haplogroup b2b

original image (585KB)

larger image (146KB)

Table 1. Diversity indices in Bolivian mtDNAs and main American and African regions.

doi:10.1371/journal.pone.0058980.t001

The most apparent geographic pattern was found when examining molecular diversity values by main ecological region. The Andean followed by the Sub-Andean regions had substantial lower diversity values than the Llanos. Therefore, the diversity was found to increase longitudinally, from the high mountains to the lowlands of the Llanos.

Diversity was particularly low in some ethnic groups compared to general urban and rural populations. For instance, the Ayoreo (from Bolivia and Paraguay) and the Aymara had extremely low diversity values compared to the average values observed in Native Bolivians and Bolivians in general (Table 1), a fact that is most likely due to strong founder effect in the case of the Ayoreo [19], while in the Aymara discussed by Corella et al. [20] it was most likely due to the small sample size analyzed (the values were compared with those obtained with the Aymara population from Gayá-Vidal et al. [22] using a larger sample size). Our sample from general rural and urban Beni showed significantly higher diversity values than those observed from other Native American groups from the same department (e.g. the Piedmont populations of Moseten, Chimane, etc [20], the Llanos populations analyzed in [18]; see Table 1).

There are only few haplotypes in Bolivia that appears more than two or three times in the whole dataset, again indicating the high mtDNA diversity of the country (Figure S1 and Figure S2). When evaluating the differentiation between departments, it was again observed that comparisons of other departments with Santa Cruz were the ones that displayed the highestFST values (Figure S3). The influence of this department is also indirectly observed when studying FST values between the main ecological regions, indicating that the comparisons between Llanos and the other two areas showed the largest values (Figure S4).

Finally, Figure S5 and Figure S6 show the expected (virtual) heterozygosity in the main departments and in the main regions, respectively; the dispersion of values indicated the existence of substantial mtDNA diversity between departments.

Phylogeography

The mtDNA Native American component predominates in the Bolivian population (98.4%); most of the variation can be classified into one of main Native American haplogroups, (A: 14.3%, B: 52.6%, C: 21.9%, D: 9.6%), Figure 1.

Download:

original image (674KB)

larger image (408KB)

Figure 1. Map of Bolivia showing the location of the samples collected in the present study.

The pie charts represent the distribution of basal Native American haplogroup frequencies in the region.

doi:10.1371/journal.pone.0058980.g001

There were remarkable differences in haplogroup frequencies between the main Bolivian departments (Figure 1). For instance, the department of La Paz had the following haplogroup composition, A: 10%, B: 71%, C: 12%, and D: 7%, which was in sharp contrast with the composition observed in the department of e.g. Beni, A: 12%, B: 30%, C: 48%, and D: 10%. The haplogroup distribution found by Afonso-Costa et al. [21] in a smaller sample from La Paz was slightly different to that found in the Bolivians from La Paz studied here, but both samples agreed regarding the high proportion of haplogroup B in this area. This geographic pattern mirrors in reality the location of the departments at different altitudes; thus, the most important differences are observed between Andean and Sub-Andean populations versus Llanos. For instance, haplogroup B is the predominant haplogroup in the Altiplano (71%), which decreases to 61% in Sub-Andean populations and 32% in Llanos (Figure 1).

As expected, Bolivia shares more haplotypes with South America than with Meso and North America (Table 2; Table S2), although these values have to be interpreted with care due to the different sample sizes. Note however that the largest sample size is in North America, but Bolivia only shares 18% of the haplotypes with this region but 49% with other South American populations (Table 2). It is also worth mentioning that 48 haplotypes are shared between the three American regions; therefore, of the 68 haplotypes shared with North America, most of them are also shared with South America (Table 2, Table S2). A large amount of haplotypes has been only observed in Bolivia (250 out of 383 different haplotypes; 65%).

Download:

original image (71KB)

larger image (27KB)

Table 2. Shared haplotypes between Bolivia and other American regions.

doi:10.1371/journal.pone.0058980.t002

There are eight mtDNAs that most likely belong to haplogroup C1d (all carry the characteristic transition 16051); there is another mtDNA that could be allocated to the northern South American cluster C1d2a, which does not carry transition 16091 but carries transition 16209 instead. C1d has been reported as one of the founding Paleo-Indian mtDNA lineages of the American continent [5] and is therefore found along the double continent. As has been previously described [5], C1d did not necessarily follow a main Pacific coastal route once it reached South America; this would explain why this lineage appears more in the Llanos than in the Andean region.

The five mtDNAs belonging to D4h3a found in our Bolivian samples most likely arrived at about the same time as C1d but probably followed a different route. According to [4], D4h3a appeared to spread along the Pacific coast. In agreement with this hypothesis, the five D4h3a Bolivians were observed in the Andean region, and three out of the five D4h3a lineages belong to the Peruvian sub-clade D4h3a3.

According to expectations, no members of the Paleo-Indian founder X2a, which has a North American distribution, have been found in Bolivians. Also, in good agreement with Bodner et al.[9], we did not find any members belonging to southern cone lineages D1j and D1g, favoring the hypothesis that these two haplogroups crossed the Andes at lower latitudes (Chile towards Argentina), where they were most likely incubated.

Very recently, de Saint Pierre et al. [51] proposed two new lineages, B2i2 and C1b13, which are thought to have originated in the Southern Cone, and are closely associated with the Paleo-Mapuche people in Chile and Argentina. Members of these two lineages have only been sporadically found outside of Chile and Argentina. Here, we only found one member of C1b13, and according to the hypothesis formulated by de Saint Pierre et al. [51], this mtDNA could have arrived in Bolivia from the South instead of from the West.

Some Bolivian mtDNAs are particularly interesting given that they appear at relatively high frequencies in this region and neighboring areas but are absent in the rest of the continent. For instance, the motif 16189 16217 16290 appears six times within haplogroup B4 in different regions from Bolivia; it also appeared in four Andean Peruvian Quechua individuals [52]. The motif 16173 16192 16223 16298 16325 16327 16346 (haplogroup C) appears twelve times in Bolivia, and all of them in the Llanos. Curiously, this motif was found also in another individual from La Paz in [21] and another one from Chaco, in the North of Argentina, close to the Bolivian Llanos [11].

Haplogroup L (L referring to all mtDNA branches excluding macro-haplogroups M and N) mtDNAs represent the sub-Saharan exclusively maternal genetic component of the country, which in this case only accounts for 1% of the total mtDNA pool (Table S1). The proportion of African recent ancestry in Bolivia is very low compared to other South American countries that were more influenced by the African slave trade [16], [17], [53]. There are only seven different haplotypes belonging to some of the typical sub-Saharan sub-clades. Individual #BNI19 belongs to L0a1b2, which is spread in all of sub-Saharan Africa as well as in America, but carries a distinctive transition (within L0a1) at position 16271 (which gets an score of nine in the mutation list of Soares et al. [41]). The L1b (#BNI75) and L1c1a1 (#LaPaz467) profiles most likely come from the West-Central African region. The L1c3b1a and the L3f1b4a profiles appear in Cabinda, Angola and Mozambique [54]–[57], but also in West-Central, in Gabon [58]. As expected [53],[59], the most likely geographic origin of the African Bolivian profiles is west-central Africa, but also encompasses the Southwest and the Southeast.

It is also curious that the European mtDNA component represents less than 1% of the Bolivian population. With the level of resolution of the HVS-I segment, it is not possible to assign European lineages to a particular region in Europe; the four sequences observed here belong to haplogroups H1af, K and X1. An Asian origin for the profile C16104T C16223T cannot be disregarded given that two exact haplotype matches appeared in China [60] but any in Europe (in our in-house database of more than 26700 HVS-I profiles).

Entire genomes from Native American Bolivians

In order to investigate the Native American lineages in our Bolivian samples further, we selected those showing a distinctive pattern from the control region data. Thus, we chose two mtDNAs carrying the tandem variants T16097C A16098G on top of the sequence motif for haplogroup A2, and a group of six mtDNAs all carrying transition G16145A on top of the B2 basal motif (Table S3). Entire genome sequencing revealed some interesting phylogeographic features of these mtDNAs.

The tandem transitions T16097C A16098G conform per se a very distinctive motif that is found very rarely in mtDNA databases. Within haplogroup A2, this motif alone (without any other control or coding region variant) defines a novel branch of the Native American phylogeny, here named as A2ah (Figure 2). The two genomes show variability within this clade and, given its very restrictive location within Bolivia, this most likely point to an origin in Bolivia or the surrounding territories. We further investigated A2ah in public databases of entire mtDNA genomes. We found just one entire genome (JQ702082) that carries the transition A16098G on top of the A2 motif but it differs substantially from the two genomes found in Bolivia (apart from lacking the transition T16097C); therefore, its inclusion within A2ah should be considered to be very dubious. According to the entire genome sequences available, it can be said that haplogroup A2ah originated about 5,200 thousand years ago (kya) (although with a large 95%CI: 0.1–10.5). We additionally investigated the phylogeographic characteristics of these lineages by looking at the abundant amount of data available in the literature on control region segments. We only detected nine mtDNAs matching the motif of A2ah. Four out of ten were observed in the data reported by Behar et al. [61] but geographic information was not available. Two other HVS-I mtDNAs were found in the ‘Hispanic’ North American subset of the SWGDAM database [62]; three other A2ah mtDNAs were observed in South America, two among the Brazilian samples of Alves-Silva et al. [13], and one in the Toba from Gran Chaco (North Argentina) [63]. In Bolivia, only four A2ah members were found (0.5%), all of which were observed in the Llanos (three in Santa Cruz and one in the Beni department). In general, the HVS-I data suggest that this lineage could have originated in Bolivia or some place in Central South America. The two members observed in the ‘Hispanic’ samples from the SWGDAM could perfectly represent recent immigrations into USA. The motif T16097C A16098G was also observed in Mainland Scotland (Western Islands; Isla of Skye) and two other samples in [61], but none of the mtDNAs carried the variation defining haplogroup A2.

Download:

original image (991KB)

larger image (625KB)

Figure 2. Maximum parsimony tree of the main branches characterizing haplogroups A2 and B2, indicating the new branches generated in the present study, namely, A2ah and B2o.

The mutations are displayed along the branches: numbering is according to the rCRS[71]: a mutation alone indicates a transition while an upper case letters as suffix (A, C, G, T) indicates a transversion; “+” and “d” refer to insertions and deletions, respectively; “s” refers to a synonymous variant while non-synonymous variants are indicated by way of the amino acid change; a star indicates a variant located in the ATP6 or ATP8 gene; a prefix @ indicates a back mutation; and underlined positions indicate parallel mutations. Sample ID (see Table S3) for the entire genomes generated in the present study are as follows: #1: COBIJA577, #2: COBIJA604, #3: STACRUZ261, #4: BENI86, and #5: STACRUZ221.

doi:10.1371/journal.pone.0058980.g002

A phylogenetic conflict exists when trying to reconstruct the most parsimonious tree of the six B2 mtDNAs carrying transition G16145A. Three of the genomes (#6, #7 and #8 in Figure 3) carried the synonymous transition G6755A, while three other genomes (#1, #2 and #3) lacked G6755A but carried the non-synonymous transitions T7270C and T16092C instead (Figure 2). Given that G16145A seems to have a much higher mutation rate than G6755A (22 versus 2 mutational hits in Soares et al. [41], and 24 versus 5 in Phylotree; respectively), we decided to resolved the phylogeny as shown in Figure 2 and Figure 3.

Download:

original image (939KB)

larger image (609KB)

Figure 3. Maximum parsimony tree of haplogroup B2b.

The map on the top right inset indicates the geographic location of the B2b genomes represented in the tree; circles are proportional to the sample size. The arrows indicate the two tentative migration routes (a and b in the map) of this lineage (see main text) along the American continent. References for entire genomes are as follows (see Table S1 and S3): #6: BENI90, #7: STACRUZ258, and #8: COBIJA110, #9 Mco-10 (present study, Table S1 and S3). For more information, see the legend of Figure 2.

doi:10.1371/journal.pone.0058980.g003

Thus, the motif T7270C T16092C G16145A defines a new branch of the Native American phylogeny named here as B2o, which is represented by three Bolivian genomes (Figure 2). No other entire genome was observed in public resources belonging to B2o. Dating the phylogeny of B2o (Figure 2) indicates that this haplogroup originated about 2.6 kya (95%CI: 6.8–13.2). Searching control region databases, we observed only two HVS-I candidate B2o sequences: one was observed in San Martin de Pangoa in Perú, a small town on the eastern slope of the Andes inhabited by Quechua and Nmatsiguenga people [64], while the other B2o sequence was observed in Guam (an island located in the western Pacific Ocean, territory of USA). The latter, however, carries the motif T16189C T16217C A16247G C16261T, which is very common in the Micronesia, Australia, etc. [65], [66] and points to a different clade of the worldwide phylogeny, haplogroup B4a. In Bolivia, we found seven B2o representatives (1%), most of them in the Llanos (five in Pando and one in Santa Cruz departments), and one in a Sub-Andean locality of La Paz. Therefore, altogether the data suggest that this lineage has also originated in Bolivia or some nearby region, most likely located in the Andes.

The entire genomes of the three Bolivian mtDNAs carrying G16145A but lacking T16092C, all share the transition G6755A. By searching the databases on entire genomes for the transition G6755A, we observed 10 additional mtDNAs belonging to this branch. In reality, this clade had already been described in the recent literature [6], [7] and was named B2b. However, its internal variation was never analyzed in detail. Figure 3 shows the phylogeny of B2b and the geographical location where the entire genomes were observed. Geographic and/or ethnic affiliation was available for eleven entire genomes (including the Bolivian ones): one Pomo (North California, USA) [8], one Mexican American [67], two Venezuelans from Pueblo Llano[68], one Yanomama (Brazil, Venezuela) [8], one Cayapa (Equador) [6], one Kayapó/Kubemkokre (South Amazonia, Brazil) [8], one Xavante (Brazil) [8], and three Bolivians (present study). We additionally detected one entire genome in our DNA databank in a Mataco Native from North Argentina, which was added to the phylogeny of Figure 3 (#9). According to the phylogeny of B2b in Figure 3, B2b appeared about 21.4 kya (95% CI: 4.4–26.7). The Bolivian sub-clade B2b2 is much younger, approximately 15.2 kya (95%CI 4.4–26.7). In the large HVS-I database, we only observed 14 B2b candidates, almost overlapping the territories represented by the entire genomes. In Bolivia, we counted seven B2b candidates: one Andean (La Paz), four Sub-Andean (Cochabamba, La Paz and Chuquisaca), and two in the Llanos (Beni and Santa Cruz).

AMOVA analysis of mtDNA profiles

AMOVA was undertaken considering different hierarchical levels; by Department, Rural vs. Urban, and Andean vs. Sub-Andean vs. Llanos. As expected, the within-population variation accounted for most of the variance (ranging from 93–100%; Table 3), independently of the population sub-division employed.

Download:

original image (53KB)

larger image (17KB)

Table 3. AMOVA computed based on haplotype pairwise differences of Bolivian populations (significant tests: 20,022 permutations; adjusted P-value<0.0000).

doi:10.1371/journal.pone.0058980.t003

AMOVA carried out by departments and ecological region provided the highest values of among-population variance (ranging from 10.48 to 11.58% of the total variance). The values are similar given the correlation that exists between these two classification criteria. The rural vs. urban division indicates virtually no correlation among group subdivision. The analysis carried out by province did not increase the among-population variance, indicating again that the main factor influencing within-population variation was the different altitudes in the country.

Continental ancestry in Bolivians

A panel of 46 AIMS was genotyped in Bolivians in order to infer their main continental ancestry. Data from main continental reference samples (Africa, East Asia, Europe and America) were used as classification sets.

Analysis of ancestry was carried out using STRUCTURE. Figure 4 summarizes the average continental ancestries observed in Bolivians under different grouping schemes, while Figure 5shows the STRUCTURE bar-plots. The analysis showed that, on average, 71% of the component in the total Bolivian sample is Native American, followed by 25% of European ancestry. When examining the ancestry by departments, La Paz was the region that showed the highest Native American ancestry (79%) (Figure 4 and Figure 5), and, therefore, the lowest European component (19%). On the other side is the department of Santa Cruz, with 57% of Native American ancestry and 39% European. The African component was very low in all of the departments, showing the highest value in the department of Pando (2.5%), in the North. When examining by ecological regions, it was also evident that the Native American component is higher in the mountainous West (80%), and decreases progressively eastward: sub-Andean (70%) and Llanos (64%). The differences are less apparent when examining ruralvs. urban areas (74% and 69% of Native American component, respectively).

Download:

original image (848KB)

larger image (268KB)

Figure 4. Average continental ancestry of Bolivians analyzed using a panel of 46 AIMs.

These values are obtained from the STRUCTURE analysis and the optimal value K = 4.

doi:10.1371/journal.pone.0058980.g004

Download:

original image (3.53MB)

larger image (2.63MB)

Figure 5. STRUCTURE analysis of Bolivians based on the 46 AIMs panel genotyped in the present study.

Each bar plot represent analysis using different grouping schemes for the Bolivian samples. Only the results for the optimal K = 4 are represented (see complete analysis in Text S1).

doi:10.1371/journal.pone.0058980.g005

The STRUCTURE bar-plot of Figure 5 indicates the ancestral membership for each individual in the source population and the Bolivian samples. It clearly shows that most of the individuals have a Native American ancestry, with only a few exceptions. The high European component of Santa Cruz compared to other departments is also evident in this bar-plot, and is also evident in the Andeans. When looking at the most extreme values, we found that there are only a few that have European ancestry above 50% in most of the departments. It is, however, rare to see individuals with high African membership; the highest values correspond to 23% in one individual from La Paz and another one from Pando (17%).

The membership of individuals into the East Asian cluster is significantly higher in some individuals, as is also evident in the STRUCTURE plots. However, this perhaps mirrors the fact that the separation between the Native American component and the East Asian one is not perfect. The overlap between Native Americans and East Asians can be observed more clearly in the PCA analysis (see below). This fact has been already observed in Pereira et al. [37]. Thus, most of the East Asian component would be captured by the Native American cluster in cases using a three-group structure analysis (Native American, African, and Europe).

The PCA agrees well with the results observed in STRUCTURE. The Bolivian profiles group mainly with the Native American ones from HapMap, with only some samples showing a projection towards the European cluster (Figure 6). The minor African component detected in the STRUCTURE analysis seems not to be relevant in the PCA; only one sample show some affinity to the African cluster; it is in fact the same sample with a 23% African ancestry in STRUCTURE. The PCA carried out by departments (Text S1) showed that the department of La Paz (Andes) was more tightly grouped than the profiles from other departments; in the other pole are the departments of Santa Cruz and Pando, showing more dispersed patterns (Llanos). This west-east pattern is more evident when observing the PCA by main regions (Figure 6): the Andean profiles are more tightly grouped in one pole of the plot, while the profiles from Llanos are in the other pole and are more dispersed, with the sub-Andean profiles occupying an intermediate position between East and West. This dispersion is also mirrored when examining the standard deviation (SD) of the Native American membership values in the main regions: (i) Andean: mean = 79.7%, SD = 0.086%; (II) sub-Andean: mean = 69.6%, SD = 0.09; and (iii) Llanos: mean 64.2%, SD = 0.137% (Table S4); given that the department of La Paz is basically an Andean department, it also shows the lowest SD (0.85%).

Download:

original image (1.58MB)

larger image (1.16MB)

Figure 6. PCA of Bolivian profiles based on the 46-AIMs panel genotyped in the present study.

Each figure shows different grouping schemes mainly aiming to show the within-Bolivian diversity.

doi:10.1371/journal.pone.0058980.g006

Finally, measuring ancestry using AIMS depends on many factors, such as the use of different panels of SNPs (not only different SNPs, but also different numbers of them), different classification test samples, sample sizes, etc. Twelve of the samples genotyped in the present study for the 46-indel panel were also genotyped in Galanter et al. [23], thereby offering a good opportunity to investigate the ability of different AIM panels to estimate percentages of continental ancestry. As shown in Figure 7, the ancestry memberships obtained using Indels lead to a decreased Native American ancestry (and proportional increased European ancestry) compared to the estimates obtained using the LACE panel (on average, the difference is about 15%). Given the way in which the LACE panel was designed (in order to balance the informative content of the different AIMS included in the panel), it seems logical to believe that the Indels panel would tend to over-estimate the proportion of European ancestry in Bolivians. Note, however, that the genome-wide study of Watkins et al. [24] carried out in few Bolivians estimated a higher European component (12%) compared to the LACE. Further genome-wide studies on a larger Bolivian sample will allow better estimates of continental ancestries to be obtained.

Download:

original image (672KB)

larger image (323KB)