Asymmetric Male and Female Genetic Histories among Native Americans from Eastern North America

Previous studies have investigated the human population history of eastern North America by examining mitochondrial DNA (mtDNA) variation among Native Americans, but these studies could only reconstruct maternal population history. To evaluate similarities and differences in the maternal and paternal population histories of this region, we obtained DNA samples from 605 individuals, representing 16 indigenous populations. After amplifying the amelogenin locus to identify males, we genotyped 8 binary polymorphisms and 10 microsatellites in the male-speciﬁc region of the Y chromosome. This analysis identiﬁed 6 haplogroups and 175 haplotypes. We found that sociocultural factors have played a more important role than language or geography in shaping the patterns of Y chromosome variation in eastern North America. Comparisons with previous mtDNA studies of the same samples demonstrate that male and female demographic histories differ substantially in this region. Postmarital residence patterns have strongly inﬂuenced genetic structure, with patrilocal and matrilocal populations showing different patterns of male and female gene ﬂow. European contact also had a signiﬁcant butsex-speciﬁcimpactduetoahighlevelofmale-mediatedEuropeanadmixture.Finally,thisstudyaddresseslong-standingquestionsaboutthehistoryofIroquoianpopulationsbysuggestingthattheancestralIroquoianpopulationlivedinsouth-easternNorthAmerica.


Introduction
To date, most phylogeographic studies have used mitochondrial DNA (mtDNA) to reconstruct the demographic and evolutionary histories of populations (Avise 2000;Hare 2001;Ballard and Whitlock 2004). Although this approach has yielded many important insights, inferences from mtDNA may not always be congruent with those based on other genetic loci. Stochastic processes and natural selection can affect individual loci differently than the genome (or population) as a whole, so analyses of only a single locus may result in false inferences about population history or the relationships between populations (Moore 1995;Hare 2001;Williams et al. 2002;Ballard and Whitlock 2004). In addition, because mtDNA is maternally inherited, it reflects only maternal relationships and female patterns of gene flow and migration (Avise et al. 1987). If male and female demographic histories have differed in any way, mtDNA will not tell the complete story.
Several recent studies have investigated whether male and female demographic histories differ in humans. Although one multilocus study detected no male-female differences on a global scale (Wilder et al. 2004), most comparisons of mtDNA and Y chromosome variation have identified sex-specific patterns of genetic diversity. Many of these comparative studies found less mtDNA differentiation than Y chromosome differentiation, suggesting that females have experienced higher rates of migration and gene flow compared with males and/or lower rates of genetic drift due to sex differences in effective population sizes (Seielstad et al. 1998;Perez-Lezaun et al. 1999;Fagundes et al. 2002;Oota et al. 2002;Dupanloup et al. 2003;Malyarchuk et al. 2004;Nasidze et al. 2004). However, some South American populations exhibit no male-female differences (Bortolini et al. 2002;Fuselli et al. 2003), and some North and Central American populations exhibit the opposite pattern, indicating higher male migration and/or larger male effective population sizes (Bortolini et al. 2002). A variety of factors may contribute to these sex-specific patterns, including sex-biased admixture among populations (Mesa et al. 2000;Wilson et al. 2001;Wen et al. 2004;Wood et al. 2005) and population-specific aspects of social structure, such as postmarital residence patterns (Seielstad et al. 1998;Oota et al. 2001;Malyarchuk et al. 2004;Nasidze et al. 2004;Hamilton et al. 2005).
To further elucidate the causes of sex-specific diversity patterns in humans, we compare mtDNA and Y chromosome variation among Native Americans from eastern North America. Eastern North America is an ideal setting for this investigation because the indigenous populations exhibit diverse cultures, languages, and histories. Both matrilineal and patrilineal social systems are common, and 4 principal language families (Algonquian, Siouan, Muskogean, and Iroquoian) are represented (Campbell 1997).
Comparative linguistic and archeological records suggest that Algonquian-speaking people moved into the Great Lakes region from western North America before expanding across eastern North America (Siebert 1967;Denny 1991;Fiedel 1991;Goddard 1994). Siouan populations likely lived in the Southeast or Midwest at first, but the Algonquian expansion pushed them into the Plains (Chafe 1976;Campbell 1997). Muskogean populations are found only in southeastern North America (Haas 1941), whereas Iroquoian-speaking populations are more geographically dispersed. The Iroquoian language family contains a northern branch of several closely related languages that are spoken in the lower Great Lakes region and a distantly related southern branch (Cherokee) in southern Appalachia. Given this geographic distribution, the history of the Iroquoianspeaking populations has been the subject of particularly intense debate. Some scholars cite linguistic and archeological evidence for ancestral Iroquoian lands in the Southeast and a subsequent migration to the Northeast (Parker 1916;Snow 1980Snow , 1995Snow , 1996, whereas others favor a model of in situ development for the northern Iroquoians and a later southward migration of the Cherokee (MacNeish 1952;Lounsbury 1978;Crawford and Smith 1996).
Although mtDNA studies have not resolved questions about the history of Iroquoian populations, they have helped reconstruct other aspect of eastern North American prehistory. Malhi et al. (2001) found evidence of extensive gene flow and admixture among Algonquian and Siouan populations in the Northeast, and ancient DNA studies of prehistoric burial populations confirmed that this pattern of regional continuity predates European contact (Stone and Stoneking 1998;Mills 2003;Bolnick 2005;Shook 2005). In contrast, Native American populations from southeastern North America exhibitgreater geneticdifferentiation (Bolnick and Smith 2003). Because the Southeast also exhibits lower overall levels of mtDNA diversity, this differentiation likely reflects genetic drift due to the population decline following European contact (Bolnick and Smith 2003). However, because these inferences are all based only on mtDNA, it is important to assess any differences in male and female demographic histories before drawing firm conclusions about the population history of eastern North America.
In this study, we examine variation on the Y chromosome to explore such differences and to improve our understanding of eastern North American prehistory. The male-specific region of the Y chromosome is strictly paternally inherited, so it can be used to assess paternal relationships and male patterns of gene flow and migration (Hurles and Jobling 2001). Previous studies of Native American Y chromosome variation included few eastern North American populations, and virtually all such studies focused on issues related to the initial peopling of the continent (Underhill et al. 1996;Bianchi et al. 1998;Karafet et al. 1999;Ruiz-Linares et al. 1999;Santos et al. 1999;Lell et al. 2002;Zegura et al. 2004) rather than on the subsequent, region-specific history of the past 10,000 years. This study therefore characterizes Y chromosome variation among Native Americans from eastern North America to investigate 1) the population history of the last 10,000 years, 2) the relative influences of geography, language and culture on genetic structure in this region, and 3) male-female differences in gene flow and migration.

Population Samples
We obtained samples from 605 individuals, representing 16 eastern North American populations ( fig. 1 and table 1). All individuals reported having at least 75% ancestry from one of these populations. The samples were previously analyzed for mtDNA markers Smith 1996, 1997;Smith et al. 1999;Malhi et al. 2001;Bolnick and Smith 2003;Shook 2005), and detailed information about their sources can be found in those publications. The Office of Human Research Protection at the University of California, Davis approved all sampling protocols.
The populations included in this study can be divided into 2 culture areas, the Northeast and the Southeast. Populations from the northeastern culture area generally exhibited patrilineal kinship systems and patrilocal postmarital residence (male philopatry), whereas populations from the southeastern culture area exhibited matrilineal kinship systems and matrilocal postmarital residence (female philopatry). Historically, these populations also spoke languages belonging to the 4 principal language families in eastern North America (Algonquian, Siouan, Iroquoian, and Muskogean).

Genetic Analyses
We extracted DNA from 200 ll of serum using the QIAamp DNA Blood Kit (Qiagen, Valencia, CA). Because the sex of each individual was not recorded when the samples were collected anonymously, we identified males by screening part of the amelogenin gene that differs in size between the X and Y chromosomes (Sullivan et al. 1993). We then amplified 8 binary polymorphisms (M19, M3, M242, M173, M45, M130, YAP, and Tat) to define Y chromosome haplogroups. Polymerase chain reaction (PCR) amplifications were performed in a 25 ll volume with 1-5 ll of DNA template, 67 mM Tris-Cl (pH 8.8), 16 mM (NH 4 ) 2 SO 4 , 0.01% Tween-20, 0.05 mM of each dNTP, 1.7 mM MgSO 4 , 0.2 mM of each primer, and 0.025 units/ll of Platinum Taq (Invitrogen, Carlsbad, CA). All PCRs were performed with an initial denaturation at 95°C for 3 min and a final extension at 72°C for 5 min. Interim cycling conditions and primer sequences are provided in the Supplementary Material online. PCR products containing the M3, M242, M45, M130, and Tat polymorphisms were digested with the appropriate restriction enzyme (Supplementary Material online), and digestion products were visualized with ethidium bromide on 6% polyacrylamide gels. PCR products containing the M19 and M173 polymorphisms were purified with ExoI (0.25 units/ll) and Montage PCRl96 plates (Millipore, Billerica, MA) before being sequenced at the CA&ES Genomics Facility at the University of California, Davis.
These binary polymorphisms define 7 haplogroups, which are named using the nomenclature recommended by the Y Chromosome Consortium (2002). The phylogenetic relationships among these haplogroups are shown in figure 2. Q-M3*, Q-M242*, and C-M130 are thought to be founding Native American haplogroups (Zegura et al. 2004), whereas Q-M19 represents a post-colonization mutational event that occurred in the Americas (Bortolini et al. 2003). The other haplogroups likely represent recent non-Native American admixture (Zegura et al. 2004 and discussed below).

Statistical Analyses
Unless otherwise noted, we excluded individuals with suspected non-Native American admixture from statistical analyses and performed analyses only on population samples containing more than 2 individuals. Some populations were therefore represented by small samples, but because most statistical analyses take sample size into account, any significant results should be robust to low sample sizes.
Population haplogroup frequencies were compared in pairs using exact tests of population differentiation to test the null hypothesis of identical haplogroup frequencies in the 2 populations being compared. Exact tests were performed using the Arlequin 2.000 software package (Schneider et al. 2000), and P values based on 10,000 Markov steps are reported. P values for the pairwise exact tests were corrected for multiple comparisons with a sequential Bonferroni correction (Rice 1989). We also used Arlequin 2.000 to estimate haplogroup and haplotype diversities (h), mean numbers of pairwise differences among haplotypes (p), and pairwise R ST values. Both h and p were estimated as measures of microsatellite haplotype diversity to ensure that both haplotype frequencies and the molecular differences among haplotypes were taken into consideration. Correlations among the 3 estimates of genetic diversity were measured using Pearson correlations with Bonferroni-corrected probabilities, and the values of h and p for the 2 culture areas were compared using 2-sample t-tests (Zar 1999).
Following Qamar et al. (2002), we calculated weighted means of within-haplogroup R ST (WMWH-R ST ) values as a measure of population differentiation that is less influenced by differences that accumulated between haplogroups before colonization of the Americas. Multidimensional scaling (MDS) analysis was performed using the SYSTAT 9 software package (SPSS), with WMWH-R ST values as variables, to investigate the genetic relationships among populations.
To assess the relative influences of language and culture on genetic structure, we performed an analysis of molecular variance (AMOVA) using Arlequin 2.000. We also performed Mantel tests in Arlequin to assess the impact of geography and language on genetic structure. Great-circle geographic distances between populations were calculated from latitude-longitude data, and linguistic distances were calculated following the method of Zegura et al. (2004) but with the language classification of Campbell (1997). Geographic and linguistic distances are given in the Supplementary Material online. It should be noted that significant P values for the partial Mantel tests should be treated with caution because there is some debate over their accuracy (Raufaste and Rousset 2001;Castellano and Balletto 2002;Rousset 2002).
Differences in the genetic structure of males and females were assessed using the parameter Nm, which is calculated as (1/U ST ) ÿ 1 according to the island migration model for haploid systems (Cavalli-Sforza and Bodmer 1971). This parameter incorporates migration rate, mutation rate, and effective population size, but because the contribution of mutation rate to the Nm parameter may be considered negligible for the genetic systems studied here (Destro-Bisol et al. 2004), different Nm values reflect different migration rates and/or effective population sizes between populations.
In addition, maximum likelihood estimates of male and female migration rates between populations were calculated using the program MIGRATE (Beerli and Felsenstein 2001). We estimated male migration rates using the Y chromosome microsatellite data and a Brownian motion approximation of the stepwise mutation model for microsatellites. MIGRATE performed 10 replicate runs using the adaptive heating scheme with 5 heated chains. Each run contained 10 primary short chains (4,000 genealogies per chain) and 3 primary long chains (40,000 genealogies per chain), and migration rate estimates were averaged across runs. We estimated female migration rates using the previously published mtDNA hypervariable region I sequences from these populations (Lorenz and Smith 1997;Malhi et al. 2001;Bolnick and Smith 2003). Similar MIGRATE conditions were used, but with 15 short chains of 100,000 genealogies each and 3 long chains of 1,000,000 genealogies each. For each pair of populations, we calculated the effective number of migrants per generation (N e m) by summing the unidirectional N e m estimates provided by MIGRATE.

Y Chromosome Haplogroups
Of the 605 samples examined, 261 males and 344 females were identified. Twenty-four male samples contained insufficient quantities of DNA for further analysis, so we analyzed 237 Y chromosomes. We identified 6 of the 7 possible haplogroups, with 37 individuals falling into the ''Other'' category. The remaining 200 individuals represent 59 members of haplogroup Q-M3*, 39 of Q-M242*, 15 of C-M130, 80 of R-M173, 3 of P-M45*, and 4 of DE-YAP. Haplogroup frequencies for all eastern North American populations are given in table 1, and haplogroup diversities (h) are given in table 2. The Northeast exhibits significantly higher levels of haplogroup diversity than the Southeast (P , 0.001).
Four haplogroups occur at frequencies greater than 5% (Q-M3* 5 23.5%, Q-M242* 5 21.0%, C-M130 5 7.8%, and R-M173 5 31.0%), and together they comprise 83.3% of the eastern North American sample. Haplogroup R-M173 likely represents recent (post-1492) European admixture, as may P-M45* (Tarazona- Santos and Santos 2002;Bosch et al. 2003;Zegura et al. 2004). Y chromosomes belonging to haplogroup DE-YAP probably result from recent admixture with individuals of African or European ancestry (Karafet et al. 1999;Lell et al. 2002). Following Bosch et al. (2003), these findings suggest that at least 34.2 6 3% of eastern North American Y chromosomes result from recent admixture. This figure may actually be as high as 47.7 6 3%, if the 37 ''Other'' Y chromosomes represent additional European or African lineages rather than previously unidentified founding Native American lineages. Admixture estimates for the Northeast (36.6-52.5% depending on whether the ''Other'' lineages represent admixture) are higher than those for the Southeast (29.6-38.8%) due to extremely high estimates for 2 of the Chippewa populations (56.9-94.1% for the Turtle Mountain Chippewa and 56.8-78.4% for the Wisconsin Chippewa).
Frequencies of the founding Native American haplogroups (Q-M3*, Q-M242*, and C-M130) also differ between the 2 areas. Populations in the southeastern culture area exhibit similar haplogroup frequencies, with both Muskogean and Iroquoian populations showing high frequencies of haplogroup Q-M3*, moderate frequencies of haplogroup Q-M242*, and low frequencies of haplogroup C-M130 (if present at all). In the northeastern culture area, populations generally exhibit lower frequencies of Q-M3* and higher frequencies of C-M130.
Exact tests of population differentiation confirm these patterns. The haplogroup frequencies of the northeastern and southeastern culture areas are significantly different (P 5 0.001). When individuals are grouped by language family, all comparisons yield statistically significant differences (P 0.010) except the Muskogean-Iroquoian (P 5 0.891) and Algonquian-Siouan (P 5 0.155) comparisons. The Wisconsin Chippewa also differ significantly from the Seminole (P , 0.001), and the Cheyenne/Arapaho differ significantly from the Stillwell Cherokee (P , 0.001). Other comparisons of individual populations are not statistically significant, perhaps because the small sample sizes for some populations provide relatively low statistical power. Alternatively, the Bonferroni correction for multiple comparisons may be too conservative: it assumes that all tests are statistically independent, but these exact tests involve the repeated use of each population.

Y Chromosome Microsatellite Haplotypes
To obtain a more detailed picture of Y chromosome variation in eastern North America, we also genotyped 214 Y chromosomes for 10 Y-specific microsatellite loci. These microsatellites define 173 different haplotypes, of which 147 (84%) are unique. Two haplotypes are shared by Y chromosomes belonging to different haplogroups, resulting in 175 combination haplotypes (see Supplementary Material online). Haplotype diversity (h) is generally high (table 2), with the Southeast exhibiting a statistically higher estimate than the Northeast (P , 0.001). In contrast, the mean number of pairwise differences among haplotypes (p) is higher in the Northeast, but not significantly so (P 5 0.494). Overall, number of pairwise differences is significantly correlated with haplogroup diversity (r 5 0.828, P 5 0.010), but haplotype diversity is not (r 5 0.570, P 5 0.404).
Haplotype median-joining networks show that some population-specific substructure exists, but haplotypes generally cluster by culture area, with different populations from the same culture area sharing closely related haplotypes. For example, the haplotype network for haplogroup Q-M3* ( fig. 3) exhibits one population-specific cluster, containing 2 Sioux haplotypes and several areaspecific clusters of haplotypes. In the haplotype network for haplogroup Q-M242* ( fig. 4), all 19 southeastern individuals cluster together, as do all but 3 northeastern individuals. The haplogroup C-M130 network ( fig. 5) exhibits similar groupings: the haplotype shared by 2 southeastern populations is 10 mutational steps away from a cluster of 7 closely related northeastern haplotypes. This northeastern cluster contains 2 smaller population-specific clusters, one representing 8 Sioux and the other representing 4 Minnesota Chippewa.
In contrast to the Q-M3*, Q-M242*, and C-M130 networks, which all exhibit area-specific clusters of haplotypes,  . 6a). This lack of structure is consistent with the hypothesis that haplogroup R-M173 represents recent (post-1492) European admixture in eastern North America rather than a founding Native American lineage (Tarazona- Santos and Santos 2002;Bosch et al. 2003;Zegura et al. 2004). Comparative data from the YHRD, Zegura et al. (2004), and Bortolini et al. (2003) further support this conclusion: the most common R-M173 haplotype in eastern North America is also the most common R-M173 haplotype in Europe, but this haplotype is rare in Asia (and therefore unlikely to be a founding lineage). In addition, the 62 other R-M173 haplotypes found in eastern North America appear to be rare or absent in Asia, whereas 43 of them are common in Europe and 14 others are one mutational step away from known European haplotypes ( fig. 6b). Five haplotypes are 4-8 mutational steps away from European haplotypes, but they probably also stem from recent European admixture: the median-joining network suggests that they derive from separate European haplotypes ( fig. 6b), whereas one would expect them to cluster together in a distinct clade if they represented a founding lineage from Asia. Thus, few, if any, R-M173 haplotypes in eastern North America are likely derived from founding lineages of the Americas. Comparative data from the YHRD also suggest that most individuals in the ''Other'' category have European Y chromosome haplotypes. Seventeen of the 27 ''Other'' haplotypes are shared with Europeans, and 8 additional haplotypes are one mutational step away from known European haplotypes. These findings indicate that the ''Other'' lineages most likely represent recent non-Native American admixture rather than additional founding Native American lineages. Similarly, the haplotypes in haplogroup DE-YAP match or are closely related to African and European haplotypes, suggesting that they also represent recent non-Native American admixture. fit between the 2-dimensional plot and the source data (WMWH-R ST values). All 5 populations from the southeastern culture area (Choctaw, Chickasaw, Creek, Stillwell Cherokee, and ORC Cherokee) cluster together in the MDS plot, whereas only 2 populations from the northeastern culture area do (the Sioux and Wisconsin Chippewa; the other 3 northeastern populations appear as separate outliers). The MDS analysis therefore suggests that populations in the southeastern culture area share closer paternal relationships than populations in the northeastern culture area.

Analysis of Molecular Variance and Mantel Tests
We performed an AMOVA to investigate the relative influences of culture and language on the genetic structure of eastern North American populations. AMOVAs based on haplogroup frequencies are shown in table 3. When the populations are grouped by culture area (2 groups), most genetic variation is found within populations, but a significant portion of the total genetic variance is due to differences between the 2 culture areas. Variation among populations within the same culture area is not statistically significant. When the populations are grouped by language family (4 groups), within-population variation is again significant, but the language groupings themselves do not account for a statistically significant portion of the genetic variance in eastern North America. AMOVAs based on microsatellite haplotypes generally yield similar results (table 3), but they differ in suggesting that a significant amount of variation exists among populations within the same culture area or language family.
Because culture areas also comprise geographic regions in this study, the effects of culture and geography may be conflated in these AMOVAs. Consequently, we performed Mantel tests based on WMWH-R ST values to calculate the partial correlation between genetic and geographic distances while controlling for the confounding effects of both language and culture (table 4). Genetics and geography are significantly correlated across eastern North America when language is taken into account. However, when culture is also taken into consideration by performing this analysis separately for each culture area, no significant correlations are observed. Consequently, no isolation by distance is detectable within either culture area that cannot be explained by linguistic or cultural differences. Mantel tests also suggest that genetics and language are not correlated in eastern North America (table 4).

Sex-Specific Estimates of Migration and Effective Population Size
In eastern North America, paternally and maternally inherited loci show opposite patterns of genetic differentiation (U ST ) among populations. Y chromosomes indicate high differentiation in the Northeast (U ST 5 0.124) and low differentiation in the Southeast (U ST 5 0.044), whereas mtDNA indicates low differentiation in the Northeast (U ST 5 0.062) and high differentiation in the Southeast (U ST 5 0.192). The U ST values for the Northeast produce a ratio of mtDNA to Y chromosome Nm of 2.142, indicating a female migration rate and/or effective population size that is more than twice that of males. In contrast, the U ST values for the Southeast produce a Nm ratio of 0.194, which suggests that males in the Southeast have had a migration rate and/or effective population size that is approximately 5 times greater than that of females.  We also calculated maximum likelihood estimates of migration rates (table 5). Although the migration rates between specific pairs of populations are quite variable, male migration is generally higher in the Southeast than in the Northeast (average Southeast N e m 5 1.429 vs. average Northeast N e m 5 0.768), whereas female migration is generally higher in the Northeast than in the Southeast (average Northeast N e m 5 11.071 vs. average Southeast N e m 5 3.860).

Discussion
This study presents the first detailed survey of Y chromosome variation among Native Americans from eastern North America. The populations analyzed here represent both the northeastern and southeastern culture areas, as well as the 4 major language families found in eastern North America. When combined with mtDNA data previously collected from these samples Smith 1996, 1997;Smith et al. 1999;Malhi et al. 2001;Bolnick and Smith 2003;Shook 2005), these Y chromosome results shed new light on 1) the effects of language, culture, and geography on genetic variation in this region, 2) the differences between male and female patterns of gene flow and migration, 3) the history of Iroquoian populations, and 4) the impact of European contact in eastern North America.

Effects of Language, Culture, and Geography on Genetic Variation
Because Native American populations from eastern North America can be divided into 4 language families and 2 culture areas, we examined the relative importance of language and culture in shaping the observed patterns of genetic variation. Although languages clearly evolve differently than genes (Bateman et al. 1990;Renfrew 2000), similar languages may indicate common ancestry between populations, and language differences can act as barriers to gene flow (Barbujani 1991). However, based on the AMO-VAs and Mantel tests, language does not seem to have had a significant impact on the structure of genetic variation in eastern North America. Four of the 6 exact tests did indicate statistically significant differences between language family groupings, but those 4 comparisons all involved populations from different culture areas. Language family groupings from the same culture area were not statistically different, suggesting that the effects of language and culture may have been confounded in these tests. Consequently, these analyses provide no clear evidence that language has influenced genetic variation in eastern North America.
Culture, on the other hand, has played an important role in shaping the observed patterns of variation. Populations from the same culture area exhibit more similar Y chromosomes than populations from different areas, and differences between the 2 culture areas account for a significant portion of the total genetic variance in eastern North America. Although culture and geography are partially conflated in this study, the Mantel tests suggest that geography has had no clear effect on the genetic structure of eastern North American populations that is independent of culture and language.
Within the southeastern culture area, populations exhibit similar Y chromosome haplogroup frequencies, share identical or closely related haplotypes, and cluster together in the MDS plot, indicating close paternal relationships among them. A close relationship among the Muskogean populations is not unexpected: archeological, linguistic, and ethnographic evidence suggest that they formed from people who had been living in the Southeast for thousands of years, and they share a variety of cultural traits, closely related languages, and a high level of intermarriage   (Swanton 1922(Swanton , 1946Haas 1941;Hudson 1976Hudson , 1997Knight 1994;Galloway 1995). Within the northeastern culture area, populations exhibit fewer similarities: some Y chromosome haplotypes are closely related, but the populations do not all cluster together in the MDS plot. These results suggest that Native American populations from the Northeast do not share as close paternal relationships.

Male-Female Differences in Gene Flow and Migration
With both Y chromosome and mtDNA data from these populations, it is possible to directly evaluate sex-specific patterns of gene flow and migration. These data show that male and female demographic histories have clearly differed in eastern North America but in opposite ways in the 2 culture areas. In the southeastern culture area, males have experienced higher rates of gene flow/migration than females, whereas females have experienced greater gene flow/migration than males in the northeastern culture area.
Significantly, these patterns of genetic variation correlate precisely with past postmarital residence patterns in eastern North America. Populations from the southeastern culture area exhibited matrilocality (Hudson 1976), in which males moved from their own household or community to that of their wife after marriage. Matrilocal systems therefore facilitated male movement and gene flow each generation while fostering mtDNA differentiation. On the other hand, the populations from the northeastern culture area that we sampled all exhibited patrilocality (Mason 1981), which facilitated female movement and gene flow and augmented Y chromosome differentiation. Thus, postmarital residence clearly had a significant impact on the genetic structure of eastern North American populations. Because the postmarital residence practices of many populations changed during the 19th and 20th centuries (Eggan 1937;Perdue 1989), these results suggest that the effects of past postmarital residence systems may be detectable for at least 5-6 generations after patterns have shifted (assuming a generation length of 30 years). It is important to note, though, that the relationship between postmarital residence and genetic variation will vary across human populations: a clear correlation may be observed in some regions, such as eastern North America, but the pattern of genetic variation in other regions may instead reflect other factors, such as local cultural practices like endogamy (Kumar et al. 2006).

Iroquoian Population History
Although Cherokee mtDNAs differ from those of other southeastern populations (Bolnick and Smith 2003), the Y chromosome data presented here show broad similarities between the Cherokee and the Muskogeanspeaking populations of the Southeast. Both mtDNA and Y chromosome data show differences between the Cherokee and northeastern populations. Because Cherokee Y chromosomes are closely related to those of populations that have lived in southeastern North America for several thousand years, the Cherokee must have lived in the Southeast long enough to develop such extensive similarities through gene flow. These findings suggest that Iroquoian populations may have first lived in the Southeast, with a later Northern Iroquoian migration to the Northeast. mtDNA differences between the Cherokee and other southeastern populations are compatible with this hypothesis because the Cherokee's matrilocal social structure would have restricted female gene flow with other populations and increased mtDNA differentiation. However, to confirm this hypothesis, Northern Iroquoian Y chromosomes would have to be studied to detect parallel similarities with the southeastern Y chromosomes as well.

Impact of European Contact
European contact affected genetic variation in eastern North America in at least 2 different ways: by causing a population decline that resulted in genetic drift and by introducing non-Native American lineages through admixture. Earlier studies suggested that the population decline following European contact did not significantly impact genetic variation in North America (Stone and Stoneking 1998;O'Rourke et al. 2000;Kaestle and Smith 2001), but 2 recent studies concluded otherwise. Bolnick and Smith (2003) found evidence that a genetic bottleneck associated with this decline altered patterns of mtDNA variation in southeastern North America, and Wang et al. (2004) identified further evidence of a relatively recent bottleneck in the southeastern Choctaw population by analyzing 175 genome-wide markers. This bottleneck does not seem to have drastically affected Y chromosome variation in southeastern North America, but the Y chromosome data presented here are nevertheless compatible with such an event for 2 reasons. First, bottlenecks are not expected to affect all loci equally (Hoelzel 1999). Second, because the southeastern populations are matrilocal, effective population sizes for their Y chromosomes are much larger than those for their mtDNA, making genetic drift less likely to affect Y chromosome variation than mtDNA variation in this region.
European contact also influenced genetic variation in indigenous eastern North American populations by introducing non-Native American lineages. In most cases, there is widespread agreement about whether a particular haplogroup represents an ancient Native American lineage or post-1492 admixture, but the status of haplogroup R-M173 has recently been subject to some debate. Some authors have argued that this haplogroup represents a founding Native American lineage (Lell et al. 2002;Bortolini et al. 2003), whereas others suggest that it instead reflects recent European admixture (Tarazona- Santos and Santos 2002;Bosch et al. 2003;Zegura et al. 2004). In eastern North America, the pattern of haplotype variation within this haplogroup supports the latter hypothesis: R-M173 haplotypes do not cluster by population or culture area, as haplotypes in the other founding haplogroups do, and most match or are closely related to R-M173 haplotypes that are common in Europe but rare in Asia. This pattern is opposite than expected if the Native American R-M173 haplotypes were descended from Asian haplotypes and suggests that recent European admixture is responsible for the presence of haplogroup R-M173 in eastern North America. This conclusion is consistent with 2170 Bolnick et al. evidence that European contact also introduced the DE-YAP and ''Other'' lineages into eastern North America.
If this interpretation is correct, eastern North America exhibits a high level of non-Native American admixture (47.7%). This estimate reflects the extremely high level of admixture in the Chippewa populations, which is consistent with ethnographic evidence indicating frequent interactions between the Chippewa and European traders (Rhodes 1982;Camp 1990). Because of the high representation of Chippewa samples in our Y chromosome data set, this estimate of non-Native American admixture in eastern North America may be somewhat exaggerated. Our estimate is substantially greater than the estimate of 17 6 2% paternal non-Native American admixture of Zegura et al. (2004) across North, Central, and South America and is particularly surprising because almost all of the sampled individuals reported being full blooded for a single Native American population. Of the 134 individuals with European, African, or ''Other'' Y chromosomes, only 3 (all R-M173) reported having less than 100% Native American ancestry.
Interestingly, the Y chromosome estimates of non-Native American admixture differ sharply from those based on mtDNA. mtDNA studies of the same samples revealed that few exhibit non-Native American mtDNA (Lorenz and Smith 1996;Smith et al. 1999;Malhi et al. 2001;Bolnick and Smith 2003;Shook 2005). Because the vast majority of non-Native American Y chromosomes appear to be European in origin, this difference suggests a high level of sex-biased admixture in eastern North America, involving European males and Native American females. These results are consistent with ethnographic evidence and with previous comparisons of Y chromosome and mtDNA variation in South America, which identified a similarly asymmetric pattern of mating (Carvajal-Carmona et al. 2000;Mesa et al. 2000;Carvalho-Silva et al. 2001).
Because many Native Americans in eastern North America did not acknowledge their paternal European ancestry, such admixture may predate current genealogical records. Male-mediated European gene flow may have therefore had a greater impact on eastern North American populations than was previously thought. Alternatively, some of the Native Americans who provided samples for this study may have been aware of such ancestry but interpreted phrases like ''full bloodedness'' and ''percentage ancestry'' as reflecting their level of social identification with a particular tribe rather than their genetic history (K Tall-Bear, personal communication). Consequently, differences between the Native American and scientific concepts of kinship and ancestry should be explored further before reaching any firm conclusions about the discrepancy between reported and actual genetic ancestry in this study.
Finally, present-day genetic variation in eastern North America may also reflect other events from the centuries following European contact, such as forced migrations, the formation of new Native American communities, and the establishment of the reservation system. Some populations became more isolated as a result of these events: the rigidity of the reservation system, for example, restricted gene flow among the various Chippewa reservations (Shook 2005). In other cases, historical events facilitated population interactions: new coalescent societies formed in both the northern Plains and the Southeast following periods of population decline (Taylor 1977;Hudson 1997;Bolnick and Smith 2003), and the geographic proximity of many reservations in Oklahoma may have increased gene flow among the southeastern populations. Historical events may have therefore contributed to the observed patterns of Y chromosome variation (e.g., closer paternal relationships among populations from the Southeast than among those from the Northeast). However, such events cannot explain the opposite pattern observed in the mtDNA data (Bolnick and Smith 2003), so they must have had less effect on eastern North American genetic variation than past patterns of postmarital residence.

Conclusion
This study demonstrates the importance of using multiple loci when reconstructing population history. In eastern North America, male and female demographic histories differ significantly as a result of postmarital residence patterns and European influences. Geography may have also played a role in shaping population interactions. Although the correlation between geography and culture area in this study makes it difficult to completely separate the two, this research suggests that sociocultural factors have had the greatest impact on the genetic structure of indigenous eastern North American populations.

Supplementary Material
1) The primers, PCR conditions, and restriction enzymes used to analyze the Y chromosome binary markers, 2) the geographic and linguistic distances used in the Mantel tests, and 3) the binary marker/microsatellite haplotypes observed in each population are available at Molecular Biology and Evolution online (http://www.mbe. oxfordjournals.org/).