Principles of Plant Genetics and Breeding

422 CHAPTER 23
Yield
(a)
Yield
(c)
Figure 23.1 Graphical presentation of: (a) genotype × environment (G × E) interactions, (b) heterogeneity, (c) crossover
interactions, and (d) combined interactions.
different genotype (B) is more productive in another
environment. The basic test for crossover interaction
(also called qualitative interaction) is to compare the
performance of two genotypes in two environments
and to determine if the difference in performance is
significantly less than zero in one environment and
significantly greater than zero in the other.
4 Combined G × E interaction. The first three interactions
previously described are the ones commonly
discussed. If one of the factors considered on the axes
increases for one genotype and reduces for the other
genotype, there is a combined G × E interaction.
The axes in the graph may be for any relevant factor
of interest to the breeder. For example, the x-axis may
be rainfall, while the vertical axis (y) may be grain yield.
In spite of the complexity of the environment, sometimes
one factor may predominate to characterize the
environment (or may be imposed by design). It should
be pointed out these four graphs are only a selected
unique few of the numerous patterns that may occur
in reality. The breeder is most interested in repeatable
G × E interactions.
Measurement of G × E interactions
Interactions occur at various biological levels, such as
genotypic, QTL (quantitative trait locus), and phenotypic
(b)
Soil fertility
(environmental mean) (d)
Soil fertility
(environmental mean)
levels – the first two requiring genetic analysis. G × E
interaction at the phenotypic level requires observations
at the plant or crop level. The G × E interactions can
also be partitioned into linear trends (e.g., G × location,
G × year, G × time). Statistical methods are used to
assess G × E interactions. Consequently, the proper field
plot design and analysis are required for an effective
assessment of the interactions. These methods include
both parametric and non-parametric procedures – partitioning
of variance, regression analysis, non-parametric
methods, and multivariate techniques.
Analysis of variance (ANOVA)
To ascertain the presence of a G × E interaction, breeders
conduct a network of comparative trials, as previously
described, in which prospective cultivars are compared
with standard cultivars at multiple locations or agroecological
regions. The premise for such trials, according to
Mather and others, is expressed by a linear equation:
X =µ+g + e + ge
where X = yield of some other quantitative traits, µ =
mean value of the population (trial), g = value of the
genotype (cultivar), e = value of the environmental
effect, and ge = genotype × environment interaction.