New Research In
Physical Sciences
Social Sciences
Featured Portals
Articles by Topic
Biological Sciences
Featured Portals
Articles by Topic
- Agricultural Sciences
- Anthropology
- Applied Biological Sciences
- Biochemistry
- Biophysics and Computational Biology
- Cell Biology
- Developmental Biology
- Ecology
- Environmental Sciences
- Evolution
- Genetics
- Immunology and Inflammation
- Medical Sciences
- Microbiology
- Neuroscience
- Pharmacology
- Physiology
- Plant Biology
- Population Biology
- Psychological and Cognitive Sciences
- Sustainability Science
- Systems Biology
Evaluating the Farming/Language Dispersal Hypothesis with genetic variation exhibited by populations in the Southwest and Mesoamerica
Edited by Francisco Mauro Salzano, Instituto de Biociencias, Porto Alegre, Rio Grande do Sul, Brazil, and approved February 25, 2010 (received for review June 23, 2009)

Abstract
The Farming/Language Dispersal Hypothesis posits that prehistoric population expansions, precipitated by the innovation or early adop-tion of agriculture, played an important role in the uneven distribution of language families recorded across the world. In this case, the most widely spread language families today came to be distributed at the expense of those that have more restricted distributions. In the Americas, Uto-Aztecan is one such language family that may have been spread across Mesoamerica and the American Southwest by ancient farmers. We evaluated this hypothesis with a large-scale study of mitochondrial DNA (mtDNA) and Y-chromosomal DNA vari-ation in indigenous populations from these regions. Partial correlation coefficients, determined with Mantel tests, show that Y-chromosome variation in indigenous populations from the American Southwest and Mesoamerica correlates significantly with linguistic distances (r = 0.33–0.384; P < 0.02), whereas mtDNA diversity correlates significantly with only geographic distance (r = 0.619; P = 0.002). The lack of correlation between mtDNA and Y-chromosome diversity is consistent with differing population histories of males and females in these regions. Although unlikely, if groups of Uto-Aztecan speakers were responsible for the northward spread of agriculture and their languages from Mesoamerica to the Southwest, this migration was possibly biased to males. However, a recent in situ population expansion within the American Southwest (2,105 years before present; 99.5% confidence interval = 1,273–3,773 YBP), one that probably followed the introduction and intensification of maize agriculture in the region, may have blurred ancient mtDNA patterns, which might otherwise have revealed a closer genetic relationship between females in the Southwest and Mesoamerica.
Scholars have long noted the close connections, both past and present, between indigenous cultures of the U.S. Southwest and Mesoamerica (1), which are exemplified by the spread of maize from Mesoamerica to the Southwest and the distribution of speakers of Uto-Aztecan (UA) languages across the regions (2). It has been proposed that UA speakers were, in fact, responsible for the northward spread of maize cultivation (3–5). However, previous genetic and morphologic studies have failed to provide direct evidence for an ancient spread of UA speakers (6–8).
The American Southwest is both a cultural and geographic region that spans from the Mexican states of Baja California and Durango in the south to Southern Utah and Colorado in the north and west to east from approximately Las Vegas, NV to Las Vegas, NM (9). The southern boundary, however, may extend as far south as the Mexican states of Nayarit and Jalisco, encompassing an area called the Greater Southwest (10). Mesoamerica is neither a geographic region nor a socio-political unit (11), but rather, it is an area occupied by populations that shared a suite of cultural characteristics, which was first defined by Kirchhoff (12). The northern limit of Mesoamerica has fluctuated throughout prehistory, both expanding opportunistically and contracting under the threats of nomadic tribes to the north, and at its peak, it overlapped the southern frontier of the Southwest (11, 12).
The earliest widely accepted evidence of maize comes from San Marcos Cave in the Tehuacán region of Oaxaca, accelerator mass spectrometry dated to ∼5,600 years before present (YBP) (dates are presented as calibrated calendar years unless otherwise noted) (13). However, recent phytolith data have pointed to an origin in the lowland tropics of Tabasco more than 7,000 YBP (14). Whereas the origin and timing of domestication continues to be debated in Mesoamerica, the introduction of maize into the Southwest before 4,000 YBP is no longer disputed (15, 16). The early appearance of this cultigen in the Southwest, soon after its widespread use in Mesoamerica, has been used to support the hypothesis that maize cultivation spread with humans from Mesoamerica and therefore, was not spread primarily through cultural diffusion (3–5).
Because UA is the only language family to extend across Mesoamerica and the Southwest, members of this language family may have played a prominent role in the interactions that took place between the two regions. The structure of the language family and its diversity as well as reconstructed Proto-Uto-Aztecan (PUA) vocabulary suggest a northern origin for UA somewhere in the Southwest or Southern California (2, 17–20). Recently, Merrill et al. (16) used linguistic data to support a PUA homeland in the Great Basin, which also provides additional support for a northern origin. In contrast, Hill (3) argued that UA originated in the vicinity of where maize was domesticated, perhaps in Central Mexico. As part of this southern-origin hypothesis, Hill (3) argued that the innovation of maize agriculture precipitated a population expansion northward, bringing maize agriculture and UA languages to the Southwest. This is a prediction of the Farming/Language Dispersal Hypothesis (FLDH) that has been put forth to explain how the most widely spread language families came to be distributed at the expense of those that have more restricted distributions today (21–23).
Additional cultural similarities attest to similar world perspectives held by individuals residing in both areas (1, 10, 24–27). Moreover, turquoise mined in the Southwest was traded southward to Mesoamerica, and goods of Mesoamerican origin were traded northward to the Southwest along the trade route known as the Turquoise Road (1, 24, 26, 28, 29).
With the wealth of cultural and linguistic similarities between the regions and especially the hypothesis of whole-scale population movement from Mesoamerica to the Southwest, it is reasonable to hypothesize close genetic ties between populations in the two regions. Although studies of protein polymorphisms have identified similarities in gene frequencies between populations in the two regions (7, 30–33), previous mitochondrial DNA (mtDNA) data provided no evidence of a population expansion northward from Mesoamerica (6, 7). Southwest populations are characterized by high frequencies of mitochondrial haplogroup B and very low frequencies, or the complete absence, of haplogroup A. The exception to this pattern is the high frequency of haplogroup A found among Southwest Athapaskan populations (Navajo and Apache) that entered the region from the north within last ∼500 years (6). As found in Results and Discussion, this population movement played little role in our understanding of the relationships between populations of the Southwest and Mesoamerica. In contrast to non-Athapaskan Southwest populations, Mesoamerican populations exhibit high frequencies of haplogroup A with moderate to low frequencies of haplogroup B (6, 7, 34–37). These haplogroup patterns crosscut linguistic and geographic boundaries within both regions, and predate European contact in the Southwest (38, 39) and Central Mexico (40, 41). In addition, phylogeographic analyses of mtDNA haplotypes have not supported a close relationship among UA-speaking populations in the Southwest and Central Mexico (6). Together with the mtDNA variation exhibited by populations in the Great Basin (42) and California (43), one finds that UA-speaking populations exhibit greater similarity within than across these regions.
Based on mtDNA patterns and the presence of the rare variant Albumin*Mexico in populations of Mesoamerica and the Southwest (7), it has been hypothesized that male movements have been the source for transmission of Albumin*Mexico, maize cultivation, and the UA language(s) across the regions but not mtDNA, which is maternally inherited (6, 7). Bellwood (23), a proponent of the FLDH, was not entirely convinced by this conclusion and stated that the “prospect of females staying close to home and males migrating makes a degree of sense, but one wonders how the claimed results reflect sampling and other factors” (23). Indeed, mtDNA of Mexican populations north of Mexico City had previously only been sampled from two cities that border the United States (44) and the Seri (6). Moreover, mitochondrial haplogroup affiliation of only 120 Native Americans from Mesoamerica had been identified before the comment made by Bellwood (23), and hypervariable region I (HVRI) haplotypes of only 15 of these individuals have been determined (6, 34, 35, 45–47). To date, no analyses of Y-chromosome variation have been conducted to directly investigate prehistoric movement of males between these regions.
If the conditions behind the FLDH explain how UA was spread across these regions, it is first predicted that UA speakers in the Southwest and Mesoamerica should be more closely related to each other than with non-UA speakers. In other words, genetic distances should correlate in a positive manner with linguistic but not geographic distances. Moreover, shared derived variation should be identified among UAs. Although these predictions might hold for a Southwest origin of UA as well, they are essential to the FLDH. It is these predictions of the hypothesis that are evaluated here with a large-scale examination of mtDNA and Y-chromosome variation.
Results and Discussion
In general, mitochondrial haplogroup frequencies differ substantially between Southwest and Mesoamerican populations; the former rarely exhibits mitochondrial haplogroup A, but it predominates in the latter (Table S1 and Fig. S1). In contrast, haplogroup B is very common in Southwest populations, but it is much less common in Mesoamerican populations. Two exceptions to this pattern are (i) the Nahua–Atocpan, a Mesoamerican population that exhibits slightly more haplogroup B than haplogroup A, and (ii) the Tarahumara, a Southwest population that exhibits a higher frequency of haplogroup A than haplogroup B (Table S1 and Fig. S1). The Cora and Huichol also exhibit over 20% greater frequencies of haplogroup B than haplogroup A. Because these two populations are grouped in Mesoamerica by some (11) and in the Southwest by others (10), the geographic distributions of haplogroup A and B across the Southwest and Mesoamerica are more clinal than previously described (6), but this does not provide any support for particularly close relations among UA speakers across the regions, a prediction of the FLDH.
The principal coordinates analysis (PCoA) plot of FST values based on mtDNA haplogroup frequencies (SI Text) and haplotypes (Fig. S2B) reflects the same geographic division of haplogroups A and B as just described, dividing the Southwest from Mesoamerica on the first principal coordinate. The second principal coordinate reflects substructure in the Southwest between Pueblo groups (Anasazi, Jemez, and Zuni) and Pimans/Yumans (Akimel O'odham, Hualapai, and Tohono O'odham) (Fig. S2 A and B).
The Y-chromosome haplogroup frequencies exhibited by the populations in this study are provided in Table S1. Haplogroup Q-M3 was detected in all populations and ranged in frequency from ∼43–100%. Haplogroup Q-M242, ranging in frequency from ∼5–43%, was in UA populations in both the Southwest and Mesoamerica as well as the Jemez. Non-Native American admixture (i.e., PM-45, M-173, and Y-chromosome Alu Polymorphism) ranging from ∼5–45% was detected in five populations.
The vast majority of mtDNA and Y-chromosome haplotypes were unique to individuals and/or shared by individuals from the same population. Three hundred twenty-three mitochondrial haplotypes were identified among the 689 sequences belonging to haplogroups A, B, C, D, and X (SI Text). The majority of these haplotypes (60.4%), equally distributed among the haplogroups, are unique to individuals, whereas 30.3% of them are shared among individuals of the same population. Thus, 90.7% of the haplotypes are population-specific, and 9.3% are shared between or among populations. Ninety-eight haplotypes were identified among the 121 Y chromosomes studied for their short tandem repeat variation (SI Text). The majority of these haplotypes (83.7%) are unique to individuals; 10.2% were shared among individuals within the same population, and only 6.1% were shared between populations. The PCoA plot based on the Y-chromosome data shows that the closest Y-chromosome relationships between the Southwest and Mesoamerica predominantly involve UA speakers, consistent with predictions of the FLDH (Fig. S2C).
The mtDNA haplotype networks are found in Fig. 1 and SI Text. The vast majority of subclades within the networks were specific to regions but not language families. In particular, few UA-specific mtDNA clades were seen in the networks, contrary to expectations of the FLDH.
Haplogroup B network containing all haplotypes that exhibit an adenine (A) at np 16483. The central node (shaded gray) exhibits the following mutations relative to the Cambridge reference sequence (75, 76): 16111, 16189, 16217, 16483, 16519, 00073, 00263, and 00499. Mutational positions from this haplotype are noted in small print. The haplotype marked with an asterisk links to the gray-shaded haplotype in Fig. S3 by the transition at np 16483. Black circles represent median vectors, haplotypes that existed at one time but are now extinct, or haplotypes that were simply not sampled in this study.
None of the haplogroup A haplotypes sampled here in the Southwest exhibited the characteristic Athapaskan markers (16233G and 16331G), which confirms that little mtDNA gene flow occurred from Navajo and Apache into other populations in the Southwest and supports previous studies (6, 7, 48). Of further note, within haplogroup A, one clade containing a Tohono O'odham and Zuni [derived by nucleotide positions (nps) 16257T and 16263A] is of interest, because this form of haplogroup A is only found elsewhere among the Chumash of Southern California (43). Combined with the absence of shared derived forms of haplogroup A across the regions, this cautions against the interpretation that haplogroup A in the Southwest was introduced by farmers from Mesoamerica (49), where haplogroup A is far more common.
Within haplogroup B, 130 of 250 individuals (52%) were derived at np 16483A, and most of these individuals were also derived at np 16111T. This subhaplogroup is termed B2a (50), and the network depicting the relationships between the B2a lineages is particularly interesting (Fig. 1). First, these derived lineages are found in every Southwest population sampled in addition to the Cora and Huichol. Although 126 of 246 individuals (51.2%) that belonged to haplogroup B in the Southwest belonged to this clade (or ∼29% of all of the individuals of these populations), not a single Mesoamerican individual exhibited the 16483A transition. This form of haplogroup B has previously been detected in low frequency among Yavapai, Kumeyaay, Cochimi (6), the Washo (51), and the Turtle Mountain Chippewa (52). It might also be present in various Yuman and UA populations from Southern California (43) and the Apache and Navajo of the Southwest (53), because they exhibit the 16111T mutation; however, np 16483 was not sequenced for these populations.
Second, available D-loop sequences indicate that haplogroup B haplotypes derived at np 16483A are not found in Asian populations [based on the samples screened by Kemp et al. (54)], and whole-genome sequences reveal that the mutation is unique to Native Americas (55). This suggests that subhaplogroup B2a evolved in the Americas. Nucleotide diversity (π) within our samples of B2a was estimated in Mega 3.0 (56) with 10,000 bootstraps of the data to be 0.0020 [95% confidence interval (CI) ± 0.000]. The approximate age of this clade is 2,105 YBP (99.5% CI = 1,273–3,773 YBP), which was estimated by using Howell et al.’s (57) average rate of 47.5% per site/myr (99.5% CI = 26.5–78.5% per site/million years) for the evolution of the D-loop (nps 16024–00576). This pedigree-based rate is appropriate for estimating the age of events that have occurred within the past 15,000 years (58), a period that encompasses most, if not all, of the occupation of the Americas (59). The estimated date of this expansion closely coincides with the date of dramatic increase in population size in the Southwest estimated from archaeologic evidence (15).
Third, these data indicate that there has been substantial interlanguage family admixture in the Southwest, possibly since (and probably during) the time of the initial expansion. The expansion seems to have been one that was region-wide, encompassing all of the diverse populations within the greater Southwest. It is possible that risk management associated with dry farming of maize may have lead to substantial social and political reformation (60), including increased reliance on long-distance relationships that facilitated migration (61) and presumably gene flow as a consequence. In addition, increase in population size could have lead to more interpopulation contact and as a consequence, increased admixture.
Fourth, it is possible that because of the magnitude of this recent expansion, genetic patterns of greater antiquity have been obscured. If an in situ Southwest expansion did occur within the past 4,000 years, it may inform us little, if at all, about the proposed earlier movement of UA females from central Mexico into the Southwest (3). However, if UA originated in the Southwest, this population expansion may have spread the language family south as far as Nayarit and Jalisco where the Cora and Huichol reside. Currently, the only available ancient DNA (aDNA) evidence from the Southwest cannot address this issue, because the oldest samples analyzed are only ∼1,600 years old (38, 39). These data confirm only that haplogroup B has been the most common haplogroup in the region for at least 1,600 years. The analysis of aDNA from early farmers and populations that predate the expansion will be required to characterize the gene pool of the Southwest at these times in relationship to Mesoamerica and provide a further test of this hypothesis.
It is unclear why this expansion is largely detectable in only mitochondrial haplogroup B and does not eliminate correlations between Y-chromosome variation and linguistic distances. It is possible that members of mitochondrial haplogroups A and D in the Southwest are so infrequent (Table S1) that they do not display a similar pattern. Although mitochondrial haplogroup C is relatively common, the haplotype network of this haplogroup (SI Text) may be confounded by containing members of at least four newly described founding lineages that cannot be differentiated by D-loop sequence alone (59, 62). As to the difference in pattern between the uniparentally inherited markers, Balaresque et al. (63) recently discovered that the advent of farming allowed for an expansion of Neolithic males in Europe, but a similar expansion was not detected in the mtDNA. Our case seems to be just the opposite.
The results of the Mantel tests show that mtDNA distances across the Southwest and Mesoamerica are positively and significantly correlated at the 0.05 level of probability with geographic distance but not with linguistic distances (Table 1). This result is in support of the recent report that linguistic and mtDNA diversity are not correlated among indigenous Mexican populations (36). The results of the Mantel tests support the overall trend of a major subdivision between populations from Mesoamerica and the Southwest on the direct maternal line.
Results of the Mantel tests
In contrast, Y-chromosome diversity correlates positively and significantly with linguistic distances when the internal structure of UA was considered (i.e., when using the Miller and Hale/Hill models) but not when it was ignored (i.e., when using the Simple model) (Tables 2 and 3). Previous research has also indicated some agreement between Y-chromosome variation and membership in UA (64), supporting the notion that the internal structure is correlated with genetic distances between populations. However, it is clear from the present study that Y-chromosome diversity is not correlated with either geography or mtDNA distances (Tables 3 and 4). Overall, these tests elude to differing population histories of males and females in these regions, especially with regards to UA prehistory. In other words, the Y-chromosome variation is predicted by the structure of UA and its relationship to non-UA language families, whereas mtDNA variation is not. Interestingly, because the correlation between linguistic distances and Y-chromosome variation disappears as the internal structure of UA is ignored (i.e., when using the Simple model), it seems that the degree of relation between UA males is as predicted by linguists. It does not, however, provide any sense of direction for the movement of UA. These results are also interesting with regards to Belle and Barbujani's (65) discovery that, at a global level, linguistic patterns have a nonnegligible, albeit small, correlation with autosomal genetic variation. Although their data cannot point to any sex bias, we detected one in our study area.
Results of the Mantel tests
Results of the Mantel tests
Results of the Mantel tests
Because of the number of cultural and linguistic connections between populations of the Southwest and Mesoamerica, it was hypothesized that populations in these two regions would exhibit close genetic ties as well. In this study, we found evidence that males in the Southwest and Mesoamerica are as genetically related to one another as predicted by the proposed relationships of the languages that they speak. Yet, despite a substantial increase in sampling, mtDNA variation across these regions remains strongly correlated with geography and not with language family. However unlikely, these data suggest that if a migration of UAs was responsible for introducing maize agriculture to the Southwest as previously hypothesized by linguists and archaeologists (3, 4, 66), it was predominantly comprised of males and likely not as a result of a demographic expansion.
This scenario would explain why previous mtDNA studies failed to find a connection between the regions, whereas a rare nuclear DNA maker, Albumin*Mexico, clearly unites them (6, 7). However, it is also possible that the recent expansion of mtDNA haplogroup B within the Southwest blurred the preexisting mtDNA structure of Southwest populations (i.e., >4,000 years ago) that may have evinced a genetic relationship between the mtDNA of the two regions at one time. Thus, future studies of aDNA may alter our understanding of the prehistory of Southwest and Mesoamerican populations.
Materials and Methods
Samples.
MtDNA variation was studied in 848 individuals from 13 populations from Mesoamerica and the American Southwest (SI Text provide sources of samples). These data were combined with those from seven previous studies (Table S1) totaling 960 individuals. Y-chromosome variation was studied in 178 males belonging to 11 of these populations (Table S1). Although the sampling was focused on populations that speak languages belonging to the UA language family, non-UA samples were included for comparative purposes (Table S1). A single sample from the Tepehuan, a UA population, was included only in the network analysis. Populations were assigned to the Southwest or Mesoamerica (Table S1). The Cora and Huichol, because of their intermediate geographic locations and their previous placement in either the Southwest (10) or Mesoamerica (11), are considered as special cases.
DNA Extraction, Haplogroup Determination, and Sequencing.
DNA was extracted from buccal swabs and blood samples with the Qiagen Blood Amp Kit. All of the samples were screened for the polymorphisms that define Native American mitochondrial haplogroups A, B, C, D, and X. nps 16011–00684 (SI Text and Table S4) of the mitochondrial genome, representing the entire D-loop, was determined for 716 individuals. The 178 male samples were screened for six Y-chromsome binary polymorphisms: M3, M242, RPS4Y711, M45, M173, and YAP (DYS287). In addition, eight Y-STR loci (DYS19, DYS390, DYS391, DYS392, DYS393, DYS389I, DYS389II, and DYS439) were typed in 121 males belonging to haplogroups Q-M3 and Q-M242 (SI Text has detailed laboratory methods).
Data Analysis.
For mtDNA analysis, all individuals that do not belong to haplogroup A–D or X (Table S1) were excluded, because they most likely represent non-Native American admixture.
FST values were calculated for all pairs of populations in Arlequin (version 2.000) (67). A distance matrix was constructed of all pair-wise FST values and used to conduct PCoA performed in the program DistPCoA using the Caillez method to correct for negative eigenvalues (68). Mantel tests were performed in Arlequin (version 2.000) (67) with 100,000 permutations of the data to test for correlations between genetic, geographic, and linguistic distances between populations (SI Text) at the 0.05 level of probability. The genetic distances were estimated as pair-wise FST values. The geographic distances were calculated in GenAlEx (69) using latitude and longitude coordinates (SI Text) determined to be central to precontact population ranges or from the locations where the samples were collected. Precontact ranges of the Southwest populations were inferred from the Handbook of North American Indians (2). The precontact ranges of the Mixe, Mixtec, and Zapotec are found in Hollenbach et al. (70).
Linguistic distances were estimated in three ways. First, the Miller estimate was based on time estimates of UA language splits taken from Fig. S2B of Miller (2). A 5,500-year age was assigned to the Mixtec–Zapotec split (71), and language-family splits were assigned time depths of 8,000 years. Second, the Hale/Hill estimate was taken from Hale (72) with the Nahua de Mecayapan used to represent both Nahua–Atocpan and Nahua–Cuetzalan; although Hale (72) noted some problems with the Tarahumara dates, we have taken his numbers exactly. The Cora–Huichol and Mixtec–Zapotec splits were assigned time depths of 1,350 (72) and 5,500 (71) years, respectively, and language-family splits were assigned time depths of 6,000 years, reflecting the shorter time depth of UA proposed by Hale (73) compared with Miller (2). Lastly, the Simple estimate was made by assigning values of 0 (minimum) to intralanguage family distances and 1 (maximum) to interlanguage family distances. This estimate ignores any intralanguage family structure.
Median-joining haplotype networks were constructed in Network (version 4.1.1.2) (74) separately for haplogroups A, B, C, and D. No network was constructed for haplogroup X, because only one haplotype of this haplogroup was detected in this study (exhibited by eight Jemez). SI Text has details of network construction.
Acknowledgments
We thank all the individuals who provided and authorized use samples in this study. We are grateful to Jane Hill for proving us expert advice on the Uto-Aztecan language family. Thanks to Tim Kohler for some important and compact advice on the possible link between dry farming and gene flow. This project was funded by The University of California Institute for Mexico and the United States and Wenner–Gren grants (B.M.K.). Additional support was provided by Grant BCS 0422144 from the National Science Foundation (to B.M.K., R.S.M., A.R., and D.G.S.) and Grant R24RR005090 from the National Center for Research Resources of the National Institutes of Health (to D.G.S.). A.G.-O. was supported by Dirección General de Asuntos del Personal Académico, Universidad Nacional Autónoma de México.
Footnotes
- 1To whom correspondence should be addressed. E-mail: bmkemp{at}wsu.edu.
Author contributions: B.M.K., A.G.-O., R.S.M., and D.G.S. designed research; B.M.K., A.G.-O., R.S.M., C.M., K.B.S., J.M., G.R., A.R., R.I.P.-E., L.B.-M., and C.G. performed research; B.M.K., A.G.-O., and R.S.M. analyzed data; and B.M.K., A.G.-O., R.S.M., C.M., K.B.S., and D.G.S. wrote the paper.
↵2Deceased September 2, 2005.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Data deposition: The mitochondrial DNA sequences reported in this paper have been deposited in the Genbank database (accession nos. EU719668–EU720177, EU720188–EU720206, and EU20221–EU720379).
This article contains supporting information online at www.pnas.org/cgi/content/full/0905753107/DCSupplemental.
References
- ↵
- Hedrick BC,
- Kelley JC,
- Riley CL
- ↵
- Sturtevant WC
- Miller WR
- ↵
- ↵
- Renfrew C,
- McMahon AMS,
- Trask RL
- Bellwood P
- ↵
- Bellwood P,
- Renfrew C
- Matson RG
- ↵
- ↵
- ↵
- González-José R,
- et al.
- ↵
- Cordell LS
- ↵
- Hedrick BC,
- Kelley JC,
- Riley CL
- Beals RL
- ↵
- Suárez JA
- ↵
- Tax S
- Kirchhoff P
- ↵
- Long A,
- Benz BF,
- Donahue DJ,
- Jull AJT,
- Toolin LJ
- ↵
- Pohl ME,
- Piperno DR,
- Pope KO,
- Jones JG
- ↵
- Kohler TA,
- Glaude MP,
- Bocquet-Appel J-P,
- Kemp BM
- ↵
- Merrill WL,
- et al.
- ↵
- ↵
- Romney AK
- ↵
- ↵
- ↵
- Bellwood P,
- Renfrew C
- ↵
- Diamond J,
- Bellwood P
- ↵
- Bellwood P
- ↵
- Turner CG,
- Turner JA
- ↵
- Hedrick BC,
- Kelley JC,
- Riley CL
- Parsons EC
- ↵
- Reyman JE
- Kelley JC
- ↵
- Hedrick BC,
- Kelley JC,
- Riley CL
- Haury EW
- ↵
- Coe MD
- ↵
- Sturtevant WC
- Di Peso CC
- ↵
- ↵
- Lauglin WS,
- Harper AB
- Spuhler JN
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- Reed DM
- Kemp BM,
- Resendez A,
- Román Berrelleza JA,
- Malhi RS,
- Smith DG
- ↵
- ↵
- Johnson JR,
- Lorenz JG
- ↵
- ↵
- Torroni A,
- Neel JV,
- Barrantes R,
- Schurr TG,
- Wallace DC
- ↵
- ↵
- ↵
- ↵
- LeBlanc SA,
- et al.
- ↵
- ↵
- Kaestle FA
- ↵
- ↵
- ↵
- ↵
- Ingman M,
- Gyllensten U
- ↵
- Kumar S,
- Tamura K,
- Nei M
- ↵
- ↵
- Henn BM,
- Gignoux CR,
- Feldman MW,
- Mountain JL
- ↵
- ↵
- ↵
- Cordell LS,
- Van West CR,
- Dean JS,
- Muenchrath DA
- ↵
- ↵
- ↵
- ↵
- ↵
- Matson RG
- ↵
- Schneider S,
- Roessli D,
- Excoffier L
- ↵
- Legendre P,
- Anderson MJ
- ↵
- ↵
- ↵
- Flannery KV,
- Marcus J
- ↵
- ↵
- Kemp BM
- ↵
- ↵
- ↵
Citation Manager Formats
Sign up for Article Alerts
Article Classifications
- Biological Sciences
- Anthropology