The arid Arabian Peninsula can be viewed as a geographic cul de sac and passive recipient of Near East cultural and demic expansions since the Bronze Age. However, southern Arabian late Neolithic excavations revealing sorghum and date palm cultivation attest to earlier influences from East Africa and South western Asia 1. However, after the southern dispersal route of modern humans across the Bab el Mandeb Strait was proposed 2 and further developed 3, multidisciplinary interest on this region has dramatically increased in the search for traces of such putative early southern dispersals across the Arabian threshold. From an archaeological perspective, Lower Paleolithic Oldowan industries found in the Arabian Peninsula indicated that, presumably, H. erectus or H. ergaster may have been making forays into this region 4. There is also evidence of Middle Paleolithic Mousterian and Aterian technologies in Arabia suggesting the possibility of an expanded southern border for H. neanderthalensis and potential links with populations from Northern Africa 5. More recent field work has sampled new Palaeolithic sites in Oman, under covering lithic assemblages with typological affinities to industries in the Levant, India and the Horn of Africa, suggesting there were a series of hunter-gathers range expansions into southern Arabia from all three refugia over the last quarter of a million years 6. However, there are few reliable age estimates for these industries. Furthermore, the passage of these hominids to Arabia could be either through an overland route or across a waterway.

From a population genetics perspective, the most recent studies on the Arabian Peninsula have been carried out using mainly uniparental, mitochondrial DNA (mtDNA) and Y-chromosome, markers. Analysis based on maternal lineages in the region has been interpreted to represent traces of late Palaeolithic, Neolithic and more recent inputs from nearby regions into Arabia as well as signatures of authochtonous expansions 789101112. However, the lack of deep rooting M and N sequences in the contemporary Arabian mtDNA pool leaves the proposed southern coastal route without empirical support. Although Y-chromosome basal lineages remain undetected, phylogeographic patterns indicate that the Levant was an important bidirectional corridor of human migrations 1314. Moreover, the Levant appeared as the main source of male lineages to the Arabian Peninsula 15. However, Saudi Arabia, a country that occupies about 80% of the Arabian Peninsula, was not directly included. In order to fill this void, we performed a high resolution Y-chromosome SNP analysis of 157 Saudi Arabian males and a STR-based analysis of J1-M267, the most frequent Y-chromosome haplogroup in Saudi Arabia. Comparisons to other published Arabian Peninsular populations and nearby regions were made to further explore paternal traces of the modern human transit across Arabia.

Consistent with the rest of the Arabian Peninsula, haplogroup J is the most abundant component in Saudi Arabia embracing 58% of its Y-chromosomes. Its two main subclades (J1-M267 and J2-M172), show opposite latitudinal gradients in the Middle East. J1-M267 is more abundant in the southern areas, reaching a frequency around 73% in Yemen, whereas J2-M172 is more common in the Levant. Most probably, the significant higher presence of J2-M172 in Saudi compared to other Arabian populations is due to the larger northern boundary that Saudi Arabia shares with the Levant. The Fertile Crescent region has been considered the most probable origin of the earliest dispersions of both subclades 213233. Further subdivisions of J1-M172 have uncovered more recent Bronze age expansions from Turkey and the Balkans traced by the J2-M67/M92 and J2-M12 subgroups 213435.

In a similar way, it is possible that J2-M47 signals a more recent expansion from the Levant that also affected the Arabian Peninsula. The peculiar distribution of J2-M67 in Arabia could be explained assuming maritime contacts from classical Mediterranean cultures. The presence in Saudi Arabia of three males harbouring underived J1-M304 chromosomes is intriguing. It could be that they came together with the J1-M267 or J2-M172 expansive waves, or they could represent the remnants of an old and geographically widespread Palaeolithic substrate. This type of underived chromosomes has been detected rarely in Turkey 21, in Oman and in the eastern Mediterranean area 34. However, as the critical Levantine region has not yet been adequately dissected for J1, it seems premature to favor any of these hypotheses. The geographic pattern and most probable origin of the Y-chromosome haplogroup J in Arabia faithfully mirrors those found for the most prevalent J and R0a mtDNA haplogroups in the same region 78912. In addition, J1-M267 divergence age calculated for Saudi Arabia (11.6 ± 2 kya) and Yemen (11.3 ± 2 kya) are also very coincidental with those calculated for J1b (11.1 ± 8.4 kya) and R0a1 (9.6 ± 2.9 kya) in Saudi Arabia 78. It is worth mentioning that J1-M267 ages in Saudi Arabia and Yemen are significantly older than those obtained for UAE and Qatar (Table 3)15 and for Oman 14 pointing to a terrestrial more than to a maritime colonization. It has been suggested that Yemen could be a center of expansion for mtDNA haplogroup R0a 9.

Although the comparatively high divergence calculated for J1-M267 Y-chromosomes in Yemen 15 could be in support of primary or, at least, secondary human expansions from southern Arabia, it could also be explained as result of successive arrivals of J1 chromosomes from different source regions. Furthermore, J 1-M267 diversities found here for Saudi Arabia are of the same range than in Yemen and both smaller than the one estimated for Turkey 21. This southwards decreasing trend is more compatible with a Neolithic arrival to Arabia via the Levant as proposed by others 15. In fact, the coalescence ages calculated in Saudi Arabia for their most prominent mtDNA and Y-chromosome lineages merged around 10 kya. This period is coincidental with an improvement of the climatic conditions in the area that facilitated the spread of Natufian and Neolithic cultures from the Fertile Crescent north and southwards. The haplogroup E1-M123/M34 has an extended but sparse geographic distribution in Eastern Africa, the Middle East and the Mediterranean basin 143235. However, its frequency rises considerably in some populations, most probable because of isolation and genetic drift effects. This would explain frequencies as high as 11% found in some Ethiopian samples 35 or the highest (31%) found, until now, in the Dead Sea region of Jordan (Flores et al. 2005). In the Arabian Peninsula this clade is, in general, well represented but reaches significant higher frequencies in the southern countries of Yemen and Oman compared with northern areas. An East African origin, and posterior spread to the Near East through the Levantine corridor, of this lineage was proposed based on the much larger variance of this clade in Egypt (0.5) versus Oman (0.14) 14. On the contrary, other authors have suggested that E1-M123 may have originated in the Near East because of its generalized implantation there 21 compared to its presence in Eastern Africa, mainly just localized in Ethiopia 32. Recent E-M123 haplogroup variances calculated for Yemen (0.14) and UAE (0.25) were also lower than the Egyptian one 15. In addition, haplotypic differences found for those two Arabian countries indicated that they do not share a common ancestry 15. So, from an Arabian Peninsula perspective, E1-M123 could have come from Ethiopia, across the Horn of Africa, or from the Levant, or even from both sources, forming independent isolates. Global male inputs from Sub-Saharan Africa and Asia across Iran, not the Levant, into the Arabian Peninsula have been estimated in this study, as 13.4% and 16.6% from both source areas respectively. Recent mtDNA studies on the same Arabian Peninsula countries 78912 have confirmed a notable female-driven sub-Saharan African input with a mean value around 15% for all the Peninsula, although frequencies as high as 60% have been detected in Hadramawt populations of Yemen 9. Curiously, the Iranian female flow (18%) was also rather similar to that calculated for Africa. Although a slight ratio excess of Sub-Saharan African female versus male gene flow is detected (1.12) we do not found the strong sexual bias proposed by other authors for Arabian populations and attributed to the peculiarities of the recent slave-trade 1236. Without dismissing the role mediated by slavery, the geographical distribution of these sub-Saharan African lineages in the Arabian Peninsula seems to indicate a prehistoric entrance of a noticeable portion of these lineages that participated in the building of the primitive Arabian population 89. The presence of two underived E-M96 Saudi lineages raises interesting questions related to the macrohaplogroup DE-YAP phylogeography. The recent resolutions of the CDEF-M168 tripartite structure to the bipartite DE-YAP and CF-P143 1631 extends the conversation regarding the early successful colonization of Eurasia. While several scenarios remain potentially possible the most parsimonious model is the most prudent. This model proposes the successful colonization of Eurasia by migration(s) of populations containing precursor Y-chromosome founder macrohaplogroup CDET-M168 and basal mtDNA L3 representatives. Regions near but external to northeast Africa, like the Levant or the southern Arabian Peninsula could have served as an incubator for the early diversification of non-African uniparental haplogroup varieties like Y chromosome DE-YAP*, CF-P143* and mtDNA M and N molecular ancestors. These would have spread globally and diversified over time and space. This model would imply that both CF-P143 and the DE-YAP evolved nearby but outside Africa. One DE-YAP* ancestor would have spread to Asia and evolved to haplogroup D while another DE-YAP* returned to northeast Africa and evolved into hg E. It is noteworthy that DE-YAP* has been detected at low frequency in Africa 37. Again, this hypothesis has its mtDNA counterpart as it is well documented that, in the Palaeolithic, at least three clades (X1, U6, M1) derived respectively from the three main Eurasian macrohaplogroups (N, R, M) came back to North Africa from Asia 3839404142.

Within this frame, it should be expected that E-M96* types appear in Africa although its presence in the Arabian Peninsula instead Eastern Africa would not compromise the last proposed model. It could be suggested that these E-M96 Saudi lineages have a sub-Saharan Africa ancestry. However, at least for one of them, all their known male ancestors belong to a big Shammar Arab tribe that ruled much of central and northern Arabia from Riyadh to the frontiers of Syria and northern Iraq. In addition, it might be present in Lebanon 18. However, as the authors did not type more markers derived within the E-M96 background such as P147, P177, P2, P75 or M329, more comprehensive phylogenetic resolution of YAP derived Y-chromosomes in the Middle East and North and East Africa are necessary to explore the topic further. In any case, the presence of E-M96* in Saudi Arabia should not be taken as support of the southern exit of modern humans across the Srait of Bab el Mandab as no D nor CF underived lineages have been yet found in this area.

The Y-chromosome genetic structure of the Arabian Peninsula seems to be mainly modulated by geography. The data confirm that this area has mainly been a recipient of gene flow from its African and Asian surrounding areas, probably mainly since the last Glacial maximum onwards. Although rare deep rooting lineages for Y chromosome haplogroups E and J have been detected, the presence of more basal clades supportive of the southern exit route of modern humans to Eurasian, were not found.

KKA and AH was in charge of carrying out the sequences, collection of samples, haplogrouping and writing part of the manuscript. AMG, JML, VMC and PAU were in charge of design of the experiments, analysis of the data and writing manuscript. All authors read and approved the final version of the manuscript.