Sardinian R-M269 casts doubt on its Neolithic arrival into Europe (or does it?)

I had previously posted on a paper which proposed a Neolithic origin of European R1b1b2 (R-M269) chromosomes. In my criticism for that paper I noted that:

Equally flawed is the inference that R1b1b2 is clinal (Figure 2A). Microsatellite variance is not significantly higher in Turkey than in Europe -- even if one makes the questionable questionable assumption that modern Anatolian Turks are patrilineal descendants of Neolithic Anatolians. The significance of the regression line disappears if 1 or 2 data points are excluded, and the plot has a quite visible "gap" between Turkey and Italy corresponding to the entirety of eastern Europe and the Balkans, i.e. the routes that any putative Neolithic lineages would have entered Europe.

The current paper does not sample the gap I noted in my earlier post, but it looks at Sardinian Y chromosomes, casting doubt on the alleged reduction of diversity from West Asia to Western Europe.

The new paper however has flaws of its own. Its most significant error is its use of the infamous "evolutionary" mutation rate.

Second, Balaresque et al. [17] used a STR specific germ-line mutation rate that placed the TMRCA in the Neolithic age. In contrast we used a unique prior for the microsatellite mutation rate estimates as 6.9×10−4 as recommended by Zhivotovsky and co-workers [24]–[26], see also [27]–[31], that, as reported above, placed the haplogroups TMRCA values in pre-Neolithic times. The difference between the former, evolutionarily effective, and the latter, germ line mutation rates is critical. In fact the haplogroups that survive the stochastic processes of drift and extinction accumulate STR variation at a lower rate than predicted from corresponding pedigree estimates. In particular, under constant population size, the accumulated variance is on average 3–4 times smaller [26]. Hence germ line mutation rates provide evolutionary estimates for haplogroups biased toward much younger age [26].

The degree of blindness required to use the evolutionary rate never ceases to amaze me. I will direct the reader to my earlier post on the evolutionary rate, but let us consider what the assumption of "constant population size" means. Sardinia has a male effective population size numbering in the hundreds of thousands, while the Paleolithic population of the entire European continent numbered at most to a few tens of thousands.

We can thus safely assume 2-3 orders of magnitude of population growth since the Paleolithic. Thus, the assumption of "constant population size" is nonsense, population increase occurred at a faster rate, and correspondingly Y-STR variance accumulated at a faster rate, close to the germline rate.

It is sad to see that geneticists have disagreed for years about what the proper mutation rate should be for evolutionary studies. Most of the work on how the evolutionary rate changes with population growth dynamics was already done by Zhivotovsky et al. (2006), but a group of geneticists seem to have discarded all those observations and uncritically used the simple case of constant population size which corresponds to the "slow" rate of Zhivotovsky et al. (2004) and leads to substantial age overestimates.

It's downright bizarre how neither authors nor peer reviewers can put 2+2 together and figure out that applying a model of constant population size for a population that grew 1,000-fold is muddle-headed oversimplification.

In conclusion, the age estimates of the new paper are wrong and should be divided by 3 or so to provide better estimates. However, I should also point out that age estimation with Y-STRs is associated with wide error margins, even if the number of Y-STRs is quite large. The authors are quite right to point out that the inclusion of an additional Y-STR marker upsets the orderly westward diminution of diversity observed in the previous paper; however, neither they (with their pre-Neolithic estimate), nor the authors of the previous paper (with their Neolithic one) have much of a case.

Where does that leave us? We have no clue as to the origin of R-M269 in Western Europe. I can think of several reasons why this is likely to remain the case:

1. Modern populations are not good representatives of ancient, Neolithic, let alone pre-Neolithic populations from the same vicinities as ancient mtDNA studies have repeatedly shown. Y-chromosome studies are more scant, but there is no reason to think that Y-chromosome distributions are more geographically stable; if anything, strong regional differentiation, and greater historical male mobility may suggest that the opposite is the case.
2. Inferences of diversity clines are based on a patchy collection of convenience samples where most of Europe (let alone West Asia) is underrepresented or sampled non-systematically and at a very small number of markers.
3. Indeed, confidence intervals of Y-STR based age estimates are so wide, that small differences in Y-STR diversity are almost never sufficient to infer greater or lesser antiquity of a population: in short, the odds that a population exhibiting lower Y-STR variance may be older than one exhibiting higher variance are too high to ignore, even with many Y-STR markers, let alone the 10 or so of most scientific studies.

It's about time that geneticists face up to the limitations of their craft. I suppose it's more impressive to make grand statements about European prehistory than to admit to the limitations of the available data.

So, is R-M269 Neolithic or Paleolithic in Europe? I have no idea and I haven't seen any data to convince me one way or another. So, while I welcome new data on its distribution and diversity, and acknowledge that all new data is useful, my own position remains agnostic. A single validated ancient DNA sample would mean more to me than all modern population studies put together.

Related: my Y-STR series.

PLoS ONE doi:10.1371/journal.pone.0010419

A Comparison of Y-Chromosome Variation in Sardinia and Anatolia Is More Consistent with Cultural Rather than Demic Diffusion of Agriculture

Laura Morelli et al.

Abstract

Two alternative models have been proposed to explain the spread of agriculture in Europe during the Neolithic period. The demic diffusion model postulates the spreading of farmers from the Middle East along a Southeast to Northeast axis. Conversely, the cultural diffusion model assumes transmission of agricultural techniques without substantial movements of people. Support for the demic model derives largely from the observation of frequency gradients among some genetic variants, in particular haplogroups defined by single nucleotide polymorphisms (SNPs) in the Y-chromosome. A recent network analysis of the R-M269 Y chromosome lineage has purportedly corroborated Neolithic expansion from Anatolia, the site of diffusion of agriculture. However, the data are still controversial and the analyses so far performed are prone to a number of biases. In the present study we show that the addition of a single marker, DYSA7.2, dramatically changes the shape of the R-M269 network into a topology showing a clear Western-Eastern dichotomy not consistent with a radial diffusion of people from the Middle East. We have also assessed other Y-chromosome haplogroups proposed to be markers of the Neolithic diffusion of farmers and compared their intra-lineage variation—defined by short tandem repeats (STRs)—in Anatolia and in Sardinia, the only Western population where these lineages are present at appreciable frequencies and where there is substantial archaeological and genetic evidence of pre-Neolithic human occupation. The data indicate that Sardinia does not contain a subset of the variability present in Anatolia and that the shared variability between these populations is best explained by an earlier, pre-Neolithic dispersal of haplogroups from a common ancestral gene pool. Overall, these results are consistent with the cultural diffusion and do not support the demic model of agriculture diffusion.

Link