Admixture Atavism

A hypothesis.

Consider a population that has low level admixture from some intrusive racially distinct population or populations. If the admixture is low level, then it is possible that the majority of the population in question will not exhibit any obvious phenotypic signs of that admixture (*). However, certain individuals in that population may be “throwbacks” – “atavisms” – to the intrusive stock, looking similar to that intrusive population, likely because that individual happened to inherit a greater concentration of functional genes (e.g., for physical appearance) from intrusive groups. This greater concentration may be due to certain families being derived to a greater extent from the intrusive groups and/or the effects of meiotic recombination and independent assortment resulting in some individuals inheriting a greater proportion of certain genes compared to others in their families (and of their ethny).

The reappearance of traits associated with the phenotypes of the intrusive population(s) that contributed to the admixture is what I call an “admixture atavism.” Thus, the average person of the population will not show these traits, or will not show them to any marked degree, but these atavistic individuals will recapitulate, to a greater or lesser extent, the phenotype of one of the populations that became assimilated into the mass (via a genetic admixture event or a series of events).  Of course, the greater the extent of the admixture in the population, then the more frequent these atavisms will be and, perhaps, the stronger will be the recapitulated phenotype.

Would the phenotype be correlated to a genetic ancestry profile significantly different from co-ethnics that have the mainstream phenotype?  In some cases yes, but not always. In some cases (most cases?), there may be little to no correlation at all between the phenotypic atavism and the overall genetic ancestry. The farther back the admixture event occurred, the greater the chance that the functional genes will no longer will associated with markers of ancestry; haplotypes will be broken up through recombination, etc. Thus, if the functional genes become uncoupled from SNPs or SNP haplotypes that are markers of ancestry, then there may not be any real correlation between the atavism and overall genetic ancestry. 

In addition, if the admixture of the population is subsumed in that population’s use as a parental reference in admixture analysis (*), then an individual’s greater concentration of that admixture may not be readily identified, unless it was significantly higher than the mass of the population (this is assuming that functional genes and overall ancestry have not already been uncoupled from centuries of meiotic recombination and independent assortment). 

Does the phenomenon of admixture atavism explain individuals such as David Bromstad and Enrico Lo Verso?

*Consider - if “testing” uses the main population as a parental reference, so that members of that population are essentially being compared to themselves, then the admixture (particularly low level and ancient) is not going to show up, since that admixture is being considered as the baseline characteristics of the population itself. As a crude example, using a population that is heavily admixed, consider if Central Asians are evaluated against European and East Asian parentals – then the Central Asians would show up as an admixed, relatively equal blend of both parental populations. On the other hand, if the Central Asians themselves were a parental reference population, then Central Asians, compared to Central Asian parentals, would test out at, say, 95-100% Central Asian (5% differences due to individual variation and statistical error).  In this latter case, the same population (i.e., Central Asians), with the same genetics, would be "not admixed" simply because they are essentially being compared to themselves (i.e.,the admixed European-East Asian genepool is being "fixed" as "100% Central Asians").

In other words, if population "X" is used as a parental reference population for admixture analysis, then any prior admixture involved in the ethnogenesis of "X" will simply be considered "X" and not be detected in any tested individuals who are close to the reference "X" genotype. If a member of "X" has a significantly greater proportion of that admixture than dothe "X" parental samples, then it may be detected, but at low levels (or maybe not at all), since the actual level of admixture is being not fully detected since most of it is simply being considered as "X" and not as something else. For example, imagine that "X" is actually a mix of "Y" and "Z." If "Y" and "Z" but not "X" are the parental samples, then the typical "X" may be 90% "Y" and 10% "Z" and then perhaps an "admixture atavism" would be 85% "Y" and 15% "Z." But if "X" is itself the parental, so that a typical "X" tests out at close to or equal to 100% "X," then the atavistic individual may show only 5% or so "Z" - and if this falls within the range of statistical error, then it may not show up at all. But, as alluded to above, if the phenotypic genes are uncompleted uncoupled from those alleles used to estimate ancestry, then there wouldn't be any differences detected in the admixture analysis is any case.