Four inbred lines of maize (Zea mays L.) were selected from 3 previous eval
uation for resistance to tropical rust caused by Physopella zeae. Lines L42
6 and L82 were selected as resistant and the lines L47 and L37 as susceptib
le ones. Lines were crossed in a diallel scheme and six sets were obtained.
Each set comprised the two parents (P-1 and P-2) and derived generations,
F-1, F-2 and backcrosses (B-1 and B-2) to both parents, and two hybrid chec
ks, AG 951 (resistant) and P 3210 (susceptible). The experimental design wa
s in randomized complete blocks, in a split-plot arrangement with three rep
lications. The sets were allocated in the whole plots and generations withi
n sets (including checks) in the subplots. The severity of the disease was
evaluated by a discrete scale of 1 to 9, representing the percentage of the
whole-infected area of the plant. Generation mean analysis indicated that
additive effects are the most important and significant source of variation
in contrasting crosses. The dominance effects were less expressive and sig
nificant in only one instance in this study; they tended to be positive (do
minance for susceptibility) in contrasting crosses. Deviations from the mod
el were significant (P<0.05) in only one instance. General results lead to
conclude that crosses between contrasting lines tend to be nearly intermedi
ate between parents, but dominance deviation also may occur.