Agriculture Reference
In-Depth Information
collection there is a high chance of
finding many alleles
per
loci.
NAM populations have this advantage too as do recurrent selection
and multigeneration pedigree populations from breeding programs.
Meanwhile, in biparental populations for inbred crops just two alleles
are going to be found, while in other types of crosses such as MAGIC
populations, the amount of alleles depends on the number of
parents used. A shared advantage of AM, MAGIC, NAM, and RIL
populations alike is that several phenotypes can be tested for associa-
tion with the same GWA genotyping, which makes this a less expen-
sive strategy to do associations with different sets of phenotypic
data. On the other hand, a population that was constructed with a
limited number of parents exhibits less phenotypic variation, so the
phenotyping challenge is less dif
cult and statistical analyses can be
simpli
ed.
A. Limitations
Detection power in AM depends on the magnitude of the effect or
association of a marker locus and allele on the phenotype under study.
Moreover, the detection power is affected by gene and allele frequency
and the consideration or not of rare alleles. Infrequent but important
alleles are found in most germplasm collections and are often very
important and among the desirable types of alleles (Rafalski 2010).
Even though AM increases the amount of alleles that can be probed,
the effect of rare alleles cannot be detected unless they have large
penetrance and signi
cant effects.
Therefore, in some cases segregating biparental populations is still
more appropriate than AM for mapping rare alleles in germplasm of
interest. Likewise, the inclusion of different individuals with different
growing conditions may present limitations on the use of certain germ-
plasm. This is especially true when working with more delicate, wild
germplasm accessions that often require different
field conditions than
the standard agronomic conditions of their equivalent cultivated acces-
sions. The phenotypic evaluation for different germplasm should be
considered when designing an association study, for example with
strati
cation (Myles et al. 2009).
Confounding effects of unmeasured variables is a signi
cant factor
that adds false positives to AM. The most common factor that causes
confounding effects is population structure as already discussed. How-
ever, even after using population structure to correct or false positive
associations, some false negatives can still appear in the analysis.
This issue makes the validation of marker
×
trait associations through
