Was Pale Skin an Elite Trait?
Skin pigmentation in Europe is one of the best-known cases of recent human evolution. The prevailing explanation holds that in low-UV environments, lighter skin allows more efficient vitamin D synthesis, so alleles associated with depigmentation rise over time. This core mechanism is supported by the data. In ancient individuals from western Eurasia in the AADR dataset, annual UVB radiation strongly predicts darker pigmentation scores, and older individuals tend to have darker pigmentation scores as well.
However, climate may not be the only factor shaping pigmentation. As societies became more complex during later prehistory and early history, social hierarchy and occupational specialization also increased. Elites, administrators, and many skilled craftsmen likely spent more of their lives indoor, protected from direct sunlight. If that buffering from UV exposure persisted across generations, it could potentially create a socially structured pattern in pigmentation.
To explore this possibility, I tested whether alleles associated with lighter skin are disproportionately found in individuals carrying higher educational-attainment polygenic scores, which may serve as a rough proxy for social status or cognitively demanding roles in ancient populations.
Educational-attainment polygenic scores are not a direct measure of ancient social status. However, they capture a set of genetic variants associated with cognitive and educational outcomes in modern populations. If socially buffered lifestyles—such as elite status, specialized craftsmanship, or other indoor occupations—were partly associated with cognitive specialization or assortative mating in complex societies, EA PGS may indirectly correlate with those social positions. For that reason, it can serve as a tentative proxy for testing whether pigmentation alleles were socially stratified.
After controlling for UVB, time, and ancestry, individuals with higher educational-attainment polygenic scores also tend to carry lighter skin pigmentation alleles. The effect remains when I add winter temperature. It remains when I model broad ancestry with ADMIXTURE components. And it even survives when I include archaeological population as a random intercept in a linear mixed model.
This pattern raises a provocative possibility: in later prehistoric and historic societies, lighter skin may have been socially stratified before it became population-wide.
To visualize the pattern, the figure below shows three views of the data. The top panel plots skin pigmentation scores through time, while the middle panel shows the corresponding trajectory for educational-attainment PGS. The bottom panel examines the relationship directly: skin pigmentation residualized on UVB, climate, ancestry, and population structure plotted against EA PGS. Even after removing those factors, individuals with higher EA scores tend to carry alleles associated with lighter pigmentation.
The baseline pattern
First, the standard evolutionary signal is clearly present.
In a baseline model with skin pigmentation PGS as the dependent variable, annual UVB is strongly positive, meaning that higher-UV regions are associated with darker pigmentation alleles. There is also a clear temporal trend: individuals deeper in the past have darker pigmentation scores. That is exactly what we would expect if depigmentation increased over the course of the Holocene.
Deep ancestry also plays a role. Steppe-related ancestry has the largest negative coefficient. This fits the known role of steppe expansions in reshaping European pigmentation.
The table below reports the main regression results. The expected variables behave as they should: UVB predicts darker pigmentation, older samples have darker scores, and deep ancestry components matter. But another coefficient appears repeatedly across the models: higher educational-attainment PGS is associated with lighter pigmentation alleles.
The unexpected result: EA predicts lighter skin
When EA PGS is added to the model, a significant negative coefficient emerges.
Even after controlling for UVB, winter temperature, time, and ancestry, the coefficient on EA remains strongly negative. In other words, individuals with higher EA PGS tend to have lighter pigmentation alleles.
This relationship is not an artifact of broad geographic or cultural contrasts. In the main regression, the EA coefficient is about -0.08/-0.09 and highly significant. In the mixed model with a random intercept for archaeological group, it remains around -0.075.
That means the relationship is not just driven by comparing northern Europeans to southern Europeans, or steppe-rich individuals to farmer-rich ones, or one archaeological culture to another. The association persists even after those broad differences are accounted for.
Temporal change in the association
When I estimate a model with an interaction between educational-attainment PGS and time, the interaction term is positive and strongly significant. Because older individuals have larger YearsBP values, this implies that the negative association between EA and lighter skin is weaker deeper in the past and becomes stronger closer to the present.
In other words, the EA–skin relationship intensifies over time.
To make this pattern easier to interpret archaeologically, I also estimated a model using a broad post-2800 BP indicator labeled IronAge. This groups together individuals dating to roughly 800 BCE and later, when social complexity, urbanization, hierarchy, and occupational specialization were already much more developed than in earlier prehistoric societies.
The interaction term between EA and this later-period indicator is negative and significant, indicating that the EA–skin relationship is stronger after about 2800 BP than before.
Taken together, the two approaches point in the same direction: the association between higher EA PGS and lighter pigmentation becomes stronger in later populations.
What about endogamy?
I also tested whether the EA–skin relationship is stronger in individuals with more intermediate-length runs of homozygosity, using hapROH-derived 8–20 cM segments. If social closure and endogamy were a major part of the mechanism, we might expect the EA–skin relationship to strengthen in more inbred or more socially bounded groups.
No such interaction was observed. The EA × hapROH term is essentially null.
So the current evidence does not support a model in which elite endogamy drove the whole pattern.
Results from the linear mixed-effects model
The most stringent test is the linear mixed-effects model. Here, Group is included as a random intercept, which means each archaeological population is allowed its own baseline pigmentation level. This is a tougher test than a pooled regression, because it asks whether the EA–skin association survives after accounting for population-level structure.
The EA coefficient remains negative and substantial. UVB is still positive. The time trend is still there. Winter temperature adds little.
If this result holds up, it suggests that the standard climate-only story is incomplete.
Conclusion
Lighter skin in western Eurasia was clearly favored in low-UV environments. The classic ecological explanation of selection for more efficient vitamin D synthesis at high latitudes is strongly supported by the data. Annual UVB radiation predicts pigmentation in the expected direction, and older individuals consistently show darker pigmentation scores.
However, climate alone does not fully explain the observed pattern. Even after controlling for UVB, temperature, time, ancestry, and population structure, individuals with higher educational-attainment polygenic scores tend to carry alleles associated with lighter pigmentation. This association survives multiple model specifications, including a mixed model with archaeological group structure.
These results do not demonstrate a direct social-class mechanism. But they do suggest that pigmentation cannot be reduced to UV alone. One possibility is that as societies became more stratified, lighter skin became associated with lifestyles that involved less direct exposure to sunlight—through indoor work, specialized occupations, clothing, or other forms of environmental buffering.
If that association emerged, selection on pigmentation may have been reinforced or socially sorted rather than driven purely by ecological pressures.
In that sense, pale skin in western Eurasia may not have been only an adaptation to northern environments. For a time, it may also have functioned as a marker of who was most shielded from the sun.
Overall, the emergence of lighter skin in western Eurasia likely reflects the combined influence of ecological, demographic, and social processes rather than a single climatic variable.