AMERICAN JOURNAL OF HUMAN BIOLOGY 14:429439 (2002)

Maximum Likelihood Estimates of Admixture in Northeastern

Mexico Using 13 Short Tandem Repeat Loci
RICARDO M. CERDA-FLORES,1,2 BRUCE BUDOWLE,3 LI JIN,2,4 SARA ANN BARTON,2*
RANJAN DEKA,4 AND RANAJIT CHAKRABORTY2,4
1
Departamento de Genetica de Poblaciones, Centro de Investigacion Biomedica del Noreste
(CIBIN), Instituto Mexicano del Seguro Social (IMSS), Monterrey Nuevo Leon, Mexico 64720
2
Human Genetics Center, School of Public Health, The University of Texas, Houston, Texas
77225, USA
3
FBI, Laboratory Division, Washington, DC 20535, USA
4
Center for Genome Information, Department of Environmental Health, University of
Cincinnati, Kettering Lab, Cincinnati, Ohio 45267, USA

ABSTRACT Tetrameric short tandem repeat (STR) polymorphisms are widely used in population
genetics, molecular evolution, gene mapping and linkage analysis, paternity tests, forensic analysis,
and medical applications. This article provides allelic distributions of the STR loci D3S1358, vWA,
FGA, D8S1179, D21S11, D18S51, D5S818, D13S317, D7S820, CSF1PO, TPOX, TH01, and D16S539
in 143 Mestizos from Northeastern Mexico, estimates of contributions of genes of European (Spanish),
American Indian and African origin in the gene pool of this admixed Mestizo population (using 10 of
these loci); and a comparison of the genetic admixture of this population with the previously reported
two polymorphic molecular markers, D1S80 and HLA-DQA1 (n = 103). Genotype distributions were
in agreement with Hardy-Weinberg expectations (HWE) for almost all 13 STR markers. Maximum
likelihood estimates of admixture components yield a trihybrid model with Spanish, Amerindian, and
African ancestry with the admixture proportions: 54.99% 3.44, 39.99% 2.57, and 5.02% 2.82,
respectively. These estimates were not significantly different from those obtained using D1S80 and
HLA-DQA1 loci (59.99% 5.94, 36.99% 5.04, and 3.02% 2.76). In conclusion, Mestizos of
Northeastern Mexico showed a similar ancestral contribution independent of the markers used for
evolutionary purposes. Further validation of this database supports the use of the 13 STR loci along
with D1S80 and HLA-DQA1 as a battery of efficient DNA forensic markers in Northeastern Mestizo
populations of Mexico. Am. J. Hum. Biol. 14:429439, 2002. 2002 Wiley-Liss, Inc.

Cosmopolitan Mestizo populations from search aimed to 1) document the summary

large metropolitan regions of Mexico are
described to have been derived from Euro-
peans (Spanish), American Indians, and
Africans (Lisker, 1981). Data from blood
groups, serum protein, as well as some of
the recently characterized DNA markers
support this ethnohistory of the Mestizo
populations of Mexico (Lisker and Babin-
sky, 1986; Cerda-Flores et al., 1991, 2001;
see also Crawford et al., 1976). In view of
this admixed ancestry, the Mestizo popula-
tions of Mexico serve as suitable examples
for examining the robustness of assump-
tions of random association of alleles within
and between loci that are used in DNA fo-
rensics. Further, characterization of allele
frequency distributions in such populations
of Mexico enables comparison of these pop-
ulations with those from the united states,
labeled as Hispanics or Mexican Americans
(Edwards et al., 1992; Budowle et al., 1999).
With a random sample of unrelated indi-
viduals from Northeastern Mexico, this re-
statistics of genotype and allele frequency
distributions of this Mestizo population at
the 13 short tandem repeat loci (D3S1358,
vWA, FGA, D8S1179, D21S11, D18S51,
D5S818, D13S317, D7S820, CSF1PO,
TPOX, TH01, and D16S539) used in the
Combined DNA Index System (CODIS) of
the US DNA forensics community; 2) esti-
mate the relative contributions from the
presumed ancestral populations of Euro-
pean, American Indian, and African origin
Contract grant sponsors: CONACYT, Mexico, scholarship
113518 (to R.M. Cerda-Flores), U.S. Public Health Service Re-
search; Contract grant number: GM41399; cgs: U.S. National
Institute of Health, cgn: 1996 IJ-CX-0023, Cgs: National Insti-
tute of Justice.
*Correspondence to: Sara Ann Barton, Human Genetic Cen-
ter, University of Texas, School of Public Health, P.O. Box
20186, Houston, TX 77225. E-mail: sabarton@sph.uth.tmc.edu
Received 10 October 2001; Revision received 16 January 2002;
Accepted 28 January 2002
Published online in Wiley InterScience (www.interscience.
wiley.com). DOI: 10.1002/ajhb.10058

2002 Wiley-Liss, Inc.

430 R.M. CERDA-FLORES ET AL.
in the present-day Mestizo population of
this region; 3) validate the robustness of
assumptions of allelic independence within
and across loci, in spite of the admixed ori-
gin of the population, and 4) examine the
appropriateness of recommendations of the
National Research Council (NRC, 1996) for
interpreting the statistical strength of DNA
matches in forensic case analyses.
The state of Nuevo Leon, in Northeastern
Mexico (which includes the states of Nuevo
Leon, Coahuila, Tamaulipas, San Luis Pot-
osi, and Zacatecas) has an area of 64,555
Km2 and had a population of more than 3.8
million inhabitants in 2000. It is a young
population: 83.8% are under 44 years of age
(Census, 2000). Nearly 82% of the state's
population resides in the Monterrey Me-
tropolitan Area (MMA) in the central
Western section of the state of Nuevo Leon
(DEPD, 1978). In 1960, Nuevo Leon had
1,078,848 inhabitants, with almost 50% of
this population having migrated to the
MMA, primarily from the adjoining states
of San Luis Potosi and Zacatecas (Monte-
mayor-Hernandez, 1971).
Admixture estimates based on blood
group data indicate that the populations of
Northeastern Mexico are similar in terms of
the contribution of Spanish (60%), Amerin-
dian (37%), and African (3%) genes. More
than 96% of the total genetic diversity could
be attributed to individual variation within
the populations defined by birthplaces and/
or birthyear. There is no nonrandom asso-
ciation of alleles among the genetic marker
systems, despite the admixed origin of this
Mestizo population (Cerda-Flores et al.,
1991, 2001). However, it is not known
whether the recently introduced DNA
markers reveal the same picture.
The introduction of the core 13 CODIS loci
(used in the Combined DNA Index System of
the US Federal Bureau of Investigation) in
most countries of the world resulted in
compilations of allele frequency data at
these loci in various parts of the world (see,
e.g., http://www.uni-Duesseldorf.de/WWW/
MedFak/Serology; http://www.cstl.nist.gov/
biotech/strbase). As a consequence, with
genotyping surveys in appropriate popula-
tions these public domain databases can be
used to address a number of issues relevant
to forensics and population structure ana-
lyses in evolutionary studies. The issues
addressed in this article accomplish such a
task with multilocus genotype data from 143
unrelated Mestizo individuals sampled from
the greater MMA of Nuevo Leon, Mexico.
MATERIALS AND METHODS
The data from this population were col-
lected as part of a larger investigation of the
genetic structure of the Mexican Mestizo
populations in Northeastern Mexico. Whole
blood (5 ml) was collected into tubes con-
taining EDTA by venipuncture from unre-
lated, healthy individuals. After the tubes
were gently rotated to ensure that the
EDTA and blood were well mixed, plastic
pipettes were used to extract blood from the
tubes and a large drop (125 ll) was placed
onto the center of each of four circles on the
commercially prepared FTA cards. When
blood is spotted onto the FTA paper, cells
are lysed and DNA from the nuclei of the
white cells is immobilized within the matrix
of the paper. An FTA paper punch was used
to cut a 1-mm diameter of paper and the
paper was washed according to the proce-
dure specified in Budowle et al. (2000) to
prepare it for PCR.
STR amplication
The 13 CODIS loci (D3S1358, vWA, FGA,
D8S1179, D21S11, D18S51, D5S818,
D13S317, D7S820, CSF1PO, TPOX, TH01,
D1S539) were typed using the Amp-
FISTRTM Profiler PlusTM kit and Amp-
FISTRTM CofilerTM kit (Applied Biosys-
tems, Foster City, CA). The manufacturer's
recommended protocols were followed.
STR typing of proler plus and coler
amplied samples
Samples were analyzed using the ABI
PrismTM 310 Genetic Analyzer (Applied
Biosystems). The samples were analyzed
using the separation medium performance
optimized polymer (POP) 4TM (PE Biosys-
tems, Foster City, CA). The genotypes were
recorded with allele designations deter-
mined by comparison of the sample frag-
ments with those of the allelic ladders (in
units of repeat sizes of alleles).
Statistical analysis
Statistical analyses included the follow-
ing steps: (1) Allele frequency computations
(by the gene count method; Li, 1976), tests
of Hardy-Weinberg expectations of geno-
type frequencies (by three tests: homozy-
 STR IN A MEXICAN MESTIZO POPULATION
gosity test of comparison of the observed
frequencies of homozygotes/heterozygotes
with their respective unbiased estimates
(Nei, 1978); likelihood ratio test (Weir,
1996); and the exact test of Guo and
Thompson (1992), and tests of pairwise
linkage disequilibria (LD) between all pairs
of loci (in which again the likelihood ratio
test, LRT, and exact test criteria were used
(Zaykin et al., 1995). Empirical level of
significance was determined in each of the
tests through permutation of alleles with
10,000 replicates.
(2) Test of mutual independence of loci
through a comparison of observed and ex-
pected distributions of the number of alleles
shared and the number of loci exhibiting
genotypic identity in comparison with 13-
locus DNA profiles of all pairs of the sampled
individuals (in all 10,153 pairwise compari-
sons of profiles). The expected distributions
of these statistics under the assumption of
mutual independence of alleles within and
across loci were computed based on the al-
gorithm of Chakraborty and Jin (1992).
(3) Evaluation of summary statistics of
levels of polymorphism and their utility
through estimation of locus-specific hetero-
zygosities, power of exclusion (for paternity
testing Ohno et al., 1982), and power of
discrimination (Jones, 1972).
(4) Estimation of admixture components
by using the maximum likelihood method of
Elston (1971). For this purpose, the trihy-
brid model of admixture for the Monterrey
population was fitted with contributions
from three parental populations: Spanish,
Amerindian, and African. As representative
of ancestral populations, gene frequency
data from the Southwestern region of Spain
were obtained from Gamero et al. (2000).
Pooled allele frequencies from several
Southwestern Amerindian communities
(Apache, Navajo, and Pueblo) represented
the Amerindian ancestral allele frequencies
(unpublished data from U.S. Scientific
Working Group of DNA Analysis Methods,
[SWGDAM]; see also Budowle et al., 2001).
For Africans allele frequencies from Nigeria
were used (Deka et al., unpublished data).
The admixture estimates are based on 10
loci (TPOX, CSF1PO, and D16S539 are ex-
cluded), since these three loci were not
typed in the Nigerian database.
5) Comparison with the U.S. Hispanic
database (Budowle et al., 1999) using two
different types of analyses. First, the allele
431
frequencies for each locus for the Monterrey
population were compared with the same of
the U.S. Hispanic database. The R C
contingency chi-square test was used with
its empirical significance level determined
by permutations, following the algorithm of
Roff and Bentzen (1989). Second, multilocus
profile frequencies of 143 individuals were
computed based on allele frequencies of this
database and those from allele frequencies
in U.S. Hispanics. The multilocus profile
frequencies of 209 U.S. Hispanics (Budowle
et al., 1999) were also computed based on
the same two sets of allele frequencies.
Scatter plots of the computations are pre-
sented in three steps. First, the strict
product rule was followed, i.e., Hardy-
Weinberg expectation for each locus and no
linkage disequilibria. Alleles not observed
in one database, but seen in profiles in the
other database, were assigned frequencies
of 0.5/2n, n being the number of individuals
sampled in the database in which the profile
is not observed. The second panel of com-
putations used the recommendation 4.1 of
the National Research Council (NRC, 1996),
in which homozygote profile frequencies are
estimated with adjustment for population
substructure (i.e., p2 + hp(1)p), with h =
0.01), but the estimates of heterozygote
frequencies required no adjustment (for
attaining conservativeness). In these com-
putations, all allele frequencies below a
minimum threshold value were replaced by
threshold values as determined from the
theory of Budowle et al. (1996). Third, the
conditional probabilities (formulae 4.10a
and 4.10b of NRC, 1996) were evaluated, in
which again the minimum threshold allele
frequencies were used.
RESULTS
The distributions (%) of the observed
alleles for all 13 STR loci are shown in
Table 1. Also shown are estimates of the
minimum threshold allele frequencies for
each locus. No off-ladder allele was found in
this sample and the estimates of the mini-
mum threshold allele frequencies are with-
in the bounds observed in other databases
of similar size (Budowle et al. 1996, 2001).
The sample size of the database is also ad-
equate for using these allele frequencies for
the purposes of DNA forensic and parentage
testing in the Monterrey population (Evett
and Gill, 1991; Chakraborty, 1992).
 TABLE 1. Allele frequencies for 13 STR loci in a sample of 143 individuals from a Northeastern Mestizo Mexico population
Allele D3S1358 VWA FGA D8S1179 D21S11 D18S51 D5S818 D13S317 D7S820 CSF1PO TPOX THO1 D16S539
N 143 143 143 143 143 143 143 143 143 143 143 143 143
6 0.35 28.67
7 7.69 1.40 0.35 30.77
8 0.70 0.35 9.09 8.39 0.35 49.30 10.49 0.35
9 4.55 23.08 8.74 0.70 8.74 9.44 9.79
9.1 0.35
9.3 19.23
10 12.59 1.05 6.29 6.99 29.02 24.13 2.80 1.40 19.58
11 7.34 2.10 38.81 19.58 27.97 28.67 31.12 31.82
12 0.35 11.19 9.09 29.37 25.17 19.93 39.86 7.34 26.92
13 1.05 32.52 11.19 11.54 12.24 3.85 4.20 0.35 9.79
14 7.34 8.74 22.38 17.13 1.05 3.85 0.35 1.75 1.75
15 40.91 10.49 10.84 16.78 0.35
16 23.78 31.82 2.45 12.24
17 14.69 29.72 15.38
18 11.54 13.64 1.05 8.04
19 0.35 5.59 8.74 1.75
20 10.49 3.50
21 11.54 0.70
22 11.89 0.70
23 15.73
23.2 0.35
24 17.48 0.35
25 12.24
26 7.34
27 2.10 0.35
28 0.70 10.49
28.2 0.35
29 20.98
29.2 0.70
30 0.35 27.97
30.2 3.15
31 6.64
31.2 11.54
32 0.70
32.2 12.94
33.2 3.85
34.2 0.35
MAF (%) 1.77 1.81 1.99 1.85 1.90 1.98 1.77 1.88 1.83 1.72 1.66 1.81 1.81
MAF = minimum allele frequency, recommended for forensic use of this allele frequency database (see text for details).
 STR IN A MEXICAN MESTIZO POPULATION 433

TABLE 2. Observed significant deviations and P-values for Hardy-Weinberg and Linkage Equilibrium expectations
HWE testa p-value LD testb p-value
c
Loci Exact LRT Loci Exact LRTc
D2S1358 0.019 0.018 D3S1358-D21S11 .044
D3S1358-D13S317 .023 .040
D3S1358-CSF1PO .018 .009
D5S818-D16S539 .020 .024
a
Test for Hardy-Weinberg equilibrium. Empirical levels of significance (P-values) are shown only for the loci where departures are
significant (P < 0.05). Exact refers to the exact test of Guo and Thompson (1992) and LRT is the likelihood ratio test (Weir, 1996).
b
LD test refers to tests of linkage disequilibrium between all pairs of loci (Zaykin et al., 1995). Only the pairs showing significant
departures from linkage equilibrium (P < 0.05) are shown with their respective P-values. P-values for both tests are based on 10,000
replication of permutations.
c
LRT = likelihood ratio test.

TABLE 3. Locus-specific heterozygosity, exclusion probability, and power of discrimination at the 13 CODIS STR
loci in the Mestizo population of Northeastern Mexico
Number Heterozygosity Exclusion Power of
of probability Discrimination
Loci alleles Obs. Exp. (PE) (PD)
D3S1358 8 0.6783 0.7383 0.5145 0.8922
VWA 6 0.7552 0.7728 0.5611 0.9128
FGA 13 0.9441 0.8807 0.7515 0.9724
D8S1179 8 0.8531 0.8008 0.6116 0.9227
D21S11 13 0.8881 0.8326 0.6670 0.9506
D18S51 14 0.8741 0.8775 0.7461 0.9712
D5S818 9 0.6853 0.7403 0.5194 0.8928
D13S317 7 0.7902 0.8183 0.6342 0.9405
D7S820 9 0.8112 0.7842 0.5768 0.9194
CSF1PO 8 0.7133 0.7010 0.4401 0.8537
TPOX 7 0.6853 0.6485 0.3946 0.8183
TH01 6 0.7832 0.7687 0.5474 0.9078
D16S539 7 0.7972 0.7711 0.5526 0.9099
Combined 18.90 10)6 12.83 10)15
Observed (Obs.) heterozygosity values are the proportion of all heterozygotes at the locus, while expected (Exp.) heterozygosity is
computed as the complement of the sums of squares of allele frequencies, corrected for bias of estimation (Nei, 1978).

Table 2 lists the loci and pairs of loci which
showed significant departures from Hardy-
Weinberg expectations (HWE) and linkage
equilibria between pairs of loci. Of the 13
locus-specific HWE tests, only the D3S1358
locus deviated from the HWE assumption by
both tests (exact and LRT). Four of the 78
pairs of loci showed departures from linkage
equilibria (LE), of which three pairs in-
volved the same locus (D3S1358) that devi-
ated from HWE. Thus, even without any
adjustments for multiple testing (e.g., Bon-
ferroni correction of p-values), there is no
evidence of any gross departures (i.e., de-
partures exceeding the nominal level of
significance) from HWE and LE in the
Monterrey Mestizo population.
Table 3 presents locus-specific values of
observed and expected (unbiased) hetero-
zygosity, exclusion probability, and power of
discrimination based on the allele frequency
data of Table 1. Clearly, each of the 13 loci is
highly polymorphic in this Mestizo popula-
tion and the exclusion probability (PE) and
power of discrimination (PD) values are
comparable with other cosmopolitan data-
bases (Budowle et al., 2001). The combined
average exclusion probability and the com-
bined power of discrimination are shown in
a format that also reflects the cumulative
efficiency of these loci in parentage and fo-
rensic analyses. For example, the combined
PE value of 1 8.9 10)6 suggests that, on
average, roughly 1 of every 112,360 wrongly
accused males will not be excluded as a
possible father when a mother and her child
are typed for all 13 loci. Likewise, the
combined PD value, 12.83 10)15, implies
that the coincidental match probability for
these 13 loci is approximately 1 in over 353
trillion tests. Of course, these are average
values for the set of 13 loci as a whole and,
hence, for specific genotypes the numbers
will vary. Nonetheless, these average esti-
mates lead to the conclusion that the
CODIS 13 loci should be a useful battery of
markers for DNA testing in the Monterrey
Mestizo population.
434 R.M. CERDA-FLORES ET AL.

Fig. 1. Observed (histogram) and expected (line graph) distributions of the number of loci with genotypic
identity (a) and number of alleles shared (b) between 13 locus DNA profiles of all pairwise comparisons of profiles in
the sample. The expected distributions are computed under the assumption of mutual independence of loci.

Since the combined probability calcula-
tions for all 13 loci involve the assumption
of mutual independence of loci (in addition
to the assumptions of HWE and LE),
Figure 1 shows the conformity with this
assumption. The observed (histogram) and
expected (line graph) distributions of a
number of loci exhibiting genotype identity
in all possible comparisons of pairs of 13-
locus profiles (Fig. 1a) and that of allele
sharing (Fig. 1b) are virtually identical. In a
comparison of two random 13-locus profiles
in this population, the observed mean
number of alleles shared was 8.54 (SD =
2.13), while under the mutual independence
of alleles the expectation is 8.57 (SD = 2.10).
Likewise, the observed and expected num-
ber of loci showing genotypic matches in two
random profiles are 1.12 and 1.13 (SD 0.99
and 1.00, respectively). Thus, for these two
forensically meaningful summary statistics
the data on 13 locus genotype profiles are in
almost perfect concordance with the pre-
dictions of mutual independence of alleles
within as well as across loci.
The results of comparison of this Mestizo
database with that for U.S. Hispanics
(Budowle et al., 1999) are shown in Figure
2. First, the R C contingency table anal-
ysis of comparison of allele frequencies de-
tected a significant difference (p = 0.021)
only at the vWA locus. The profile frequency
estimates based on the allele frequency es-
timates of the two databases for 143 Mestizo
individuals and 209 U.S. Hispanic individ-
uals are generally across the 45 line (Fig.
2ac). A closer examination reveals that
when the strict product rule (i.e., HWE and
LE) is used the profile frequency estimates
based on the cognate (same sample) allele
frequencies are generally higher (Fig. 2a),
suggesting that the criticism of Krane et al.,
(1992) is supported. However, when two
stages of conservativeness are invoked in
Figure 2b (recommendation 4.1 of NRC,
(1996); making adjustment for population
substructure effect and invoking a minimum
threshold allele frequency for rarer alleles),
and Figure 2c (formulae 4.10a,b of the NRC
(1996), conditional probabilities invoking the
minimum threshold allele frequencies), the
scatter plots are much closer to the 45 line,
with separation of the population samples
almost disappearing. This demonstrates an
empirical validation of the NRC (1996) rec-
ommendations, suggesting that the cur-
 STR IN A MEXICAN MESTIZO POPULATION 435

Fig. 2. Comparison of the estimates of 13 locus DNA profile frequencies for 143 Mestizo individuals of Mont-
errey, Mexico, and 209 U.S. Hispanic individuals surveyed by Budowle et al. (1999) using three types of estimation:
strict product rule (A), population substructure adjustment as prescribed in recommendation 4.1 of NRC, 1996 (B),
and population substructure adjustment using formulae 4.10 (A) and (B) of NRC, 1996 (C). The open squares refer
to the estimates for the Mestizo individuals of Monterrey and the triangles refer to the U.S. Hispanics.

TABLE 4. Percentage contribution from Spanish, Amerindian, and African gene pools to the Mestizo population
in Northeastern Mexico
Marker Spanish Amerindian African
HLA-DQA1 and D1S80a 59.99 5.94 36.99 5.04 3.02 2.76
10 STRb 54.99 3.44 39.99 2.57 5.02 2.82
10 STRc 50.00 2.64 44.00 2.17 6.00 2.13
The 10 STR loci are D3S1358, vWA, FGA, D8S1179, D21S11, D18S51, D5S818, D13S317, D7S820, and THO1. The admixture
estimates based on D1S80/HLA-DQA1 and 10 STR loci for the Monterrey population are not significantly different (P > 0.46 for all
three components) and so are the ones for Monterrey and U.S. Hispanic populations for the same 10 STR loci (P > 0.25 for all three
components).
a
Mestizo population of Monterrey, Mexico (Cerda-Flores et al., 2001).
b
Mestizo population of Monterrey, Mexico (present study).
c
U.S. Hispanics (Budowle et al. 1999).
436 R.M. CERDA-FLORES ET AL.
rently employed conservative means of esti-
mating multilocus profile frequencies
achieves the purpose of virtually eliminating
the effect of allele frequency differences in
subpopulations (or samples thereof).
Finally, the admixture estimates (contri-
butions of the three presumed ancestral
gene pools) are shown in Table 4. As shown
in the Appendix, allele frequencies for the
Nigerian sample are not available for three
loci (TPOX, CSF1PO, and D16S539) and,
hence, the admixture estimates for the
CODIS STR loci are based on 10 loci. The
relative contributions from Spanish, Amer-
indian, and African sources in the present
Mestizo population of Nuevo Leon are ap-
proximately 55%, 40%, and 5%, respective-
ly; not significantly different (p > 0.25) from
those (50%, 44%, and 6%, respectively) in
the U.S. Hispanic population. Also, the
STR-based admixture estimates are similar
to these (60%, 37%, and 3%) based on two
other DNA markers (D1S80 and HLA-
DQA1) in the Monterrey Mestizo population
(Cerda-Flores et al., 2001).
DISCUSSION
This is the first study of 13 CODIS STR
loci in a cosmopolitan Mestizo population of
Mexico, in which 13 locus genotype data
have been presented in 143 unrelated indi-
viduals from the greater Monterrey Metro-
politan area. Concordant with ethnohistory,
these DNA markers show that the popula-
tion is of admixed origin, with contributions
of genes from the Spanish, American Indi-
ans, and Africans in approximate propor-
tions of 55%, 40%, and 5%, respectively.
While the choice of samples that repre-
sent the ancestral gene pools may influence
such estimates (Chakraborty, 1986), we ar-
gue that these estimates are probably reli-
able for several reasons. First, with al-
ternative choice of African samples (e.g.,
Angola and Mozambique [http://www.uni-
Duesseldorf. de/WWW/MedFak/Serology]),
and other Spanish surveys (e.g., Galicia, and
pooled data from Spain, Canary Islands not
included [http://www.uni-Duesseldorf.de/
WWW/MedFak/Serology] the admixture es-
timates (Table 4) do not change appreciably
(data not shown). Second, those admixture
estimates are consistent with those based
on D1S80 and HLA-DQA1 (Table 4), and
also with these obtained from blood groups
(Cerda-Flores et al., 1991). Third, the trihy-
brid admixture model is also supported by the
ethnohistory of the Mexican Mestizo popu-
lations (Lisker, 1981; Crawford et al., 1976).
Finally, the adequacy of these 13 STR loci
for admixture proportion estimation may be
judged through a computation of the com-
posite d-score (one-half of the summed ab-
solute allele frequency differences between
parental populations; Shriver et al., 1997) of
allele frequency differences among the three
ancestral populations. Only the D8S1179
locus had a composite d-score below 0.27 for
Spanish-Amerindian allele frequency dif-
ferences; three (vWA, FGA, and D5S818)
fell in this category for Spanish-African
differences and none for Amerindian-
African differences. Thus, this exercise
also establishes that the CODIS 13 STR loci
are useful for estimating admixture pro-
portions in the Continental populations of
the Americas (in most of whose admixture
occurred from the same three sources
Europeans, American Indians, and Afri-
cans).
In view of the admixed origin of this pop-
ulation, the summary features of genetic
variation at the 13 loci in this Mestizo popu-
lation (Tables 13, Figs. 1, 2) are even more
interesting. With the history of admixture
not more than 400 years (or 20 generations,
approximately), composite d-scores of the
magnitude summarized above should have
generated considerable linkage disequilibria
between these markers, even though they
are unlinked (Chakraborty and Weiss, 1988).
Yet this has virtually no impact on depar-
tures from HWE, LE, and mutual indepen-
dence of alleles within and across loci (Table
2, Fig. 1). One might argue that this is due to
the low statistical power of detecting dis-
equilibria through these tests. Indeed, these
tests have limited power (Ward and Sing,
1970; Zaykin et al., 1995). Nonetheless, the
low level of allele frequency differences be-
tween the Monterrey Mestizo (present sur-
vey) and U.S. Hispanic samples (Budowle et
al., 1999), together with the close similarity
of profile frequency estimates (Fig. 2b,c),
suggest that only a minimal level of adjust-
ment (i.e., h = 0.01) is needed to account for
any effect of recent admixture in this popu-
lation. In this context, it is worth noting that
a gene diversity estimate of h based on re-
gional allele frequency differences of the U.S.
Hispanic populations is 0.21% (Budowle
et al., 2001), much smaller than the 1% value
we used in our computations (Fig. 2).
 STR IN A MEXICAN MESTIZO POPULATION
Finally, this survey illustrates the need
to establish corresponding databases else-
where in Mexico. Regional differences in
contributions from different gene pools can
be detected through these loci and their
impact on forensic computations can be
evaluated. For three of these loci (vWA,
THO1, and CSF1PO), a recent survey
presented allele frequencies from a Mesti-
zo population of Jalisco, a Western state of
Mexico (Rangel-Villalobos et al., 1999).
Minor differences of allele frequencies
were noted at the vWA locus (R C con-
tingency chi-square 12.32, p = 0.03 by the
algorithm of Roff and Bentsen, 1989) and
none at the other two. This is consistent
with almost no difference of admixture
components in Jalisco and Nuevo Leon
based on D1S80 and HLA-DQA1, which
was reported elsewhere (Cerda-Flores et
al., 2001). Thus, further availability of
genetic data on these loci from Mexico
should increase our understanding of the
population structure in the various Mexi-
can populations and improve the statisti-
cal interpretation of DNA forensic and
parentage testing results.
ACKNOWLEDGMENTS
The authors thank the volunteer partici-
pants and the IMSS and Universidad
Autonoma de Nuevo Leon facilities in
Monterrey, Nuevo Leon, for data collection
and sample preparation. Institutional Re-
view Board approval was obtained for all
sample collections.
LITERATURE CITED
Budowle B, Monson KL, Chakraborty R. 1996. Esti-
mating minimum allele frequencies for DNA profile
frequency estimates for PCR-based loci. Int J Leg
Med 108:173176.
Budowle B, Moretti TR, Baumstark AL, Defenbaugh
DA, Keys KM. 1999. Population data on the thirteen
CODIS core short tandem repeat loci in African Am-
erindians, U.S. Caucasians, Hispanics, Bahamians,
Jamaicans, and Trinidadians. J Forensic Sci
44:12771286.
Budowle B, Smith J, Moretti T, DiZinno J. 2000. DNA
Typing protocols: molecular biology and forensic
analysis. Natick, MA: Eaton. p 4142.
Budowle B, Shea B, Niezgoda S, Chakraborty R. 2001.
CODIS STR loci data from 41 sample populations. J
Forensic Sci 46:453489.
Census. 2000. Instituto Nacional de Estadstica Geog-
rafa e Informatca. 2000. XII Censo General de Po-
blacion y Vivienda. Mexico, DF: Institute Nacional de
Estadistca, Geografa e Informatica (http://www.
INEGI.gob.mx).
437
Cerda-Flores RM, Kshatriya GK, Barton SA, Leal-
Garza CH, Garza-Chapa R, Schull WJ, Chakraborty
R. 1991. Genetic structure of the populations mi-
grating from San Luis Potosi and Zacatecas to Nuevo
Leon, Mexico. Hum Biol 61:249261.
Cerda-Flores RM, Villalobos-Torres MC, Barrera-Sal-
dana HA, Cortes-Prieto LM, Barajas LO, Rivas R,
Carracedo A, Zhong Y, Barton SA, Chakraborty RC.
2001. Genetic admixture in three Mexican Mestizo
1 populations based on D1S80 and HLA-DQA1 loci. Am
J Hum Biol 14:257263.
Chakraborty R. 1986. Gene admixture in human pop-
ulations: Models and predictions. Yearbook Phys
Anthropol 29:143.
Chakraborty R. 1992. Sample size requirements for
addressing the population genetic issues for forensic
use of DNA typing. Hum Biol 64:141159.
Chakraborty R, Jin L. 1992. Heterozygote deficiency,
population substructure and their implications in
DNA fingerprinting. Hum Genet 88:267272.
Chakraborty R, Weiss KM. 1988. Admixture as a tool
for finding linked genes and detecting that difference
from allelic association between loci. Proc Natl Acad
Sci USA 85:91199123.
Crawford MH, Workman PL, McLean C, Lees FC. 1976.
Admixture estimates and selection in Tlaxcala. In:
Crawford MH, editor. The Tlaxcaltecans, prehistory,
demography, morphology and genetics. Publications
in Anthropology 7. Lawrence, KS: University of
Kansas, p 161168.
Deka R, Shriver MD, Yu LM, Heidreich EM, Jin L,
Zhong Y, McGarvey ST, Agarwal SS, Bunker CH,
Miki T, Hundrieser J, Yin S-J, Raskin S, Barrantes R,
Ferrell RE, Chakraborty R. 1999. Genetic variation at
twentythree microsatellite loci in sixteen human2
populations. J Genet 78:99121.
DEPD. 1978. Direccion de Estadstica y Procesamiento
de Datos del Gobierno de Nuevo Leon. Monterrey,
Nuevo Leon, Mexico.
Edwards A, Hammond H, Jin L, Caskey CT, Chakra-
borty R. 1992. Genetic variation at five trimeric and
tetrameric repeat loci in four human population
groups. Genomics 12:241253.
Elston RC. 1971. The estimation of admixture in racial
hybrids. Ann Hum Genet 35:917.
Evett IW, Gill P. 1991. A discussion of the robustness of
methods for assessing the evidential value DNA sin-
gle locus profiles in crime investigations. Electro-
phoresis 12:226230.
Gamero JJ, Romero JL, Gonzalez JL, Arufe MI, Cuesta
MI, Corte-Real F, Carvalho M, Anjor MJ, Vieira DN,
Vide MC. 2000. A study of ten short tandem repeat
system: African immigrant and Spanish population
data. Forensic Sci Int 110:167177.
Guo SW, Thompson EA. 1992. Performing the exact test
of Hardy-Weinberg preparation for multiple alleles.
Biometrics 48:361372.
Jones DA. 1972. Blood sample: probability of discrimi-
nation. J Forensic Sci Soc 12:355359.
Krane DE, Allen RW, Sawyer SA, Petrov DA, Hartl
DL. 1992. Genetic differences at four DNA typing
loci in Finnish, Italian, and mixed Caucasian pop-
ulations. Proc Natl Acad Sci USA 89:10583
10587.
Li CC. 1976. First course in population genetics. Pacific
Grove, CA: Boxwood Press.
Lisker R. 1981.Estructura genetica de la poblacion
Mexicana: Aspectos medicos y antropologicos. Mexico,
DF: Salvat Mexicana de Ediciones.
Lisker R, Babinsky V. 1986. Admixture estimates in
nine Mexican Indian groups and five East coast lo-
calities. Rev Invest Clin 38:145149.
438 R.M. CERDA-FLORES ET AL.

Montemayor-Hernandez A. 1971. Historia de Monter- 18 0.1739 0.1453 0.0870

rey. Mexico: Asociacion de Editores de Monterrey, AC.
Nei M. 1978. Estimation of average heterozygosity and 19 0.0761 0.0664 0.0652
genetic distance from a small number of individuals. 20 0.0254 0.0137 0.0109
Genetics 89:583590. 21 0.0000 0.0000 0.0217
NRC. 1996. National Research Council. The evaluation 22 0.0000 0.0000 0.0109
of forensic DNA evidence. Washington, DC: National
Academy Press. (n) 138 874 46
Ohno Y, Sebetan JM, Akaishi S. 1982. A simple method
for calculating the probability of excluding paternity FGA
with any number of codominant alleles. Forensic Sci 18 0.0145 0.0092 0.0000
Int 19:9398. 19 0.0688 0.1785 0.0761
Rangel Villalobos H, Rivas F, Torres-Rodrguez M, Ja-
loma-Cruz AR, Gallegos-Arreola MPO, Lopez-Satow 20 0.1413 0.1076 0.0326
J, Cantu JM, Figuera LE. 1999. Allele frequency 21 0.1522 0.1144 0.1087
distributions of six Amp-FLPS (D1S80, APO-B, VWA, 22 0.1848 0.0904 0.1848
TH01, CSF1PO and HPRTB) in a Mexican popula- 23 0.1703 0.0664 0.1848
tion. Forensic Sci Int 105:125129.
Roff DA, Bentzen P. 1989. The statistical analysis of 24 0.1667 0.1316 0.1739
mitochondria DNA polymorphisms: v2 and the prob- 25 0.0652 0.1808 0.1196
lem of small samples. Mol Biol Evol 6:539545. 26 0.0290 0.0881 0.0326
Shriver MD, Smith MW, Jin L, Marcini A, Akey JM,
Deka R, Ferrell RE. 1997. Ethnic-affiliation estima-
27 0.0036 0.0332 0.0326
tion by use of population-specific DNA markers. Am J 28 0.0036 0.0000 0.0544
Hum Genet 60:957964. 30 0.0000 0.0000 0.0000
Ward RH, Sing CF. 1970. A consideration of the power (n) 138 874 46
of the v2 test to detect inbreeding effects in natural
populations. Am Nat 104:355365. D8S1179
Weir BS. 1996. Genetic data analysis. II: Methods for
discrete population genetic data. Sunderland, MA: 8 0.0072 0.0000 0.0000
Sinauer Associates. 9 0.0181 0.0080 0.0000
Zaykin D, Zhivotovsky L, Weir BS. 1995. Exact tests for 10 0.0978 0.1373 0.0000
association between alleles at arbitrary number of
loci. Genetica 96:169178.
11 0.1087 0.0503 0.0106
12 0.0978 0.0927 0.1170
13 0.2645 0.3249 0.2553
14 0.2391 0.2723 0.3723
APPENDIX 15 0.1014 0.0995 0.2128
16 0.0543 0.0137 0.0319
Allele frequencies for 10 STR loci in three 17 0.0109 0.0000 0.0000
ancestral populations that contributed to 18 0.0000 0.0011 0.0000
the current gene pool of the Mestizo popu- (n) 138 874 47
lation of Nuevo Leon, Mexico
D21S11
Spain Amerindians Nigerian 24.2 0.0000 0.0000 0.0000
D3S1358 25 0.0000 0.0000 0.0104
11 0.0000 0.0000 0.0104 26 0.0000 0.0011 0.0104
12 0.0000 0.0000 0.0000 27 0.0072 0.0103 0.0417
13 0.0109 0.0080 0.0000 28 0.1667 0.0526 0.2813
14 0.0797 0.0606 0.1354 29 0.2246 0.1533 0.1771
15 0.2283 0.6533 0.3125 30 0.2971 0.4966 0.1771
16 0.2427 0.1682 0.3125 31 0.1521 0.1773 0.1250
17 0.2174 0.0767 0.1771 32 0.1051 0.0870 0.0938
18 0.1993 0.0320 0.0521 33 0.0072 0.0000 0.0000
19 0.0217 0.0011 0.0000 33.2 0.0398 0.0000 0.0208
20 0.0000 0.0000 0.0000 34 0.0000 0.0217 0.0104
(n) 138 874 48 34.2 0.0000 0.0000 0.0000
35 0.0000 0.0000 0.0313
VWA 36 0.0000 0.0000 0.0208
11 0.0000 0.0000 0.0000 (n) 138 0.874 48
13 0.0000 0.0011 0.0000
14 0.1159 0.0446 0.0870 D18S51
15 0.1667 0.0355 0.3044 9 0.0072 0.0000 0.0000
16 0.2681 0.4394 0.2391 10 0.0326 0.0023 0.0000
17 0.1739 0.2540 0.1739 11 0.0145 0.0080 0.0106
 STR IN A MEXICAN MESTIZO POPULATION 439

Spain Amerindians Nigerian 13 0.1159 0.1384 0.1277

14 0.0326 0.0160 0.0638
12 0.1811 0.1087 0.0957 15 0.0000 0.0011 0.0106
13 0.0978 0.2391 0.0319 (n) 138 874 47
14 0.1522 0.1590 0.0532
15 0.1196 0.0824 0.2021 D7S820
16 0.1304 0.2059 0.0957 6 0.0000 0000 0.0106
17 0.1232 0.1178 0.2234 7 0.0254 0046 0.0106
18 0.0471 0.0378 0.1064 8 0.1123 0950 0.2340
19 0.0507 0.0183 0.0851 9 0.1449 0.0286 0.1064
20 0.0217 0.0092 0.0638 10 0.2572 0.1773 0.3404
21 0.0145 0.0092 0.0213 11 0.2246 0.4073 0.2128
22 0.0072 0.0011 0.0000 12 0.1739 0.2643 0.0638
23 0.0000 0.0011 0.0106 13 0.0580 0.0217 0.0213
24 0.0000 0.0000 0.0000 14 0.0036 0.0011 0.0000
(n) 138 874 47 (n) 138 874 47
D5S818 TH01
7 0.0000 0.1304 0.0104 5 0.0008 0.0011 0.0000
8 0.0109 0.0206 0.0417 6 0.2247 0.2174 0.1280
9 0.0398 0.0114 0.0208 7 0.1517 0.4805 0.5290
10 0.0688 0.0824 0.1563 8 0.1382 0.0492 0.1570
11 0.3043 0.5515 0.2188 9 0.4709 0.0458 0.1860
12 0.3732 0.1236 0.3333 10 0.0126 0.2059 0.0000
13 0.1956 0.0778 0.1979 11 0.0008 0.0000 0.0000
14 0.0072 0.0023 0.0208 (n) 1881 874 102
15 0.0000 0.0000 0.0000 Number of individuals (n) scored for the
(n) 138 874 48 locus. Spanish data are from Gamero et al.
D13S317 (2000) and http://www.uni-Duesseldorf.
8 0.1340 0.0240 0.0213 de/WWW/MedFak/Serology/database.html.
9 0.0725 0.2529 0.0000 Amerindian data are from Budowle et al.
10 0.0725 0.1030 0.0213 (2001) Nigerian (African) data are from
11 0.2754 0.2334 0.2447 Deka et al. (unpublished); see also Deka
12 0.2971 0.2311 0.5106 et al. (1999).