|dc.description.abstract||Inheritance studies have previously been undertaken at the South African Sugar Association
Experiment Station (SASEX) under irrigated conditions. Since most sugarcane is grown in
South Africa under dryland (raingrown) conditions, heritability estimates were calculated under
these conditions in this study and compared to those previously obtained under irrigated
conditions. A sugarcane population consisting of 12 crosses, 32 offspring in each cross, and
their parents were planted in the first two selection stages of the SASEX selection programme
to ascertain which stage provided the most useful information when selecting parent cultivars.
Data collected from Stage 2 was more reliable than data collected from Stage 1. Variance
components, narrow and broad sense heritabilities, correlations among traits, and clonal
repeatabilities between seasons were determined for 11 sugarcane traits at Stages 1 and 2.
These traits studied included: stalk population; stalk diameter; stalk height; cane mass;
dry matter % cane; fibre % cane; brix % cane; brix % dry matter; purity; pol % cane; and
ers % cane. Narrow sense heritabilities of the sugarcane traits were estimated by mid-parent
offspring regression . Alternative heritability estimates were obtained through restricted
maximum likelihood (REML) analysis of the unbalanced North Carolina design II at Stage 2.
Although narrow sense heritabilities determined by mid-parent-offspring regression were
comparable with those previously determined at SASEX and by other workers, REML was
more efficient than regression. Use of REML enabled additive and non-additive genetic
variance components to be estimated by allocating degrees of freedom to treatments and the
interactions between the different treatments. Heritability estimates varied for different traits
and compared favourably with those obtained under irrigated conditions and by other workers.
Additive genetic variance was more important than non-additive genetic variance for some
characters, but not for stalk population, cane mass, and dry matter % cane, for which both
variances were important. Selection of parent cultivars for all sucrose-related traits, fibre %
cane, and stalk diameter should be as successful under raingrown as under irrigated conditions,
provided that the environmental variation is determined efficiently under raingrown conditions.
Environmental correlations were observed between some traits, particularly between the yield related
traits, and may have influenced heritability estimates for those traits determined by
mid-parent offspring regression. Stalk diameter, fibre % cane, and brix % dry matter were the most repeatable traits between seasons. Cane mass was the least repeatable trait between
Stages 1 and 2 but was highly repeatable between plant (-P) and ratoon (-R) crops of Stage 2.
Stalk diameter was positively correlated with brix % dry matter (0.457-P and 0.623-R) and
strongly negatively correlated with stalk population (-0.790-P and -0.711-R) and fibre % cane
(-0.628-P and -0.651-R). Cane mass was strongly positively correlated with brix % dry matter
(0.638-P and 0.679-R). By selecting for brix % dry matter and stalk diameter, indirect
selection for cane mass would be possible. Brix % dry matter was determined as the most
reliable trait on which to base parental and commercial cultivar selection because it was highly
heritable, highly repeatable and highly positively correlated with stalk diameter and cane mass.||en