“…The analysis was performed in two steps; first, for each % suppression value the likelihood (predictive density) that a UV suppression value belonged to each genotypic group separately was computed, using the mean and standard deviation of the group; predictive densities were calculated as follows (Geisser 1964;Heston et al 1%5): E where z is the UV % suppression value, N1, RI, and Sl are the sample stze, mean, and standard deviation, respectively, for the genotypic group G,. ~Ihe predictive density for each UV suppression value was then combined with the a priori genetic probability of a segregating class to obtain the probability (posterior probability) that the UV-suppression value belonged to a specific genotypic group and segregating class; posterior probabilities were calculated as follows (Geisser 1964;Heston et al 1965): , 8~, c,) q,f (z/z,, 8,, c,) t where ql is the a priori probability of a segregation class and Eqi = 1. High and low UV % suppression cut-off values were set where the likelihood ratios of posterior probabilities were maximum for each genotypic group.…”