Professional Documents
Culture Documents
Objectives: The ancient Chachapoya were an aggregate of several ethnic groups that shared a common language,
religion, and material culture. They inhabited a territory at the juncture of the Andes and the Amazon basin. Their
position between those ecozones most likely influenced their genetic composition. We attempted to better understand
their population history by assessing the contemporary genetic diversity in the Chachapoya and three of their immedi-
ate neighbors (Huancas, Jivaro, and Cajamarca). We inferred signatures of demographic history and genetic affinities,
and contrasted the findings with data from other populations on local and continental scales.
Methods: We studied mitochondrial DNA (mtDNA; hypervariable segment [HVSI and HVSII]) and Y chromosome
(23 short tandem repeats (STRs)) marker data in 382 modern individuals. We used Sanger sequencing for mtDNA and
a commercially available kit for Ychromosomal STR typing.
Results: The Chachapoya had affinities with various populations of Andean and Amazonian origin. When examining
the Native American component, the Chachapoya displayed high levels of genetic diversity. Together with other parame-
ters, for example, large Tajimas D and Fus Fs, the data indicated no drastic reduction of the population size in the past.
Conclusion: The high level of diversity in the Chachapoya, the lack of evidence of drift in the past, and genetic affin-
ities with a broad range of populations in the Americas reflects an intricate population history in the region. The new
genetic data from the Chachapoya indeed seems to point to a genetic complexity that is not yet resolved but beginning
to be elucidated. Am. J. Hum. Biol. 00:000000, 2016. C 2016 Wiley Periodicals, Inc.
V
The Peruvian cloud forest and neighboring Amazonia as a confederation of several ethnic groups that shared a
host a number of multicultural populations who speak over common language, religion, and cultural traits (Espinoza,
30 different languages (Adelaar, 2004). However, genetic 1967). They were clearly distinguished from other popula-
studies of the indigenous populations in these areas are still tions such as those from the central Andes. Their location
scarce. This situation contrasts with the Peruvian highlands between two major ecological zones (Andes-Amazon) and
or the coastal area, where some genetic data have recently the narrowing of this horizontal stretch of the Andes in the
been published (Cabana et al., 2014; Sandoval et al., region allowed the Chachapoya to maintain long-standing
2013a,b) and especially archaeological studies are abundant interactions with highland, tropical rainforest, and even
(Church and von Hagen, 2008; Guengerich, 2014). coastal societies. This feature makes the Chachapoya an
Even though the last decades have seen a substantial attractive target for population genetic studies.
increase in archaeological studies, the emergence and In this study, we have explored the demographic history
decline of civilizations in the cloud forests of the north- of a region between two major ecosystems, the Andes and
eastern Peruvian Andes remains poorly understood. the Amazonian rainforest, by assessing the Y-chromosomal
Human activity in this area is reported as early as 11 kya and mitochondrial DNA (mtDNA) diversity in the present-
(Church and von Hagen, 2008), although it is not until day Chachapoya in the northeastern Peruvian cloud forest.
much later (4 kya) that more complex societies emerged To date, very few of these transitional area populations
(Olivera, 2012). have been studied (Barbieri et al., 2014; Corella et al.,
One of the most prominent cultures in this region dur- 2007, 2008; Scliar et al., 2014), with these tending to show
ing pre-Hispanic times was the Chachapoya. This society a complex genetic makeup (Tarazona-Santos et al., 2001).
flourished during the Late Intermediate Period This is the first large scale study of the extant Chachapoya.
(8001470 C.E.) and had an area of influence of approxi- It aims to provide a better understanding of the population
mately 22,000 Km2 in the cloud forests of todays political
region Amazonas (Guillen et al., 2011). During the first
half of the 16th century, the Inca gradually annexed most
of the Chachapoya territory. This phenomenon triggered Additional Supporting Information may be found in the online version
of this article.
a sociopolitical restructuring that most likely altered the Contract grant sponsors: Provincial Municipality of Chachapoyas and
original population composition, as well. The arrival of the Regional Government of Amazonas in Peru; Contract grant sponsor:
Spaniards c. 1,532, deeply affected the biological and Center for International Mobility (CIMO); Contract grant number: TM-12-
8372; Contract grant sponsor: Paulo Foundation (A.S.).
cultural features of the indigenous human populations in
*Correspondenc to: Evelyn K. Guevara, Department of Forensic
the region, causing substantial population declines (Cook, Medicine, PO BOX 40, FIN-00014, University of Helsinki, Finland.
2010). E-mail: evelyn.guevara@helsinki.fi
Although most of the nonmaterial aspects of the Chacha- Received 9 June 2015; Revision received 11 May 2016; Accepted 13 May
2016
poya culture faded away, a rich archaeological, linguistic,
DOI: 10.1002/ajhb.22878
and most likely also genetic, heritage persist in the region. Published online 00 Month 2016 in Wiley Online Library
Ethnohistorical accounts describe the ancient Chachapoya (wileyonlinelibrary.com).
Fig. 1. Map of Amazonas political region showing the location of the study populations.
METHODS
history of this intermediate region by (1) assessing the
Samples and reference data
amount and nature of genetic diversity in the Chachapoya
and its immediate neighbors (Huancas, Jivaro, and Saliva samples from a total of 382 individuals were col-
Cajamarca), (2) evaluating the extent to which the modern lected from four populations in the northeastern Peruvian
Chachapoya population is related to these neighboring Andes (Fig. 1). The total sample includes individuals pro-
groups and to populations from the Andean pool, and (3) posed as descendants of the ancient Chachapoya (N 5 277),
exploring the past demographic history of the Chachapoya the Amazonian Jivaro (N 5 47), the Huancas (N 5 21), and
and comparing these patterns of genetic variation at the Cajamarca (N 5 37; for a detailed explanation of the
broader scales (Andes, the Americas). Since most of the populations see Supporting Information file Populations.-
questions discussed are concerned with the pre-Hispanic docx). Samples were collected after informed consents were
history of the region (see Discussion), we restricted our obtained. We followed ethical guidelines (e.g., Declaration
observations to the Native American genetic component in of Helsinki) and obtained the approval of local authorities
these populations. (e.g., Regional Government of Amazonas). Regional and
local authorities, communal representatives and each indi- cytosine (C) at nucleotide position (np) 16,189 and the
vidual were informed about nature and the objectives of resulting C-homopolymer in this region produces unreli-
the study. The project was approved by the Hospital Dis- able sequence reading. To overcome this, samples were
trict of Helsinki and Uusimaa Ethical Committee with per- additionally sequenced with primers L16209 (50 -CCCCAT
mit #329/13/03/00/13. GCTTACAAGCAAGT-30 ) and H16164 (50 -TTTGATGTGGA
Samples were collected into Eppendorf tubes containing TTGGGTTT-30 ) for HVSI, and H285 (50 -GGGGTTTGGT
1 ml of fieldwork buffer (0.05 M TRIS, 0.05 M EDTA pH GGAAATTTTTTTG-30 ), for HVSII. Sequencing was per-
8.0, 0.05 M sucrose, 0.1 M NaCl, and 1% SDS 50 ml/L), formed using BigDye v1.1 Cycle Sequencing Ready Reac-
finally stored at 148C (Quinque et al., 2006). Demo- tion Kit (Life Technologies, Foster City, CA), according to
graphic information, including surnames, place of birth, suppliers recommendations. Sequencing reactions were
mother tongue, and ethnicity up to the grandparents gen- performed in both directions, and the products were puri-
eration, was recorded. Samples were anonymized for fur- fied with BigDye XTerminator and analyzed using an ABI
ther data processing. Prism 3130xl automated sequencer (Life Technologies, Fos-
In addition to these new datasets, reference mtDNA ter City, CA). The data were compiled using Sequencher
data for the hypervariable segment I (HVSI) sequences 4.8 software (GeneCodes, Ann Arbor, MI). All sequences
from 99 populations and Y chromosome STRs (Y-STR) were first aligned using Sequencher and inconsistencies
haplotypes from 6 to 17 loci from 68 Native American pop- were resolved manually in Bioedit (Hall, 1999). Sequences
ulations were gathered from published sources (Support- have been deposited in GenBank under accession numbers
ing Information Table S1). Populations with small sample KR870439KR870819.
sizes (28), geographic proximity and with genetic distan-
ces not significantly (P > 0.05) larger than zero were Y-STR genotyping
pooled for analysis. For the genotyping of Y-STRs, the PowerPlex Y23 kit
(Promega, Madison, WI) was used. The kit co-amplifies 23
DNA extraction
loci (DYS576, DYS389I, DYS448, DYS389II, DYS19,
DNA extraction was performed applying a modified DYS391, DYS481, DYS549, DYS533, DYS438, DYS437,
high-salt DNA extraction protocol (Quinque et al., 2006). DYS570, DYS635, DYS390, DYS439, DYS392, DYS643,
Into 750 ml buffer-saliva mixture aliquot, 15 ml Proteinase- DYS393, DYS458, DYS385a/b, DYS456, and Y-GATA-H4).
K (20 ng/ml), and 40 ml 20%SDS were added and incu- PCR amplification was performed according to the suppliers
bated overnight at 1568C in shaking water bath. For puri- recommendations with the following cycling conditions: 2
fication 300 ml 6 M NaCl was added, followed by 10 min min of initial denaturation at 968C, followed by 26 cycles of
incubation on ice, and 10 min centrifugation step at 948C for 10 s, 1 min at 618C and 30 s at 728C. Final exten-
13.3 g. The supernatant was transferred into a new tube sion consisted of 20 min at 608C. Products were resolved and
R
and the DNA was precipitated by adding 1 ml of ice-cold analyzed using an ABI Prism 3130xl and GeneMapperVID
isopropanol. After 1 h incubation at 208C the DNA was v. 3.2. Haplotypes from all study populations are provided in
pelleted by centrifuging at full speed in a tabletop centri- Supporting Information Table S2.
fuge for 15 min. The supernatant was discarded and the
pellet washed with 500 ml 70% ETOH. The pellets were Haplogroup assignment
dried at room temperature and then resuspended in 100 The mitochondrial haplogroup was estimated for each
ml of ddH2O. The DNA was additionally purified and con- sample using Haplogrep Build 16 (http://haplogrep.uibk.ac.
centrated using QIAQuick Purification Kit (Qiagen, Hil- at/), which uses data from Phylotree Build 16 (Kloss-Brand-
den, Germany) following manufacturers protocol. staetter et al., 2011; van Oven and Kayser, 2009). Hap-
logroup designations were verified manually by using the
MtDNA sequencing information from Phylotree (mtDNA tree Build February
Mitochondrial HVSI (16,024-16,385) and HVSII (72- 16, 2014). For the Y-chromosomes, haplogroups were tenta-
340) were sequenced in all samples. The mitochondrial tively predicted using Whit Atheys Haplogroup Predictor
control region was amplified using primers L15900 (50 (http://www.hprg.com/hapest5).
TACACCAGTCTTGTAAACC30 ) and H00599 (50 TTGAGG
AGGTAAGCTACATAA30 ). Amplification was performed Data analyses
in 15 ml reactions containing 3.8 ml 5X GoTAQ FlexiBuffer; Basic diversity indices, haplotype diversity (H), nucleo-
1.1 ml 25 mM MgCl2 (Life Technologies, Foster City, CA); tide/locus diversity (mtDNA: p, Y-STR: Hd), and mean
0.6 ml BSA (5 mg/ml, MBI Fermentas (Vilnius, Lithuania) number of pairwise differences (MNPD) for all study popu-
1.9 ml dNTPs (2.0 mM); 0.4 ml Primer L15900 (10 mM); 0.4 lations and reference data were obtained using Arlequin
ml Primer H00599 (10 mM); 0.75 ml GoTaqDNA Polymerase ver. 3.5.1.2 (Excoffier et al., 2005). To compensate for the
(5 U/ml); 8.3 ml ddH2O and 1 ml of DNA template. PCR variability in sample sizes, haplotype diversities (hr), were
cycling conditions were: 2 min at 958C followed by 31 cycles also estimated after rarefaction using Contrib ver. 1.02
of 958C 15 s, 568C 30 s, and 728C 1 min 30 s; and a final (Petit et al., 1998). Genetic distances (UST), mismatch dis-
extension at 728C for 10 min. PCR products were purified tributions, AMOVA as well as neutrality tests, Tajimas D
using 0.25 ml Exonuclease I (10 U/ml), 0.25 ml Shrimp Alka- (Tajima, 1989)and Fus Fs (Fu, 1997), were also computed
line Phosphatase (1 U/ml), and 4.5 ml of ddH2O. HVSI was in Arlequin. Pairwise UST distances were summarized with
sequenced using primers L15997 (50 -CACCAT TAGCACCC neighbor-joining (NJ) trees using MEGA 5 software
AAAGCT-30 ) and H16391 (50 -GAGGAT GGTGGTCAAGGG (Tamura et al., 2007). Trees included an African sequence
AC-30 ) and for HVSII primers L048 (50 -CTCACGGGAG (EU092764.1) (Behar et al., 2008) as an outgroup in the
CTCTCCATGC-30 ) and H408 (50 -CTGTTAAAAGTGCATAC case of mtDNA data and set of African genotypes (belong-
CGCCA-30 ) were utilized. Most of the samples exhibited a ing to haplogroups B2a, B2b, and E3a) for Y-chromosome
TABLE 1. Sample sizes and mtDNA haplogroup frequencies for all study populations
Population/Haplogroup N A B C D Other
Corobamba 27 14.9% (4) 40.7% (11) 22.2% (6) 18.5% (5) 3.7% (1)
Pomacochas 49 10.2% (5) 53.1% (26) 16.3% (8) 10.2% (5) 10.2% (5)
Rodrguez de Mendoza 28 0% 35.7% (10) 21.4% (6) 25.0% (7) 17.9% (5)
Chillao 40 22.5% (9) 17.5% (7) 25.0% (10) 20.0% (8) 15.0% (6)
Chachapoyas 50 18.0% (9) 32.0% (16) 16.0% (8) 24.0% (12) 10.0% (5)
La Jalca 33 18.2% (6) 48.5% (16) 18.2% (6) 3.0% (1) 12.1% (4)
Leymebamba 38 13.1% (5) 21.1% (8) 31.6% (12) 21.1% (8) 13.1% (5)
Chilchos 12 0% 33.3% (4) 41.7% (5) 25.0% (3) 0%
Chachapoya (pooled) 277 13.7% (38) 35.4% (98) 22% (61) 17.7% (49) 11.2% (31)
Jivaro 47 12.8% (6) 59.6% (28) 25.5% (12) 0% 2.1% (1)
Huancas 21 42.9% (9) 42.9% (9) 4.7% (1) 9.5% (2) 0%
Cajamarca 37 10.8% (4) 32.4% (12) 32.4% (12) 16.3% (6) 8.1% (3)
Total 382 15.9% (66) 38.2% (158) 23.2% (96) 14.2% (59) 8.5% (35)
TABLE 2. MtDNA diversity indices for the study populations Chachapoya (Table 1). Among the study populations, the
considering HVSI and HVSII Chachapoya showed the highest proportion of non-native
Population N A h SDh p SD p MNPD SDMNPD haplogroups (11.2%), which were absent or very low (2%)
among the Huancas and Jivaro, respectively. Haplogroups
Chachapoya 245 99 0.9671 0.0051 0.0157 0.0079 9.8241 4.5125 B and C were the most common lineages in all studied
Jivaro 46 19 0.9343 0.0178 0.0128 0.0067 8.0141 3.7916 populations except the Huancas, which showed high fre-
Huancas 21 12 0.9286 0.0325 0.0164 0.0087 10.2445 4.8721
Cajamarca 34 22 0.9715 0.0137 0.0141 0.0074 8.8323 4.1756 quency of haplogroup A, the least frequent native hap-
logroup in the Chachapoya. Haplogroup B reached 60% in
the Jivaroan set, in which haplogroup D was not
(Coelho et al., 2009). In addition, the interpopulation differ- observed. The Cajamarca had the same proportion of hap-
ences in the mtDNA haplogroup frequencies were summar- logroups B and C mtDNAs.
ized by principal component analysis (PCA) conducted
using XLSTAT (Addinsoft Corp. 02.03. 2014). The mito-
chondrial sequence analyses were performed assuming Native American haplotype-level diversity. Haplotypes
Tamura-Nei mutation model (Tamura and Nei, 1993). that did not contain characteristic HVSI and HVSII point
The different statistics were calculated on the popula- mutations for Native American haplogroups were
tion level for the Chachapoya and neighboring popula- excluded in subsequent analyses since the main focus of
tions. The mitochondrial and Y-chromosomal diversity in this study was on the Native American component. All
the main study population was compared with that of the sampled populations exhibited high levels of haplotype
Jivaro, the Huancas and the Cajamarca groups in the diversity (Table 2) ranging from H 5 0.9286 (Huancas) to
regional context. Comparisons at this level were accom- H 5 0.9715 (Cajamarca), with the Chachapoya close to the
plished using both mtDNA HVSI and HVSII and 23 Y- high end (H 5 0.9671). Despite this level of variation in
STR haplotypes. At a broader scale (the Americas) com- the point values, no statistically significant differences
parisons included only HVSI sequences and a set of Y- among the populations were observed. Within each Native
STRs (six, nine, fifteen, and seventeen), based on the American haplogroup (AD), the haplotype diversity also
availability of reference data. In these comparisons, popu- exceeded H 5 0.95 in the Chachapoya (data not shown).
lations with N < 20 and mtDNA data C-stretch length The high mtDNA diversity in the Chachapoya was also
polymorphisms (positions 16,18216,190 and 302315) apparent at a continental level comparison-based on
were not included in the analyses. HVSI sequence data. Among studied populations, where
To explore the genetic structure in the cloud forest allelic richness after rarefaction (assuming N 515) ranged
region, the Chachapoya sample set was further divided from 0 to 13.671, the study populations Chachapoya
into eight subunits that roughly represent the ancient (hr 5 11.227) and Cajamarca (hr 5 11.356) still main-
geographic distribution of Chachapoya ethnic groups, tained relatively high levels of diversity (Supporting
namely, Corobamba, Pomacochas, Chillao, Chachapoyas, Information Table S3). Nucleotide diversities ranged from
La Jalca, Leymebamba, Chilchos, and Rodriguez de Men- 0.0129 (Jivaro) to 0.0164 (Huancas), which were interme-
doza, as suggested by ethnohistorians (Espinoza, 1967; diate when compared to other Native American popula-
Schjellerup, 2005). This approach allowed us to explore tions. Similarly, the high diversity in these populations
variability within this population. (For the distribution of was also reflected in the MNPD (Supporting Information
these subunits, see the Supporting Information map, Cha- Table S3).
chapoya subunits.tiff).
Fig. 3. PCA plot with a subset of reference populations. The first principal component (PC 1) is responsible for 47% of the total variation and
correlates with the frequency of haplogroup B. The second principal component (PC 2) represents 29.79% of the variance and correlates with
the frequency of haplogroup C (haplogroups AD with positions relative to the axis indicated).
around one to three differences. However, very few, Jivaro showed a slight reduction in FSC values from 3.20
mostly of Andean origin, showed a well-defined peak at to 2.56 (P < 0.0001), when the Pomacochas subunit was
around sixseven differences and no signs of drastic grouped with the Jivaro.
reduction in their diversity (e.g., Tayacaja, Northern
Chile, Supporting Information Table S6). Y-chromosome
Compatible with the high diversity and mismatch distri- Haplotype diversity. In all of the study populations, the
bution, the Chachapoya exhibited large and statistically native haplotypes were less common in the Y-chromosomal
significant negative Fs (Fs 5 224.11948, P 5 0.00100) and than in the mtDNA data (Table 3). The Native American Y-
Tajimas D values (D 5 21.56347, P 5 0.02900), which are component (Q) in the Chachapoya was only 60% but sub-
considered to indicate population expansion. At a conti- stantially higher in the Jivaro and Huancas (96% and 92%,
nental level, few other populations showed significant respectively). The Cajamarca, on the other hand, had the
values for both indices (Supporting Information Table S7). smallest Native American component (46%).
Looking at the full set of 23 Y-STRs, all of the study pop-
ulations exhibited relatively high haplotype diversity
Substructure in the Chachapoya area. To explore differen- (H > 0.92) when compared to most Native American popu-
tiation within the cloud forest region, pairwise compari- lations. The Chachapoya (H 5 0.9974) and the Cajamarca
sons among the Chachapoya subunits (see Materials and (H 5 1) displayed the highest values (Table 4). Again,
Methods) were estimated (see Supporting Information since we focused on the Native American component,
map: Chachapoya subunits.tiff). Interestingly, we found nonindigenous haplotypes were excluded for subsequent
variability in haplogroup frequencies between the Cha- analyses. Similarly, in continental comparisons for a
chapoya subunits, such as the absence of haplogroup A in subset of relevant additional populations from Peru with
Rodriguez de Mendoza and Chilchos and the near absence nine loci (DYS19-DYS385 I-DYS385 II-DYS389 I-DYS389
of haplogroup D in La Jalca (3%, 1 individual). There II-DYS390-DYS391-DYS392-DYS393), the Chachapoya
were also relatively large and significant genetic distances (H 5 0.9833 6 0.0066), and the Cajamarca (H 5 1 6
between the Pomacochas and the Chillao (UST 50.074, 0.0388) exhibited high haplotype diversity similar to the
P 5 0.00050) and also between Chillao and Rodrguez de population from Junn (H 51) and the Quechua-speaking
Mendoza (UST 50.070, P 5 0.0009). Further hierarchical population from Cusco (H 5 0.9613). The other two study
analyses, AMOVA tests, among these subunits and the populations, the Jivaro (H 5 0.8972) and the Huancas
Hd SD Hd
Population N A h SDh loci loci MNPD SDMNPD
enough to greatly reduce the genetic diversity. The genetic from the human groups that first colonized South Amer-
effects of the bottleneck are determined by a number of ica, high gene flow into the region from different direc-
factors, such as the minimum number of individuals dur- tions at different time points, or both.
ing the bottleneck as well as its duration and rebounce Since there is no historical evidence for substantial
dynamics, and genetic diversity may survive in situations migration into the Chachapoya territory, we are more
that appear severe (for instance, see Kekkonen et al., inclined to consider that the Chachapoya people have
2012). However, as the diversity levels observed in the retained ancient Native South American haplotypes. This
region are higher than in most other native populations in view recalls an old hypothesis proposing a dispersal route
South America, we must also consider other factors southwards, through the eastern slopes of the Andes in
beyond the bottleneck dynamics alone. northern Peru and South Ecuador used by the first
First, the Chachapoya could have harbored higher Andean settlers (Hester, 1966; Lothrop, 1961; Sauer,
genetic diversity in the remote past. The consensus is that 1944), one that was revised relatively recently based on
the ancient Chachapoya sustained large population sizes early lithic assemblages found in the region (Church and
prior to the arrival of the Incas and the Spaniards von Hagen, 2008). This area could indeed have served as
(Church and von Hagen, 2008; Schjellerup, 2005). How- a corridor because of two important factors, the drop in
ever, documentation is scarce and settlement density altitude of the mountain range and the shorter distances
alone is not enough to give support to this idea because of to reach Amazonia from the coast, which probably facili-
the lack of a chronological framework for the whole tated, among other things, the horizontal mobility of peo-
region. Additional archaeological work in the region may ple. Certainly, this territory has been considered a
shed light on the issue of population movements into this crossroads (Church and von Hagen, 2008), a feature that
area. probably contributed to further enrich the complex gene
Another possibility is that the policy of Indian reduc- pool of the Chachapoya through time.
tions (reducciones) implemented in colonial Peru (Medina, The Y-chromosome data indicate a slightly different pat-
1975) may have had a different impact on the local popu- tern. The Chachapoya have affinities with very few Peru-
lations, increasing the amount of diversity. Based on docu- vian populations, and in South America only with the
mentation from the 16th and 17th centuries, Espinoza Mapuche. Unlike mtDNA data, the main study population
drafted a map of more than ten ethnic groups that occu- does not assume a basal position and does not cluster with
pied the neighboring area to the northwest of the Chacha- Andean or Amazonian populations in phylogenetic trees. It
poya, in a short segment of about 150 Km from north to was mentioned above that depopulation probably had a
south (Espinoza, 1973). These populations, along with more dramatic effect in the Chachapoya males. Possibly
other groups to the southeast of the Chachapoya territory much of the pre-existent Chachapoya genetic diversity was
(Hivito and Cholon), show the high ethnic diversity in a lost due to the series of transformations that began with
region that could have served as a source of migrants. the arrival of the Incas in the region and the application of
Last, probably due to the singular geography and topogra- their mitmaq policy, resulting in affinities restricted to
phy of this stretch of the Andes, the diversity in the area very few populations in South America in general.
could have been continuously enhanced by the movement
of people through the region during and after the histori- Regional patterns and differentiation within the Chachapoya
cal demographic collapse reported for this area. To understand population interactions in this transi-
We cannot assess unequivocally which factor contrib- tional region, we also sampled immediate neighbors of the
uted the most to the high levels of diversity observed in Chachapoya. The overall picture indicates that the Cha-
the Chachapoya. However, the data provide evidence of chapoya are not clearly differentiated from the Huancas
the Chachapoya being a very different population from and the Cajamarca. Unlike these two Andean groups, the
most indigenous groups in South America, one showing Jivaro from the tropical rainforest in the north remains
low levels of genetic drift. distant from the Chachapoya, although a northern Cha-
chapoya subunit (Pomacochas) seems to cluster better
Genetic affinities
with the Jivaro (see below).
One of our aims was to situate the Chachapoya in a The Huancas were considered an enclave population
broader context, comparing it with populations from dif- originally from the central Andean area that specialized
ferent origins and diverse genetic composition. For in manufacturing pottery (Colin, 1907). Their arrival is
mtDNA data, the Chachapoya have, as expected, genetic estimated to have been around the time the Incas were
affinities with several Peruvian populations (e.g., Ancash, expanding in central Peru. We expected to find small
Tayacaja, San Martn, Chiribaya), but also relatively genetic distances between the Chachapoya and the Huan-
short genetic distances to populations from both the cas since, for the purpose of admixture, more than 20 gen-
Andean (e.g., Chile, Coya) and Amazonian groups such as erations have elapsed. The Y chromosome data support
the Kalina. Due to this pattern, the Chachapoya do not this idea, while the mtDNA information suggests they are
show any particular population affinity in the phyloge- relatively distant from each other (UST 5 0.060, P < 0.01).
netic trees but assume a basal position and a central loca- The Chachapoya people spoke a language that did not
tion in the PCA plot based on haplogroup frequencies. survive the series of biological and cultural transforma-
Interestingly, another population showing this pattern is tions that occurred during the onset of the second half of
the Tayacaja, which also exhibited diversity levels compa- the 16th century. Guffroy (2006) points out that the his-
rable to the Chachapoya. This finding suggests that the torical Jivaro and the ancient Chachapoya shared some
Chachapoya have retained mtDNA diversity common to toponyms as well as other cultural features (Guffroy,
many populations from different regions in South Amer- 2006; Taylor, 1996, 2000). These and other similarities in
ica which, in turn, may indicate the survival of diversity their culture led scholars to suggest that there might
have been connections between them in the past. Our resemble that of populations from other regions in South
genetic data do not support this notion, although we America, probably because of the unique environmental
should bear in mind that genes and cultural traits are and topographic features of this part of the Andean Moun-
transmitted in different ways. tain Range. We are aware of the difficulties in assessing
The archaeological record has shown that the Chacha- whether this distinct pattern reflects very early genetic sig-
poya maintained interactions with societies as far as the natures or the action of different evolutionary forces
northern Peruvian coast during the Late Intermediate through time. However, a step forward is to genetically
period (10001470 C.E.)(Church and von Hagen, 2008). characterize populations from the Andes-Amazon divide to
The Cajamarca, located to the west of the Chachapoya ter- better understand broader processes such as the early colo-
ritory, were the people who, probably by proximity, influ- nization of the northeastern Peruvian Andes and Amazo-
enced more the Chachapoya society. This is evident, for nia. The genetic profile of the Chachapoya indicates that
example, in the abundance of Cajamarca pottery in most populations that developed in intermediate ecological
archaeological sites investigated until now (Church and regions have a complex genetic composition which has
von Hagen, 2008; Ruiz, 1972; Schjellerup, 2005). More- been influenced by their position with regard to other civili-
over, during Inca times, the Cajamarca people were zations in the northeast Andes. In fact, the cultural evi-
among the few mitmaq present in the region (Schjellerup, dence points to the vital role of the societies in this area, as
2005). The mtDNA data support the close affinity of the mediators of various cultural interaction spheres since
Chachapoya and the Cajamarca people. However, since very ancient times, and also to their importance in the for-
there has been a strong migration from Cajamarca to mation and transformation of complex societies (Church
Amazonas (but not vice versa) during the second half of and von Hagen, 2008; Guengerich, 2015). Studies of extinct
the last century, we cannot assess with certainty whether and extant populations in similar transitional zones have
our observations reflect an ancient pattern or a more also found complex genetic configurations with high levels
recent one. of genetic diversity (Casas-Vargas et al., 2011; Corella
To study the genetic heterogeneity within the Chacha- et al., 2007). Overall, even though our study shows some
poya region, the samples were divided geographically into general patterns in the populations from transitional
sub-areas. We adopted a division based on the ancient dis- zones, the distinctiveness of the Chachapoya suggests that
tribution of ethnic groups (see Methods). For mtDNA, we extrapolations to similar regions (e.g., Scliar et al., 2014)
detected the absence (or very low frequency, N 5 1) of hap- must be considered with care. Additional data from this
logroups in specific subunits. Also, we found that the population and others surely will help to broaden our
Chillao mtDNA data differentiated from the others. On the knowledge on the population history of the northeastern
other hand, for the Y chromosome, the group that differen- Peruvian Andes and ultimately of the Americas.
tiated the most was La Jalca. Similarly, although with a
very weak signal and based only on the AMOVA test for ACKNOWLEDGMENTS
mtDNA, the northern subunit Pomacochas could better be
We are very grateful to all participants in this study. We
grouped with the Amazonian Jivaro population than with
thank also many members of each community (Chachapoya
any other neighboring Chachapoya subunit. However, Y-
and non-Chachapoya people) visited during the sampling in
chromosomal data do not differentiate this subgroup. Par-
Amazonas for their assistance and interest in the study. We
tially in line with early documents depicting the Chacha-
are grateful to Paula Lehtoaho and Kirsti Hook for labora-
poya as an aggregate of several ethnic groups, the patterns
tory assistance. We are very thankful to Warren Church for
of local variation suggest genetic heterogeneity within
insightful comments on part of the ethnohistorical perspec-
the region. Indeed, new archaeological evidence points to
tive of an earlier version of this manuscript as well as to the
dissimilarities in material attributes in this territory,
two anonymous reviewers for their help in improving the
although scholars do not agree on the number of pre-Inca
manuscript. Contributions: All authors designed the
subunits or their geographic boundaries (Church, 1996;
study. EG collected and typed the samples. EG and JUP
Church and von Hagen, 2008; Espinoza, 1967; Schjellerup,
analyzed the data. All authors drafted the manuscript.
2005; Zevallos, 1995). Conclusive genetic evidence for this
heterogeneity can, however, only be attained through anal- LITERATURE CITED
yses of ancient DNA and depend on the availability of rep-
resentative archaeological sample collections. Adelaar WFH. 2004. Languages of the Andes. West Nyack, New york:
Cambridge University Press.
Concluding remarks Barbieri C, Heggarty P, Yang Yao D, Ferri G, De Fanti S, Sarno S, Ciani G,
Boattini A, Luiselli D, Pettener D. 2014. Between Andes and Amazon:
Even though there have been efforts to acquire samples the genetic profile of the Arawak-speaking Yanesha. Am J Phys Anthro-
of significant size for population genetic studies in South pol 155:600609.
Behar DM, Villems R, Soodyall H, Blue-Smith J, Pereira L, Metspalu E,
America, many regions remain unexplored. Recently, high Scozzari R, Makkan H, Tzur S, Comas D, Bertranpetit J, Quintana-
throughput genotyping technologies have been utilized to Murci L, Tyler-Smith C, Wells RS, Rosset S. 2008. The dawn of human
retrieve large amounts of data from existing samples (Per- matrilineal diversity. Am J Hum Genet 82:11301140.
ego et al., 2010; Tamm et al., 2007). However, other than Bolnick DA, Bolnick DI, Smith DG. 2006. Asymmetric male and female
genetic histories among Native Americans from Eastern North America.
major consortia efforts (e.g., Genographic Project), there Mol Biol Evol 23:21612174.
has been little emphasis on new sample collections from Cabana GS, Lewis CM Jr, Tito RY, Covey RA, Caceres AM, De La Cruz AF,
unstudied populations. In some areas, this incomplete Durand D, Housman G, Hulsey BI, Carlo Iannacone G, Lopez PW,
genetic landscape oversimplifies the interpretation of pop- Martinez R, Medina A, Ortega O, Osorio KP, Polo SI, Rojas P, Rubel M,
Smith HF, Smith SE, de Celis Massa VR, Lizarraga B, Stone AC. 2014.
ulation patterns at broader geographical scales. Population genetic structure of traditional populations in the Peruvian
The Chachapoya from the cloud forest of Northeast Central Andes and implications for South American population history.
Peru have shown a peculiar genetic pattern that does not Hum Biol 86:147165.