W. Amos, E. Elhaik
The D statistic is widely used to infer introgression and played a pivotal role in establishing as fact that humans interbred with Neanderthals, leaving a lasting genetic legacy. However, D relies on two untested assumptions: mutation rate constancy between groups and a negligible contribution from recurrent mutations. Together, these preclude a previously ignored alternative explanation of the data based on mutation-rate variation across human populations. Here we show that neither assumption is valid and infer that essentially all sites contributing to non-zero D carry recurrent mutations. When D is calculated separately for different sequence motifs, it varies far more than can be explained using a model based on introgressed haplotypes, where non-independence due to being located on the same haplotype largely prevents different motifs from generating different values of D. Large variation in D between triplets appears only possible if D is driven by processes acting at the level of individual bases. We further show that D is dominated by sites where the Neanderthal allele matches the human major allele and, more specifically, where the Neanderthal allele is fixed in humans outside Africa. This is the exact opposite of expectations under the introgression hypothesis, where Neanderthal alleles should mostly be rare and found only outside Africa. Finally, we tested a new model in which heterozygosity modulates mutation rate. Remarkably, heterozygosity differences, relative mutation rate, and D are all strongly inter-correlated, not only in African - non-African comparisons but also in comparisons among African and among non-African population pairs, and even across the genome. We conclude that the mutation rate variation hypothesis provides a more compelling explanation for the observed patterns in human-Neanderthal genetic relationships than the introgression hypothesis and that the concept of archaic introgression into humans and its implications for hominin-derived traits must be rethought.