Dark Skin Evolution in Early Humans: Revisiting the Skin Cancer Hypothesis Through Migration-Related Mismatch.

Introduction
Skin cancer is one of the most common malignancies in humans today, and yet the use and appropriation of evolutionary reasoning to understand why this is so has only gained more attention in dermatology more recently (Jablonski 2021). By contrast, researchers in other fields of biology have long embraced evolutionary, ecological, and principles of comparative biology to better understand, predict, and possibly treat cancers (Greaves 2000a; Maley and Greaves 2016; Ujvari et al. 2017; Aktipis 2020; Vincze et al. 2025).
Two key concepts from the field of cancer evolution and ecology are cancer selection and evolved tumor suppression. Cancer selection refers to the idea that, rather than being selectively neutral, cancer can act as a force of selection through deleterious effects on host fitness, favoring the evolution of some traits (Leroi et al. 2003). Evolved tumor suppression refers to the idea that some of these traits have evolved under cancer selection, namely, mechanisms, behaviors, and changes to life history strategies that avoid, prevent, or mitigate the selective pressure exerted by cancer, sometimes also referred to as “anti-cancer adaptations” (DeGregori 2011; Boutry et al. 2020).
Cancer selection is held to have played a major role in shaping the evolution of complex multicellular organisms, roughly appearing 800 to 600 million years ago (mya). These newly emergent organisms had to evolve mechanisms to suppress, detect, and police selfish cancer cells from gaining a fitness advantage at the cellular level at the expense of the organism's cooperative integrity (Aktipis et al. 2015). Today, most species across the tree of life are susceptible to cancer to varying degrees (Aktipis et al. 2015; Albuquerque et al. 2018; Vincze et al. 2022). This variation in cancer susceptibility is thought to reflect differences in species' investment in evolved tumor-suppressive mechanisms, as exemplified by what is known as Peto's paradox: the observation that larger animals, despite having more cells (i.e. more chances of tumorigenesis) than smaller animals, do not have correspondingly higher rates of cancer (Caulin and Maley 2011; Vincze et al. 2022).
Cancer selection has influenced ecosystem dynamics in wildlife populations. One of the most striking cases is Tasmanian devils. Since the 1990s, these populations have faced the threat of extinction due to early-life mortality from a transmissible cancer, called the devil facial tumor disease (DFTD) (McCallum et al. 2009). In response to this selective pressure, field observations suggest that Tasmanian devils have undergone a marked shift toward a faster life history strategy, characterized by earlier sexual maturity and semelparous reproduction in a manner of couple of decades (Jones et al. 2008). Tasmanian devils have also shown genomic adaptations associated with immune-modulated resistance to DFTD, indicating that the cancer selection also is driving evolutionary changes in immune function (Epstein et al. 2016). Similar life history responses to cancer have been observed in experimental evolution studies on Drosophila melanogaster and in freshwater Hydra oligactis (Arnal et al. 2017; Boutry et al. 2022).
In a perspective paper by 33 leading experts in the field of cancer evolution and oncology, it was suggested that “just as with other animals, the evolution of humans was likely shaped by cancer” (Dujon et al. 2021: 879, 885). This claim can be interpreted in two ways: either it is trivially true, as human evolution is part of the broader metazoan lineage, or else it implies that cancer specifically acted as a selective pressure in human evolution, somewhere after the hominin-panin split, roughly occurring 10 to 5 mya (see Box 1).
Box 1 A primer on early human skin evolutionThe lineages of humans and the lineages of chimpanzees and bonobos, collectively referred to as “hominins,” share a last common ancestor (LCA) from which they diverged approximately 7 to 10 mya, likely in East Africa (Scally et al. 2012; Almécija et al. 2021). Phylogenetic analysis suggests that the ancestral condition of skin of LCA probably was pale, covered with dark body-wide hair, mainly had apocrine sweat glands and was capable of developing melanin pigment on hairless areas (e.g. hands and face) in response to occasional sun exposure (Jablonski 2013). Climate-driven changes gradually reshaped the habitats of early hominins, shifting them from dense tropical forests to predominantly more open and arid mosaic environments, often termed “savanna” (deMenocal 2004 ; Cerling et al. 2011). Early versions of the savanna hypothesis assumed largely treeless, grass-dominated landscape, which has since then been discredited. More recent definitions, informed by paleoecological evidence, describe the savanna as a mosaic of seasonal open and wooded habitats, including grasslands, scattered trees, and patches of denser woodland or forest (Domínguez-Rodrigo 2014). This transition imposed novel selection pressures, including increased exposure to solar radiation, heat, and seasonality and led to a series of adaptations in the skin and integumentary system of ancestral humans. Among these changes was a reduction in body-wide hair coverage, inferred to have unfolded roughly between 3 and 1 mya based on molecular evidence from lice divergence (Reed et al. 2007) and variation at the MC1R locus (Rogers et al. 2004). Concomitant with hair loss were changes in sebaceous gland distribution and secretion, a wider dispersal of eccrine sweat glands, and a redistribution of melanocytes within the interfollicular epidermis (Zihlman and Cohn 1988; Elias et al. 2009; Jablonski 2021). The fitness benefit of reduced body hair and enhanced eccrine sweating was likely improved thermoregulation, allowing more efficient heat dissipation and reducing the risk of hyperthermia during prolonged activity on the savanna while foraging and hunting prey (Wheeler 1991; Ruxton and Wilkinson 2011; Lieberman 2014). Population genetic analysis suggests that the gene encoding the melanocortin 1 receptor (MC1R) swept to fixation around 1.2 mya in ancestral humans, favoring variants that primarily produced dark (eumelanin) pigmentation (Rogers et al. 2004). This not only provides a timeframe for onset of darker skin pigmentation in ancestral humans but also sets an upper boundary for the migration of melanocytes to the interfollicular epidermis. Dark skin pigmentation most likely evolved as an adaptation to mitigate the harmful effects of UVR from prolonged sun exposure (Jablonski and Chaplin 2010; Elias and Williams 2013; Greaves 2014a), possibly in combination with other environmental pressures, such as desiccation stress from the savanna (Elias et al. 2009).
Cancer selection in human evolution remains sparsely explored with one exception being human menopause. Originally, Williams (1957: 408) proposed that menopause may be a reproductive adaptation to a life cycle already marked by senescence, higher mortality risks during pregnancy and childbirth, and an extended period of juvenile dependence. More recently, Thomas et al. (2019) proposed that an overlooked factor here may have been the balance between cancer risk and cancer defense. They argue that the fitness benefits of early reproductive cessation may have evolved to reduce the risk of exacerbating pre-cancerous lesions in late pregnancy, based on evidence that women kept in a premenopausal state (via hormone replacement therapy) have increased risk of hormone-sensitive cancers compared to those who undergo menopause (Thomas et al. 2019).
Here, we want to draw attention to another possible case of cancer shaping human evolution: the evolution of darker skin pigmentation in early hominins as an adaptation to increased UVR exposure, following climate-driven changes to their ancestral environment, roughly occurring 2 to 1 mya (see Box 1). The selective pressure exerted by UVR-induced skin cancers in this context—sometimes referred to as the skin cancer or “genotoxic hypothesis” (Elias and Williams 2013)—has been defended using cases of evolutionary mismatch, in which individuals with albinism and fair-skinned migrants are exposed to high-UVR environments (Greaves 2014a ,b; Osborne and Hames 2014).
In this paper, we begin by providing an overview on skin cancer, UVR, and melanin biology. We then situate the skin cancer hypothesis in the context of ancestral human dispersal onto the savanna and the use of modern evolutionary mismatch scenarios to infer putative selection pressures exerted by skin cancers, finding that migration-related mismatch largely remains largely unexamined. Taking a charitable case of migration-related mismatch, we test data on skin cancer rates in modern Australians and find that, although skin cancers with metastatic potential are common, their onset and related mortality typically occur after reproductive age, rendering the inference for cancer selection weak. We also discuss several ad hoc explanations to preserve the skin cancer hypothesis, along with criticisms concerning the selective pressure of melanoma, inferencing life history traits from extinct populations, and inclusive fitness models and the role of the grandmothers .
We conclude that in ancestral humans, the protective effects of melanin pigmentation were most likely selected for preventing more acute UV- and thermoregulation-related pathologies associated with life on the savanna mosaics than selected as an anti-cancer adaptation. However, lethal skin cancers may have influenced human evolution indirectly if post-reproductive elders played a key role in ancestral social structures by providing alloparental care to their kin.

Skin Cancer, UVR, and Melanin Biology
Skin cancer is the most common form of cancer in humans, comprising three main types: basal cell carcinoma (BCC), squamous cell carcinoma (SCC)—together referred to as keratinocyte cancers—and cutaneous melanoma (CM), derived from the melanocyte lineage. The predominant cause of skin mutagenesis is solar UVR, a fact well-supported by epidemiological and genetic studies, as well as evidence from animal models (Garbe et al. 2024). SCC development is mainly associated with chronic and cumulative sun exposure over decades, whereas CM are linked to intermittent, high-intensity sun exposures, particularly episodes of sunburn during childhood and adolescence. BCC development involves a mixture of both cumulative sun exposure and intermittent sunburn early in life, depending on site, genetic predisposition, and skin phenotype, among other factors.
Solar radiation comprises approximately 50% infrared, 40% visible, and 5% ultraviolet light. Ultraviolet light is categorized by wavelength into UVC (100 to 280 nm), UVB (280 to 315 nm), and UVA (315 to 400 nm). While UVC is effectively absorbed by the ozone layer (see below), UVB and UVA reach the Earth's surface and penetrate the skin to varying depths. UVA reaches both the epidermal and dermal layer and mainly causes indirect DNA damage through the generation of reactive oxygen species (ROS). UVB, though partially filtered by the ozone layer, mainly affects the epidermis and can directly induce DNA lesions such as cyclobutane pyrimidine dimers (CPDs) and pyrimidine(6–4)pyrimidone photoproducts (6–4PPs). These lesions can cause errors in cell replication and repair, leading to characteristic C>T and CC>TT transitions, also known as UV signature mutations, that frequently affect the TP53 tumor suppressor, and other oncogenes, contributing to the development of BCC and SCC (Nakazawa et al. 1994; Pfeifer et al. 2005).
The amount and intensity of solar UVR transmitted to the Earth's surface with the potential to drive evolutionary change in human skin is influenced by a combination of factors operating at planetary, atmospheric, and geographical levels. At the planetary level, the Earth's distance from the Sun and its axial tilt (approximately 23.5 degrees relative to its orbital plane) are fundamental constants underlying seasonal and latitudinal variation in solar UVR exposure. At the atmospheric level, several factors beyond ozone absorption influence UVR transmission, including the scattering properties of clouds, air mass, and various aerosols (e.g. dust, pollutants, and fog). At the geographical level, key factors include altitude, latitude, and surface reflectivity. UVR intensity tends to be higher at high altitudes due to thinner atmospheric layers and lower at low altitudes where denser air absorbs and scatters more radiation. Similarly, latitude influences the angle of solar incidence, with low-latitude regions receiving more direct UVR and high-latitude regions receiving less on an annual average basis (Jablonski and Chaplin 2000). Moreover, reflective surfaces such as snow, sand, and water can enhance local UVR exposure by bouncing radiation back toward the body, also known as reflective UVR.
Human skin phenotypes are determined largely by melanin content, types, and its distribution and provide another mediating layer through which solar UVR (primarily UVB) must pass to cause DNA damage (Brenner and Hearing 2008). Generally, melanin is a pigment synthesized by melanocytes in the basal layer of the epidermis. Several protective roles of melanin against UV-induced damage have been identified, including the absorption and scattering of UVR, induction of melanin-mediated apoptosis, and antioxidant activities with ROS scavenging (Brenner and Hearing 2008; Yamaguchi et al. 2008). These properties shift the UVB dose–response curve from a deleterious toward a beneficial range, while also stimulating water barrier function and antimicrobial peptide production (Gunathilake et al. 2009; Elias et al. 2010), extending the role of pigmentation beyond photoprotection. Importantly, though, these effects occur to different degrees depending on melanin type.
Melanocytes produce two main types of pigment: dark brown eumelanin and red yellowish pheomelanin. Both are present in the human epidermis in varying proportions (Thody et al. 1991), reflecting different skin phenotypes and often classified clinically using the Fitzpatrick scale (see Table 1). Most of the photoprotective effects of melanin are attributed to eumelanin, while pheomelanin is considered photosensitive and a carcinogenic factor in the formation of CM (Brenner and Hearing 2008; Morgan et al. 2013). Moreover, eumelanin and pheomelanin melanosomes are distributed differently within the epidermis. Eumelanin tends to accumulate in larger, individually dispersed melanosomes that are more broadly distributed, including in the upper epidermal layers. In contrast, pheomelanin is packaged into smaller, aggregated melanosomes that are typically confined to the basal and immediate suprabasal layers of the epidermis. In basal keratinocytes, both eumelanin- and pheomelanin-containing melanosomes can form supranuclear caps, which help to filter UV radiation (Kobayashi et al. 1998).
Human skin phenotypes can be classified using both quantitative and qualitative methods, including reflectance spectrophotometry and clinical tools such as the Fitzpatrick scale. The Fitzpatrick scale, based on “sun-reactive skin typing” categorizes skin types according to its response to UVR, particularly in terms of burning and tanning behavior (Fitzpatrick 1988). It ranges from Type I to Type VI (see Table 1), with higher types correlating with a greater concentration of eumelanin-containing melanosomes (Thody et al. 1991).
Human skin pigmentation varies strongly with latitude and UVR levels and is today widely accepted as an adaptation shaped by natural selection to balance the competing demands of UV-protection and vitamin D synthesis (Jablonski and Chaplin 2010; Lucock et al. 2022). In other words, lighter skin phenotypes evolved in high-latitude environments with lower UVR to enhance sufficient cutaneous vitamin D production, while darker skin phenotypes were selected in equatorial regions with high, year-round UVR to protect against UV-related damage to the skin and integumentary system. Exceptions to this pattern between latitude and pigmentation include northern-dwelling Inuit, Eurasians, and highland Tibetans, all of whom retain darker skin than expected for their latitude, likely related to vitamin D–rich diets, reflective UVR from snow, and intense high-altitude UVR (Deng and Xu 2017; Yang et al. 2022). These exceptions illustrate the complex interplay of genetic, environmental, and cultural factors that shape pigmentation. The question is whether skin cancer may have mediated part of the selective pressures exerted by UVR, contributing to the development of dark skin pigmentation in human evolution? While often overlooked, one particular historical and ecological context has been proposed in which this may have occurred, namely, as early hominins dispersed onto the savanna (see Box 1).

The Skin Cancer Hypothesis
The skin cancer hypothesis posits that UVR-induced skin cancer and resultant metastasis had a detrimental impact on hominin fitness, contributing in part to the evolution of darker skin pigmentation as a protective response. This hypothesis is an adaptive explanation for why early ancestral humans, characterized by recent pelage loss, fair skin, and a lack of clothing, developed darkly skin while living on the savannas of East Africa approx. 2 to 1 mya. To provide additional context for the skin cancer hypothesis, we include a primer on human skin evolution (see Box 1) and with a timeline of some key cutaneous events in human evolution discussed in this paper (see Fig. 1).
The role and significance of skin cancers in the context of savanna-dwelling hominins has been debated among dermatologists, photobiologists, and anthropologists since at least the 1960s (Blum 1961; Daniels et al. 1972). A contributing factor to this debate is that several other adaptive explanations based on solar UV light, as well as other selective pressures, have been proposed to account for dark skin in ancestral humans (see Table 2 and Supplementary material for further details). Given the range of factors that may have affected the survival and reproductive success of ancestral humans, these hypotheses have sometimes been framed as being in competition. Importantly, though, many of these explanations are not necessarily mutually exclusive and may have operated in concert. Moreover, some have been discredited today (e.g. the vitamin D hypothesis for eumelanin), while others are regarded as speculative because they draw on ecological contexts other than the savanna mosaic habitats (e.g. the threat display and camouflage hypotheses) and lack empirical support (see Supplementary material for details).
The skin of ancestral humans, including its composition and type of pigmentation degree relative to the Fitzpatrick scale, is almost entirely absent from the fossil record, as soft tissues usually decay quickly after death. Thus, evaluating the putative selective pressure of skin cancers—like any other hypothesis concerning epidermal pigmentation in early humans—must rely on inference to the best explanation. This involves contextual interpretation and integration of diverse lines of evidence, including comparative analysis, ethnographic/anthropological observations, paleoecology, medical sciences (i.e. dermatology) and evolutionary theory to build a plausible model of past human behavior and adaptation. Box 1, while obviously not a complete natural history of human skin evolution, illustrates the complexity of this task.
An important question remains: How can we design a situation capable of testing the skin cancer hypothesis and generate evidence that allows us to infer whether skin cancer was among the UV-related pathologies that exerted selective pressure for dark skin in ancestral humans? Addressing this question is crucial, as it goes to the heart of how to evaluate adaptive explanations in the absence of direct evidence. To proceed, we turn to Darwinian medicine and the concept of evolutionary mismatch as a conceptual tool to bridge past selective environments with present-day observations.

Darwinian Medicine and Evolutionary Mismatch
Evolutionary mismatch refers to situations in which the pace or scale of environmental change exceeds an organism's capacity to adapt, resulting in maladaptation, also known as “adaptive lag” (Williams and Nesse 1991; Bourrat and Griffiths 2021). The concept most often is used to explain why certain diseases in humans are common by being caused, or exacerbated, by mismatch between contemporary environments and genotypes selected for ancestral environments—for example, the increased availability of calorie-dense foods and reduced pathogen exposure have been linked to obesity, type 2 diabetes, allergies, and autoimmune disorders (Brüne and Schiefenhövel 2019). Beyond explaining “diseases of modernity,” the evolutionary mismatch framework can also be used as a tool to generate indirect evidence for selective pressures by assessing whether the maladaptations they produce also impacts the survival or reproductive success in affected individuals. The skin cancer hypothesis relies on this approach and has been defended using two distinct scenarios: individuals with albinism living in high-UVR environments (Greaves 2014a,b) and fair-skinned migrants who have settled in high-UVR environments (Osborne and Hames 2014).

Evolutionary Mismatch Models: Albinos and Migrants
Oculocutaneous albinism (OCA) refers to a group of autosomal recessive disorders characterized by absent or reduced melanin production due to mutations in genes encoding proteins that are involved in the melanin synthesis pathway, such as TYR or OCA2 (Ma et al. 2023). Individuals with albinism have generalized hypopigmentation of the skin, hair, and eyes, limited tanning capacity, and therefore predisposed to UV damage, sunburns, and risk of skin cancers, particularly SCCs (Juan et al. 2023). Although OCA affects individuals globally, it is more common in certain African populations (Ma et al. 2023), particularly those residing in high-UV regions. While OCA is caused by mutations in melanin-related genes, its effects on individuals living in high-UV environments represent a model of evolutionary mismatch, where their skin phenotype is poorly suited to its environment, relative to individuals in the same region, who have darker skin pigmentation. Other inherited skin diseases with heightened sun sensitivity, such as xeroderma pigmentosum (XP), may likewise serve as models for testing skin cancer's impact on survival and reproductive success.
Considering the mismatch scenario involving albinism: Does skin cancer affect survival and reproductive potential such that we can infer a similar selective pressure in early savanna-dwelling hominins? Greaves (2014a) synthesized clinical reports on the prevalence, onset, causes, and mortality patterns in individuals with OCA2, a subtype of albinism common in African populations. He found that SCCs and their precursor, actinic keratoses, often develop in sun-exposed areas during adolescence and are a leading cause of premature mortality during the reproductive years. Based on this, Greaves (2014a) argues that skin cancers may have mediated part of the selective force of UVR, driving the evolution of darker pigmentation in ancestral humans. Other clinical reports confirm that African albinos without adequate photoprotection tend to die in the third and fourth decade of life (Mabula et al. 2012; Kiprono et al. 2014; Nakkazi 2019). Greaves (2014a, 7) notes that XP patients show even more marked shifts in skin cancer onset. In XP patients, median age of diagnosis is 7 and 22 years for NMSC and MC, respectively, with death at a median of 32 years, primarily from skin cancer (Bradford et al. 2011). Similar patterns of morbidity and mortality are observed in larger XP cohorts and for high-UV environments (Kraemer et al. 1987; Ben Rekaya et al. 2009).
The mismatch argument based on albinism has been criticized for relying on an inappropriate model to infer the selective forces acting on early hominin skin. Generally, albinism is characterized by extremely pale skin and an irreversible loss of melanocyte function, which predispose individuals to high UVR-induced DNA damage. These features do not represent the integumentary system of ancestral humans but instead reflect a highly derived condition that evolved much later in the history of human skin pigmentation and has arisen multiple times. For example, one mutation in the OCA2 gene is estimated to have appeared only 3,000 to 2,000 years ago in certain African populations (Stevens et al. 1997), whereas savanna-dwelling hominins lived around 1 mya (see Fig. 1). These hominins more likely resembled extant catarrhine primates and African apes, which possess intact pigmentary systems, the ability to develop facultative pigmentation (tanning), and melanin-rich skin on sun-exposed body parts—traits that are not functionally analogous to those seen in albinism (see also reply by Greaves 2014b; Jablonski and Chaplin 2014; Elias and Williams 2016; Jablonski 2021).
By extension, this criticism of inappropriate model choice would also apply to using patients with XP. Their photosensitivity results from defects in DNA repair mechanisms and is likewise a derived condition that bears little resemblance to the pigmentary system and adaptations of early hominins. A founder mutation in the XPC gene associated with XP in North African populations is estimated to have arisen only around 1,250 years ago (see Fig. 1) (Soufir et al. 2010).

Osborne and Hames (2014) proposed a second model for the skin cancer hypothesis using migrant populations. Specifically, they suggest examining skin cancer rates in fair-skinned individuals “whose ancestral homelands were in low UV environments but who currently grew up and live in higher UV environments” (Osborne and Hames 2014: 5). This scenario offers a testbed for assessing whether skin cancer affects survival or reproductive success in migrants used as a proxy for early hominins, while avoiding the criticism of inappropriate model choice by not relying on data from albinism or XP.
In the context of migration-related mismatch, Osborne and Hames (2014) draw on evidence of disparities in skin cancer risk between fair- and dark-skinned individuals living in Texas and between non-Māori and Māori in New Zealand. Moreover, they compare mortality data for fair-skinned populations in northern Europe with those in New Zealand and Australia, noting that skin cancer deaths were unexpectedly lower, not higher, in the latter, contrary to the skin cancer hypothesis. To account for this, Osborne and Hames (2014) suggest that higher vitamin D synthesis in fair-skinned populations in high-UV environments may confer some cancer-protective effects (Moan et al. 2008; Bikle 2012).
While important, the evidence offered by Osborne and Hames (2014) can be challenged on several grounds. First, a comparison of skin cancer mortality between populations residing in low-UV (high-latitude) versus high-UV (low-latitude) regions is not equivalent to a comparison between fair-skinned migrants to high-UV environments and long-term residents of those same environments. The critical distinction here is the role of migration and the sudden change in sun exposure duration and intensity it confers on the integrity of the skin. Second, the cancer-protective effects of vitamin D remain controversial, as observational and interventional studies have shown mixed results (Reichrath 2014; Manson et al. 2019; Zhang et al. 2019), making it difficult to attribute lower skin cancer mortality to increased vitamin D synthesis specifically. Third, since lower latitudes are associated with stronger UV radiation levels, it is possible that combining data from New Zealand with Australia may dilute important variations in skin cancer trends. For these reasons, we propose to reevaluate the case of migration-related mismatch using data on skin cancers in Australia specifically, for reasons we explain below.

Australia: Migration Policies and the Anglo-Celtic Heritage
Historically, Australia had a substantial influx of European migrants, particularly of Anglo-Celtic background, alongside a troubling legacy of racial immigration policies. These policies generally favored fair-skinned populations while restricting non-European immigration, especially from China and the Pacific Islands, and marginalizing Indigenous Australians on the basis of their darker skin (Jupp 2002; Carey and McLisky 2009). A significant early part of the European migration came through the British penal system, which between 1788 and 1868 transported over 150,000 convicts from Great Britain and Ireland to newly established Australian colonies as an alternative to execution or long-term imprisonment (Shaw 1977; Hughes 1987; Greaves 2000b: 182). By the first national census in 1901, the vast majority of the non-Indigenous population of Australia was of Anglo-Celtic origin, a demographic pattern that persisted into the early 20th century (Jupp 2001). Other significant migrant influxes included Germans, mainly before World War II, and postwar arrivals from the Netherlands, Italy, and Greece. Immigration from Asia only began after the 1970s, following the abolition of racially based immigration laws in Australia. According to the latest census in 2021, Australians who self-report European ancestry still constitute the majority, accounting for 57% of the population.
In sum, Australia's history of preferential immigration policies favoring fair-skinned migrants provides a salient example of migration-induced mismatch. This context enables examination of whether the resulting “adaptive lag” in pigmentation degree contributes to elevated skin cancer morbidity and mortality. More broadly, such findings may offer indirect evidence for skin cancer as a selective pressure among early savannah-dwelling hominins. In the following section, we review the literature on skin cancer rates in Australia.

Results: Skin Cancer in Australia
Generally speaking, the literature on skin cancers can be grouped into three categories: studies in migrants, high-risk populations, and the general population. Here, we focus on data for incidence, mortality, and, when available, age-relevant information (e.g. age of diagnosis) for BCC, SCC, and CM, and at times using age-standardized rates where relevant.Age-standardized incidence rates refer to the number of new skin cancers occurring in a population within a specific timeframe, adjusted to a standard age distribution, usually the 2001 Australian standard population. The main advantage here is that, by using the same standard, it allows for more accurate comparisons across populations with differing demographic patterns. Studies on actinic keratosis and other skin cancer types will not be considered.
Several studies have examined keratinocyte cancers in mainly Anglo-Celtic migrants to Australia. All show that incidence rates among migrants were consistently lower, sometimes even as much as half, compared to those in the Australian-born population (Giles et al. 1988; Kricker et al. 1991; Marks et al. 1993; English et al. 1998). Moreover, Kricker et al. (1991) and English et al. (1998) found that age at arrival was an important factor: Only migrants who arrived later in life had lower risk, whereas those who migrated during childhood had a similar risk to Australian-born individuals. Migrants from Southern Europe, with darker skin phenotypes, also showed a decreased risk of skin cancers (Kricker et al. 1991).
Beyond migrant studies, Australians of European ancestry residing at high-risk areas, such as Queensland, the northeastern state of Australia, represent another relevant category. In the tropical city of Townsville with around 6% of resident being of non-Caucasian origin, the age-standardized incidence rates for all skin cancers together were 3,439.7 and 1,991.0 per 100,000 annually for men and women, respectively, with most cases being of keratinocyte origin (Buettner and Raasch 1998). For subtropical Nambour with similar ethnic composition, largely similar incidences in skin cancers were found (Green et al. 1996). By contrast, in the southernmost state of Tasmania, whose population is predominantly of European ancestry, rates of keratinocyte cancers are less than half those in Queensland, indicating a strong latitude-related gradient in skin cancer risk within a genetically similar population (Ragaini et al. 2021). The age-standardized incidence rate of CM in Queensland as a whole is 72 cases per 100,000 person-year between 2010 and 2014 and notably the highest rate ever recorded globally (Aitken et al. 2018). CM-related mortality was reported by Aitken et al. (2018) at seven deaths per 100,000 person-year in 2014, also the highest record globally, and this is the only data available for high-risk areas.
The third, and most broad, category to examine skin cancer rates in Australians reflects the national level. Here, the most recent report found that person-based incidence rate of keratinocyte cancers was 1,531 cases per 100,000 person-annually, based on Medicare data (Medicare is the main government-subsidized health care insurance system for Australia and comes most essential health care services, including surgical treatments of skin cancer) for treatment services between 2011 and 2014 (Pandeya et al. 2017). Nationally, the incidence rate of CM reported at 50.3 per 100,000 person-year in 2015, representing the highest among seven fair-skinned populations studied (Whiteman et al. 2016; Olsen et al. 2019). Importantly, a recent modeling study projects age-standardized CM incidence rates at 82.2 per 100,000 for males and 58.5 per 100,000 for females among Australians with self-reported European ancestry. In contrast, the rate among individuals with low-risk ancestry was estimated to just 0.8 per 100,000 for both sexes (Whiteman et al. 2024).
How often is skin cancer lethality recorded at the national level? Concerning BCC, while recurrence and even locally aggressive behavior are not uncommon, metastatic spread is rare, with fewer than 400 cases in total reported worldwide (Laga et al. 2019). The incidence of metastatic BCC has been estimated to occur in an order of 0.0028% of cases, based on survey of dermatologists and known metastatic cases (Paver et al. 1973). In other words, roughly one BCC metastasizes per 35,000 cases observed in Australia (and New Zealand), making BCC lethality exceedingly rare.
SCC carries a much higher risk of metastasis than BBC. Although estimates vary, the metastatic rate for SCC in immunocompetent patients in Australia is approximately 5% (see also Czarnecki et al. 1994 and Czarnecki 2024). SCC is the main cause of death of all NMSC, accounting for roughly 70% of the cases (Girschik et al. 2008). Accurate mortality rates of SCC for Australia nationally are not available. Diagnoses of keratinocyte cancers are not legally required to be reported to cancer registries in Australia (except in Tasmania), making data collection and analysis more difficult. There have also been reports of misclassification of death certificates attributed to NMSC (Rosenblatt and Marks 1996; though see also Girschik et al. 2008). It has been estimated that the age-standardized rate for “skin cancers (excluding melanoma)” was 2.8 deaths in men and 1.1 deaths in women per 100,000 person in 2016 (Stang et al. 2019). Recently, Czarnecki (2024) has calculated that 1 in 275 cases of SCC among Australians resulted in death from 2019 to 2021. This was based on assumptions about SCC-related deaths and the ratio of BCC to SCC surgical treatments recorded by Medicare during the period, and Medicare data on surgical treatments does not differentiate between types of keratinocyte cancer excised. The crude death rate from NMSC for Australia in 2021 was calculated to be 3.0 per 100,000 total population and 3.9 for the susceptible (i.e. fair-skinned) populations (Czarnecki 2024).
CM is the deadliest skin cancer type due to its high metastatic potential. Its likelihood of spreading depends on several factors, including stage at diagnosis, presence of ulceration, and tumor depth (Breslow thickness) being key predictors of survival. Generally, while thick melanomas (typically >4 mm) are associated with the poorest long-term survival, most Australians diagnosed with CM are thin (≤1 mm), and about 23% melanoma deaths are attributable to thin tumors (Whiteman et al. 2015; Lo et al. 2025). National trends in CM mortality have steadily increased from the 1930s, peaking in 1985 with an age-standardized rate of 4.82 per 100,000 persons, with some intra-state variability (Giles et al. 1996). The stability in the CM mortality rate observed since 1994 has persisted or decreased in all groups except males aged ≥60 (Baade and Coory 2005). The age-standardized rate of mortality from CM for Australia nationally was 6 deaths per 100,000 person-year in 2014 (Aitken et al. 2018).
Concerning age of skin cancer diagnosis, keratinocyte cancers are most commonly diagnosed in individuals aged 60 and over, with incidence rates increasing rapidly with higher age (Raasch and Buettner 2002; Pandeya et al. 2017). CM largely follows this, but with a trend of diagnosis slightly earlier: In Queensland, Green et al. (2012) found the average age diagnosis with thin, invasive CM was 52.7, ranging from 15 to 89 years of age. Importantly, CM remains one of the most common cancers diagnosed in Australians under the age of 35, with incidence rates in young and adolescents in Queensland being three to five times higher than corresponding rates in United States and Europe (Iannacone et al. 2015).
In summary, skin cancers rates in Australia are consistent with an adaptive lag from mismatch-related migration events and the country's past preferential treatment for fair-skinned populations. Incidence rates of BCC and SCC are frequent but rarely fatal, whereas CM is relatively less common but associated with higher mortality risk. Skin cancers with metastatic potential peak after reproduction, but diagnosis is not exclusive to the elderly, particularly in CM, and childhood migration is associated with considerable skin cancer risk. Based on this, we infer that the Australian case of mismatch-by-migration does not present sufficiently strong evidence for skin cancer as a selection pressure in early hominins living on the savanna.

Discussion
Our conclusion, based on the Australian case of mismatch-by-migration, is largely consistent with Blum's (1961) cursory remark that onset and lethality of skin cancer usually occur after reproductive cessation and therefore are unlikely to act as significant selection pressure. The data used are limited by several factors, including advances in medical diagnosis and intervention and the influence of public health campaigns (e.g. sunscreen use), all of which modulate skin cancer risk in contemporary populations. Importantly, part of the adaptive lag may also reflect a vacation effect (Diffey 2002), in which sudden intense UV exposure from sun-seeking behavior during leisure time contributes to the patterns of skin cancer risk observed, rather than evolutionary factors alone. Moreover, such sudden, high-intensity UV exposure experienced by modern vacationers and travelers is also likely to differ significantly from the more gradual changes in sun exposure from tropical to savanna habitats over longer timescales that shaped human skin evolution. More fundamentally, any extrapolation from present-day populations to early hominins is inherently uncertain, as contemporary epidemiological patterns may not accurately reflect selective pressures of the distant past. Furthermore, our knowledge of ancestral human biology, their skin, and behavior remains incomplete.
Beyond Blum's criticism of the skin cancer hypothesis, several other have since then been raised, along with ad hoc explanations to recover a role for skin cancer in the evolution of dark skin. Below, we review and discuss several of these arguments, which broadly pertain to the use and appropriation of evolutionary reasoning particularly in relation to CM, life history theory, and inclusive fitness modeling.

Melanoma Skin Cancers
The selective pressure exerted particularly by CM has previously been questioned on the basis of its low incidence and tendency to occur post reproduction (Jablonski and Chaplin 2010; Elias and Williams 2016), largely consistent with our results. But the selective pressure of CM has been debated in relation to two other related aspects: first, whether they can play an evolutionary role at all, and second, if so, how best to operationalize their putative impact, that is, what measure is appropriate to gauge its selection pressure.
Concerning the first point, Elias and Williams (2016) have argued that attributing a selective pressure to melanoma skin cancers is “fundamentally illogical,” the reason being that “without the evolution of interfollicular melanocytes, melanoma would not occur.” To clarify their argument a bit further, melanocytes were at first confined to the base of hair follicles and only later migrated into the interfollicular epidermis, and this migration from intra- to interfollicular localization was a critical evolutionary step that enabled CM development (Pozdnyakova et al. 2009). Thus, CM could not have exerted selective pressure on pigmentation, since in the absence of interfollicular melanocytes producing melanin, melanoma could not have arisen in the first place.
There are two caveats to consider in this argument, one conceptual and one empirical. Conceptually, it is important to be clear about the use of teleological reasoning when discussing “the function” of melanocytes. These cells may have evolved initially for one purpose (e.g. pigmentation for camouflage), but only later to be co-opted for another (UV protection), much like protofeathers in avian ancestors likely evolved first for thermoregulation and were later exapted for flight (Gould and Vrba 1982). There is nothing “illogical” about skin pigmentation as an adaptation of the integument to prevent melanomas, among other UV-related pathologies, and the evolved function of melanocytes may have shifted once conditions like hairlessness and increased UV exposure from the savanna became sufficiently strong (see Box 1).
Empirically, it remains unclear whether the presence of interfollicular melanocytes is strictly necessary (“fundamental”) for CM development, as Elias and Williams imply, or whether they simply influence cancer likelihood. While it is true that most CM originate in epidermal melanocytes (Tobin 2011), follicular involvement is well documented (Pozdnyakova et al. 2009) and cases of CM with follicular origins have been reported (Machan et al. 2015; Tjarks et al. 2017), although the limited UV penetration to follicular depths suggests these cases may not result from direct UV damage. Experimental studies in mice have also shown that melanocyte stem cells in hair follicles can give rise to melanomas that closely resemble those seen in humans (Sun et al. 2019). Moreover, the presence of melanized skin in hairless areas of non-human primates suggests that the migration from follicular to interfollicular melanocytes and their functionality was likely a gradual, rather than a discrete evolutionary leap.
Concerning the second point, once we grant that CM cancers can act as a selective pressure, the next debated issue is how to gauge their putative impact. Instead of the standard measures of fitness (i.e. impact on survival and reproductive success), Osborne and Hames (2014) proposed that “the selective potential of CMM [cutaneous malignant melanoma] is best appreciated through an understanding of survival time following diagnosis.” They cite studies indicating that late-stage melanoma was previously associated with a survival time, ranging from 5 and 9 months. According to Osborne and Hames (2014, 3), these and similar studies “assume some level of medical intervention, implying that CMM would have been a strong source of selection during the evolution of skin pigmentation.”
This proposal risks conflating a clinical outcome with evolutionary relevance. Survival time following diagnosis is hardly a reliable proxy for evolutionary fitness—particularly if diagnosis typically occurs after reproductive age, as the Australian case of mismatch-by-migration illustrates. Put differently, evolutionary selection is most effective when deleterious traits affect early survival or during reproduction, as it would impact the entire reproductive output possible. In contrast, late-onset traits are subject to much weaker selection (Medawar 1952; Williams 1957). When assessing selective pressure of CMs and other skin cancers, the best available—though hardly the only—indicators of parental fitness cost remain incidence and mortality rates during the reproductive years.
Other fitness-related components may provide more informative measures than post-diagnosis survival time. For example, nonlethal melanomas and other skin cancers may have affected social or sexual competitiveness in ancestral humans by causing visible disfigurement or facial scarring. Such skin injuries could theoretically have reduced mating success in affected individuals, thereby exerting an indirect selective pressure. This broader approach to estimating fitness costs in the context of the evolution of dark pigmentation in ancestral humans has previously been proposed by Robins (1991) .
Box 2 Diamond's test for the skin cancer hypothesisIn some hunter-gatherer societies, post-reproductive elders play crucial roles in post-weaning childcare, food provision, and knowledge transmission, collectively also referred to as the grandmother effect (O’Connell et al. 1999). Drawing on these roles, Diamond (2005) proposed the skin cancer hypothesis could be tested by combining data on skin cancer in populations of varying skin phenotypes with estimates of the contributions of the elderly to the survival and well-being of their descendants. If these contributions were found to correlate with the lethality of skin cancer, it would support the idea that skin cancer exerted evolutionary pressure favoring darker skin indirectly, by enhancing the inclusive fitness of kin. Evidence for the presence of kin, including grandmothers, affecting child survival rates is now well-established (Sear and Mace 2008), and the main causes of death in hunter-gatherers appear to be infections, violence, and accidents (Gurven 2024). Yet, Diamond's test for the skin cancer hypothesis using inclusive fitness remains to be studied more systematically.

Life History Theory
The skin cancer hypothesis is also tied to discussions around life history theory, such as the following: At what life stage should we consider the selective pressure of skin cancer to be operative? And did early hominins live long enough to develop skin cancer, or at least long enough for it to manifest during their reproductive years?
Concerning the first issue, Osborne and Hames (2014), following Robins (1991), argued that the selection for dark skin pigmentation should not be considered solely in the context of adults of reproductive age, but also in infants, who possess limited thermoregulation, thinner epidermis, and less melanin and thus more susceptible to UV damage. Consistently, we also noted that fair-skinned individuals that migrate to Australia during childhood had a higher skin cancer risk. Robins' point is especially relevant for modeling purposes, but it is important to note that his argument was made in support of sunburn as a selective pressure for pigmentation and the thermoregulation hypothesis (see Supplementary material). While it is impossible to determine whether infants of ancestral humans were affected by sunburns, clinical reports document pediatric sunburns in modern populations living in UV-intense regions such as New Zealand and Australia (Mah et al. 2013; Connolly et al. 2021).
Evidence of infant mortality from skin cancers (excluding albinism) is sparse, both in modern clinical records and in the anthropological literature (though see Greaves 2000b, 183). More generally, any cancers occurring in earlier life stages are rare, and when they do occur, they are typically linked to developmental processes associated with hormones and growth. These processes are largely not implicated in the etiology of skin cancers. One notable (albeit controversial) exception may be pregnancy-associated melanomas that involve hormonal interactions, which possibly are linked to naevus/melanoma biology (Richtig et al. 2017).
Hyperthermia or impaired thermoregulation, by contrast, is a well-documented cause of child mortality (Booth et al. 2010). It is likely that sunburn-induced hyperthermia exerted a more immediate selective pressure on ancestral infants than lethal skin cancers, due to its acute onset and rapid, potentially fatal effects, in contrast to the typically slower progression of most skin cancers. Incorporating life stages other than the reproductive adult as operative units of selection for UV-related damage is crucial for accurately modeling the evolution of pigmentation, though this does not entail that skin cancer lethality was the most acute or potent threat across all life stages.
A second line of argument against the skin cancer hypothesis concerns life history states like lifespan and reproductive timing. Jablonski and Chaplin (2010) argue that early hominins reproduced at relatively young ages, and since extension of the average human lifespan is a modern phenomenon, skin cancer would have had little impact on reproductive success. Similarly, Elias and Williams (2016, 191) note that the peak incidence of SCC being the most common potentially lethal skin cancer “occurs above the age of 70, well past both the reproductive years and life expectancy of ancestral hominins.”
Life expectancy and reproductive patterns—like most other life history traits—are not directly observable in extinct populations (Robson and Wood 2008) but can be inferred from paleodemographic data, like dental and skeletal remains (see also Lequin et al. 2025). These samples can be used to estimate age-at-death and categorize individuals into broad age groups (e.g. infant, adolescent, older adult), enabling analysis of age distribution and mortality patterns. Based on this, the evidence indicates that hominin life expectancy, even until the late Pleistocene (approx. 130,000 to 12,000 years ago), was relatively short, marked by high infant mortality and low survival into older adulthood (Caspari and Lee 2004; Trinkaus 2011). Consequently, ancestral humans had evolve a fast life history strategy, characterized by early sexual maturation and onset of reproduction (much like Tasmanian devils today; see Introduction), a claim that is largely interpretable though fossil remains (Kennedy 2003).
Paleodemographic approaches to inferring hominin life history traits face well-known challenges, particularly associated with interpreting health from skeletal remains, also known as the osteological paradox (Wood et al. 1992). More to the point: Paleodemography is not the only approach available. Hominin lifespan and reproductive patterns can also be inferred from comparative life history data, using either extant primates or modern hunter-gatherer populations. Based on such evidence, life expectancy from adolescence (i.e. age 15) among most hunter-gatherers reaches around 45 years, with many individuals living an additional two decades beyond that (Gurven and Kaplan 2007: 236). Similarly, great ape species show maximum lifespans approaching five decades (Robson and Wood 2008: 398).
As for reproductive patterns, hunter-gatherer females typically have their first child around 19 to 20 years of age, bear around four children, with interval births each 3 to 4 years and a reproductive span of roughly two decades (Kaplan et al. 2000: 158). Most female great apes begin reproducing around 10 to 15 years of age, produce three to six offspring, have interbirth intervals of 4 to 8 years, and typically continue reproducing until their 40s (Robson and Wood 2008: 398). Based on this, the survival and reproductive lifespan of ancestral humans may have been longer than paleodemographic data alone suggest, potentially allowing skin cancer to impact not only parental fitness but also the reproductive success of close relatives (Greaves 2014a; Osborne and Hames 2014). We will discuss this last option concerning inclusive fitness models separately below.

Inclusive Fitness

Hamilton's (1964) proposed that evolutionary fitness is influenced not only by direct (parental) reproductive success but also by indirect genetic transmission via social behaviors that enhance the survival and reproduction of close kin. Based on this model of inclusive fitness and given that ancestral humans may have lived longer than paleodemographic estimates suggest, it has been argued that skin cancers may have exerted selection pressure on post-reproductive elders, thereby limiting their ability to provide alloparental care to their grandchildren (see Box 2). In this way, skin cancer mortality risk in later life could have acted as a selective pressure by reducing indirect fitness benefits, thereby shaping the evolution of traits like skin pigmentation (Diamond 2005; Greaves 2014b; Osborne and Hames 2014). Put differently, this argument proposes that dark skin pigmentation evolved in service of the grandmother effect, or perhaps a “great-grandmother effect,” as Elias and Williams (2016) put it.
One argument against skin cancer's impact on inclusive fitness goes as follows: The ethnographic observations of elders' roles in reducing mothers' time allocation between children and in influencing children's nutritional status (e.g. Hawkes et al. 1997) should not be considered to be a “cultural absolute even in recent hunter-gather cultures” (Elias and Williams 2016). Put differently, the value of elders for kin survival and health status is context-dependent and likely varies across environments and populations.
We posit that the fact this value is not universally observed across all hunter-gatherer populations is, to some extent, irrelevant—the key question is whether such a role in ancestral humans living on the savanna is plausible and coherent with other lines of evidence. To our knowledge, no evidence from paleoecology, dermatology, evolutionary theory, or related fields precludes the possibility that the grandmother effect was operative. However, direct evidence supporting its role in ancestral humans—specifically in conjunction with skin cancer risk, as proposed by Diamond (2005) (see Box 2)—remains, admittedly hypothetical. Relatedly, Jablonski and Chaplin (2014) note that, based on a theoretical model using life history data from Gambian villages (Shanley et al. 2007), grandmaternal survival has a fitness effect, though its impact on child mortality is about five times less than that of maternal survival. While modest, this nonetheless supports a potential fitness role for grandmothers. However, further modeling and mortality data are needed to determine the specific contribution of skin cancer to this fitness effect.

Conclusion
In this review, we have introduced a putative case of cancer selection in human evolution within the field of cancer evolution and ecology, examining whether UVR-induced skin cancers in early savanna-dwelling hominins contributed to the evolution of dark epidermal pigmentation. Drawing on the model of evolutionary mismatch-by-migration, we used skin cancer rates in modern Australians of high-risk ancestry as a proxy for the skin phenotype and solar UV-selection pressures faced by early hominins transitioning from tropical forests to savanna mosaics. Our analysis indicates that skin cancer onset of incidence and mortality occurs mostly after reproduction, leaving the skin cancer hypothesis unsupported. This suggests that the protective effects of dark skin in ancestral humans were likely not selected as an anti-cancer adaptation, but rather to mitigate other risks associated with a UV-saturated, desiccating environment, including folate loss, sunburn, and thermoregulatory stress as well as disruption of barrier function. While the skin cancer hypothesis could be sustained if post-reproductive elders contributed to kin fitness through alloparental caregiving, this possibility remains to be tested.

Supplementary Material
msaf306_Supplementary_Data