(go to KOMUGI Home) (go to WIS List) (go to NO.91 Contents)

Results and discussion

Although GI was measured only G% data are presented here. The two measurements were closely associated. Correlation coefficients between G% and GI ranged from 0.85 to 0.98 (P<0.001) among F4 progeny in 1989 and 0.77 to 0.99 (P<0.001) among F5 progeny in 1993. Hence using GI was unwarranted.

The four parents showed. distinctly different (P<0.05) levels of dormancy when their G% values were combined over the five 1993 population sets. The overall G% means of LS, CC, BV and Gr were 11, 23, 39 and 90%, respectively. Among the five population sets, LS had G% means lower (P<0.05) than CC in all but one set where they were equal (P>0.05); LS had lower G% means than BV in all five sets. The G% values of CC were lower than BV in four sets and equal to BV in one set.

Heritability estimates for G% were moderate (0.30 to 0.54) based on the standard unit method and low to moderately high (0.17 to 0.63) based on the regression method (Table 1). Among crosses of dormant parents with nondormant GR, h2 estimates were generally higher for crosses with LS and CC than for BV .Among crosses between dormant parents, the LS/CC cross had low heritability especially based on the regression method. These grain dormancy h2 values are similar to those obtained by Upadhyay and Pgulsen (1988), and by Allan (1993). Paterson and Sorrells (1990) obtained low regression W estimates m a cross between a nondormant parent and CC.

The G% distribution patterns of F5 progeny of crosses between the dormant parents with GR bore some similarities. None followed normal distribution (P<0.001). Rather they were skewed toward nondormancy. Very few progeny had, G% means similar to their dormant parents. Among progenies of BV/GR, CC/GR, and LS/GR populations, 3, 1 and 2% had G% means similar to their respective dormant parent (P>0.05) versus 77, 73, and 55% of the progenies with G% means comparable to GR (Table 2).

The G% values of progenies of crosses between dormant by dormant parents also were not normally distributed (P<0.01). Progenies with high G% values were not recovered in the LS/CC populations. About 12% of the progeny had G% values higher than LS while 98% of the progeny had G% values similar (P>0.05) to CC (Table 2). Apparently the dormancy traits of LS and CC have some genetic similarities. A few progeny (4%) of the LS/BV population had higher G% means than BV (P<0.05). Yet no progeny of LS/ BV and LS/CC populations had high G% values similar to GR suggesting that the three dormant parents had some genes in common controlling grain dormancy.

Earlier BV and CC were reported to have dissimilar genetic mechanisms controlling dormancy (Allan 1993). In that study GI was used to assess dormancy of F3 and F4 lines of a BV/CC population. The same conclusion was reached based on the distribution of G% values of F4 progeny of this cross. Over 20% of the progeny had G% values lower than BV and CC while 12% of the progeny had G% values greater than both parents (Table 2).

Although it is likely that CC and BV differ genetically for grain dormancy, they may not have additional dormancy genes to those occurring in LS. Anderson et al. (1993) also did not recover segregants that transgressed both parents for PHS resistance in a cross between CC and a selection with moderate PHS resistance.

Among the three sources of dormancy, LS offers the most breeding potential because it expressed the highest level of phenotypic grain dormancy. Progeny numbers greater than studied here should be screened to recover an adequate proportion of selections with the LS dormancy phenotype.

<--Back | -->Next

(go to KOMUGI Home) (go to WIS List) (go to NO.91 Contents)