Principles of Plant Genetics and Breeding

Development of hybrids
BREEDING CORN 493
Heterotic groups
Heterosis is related to the level of heterozygosity. If two inbreds are crossed, heterosis is a function of the dominance in those
loci with different alleles in the inbreds. Therefore, the identification and development of heterotic groups of elite inbreds having
different alleles at loci regulating productivity can contribute to hybrid performance. A heterotic group is germplasm that when
crossed to germplasm from another heterotic group, tends to exhibit a higher degree of heterosis than when crossed to a member
of its own group (Lee 1995). Heterotic patterns, which are composed by two reciprocal heterotic groups, were empirically established
through testing and choice of lines to start breeding populations. Performance of lines in hybrids has been the main criteria
of classification into heterotic groups: A × B hybrids were superior to A × A or B × B hybrids. Heterotic patterns for temperate
maize are well established (e.g., “Reid’s Yellow Dent” × “Lancaster Sure Crop” in the USA, and European flints × US dent in
Europe). After establishment, heterotic patterns have been enhanced and optimized through selection and recombination.
Multiple heterotic patterns have been developed as a result of intensive elite line recycling and specific emphasis across breeding
programs. Testcross performance with representative testers has been used to group large number of inbreds to known heterotic
groups. Recently, DNA molecular markers have being effective for assigning inbreds to heterotic groups (Melchinger 1999). The
enhancement of heterotic response is improved by subsequent cycles of inbred line development. Increasing degrees of heterosis
are observed after several cycles of hybrid selection due to increasing divergence of allele frequencies and selection of complementary
alleles in the heterotic groups. In line recycling and in the development of source breeding populations, crosses among
elite lines from the same heterotic group are preferred. Heterotic response is heritable and inbreds have heterotic reactions
similar to their parents (Troyer 2001).
Correlation between inbred and hybrids, and hybrid prediction
The number of potential single crosses to evaluate increases substantially with the number of parental inbreds. The possibility
of using inbred line information, as indicative of hybrid performance, is desirable to reduce the number of hybrid evaluations.
The correlation between parental inbreds and hybrids depends on the trait. In general, the correlation is relatively high for some
additively inherited traits (e.g., plant morphology, ear traits, maturity, quality characters) but is relatively low for grain yield. The
correlation for grain yield has been consistently positive and significant but not high enough to predict hybrid performance. The
correlations between parental genetic diversity estimated with molecular markers, pedigree, or phenotypic traits and hybrid
performance also have been too low to have predictive value (Melchinger 1999). Methods based on linear mixed models have
been adapted to maize to predict performance of inbreds in untested environments or hybrid combinations (Bernardo 1999).
Although these recent approaches facilitate hybrid selection, hybrid testing is required ultimately to identify the inbreds with the
best breeding values.
Hybrid testing and screening
Hybrid testing in several environments representative of the target area is executed in several testing stages. A good example of
testing stages within a commercial breeding program is outlined by Smith et al. (1999):
Stage 1 Testcross performance of experimental lines in a few locations (e.g., five).
Stage 2 Hybrid evaluation of selected lines in more hybrid combinations and locations (e.g., 20).
Stage 3 Hybrid evaluation in c. 50 locations on research plots in several hybrid combinations.
Stage 4 Evaluation of best precommercial hybrids in c. 75 research plot locations and c. 200–500 on-farm locations.
Stage 5 Hybrid performance verification in c. 75 research plot locations and 300–1,500 on-farm strip plot tests.
Efforts are allocated in preliminary tests to evaluate as many hybrids as possible in a few locations with intensive selection,
leaving relatively few hybrids to proceed to the more advance precommercial stages. As the numbers of lines to be tested at various
stages of inbreeding increase over time, their evaluation in all possible hybrid combinations is not feasible. Therefore, testcrossing
with appropriate testers has been adopted extensively to evaluate the relative combining ability of experimental inbred lines.
The most common tester used is an elite inbred line from the opposite heterotic group. Testers with low frequency for favorable
alleles (e.g., susceptible to diseases) are also used. The level of inbreeding when testcross evaluation is conducted varies among
breeders and depends on the traits under consideration and the effectiveness of visual selection. Two basic systems are used: late
testing and early testing. In late testing, hybrid evaluation is delayed until advanced generations of selfing (e.g., S 5 or S 6 ), assuming
that selection for additively inherited traits, and seed yield during inbred line development, will assist in reducing the number of
lines for testcross evaluation. In early testing, evaluation of hybrid performance is conducted in early generations of inbreeding
(S 1 –S 3 ). Approximately 60% of the maize breeders in the USA evaluate new lines in testcrosses in S 3 and S 4 (Bauman 1981).
Characterization and selection of inbreds is a sequential testing with some lines discarded either through early testing or by performance
per se in early generations, and others discarded later by general and specific combining ability in hybrid combinations.