Human Effective Population Sizes Are Such A Mess

[Uncommon Ancestor]

Hi,

I was doing some online resarch about effective population sizes for different ethnic groups in order to correlate these data with other parameters in the future. Little did I know that it would be such a tricky thing.

I would seem that the standard number for human total N_e (effective population size) is roughly 10,000, but half the papers only cite that mesure to discredit it. Most estimations vary between 20,000 and 5,000 for the whole human population, yet some publied results are really weird.

Maybe an extreme case would be Park (2014) which states that current Korean effective population size is 30 times bigger than Han Chinese's, no less (3,000 vs 91,000). Other papers also display population-specific effective sizes higher than 10,000, while some argue that 10,000 is an overestimate. Such a mess.

It appears that different methods give different results, but even the same methods over the same database give different results for different papers. As here is discussed, probably different authors mean different different things when speaking about human N_e. Most papers infer different effective population sizes for different times in human history. I confess I get lost here, as I both don't know how can we estimate ancient effective population sizes and, more importantly, don't understand why is it rellevant for discussions about current populations, as I thought current N_e did account for bottlencks and similar.

Normally I would ask for help/advice, but I hopefully will be able to reach a coherent database for myself (or maybe not, who knows), so my primary point is that I'm really interested about why does data vary so much between different studies as well as why would anyone publy an article saying Koreans have a larger N_e than Chinese people (or even humanity) without blinking an eye. I mean, there's shurely something that explains why would anyone believe that data indicating that is believable, but I can't think about anything.

Now, isn't it messy?

[Area54]

Were all the conflicting studies you referenced using the same quantity of interest? If not, surely the disparities you see are to be expected.

[Uncommon Ancestor]

Well, I'm not sure what is meant by quantity of interest. I assume (based on what I've found in Google) that it menas the measure that's actually looked to in raw data. If so, I think they all use Linkage Disequilibrium to estimate N_e, even if other papers I've not referenced use Hardy-Weimberg Equilibrium deviation, or both.

However, I don't see why such huge disparities were to be expected if different quantity of interest were used, given that the final mesure was still the same. Could you please tell me why?

[Area54]

7 hours ago, Uncommon Ancestor said:

However, I don't see why such huge disparities were to be expected if different quantity of interest were used, given that the final mesure was still the same. Could you please tell me why?

My knowledge of the work of Fisher and Haldane and Sewell Wright and all who came after is worse than rudimentary, but if the quantity of interest is different and the methodology to derive the solution is different, why would you expect comparable results?

(Hopefully someone who knows what they are talking about will drop by shortly and put us out of our misery)