A population sample from people of diverse ethnic origins living in Ne
w Zealand serves as a database to test methods for inference of popula
tion subdivision. The initial null hypothesis, that the population sam
ple is homogeneous across ethnic groups, is easily rejected by likelih
ood ratio tests. Beyond this, methods for quantifying subdivision can
be based on the probability of drawing alleles identical by descent (F
-ST), probabilities of matching multiple locus genotypes, and occurren
ce of unique alleles. Population genetic theory makes quantitative pre
dictions about the relation between F-ST, population sizes, and rates
of migration and mutation. Some VNTR loci have mutation rates of 10(-2
) per generation, but, contrary to theory, we find no consistent assoc
iation between the degree of population subdivision and mutation rate.
Quantification of population substructure also allows us to relate th
e magnitudes of genetic distances between ethnic groups in New Zealand
to the colonization history of the country. The data suggests that th
e closest relatives to the Maori are Polynesians, and that no severe g
enetic bottleneck occurred when the Maori colonized New Zealand. One o
f the central points of contention regarding the application of VNTR l
oci in forensics is the appropriate means for estimating match probabi
lities. Simulations were performed to test the merits of the product r
ule in the face of subpopulation heterogeneity. Population heterogenei
ty results in large differences in estimates of multilocus genotype fr
equencies depending on which subpopulation is used for reference allel
e frequencies, but, of greater importance for forensic purposes, no fi
ve locus genotype had an expected frequency greater than 10(-6). Altho
ugh this implies that a match with an innocent individual is unlikely,
in a large urban area such chance matches are going to occur.