Conclusions and Discussion - Genetic Diversity in Establishing Plant Populations: Founder Number and Geometry - page 318

of the outcomes produced, especially when many interacting factors are
involved. NEWGARDEN analyses report levels of variance of means based
on a resampling of the various sets of conditions being modeled.
NEWGARDEN modeling can also be instructive in cases where the
modeling statistics being calculated do not always lend themselves to
obvious interpretation. One such relevant statistic here is the notion of
Ne, the effective population size. Obviously, Ne is not the census size of a
population, but rather the theoretical size of an ideal population that will
behave with regard to loss of genetic diversity (drift) at the same rate as
an actual population. One example using Ne noted earlier is the Franklin/
Soul 50/500 rule (e.g., see Meffe 1996: 53 or Finkeldey and Hattemer 2007:
123-127), under which an Ne size of at least 50 is necessary for short-term
reduction of inbreeding, and an Ne size of at least 500 is needed for long-
term maintenance of diversity (including protection against drift). But as
is well known, and as is shown in earlier NEWGARDEN trials, virtual
populations can be subject to inbreeding and/or drift (e.g., loss of unique
alleles) over a range of census population sizes much greater than 500
individuals.
Further, the modeling of population genetic diversity is still a
fermenting discipline, with many contrasting models and approaches under
development (e.g., see Broquet et al. 2009; Harwood 2009; Waples and
Faulkner 2009). NEWGARDEN trials and output can be used to compare
new model predictions with virtual trial output generated under a range
of preserve designs, founder spationumeric distributions, and species life
history characteristics.
Loci Composed of Fewer Alleles with Higher Frequencies?
In most of the examples presented earlier to demonstrate the use of
NEWGARDEN to examine the population effects of different modes of
establishment we used loci with numerous different alleles, each with a
relatively low frequency. We argued that such loci would often provide a
more sensitive indicator of genetic diversity loss and variance than measures
of heterozygosity, F, or loci with few alleles.
We acknowledge that such high-diversity loci composed of numerous
alleles, all at low frequencies, are likely not the most abundant class of
loci and may be rather uncommon (but see the next section). Obviously,
a wide range of variation across loci as to their number of alleles, and the
frequencies of those alleles, is to be expected within and among populations,
and among species. But loci with very low diversity (e.g., only one allele,
or one allele with very high frequency and two or four alleles of extremely
low frequency) are not of central interest in the context of this topic. Low
diversity means that there is not much diversity to lose, and thus they do