The majority of fl oral meristems abort, and this excess undifferentiated tissue is particularly susceptible to attack by pathogens. Crisp asserted that it is diffi cult to imagine the curding trait evolving gradually under these conditions, but that it seemed more plausible that humans discovered a grossly mutated form that was then favored for seed production. Second, Crisp et al. (1975) discovered a single, dominant gene mutation in caulifl ower which gave rise to very small curds on vegetatively normal plants (i.e., a reproductive fi tness similar to broccoli). This mutation was thought to possibly represent a back mutation to an ancestral allele. In addition, although Crisp’s studies of crosses between broccoli and caulifl ower did not show evidence of a major gene effect in F1 or F2, selfs of F8 and F9 plants showed genetic evidence, albeit recessive, of a single locus difference responsible for curding (Crisp, 1982). The caulifl ower mutant phenotype in the genetic model Arabidopsis thaliana (L.) Heyn is reminiscent of the curding trait in B. oleracea var. botrytis, and results from combining recessive mutations in CAULIFLOWER (CAL) and APETELA1 (AP1) genes in the model species (Kempin et al., 1995). A nonsense mutation in the B. oleracea var. botrytis CAL homolog provided additional evidence that CAL and AP1 homologs BoCAL-a and BoAP1-a