Cornell researchers have developed new statistical methods based on the complete genome sequences of people alive today to shed light on events at the dawn of human history.

They applied their methods to the genomes of individuals of East Asian, European, and western and southern African descent. They analyzed only six genomes, but made use of the fact that these genomes contain traces of genetic material from thousands of human ancestors, which have been assembled into new combinations over the millennia by genetic recombination.

The main finding of the study, published Sept. 18 in Nature Genetics, is that the San, an indigenous group of hunter gatherers from southern Africa, diverged from other human populations earlier than previously thought – about 130,000 years ago. In comparison, the ancestors of modern Eurasian populations migrated from Africa only about 50,000 years ago.

Previous studies of human demography have primarily relied on mitochondrial DNA from the maternal line or Y-chromosome data passed from fathers to their sons, but those studies are limited by small numbers of genomic positions. This study uses the full genome of each individual, providing a richer, more complete picture of human evolution, according to the researchers.

"The use of genomewide data gives you much more confidence that you are getting the right answer," said Adam Siepel, associate professor of biological statistics and computational biology, and senior author of the paper. "With mitochondrial DNA, you are only looking at one family tree [the maternal line], with one pathway from each individual to its ancestors. We are sampling from all possible pathways."

"What's unusual about our methods is that, not only do they use complete genome sequences, but they consider several populations at once," said Ilan Gronau, the paper's lead author and a postdoctoral associate in Siepel's lab. "This is the first paper to put all of these pieces together," he added.

Previous studies using mitochondrial DNA, Y chromosomes and other markers have estimated that anatomically, modern humans arose roughly 200,000 years ago in eastern or southern Africa; and that the indigenous hunting-and-gathering central and southern African San people – one of the most genetically divergent human populations – diverged from other Africans about 100,000 years ago.

But this study shows that the San people split from other African populations about 130,000 years ago (somewhere between 108,000 and 157,000 years ago). The estimate of an "out of Africa" migration of about 50,000 years ago (somewhere between 38,000 and 64,000 years ago) is consistent with recent findings using other methods, the researchers said.

To conduct the study, the researchers began with a statistical approach that was originally developed to infer divergence times for related but distinct species, such as the human, chimpanzee and gorilla. They faced a number of challenges in adapting these methods for use with human genome sequences. For example, the great ape genome method assumes that gene flow stops after two species diverge, because they can no longer mate. That is not true for distinct human populations, and without accounting for gene flow, the divergence times would have been underestimated, Siepel said.

Gronau used mathematical techniques to work around that problem and then created elaborate computer simulations to demonstrate that the new method worked within known parameters of human divergence.

The study was funded by the Packard Foundation, National Science Foundation and National Institutes of Health.