Monday, April 02, 2018

Human generation times

1. The paper discussed in my previous two blog posts, "The Genomic Formation of South and Central Asia" uses a time per human generation of 28 years.

2. This time is derived from a paper by one of the authors, "A genetic method for dating ancient genomes provides a direct estimate of human generation interval in the last 45,000 years", P. Moorjani et. al. That paper comes up with an estimate of 26-30 years, from which the mid-value is used.

3. "The Genomic Formation...." uses this human generation time of 28 years in two places.
...we estimate that the time of admixture between Iranian agriculturalist-related ancestry and AASI ancestry in the three Indus_Periphery samples was 53 ± 15 generations ago on average, corresponding to a 95% confidence interval of about 4700-3000 BCE assuming 28 years per generation
....Using admixture linkage disequilibrium, we estimate a date of 107 ± 11 generations ago for Iranian agriculturalist and AASI-related admixture in the Palliyar, corresponding to a 95% confidence interval of 1700-400 BCE assuming 28 years per generation.
4.  Here we notice something.  (107 + 11) generations * 28 years =  3304 years.  The Iranian agriculturalist and AASI admixture in the (modern) Palliyar  dates to 1300 BCE to 700 BCE.

5. Moreover, if we go with the generation time measured with the !Kung, 25.5 years - the lower end of P. Moorjani's estimation, it only makes it worse for the idea that all this admixture happened around the collapse of the Indus Valley Civilization.

6. The ancestry of the Indus_Periphery samples may likewise become too late for the archaeological  record of agriculture.

PS: Note that for the Indus_Periphery, the date is 53 ± 15 generations prior to the date of the Indus_Periphery samples, i.e., prior to 3100-2200 BCE.

PPS: The life expectancy at birth in the Paleolithic is estimated to be 33 years per Wiki.  The per generation time of 28 years means that the average age of a woman relative to the birth of her children is 28 years.  E.g., if women uniformly bore children at ages 20, 24, 28, 32 the generation time would be (20 + 24 + 28 + 32)/4 = 26.  You can see that women would be bumping up into the life expectancy.  On the other hand, what is important is actually the conditional life expectancy, which is the life expectancy of women who survived up to at least one live birth, which may be better than that 33 years.  Still, one would think that in the Paleolithic, women in the 18-28 age group would have more children than the women 28-38 age group, if only for the reason that there's more of them.

PPPS: The Wiki that gives the Paleolithic life expectancy at birth also says "Based on the data from modern hunter-gatherer populations, it is estimated that at 15, life expectancy was an additional 39 years (total 54)"" Supposedly after surviving birth the probability of reach age 15 was 0.6.

A paper linked at the Wiki gives (figure 10) the yearly mortality rate ("Proportion that die") of the Aché people of Paraguay.  The yearly mortality rate starts at 0.10 around birth, and rapidly falls to 0.02 between five and ten years of age (the curves are different for males and females).  It then remains steady between 0.01 and 0.02 until age 40, after which it starts rising fairly rapidly.  A cohort of 100 women aged 20 years would reduce to 86 by age 30, and 74 by age 40 if the annual mortality rate ("proportion that die") was 0.015.

In comparison, in 2013, per WHO data, globally,  the adult mortality rate - the probability of dying between ages 15 and 60 - was for men 0.182 and for women 0.121.

Also note that if women were having a lot of babies, and if the population was growing very slowly if at all, then there must have been a corresponding number of deaths.