Europeans Didn’t Evolve as One Population
Ancient DNA shows that ancestry influences traits, but evolutionary trajectories reshape them
Human evolution can move much faster than most people assume.
Ancient DNA findings still surprise people. When results show that prehistoric Europeans had darker skin than many Europeans today, someone inevitably asks: Why? Put bluntly: If modern Europeans look white, why didn’t their ancestors?
I’ve written before that the evolution of skin pigmentation, hair and eye colour in Europe over the last ~12,000 years was significant. Alleles associated with lighter features rose dramatically in frequency. That debate is largely settled.
But there is a subtler yet more persistent source of confusion that still shapes how people think about ancestry and traits.
Many assume that populations differ today because they inherited fixed trait profiles from ancestral groups. For example:
Northern Europeans are taller because they have more Steppe ancestry.
Middle Eastern ancestry implies darker pigmentation.
Hunter-gatherer ancestry implies shorter stature.
This assumes ancestral populations were fixed trait packages that just mixed but this is not how evolution works.
Lineages mix and evolve, and they do so at different speeds and in different directions.
I called one manifestation of this the Aryan paradox: many of the populations in Europe with the lightest skin and hair today derive a large fraction of their ancestry from groups like the Yamnaya, who likely had light-brown skin and dark hair. If ancestry alone determined traits, this shouldn’t happen. But traits changed within those lineages after they spread.
The real question is whether Europe’s ancestral components followed distinct evolutionary trajectories, not just different starting points but different directions and rates of change.
The answer turns out to be yes, and not in the way most people expect.
Methods
I analyzed ancient genomes from the Allen Ancient DNA Resource (AADR), restricting the sample to individuals within Europe dated between about 200 and 45,000 years before present and excluding modern individuals so the patterns reflect prehistoric processes rather than modern sampling.
For each individual I computed polygenic scores for educational attainment (averaging EA3 and EA4), cortical surface area, brain volume, height, and skin pigmentation. All scores were standardized across the dataset.
Ancient individuals were also assigned ancestry proportions from ADMIXTURE projections. Rather than treating ancestry as a static label, I modeled how polygenic scores changed through time as a function of ancestry background.
For each trait I estimated:
PGS ~ YearsBP × ancestry + coverage
YearsBP is measured in thousands of years and centered. Coverage is included as a technical control because sequencing depth affects PGS noise. The key quantity is the interaction between time and ancestry, which captures whether polygenic scores changed at different rates across ancestry backgrounds.
To make the results interpretable, I generated predicted evolutionary trajectories for populations dominated by a given ancestry while allowing small contributions from others. From these models I extracted ancestry-specific time slopes, which measure how quickly polygenic scores change per thousand years since that ancestry enters the record. Each slope corresponds to the time coefficient associated with that ancestry in the ancestry×time interaction, evaluated under the ancestry scenarios.
Europe’s 4+1 Deep Ancestries
Before looking at how traits changed within ancestries, it helps to see what those ancestries actually look like in the ancient record. These components are not arbitrary statistical constructs.; they correspond to recurring ancestry profiles that appear across archaeological populations. In earlier posts I emphasized four deep lineages shaping Europe: Anatolian farmer–related ancestry, Western hunter-gatherer ancestry, Steppe pastoralist ancestry, and an East Mediterranean lineage related to Levantine and Iranian populations.
In this dataset a fifth component also appears clearly: an Inner Asian ancestry associated mainly with Avar-period and some Hungarian samples. This doesn’t replace the four-lineage framework — outside of those populations the 4-way structure still describes variation well — but including the Inner Asian component prevents forcing those samples into the wrong ancestry bins and lets the models account for their distinct genetic background explicitly.
The figure below shows the average ancestry composition of ancient population groups with at least twenty individuals using this 4+1 framework. Population groups are ordered by the proportion of Steppe ancestry, highlighting how Steppe ancestry spreads across time and regions.
Figure 1. Mean ancestry composition of ancient European population groups (K=5 ADMIXTURE, Europe’s “4+1” ancestries).
The five ancestry components show clear geographic and temporal structure rather than being evenly distributed across European groups. Anatolian farmer ancestry is highest in Neolithic farming populations such as Austria_LBK (≈0.83), Germany_EN_LBK (≈0.81), Slovakia_N_LBK (≈0.80), and France_N (≈0.79), while it is much lower in hunter-gatherer groups like Serbia_IronGates_Mesolithic (≈0.10). Steppe ancestry peaks in Corded Ware and later northern European groups, reaching about 0.50 in Czechia_EBA_CordedWare and around 0.43 in Viking and Anglo-Saxon contexts, while remaining minimal in early farming populations. WHG ancestry is highest in Mesolithic hunter-gatherers such as Serbia_IronGates_Mesolithic (≈0.72) and remains elevated in northern European Neolithic contexts, but is low in Near Eastern–related populations. East Mediterranean ancestry is highest in Anatolia and Levant-proximal contexts such as Turkey_Hatay_Alalakh (≈0.41) and Turkey_Aegean_Mugla (≈0.34) and much lower across most central and northern European groups. Inner Asian ancestry is negligible across most European populations but rises sharply in Avar-period Hungary (≈0.35–0.47), where it becomes the dominant component.
Steppe ancestry characterizes Bronze and later northern Europeans, Anatolian ancestry dominates early farmers, WHG marks Mesolithic populations, East Mediterranean reflects southeastern Near Eastern input, and Inner Asian ancestry is localized to specific migration events.
Once ancestry and evolutionary change are disentangled, the picture shifts. These lineages did not simply preserve ancient trait profiles but they moved in different directions.
The full breakdown of trajectories, effect sizes, and what this implies for European prehistory is below.
Europe Did Not Evolve Along a Single Path
If ancestry components were static packages of traits, evolutionary change after their arrival would look broadly similar across ancestry backgrounds. Instead, once time is aligned to ancestry entry, the trajectories diverge sharply. Different ancestries show different directions and rates of change for the same traits, meaning Europe’s evolutionary history cannot be summarized by a single continental trend.
To show this directly, I align time to ancestry entry. For each ancestry-dominant scenario (≈80%), I compute predicted polygenic scores at the point when that ancestry becomes substantial in the record and at the youngest ancient samples. The lines connecting these points therefore represent lineage-specific evolutionary change rather than differences inherited from deep time.
Once time is aligned to ancestry entry, ancestry backgrounds exhibit systematically different directions and magnitudes of change rather than parallel trends.
Figure 2. Predicted entry-to-late trajectories by ancestry lineage.
Educational attainment (EA)
All ancestry backgrounds show increases in EA polygenic scores after entry, but the magnitude differs.
Anatolian ancestry increases from −0.292 to 0.291 (+0.583; slope 0.0857 SD per kyr). East Mediterranean ancestry shows the largest total increase (−1.29 to −0.518; +0.772; slope 0.0778). Western hunter-gatherer ancestry increases more moderately (−1.41 to −0.930; +0.484; slope 0.0355), while Steppe ancestry shows the smallest EA change (−0.193 to −0.051; +0.142; slope 0.0236).
Thus, EA increases across ancestries, but most strongly for Anatolian and East Mediterranean ancestry and least for Steppe ancestry.
Cortical surface area
Cortical surface trajectories diverge strongly by ancestry.
Steppe ancestry increases markedly (0.089 → 0.779; +0.690; slope 0.110). Anatolian ancestry increases moderately (−0.376 → −0.037; +0.340; slope 0.055). Western hunter-gatherer ancestry is nearly flat overall (0.351 → 0.212; −0.139 total). In contrast, East Mediterranean ancestry declines sharply (0.791 → −0.919; −1.71; slope −0.150).
This is the clearest example of ancestry-specific directional divergence.
Brain volume
Brain volume shows the same structure as cortical surface.
Steppe ancestry increases strongly (0.364 → 1.10; +0.731; slope 0.124). Anatolian ancestry increases moderately (−0.219 → 0.078; +0.296; slope 0.056). Western hunter-gatherer ancestry increases slightly (+0.227; slope 0.0369). East Mediterranean ancestry declines substantially (0.795 → −0.389; −1.18; slope −0.090).
Thus, Steppe and Anatolian ancestry show increases in brain proxies after entry, whereas East Mediterranean ancestry shows decreases.
Height
Height displays the strongest ancestry differentiation.
Steppe ancestry increases dramatically (0.550 → 1.82; +1.27; slope 0.190). Anatolian ancestry increases modestly (−0.844 → −0.541; +0.303; slope 0.042). Western hunter-gatherer ancestry declines (0.273 → −0.612; −0.885; slope −0.046). East Mediterranean ancestry declines even more strongly (0.482 → −0.998; −1.48; slope −0.137).
Height therefore shows opposite post-entry trends across ancestries, with Steppe increasing while East Mediterranean and WHG decrease.
Skin pigmentation
Skin pigmentation shows a different pattern. All ancestries shift toward lower pigmentation scores (lighter pigmentation), but at different rates.
Steppe ancestry shows the largest decline (0.252 → −0.567; −0.820; slope −0.117). Western hunter-gatherer ancestry also declines strongly (1.38 → 0.306; −1.07; slope −0.0525). East Mediterranean ancestry declines modestly (1.44 → 1.10; −0.345; slope −0.0248), and Anatolian ancestry changes little (0.57 → 0.48; −0.09).
Unlike height and brain traits, pigmentation change is largely shared across ancestries rather than directionally opposed.
Different lineages, different directions
The clearest pattern is not just that ancestries change differently, but that they begin and end at different positions across traits.
For educational attainment, Anatolian and Steppe ancestry begin relatively higher than WHG and East Mediterranean ancestry, but Anatolian ancestry shows the strongest increase over time. WHG ancestry starts lowest and remains lowest even after increasing, while Steppe ancestry changes comparatively little. The result is that ancestry differences persist but are reshuffled by lineage-specific change rather than erased.
Brain structure proxies show an even stronger divergence. Steppe ancestry begins moderately high and increases sharply, Anatolian ancestry increases modestly, WHG changes little, while East Mediterranean ancestry begins relatively high but declines substantially, ending among the lowest. This produces a clear split between Steppe/Anatolian increases and East Mediterranean decreases.
Height shows the most dramatic reversal of expectations. Steppe ancestry starts high and increases further, while WHG and East Mediterranean ancestry both decline strongly despite the broader long-run tendency toward increasing stature in Eurasian ancient DNA. Anatolian ancestry increases only slightly. Thus, height evolution is not a uniform upward trend but depends strongly on ancestry background.
Skin pigmentation provides the clearest example of within-lineage change overturning ancestry expectations. Steppe ancestry shows the largest decline in pigmentation scores, meaning strong movement toward lighter pigmentation after entry. This aligns with what I called the Aryan paradox: many present-day Europeans with very light pigmentation derive substantial ancestry from Steppe populations that were likely darker. The paradox disappears once traits are allowed to evolve within lineages rather than being fixed properties of ancestry.
Across traits, East Mediterranean ancestry is the most consistently divergent lineage. It shows large declines in height and brain proxies despite increases in educational attainment, often moving in the opposite direction from Steppe and Anatolian ancestry. WHG ancestry tends to remain lowest or decline where Steppe increases, particularly for height.
Ancestry influences traits, but evolutionary change keeps reshaping them.
Very interesting, thanks.