Genomic Data Reveal a Complex Making of Humans

Figure 1. Phylogenetic tree of the Y-chromosome haplotypes. Haplogroup defining mutations assayed in this study are shown along branches. Population codes are as follows: Madeira-MA, Açores-AC, North Portugal-NP, Centre Portugal-CP, and South Portugal-SP.

For centuries the Iberian Peninsula was a melting pot for various populations of different origins, including Celts, Germanics or Scandinavians (Swabians and Visigoths), Phoenicians, contributions from a myriad of east Mediterranean peoples, and finally Moors and Arabs (Amaral & Amaral, 1997; Oliveira Martins, 1994). Apart from these populations, there is also a well-documented input of sub-Saharan slaves into Portugal during the 15th–16th centuries. From all these peoples, three groups have been the focus for vigorous debate over their contribution to the present day Portuguese genetic gene pool: the Sephardim Jews, the Arab-Berbers and the sub-Saharans. Jews are known to have inhabited the Iberian Peninsula since at least the 3rd century AD (Tavares, 2000). Their arrival was probably the result of migrations via North Africa, where Jewish communities were already well established. During the Visigoth and Muslim periods Jews reinforced their position and flourished in the peninsular society. Expelled from Spain and Portugal in the 15th century, the Sephardim Jews (from “Sepharad” meaning Spain or “Espaniah”) went back to North Africa or migrated, mainly to Amsterdam and Constantinople (Kayserling, 1971). However, many are supposed to have remained in Portugal under forced conversion to Catholicism, becoming known as “new-Christians” (Borges, 1998; Guerra, 2003).

Moors, or Berbers from Mauritania, started to expand into Iberia in 711 AD. However, it is generally assumed that their presence in the north of Portugal was minimal and short-lived. On the contrary, it is known that Arabs from Egypt and Yemen settled down in villages of southern Portugal (Lopes, 1928). Arab rule in Iberia ended in 1492 with the fall of Granada. Although some sub-Saharans entered the Iberian Peninsula with the Berbers and Arabs, most of this component in Portugal dates back to the 15th century when the importation of sub-Saharan slaves accelerated, following the establishment of a commercial Atlantic network. At the height of the slave trade, 10% of the population in the south of the country, and particularly in Lisbon, was composed of sub-Saharan slaves (Godinho, 1965).



The sub-Saharan component
Typically sub-Saharan lineages are represented by haplogroups A and B, which are also the basal clades in the human Y-chromosomal tree of humans. E1-M33 has a rather limited distribution in the West sub-Saharan region, reaching its highest frequency in Mali (34%, Underhill et al. 2000) and north Cameroon (especially among the Fulbe and Tali groups, Cruciani et al. 2002). This haplogroup is virtually absent in European populations (Semino et al. 2004) and Northwest Africa, although it is found among Berbers (1.6–3.2%, Bosch et al. 2001). Haplogroup E3a is especially common in Sub-Saharan West Africa (Underhill et al. 2000, 2001; Semino et al. 2002), has been associated with the dispersal of Bantu people (Underhill et al. 2001, 2001) and reaches ∼80% in Senegal (Semino et al. 2002). The combined frequency of all these haplogroups in Portugal is marginally low (∼0.7%), so it is highly unlikely it is a result of accidental sampling of African immigrants. Haplogroup E3a is notably absent from our sample except for a single E3a individual in the Açores, although this haplogroup constitutes the majority of Y-chromosomes in Guinea (>70%, Rosa et al. unpublished data) and Cabo Verde (16%, Gonçalves et al. 2003), the putative regions of origin for the first slaves brought to Portugal. These results sharply contrast with those obtained with mtDNA markers. mtDNA haplogroups L0–L3 and M1 that are characteristic to sub-Saharan populations are present at ∼12% and ∼14.8% in the south of Portugal and Madeira, respectively (Brehm et al. 2003; Gonzalez et al. 2003). This contrasting pattern of paternally and maternally inherited markers closely follows the situation in the Canary Islands where E3a is residual (0.9%, Flores et al. 2003). These Y-chromosome haplogroups have escaped detection in other populations of Iberia (Semino et al. 2000, 2004; Bosch et al. 2001), in spite of the fact that the peninsula was a recipient for sub-Saharan slaves from the 15th century onwards. These results are consistent with sex specific gene flow, probably resulting from the custom that male slaves did not mate with Iberian women while the opposite situation was common, as supported by mtDNA analysis (Brehm et al. 2003; Gonzalez et al. 2003). Three lineages belonging to haplogroup A (Y(×M94)) were found. The presence in Portugal of both the A and E1 haplogroups may be independent from the slave trade (otherwise E3a would be well represented since it comprises the majority of West Africa lineages). These findings either suggest a pre-Neolithic migration from North Africa or a more recent origin from a founder population of small size that did not carry haplogroup E3a, which is a major component in North African populations today. TMRCA for Portuguese E1 lineages estimated as 22.9 ± 7.2 ky (Table 2) favours the first scenario, a possible parallel to mtDNA U6 cited in Gonzalez et al. (2003).


Figure 2. Principal Component Analysis based on Y-chromosome haplogroups frequencies. Axis 1 and 2 extracted 37.3% and 23% of the total variation (a third axis accounts for 17%, not shown). The MST superimposed on the ordination of populations is also shown and reveals the most likely connections between populations. Populations are as follows: MA-Madeira, AC-Açores, SP-South Portugal, CP-Centre Portugal, NP-North Portugal, ME-Middle East, BI-British Isl., GD-Germany/Denmark, NO-Norway, SJ-Sephardim Jews, EG-Egyptians and NA-North Africa.

Haplogroup E3b
Haplogroup E3b (characterized by mutation M35) is widespread in Northwest Africa, East Africa, the Middle-East (Bosch et al. 2000; Semino et al. 2002; Underhill et al. 2000; Cruciani et al. 2004) and is also common in Europe, albeit at variable frequencies (Semino et al. 2000). Eastern Africa is seen as the homeland for E3b as there it has the highest number of different clades and microsatellite diversity and the almost exclusive presence of E3b*, with estimates of ∼30 ky for the age of the M35 mutation (Bosch et al. 2001; Cruciani et al. 2004). The frequency of E3b* varies from ∼3–8% in Morocco (Bosch et al. 2001; Cruciani et al. 2002; Semino et al. 2004) and reaches its highest frequency in Ethiopian Oromo (19.2%, Cruciani et al. 2002; Semino et al. 2002). The clustering of E3b* lineages in North Portugal (with 3.8 ± 2.2 kya) and low Y-STR variation (compared to North Africa) is a putative indication of a restricted but definite gene flow across the Strait of Gibraltar.

Haplogroup E3b1 has a significantly uneven distribution in mainland Portugal, occurring at 8–6% in Northern and Central Portugal versus only 1% in the South (x2, P < 0.01). Regional sub-clustering of the STR haplotypes on the background of Portuguese E3b1-M78 suggests that these lineages could have spread from the Balkans all over Europe during the Neolithic (E-M78α in Cruciani et al. 2004). This is furthermore supported by the age of STR variation (8.8 ± 4.1 ky, Fig. 2) and fits into the 6.3–9.2 ky range suggested by Cruciani et al. (2004). The pairwise divergence times between Madeira, North and Central of Portugal differ insignificantly from each other, being on average around 6 thousand years (Table 2). The timescale for the arrival of E3b1-M78 is coincident with that pointed out by Flores et al. (2003). Defined on the basis of GATA A7.1 allele 9, E-M78α follows the same pattern of north-central prevalence (3% vs 1% in the south). An increased frequency of 4.6% was detected in Madeira. This may correlate with the contribution of Italians to the archipelago, as this cluster is present in ∼7% in their homeland (Cruciani et al. 2004). Based on the same resolution level we detected one individual in the North of Portugal having the GATA A7.1 indicated by Cruciani et al. (2004) as the North African E3b1ß. All the remaining E3b1 cannot be associated to any proposed geographical origin on the basis of the additional STRs.



Population structure and Principal Component Analysis
An AMOVA between populations pooled by geographic regions (mainland Portugal, the two Archipelagos of Madeira and Açores, and data from North Africa (Bosch et al. 2001)) attributes ∼33% of variance to differences among groups, but only 1% to differences among populations within each group (overall FST value of 0.335, P < .0001). Population pairwise differences were all statistically significant (P < 0.0001). Excluding North Africans, the variance attributed to differences among insular and mainland populations drops to only ∼2%. This amount of variance is mainly due to the lack of R1b in North Africa. These results are also evident in the PCA (Figure 2). The Portuguese populations reveal themselves as a highly heterogeneous group, clearly separated from other European populations (GD, NO and BI, see Figure 2 for population codes) but also from North Africa and Mid-Eastern populations obviously denoting an intermediate status. Interestingly, both North Africans and Middle-Easterners join the Portuguese populations through the “Sephardim” Jews as indicated by the MST. The most likely lineage groups responsible for differences on the axis distinguishing European-Iberian-MidEast populations are E3b (a characteristic of NA) and R1b (typical of West Europeans). The latter haplogroup is almost absent in North Africa and Middle East.

Overall, the Portuguese show an affinity to Middle Eastern populations. So far, the degree of genetic contribution from peoples of Jewish, North African and sub-Saharan ancestry to the present day Portuguese has been uncertain. Our data suggests only minor influences from sub-Saharan males in Portugal and the Atlantic Islands, despite historical records indicating that slaves constituted at least 10% of the total population in Madeira and the South of the country in the 15th century. This contrasts mtDNA and HLA data, but provides genetic support to the view that mixing was highly asymmetrical by sex. The North African component at least for mtDNA, is mainly concentrated in the North of Portugal. The mtDNA and Y data indicate that the Berber presence in that region dates prior to the Moorish expansion in 711 AD. Our Y chromosome results are also consistent with a continuous and regular assimilation of Berbers in North of Portugal. This argues against previous interpretations of Moorish mediated contributions, based on Y chromosome data (Bosch et al. 2001; Pereira et al. 2000b; Cruciani et al. 2004) and provides an alternative view of an earlier Berber presence in the North of Portugal. Haplogroup J* distribution in Portugal runs opposite to haplogroup E3b2, corroborating the hypothesis that J1 lineages were not introduced with North African E3b2 lineages during the Arab/Berber expansion. In addition, the J1 lineages are distinct from the common Arab haplotype, consistent with an independent source, possibly Sephardim and/or other Near East peoples. Until now, the only evidence supporting the presence of Berbers in Iberia was the high frequencies of haplogroup U6 in Northern Portugal. Our data indicate that male Berbers, unlike sub-Saharan immigrants, constituted a long-lasting and continuous community in the country.

DOI: 10.1111/j.1529-8817.2005.00161. onlinelibrary.wiley.com/doi/10.1111/j.1529-8817.2005.00161.x/full