Forensic Science International 133 (2003) 260–265

Short communication
Inferring recent human phylogenies using forensic
STR technology
Diane J. Rowold, Rene J. Herrera*
a
Department of Biological Sciences, Florida International University, University Park Campus, Miami, FL 33199, USA
Received 12 February 2003; accepted 13 February 2003

Abstract

STR loci are characterized by extremely high mutation rates and thus, high levels of length polymorphism both within and
among populations. In addition, much of the observed variation is believed to be nearly selectively neutral. Because of these
features, STRs are ideal markers for genetic mapping, intra-species phylogenetic reconstructions and forensic analysis. In the
present study, we investigate the application of five STR loci (CS1PO, TH01, TPOX, FGA and vWA) routinely used in forensic
analysis for delineating the phylogenetic relationships of 10 human populations representing the three major racial groups
(African–Caribbean, Croatian from the island of Hvar, East Asian, Han Chinese, Italian, Japanese, Portuguese, UK Caucasian,
US Caucasian and Zimbabwe). The resulting tree topology exhibited strong geographic and racial partitioning consistent with
that obtained with mtDNA haplotypes, Y-chromosome markers, SNPs, PAIs (polymorphic Alu insertions) as well as classic
genetic polymorphisms. These findings suggest that forensic STR loci may be particularly powerful tools and provide the
necessary fine resolution for the reconstruction of recent human evolutionary history.
# 2003 Elsevier Science Ireland Ltd. All rights reserved.

Keywords: Recent human evolution; Microsatellite; STR; Phylogenetic analysis

1. Introduction forensic STRs and those typically employed in these phylo-

Microsatellite loci are endowed with several properties
that render them desirable markers for fine scale genetic
mapping [1–3], intra-species phylogenetic reconstructions
[4,5], maternity/paternity determination [6] and forensic
analysis [1,2]. These features include a high level of rela-
tively stable polymorphisms, a dense, uniform chromosomal
distribution as well as short sequence lengths which facil-
itates detection and analysis by PCR and sequencing [6,7].
To be effective tools for forensic applications, STR loci must
also possess high levels of heterozygosity, uniform repeat
motifs, and be compatible with rapid, multiplex PCR-based
assays that provide reliable and unambiguous detection of
alleles (http://www.cstl.nist.gov/div831/strbase).
Numerous microsatellite markers have been used in
phylogenetic analyses of extant human populations [4,5,8].
Although there exists some overlap between the sets of
*
Corresponding author. Tel.: þ1-305-348-1258;
fax: þ1-305-348-1259.
E-mail address: herrerar@fiu.edu (R.J. Herrera).
genetic investigations, it is not extensive and is more a result
of happenstance rather than design or rationale. On the other
hand, the phylogenetic analyses of both Chakraborty et al. [9]
and Budowle and Chakraborty [10] are based upon the
thirteen CODIS loci. However, the phylogenetic assessment
in these two studies are restricted to a simple neighbor joining
(NJ) or an UPGMA procedure (each yielding a single output
tree) and distance measures. Thus, in the absence of a more
extensive phylogenetic evaluation such as one which includes
different phylogenetic approaches (i.e. distance and optim-
ality criterion) and statistical bootstrapping, it is not known, at
this time, whether the use of forensic STR loci is phylogen-
etically informative or hinders meaningful analysis. The
extremely high level of intra-population variation required
of forensic systems coupled with the elevated mutation rates
and the narrow size constraints of microsatellite sequences
[11–13] suggests a rapid saturation of genetic variation at
forensic STR loci and thus, may pose a greater risk of
undetected convergent evolution among some populations.
On the other hand, the evolutionary signal produced by these
markers may be strong enough to prevail against any random

0379-0738/03/$ – see front matter # 2003 Elsevier Science Ireland Ltd. All rights reserved.
doi:10.1016/S0379-0738(03)00073-2
 D.J. Rowold, R.J. Herrera / Forensic Science International 133 (2003) 260–265
noise generated by allelic homoplasy. Furthermore, the fine
scale resolution characteristic of forensic STRs may prove
useful in delineating genetic affinities among closely related
ethnicities (within racial groups). This, in turn, may provide
a phylogenetic window to examine recent evolutionary
events.
The objective of this study is to explore the utility of
forensic STR loci in ascertaining phylogenetic relationships.
We approach this goal by compiling a geographically tar-
geted and racially diverse set of population STR databases
from the forensic literature and analyzing this data both
phenetically and phylogenetically with respect to CSF1PO,
FGA, TH01, TPOX and vWA (the largest set of forensic STR
in common to all 10 populations). The phylogenetic assess-
ment used in this study is rigorous in that it includes two
different phylogenetic approaches, a distance method and
one based upon the maximum likelihood optimality criterion
as well as a statistical bootstrapping procedure involving
1000 replications. The ensuing tree topologies and principle
component (PC) plot are then compared with results obtained
in previous phylogenetic investigations based on microsatel-
lite data as well as other markers including mtDNA, restric-
tion fragment length polymorphisims, Y haplotypes and
polymorphic Alu insertions (PAIs).
It should be mentioned that one of the five STRs included
in this study, FGA, may exhibit somewhat higher hetero-
zygosities than the rest primarily due to the larger array
of length polymorphisms (including both full and partial
numbers of repeats) observed at this locus [current study: 24
for FGA versus 11, 9, 9 and 9 for vWA, CSF1PO, TH01 and
TPOX, respectively; National Institute of Standards and
Technology (http://www.cstl.nist.gov/div831/strbase): 67
for FGA versus 20, 15, 20 and 10 for vWA, CSF1PO,
TH01 and TPOX, respectively; Perkin-Elmer Applied
Biosystems: 14 common length polymorphisms for FGA
versus 11, 10, 10 and 8 for vWA, CSF1PO, TH01 and TPOX,
respectively]. Loci with substantially lower allelic counts
may provide better phylogenetic discrimination in cases in
which there is extensive overlap in allelic types among
populations. However, high heterozygosities and/or numer-
ous observed alleles do not necessarily interfere with the
phylogenetic information content of a locus if the frequency
profiles of the populations are sufficiently different. Further-
more, multiplicity of alleles may reflect a higher incidence
of private alleles which are highly significant in segregating
populations during phylogenetic reconstruction. In the
absence of the phylogenetic analysis itself, it is difficult
to predict which of these scenarios apply.
2. Materials and methods
2.1. Population data
Ten geographically targeted populations were selected to
encompass the major biogeographical zones and represent of
261
the three major races (Africans, Orientals and Caucasians).
The populations examined include: African–Caribbean
(N ¼ 157–192 depending upon locus [15]), Croatian
(N ¼ 206 [18]), East Asian (N ¼ 183–200 depending upon
locus [15]), Han Chinese (N ¼ 500 [16]), Italians (N ¼ 223
[21]), Japanese (N ¼ 206 [17]), Portuguese (N ¼ 153 [20]),
UK Caucasian (N ¼ 192–217 depending upon locus [15]),
US Caucasian (N ¼ 151 [19]) and Zimbabwe (N ¼ 104
[14]).
The five STR loci analyzed in the current investigation
(CSF1PO, FGA, TH01, TPOX and vWA) contain tetrameric
core repeat sequences and reside on different autosomes. All
10 populations are in Hardy–Weinberg equilibrium for each
locus (data not shown).
2.2. Statistical analyses
Three separate analyses were executed, each based upon
the STR allelic frequency profiles of the 10 populations
listed above. In the first, distance values were estimated
using Nei’s formula [22], and a phylogeny was inferred by
the neighbor joining (NJ) option in PHYLIP version 3.5c
[23]. A second phylogenetic reconstruction was based on
maximum likelihood (ML) and the STR frequency distribu-
tion (CONTML in PHYLIP version 3.5c [23]). A principal
component (PC) analysis generated by NTSYSpc [24],
constituted the third analysis.
3. Results
The two resulting unrooted radial phylograms (NJ and ML)
are presented in Fig. 1 (a and b, respectively). The edge
lengths displayed in these phylograms indicate the amount of
evolutionary change occurring along each branch. The scores
next to the nodes represent the number of bootstrap replicates
(out of 1000) exhibiting these specific bifurcations.
The NJ phylogram (Fig. 1a) contains three basal groups:
(1) Zimbabwe and African–Caribbean; (2) East Asian,
Han Chinese and Japanese; and (3) the Caucasian cluster
comprised of the Portuguese branch and two monophyletic
units, Italian/Croatian and UK Caucasian/US Caucasian.
Two of the five nodes are accompanied by a bootstrap score
substantially above 50%. These are: (1) the division between
the African–Caribbean/Zimbabwe monophyly and the other
populations (bootstrap ¼ 997); and (2) the vertex separating
the Han Chinese/Japanese monophyly from the East Asians
(bootstrap ¼ 777).
The ML phylogram (Fig. 1b) displays a distinct racial
partitioning that is identical in topology to the NJ tree. As
detected in a rectangular version of the unrooted phylogram
(not shown), the Italians are the first to branch off from the rest
of the Caucasian cluster. The Croatian is the next population to
split off followed by the Portuguese. The two remaining
groups, UK Caucasian and US Caucasian form a monophy-
letic unit with a bootstrap score of 486. As in the NJ tree, the
262 D.J. Rowold, R.J. Herrera / Forensic Science International 133 (2003) 260–265

Fig. 1. (a) Neighbor joining phylogram based on Nei’s distance with bootstrap scores. Consensus bootstrap scores (1000 replicates) were
transferred to the corresponding nodes of the NJ phylogram. These topological divisions and their respective bootstrap scores (BS) are as
follows: the African–Caribbean and Zimbabwe from the Oriental and Caucasian clusters (BS ¼ 997); the Oriental groups from the African and
Caucasian clades (BS ¼ 300); the Caucasian from the African and Oriental clusters (BS ¼ 401); the Han Chinese/Japanese monophyly from
the East Asian (BS ¼ 777); the UK/US Caucasian monophyly from the Italian and Croatian groups (BS ¼ 464). (b) Maximum likelihood
phylogram with bootstrap scores. Consensus bootstrap scores (1000 replicates) were transferred to the corresponding nodes of the ML
phylogram. These topological divisions and their respective bootstrap scores (BS) are as follows: the African–Caribbean and Zimbabwe from
the Oriental and Caucasian clades (BS ¼ 828); the Oriental from the African and the Caucasian clusters (BS ¼ 452); the Caucasian from the
African and Oriental populations (BS ¼ 440); the Han Chinese/Japanese monophyly from the East Asian (BS ¼ 796); the Italian from the
remaining Caucasian populations (BS ¼ 240); the Croatian from the Portuguese, and the UK/US Caucasian monophyly (BS ¼ 214); the UK/
US Caucasian monophyly from the Portuguese (BS ¼ 486).

Han Chinese/Japanese and African–Caribbean/Zimbabwe 4. Discussion

monophylies are both supported by bootstrap values above
50% (with score of 796 and 828, respectively). Although the
arrangement of the Caucasian groups is somewhat different in
the two trees, both contain a terminal UK Caucasian/US
Caucasian subcluster.
The East Asian, Italian, Portuguese and Zimbabwe
populations display relatively short (or non-detectable)
branch lengths in one or both phylogenies. This reflects
the lesser degree of genetic differentiation sustained by these
groups.
A plot of the first and second principal components, which
together constitute 63.5% of the total variability (PC 1,
47.7%; PC 2, 15.8%) is presented in Fig. 2. There is a
substantial separation of the African groups from the two
other racial clusters (Oriental and Caucasian) along the x-
axis (PC 1). The second principle component (PC 2) exhibits
a relatively clean division between the Caucasians and the
remaining populations with the exception of the East Asians,
an Oriental group that segregates near the Caucasians. Over-
all, however, the racial assemblages are distinct and parallel
the biogeographical associations depicted in both the NJ and
ML phylograms.
The CONTML option in PHYLIP is a ML program based
upon the assumption that differences in allele frequency
arise solely from the random action of genetic drift (Brow-
nian motion model) upon the diverged populations [23]. In
contrast, the NJ algorithm constructs a branching order from
a matrix of estimated genetic distances [22]. This distance
measure, which assumes both mutation as well as genetic
drift, is expected to increase linearly with time if several
assumptions hold [22,23]. The close similarity in basal
cluster patterns (i.e., among the three racial groups versus
the more terminal intra-racial associations) between the NJ
and ML trees may indicate that, the impact of the genetic
drift component of these distance estimations eclipses that of
mutation. Although, the European branching pattern differs
slightly between the two trees, both are fully resolved
(according to the rectangular versions of the two phyloge-
nies which are not shown). The variation in the Caucasian
branching order between the two trees may also be attribu-
table, to the limited power provided by the small number of
loci and/or the limited genetic uniqueness of European
groups (Cavalli-Sforza et al., 1994).
 D.J. Rowold, R.J. Herrera / Forensic Science International 133 (2003) 260–265
Fig. 2. Plot of principle components 1 (47.7%) and 2 (15.8%).
In both trees, the length of the African–Caribbean/Zim-
babwe branch is much longer than that of any other
group. The patristic separation of the Africans from the
remaining groups is also reflected, in the PC plot (Fig. 2). In
contrast with the large inter-nodal distance of the African–
Caribbean/Zimbabwe monophyly, the East Asians display a
branch length of zero in both phylograms and lie near the
patristic origin suggesting a genetic affinity to the non-
Oriental populations. This is supported by the East Asian’s
medial position in the PC plot (Fig. 2).
The Fst [25] NJ phylogram of Perez-Lezaun et al. [26],
generated with 20 STRs, is largely consistent with that of the
current study in that it also displays the generally accepted
clustering of racial groups. On the other hand, the second
best tree, reconstructed using the Dsw distance (and the same
20 STRs), exhibits a trifurcation composed of two Caucasian
groups and the African lineage. In contrast, both the NJ and
ML phylogenies of the current study, which are based on
only five forensic STR loci versus the 20 STRs of Perez-
Lezaun et al. [26], depict a clean separation among all inter-
and intra-racial groups.
Overall, the trees generated by the five forensic STR
tetrameric loci are in accordance with those derived from
much larger sets of microsatellites (30 loci each for Bow-
cock et al. [4], Jorde et al. [5] and 60 in Jorde et al. [27]).
263
Strong racial partitioning is evident in phylogenies inferred
from these studies as well. The phylogenetic reconstruction
of Jorde et al. [5] assimilates frequency data from 30
tetranucleotide STR loci and 13 populations. The level of
intra-racial resolution is similar to the NJ and ML phylo-
grams of the current analysis. In a subsequent study, Jorde
et al. [27] performed a principle coordinates analysis based
on the mean allele sizes of 60 STR loci (including the 30
STR loci used in Jorde et al. [5]) and 15 human populations
which generated a racial cluster pattern highly concordant
with our results. The NJ phylogeny reported by Bowcock
et al. [4], a chord distance [28] reconstruction based on 30
dinucleotide loci and 10 globally dispersed populations,
exhibits a topology that is highly congruent (with respect
to the major racial groups) with the mid-rooted versions (not
shown) of both forensic STR phylogenies (NJ and ML).
The NJ and ML deudograms of the current analysis were
also compared to those obtained with other genetic markers
(mtDNA in [5] and PAIs in [29]). Overall, the strong racial
partitioning in the forensic STR phylograms are consistent
with that inferred from bi-allelic frequency data involving
six PAIs from a globally distributed set of 47 populations
[29]. However, the fit between the present results and that
derived from an analysis involving the hyper-variable
sequence-2 of mtDNA [5] is less satisfactory since the
264 D.J. Rowold, R.J. Herrera / Forensic Science International 133 (2003) 260–265
groups in the mtDNA topology do not partition as cleanly
along racial lines. This is evidenced by the placement of the
two Caucasian groups (French and northern European) in a
monophyletic clade with the Chinese.
There are several potential difficulties associated with
using STRs for phylogenetic inference. These include: the
erosion of genetic differentiation among taxa due to high
mutation rates coupled with allelic size constraints [11–13],
lack of an appropriately detailed model(s) of STR evolution
and corresponding analytical computer programs, as well as
mutation rate heterogeneity among alleles [30]. Nonetheless,
it is encouraging to note that despite these limitations, we
were able to obtain topologies with a strong racial partition-
ing and a 100% resolution of intra-racial groups using
two different methods of phylogenetic reconstruction. It is
possible that the hyper-variability exhibited by these STR loci
provides the fine resolution required to observe the phylo-
genetic events that shaped very recent human evolution.
References
[1] A. Edwards, H. Hammond, L. Jin, C. Caskey, R. Chakraborty,
Genetic variation at five trimeric and tetrameric tandem
repeat loci in four human population groups, Genomics 12
(1992) 241–253.
[2] A. Edwards, A. Civitello, H. Hammond, C. Caskey, DNA
typing and genetic mapping with trimeric and tetrameric
tandem repeats, Am. J. Hum. Genet. 49 (1991) 746–756.
[3] C. Hearne, S. Ghosh, J. Todd, Microsatellites for linkage
analysis of genetic traits, Trends Genet. 8 (1992) 288–295.
[4] A. Bowcock, A. Ruiz-Linares, J. Tomfohrde, E. Minch, J.R.
Kidd, L.L. Cavall- Sforza, High resolution of human
evolutionary trees with polymorphic microsatellites, Nature
368 (1994) 455–457.
[5] L. Jorde, M. Bamshad, W. Watkins, R. Zenger, A. Fraley, P.
Krakowiak, K. Carpenter, H. Soodyall, T. Jenkins, A. Rogers,
Origins and affinities of modern humans: a comparison of
mitochondrial and nuclear genetic data, Am. J. Hum. Genet.
57 (1995) 523–538.
[6] H. Hammond, L. Jin, Y. Zhong, C. Caskey, R. Chakraborty,
Evaluation of 13 short tandem repeat loci for use in personal
identification applications, Am. J. Hum. Genet. 55 (1994)
175–189.
[7] C. Kimpton, P. Gill, A. Walton, A. Urquhart, E. Millican, M.
Adams, Automated DNA profiling employing multiplex
amplification of short tandem repeat loci, PCR Methods
Appl. 3 (1993) 13–22.
[8] R. Deka, M. Shriver, L. Yu, R. Ferrell, R. Chakraborty, Intra-
and inter-population diversity at short tandem repeat loci in
diverse populations in the world, Electrophoresis 16 (1995)
1659–1664.
[9] R. Chakraborty, D. Stivers, B. Su, Y. Zhong, B. Budowle, The
utility of short tandem repeat loci beyond human identifica-
tion: implications for development of new DNA typing
systems, Electrophoresis 20 (1999) 1682–1696.
[10] B. Budowle, R. Chakraborty, Population variation at the
CODIS core short tandem repeat loci in Europeans, Legal
Med. 3 (2001) 29–33.
[11] M. Nauta, F. Weissing, Constraints on allele size at
microsatellite loci: implications for genetic differentiation,
Genetics 143 (1996) 1021–1032.
[12] J. Weber, C. Wong, Mutation of human short tandem repeats,
Hum. Mol. Genet. 2 (1993) 1123–1128.
[13] B. Brinkmann, M. Klintschar, F. Neuhuber, J. Huhne, B. Rolf,
Mutation rate in human microsatellites: influence of the
structure and length of the tandem repeat, Am. J. Hum. Genet.
62 (1998) 1408–1415.
[14] B. Budowle, L. Nhari, T. Moretti, S. Kanoyangwa, E.
Masuka, D. Defenbaugh, J. Smerick, Zimbabwe black
population data on the six short tandem repeat loci—
CSF1PO, TPOX, TH01, D3S1358, VWA, and FGA, Forensic
Sci. Int. 90 (1997) 215–221.
[15] J. Thomson, V. Pilotti, P. Stevens, K. Ayres, P. Debenham,
Validation of short tandem repeat analysis for the investigation
of cases of disputed paternity, Forensic Sci. Int. 100 (1999) 1–16.
[16] C. Pu, F. Wu, C. Cheng, K. Wu, C. Chao, J. Li, DNA short
tandem repeat profiling of Chinese population in Taiwan
determined by using an automated sequencer, Forensic Sci.
Int. 97 (1998) 47–51.
[17] T. Yamamoto, R. Uchihi, H. Nozawa, X. Huang, Y. Leong,
M. Tanaka, M. Mizutani, K. Tamaki, Y. Katsumata, Allele
distribution at the nine STR loci—D3S1358, vWA, FGA,
TH01, TPOX, CSF1PO, D5S818, D13S317, and D7S820—in
the Japanese population by multiplex PCR and capillary
electrophoresis, J. Forensic Sci. 44 (1999) 167–170.
[18] I. Martinovic, L. Barac, I. Furac, B. Janicijevic, M. Kubat, M.
Pericic, B. Vidovic, P. Rudan, STR polymorphism in the
population of the island of Hvar, Hum. Biol. 71 (1999) 341–
352.
[19] T. Kupferschmid, T. Calicchio, B. Budowle, Maine Caucasian
population DNA database using twelve short tandem repeat
loci, J. Forensic Sci. 44 (1999) 392–395.
[20] S. Santos, B. Budowle, J. Smerick, K. Keys, T. Moretti,
Portuguese population data on the six short tandem repeat
loci: CSF1PO, TPOX, TH01, D3S1358, VWA and FGA,
Forensic Sci. Int. 83 (1996) 229–235.
[21] L. Garofano, M. Pizzamiglio, C. Vecchio, G. Lago, T. Floris,
G. D’Errico, G. Brembilla, A. Romano, B. Budowle, Italian
population data on the thirteen short tandem repeat loci:
HUMTH01, D21S11, D18S51, HUMVWFA31, HUMFIBRA,
D8S1179, HUMTPOX, HUMCSF1PO, D16S539, D7S820,
D13S317, D7S818, D3S1358, Forensic Sci. Int. 97 (1998)
53–60.
[22] M. Nei, Genetic distance between populations, Am. Nat. 106
(1972) 283–292.
[23] J. Felsenstein, Phylogeny Inference Package (PHYLIP)
version 3.5c/ distributed by the author, Department of
Genetics, University of Washington, Seattle, 1993.
[24] F.J. Rohlf, NTSYSpc, Version 2.1 (2002), Exeter Publishing
Ltd., Setauket, NY.
[25] S. Wright, The genetical structure of populations, Ann.
Hugenics 15 (1951) 323–354.
[26] A. Perez-Lezaun, F. Calafell, E. Mateu, D. Comas, R. Ruiz-
Pacheco, J. Bertranpetit, Microsatellite variation and the
differentiation of modern humans, Hum. Genet. 99 (1997) 1–7.
[27] L. Jorde, A. Rogers, M. Bamshad, W. Watkins, P. Krakowiak,
S. Sung, J. Kere, H. Harpending, Microsatellite diversity and
the demographic history of modern humans, Proc. Natl. Acad.
Sci. U.S.A. 94 (1997) 100–3103.
 D.J. Rowold, R.J. Herrera / Forensic Science International 133 (2003) 260–265 265

[28] L.L. Cavalli-Sforza, A. Edwards, Phylogenetic analysis.
Models and estimation procedures, Am. J. Hum. Genet. 19
(Suppl.) (1967) 233.
[29] G. Antunez-de-Mayolo, P. Antunez-de-Mayolo, A. Antunez-
de-Mayolo, S. Papiha, M. Hammer, E. Yunis, J. Yunis, C.
Damodaran, M. de Pancorbo, J. Caeiro, V. Puzyrev, R.J. Herrera,
Phylogenetics of worldwide human populations as determined
by polymorphic Alu insertions (PAIs), Electrophoresis 23 (2002)
3346–3356.
[30] A. Primmer, N. Saino, A. Moller, H. Ellegren, Directional
evolution in germline microsatellite mutations, Nat. Genet. 13
(1996) 391–393.