Ancient DNA of Crusaders

Finding Europeans in the burial supports the observations from archaeology suggesting that the buried individuals could have been Crusaders. But we also found Near Eastern individuals buried in the same pit along with the Europeans. Two possible scenarios could explain this finding. (1) Following a battle between Crusaders/Europeans and a Muslim army/Near Easterners, the dead from both sides were buried in the same pit. (2) Alternatively, the Crusaders recruited local Near Easterners into their army and thus the burial was for individuals who were all fighting with the Crusaders. The historical records explain that the attacks on the Crusaders were led by Muslim armies mostly recruited from Syria, Iraq, Egypt, Turkey, and Bedouin tribes in the region. But we find the statistic f4(Lebanese_Christian, A; Lebanon_MP, Chimpanzee) is always positive when A is any Near Eastern population (Figure S6), suggesting that the present-day Lebanese Christians are genetically one of the closest populations to the Near Eastern individuals found in the pit. These results support historical records on local Christian integration into the Crusaders’ social structure including the military, with indigenous Christians joining the Crusaders as troops or becoming marshals and knights.4 However, we should note here that we are making an assumption that 800 years ago the medieval Near Easterners were already genetically structured in a way that allows us to differentiate between the different medieval populations in the region, i.e., a medieval Lebanese Christian could be genetically differentiated from a medieval Bedouin. Our tests (see below) suggest that genetic structure between the Lebanese and other Near Easterners could have existed more than 800 years ago but genetic structure between the Lebanese religious groups is probably more recent and therefore our tests might not be able to unambiguously differentiate between the Lebanese Christians and the Lebanese Muslims who lived during the medieval period.


Figure 3 Admixed Individuals in the Crusaders’ Pit

In addition to the three Europeans and four Near Easterners found in the burial, there were two individuals (SI-41 and SI-53) who we could not associate with any specific group. We consequently wanted to explore the possibility that these individuals’ particular pattern on the PCA might be due to their genomes being admixed. Therefore, we first ran an ADMXITURE24 analysis (Figure 3A) which confirmed the presence of three genetically different groups in the Crusaders' pit with SI-41 and SI-53 having intermediate ancestry compositions compared with the other ancient individuals. Next, we simulated hybrid diploid genomes by selecting pairs of individuals, one from the Near East and the second from Europe and we picked a random allele from each individual. We then projected the simulated hybrid genomes onto the PCA plot, looking for hybrid combinations that overlapped with individuals SI-41 and SI-53. We found that a mixture between a medieval Lebanese and a Croatian or a medieval Lebanese and a Hungarian could reproduce the PCA position of SI-53, while a mixture between a Saudi and a Norwegian, or a Bedouin and a Northern Spanish, could reproduce the position of SI-41 (Figure 3B). We assessed these results with formal tests using qpAdm25 and setting the source populations for SI-41 and SI-53 as pairs of populations selected from two groups from the HO set: (1) Medieval Lebanese, Lebanese Christian, Syrian, Palestinian, Assyrian, Iraqi Jew, Turkish Jew, Ashkenazi Jew, Saudi, and Bedouin and (2) French, Italian, Spanish, Basque, English, Norwegian, German, Croatian, Hungarian, Romanian, and Ashkenazi Jew. Then we selected a set of outgroups that are related differently to the source populations: Ust’-Ishim, Kostenki 14, MA1, Onge, Papuans, Chukchi, Karitiana, Eastern hunter-gatherers (EHG), WHG, Natufians, Caucasus hunter-gatherers, Neolithic Iranians, Neolithic Anatolians, and Neolithic Levantines. We found that both SI-41 and SI-53 can be modeled as mixtures of European and Near Eastern ancestries.

For individual SI-41, the best-supported model is a descent from a Near Easterner related to a Bedouin or a Saudi and a European related to Northern Spanish or Basques (Figure 3C, Table S3). For individual SI-53, the best-supported models involved a Near Easterner related to medieval Lebanese, Lebanese Christians, or Jews and a European who can be from diverse origins (Figure 3C, Table S4). We investigated the robustness of our qpAdm inferences by testing, similarly to SI-41 and SI-53, the European individuals from the Crusaders’ pit and, as expected, we found that most of their ancestry was derived from Europeans in contrast to SI-41 and SI-53 (Table S5). Individual SI-53 clustered on the PCA with Ashkenazi Jews, Sicilians, and Southern Italians and therefore we wanted to test whether SI-53 could have descended directly from one of these populations who were previously reported to be admixed,26, 27 with the implication that admixture in SI-53 could then have occurred before the Crusaders’ time and in Europe instead of in the Near East. However, our results (Table S6) show that the Europeans have in general a distinct admixture pattern from the one observed in SI-53; among the populations tested, Ashkenazi Jews’ Near Eastern ancestry is mostly related to Near Eastern Jews, Sicilians’ European ancestry is related to Italians, and Southern Italians have Northern Italians as top sources of their European ancestry. We note here that the reference populations tested are not necessarily the precise parental populations; any genetically equivalent populations could have been involved, but the admixture patterns in SI-41 and SI-53 suggest our data (1) provide direct genetic evidence of admixture between Crusaders and Near Easterners, (2) consequently show admixture could have been common (at least in Sidon), as two out of nine sequenced individuals were admixed, (3) show that admixture with the Near Easterners was not limited to only Levantine Christians but also involved people genetically related to Saudis and Bedouins, and (4) provide support for our previous assumption that 800 years ago the Near Easterners were already genetically structured in a way that allows us to differentiate coastal Levantines from inland people such as Bedouins and Saudis.



Next, we analyzed the variations on the Y chromosomes28 of all ancient males and the mtDNA genomes29 of all ancient individuals and determined the respective haplogroups, which can provide information on the place of origin of the paternal and maternal lineages. All ancient individuals assigned in this work as either European or Near Eastern had Y and mtDNA haplogroups that reflected this ancestry, i.e., Europeans had Y-haplogroup R1b and mtDNA-haplogroups H or U5, while the Near Easterners had Y-haplogroups E, T, J, and Q and mtDNA-haplogroups J1 or HV (Table 1). However, the admixed individuals SI-41 and SI-53 had Y-haplogroup R1b-P312 typically found in European males and mtDNA haplogroups HV0 and T2 present in both Europe and the Near East. In particular, SI-41 carried the DF27 lineage, which is highly prevalent in Iberia (up to 70% of males in Basque) and rare elsewhere,30 supporting our previous results from the autosomal variants that this individual descended from a European related to Northern Spanish or Basques. The combined results from the Y and mtDNA haplogroups suggest that SI-41 and SI-53 possibly had a European father and a Near Easterner mother, but a more complex admixture scenario, where the parents themselves are admixed, could also produce this ancestry pattern. Our tests31 on the X chromosome were inconclusive about the ancestry of SI-41 and SI-53, probably because the reduced sequencing coverage and diversity compared with the autosomal genome make them less powerful (Figure S7).

Our data suggest that admixture occurred in Lebanon during the Crusaders’ time; therefore, we sought to investigate the consequences of this admixture for the genomes of the Lebanese by tracking the genetic changes that occurred in Lebanon before and after the Crusades. We started by using the SGDP set to compute the statistic f4(Lebanon_BA, Lebanon_RP; Ancient West Eurasian, Chimpanzee) which quantifies the genetic changes related to West Eurasians that occurred in Lebanon between the Bronze Age and the Roman period. We found there was an increase in Eurasian hunter-gatherer and Steppe population ancestry in Lebanon after the Bronze Age (Figure S8A), which provides direct evidence for our previous inference that this Eurasian ancestry in the Levant predates both the Crusaders and the Roman period.6 Next, we computed f4(Lebanon_RP, Lebanon_MP; Ancient West Eurasian, Chimpanzee) and found that the statistic is not significantly different from zero for any Ancient West Eurasian tested, thus indicating that there were no significant genetic changes between the Roman period and the medieval period in Lebanon (Figure S8B). We then tested f4(Lebanon_MP, Lebanese_Christians; Ancient West Eurasian, Chimpanzee) and f4(Lebanon_MP, Lebanese_Muslims; Ancient West Eurasian, Chimpanzee); there were no significant genetic differences between medieval Lebanese and present-day Lebanese Christians (Figure S8C), but we found that Lebanese Muslims had significantly lower genetic affinity to West Eurasians (Figure S8D). This genetic change in the Lebanese Muslims could potentially be a result of gene flow from a population genetically distant from West Eurasians. We investigated this possibility by computing f3(Lebanese_MP, A; Lebanese_Christians) and f3(Lebanese_MP, A; Lebanese_Muslims), which test whether the Lebanese groups descended from a mixture between medieval Lebanese and another population. We found that Lebanese Christians cannot be modeled in this way (Figure S9A), but Lebanese Muslims had negative f3-statistics when A was an African or a Central/East Asian population (Figure S9B), indicating that they are admixed from these sources. We confirmed these results by analyzing the 1000GP set using rarecoal-tools32 which identifies genetic ancestry using rare variants and thus complements the low sensitivity of the f-statistics to detect admixture when the ancestral mixing fractions are small. We find an enrichment of African and East Asian rare alleles in the Lebanese Muslims compared with the Lebanese Christians, but we found no substantial differences related to their European ancestry (Figure S10).

Therefore, our results suggest that the genome-wide genetic legacy of the Crusaders cannot be observed today in the Lebanese even though the Lebanese ancestors have admixed with the Crusaders—as it is evident from the genome of individual SI-53. We propose that admixture during the Crusaders’ time was not widespread enough to leave genetic traces quantifiable in the present-day populations. A shared recent genomic history between Europeans and Near Easterners also contributes to rapidly diluting a specific genetic signal of gene flow from the Crusaders. For example, the Neolithic Anatolians and Bronze Age steppe people, who contributed ancestry to the Europeans,17 were themselves a mixture of populations that included Neolithic Levantines and Chalcolithic Iranians18 who were also major contributors to the genetics of Near Easterners.6 We show by simulating hybrid genomes of decreasing European ancestry that a Lebanese who had a French great grandparent cannot be distinguished—by their genetic relationship to ancient West Eurasians—from a Lebanese who did not have admixed ancestors (Figure S11). Thus our data highlight the power of aDNA in informing on past events that would be otherwise undetectable from modern autosomal DNA alone.



Remarkably, the significant genetic changes in the Lebanese following the Crusades appear to have been not from the Crusaders and were not in the Lebanese Christians, but were mostly restricted to the Lebanese Muslims. We analyzed admixture LD33, 34 in the Lebanese in order to date the time when admixture occurred in their history. We found two signals; the first admixture can be detected with overlapping dates in both Lebanese Christians (850–150 BCE, Z = 6.95) and Lebanese Muslims (900 BCE–500 CE, Z = 3.02) and is consistent with finding this admixture in our ancient Roman period samples 240–400 calCE. However, the second admixture was specific to Lebanese Muslims (Figures S12A and S12B) occurring around 1550–1700 CE (Z = 3.69) and can be detected when Africans and East Asians are used as reference populations. The time of this admixture coincides with the Ottoman Turkish rule over Lebanon and we propose that the Turks, who themselves carry East Asian ancestry (Figure S12C) from their Seljuk ancestors,35 brought this ancestry to the Lebanese Muslims. The African ancestry was introduced into the Lebanese Muslims most likely via the slave trade in the Ottoman Empire and the prohibition of non-Muslims from owning slaves during this period.36

In this report, we have presented new whole-genome sequence data from ancient individuals who lived during times when major historical events were unfolding in the Near East. Our samples from the Crusaders period are evidence of a remarkable genetic diversity that coexisted in this region—Europeans, Near Easterners, and their mixed descendants—but this heterogeneity was transient in the genomic history of the Near East, since, with the exception of some Y chromosome lineages, present-day populations derive most of their ancestry from local people who preceded the Crusades. Ancient DNA from the warm Near East is still problematic to retrieve, and in addition our samples from the Crusaders’ pit were crudely buried and partly burned, but our study shows that recovering such aDNA is feasible and that the historical and genetic insights from it are exceptional and not possible from modern DNA alone.

A Transient Pulse of Genetic Admixture from the Crusaders in the Near East Identified from Ancient Genome Sequences
Published:April 18, 2019