|Robbins, Matthew - UNIV OF WISC|
Submitted to: Gordon Research Conference Proceedings
Publication Type: Abstract Only
Publication Acceptance Date: December 15, 2004
Publication Date: February 13, 2005
Citation: Robbins, M.D., Staub, J.E. 2005. Comparing selection by molecular markers to phenotypic selection for multiple, quantitatively inherited traits related to yield in cucumber. Gordon Research Conference Proceedings. p. 15. Technical Abstract: Cucumber (Cucumus sativus L.) is an important vegetable crop whose yield has reached a plateau in the last 15 years. Factors such as fruit-set inhibition, source-sink relationships, a narrow genetic base, and negative correlations among yield-related traits must be overcome to increase yields. Such correlations exit among multiple lateral branches (MLB), gynoecious sex expression, earliness, and fruit length to diameter ratio (LD) which are four important yield-related traits. All of these characters are under genetic control from relatively few quantitative trait loci (QTL; 2-6 for each trait), and have been placed on a moderately saturated molecular linkage map. Molecular markers from this map have previously been used to demonstrate that selection for MLB by markers alone is as effective as phenotypic selection in backcross breeding. In an effort to increase yield in cucumber, we designed a recurrent selection program utilizing two independent selection methods, selection by molecular markers only (MAS) and phenotypic selection (PHE). Four inbred lines, which complemented each other for MLB, gynoecy, earliness and LD, were intermated and maternally bulked to create four base populations. These four populations independently underwent intrapopulation, half-sib recurrent selection by two separate methods (MAS and PHE), as well as a random mating control (RAN), each over three cycles (400 plants/population in cycle 1 and 600 plants per population in cycles 2 and 3). In each method, selection was performed simultaneously for the highly branched, gynoecious, early, and long fruited (high LD) ideotype. Phenotypic selections were made visually based on the combination of all four traits at the whole plant level. Selections based on markers contained the highest number of desired marker genotypes from an array of 17 marker loci. Using the same selection scheme and selection intensity (plants selected:population size) of 20:400, 20:600, and 20:600 for cycles 1, 2, and 3, respectively in both methods allowed us to directly compare the effectiveness and efficiency of MAS to PHE for multiple, quantitatively inherited traits. The most effective selection method varied by trait and between populations. Using population 1 as an example, selection for MLB and mean LD over 3 harvests was equally effective using both MAS (3.53 branches/plant; 3.02 LD) and PHE (3.56 branches/plant; 2.99 LD), which were both superior to RAN (2.78 branches/plant; 2.78 LD). For earliness and gynoecy, MAS (41.8 days to anthesis; 26.6% gynoecious) was less effective than both PHE (40.5 days to anthesis; 81.8% gynoecious) and RAN ( 41.0 days to anthesis; 80.9% gynoecious), which were equal. Although fruit per plant was not directly selected, RAN (1.90 fruit/plant) and MAS (1.88 fruit/plant) were equal, but less than PHE (2.15 fruit per plant). Data over populations suggest that the value of MAS depends on genotypic background and numbers of selection cycles. The development of effective selection strategies, therefore, requires an understanding of phenotypic characteristics (source-sink relationships, heritabilities, and number of genes conditioning target traits) in cucumber.