October 26, 2012

IBD length distribution and demographic history (Palamara)

Another interesting new paper in AJHG deals with the problem of inferring the demographic history of a population from the length distribution of IBD segments. I wonder how admixture (which is likely to have occurred in both chosen real-world examples used in the paper, the Ashkenazi Jews and the Maasai) may affect the accuracy of the reconstructed demographic history.

In any case, the conclusions are worth mentioning in themselves. For the Ashkenazi:

We obtained an improved fit for a population composed of ~2,300 ancestors 200 generations before the present; this population exponentially expanded to reach ~45,000 individuals 34 generations ago. After a severe founder event, the population was reduced to ~270 individuals, which then expanded rapidly during 33 generations (rate r ~ 0.29) and reached a modern population of ~4,300,000 individuals.
And, for the Maasai:

Optimizing a model of exponential expansion and contraction (Figure 1A), we obtained a good fit to the observed IBD frequency spectrum (Figure 6), suggesting that an ancestral population of ~23,500 individuals decreased to ~500 current individuals during the course of 23 generations (r ~ -0.17). We note that this result might not be driven by an actual gradual population contraction in the MKK individuals, but it most likely reflects the societal structure of this seminomadic population. ... We thus used the village model to analyze the MKK demography and relied on coalescent simulations to retrieve its parameters: migration rate, size, and number of villages that provide a good fit for the empirical distribution of IBD segments.We observed a compatible fit for this model, in which 44 villages of 485 individuals each intermix with a migration rate of 0.13 individuals per generation (Figure 6).
If I understand this correctly, it appears that Maasai (MKK) individuals share long IBD segments not because their population has contracted (and hence they're all descended from a limited number of founders, as is the case for Ashkenazi Jews), but rather because their social structure follows the "village model" in which people share shallow ancestry (and hence long IBD) with other people in their "village" and exchange genes with other "villages".

The American Journal of Human Genetics, 25 October 2012 doi:10.1016/j.ajhg.2012.08.030

Length Distributions of Identity by Descent Reveal Fine-Scale Demographic History

Pier Francesco Palamara et al.

Data-driven studies of identity by descent (IBD) were recently enabled by high-resolution genomic data from large cohorts and scalable algorithms for IBD detection. Yet, haplotype sharing currently represents an underutilized source of information for population-genetics research. We present analytical results on the relationship between haplotype sharing across purportedly unrelated individuals and a population’s demographic history. We express the distribution of IBD sharing across pairs of individuals for segments of arbitrary length as a function of the population’s demography, and we derive an inference procedure to reconstruct such demographic history. The accuracy of the proposed reconstruction methodology was extensively tested on simulated data. We applied this methodology to two densely typed data sets: 500 Ashkenazi Jewish (AJ) individuals and 56 Kenyan Maasai (MKK) individuals (HapMap 3 data set). Reconstructing the demographic history of the AJ cohort, we recovered two subsequent population expansions, separated by a severe founder event, consistent with previous analysis of lower-throughput genetic data and historical accounts of AJ history. In the MKK cohort, high levels of cryptic relatedness were detected. The spectrum of IBD sharing is consistent with a demographic model in which several small-sized demes intermix through high migration rates and result in enrichment of shared long-range haplotypes. This scenario of historically structured demographies might explain the unexpected abundance of runs of homozygosity within several populations.



pconroy said...

So what was the second population constriction for AJ's? If it was 33 generations ago, that would yield a figure of 957 yo at 25 years per generation, or 1055 AD.

That date would coincide almost exactly with the First Crusade 1095–1099

Tom Bridgeland said...

Or roughly 1330 for a 20 year generation time. The plague years in Europe. Since the Jews were sometimes blamed for spreading the plague, I can see why many might fall away and disappear from the record, besides those that just died, of course.

Unknown said...

It does not have to be a death event. It could just be that being Jewish became unpopular within the Jewish community, and folk dispersed into the general population leaving behind a small hard core Jewish closed breeding pool.

It could also be that this was the time of Out-of-Israel. The time when that small group left the larger Israel breeding population to settle somewhere in Europe. It would look like a sudden contraction in population.

Ted Kandell said...

Medieval Jews did not "fall away and disappear from the record".

They were murdered. The victims of mass murder, on an unprecedented scale before the Age of Exploration and the recent genocides.

Jews during the Crusades

"The Jews of the Rhine district were decimated: it has been calculated that about 4,000 were killed or slew themselves."

Mass Murder of Jews during the Black Death

"At Strasburg the mayor refused credence to the rumors, and declared his intention of sustaining the Jews; whereupon he was removed from his post, and more than 2,000 Jews of the city were put to death (Feb. 16, 1349)"

"The largest number of victims is recorded at Mayence, where no less than 6,000 are said to have been slain Aug. 22, 1349."

The Prague massacre of the Jews in 1389

"The populace thereupon attacked the ghetto (April 18, 1389) and killed about 3,000 Jews."

The Seville Massacre of the Jews of 1391

"A riot finally broke out on March 15, 1391, during which several Jews were slain; but the nobles, who protected them, soon quelled the uprising. Three months later, on June 6, the persecution was renewed. The infuriated populace attacked the ghetto from all sides, plundering and burning the houses. More than 4,000 fell victims to the mob's fury, although most of the Jews accepted baptism to save their lives."

Chemielnitzki's Pogrom of 1648 in the Polish-Lithuanian Commonwealth

Map of the Massacres of Jews in Poland and Russia 1648-1656"

"The horrors of the war were brought to a climax by the outbreak of the plague in Poland. The Jews in the provinces of Cracow, Posen, Kalish, Piotrkov, and Lublin perished in large numbers, both by the sword of the enemy and by disease. Only after 1658 did the disturbance caused by the war begin to subside. According to the chronicles, the number of Jews who perished during this time (1648-58) exceeded half a million. Over three hundred Jewish communities (740, according to the unreliable Samuel Phoebus in "Ṭiṭ ha-Yawen") were massacred and sacked. Approximately only one-tenth of the Jewish population remained in Polish Ukraine, Volhynia, and Podolia. The remainder had either perished or had emigrated into Lithuania, Poland proper, and the states of western Europe. Jewish fugitives from Poland, and captives ransomed from Tatar bondage, could at that time be met with in all the countries of Europe and Asia."

(The present-day most conservative estimate is a minimum of 50,000 Jews killed out of a total population of approximately 100,000 Jews in Eastern Europe in 1648.)

Genetics does not happen in a vacuum.
The tragic history here has to be considered in any discussion of Ashkenazi population genetics. An understanding of the causes of this mass murder will help explain why we see large numbers of Hispanics (and others, like French Huguenots, Swiss Mennonites, Cajuns, and people from the British Isles) who not only share "Ashkenazi" Y and mtDNA unparental clades, but also carry what are thought to be "Ashkenazi" genetic diseases, but not commonly recognized or screened in these populations. Knowinng about this tragic history can help save lives today.

Unknown said...

I beleive more than 270 people collectively survived those genocides Ted. This appears to be an earlier event.

Ted Kandell said...

Annie, it isn't that 270 survived at a single time. What happened was that the successive massacres eliminated lineages, and that others from outside the population (Jews from other regions such as Spain, and known converts) replaced those ancestors. The end result was that there was a total of 270 effective ancestors circa 1656 from asynchronous times in the past.

Also, please be aware of the following:

1. The total number of Judeans exiled by the Babylonians in 598 and 589 BCE was just 4000 men women and children.
This is well in accord with the estimates of the carrying capacity of the land in the Jerusalem region during the Iron Age.

2. Clearly these Iron Age Judean ancestors were supplemented to a degree by Mesopotamian and North Syrian admixture during the return in 444 BCE, but also a continuing process of conversion over the centuries, including that of the Idumeans in the 1st century BCE.

3. Ashkenazi Jews seem to generally descend from Second Temple era Judeans who were captured and exiled in 70 CE after the First Jewish War. The tradition is that 4000 young Judean men were taken prisoner and settled throughout Southern Italy (Rome, and Naples to Bari, Oria and Calabria, but not Sicily).

4. The process of the imposition of State Zoroastrianism and Christianity in the 3rd century in Persia and the Roman Empire halted the large-scale conversion to Judaism which occurred before. However, this large-scale conversion continued to take place in Arabia and North Africa until the advent of Islam. Based on the Y data we see a non-trivial amount of Persian Gulf Arabians and North Africans who converted and admixed with existing Jewish population in the pre-Islamic Era, as was documented in the Babylonian Talmud, Tractate Kiddushin, and Ibn Khaldun's history of North Africa.

5. As the Germanic states converted from Arian non-Trinitarian Christianity to Catholic Orthodoxy, each in turn expelled the Jews. Hispania, "Francia", and Lombard Italy. Jews were expelled from Alexandria and Egypt in 415.The Byzantine Empire expelled the Jews in 712, just at the time of the Muslim Conquest of Spain. The only areas that had a continuous Jewish settlement were Calabria (under the Emirate of Bari) and Septimania (Narbonne and the surrounding region) from 640 CE onward.

6. Jews were successively driven out of North Africa and Spain in 1147-60 by the Almohades, and the Kingdom of France in 1182 (Northern France). It would lseem that the only larger region where Jews cold find refuge north of Italy and Spain during the Black Death was the Papal State of Avignon, where Jews were protected by the Pope.

It may be that Ashkenazi ancestors were the Jews of the City of Rome and Southern Italy (Calabria and Apulia) after 70 CE, and the Jews of Narbonne starting around 640 CE. The Sephardi population may have included large numbers of Near Eastern Jews who began to migrate to Muslim Spain after 711 with the Umayyad Conquest. The Iberian Jews may have included a substantial proportion of non-Second Temple Judean ancestry, both Mesopotamian Jews and the many converts from Eastern Anatolia, Central Iraq, and the Persian Gulf region.

Ted Kandell said...

Annie, please see the following paper by Montgomery Slatkin of Stanford about the historical effective population sizes of the Ashkenazi ancestors:

A Population-Genetic Test of Founder Effects and Implications for Ashkenazi Jewish Diseases

This paper used several rare AJ Medelian disease alleles which passed the neutrality test to create a model of historical population growth. It does not depend on LD and LD decay.

The conclusion is that the AJ ancestral population was (N0) 3000 at 70 CE and 600 (N1) at 1350 CE.

600 individuals in 1350 in fact is not the breeding population but the total population. The actual set of ancestors would have been around 300.

As an aside, I carry APC I1307K (rs1801155=A) but I have no family or personal history of colorectal cancer.
Ted Kandell Rare SNPs (extracted by Ian Logan from OpenSNP.org data

"Table 3 shows that a founder event in a.d. 70 is likely for the smallest assumed population size, N0=150, but, even so, there is an ∼10% chance that two or more lineages carried I1307K. For N0=150 and 600, though, the inferred average number of lineages at a.d. 75 is small, suggesting that the allele had a low frequency before the founder event. Only if N0=3,000 is the average higher."

"N0=3,000 and N1=600 are most consistent with these two data sets considered together. The demographic model cannot be taken literally, but these results suggest that there was a severe reduction in population size during the period between a.d. 1100 and 1400 and a less severe reduction in size after a.d. 70. A severe bottleneck in a.d. 70 or additional bottlenecks between a.d. 70 and 1348 would make a founder effect in Ashkenazi Jews more likely and, hence, would make it more difficult for I1307K at APC not to have undergone a founder effect in a.d. 70."

"Figure 4 shows the results for two different population sizes in a.d. 1348 (200 and 1,000, corresponding to census sizes of 600 and 3,000). Even with N1=200, there is only a very low probability (<0.02) that an allele carried by one of the 400 founder chromosomes could reach a frequency >0.015."

The following paper published earlier in 2003 found APC I1307K (rs1801155=A) in Ashkenazim (6%) Sephardim (1%-2%) and came to the conclusion that this allele arose on a haplotype background common to Jews and Arabs, but the mutation occurred at the start of the Jewish Diaspora in 70 CE:

Genetic Anthropology of the Colorectal Cancer–Susceptibility Allele APC I1307K: Evidence of Genetic Drift within the Ashkenazim

"A common progenitor haplotype spanned across APC I1307K from the centromeric marker D5S135 to the telomeric marker D5S346 and was observed in individuals of Ashkenazi, Sephardi, and Arab descent. The ancestor of modern I1307K alleles existed 87.9–118 generations ago (∼2,200–2,950 years ago). This age estimate indicates that I1307K existed at about the time of the beginning of the Jewish diaspora, explaining its presence in non-Ashkenazi populations. Our data do not indicate that selection operated at I1307K (D5S346, P=.114; D5S135, P=.373), providing compelling evidence that the high frequency of disease-susceptibility alleles in the Ashkenazim is due to genetic drift, not selection."

The result is that papers using different methodologies, linkage decay, and rare allele frequencies along with neutrality tests, come to a common conclusion about the population history of Ashkenazi Jews.

Ted Kandell said...

If anyone needs more historical "proof" of these 270-300 ancestors, here it is:

Rabbi Mattityahu Ashkenazi-Treves, b. 1323 Marseilles France

Early Ashkenazi-Treves family genealogical tree

(Please disregard all the idiocy out there about "Davidic Descent" of this family, and about any proven connections earlier than his father who was a Jewish refugee from Trier in the Rhineland who went to Marseille in 1306.)

The evidence:

1. The Treves family of Rhodes matches:
2. The Ashkenazi family of Aleppo
3. The Eventov/Evenchik family of Belarus who descend from one "Yaakov Ashkenazi"

There are other interesting Y connections:

1. The Zemanis family of Vilnius Lithuania who claim to have been "Ximienes" (Jimenez)
2. The Sephardic Toledano family of Morocco
3. The Martinez-Loera family of Nuevo Leon Mexico
4. The De Cubilla family of Boqueron Panama
5. The Columbian descendants of the Conquistador Fernando Cardenas y Zapata who accompanied Viceroy Antonio de Mendoza to Peru in 1550

It may also be that Rabbi Asher ben Jehiel, also a 1306 refugee from the Rhineland to Marseilles who later settled in Spain (as did Rabbi Mattityahu Treves for 20 years)was a very close relative of Rabbi Johanan "Ashkenazi" (the German) Treves (of Trier). He had 8 sons in Spain who all became Rabbis, including the Chief Rabbi of Toledo.

Rabbi Asher ben Jehiel b. circa 1250

In turn he was a genealogical descendant of the famous Kalonymos family of Bari Apulia.

Rabbi Eliezer ben Nathan of Mainz b. circa 1040

Rabbi Eliakim ben Joseph of Mainz family tree

In the 970 CE battle with the Emirate of Bari, one "Kalonymos" aided the German Emperor Otto III and was granted "privileges" in Mainz. The trade fairs of the "Cathedral Age" in Mainz and Troyes attracted Jews northward from Southern Italy and Narbonne.

The Kalonymos family of Bari, Rome and Mainz

Be aware the most members of the Kalonymos family of Mainz were MASSACRED during the First and Second Crusades along with most other Jews of the Rhineland.

mtDNA K1a1b1a is found in 19.7% of the Ashkenazi population. N1b2 is found in another 9.9%. These two recent mtDNA clades alone account for 30% of all Ashkenazi mtDNAs, but are also found among Sephardic Jews, in Northern Mexico, and as far afield as Iranian and South Indian Jews. (We know that one of the 7 families of the so-called "White Jews" of Cochin was named "Ashkenazi" and another was named "Rotenberg". They arrived in the 17th century with the Dutch. No doubt they mixed with the existing earlier Cochin Jewish population. )

Notice that the famous Rabbi Solomon ben Isaac of Troyes (Rashi) whose mother was from the Kalonymos family of Bari and Mainz had three daughters, and Eliezer ben Nathan had four daughters. These women were ancestresses of many Jews today. It may very well be that these were quite closely related. Were these Rhineland ("Ashkenaz") Jewish women the source of K1a1a1b1a?

Documented history also seems to support the very limited number of ancestors of Ashkenazi Jews at 1350 CE.

Eran Elhaik said...

Actually, the 13th century is the time when what you call "Ashkenazi Jews", but actually are Eastern European Jews fled from Khazaria into Eastern and Central Europe and established the majority of Eastern European culture. These findings are in agreement with my findings ("The Missing Link of Jewish European Ancestry: Contrasting the Rhineland and the Khazarian Hypotheses") published in:

If these findings are correct, they relate to one of the most interesting questions regarding the Khazarian Hypothesis. While there is no disagreement about whether Khazars converted to Judaism and maintained it as well as fleeing from their kingdom in the 13th century, it is unclear how many Jews escaped and whether they largely or marginally supported the formation of Eastern European Jewry. My findings support a major contribution. The findings of Palamara et al. (2012) indicate that the effective population size was quiet small. One reason for that is that only the elite of the population escaped. But regardless, we both agree that these people later expanded and established the majority of Eastern European Jewry (or else the signal would have been different).