Agriculture Reference
In-Depth Information
will be considered for cultivar release. At this stage, and
before, it is common practice to examine the perfor-
mance averaged over 3-4 years of advanced testing and
to compare this performance to standard control culti-
vars that were included in the trials. The performance of
the breeding lines is likely to be somewhat biased as they
had specifically been chosen in the previous years test-
ing because they
had better than average
performance.
In order to get a true representation of the new culti-
vars worth, it is common to evaluate the newly released
cultivar for several years in breeder trials after selection
is complete. These post selection trials are commonly
conducted on a large scale, large (on-farm) plots and uti-
lizing farm-scale equipment (planters, harvesters, etc.).
Often it is the produce from these on-farm tests that
offers breeders the first opportunity to have sufficient
volume of material for actual quality evaluations.
No interaction
Yield
Line A
Line B
Low nitrogen
High nitrogen
Interaction
Yield
Line A
Line B
Experimental design
Limited number of test entries combined with large
amounts of planting material allows the use of the
most sophisticated experimental design at the advanced
stages. Therefore it is common to use lattice squares or
rectangular lattice designs for location trials. It should
be noted that such trials are often managed by collabo-
rators who are inexperienced in handling trials and the
more highly sophisticated the designs the more easily
can it be planted or harvested incorrectly. With this
in mind, randomized complete block designs may still
offer the most practical design for advanced trials.
Low nitrogen
High nitrogen
Figure 7.10
Genotype by environment relationships:
where there are no genotype by environment interactions
(top) and where the two genotypes (A and B) respond
differently to different environments (nitrogen levels).
However, it is not usually possible to fully evaluate the
G
×
E phenomenon until the advanced stages.
×
E interaction based on only two genotypes is
shown diagrammatically in Figure 7.10. It should be
noted that a significant interaction in an analysis of
variance can be obtained even when there is no change
in ranking of genotypes under study (no cross overs
in performance). Consider the two genotype cases in
Figure 7.10. On top both genotypes perform relatively
similarly (i.e. the two response lines are parallel). Inter-
actions can, however, appear significant if the lines
converge, but do not cross (i.e. no change in ranking
of the genotypes). If there is no change of ranking,
the interaction can usually be designated as a
scalar
effect
and is not considered by many as a true interac-
tion. Some would argue that the lines would converge
and eventually cross either above or below the range
of the data set collected, but there is no evidence that
this is true. Interactions caused by scalar effects can
often be removed by transformation of the data. True
interactions, where there are changes in ranking of lines
in differing environments, cannot be eliminated by data
transformation.
G
Genotype by environment interactions
A major goal of advanced selection is to determine the
response of selected lines over differing environments.
It is therefore difficult to consider this aim with-
out specifically considering genotype by environment
interactions.
As noted earlier, genotype
E)
interactions occur because some genotypes perform to
a high degree under some environmental conditions
while others perform poorly in that same environ-
ment, conversely the lower yielding lines may exceed the
higher yielding genotypes when grown under different
conditions.
G
×
environment (G
×
×
E interactions affect traits throughout all stages
of a plant breeding selection programme, unless molec-
ular markers or other similar techniques are used to
evaluate genotypes free from environmental influences.
