The Sad Case of the Orthodoxy and the Posth Article on Pleistocene Demographics

Just a couple months ago, in the context of the peopling of Ireland, I emphasized on Eupeida (and here) how important it is to put all the Theories Du Jour that are based on modern uniparental distributions through a model based on population demographics and sound logic.

Specifically, I emphasized that ancient population sizes were minuscule compared to modern ones, and that if a population started a long long time ago, with a size that was way way small -- compared to subsequent waves -- that it would give a false signal that the original population was "conquered" or "outcompeted" or "never existed" or originated somewhere incorrect.   I cautioned against those four errors.  

This engendered quite the debate on Eupedia forums.  When backed into a corner and shown the weakness of his "R1b Were Studly Conquerors Theory," the "blindly following the current orthodoxy" folks react badly.
 
Many "Interwebz Scientistz" fail to grasp these concepts.  They favor their own wacky, biased theories based on what they see today only.  If a land is populated by one people, they must be all conquering studs, right?

Today, Posth et. al put out an extensive paper on Pleistocene demographics.  

Its shocking discovery?  Just like Y DNA Hg C existed in Europe in tiny numbers among the very first Europeans, so did mtDNA Hg M.

M disappeared eventually, due to the simple fact that its initial population size was tiny, and that because it had been there so long, the odds that certain women didn't have daughters, each generation, eventually meant it was not passed on.  Remember, we're talking uniparental markers here.  

The authors commented exactly as I did: up to now, people mistakenly believed that Hg M never set foot in Europe -- or that if it did, it was killed off or whatever by a new wave.  Sorry, both theories are wrong.

It is WONDERFUL to see another peer-reviewed, scholarly paper making this exact same point, and backing it up with newfound data.

As the paper indicates:

-These first hunter gatherers started with a TINY initial population size.

-There is a loss every generation of males having males or females having female offspring.

-I've calculated the approximate odds of a male not having a male child or a female not having a female child (i.e. looking like their uniparental marker was "conquered") at 12.5%, each generation, totally random.

-The longer a population has existed in a locale (and being free of mutations), the more generations go by, the greater the chance that random happenstance, chance, etc. will make it appear that a Hg either never existed or was slaughtered in a mass killing/enslavement/mate preference.

Now you have further proof of it.

I'm waiting to hear how Hg M died out because of some studly new more beautiful females who moved in.  Oh woops, Maciamo doesn't post here.  And he doesn't himself bear Hg M.  And M is not linked to R1b...

A Proposal for a New Lexicon for Ancient DNA "Components" Like WHG, EHG, EEF, ANE, and CHG

Some of us a few years back started to decry the ever-ongoing ISOGG renaming process, which coupled with the discovery of new subclades, meant that one year, someone might be deemed R1b1b1a2bab2ba11babd12ba2b1c, and the next year R1b1b2bab2f1faf1fafaf1f1f1a. 

People started saying that it would probably be better to say the first couple letters and the major terminal SNP. For example, R1b-U106 or I2-M26. This was logical and good.  Unlike the terminology, the SNPs never change. And they're shorter to write.
  
Here I humbly propose a new terminology for ancient autosomal samples. I think picking terms like, "WHG" was a mistake, and now that I read about EHG and CHG, I really think so. For the uninitiated, these acronyms stand for "Western Hunter Gatherer," "Caucasus Hunter Gatherer," etc.

People compare their modern genomes, or the genomes of modern populations or ethnic groups, to these ancient samples. And then they use the shorthand, like, "Scottish average 19% CHG." This is highly misleading.
 
Let me give the reasons why I think it is deficient, and tell me if you disagree.
 
1. As we get more samples over time, it will be hard to keep renaming the different samples, if they form a different component. We just saw this with the recent CHG finds. Imagine if we find a detectable signal of ancient genes from Iberia. What will we call that component? "Really Western Hunter Gatherer?"
 
2. The shorthand is deeply misleading (i.e., "Scottish are 19% CHG.") This to me is the most important point. Most people reading this are experts. But I see on so many other boards people who seem to think that some scientist somewhere took a survey of a bunch of ancient samples, "averaged" it, and that we are comparing populations to populations.

We're not. We are not comparing Scots to Western Hunter Gatherers. We are comparing Scots (or any other modern individual or group) to ONE SAMPLE. For WHG, it's Loschbour. For EEF, it's Stuttgart. For ANE, it's Mal'ta. Etc.

3. We don't know that that one sample will turn out to be representative of "Western Hunter Gatherers" any more than we know that taking Danny Devito or the harlequin model Fabio is a representative of a modern Italian. Indeed, as the number of samples we get grows, we know the situation is infinitely more complex.

We all remember, for example, when the first farmers sampled had very unique mtDNA. For a while, people tried to read too much into it. "OMG, what if all farmers bore this odd mtDNA?" was the refrain. But it turned out to be a one-off. This can and will happen again and again as we get more samples over time.

4. The acronyms will get repetitive real fast. We are talking about aDNA, remember? Before farming, the whole world were hunter gatherers. So, many (most) aDNA samples will eventually have -HG after them, if we follow the current convention.

I imagine a world where we have found 26 slightly different hunter gatherer samples, and thus we have one different -HG for every letter in the alphabet! That'd be just silly.
 
For these reasons, but primarily numbers 2 and 3, I think the current practice is misleading and doomed to failure. Europe is a very complicated place. We will find ancient samples with very unique genomes, which are detectable in modern populations. They will all be slightly different from one another, because one sample is, well, one sample... It is highly misleading to say that "John Smith..." or "Estonians are more Western Hunter Gatherer than..." because we have not sampled all, most, or even many Western Hunter Gatherers. (I don't mean to pick on WHG. This applies equally, indeed MORE, with EEF and ANE!)

So, what is the solution?

I think if we purport to be scientific, we need to speak with scientific precision.

If an individual or a modern population bears resemblance to an ancient genome, we should state that it has a percentage similarity to that one sample. And not try to make it more than it is, like the very official and extensive term like, "Eastern Hunter Gatherers."

As for the sample, we should also include the year discovered, the situs of the discovery, and the years Before Present (BP). 

Remember, many of these sites are caves where there have been and will be more discoveries. In other words, I expect there will be many more Loschbours, more Stuttgarts, etc., and it will get quite confusing unless we speak with specificity about when something was discovered and when in time it came from.
 
Let's avoid a situation like we had with terms like R1b1b1b1a2a1b2bc3d, which lose meaning. Let's refer to things with scientific precision.

Examples:

Instead of, "Scots are 19% Ancient North Eurasian."

SAY: "On average 19% of the genes of the modern Scottish population match 2013Mal'ta-24,000BP."

Instead of, "Southern European populations have a lot more CHG blood than I expected."

SAY: "Southern European populations bear many genes matching 2015Kotias-10,000BP."

Instead of, "Sardinians are 45% WHG."

SAY: "Approximately 45% of the genes in the modern Sardinian population resemble 2013Loschbour-6000BP."


This convention is much more accurate.