When Sunlight Shaped Our Genome: How Vitamin D Drove the Evolution of Light Skin and Milk Drinking

Of all the forces that have sculpted the human genome over the past ten millennia, one stands out for its sheer consistency: the need for vitamin D. As populations migrated to higher latitudes, left behind equatorial sunlight, and adopted agricultural diets poorer in this critical nutrient, their genomes were placed under relentless selective pressure. The result was not one adaptation but a suite of convergent solutions, lighter skin to let more ultraviolet radiation through, the ability to digest milk into adulthood to access a dietary vitamin D source, and metabolic adjustments to maximise what little sunlight was available.

What makes this story extraordinary is that ancient DNA now allows us to watch these adaptations unfold in real time. Thanks to datasets of thousands of ancient genomes spanning 14,000 years, we can track allele frequency changes generation by generation, identify exactly when selection intensified, and reveal surprises that no amount of modern population comparison could have predicted. In this article, we draw on seven recent studies to reconstruct how vitamin D shaped the European genome, and why the story extends far beyond Europe itself.

I. The Vitamin D Imperative

Vitamin D occupies a unique position among nutrients: it is the only one that humans can synthesise endogenously, using ultraviolet B radiation (UVB, wavelengths 290, 315 nm) to convert 7-dehydrocholesterol in the skin into previtamin D₃. This molecule then undergoes hydroxylation in the liver and kidneys to produce the active hormone, calcitriol (1,25-dihydroxyvitamin D), which regulates calcium absorption, bone mineralisation, immune function, and dozens of other physiological processes.

The critical variable is UVB availability. At the equator, UVB is intense year-round, and dark skin, rich in eumelanin, provides necessary protection against UV-induced folate degradation and DNA damage while still allowing adequate vitamin D synthesis. But as humans migrated northward after the Last Glacial Maximum, UVB intensity dropped sharply: at latitudes above 40°N, there is essentially no UVB capable of triggering vitamin D synthesis during winter months. At 55°N (roughly the latitude of Edinburgh or Copenhagen), this "vitamin D winter" extends from October through March, more than half the year.

The consequences of chronic vitamin D deficiency are severe. Rickets, the softening and deformation of growing bones, was the most visible manifestation, but vitamin D insufficiency also impairs calcium metabolism, compromises immune defence against tuberculosis and other infections, and increases maternal and infant mortality. In a pre-modern population at 55°N, an individual who could synthesise even marginally more vitamin D during the short summer months would have had a meaningful survival and reproductive advantage.

This is the selective pressure that reshaped the genome. But what ancient DNA now reveals is that the response was neither instantaneous nor simple. It unfolded over thousands of years, involved multiple independent genetic loci, and varied in intensity depending on cultural and dietary context.

II. The Slow March Toward Lighter Skin

Dark-skinned Europeans persisted into the Bronze Age

One of the most striking findings from ancient DNA research is how recently light skin pigmentation reached high frequency in Europe. The popular assumption that European hunter-gatherers were light-skinned is simply wrong. Mesolithic individuals from sites across Europe, from the famous "Cheddar Man" in Britain (dated to ~7100 BCE) to the "La Braña" hunter-gatherer from Spain (~5000 BCE), carried ancestral (dark) alleles at the key pigmentation loci SLC24A5 and SLC45A2, while often possessing derived (light) alleles for eye colour.^[6] These were, in all likelihood, dark-skinned, blue-eyed individuals, a combination that no longer exists in modern Europe.

The Neolithic diffusion of Anatolian farmers beginning around 7000 BCE introduced the first wave of lighter pigmentation alleles into Europe, as these farming populations had already undergone some degree of selection for lighter skin in the Near East. But even after this, pigmentation diversity in Europe remained remarkably high. A 2025 study by Perretti and colleagues, which applied a novel probabilistic method for pigmentation inference from low-coverage ancient DNA, concluded that many Europeans retained dark skin tones well into the Bronze and Iron Ages.^[6]

The landmark study by Allentoft et al. (2015), which sequenced 101 Bronze Age genomes from across Eurasia, confirmed that light skin pigmentation was already present at high frequency by the Bronze Age, but critically, they found that another vitamin D, related adaptation, lactose tolerance, was not yet common, pointing to a more recent onset of positive selection on that trait.^[2] This temporal mismatch between skin lightening and lactase persistence is a crucial clue to how natural selection operates: different adaptations to the same underlying pressure (vitamin D deficiency) were selected at different rates and different times.

The seven loci that matter most

In 2022, Mathieson and Terhorst developed a novel method for estimating time-varying selection coefficients from ancient DNA and applied it to a large dataset from Britain spanning the past 4,500 years. Their findings were remarkable: of the seven genomic regions with genome-wide significant evidence of selection, almost all were related to increased vitamin D or calcium levels.^[1] Furthermore, among 28 complex anthropometric and metabolic traits tested, skin pigmentation was the only one with significant evidence of polygenic selection.

The selection coefficient at DHCR7, the gene encoding 7-dehydrocholesterol reductase, which controls the availability of vitamin D's precursor molecule in the skin, tells a particularly compelling story. Mathieson and Terhorst found that the derived allele (tagged by rs7944926) was under weak or no selection during the earliest part of the period, but selection intensified dramatically around 3,000 years ago, with the coefficient reaching approximately 0.06, an exceptionally high value for a single locus. This drove the allele from roughly 20% to 60% frequency in just 3,000 years.^[1]

Why did selection intensify? One possibility is that the transition from pastoral to more sedentary agricultural lifestyles reduced dietary vitamin D intake (less game, less fish, more cereals), making endogenous synthesis even more critical. Another is that population density increases associated with the Bronze Age amplified the consequences of vitamin D deficiency, particularly through tuberculosis, against which vitamin D plays a documented protective role.

Convergent evolution, different paths to the same solution

A critical insight from the genetics of skin pigmentation is that light skin evolved independently in European and East Asian populations through different genetic mechanisms. While SLC24A5 and SLC45A2 are the primary loci driving skin lightening in Europeans, East Asian populations achieved lighter pigmentation through variants in different genes, including OCA2, MC1R, and others. This convergent evolution, arriving at a similar phenotype through different genetic paths, is strong evidence that the selective pressure was real and powerful, rather than the result of drift or sexual selection alone.

The DHCR7 ancestry signal at Sint-Truiden

A fascinating 2025 study by Beneker and colleagues provided striking evidence of how vitamin D, related selection can differ between ancestral groups living in the same location. At the medieval city of Sint-Truiden in Flanders (Belgium), the team analysed 338 ancient genomes spanning the 8th to 18th centuries and discovered that carriers of the DHCR7 rs7944926-G allele, associated with higher circulating levels of the vitamin D precursor 25(OH)D₃, had significantly higher proportions of Gaulish (rather than Germanic) ancestry. Meanwhile, carriers of red hair, causing MC1R alleles, which also increase vitamin D production efficiency, had higher Germanic ancestry.^[4]

The Faroe Islands provide another powerful case study. Hamid et al. (2025) analysed 40 whole genomes from Faroese individuals and identified positive selection signals at loci involved in vitamin D absorption and dietary fat processing, alongside signatures of increased diversity at the lactase persistence locus.^[5] The Faroe Islands sit at 62°N latitude, even further from the equator than most of Scandinavia, making them an extreme test case for vitamin D, related selection. The results confirmed that this selective pressure operated most intensely at the highest latitudes.

III. Lactase Persistence: The Other Vitamin D Adaptation

A textbook example, with a plot twist

Lactase persistence, the ability to digest the milk sugar lactose into adulthood, is the textbook example of recent natural selection in humans. In most mammals, including most humans, the enzyme lactase is downregulated after weaning, making adult milk consumption uncomfortable or impossible. But in populations with a long history of dairy herding, particularly in northern Europe, parts of Africa, and the Middle East, derived mutations upstream of the LCT gene maintain lactase expression throughout life.

The European variant, -13,910:C>T, is perhaps the strongest single-locus signal of positive selection in the entire human genome, with estimated selection coefficients of 0.04, 0.10. But ancient DNA has revealed a startling timeline: this allele was virtually absent in European populations before 4,000, 5,000 years ago. Allentoft et al. (2015) found that while light skin pigmentation was already common in Bronze Age Eurasia, lactase persistence was not, indicating a far more recent selective sweep than previously assumed.^[2]

The timing is crucial. Dairying began in Europe around 7,500 years ago, archaeological lipid residues in pottery confirm early Neolithic milk use across a wide swath of the continent. Yet the LP allele remained at very low frequencies for the first 3,000, 4,000 years of European dairy farming. Ségurel and Bon (2020), synthesising a large body of ancient DNA data, showed that the -13,910*T allele appeared at appreciable frequencies in Central Europe only around 4,000, 3,000 BP, and then spread rapidly across Eurasia in the following millennia.^[3]

This lag between dairying and LP selection poses a genuine puzzle. Why would a population farm dairy for millennia before the gene to digest it efficiently rose in frequency? Several hypotheses have been proposed: that fermented dairy products (cheese, yoghurt) were the primary products and did not require LP; that LP became advantageous only when combined with other dietary changes (the "famine hypothesis" suggests LP was selected during periodic crop failures when milk was the only available food); or that LP's benefits went beyond calories to include vitamin D and calcium from unfermented milk, nutrients that became increasingly critical as diets shifted toward cereals.

Beyond Europe: convergent evolution in Africa and the Middle East

Lactase persistence is not a uniquely European phenomenon. In East Africa, pastoralist populations like the Maasai and Tutsi carry LP alleles at high frequencies, but these are different mutations from the European -13,910*T variant. The African alleles (including -14,010*G>C and -13,915*T>G) arose independently, providing another striking example of convergent evolution driven by dairying culture. In the Arabian Peninsula, yet another variant predominates. Each of these represents an independent molecular solution to the same selective pressure.

What makes this convergent evolution so powerful as evidence for selection is that all LP-associated alleles cluster in the same regulatory region upstream of LCT, and all achieve the same functional outcome: maintenance of lactase expression in adulthood. The probability of this happening by chance in multiple independent populations is vanishingly small.

Central Asia: herders who never needed the gene

Central Asian pastoralists present a fascinating counterpoint. As Ségurel and Bon (2020) documented, populations in Kazakhstan and Mongolia have practised intensive dairy pastoralism for thousands of years, yet LP frequencies remain low (~30% or less). How did they manage? The answer lies in cultural adaptation: these populations process milk into fermented products (kumis from mare's milk, various cheeses and curds) that contain significantly less lactose than fresh milk. They also found that the low LP prevalence in modern Central Asia may partly result from the replacement of early LP-carrying populations by eastern neo-pastoralists who were lactase non-persistent.^[3]

The Neanderthal twist: a different selection story in East Asia

In March 2025, Ma and colleagues published a study in PNAS that sent shockwaves through the field. They demonstrated that approximately 25% of East Asian individuals carry a haplotype at the LCT locus that is absent from European and African populations, and that this haplotype derives from Neanderthals.^[8]

The Neanderthal-derived haplotype spans roughly 467 kilobases at the 2q21.3 locus and has been under positive selection in East Asian populations. It is associated with altered LCT expression and promoter methylation in certain cell types, raising the possibility that it could contribute to lactase persistence in some carriers. The frequency of this haplotype (~20, 25%) is intriguingly close to the estimated LP prevalence in East Asian populations.

The key evidence against a milk-related function is temporal: this haplotype was already present in early East Asian populations before the domestication of dairy animals, ruling out milk consumption as the original selective driver. Ma et al. found that the Neanderthal haplotype alters expression of genes in immune cells, particularly affecting neutrophil and white blood cell counts. They propose that it was selected because it enhanced pathogen resistance in East Asian hunter-gatherers who inhabited Eurasia for tens of thousands of years before agriculture.^[8]

This finding has profound implications for how we understand selection at the LCT locus globally. If selection at 2q21.3 in East Asia was driven by immune function rather than LP, it raises the possibility that even the European and African selective sweeps at this locus were not solely about milk digestion. The "textbook example" of gene-culture coevolution may be more complicated than we thought, and the involvement of Neanderthal introgression adds yet another layer to an already complex story.

South Asia: the world's largest dairy producer, and a new aDNA frontier

The most recent addition to the LP story comes from the 2025 preprint by Hamid, Mortensen, and colleagues on lactase persistence in South Asia, the world's largest producer and consumer of dairy products. Assembling genome-wide data from approximately 8,000 present-day and ancient genomes from India, Pakistan, and Bangladesh, spanning from ~3300 BCE to 1650 CE, the study found that the European -13,910*T variant is widespread across South Asia but exhibits clinal variation along both north-south and east-west gradients, consistent with its introduction through steppe-related (Indo-European) gene flow from the northwest.^[9]

This South Asian story connects directly to the Bronze Age Eurasian population dynamics revealed by Allentoft et al. (2015): the Yamnaya-related steppe pastoralists who expanded into both Europe and South Asia during the 3rd millennium BCE likely carried early LP alleles with them. The subsequent rise in LP frequency in both regions then reflects parallel selective sweeps on the same variant in different geographic contexts, yet another instance of how a single genetic adaptation was spread and amplified by the interplay of migration and natural selection.

IV. Synthesis: Vitamin D as the Master Driver

The evidence assembled here converges on a single, powerful conclusion: the need for vitamin D was the dominant selective pressure shaping European genomes over the past 10,000 years. It operated through multiple independent pathways simultaneously, skin pigmentation genes that increased UVB penetration, metabolic genes that optimised vitamin D synthesis and transport, and dietary genes that unlocked a new nutritional source of vitamin D through fresh milk consumption.

The 2024 study by Akbari et al. dramatically expanded this picture, identifying 347 independent loci with >99% probability of directional selection in West Eurasians over the past 14,000 years, an order of magnitude more signals than any previous scan. Many of these relate to immune function, metabolism, and diet, but the strongest and most consistent clusters revolve around vitamin D and calcium homeostasis.^[10]

What is perhaps most remarkable about this story is its recency. The selective pressures we have described did not operate over deep evolutionary time. They have been shaping European genomes within the past few hundred generations, an eyeblink in evolutionary terms. The light skin, the ability to drink milk, the metabolic optimisations for vitamin D, all of these are recent adaptations, still incomplete in many populations, still varying across geography. When we look at a modern European face, we are looking at the product of a selective process that is, in evolutionary terms, still underway.