Multi-regionalism or assimilation?

In Is Multi-regional evolution dead? I argue that recent work on Neandertal/Denisovan admixture in modern humans has been misinterpreted as signifying only a limited contribution of archaic hominins to the dominant "Out of Africa" population. The main points of my argument are:

• Each non-modern specimen so far has yielded evidence of admixture; and since we have only examined only 2 types of archaic hominins in a relatively small part of Eurasia, it stands to reason that additional hominin populations (in Africa and the rest of Eurasia) may have also contributed their DNA to the modern human gene pool
• The uniformity of the common population element in modern humans should not be interpreted as exclusively African in origin. In a multi-regional model with gene flow we expect parts of the genome to be derived from different regions. The residual "admixture" with archaics may represent the constituent elements that have not gone global.

Ann Gibbons, writing in Science repeats the claim in favor of assimilationism:

New DNA data from archaic human species are providing a much higher resolution view of our past. When compared with the genomes of living people, the ancient genomes allow anthropologists to thoroughly test the competing models of human origins for the first time. The DNA data suggest not one but at least two instances of interbreeding between archaic and modern humans, raising the question of whether Homo sapiens at that point was a distinct species (see sidebar). And so they appear to refute the idea that modern humans came out of Africa, spread around the world, and completely replaced the archaic humans they met. But the genomic data also don't prove the classic multiregionalism model, which argues that a single, worldwide species of human, including archaic forms outside of Africa, met, mingled, and had offspring, and so produced Homo sapiens. They suggest only a small amount of interbreeding, presumably at the margins where invading moderns met archaic groups. The new picture most resembles so-called assimilation models, which got relatively little attention over the years.

This is a fallacious argument: even if we assume that the sub-10% admixture represents the entirety of non-African archaic admixture, this does not mean that there was a "small amount" of interbreeding "at the margins".

It may very well reflect differences in population sizes of different Pleistocene hominin populations.

Consider two species A and B, and a composite species C having 95% of genes derived from A and 5% of genes derived from B. This may either mean that there was sporadic and exceptional intermixture between A- with B-poeple. Alternatively, it may mean that there were many more A than B people, and even though nearly all B people were absorbed, they still did not affect the major group substantially.

This is not an insignificant point: occasional admixture at the fringes is quite a different phenomenon than wholesale admixture between demographically unequal populations.

I don't have any data on population sizes in the last 100-200 thousand years, but I would venture that before the technological advances of the last 50 thousand years ago, when full behavioral modernity arrived, northern Eurasians (Neandertals or Denisovans) did not have as great population densities and sizes as could be maintained in either the African homeland of Homo or more temperate climate zones in general. So, I would lean towards the idea that the observed levels of admixture signify substantial admixture between demographically unequal populations, rather than exceptional admixture.