Founder Number and Allelic Diversity - Genetic Diversity in Establishing Plant Populations: Founder Number and Geometry - page 81

When all unique alleles at each of the 20 loci are at equal frequencies, as
in the previous example, the above code could be abbreviated using the
following “auto_alleles_per_locus” statement:
<LOCI number_loci=“0” auto_alleles_per_locus=“10”
number_automatic_loci=“20”/>
In the previous example with one locus with 10 alleles of equal frequency,
the maximum number of unique alleles that could be retained in the founding
population was 10 (see Fig. 7.5). Consider now the just described example
with a source population that has 20 replicates of that locus. In this case, 200
unique alleles are present in the infi nite source population (in 20 loci, each
with 10 unique alleles, all of frequency 0.1). That is the maximum number
of source population unique alleles that can be retained in the founding
population. This is equivalent to sampling the one locus of the previous
example 20 times, or for a trial with 25 runs, sampling that locus 500 times.
The increased replication sampling should act to decrease the variation
around the “expected” mean value. This effect can be seen in Fig. 7.9, which
compares results of the just described set of trials in which one locus with 10
alleles all of equal frequency (0.1; “B” in the fi gure) was analyzed for unique
allele retention as the number of founders increases, with results from re-
conducting the trials in the exact same fashion, except that in each trial, 20
loci, each with 10 alleles of equal frequency, are included (“C”). Since, in the
former, a maximum of 10 unique alleles can be retained among the founders,
while in the 20 loci case, 200 unique alleles is the maximum, the results in
Fig. 7.9 are shown standardized as the percentage of the possible maximum
of unique alleles retained. Note that, since there is more repeated sampling
of the same type of locus in the 20 loci case (letter C), that variation about the
mean value is reduced for each increasing level of number of founders. This
effect of reduced variation also applies to analysis of heterozygosity (see Fig.
7.10). These results illustrate that the amount of variation about mean values
will be infl uenced by differences in the balance of loci types in comparing
NEWGARDEN runs with different loci arrays, and also demonstrate how
this principle should be considered when comparing genetic variation across
natural organisms. Analyses of small populations will generally yield more
accurate mean values if multiple copies of the number of loci of each type
are included. Note, however, that the relative balance of loci of each type (e.g.,
4 unique allele loci, 10 allele loci) should be refl ected in the input loci array,
when information on the balance is available. For example, loci with 10 unique
alleles of approximately equal frequency may account for 14% of the genome,
while loci with 5 unique alleles, each with decreasing frequency, may account
for 42% of the genome. The number of repetitions of each locus type submitted
in the input fi le should proportionately refl ect those percentages.