The Canaanites walk among us: ancient DNA edition

Ancient DNA from here to there:

Ancient DNA has illuminated many things, but there is a logic as to what topics and questions it tackles. The focus on northern Eurasia is clearly a function of the probability of preservation, though techniques of extraction are getting better and better. I can’t imagine how we’d ever get a sample out of a moist tropical environment, but I won’t be surprised if something is obtained from a cave in southern Africa or high in the Tibesti in the near future.

But another parameter is time since the demographic events in question. Too ancient, and the probability of success is too low(ok, time is a parameter in much of science!). It seems plausible that in idealized circumstances we’re going to push beyond the one million year barrier. And yet too recent is also a problem (or not a problem!). For humans and even non-humans we have lots of corroboration about questions we might ask about the recent past. You could use “ancient DNA” to trace the migration of Mormons across the Intermontane West, but why would you?

So you see the earliest ancient DNA work on humans was biased toward testing models about gene flow and ancestry tens of thousands of years in the past, between modern humans and archaic lineages. Obviously we don’t have oral history or written texts from this period, and archaeology will only get us so far.

More recently the time depth has been getting shallower and shallower. Both David Reich and Eske Willerslev’s work on European prehistory is liminally historical. By this, I mean that what is prehistory in Europe is a historical period in the Near East. We may not have written records from the Corded Ware or Bell Beaker cultures, but we do have plenty of them from contemporaneous Near Eastern groups.

The Cauldron of Peoples:

There are still questions to be asked about European prehistory, but the gaps are getting narrower and narrower. Scholars are finally devoting resources to other regions of the world. Last year Iosif Lazaridis’ The genetic structure of the world’s first farmers finally opened up the box that was the prehistory of the Near East. This was important, because much of prehistory and history began in the Near East. Farmers from this region seem to have moved into Europe, South Asia, Central Eurasia, and Africa. To understand the population histories of these areas one needs to understand the population history of the Near East.

What Lazaridis et al. found this that there were at least two major groups of very genetically distinct Near Eastern farmers at the dawn of agriculture. Once group faced the eastern Mediterranean, while the other seems to have flourished on the slopes of the Zagros. Western and eastern farmers respectively. It is important to note that these two groups were very genetically distinct. If we sampled these two groups of farmers, who faced each other across northern Mesopotamia, in any modern population survey we’d assume that the genetic distance meant that they were sampled from different continents or very distant regions of Eurasia.

This finding suggest that the clinal patterns of variation in much of today’s world may be a consequence of massive population admixture between groups which had heretofore exhibited deep population structure. Why such deep structure existed and persisted is an interesting question, but at this point it is important to note descriptively that the past 10,000 years have seen a massive reduction of this structure due to gene flow between populations.

In the Near East Lazaridis et al. found that there was significant reciprocal gene flow between the western and eastern regions of the Near East after the emergence of farming, down to the historical period. This is one reason that estimates of “farmer” ancestry in modern Europeans always gave very low estimates: the reference populations no longer existed in unmixed form in the Near East. The peoples who brought agriculture to Southern Europe were related exclusively to the western farmers of the Near East, a population which no longer exists in unmixed form in that region of the world (ergo, among modern groups Sardinians are the closest proxies we have).

The Age of Bronze:

But there is much that occurred after prehistory in the Near East. We know this because we have extensive records going back 4,500 years, and even earlier. And though put into written form in the first millennium before Christ, the Hebrew Bible also records the deeds and names of people who have come and gone well before the Classical Age.

A new preprint on biorxiv sheds some light on a critical transitory period, Continuity and admixture in the last five millennia of Levantine history from ancient Canaanite and present-day Lebanese genome sequences:

The Canaanites inhabited the Levant region during the Bronze Age and established a culture which became influential in the Near East and beyond. However, the Canaanites, unlike most other ancient Near Easterners of this period, left few surviving textual records and thus their origin and relationship to ancient and present-day populations remain unclear. In this study, we sequenced five whole-genomes from ~3,700-year-old individuals from the city of Sidon, a major Canaanite city-state on the Eastern Mediterranean coast. We also sequenced the genomes of 99 individuals from present-day Lebanon to catalogue modern Levantine genetic diversity. We find that a Bronze Age Canaanite-related ancestry was widespread in the region, shared among urban populations inhabiting the coast (Sidon) and inland populations (Jordan) who likely lived in farming societies or were pastoral nomads. This Canaanite-related ancestry derived from mixture between local Neolithic populations and eastern migrants genetically related to Chalcolithic Iranians. We estimate, using linkage-disequilibrium decay patterns, that admixture occurred 6,600-3,550 years ago, coinciding with massive population movements in the mid-Holocene triggered by aridification ~4,200 years ago. We show that present-day Lebanese derive most of their ancestry from a Canaanite-related population, which therefore implies substantial genetic continuity in the Levant since at least the Bronze Age. In addition, we find Eurasian ancestry in the Lebanese not present in Bronze Age or earlier Levantines. We estimate this Eurasian ancestry arrived in the Levant around 3,750-2,170 years ago during a period of successive conquests by distant populations such as the Persians and Macedonians.

The period between 1700 and 1800 BCE in the Near East saw many changes and was a sort of nexus. Sumer had fallen, the Hittites had not emerged as a superpower, while Egypt was not heavily involve in the game of kings as of yet. The system of international relationships described in had not crystallized. That was for the late Bronze Age.

But some of the pieces we were to recognize were already in place. An Amorite Babylon under Hammurabi established the contours of the culture and polity we’d recognize down to the Persian conquest. In Egypt the Middle Kingdom was going into decline, and the Hyksos interregnum would give rise to the New Kingdom, which would become a major player in the Levant (and probably is the model for much of the Egypt we see described in the Bible).

The admixture plot above reflects the five individuals from Sidon dating to about ~1750 BCE. They are about a 50:50 mix of western and eastern farmer. Though they seem to be genetically rather similar to modern Lebanese (the authors sampled Lebanese Christians in particular), there have been some changes between the Bronze Age and the modern period. In particular, a genetic component that seems to be related to the Eurasian steppe is present in modern Lebanese. Explicit admixture estimates give a range of 5-10% mixing into a ~90-95% Bronze Age ancestral background.

This seems to establish basic continuity between the Bronze Age and the modern period. Totally unsurprising. Remember that Italy exhibits deep population structure that dates back to at least 2,000 years ago, and probably earlier. It is likely that much of the same applies to the Near East. Though looking at Muslim populations one can see minor and non-trivial contributions of populations which moved in after Islam (Sub-Saharan and East Asia segments are clear signs of slavery impacting Muslims that would not apply to ethno-religious minorities), most of the ancestry broadly is deeply rooted back to antiquity.

Because of sampling issues one can’t estimate admixture between eastern and western farmers just from looking at ancient DNA transects. We don’t have the density that we have in Europe (yet). So the authors used a more classic inference technique looking at decays of linkage disequilibrium in the genome. In short you can see how many generations that a pulse admixture between two populations occurred by looking at correlations of variants across the genome. The authors arrive at the intervals above, and in particular focus on the period that seems to overlap with the rise and fall of the empire of Sargon of Akkad and correlated with a climatic disruption.

I suspect they are wrong here. First, it seems pretty clear to me that LD based admixtures assuming a pulse event have a bias toward underestimating values. There are theoretical reasons for this. So usually I pad the mid-point value across the interval on these estimates.

One thing that ancient DNA has told us is that often the less complex the society, the more demographic turnover you have. All things equal then we would expect turnover to be an older event, as simple societies are succeeded by complex ones. The succession of complex societies by other complex societies is often less disruptive for the masses because this transformation is more a matter of elite replacement.

By ~2200 BCE the Near East was already quite complex. I believe that the massive western-eastern farmer admixture occurred between 3600 and 3100 BCE, during the Uruk Expansion. The evidence of lower Mesopotamian influence and demographic settlement in places as far afield as Anatolia, the Caucasus, and Syria, are well attested from the archaeology of this period. This was was a time when a very complex and sophisticated civilization emerged almost de novo across much of the Near East. I believe that a prehistoric expansion of Sumerian civilization mediated the merging of eastern and western farmers, though some of the mixing pre-dates and post-dates the Uruk Expansion and collapse (e.g., the movement of western farmer ancestry into Mesopotamia seems certain to have occurred through the arrival of groups like the Amorites).

Additionally, buried in this preprint is evidence of major Y chromosomal turnover. We’ve seen this  before. The prominence of haplogroup J in Bronze Age and modern Levantines seems to be due to eastern farmer migration. In fact, adding haplogroup J and R together we get the inference that more than half the paternal lineages of Lebanese today are not from western farmers native to the area.

Beyond the Bronze Age:

What about the second ancestral component? Drilling down on the Y chromosomes of the Levant, R1b seems to far outnumber R1a, though the R1a clades are all of the Asian/Scythian Z-93 branch which is dominant in Central Asia and the Levant. The R1a may have come with the Persians, but in region of the western Levant for several hundred years after the period of the Bronze Age Sidon samples there was a state, the Mitanni, which clearly had an Indo-Aryan ruling class.

An Aegean influence occurred multiple times. First, at the end of the Bronze Age many of the “Sea Peoples” were clearly of Aegean origin, and so may have brought steppe-like ancestry. Second, there was the long period under Hellenistic and Roman rule, when Greek and non-Greek ethnic identity existed side by side, and movement occurred in both directions. I think only ancient DNA will answer this question, and it may be that there were multiple post-Bronze Age inputs of genes which shaped modern Levantines.

After Babel:

The curious thing that many of these studies are telling us is two-fold:

  1. Most of the population genetic structure we see around us dates to the Bronze Age, on the borderlands between history and prehistory. I think we can start to set this as a strong prior. It holds true for the Near East, Africa, South Asia, Japan and Southeast Asia. We’ll see about core East Asia, but I think probably it is true there too.
  2. Selection has continued, so that alleles for lactose tolerance and lighter skin have changed in frequency even since that period. The derived allele for SLC45A2 is found at about 2/3 frequency in modern Lebanon, but was absent in these five Sidonians. Though the sample size is small, this was somewhat surprising, and suggests that they were a swarthier people than modern Lebanese.

Addendum: I have said little here about Afro-Asiatic languages, as I don’t know enough about this topic.

20 thoughts on “The Canaanites walk among us: ancient DNA edition

  1. Can we say which of these early farming populations were Semitic vs. non-Semitic. And is the problem of Sumerian origins any closer to solution?

  2. It’ll be interesting to see how Canaanites compare to the Middle Eastern Gladiator from Roman Britain and to ancient Egyptians. Leaks give the ancient Egyptians half as much Eastern farmer ancestry as the Canaanites.

    Yemanite_Jews may the most similar population to Canaanites. They have hardly any Yemanite admixture. Jordan_EBA(Jordan genomes, 2600 BC) and the British Roman_outlier are most closely related to Yemanite_Jews.

    Modern Levanties(Arabs, Samartians, Druze) have something Jordan_EBA and British Roman_outlier don’t. The Middle Eastern side of European Jews also has this something. I’m not sure what this something is, it could be an extra layer of Eastern farmer ancestry. Anyways Yemanite_Jews don’t have this something.

  3. I made a comment a few weeks back on the Southern Italy/Sicily post which got eaten, but the question on my mind seems relevant to this post as well. What is the deal with Cypriots? I could have sworn I had read in the past that they were, genetically speaking, a Near Eastern population, not a European one. But that paper the other week seemed to indicate that Cypriots were fundamentally a European population with closer links to EEF-heavy populations around the Mediterranean than to modern day populations in Anatolia. I take it then they do not show evidence of the “eastern farmer” component in appreciable amounts?

    Also, your little note on skin tone has me thinking about ancient Egypt and the “race controversy.” If skin color was darker than today in Canaan, the same may have been true in Egypt. Perhaps Egyptians lacked substantial Sub-Saharan ancestry, but also much swarthier than the average Copt is today.

  4. You could use “ancient DNA” to trace the migration of Mormons across the Intermontane West, but why would you?

    To get an idea of how accurate your methods are?

  5. O/T & FYI:

    “Re-peopling the Past: From a Stockholm-based science journalist comes a joyous and idiosyncratic history of Homo sapiens. Karin Altenberg reviews ‘My European Family’ by Karin Bojs (Bloomsbury Sigma, 400 pages, $28).”
    May 26, 2017

    “Mercifully, recent discoveries in biotechnology and DNA research are balancing forces to all this, re-opening the door to a science-based study of our common past. After reading “My European Family,” Karin Bojs’s fascinating account of the people that lived (and died) on the European continent over the past 54,000 years …

    “The book, translated from the Swedish by Fiona Graham, begins with an exploration of Ms. Bojs’s family genealogy, and part of her mission is to trace her own genetic haplogroup. … as Ms. Bojs tests her own DNA and travels to 10 countries to interview around 70 researchers, her family history becomes the history of Homo sapiens, our common ancestor, who walked out of the African continent into what we know today as Europe and the Middle East.”

  6. “Most of the population genetic structure we see around us dates to the Bronze Age, on the borderlands between history and prehistory. I think we can start to set this as a strong prior.”

    It would be interesting to see estimations of population density around the world in the Bronze Age, to figure out how large populations have to get before they become difficult to “dislodge.”

  7. Lazaridis et al. talks a lot about the Basal European hypothesis. But the intrapopulation range of alleged BE ancestry in all sorts of populations in the region ranges from the low single digits to more than 50-60% within contemporaneous populations in the same geographic area spanning almost the entire region and something like 4000-5000 years per their supplemental materials.

    This is pretty much impossible. You should only be able to see that spread of autosomal ancestry percentages right around the moment of admixture and the first few generations afterwards. With panmixia or even a condition far less rigorous than that, for not that many generations, the range of BE ancestry should be getting far narrower. A wide range in the Natufians is one thing, but that wide range of BE ancestry shouldn’t be persisting well into the Bronze Age. The only way to avoid that is some sort of rigorous caste structure lasting several thousand years and there is no other evidence of any kind suggesting that kind of sustained caste-like division in myriad different cultures over thousands of years.

    This suggests that there is something seriously wrong the BE hypothesis advanced in that paper.

  8. The western population has to be proto-Semitic. It’s intuitively demonstrable by way of elimination. The first (para)Semitic group of linguistic record in the Zagros is the Akkadian group, and they’re an outlier – the “IndoEuropean” analogue is the Anatolian group. That means the Akkadians moved in from the west.

    I am however willing to entertain theories that the Semitic group formed out of an Afro-Asiatic population, like the Berbers, subject to extreme introgression from a now-deceased eastern language family.

  9. Egyptian art by Ramesses III’s reign depicted their neighbors in the Maghreb and Levant as overall significantly lighter (ranging from yellow to beige) while Egyptians themselves ranged from brown to yellow (especially among females). It’s apparent that not only were their neighbors were sterotyped as noticeably lighter than them but that there was a population of lighter skinned citizenry in the kingdom.

    As for Copts, they aren’t just light skinned. Here’s a page with samples from Minya:

  10. Without actual 100% Basal Eurasian genomes from both the East and the West groups (and perhaps there were others), it is very hard to be positive about the percentages in their descendants.

  11. FWIW, I’m puzzled and suspect it may be some sort of methodological quirk, or is actually capturing variation in Neanderthal admixture which is low enough percentage to vary and getting amplified somehow.

  12. Re: the Steppe_EMBA pulse and modeling for modern day Lebanese and “Iron Age Levantine” relative to Bronze Age Sidon, this raises some clear questions.

    If they model, Sidon_BA as Iran_Chl plus Levant_N, then model Iron_Age_Levantine as Sidon_BA plus Steppe_EMBA, then that basically means only a Persian intermediary would be possible.

    Now if the admixture came through any Armenian or Greek intermediary, it would have to pick up mainly Anatolia_Neolithic and Caucasus ancestry….

    But on the other hand, if admixture came from Persia, it would be expected to disrupt the Iran_Chal:Levant_N ancestry ratios, so there is a question there. How could the scenario they describe actually happen?

    A pulse of steppe related admixture itself is not necessarily so doubtful, and is likely just on the grounds of isolation-by-distance alone. Comparing Armenia_MLBA to modern day Armenians, the modern day Armenians are closer to the Neolithic Levant, thus Levant_N related admixture in Armenians. But a pulse which did not alter the ratios of Iran:Caucasus:Anatolia:Levant farmer ancestry seems like an eyebrow raiser, as it implies that this migration did not go through intermediaries at all.

    And it’s very unlikely any ancestry to the Levant came direct from the steppe without picking any ancestry from up from other farming populations first.

    On another note, some issues re: measuring the closest population to Sidon_BA using f4(Lebanese_Christian, Test;Sidon_BA, Chimpanzee) stats as in their supplement. These imply that the furthest European population in their panel (Croatian) is closer to Sidon_BA than their closest Middle Eastern population (Assyrians), and all at least slightly closer than Lebanese are.

    Fst statistics for the similar Jordan_EBA panel previously published in Lazaridis 2016 paper ( do not have this issue. The top 5 closest populations by Fst to Jordan_EBA in descending order are Lebanese, Lebanese_Christian, Lebanese_Muslim, Saudi, Syrian. Unlike the f4 statistic, this actually makes sense.

  13. I was not implying that Copts are universally light skinned. Just that some people, as counterreaction against Afrocentrist nonsense about Egypt, tend to conclude that the modern phenotype of Copts (or say the Fauyum mummy portraits) are a reasonable approximation of what ancient Egyptians looked like – particularly because contemporary Egyptian SSA ancestry seems to have mostly come from the Islamic slave trade. But if skin color was notably swarthy still in Bronze age Canaan, it may well be that ancient Egyptians were fairly dark brown even if they lacked any SSA ancestry to speak of.

  14. Razib: I believe that a prehistoric expansion of Sumerian civilization mediated the merging of eastern and western farmers, though some of the mixing pre-dates and post-dates the Uruk Expansion and collapse (e.g., the movement of western farmer ancestry into Mesopotamia seems certain to have occurred through the arrival of groups like the Amorites).

    Check out – – from a webcast last week, with David Reich.

    He talks about an early Copper Age population from Northern Israel, 4000 BCE. Overlaps very earliest dates for Uruk.

    Ancestry proportion estimates, 60:20:20 from Levant_Neolithic:Iranian_Chalcolithic:Anatolian_Neolithic. Compares to 50:50 Iran_Chl:Levant_N in Sidon_BA in the model here.

    Their estimate is that this population did not contribute much ancestry to later populations of the Levant, at least from the sites they have dna from.
    (IRC, though they had some problems of different estimates of Iranian farmer ancestry into Anatolians estimated by f statistics methods vs Levant farmer ancestry via PCA and ADMIXTURE back in the paper from 2016. These may be solved.).

    Additionally, buried in this preprint is evidence of major Y chromosomal turnover. We’ve seen this before. The prominence of haplogroup J in Bronze Age and modern Levantines seems to be due to eastern farmer migration.

    Likely, but not a lot of coverage of space and time across the ancient Near East. Some J is present in earliest farmers of Europe and Anatolia (where the best coverage is going into the Neolithic Near East so far) and then previously present in Mesolithic HG of Eastern Europe, the Caucasus and Iran. Not present in the earliest Armenian or Iranian farmers (not in Iran farmer samples so far until 1/2 in the Chalcolithic, along with G). In any event, yes, likely correct came from a population who were more related to eastern farmers than the early Levant (since they’re from pretty much the westmost point of the NE), just not necessarily from one who model with eastern farmer as the majority of their ancestry.

  15. Sorry, slight mistake in how I phrased that (shouldn’t have said descending order) – the Fsts from the Lazaridis paper actually show:

    1) Joint closest populations to Jordan_EBA Fst: 0.012 – Lebanese, Lebanese_Christian, Lebanese_Muslim, Syrian, Saudi. 2) 0.013: Bedouin (Israel), Cypriot, Jordanian, 3) 0.014: Turkish_Jews, Yemeni_Jews, Palestinian, 4) 0.015: Armenian, Assyrian. (Then incrementally higher Fsts with Anatolian Turks, Southeastern Europe, Druze and so on).

Comments are closed.