Y chromosome
A total of 140 individuals were genotyped and compared with other highland populations of the South American Genographic Project database and published data (n = 376) [12,13,14,15, 17]. A list of 17 short tandem repeats (STRs) haplotypes of Q-L54 lineages (major native Y chromosome lineage) obtained for the studied populations is presented in Additional file 1: Table S1a.
We used median joining networks to infer the genetic relationships among the STRs haplotypes at the individual level that are shown in Fig. 2. The ‘Cañaris’ of Ecuador are a very heterogeneous group, similar to other Ecuadorian Quichua speakers, showing a considerable diversity of haplotypes (Fig. 2). This scenario fits with the stories suggested in the chronicles: the current Province of Cañar was a conglomerate of several chiefdoms, similar to those of Quito, and its territorial extension included the current region of Piura in Peru [7, 18]. Likewise, the results show that individuals from the Cañar region (Cañar_EC) do not share STR haplotypes with those of the Kañaris and Inkawasi communities. There are some haplotypes of the Ecuadorian ‘Cañaris’ (labelled with the letter e) closely related to some from Cajamarca, Chivay (Arequipa), Cusco and Lake Titicaca, an association supported by colonial records. Individuals from the Inkawasi and Kañaris communities are a conglomerate of at least five clans, and most of their haplotypes are shared. Some haplotypes from both localities are more related to those from Cajamarca and Chachapoyas, congruent with the results from linguistic analyses and their geographical proximity [10]. In contrast, there is a close relation between a group of related haplotypes of the Kañaris and Inkawasi and two Quichua-speaking individuals from Ecuador (labelled with the letter p, Fig. 2). However, these two individuals from Ecuador are more related to the Chachapoyas population, and it is probable that their ancestors were mitmakuna, brought from the north of Peru to Ecuador, including the individuals labelled with the letter c (Fig. 2).
At the population level, the Analysis of Molecular Variance (AMOVA) results (see Additional file 1: Table S1b), visualised using Principal Coordinates Analysis (PCoA) (Fig. 3), suggest that there is a genetic affinity between the Cañar_EC, Pastos_EC and Quichua_EC. The RST index of Cañar with the other two groups is 0.046 (p = 0.01504) and 0.047 (p = 0.0053), respectively (see Additional file 2: Table S2). This finding is consistent with observations at the individual level. However, due to the low haplotype diversity (h), the populations of the Inkawasi (h = 0.833) and Kañaris (h = 0.908) (see Additional file 3: Table S3) are separated from other populations. This observation makes sense in light of the historical data, which indicate that during the colonial and republic times, both communities, and other nearby ones such as Penachí and Salas, remained in relative geographic isolation [9]. The AMOVA analyses appear to reflect the differences between the 10 populations studied (Rst = 0.099, p < 0.001) (see Additional file 1: Table S1b). The RST value between Cañar_EC and Kañaris showed moderate difference (RST = 0.141, p = 0.001). There is also a genetic difference between Kañaris and Inkawasi (RST = 0.151, p = 0.008). A high level of difference is observed between Cañar_EC and Inkawasi (RST = 0.237, p < 0.001) (see Additional file 2: Table S2). However, due to the connection of the haplotypes of Kañaris–Inkawasi with other haplotypes from Ecuador (labelled with the letters p and c, suggested as probable mitmakuna from the Peruvian north, particularly from Chachapoyas, Fig. 2), the expected relationship is observed in the PCoA at the population level. Kañaris is more connected to the north highland populations of Peru than to the southern ones. This observation is consistent with the macro regional genomic landscape inferred using a DNA chip with about 630,000 autosomal SNP markers, which showed close genetic relationships between the autochthonous Northern Coast Highland populations from Peru [19]. Kañaris and Cajamarca have been historically associated since the Inka era, in addition to being geographically adjacent [1, 9]. It seems that some of the inhabitants of the Inkawasi community are related to individuals from southern Peru, suggesting that a proportion of their ancestors were brought as mitmakuna [7, 8]. Taking into account the geographic isolation and genetic drift, the Inkawasi community was differentiated from the rest of the populations in a similar way to the Kañaris and there are also differences between the two groups (Fig. 3). However, due to cultural issues and similar geographic niches, there was inevitably gene flow between both communities, as shown in the analysis at the individual level (Fig. 2). Another observation arising from the PCoA scatter plot is the genetic affinity between the populations of Cajamarca and Chachapoyas. The RST value is 1.3% (p = 0.079), a small difference (see Additional file 2: Table S2).
In the Amantani population there is a high haplotype diversity (h = 0.966), despite its geographical isolation on an island located in the Lake Titicaca. Part of the male population of Amantani is associated with the populations of Ecuador, Cajamarca, Chachapoyas and Cusco, suggesting that considerable gene flow took place among the Andean communities. However, the macro regional scenario depicted in the PCoA space, with the exception of Amantani, reflects a genetic gradient according to geographic location.
Mitochondrial DNA
A total of 182 mitochondrial DNA (mtDNA) control region haplotypes were assigned to four Native American maternal lineages (A2, B2, C1 and D1) and compared with other highland populations as done previously with the STR haplotypes. Genetic analysis was carried out using only the autochthonous lineages (n = 379). A list of the mtDNA control region data according to the rCRS is presented in Additional file 4: Table S4a. With some exceptions, the distribution of haplogroup frequencies showed that B2 is the most frequent lineage among the highland populations (see Additional file 4: Table S4c).
At an individual level, we inferred phylogenetic relationships using median joining networks, as with the paternal line. According to the results shown in Figs. 4, 5, 6 and 7, with the exception of one haplotype shared by the D1 lineage between Cañar and Kañaris, there are no genetic affinities between these populations. However, this is not the case when comparing Kañaris and Quichua_EC, where some haplotypes within the A2, B2 and D1 lineages are shared. Most of the Inkawasi and Kañaris individuals share haplotypes, corresponding to lineages A2, B2 and D1. In the case of lineage C1, there is a shared haplotype between several individuals from Cañar, Quichuas from Ecuador, northern Peru (Chachapoyas and Cajamarca) and Amantani (Lake Titicaca), with the exception of Inkawasi and Kañaris. The low FST value calculated using AMOVA for the 10 studied populations (FST = 0.058, p < 0.001 (see Additional file 4: Table S4b), indicates that there is a little difference between them. The ϕst genetic distances (see Additional file 5: Table S5), visualised in a two-dimensional space such as PCoA (Fig. 8), indicate that the Cañar_EC population is different from the communities of Kañaris and Inkawasi, located at the opposite side, reflecting the correlation between the genetic background and the geographic isolation. Statistical analysis showed a decrease in maternal haplotype diversity (see Additional file 6: Table S6) in the Inkawasi (h = 86.8%), Amantani (h = 87.4%), Pastos_EC (h = 93.3%), Kañaris (h = 94.3%) and Cañar_EC populations (h = 95.3%). Likewise, there is a maternal genetic affinity between the Kañaris and Inkawasi communities (ϕst = 0.0067, p = 0.292) (Fig. 5); our inference coincides with colonial records, which reported that both groups have been intimately associated since pre-Columbian times until the present [9]. Certainly, the access through the Qhapaq Ñan, or great Inka road, that crosses the Kañaris and Inkawasi localities, would have contributed, on the one hand, to the maintenance of the genetic affinity between them, and on the other hand, to their differences, reflecting the admixture of two groups of maternal founders in both the Inkawasi and Kañaris of Lambayeque. One group would be associated with the north of Peru and the other with south Peru. However, the observation of shared mtDNA haplotypes among the populations studied indicates that considerable gene flow took place in the Tawantinsuyu over generations [12, 13].
With respect to the historical association of the Cañar region (Ecuador) with Kañaris of Lambayeque (Peru), it is likely that at the beginning of the colonial period and after the collapse of the Inka Empire, most mitmakuna (or their descendants) from Tawantinsuyu, especially from Cañar, returned to their original regions. It is known that a proportion of the Ecuadorian Cañaris, including other groups subjugated by the Inkas, were allied with the Spaniards. This situation would have offered a chance to free them from the Inkas, and also to reintegrate the mitmakuna Cañaris and return them to their old home [5]. This could be the reason why we do not detect the presence of Ecuadorian male DNA in the Kañaris community, assuming that the Inkas displaced them to Lambayeque Department. A different situation would have occurred with other Ecuadorian Cañaris, part of the Inka contingent, who would have stayed in their original locations, such as in Cajamarca, Chivay (Arequipa) Cusco and Lake Titicaca. The general scenario shows that the genetic pattern of the individuals associated with the aforementioned places reflect the migrations that took place during the Inka Empire throughout the Andes. Data from ancient DNA of the ‘Cañaris’ from Ecuador as well as from Peru will undoubtedly be the key to complement their history.