Is Race Purely a Social Construct?

Please try to listen. I just explained this. If you dump all 2 billion East Asians on a small graph with 1 billion South Asians and 50 million Central Asians, the major groups will overwrite and obscure each other, and the admixed Central Asians will appear as a smudge between. That isn't controlling for population density and gives a misleading appearance that the variation is more clinal than it is. The between populations aren't even specifically labelled.

When you label all groups geographically adjacent populations such as Bengalis and Bangladeshis separate cleanly, East Asians cluster tightly, and low density mixed Central Asians such as Uyghur scatter between.

If East Asians overwrite the Central Asians it means the two have the same genetic distance to the other points in the sample on a global scale. If they didn't then they would not occupy the same position on the map. A continuous stream of genetic is the result of adding more (numerically)/more geographically representative samples to the analysis.

Graph B didn't include any samples from West, Central and South Asia or Oceania, so of course the distinction between the samples that were included is clearer. Similarly, ClickItBack's graph showed three distinct clusters because the samples were very geographically unrepresentative.

The PCA analysis in a worldwide population context still showed a cline.

This is ridiculous. You complain about inadequate sampling then present a study that includes Central and South Asians under the same label. It's useless.

It is the most geographically representative sampling procedure that I've seen so far, but there is still a long way to go. As it happens, individuals from other populations overlap with their genetic location, but that is further evidence of a continuous and overlapping distribution of qualities.

In the table I linked below it tells you which populations they classified as East Asian. You can research the population size for each one yourself if you want to know how many individual points there are.


How on earth is that representative? For the second time, the Xu and Lu (2013) graph uses the same PASNP dataset. Please tell me how one dataset is more representative than three.

So when we actually label the populations between East and South Asia, which your study fails to do, we see that they are minor admixed populations (Uyghur, Ladakh and Tharu) between massive population groups.



Your study isn't geographically representative at all. It isn't even labelled point by point.

So? Tens of millions of SNPs were sampled from thousands of people from around the world and when they are subjected to a PCA analysis these individuals exist as a continuum. How would labelling influence this? From West Europe to East Asia adjacent populations would still form a continuous line without gaps. If there were a genetic discontinuity between East and Central Asia it would have shown in the graph. The very fact that you need labels applied to the graph afterwards to distinguish them is evidence of the clinal nature of human biodiversity. Difference is individual.

The labels distinguish geographic location. Adjacent populations don't cluster. People in Bangladesh don't cluster with people in Burma. They cluster with people in Europe. You are wrong. That is all.

Geographical location is redundant. We know that samples from various geographical locations are included in the graph, yet the graph doesn't show any discontinuities that would suggest these geographical samples are distinct. According to this PCA analysis of tens of millions of SNPs, there are no global-scale clusters, which is why you are unable to distinguish between each sample based on their location on the graph in itself. There is only a continuum, with the termini signifying the end of a landmass.

The graph didn't have any west Asian samples, three central and south Asian samples, and one European (or more precisely, people from Utah with northwest European ancestry) sample. Heterogeneous sampling confers heterogeneous graphical points.

Actually I can distinguish, based on studies with labelled geographic samples.

Congratulations for being able to read. On an unlabelled graph it is clear that genetic variation is continuous over space and corresponds to sampling cover. If all East Asians were part of a discrete group then this continuum would not exist.

It doesn't matter. They comprise the continuum depicted in the PCA analysis of tens of millions of SNPs. In the PCA conducted in a worldwide population context there is still a continuum. If there were three distinct population clusters on the basis of tens of millions of SNPs they would have been visible on the analysis without additional labelling aids.

What you don't understand is that some areas of your plot represent 2 billion people, and other areas the same size represent 50 million people.

So what? If these people were genetically distinct on the basis of tens of millions of SNPs there would be a gap somewhere that represents such a genetic demarcation between East Asians and Eurasians. Across geographical space, on the basis of tens of millions of SNPs, there is not genetic lacuna between populations at the global level.

There is a gap, a big one. But when you give equal weight to low density mixed Central Asian populations you don't see it.

Does your graph tell us whether there is a major discontinuity between Burma and Bangladesh? No, it just dumps data and produces a pointless mess. My graph tells us yes, there is. You have nothing.

It isn't my fault that there is genetic continuity across tens of millions of SNPs at the global level to the point that there are no gaps between two geographical areas.

There are no distinct populations at the global scale when assessing tens of millions of SNPs. There is a continuum of variation between a single cline.

It shows me an inferior study with less genetic representation.

I don't see 'Burma' or 'Bangladesh' anywhere.

Your graph doesn't tell us that. It tells us nothing. You can't say what the populations between the East Asian and Caucasoid clusters are. So we can just throw your study in the trash.

There aren't any 'East Asian' or 'Caucasoid' clusters, which is why there are not any gaps in well-sampled areas. I'm not interested in the geographical populations, I'm interested in seeing whether or not there are genetic discontinuities between people as we travel along geographic space. I don't so any gaps in that graph (in well-sampled areas) so I am concluding there are not.

I'm growing tired of this circle.

So I guess you aren't familiar with the Mon or the Bengali?

I guess according to your detached theories these Kachin girls along the Burma/Bangladesh border don't look East Asian.

http://en.wikipedia.org/wiki/File:Kachin.JPG

You don't have any indication of geographic space. The data points are not labelled by precise geographic origin. So it's useless. Get it yet?