2.5. Assignment tests
Assignment tests are a robust method for testing structure in a
population (Paetkau et al., 1995). Assignment tests incorporate multiple
loci and therefore minimize the effect of rare alleles or low frequency
alleles, common in microsatellite data, from driving the pattern of
structure observed using other microsatellite analyses. The assignment
test calculates a reference frequency for each location while excluding
the individual in question, draws an individual from the total (all
locations combined) and calculates the likelihood that this individual
genotype belongs to any reference location, and then assigns it to the
reference location with which it has the greatest affinity (Paetkau et al.,
1995; Buonaccorsi et al., 2002). The program tests for significance using
permutation resampling to test whether the frequency of self-assignment
to a given location was greater than that expected in a panmictic
population. We used a correction factor based on Titterington et al.
(1981) when there were frequencies of zero in the data. We calculated
assignment probabilities using the Doh program (Brzustowski, 2002).