TMRCA's : Knortvedt vs 1000 Genomes

[bicicleur 2]

Knordtvedt has calculated the age of some mutations in haplogroup I

http://knordtvedt.home.bresnan.net/

I don't know his exact methods but I suspect he combines STR- and SNP-analyses
He estimates a new generation every 30 years

Data from the 1000 genomes project became available last year.
It allows to construct a full y-DNA tree and count all SNP's.

Here are 2 trees constructed like that :



This is the source :

http://www.goggo.com/terry/HaplogroupI1/

The tree is subject to the chosen samples.
Assumed mutation rate 3.0 x 10 -8.

The ages for haplo I1 come out older than the estimate from Knordtvedt.

Anyway, mutation rate has to be estimated, and also how long every generation switch lasts (this is largely influenced by the culture and habits of the tribe).

It looks like accurate TRMCA estimates are not possible for the moment.

 

[Maciamo]

The estimation of the age of haplogroups is a fiendishly complicated business, because it does not rely "simply" on mathematical calculations, but also on historical population sizes for each haplogroup and subclade, which can vary tremendously based on:

- the environment : fertile and resource-rich region like India or eastern China can support much higher population densities than cold and inhospitable Siberia or Iceland.

- climate changes in the same region : the warm and humid climate of the Neolithic allowed for population boom in Europe and North Africa, but was followed by a population collapse when the climate cooled and dried.

- evolution in technologies : farmers can sustain population sizes 10 to 100 times higher than hunter-gatherers on the same area. This was not a one time event. Many progressive and cumulative agricultural innovations in history have allowed big changes in the quantity of food produced. A population of 1 million will develop new mutations 100 times faster than a population of 10,000. Genetic diversity within one haplogroup may therefore depend more on the total, accumulated historical population of that haplogroup than on anything else.

- diseases and epidemics : the plague wiped out 30–60% of Europe's total population in the mid-14th century alone. The arrival of Europeans in the Americas brought new germs that could have killed up to 90% of North American natives. There are plenty of historical examples like that, and those that happened in prehistory are left unrecorded. Pandemics can, and certainly do change the genetic landscape of a region, wiping out entire haplogroups or subclades, and reducing diversity in others. Without knowing exactly when and where pandemics happened and how serious they were, there is not a chance to get age estimates right without ancient DNA.


Over the last decade or so, the age of I1 has been estimated anywhere between 20,000 and 4000 years old. The age of haplogroup T is still mentioned on Wikipedia in the range 19,000 to 34,000 years old, while the first tree here places the origin of LT at only 15,000 years old (meaning that T itself must be at least a few thousand years younger). There is nothing accurate about those "calculations". It's anyone's best guess against anyone else's.

 

[bicicleur 2]

Over the last decade or so, the age of I1 has been estimated anywhere between 20,000 and 4000 years old. The age of haplogroup T is still mentioned on Wikipedia in the range 19,000 to 34,000 years old, while the first tree here places the origin of LT at only 15,000 years old (meaning that T itself must be at least a few thousand years younger). There is nothing accurate about those "calculations". It's anyone's best guess against anyone else's.

You're right, LT is surprisingly young in this tree. But there are only 7 LT people in the sample, so that could be a coincidence, maybe if they would add more LT people their TMRCA would get older.
Overall the sample group should be bigger to get a more acurate view.
I guess making up such a tree involves a lot of data to be processed and checked, it's hard work.
And there is the need to 'calibrate' the tree as both mutation rate and time between 2 generations are uncertain.
Nevertheless I would be interested to see more such trees, even if it lacks accuracy, at least it gives a good perspective in comparing the ages of several clades to each other.

 

[adamo]

Genographic places t-m70 (T1a) roughly between 35,000-20,000 years ago. T1a1 is placed at 25,000 with 15,000 years youngest possible age.