Klyosov correctly criticizes the authors for using the evolutionary mutation rate:
A common ancestor of all 99 Cohanim lived 1,075 ± 130 ybp, and this timing is reproducible for 9-, 32412-, 17-, 22- and 67-marker haplotypes. A much higher values of 3,190 ± 1,090 and 3,000 ± 1,500 ybp were obtained in the cited paper (Hammer et al. 2009) using incorrect methods and incorrect mutation rates.
Please note that Klyosov uses both a mutation-counting "linear" method, as well as a "logarithmic" method which relates age to the fraction of inferred ancestral ("base") haplotypes in a collection; the "linear" method produces age estimates of comparable error as the more commonly used ASD/variance methods, but the "logarithmic" method produces sufficiently worse (larger confidence intervals) estimates. You can easily modify the Y-chromosome Microsatellite Genealogy Simulator to test the performance of Klyosov's methods.
In conclusion: Klyosov is right to criticize Hammer et al. for using the evolutionary mutation rate. However, his methods do not warrant the strong conclusion that Cohanim J-P58's share a common origin in the last 1,000 years. See my own post on the Hammer et al. paper for my thoughts on the matter.
The whole argument is moot. What proof is there that a priesthood existed among the Semitic language speaking inhabitants of Judah? As distinct from the other inhabitants who spoke other languages. The Hittites, Phoenicians and Philistines have all gone the way of the Dodo. The notion of a priesthood among the Jews comes from the religious books. Are Jews Hebrews or Israelites? All these questions have not been answered. It is odd that the Gospels speak of Pharisees, and Sadducees, not Levites or Cohens. So other than references in the Old Testament, where were these priests?
ReplyDeleteThe time calculation by Hammer et al is influenced by religious belief not by any reality. It should be taken with a large grain of salt. There are Cohens and Levites who belong to other haplogroups or have other haplotypes of J1. The Russian Klyosov himself believes a lot of the factoids surrounding Jewry. I don't. To me Jews are just European converts to Judaism with some minor admixture, mostly male, from the Middle East which got exaggerated or skewed due to restricted marriage practices and bottlenecks. No European Jew or other Jewish people can prove his or their connection to the Middle East or to Judah or to the Hebrews or another ancient people.
Any dating of haplogroups or haplotypes have to be taken with caution as not enough studies have been done on each haplogroup to show real mutation rates using a large number of people. So far, most dating has just proven the researchers' preconceptions, and prejudices regarding the origins of certain ethnic groups, races and religions.
Ponto says: “To me Jews are just European converts to Judaism with some minor admixture, mostly male, from the Middle East which got exaggerated or skewed due to restricted marriage practices and bottlenecks. No European Jew or other Jewish people can prove his or their connection to the Middle East or to Judah or to the Hebrews or another ancient people”.
ReplyDeleteThis is to speak clear. Thank GD (genetic distance) we aren’t on “genealogy-dna” or “forums-dna”.
>However, it should be noted that the +/- 130ybp limit around the estimated age of Cohanim J-P58 is too small, and gives the false impression of great confidence in the age estimate, which is not really warranted by Y-STR markers.
ReplyDeleteAll margins of error in the paper were calculated using a systematic and justified approach. They were not taken arbitrarely. Margins of error depend on a number of mutations in the series of haplotypes and on margins of error of the mutation rate constant. All of this was examined and verified.
******************
>Please note that Klyosov uses both a mutation-counting "linear" method, as well as a "logarithmic" method which relates age to the fraction of inferred ancestral ("base") haplotypes in a collection; the "linear" method produces age estimates of comparable error as the more commonly used ASD/variance methods, but the "logarithmic" method produces sufficiently worse (larger confidence intervals) estimates.
The logarithmic method was employed in the paper NOT for calculations of time spans to a common ancestor, but as a criterion that the haplotype series is descended from ONE common ancestor. In this case the linear and the logarithmic methods should give approximately the same time span. That was the whole and the only point.
*****************************
>However, his methods do not warrant the strong conclusion that Cohanim J-P58's share a common origin in the last 1,000 years. See my own post on the Hammer et al. paper for my thoughts on the matter.
In fact, J-P58 DO share a common ancestor in the last 1,000 years. One can take a qualified look at their haplotypes to say so right away. However, the "CMH" folks have a common ancestor 4,000+/-520 years back.
All margins of error in the paper were calculated using a systematic and justified approach. They were not taken arbitrarely. Margins of error depend on a number of mutations in the series of haplotypes and on margins of error of the mutation rate constant. All of this was examined and verified.
ReplyDeleteThat is false. Even if one samples the entire population and knows the mutation rate perfectly (no error), the generation length perfectly, and the mutation process behaves perfectly according to the symmetrical stepwise model, there are still wide confidence intervals around age estimates which your methodology ignores.
>"That is false. Even if one samples the entire population and knows the mutation rate perfectly (no error), the generation length perfectly, and the mutation process behaves perfectly according to the symmetrical stepwise model, there are still wide confidence intervals around age estimates which your methodology ignores".
ReplyDeleteOne can come up with all kinds of reasons that something/everything is wrong, that life on the Earth could not have possibly appear, that there is no such thing as the DNA, that mutations in the DNA depends on a diet, etc. etc.
I repeat again, that for a margin of error estimate I use four factors: (1) the average number of mutations per marker, (2) the margin of error for the average number of mutations per marker at 95% confidence (two sigma), (3) the mutation rate constant for the given haplotype format, and (4) the margin of error for the mutation rate constant.
For a large number of mutations in a given series of haplotypes a margin of error is approaching 10%, and cannot be lower in view of the above four parameters.
An example: 750 of 19-marker haplotypes of Iberian R1b1b2 (Adams et al, 2008) contain 2796 mutations, and are derived from a common ancestor who lived 3750+/-380 ybp (see below).
All 750 haplotypes contain 16 identical (base) haplotypes, which gives ln(750/16)/0.0285 = 135 generation (w/out correction for back mutations), or 156 generations (25 years per generation), that is 3900 ybp. As one see, it is WITHIN the indicated above margin of error.
By the way, 2796 mutations divided by 750x19 = 14250 markers give 0.196 mutations per marker on average, and 0.196/0.0015 = 131 generations (w/out a correction for back mutations) or 150 generation with a correction, that is 3750 ybp. An error margin here, if one knows how to calculate it, is +/-380 with 95% confidence (two sigma). That is how 3750+/-380 ybp appears.
0.0015 mut per marker per generation (25 years) and 0.0285 mut per haplotype per generation are the mutation rate constants for the 19-marker haplotype.
By the way, if to consider a slight asymmetry of mutations in this series (the degree of asymmetry equals to 0.56), it would give 3625+/-370 ybp instead of 3750+/-380 ybp.
What say you? I challenge you to calculate YOUR error margin in the given case. No empty words, please. And, yes, how about the logarithmic method? Have you ever used it before dismissing?
I use it as a secondary tool, to verify my calculations and to make sure that the haplotype series is not "distorted".
I have used both of your methods and they do not produce error margins of 10%.
ReplyDelete