Heterogeneity in Palaeolithic Population Continuity and Neolithic Expansion in North Africa
Heterogeneity in Palaeolithic Population Continuity and Neolithic Expansion in North Africa
Gerard Serra-Vidal
Marcel Lucas-Sanchez
Karima Fadhlaoui-Zid
Asmahan Bekada
Pierre Zalloua
David Comas 6
Show footnotes
Published:October 31, 2019DOI:https://doi.org/10.1016/j.cub.2019.09.050
Highlights
•
Paleolithic genetic continuity is found in extant human North African genomes
•
There is a West-to-East genetic cline of the Paleolithic component in North Africa
•
Neolithization had a larger demographic impact than Arabization
•
Differential admixture and genetic drift have modeled North African genomes
Summary
North Africa is located at the crossroads of the Mediterranean Sea, the Middle East, and the Sahara Desert. Extensive migrations and gene flow in the region have shaped many different cultures and ancestral genetic components through time [1, 2, 3, 4, 5, 6]. DNA data from ancient Moroccan sites [7, 8] has recently shed some light to the population continuity-versus-replacement debate, i.e., the question of whether current North African populations descend from Palaeolithic groups or, on the contrary, subsequent migrations swept away all pre-existing genetic signal in the region. In the present study, we analyze 21 complete North African genomes and compare them with extant and ancient genome data in order to address the demographic continuity-versus-replacement debate, to assess whether these demographic events were homogeneous (including Berber and Arabic-speaking groups), and to explore the effect of Neolithization and posterior migration waves. The North African genetic pool is defined as a melting pot of genetic components, including an endemic North African Epipalaeolithic component at low frequency that forms a declining gradient from Western to Eastern North Africa. This scenario is consistent with Neolithization having shaped most of the current genetic variation in the region when compared to posterior back-to-North-Africa migration waves such as the Arabization. A common and distinct genetic history of the region is shown, with internal different proportions of genetic components owing to differential admixture with surrounding groups as well as to genetic drift due to isolation and endogamy in certain populations.
Results and Discussion
The initial dataset was merged with other available datasets (see STAR Methods), providing ten current and four ancient North African groups (Figure S1). A first exploration of the data was performed using principal component analysis (PCA). The first component (PC1, accounting for 3.5% of the variation) captures the genetic differentiation between sub-Saharan Africans and non-African populations, with ancient and current North Africans placed in an intermediate position (Figure 1A). PC2 (0.7% of the variation) splits Middle Easterns and Europeans, with North Africans closer to the former. Regarding ancient North Africans, while Canary Guanches (∼5th century BCE) cluster with current North Africans (in agreement with their putative Berber origin [9]), Moroccan Epipalaeolithic samples from Taforalt cluster independently, while Moroccan Early (IAM) and Late Neolithic (KEB) have intermediate positions in the PC.
https://els-jbs-prod-cdn.literatumon...f27af6/gr1.jpg
Even though a sudden genetic change comparing early and late Neolithic samples in Morocco has been recently proposed [7], our data show that there are still traces of the Epipalaeolithic ancestors in the genomes of extant North Africans, as shown by the admixture components and the amount of shared haplotypes between ancient and present North African genomes. Our results confirm that gene flow in the area, coming for surrounding regions such as Europe, the Middle East or sub-Saharan Africa did not completely erase the ancient background of autochthonous North Africans in the last 15,000 years. This Palaeolithic autochthonous genetic component might correlate with the Maghrebi component as a result of a back-to-Africa gene flow defined by [6], and its presence is not distributed following a uniform pattern in the area.
Genetic Heterogeneity within North Africa
Saharawi and Berber groups show the highest outgroup-f3 values for the Epipalaeolithic Taforalt component (0.218–0.222). The ancient samples IAM (0.322), KEB (0.235), and Guanche (0.227) show a higher North African Epipalaeolithic component than current populations (in agreement with the ADMIXTURE analysis), whose decrease is compatible with a dilution of the Taforalt component through time. This component is significantly more frequent in Western (Saharawi, Moroccan, Algerian) and Berber-speaking individuals (Saharawi, Mozabites, Moroccan, and Moroccan and Tunisian Berbers) (Mann-Whitney U test, p value = 4.52e−15), suggesting a continuity of this autochthonous North African component in Berber-speaking groups. Although no perfect correlation between culture (i.e., Arabic- and Berber-speaking groups) and genetics can be claimed ([1]), the consideration of the Berber-speaking groups as the autochthonous peoples of North Africa [10] is reinforced by the present results.
Zenata, Mozabite, Saharawi, and Moroccan groups show the highest proportions of sub-Saharan ancestry, which could be attributed to the slave trade routes carried out during Roman and Arab presence mainly in northwest Africa [11, 12]. The Taforalt and Moroccan Early Neolithic have a higher sub-Saharan affinity than most current North Africans (as stated by [8]), whereas the Moroccan Late Neolithic and the Guanches have a similar level of sub-Saharan affinity to most current groups analyzed in the present study.
Egyptian and Libyan show the highest proportion of the Caucasus-Iran component, in agreement with their geographical proximity to southwest Asia. A high f3 value is also estimated for KEB (0.250), in contrast to the lower estimated values found for Taforalt and IAM (0.206, 0.216), suggesting that this component might have entered North Africa during the late Neolithic coming from Iran and may have been posteriorly diluted in western North Africa.
Admixture Models in North Africa
To determine the main Eurasian ancestry sources in North Africa during the back-to-Africa events and test whether these sources are homogeneous across North African groups, admixture-f3 tests of the form f3(OOA, SS; NA) were performed (OOA, out of Africa population; SS, Sub-Saharan population; NA, North African population). The sub-Saharan source was fixed to Yoruba (no significantly different results were found when choosing an East African group, such as Dinka, instead; data not shown). For all groups (with the exception of Tunisian Berbers, for which admixture was not detected through f3, see above), the most significant results were found for Sardinians, Basques, and North Italians (data not shown). When the analyses were repeated adding ancient Eurasian populations [13] to the dataset, the higher significant values were found with the Neolithic farmers from Europe and Anatolia and, again, Sardinian and Basques, which is consistent with their high frequency of Neolithic component [13]. These results point to the main role of the Neolithization process when shaping the current North African genetic landscape, thus supporting the PCA, ADMIXTURE, and internal f3 tests, where Eurasian gene flow after Neolithization (such as the Arabization process starting in the 7th century CE) seem to have had a lower impact. The arrival of the Neolithic to North Africa as a demic diffusion process from the Middle East with putative interactions with local groups is widely accepted [14, 15], although some hypotheses and recent genetic data also point to direct contacts with Iberia [7, 16]. Despite recent approximate Bayesian computational based analyses of genetic data suggesting that the synchronous Neolithic expansion from the Middle East through both Mediterranean shores (i.e., Europe and North Africa) had a similar demographic pattern [17], our results point to a larger demographic replacement in North African than in European populations. The genomes of current Europeans carry a larger amount of hunter-gatherer components (up to 50% in Northern Europeans, according to [18]) compared to those of North Africans, where the Palaeolithic component, although present in extant populations, it is found at much lower frequencies (from 18.1% in Western Sahara to 5.2% in Egypt, according to the crude estimates of ADMIXTURE). This might suggest that the Neolithic demographic imprint was lower in Europe than in North Africa, where fewer local hunter-gatherers were assimilated by Neolithic farmers [13].
Complex demographic scenarios were tested with qpGraph [19] on a model involving different ancestral components: sub-Saharan (Yoruba), European (Basque), Middle Eastern (BedouinB), and North African Epipalaeolithic (Taforalt) (Figure 3). All North African groups fit into the tested model, with varying proportions of the four ancestral components, yielding compatible results with the admixture-f3 tests (Figure 2A). Admixture signals in contemporary North African populations were dated using MALDER. All population triplets of the form (Ref A, Ref B, Target) were tested, taking the same populations used in the qpGraph analysis (i.e. Yoruba, Basque, BedouinB, and Taforalt) as references. Tunisian Berbers were chosen instead of Taforalt when the latter lacked power to give significant results due to their high missing data, since Tunisian Berbers have a high North African component, according to previous studies [1]. Interestingly, only tests involving the Yoruba population passed all the pre-test steps and yielded statistically significant results. Dates for these admixture times are shown in Table S2 for Saharawi and Egyptian, which account for both geographic and genetic diversity extremes in North Africa. These results (ranging from 1329 to 1643 AD) are compatible with previous estimates of the sub-Saharan introduction in North African populations in recent historical times ([6]).
https://els-jbs-prod-cdn.literatumon...a9c044/gr3.jpg
Conclusions
The intricate genome landscape in North Africa is shaped by two factors. The first one is an amalgam of genetic components resulting of extensive gene flow coming from different geographical (sub-Saharan Africa, Europe, Middle East, Caucasus, and North Africa itself) and temporal sources (Palaeolithic migrations, Neolithization, Arabization, and recent migrations). The second factor is the result of internal admixture and genetic drift, which have produced an ample genetic heterogeneity within the region. These two factors of complexity should be considered in biomedical studies in which North African samples are included in order to avoid genetic biases and artifacts.
https://www.cell.com/current-biology...showall%3Dtrue
