Human skin color diversity is highest in sub-Saharan African populations

Relethford, John H

Abstract Previous studies of genetic and craniometric traits have found higher levels of within-population diversity in sub-Saharan Africa compared to other geographic regions. This study examines regional differences in within-population diversity of human skin color. Published data on skin reflectance were collected for 98 male samples from eight geographic regions: sub-Saharan Africa, North Africa, Europe, West Asia, Southwest Asia, South Asia, Australasia, and the New World. Regional differences in local within-population diversity were examined using two measures of variability: the sample variance and the sample coefficient of variation. For both measures, the average level of within– population diversity is higher in sub-Saharan Africa than in other geographic regions. This difference persists even after adjusting for a correlation between within-population diversity and distance from the equator. Though affected by natural selection, skin color variation shows the same pattern of higher African diversity as found with other traits.


There has been considerable interest in recent years regarding the level of human genetic diversity in different geographic regions. A variety of genetic traits have shown higher diversity in sub-Saharan African populations, including mitochondrial DNA (Cann et al. 1987; Vigilant et al. 1991; Bowcock et al. 1994; Jorde et al. 1995), microsatellite DNA (Bowcock et al. 1994; Deka et al. 1995; Jorde et al. 1995, 1997; Tishkoff et al. 1996; Relethford and Jorde 1999), and craniometrics (Relethford and Harpending 1994). Alu insertion polymorphisms also show relatively high diversity in sub-Saharan African populations, although somewhat lower than found in western Asia (Stoneking et al. 1997). The situation is different for classical genetic markers and restriction fragment length polymorphisms; sub-Saharan African populations do not show higher levels of diversity for these loci, perhaps reflecting ascertainment bias (Rogers and Jorde 1996) and/or lower mutation rates (Relethford 1997).

The evidence to date points to a general finding of greater within– population diversity in sub-Saharan African populations relative to other major geographic regions such as Europe, East Asia, and Australasia. The significance of this observation has been argued in the context of the continuing debate over modern human origins. Some authors have argued that higher African diversity is a reflection of greater “age” within Africa (e.g., Stoneking and Cann 1989), whereas others have suggested that a higher level of diversity reflects larger population size in Africa (Relethford and Harpending 1994; Relethford 1998a; Relethford and Jorde 1999).

Regardless of the specific factors, it is clear that we should expect, with certain exceptions due to ascertainment bias and mutation rate, higher levels of genetic diversity in sub-Saharan African populations across the human genome. Most work to date has focused on genetic markers and DNA sequences, with the exception of Relethford and Harpending’s (1994) demonstration that craniometric variation follows the same geographic pattern. Their findings raise the question of the extent to which we might expect higher levels of African diversity for other quantitative traits. This paper investigates regional differences in within-population diversity of human skin color, showing that diversity is highest within sub-Saharan African populations.

Materials and Methods

Summary statistics on human skin color variation were obtained from the published literature. Human skin color has most frequently been measured on one of two different reflectometers, the E.E.L. and Photovolt machines. Most studies of Old World (and some New World) populations have used the E.E.L. reflectometer, while the Photovolt reflectometer has been used primarily in New World studies. Both machines function in the same manner, by measuring the percentage of light reflected back from relatively unexposed skin on the inner surface of the upper arm. However, the two machines are not directly comparable, and conversion formulae are not available for all ranges of pigmentation or validated on many samples (Garrard et al. 1967; Lees and Byard 1978; Lees et al. 1979). Consequently, the analyses reported in this paper rely entirely on data obtained for the E.E.L. machine, since more samples are available. The E.E.L. machine typically measures skin reflectance at nine different wavelengths, although a number of studies use only a subset of wavelengths. This analysis was limited to data for the E.E.L. filter number 609, which corresponds to red light at 685 nm. Previous studies have shown that this is the single best filter for analyzing population differences in skin reflectance.

The specific samples chosen have been reported previously by Relethford (1998b). Analysis was further limited to male samples, since there are more data available for males than females. Of the 102 Old World male samples in my earlier paper, I deleted eight because the variance in skin color was not reported and could not be calculated from data within the paper. Four other samples were deleted because of small sample size (n

Geographic coverage within each region is limited by available data, with the result that some regions are better represented than others. SubSaharan Africa and South Asia are well represented in terms of the number of samples and geographic coverage. North Africa, West Asia, and Southwest Asia have fewer samples, but these numbers are adequate for their geographic area. Europe has a large number of samples, but primarily from western and southern Europe. Australasia and the New World are more limited in numbers of samples, particularly the geographic coverage in the New World. The lack of suitable data from East Asia is problematic, but there still exists enough of a range of African and non-African samples to test the hypothesis of higher diversity in sub-Saharan African populations.

Two measures of within-population phenotypic diversity were examined for each local population that was sampled-the variance and the coefficient of variation. The means of these two measures were computed within each region to estimate the average within-population diversity. The null hypothesis of no regional differences in within-population variation was tested using analysis of variance (ANOVA). All computations were performed using SYSTAT for Windows, version 8 (SPSS 1998).


Mean within-population variances and coefficients of variation are reported in Table 1. The highest average within-population variance is found in sub-Saharan Africa and the lowest in Europe. This pattern is more pronounced for the mean within-population coefficients of variation (CV), where the value for sub-Saharan Africa is almost double the values for all other regions excepting Australasia, which has the second highest within-population CV. The higher CV values for sub-Saharan Africa and Australasia are a reflection of high within-population variation combined with low mean reflectances (darker populations have lower mean reflectances), since the CV is a measure of variation relative to the mean (100V/X(overscored)).

Regional differences in within-population variation are highly significant. For mean within-population variance, the ANOVA shows highly significant results (F = 3.61, df = 7 and 90, p = 0.002). The Bonferroni posthoc test reveals only one significant contrast (p

The use of ANOVA makes a number of assumptions including homogeneity of variances (variance in this sense refers to the variation in sample variances or CVs within each region). While generally robust, ANOVA can be affected by departures from this assumption. Both the Bartlett and Levene tests show significant heterogeneity across geographic regions for both within-population variances and coefficients of variation (p


Previous results obtained from mtDNA, microsatellite DNA, and craniometric variation show a pattern in which the highest level of within– population diversity occurs in sub-Saharan African populations. The current study shows the same pattern for human skin color. Mean within-population diversity, whether assessed using the variance or the coefficient of variation, is highest in sub-Saharan Africa populations and lowest in Europe.

One potential problem is the effect of natural selection on patterns of within-population diversity. Several studies have shown a very strong relationship between skin reflectance and distance from the equator, suggesting a causal link between skin color and geography (Roberts and Kahlon 1976; Tasa et al. 1985; Relethford 1998b). Could natural selection mimic the pattern of within-population variation seen in other traits? That is, is it possible that past selection intensity varied by latitude, and resulted in higher within-population variation in some parts of the world? One way of investigating this problem is to look at the relationship between measures of within-population variation and latitude. This was accomplished by regressing each measure (variance and CV) on distance from the equator. There is a slight, though significant, relationship of within-population variance and distance from the equator (R^sup 2^ = 0.08, p = 0.006). The relationship between within-population coefficients of variation and distance from the equator is higher (R^sup 2^ = 0.42, p = 0.000), as expected, since mean skin color is highly correlated with distance from the equator, and the CV includes the mean in its computation.

These results suggest that latitude may possibly have an effect on the level of within-population diversity in skin reflectance. To control for this, the ANOVA analyses were repeated using distance from the equator as a covariate. The adjusted regional means are reported in Table 2. For both within-population variance and within-population CV, the results are almost the same as in Table 1. As with the unadjusted means, the regional effect is significant for both within-population variance (F = 2.43, df = 7 and 89, p = 0.025) and within-population coefficient of variation (F = 9.61, df = 7 and 89, p = 0.000). These results suggest that within-population diversity of skin color, like mtDNA, nuclear DNA, and craniometrics, is a reflection of the past population history of the human species.

However, which history? The modern human origins debate typically centers on the relationship between older “archaic” humans living throughout the Old World during the past several hundred thousand years and more “modern” humans during the past 30,000 years or so. The recent African origin model proposes that the transition from archaic to modern humans occurred only within Africa roughly 100,000 to 200,000 years ago, followed by replacement of non-African archaic populations. An alternative, the multiregional evolution model, proposes that humans have been a polytypic species over the past 2 million years, and that the transition to modems occurred within this geographically widespread species (see Relethford 1998a and Wolpoff and Caspari 1997 for a review of modern human origin models).

The issue of regional diversity relates to these models. The finding of higher within-population diversity in sub-Saharan African populations has frequently been used as support for the recent African origin model. According to this model, within-population diversity is a function of time. When some African populations began dispersing outside of Africa, the small number of initial founders resulted in a “reset” of within-population diversity. Over time, the non-African groups began accumulating diversity through the action of mutation balanced by genetic drift, while the older African parent population retained higher levels of diversity (Stoneking and Cann 1989). There are two problems with this interpretation. First, the bottleneck required to “reset” levels of within-population diversity would have to have been more severe than is likely (Rogers and Jorde 1995; Relethford and Jorde 1999). Second, the finding of higher within-population diversity is also compatible with a multiregional model, where Africa had a larger long-term effective population size (Relethford and Harpending 1994; Relethford and Jorde 1999). If so, more of our ancestors several hundred thousand years ago may have lived in Africa than elsewhere, in which case the origin of modern humans may be mostly, but not exclusively, out of Africa. Unfortunately, the present finding of higher within-population skin color diversity in subSaharan Africa does not resolve the debate. The finding only adds to a growing literature on the special influence of African populations in the evolution of modern humans.

Literature Cited

Bowcock, A.M., A. Ruiz-Linares, J. Tomfohrde et al. 1994. High resolution of human evolutionary trees with polymorphic microsatellites. Proc. Natl. Acad. Sci. USA 368:455-457.

Cann, R.L., M. Stoneking, and A.C. Wilson. 1987. Mitochondrial DNA and human evolution. Nature 325:31-36.

Deka, R., L. Jin, M.D. Shriver et al. 1995. Population genetics of dinucleotide (dC-dA)n (dG-dT), polymorphisms in world populations. Am. J. Hum. Genet. 56:461-474. Ducros, A., J. Ducros, and P. Robbe. 1975. Pigmentation et brunissement compares d’Eskimo

(Ammassalimut, Groenland de l’est). L’Anthropologie (Paris) 79:299-316.

Garrard, G., G.A. Harrison, and J.J. Owen. 1967. Comparative spectrophotometry of skin colour with EEL and Photovolt instruments. Am. J. Phys. Anthropol. 27:389-396.

Harrison, G.A., and F.M. Salzano. 1966. The skin colour of the Caingang and Guarani Indians of Brazil. Hum. Biol. 38:104-111.

Jorde, L.B., M.J. Bamshad, W.S. Watkins et al. 1995. Origins and affinities of modern humans: A comparison of mitochondrial and nuclear genetic data. Am. J. Hum. Genet. 57:523538.

Jorde, L.B., A.R. Rogers, M. Bamshad et al. 1997. Microsatellite diversity and the demographic history of modem humans. Proc. Natl. Acad. Sci. USA 94:3100-3103.

Lees, F.C., and P.J. Byard. 1978. Skin colorimetry in Belize. I. Conversion formulae. Am. J. Phys. Anthropol. 48:515-522.

Lees, F.C., P.J. Byard, and J.H. Relethford. 1979. New conversion formulae for light-skinned populations using Photovolt and E.E.L. reflectometers. Am. J. Phys. Anthropol. 51:403408.

Parsons, P.A., and N.G. White. 1976. Variability of anthropometric traits in Australian aboriginals and adjacent populations. In The Origin of the Australians, R.L. Kirk and A.G. Thorne, eds. Atlantic Highlands, NJ: Humanities Press, 227-243.

Relethford, J.H. 1997. Mutation rate and excess African heterozygosity. Hum. Biol. 69:785792.

Relethford, J.H. 1998a. Genetics of modern human origins and diversity. Annu. Rev. Anthropol. 27:1-23.

Relethford, J.H. 1998b. Hemispheric difference in human skin color. Am. J. Phys. Anthropol. 104:449-457.

Relethford, J.H., and H.C. Harpending. 1994. Craniometric variation, genetic theory, and modem human origins. Am. J. Phys. Anthropol. 95:249-270.

Relethford, J.H., and L.B. Jorde. 1999. Genetic evidence for larger African population size during recent human evolution. Am. J. Phys. Anthropol. 108:251-260.

Roberts, D.F., and D.P.S. Kahlon. 1976. Environmental correlations of skin colour. Ann. Hum. Biol. 3:11-22.

Rogers, A.R., and L.B. Jorde. 1995. Genetic evidence on modern human origins. Hum. Biol. 67:1-36.

Rogers, A.R., and L.B. Jorde. 1996. Ascertainment bias in estimates of average heterozygosity. Am. J. Hum. Genet. 58:1033-1041.

SPSS. 1998. Systat 8.0 for Windows. Chicago, IL: SPSS, Inc.

Stoneking, M., and R.L. Cann. 1989. African origin of human mitochondrial DNA. In The Human Revolution: Behavioral and Biological Perspectives on the Origin of Modern Humans, P. Mellars and C. Stringer, eds. Princeton, NJ: Princeton University Press, 1730.

Stoneking, M., J.J. Fontius, S.L. Clifford et al. 1997. Alu insertion polymorphisms and human evolution: Evidence for a larger population size in Africa. Genome Res. 7:1061-1071. Tasa, G.L., CT Murray, and J.M. Boughton. 1985. Reflectometer reports on human pigmentation. Curr. Anthropol. 26:511-512.

Tishkoff, S.A., E. Dietzsch, W. Speed et al. 1996. Global patterns of linkage disequilibrium at the CD4 locus and modern human origins. Science 271:1380-1387.

Vigilant, L., M. Stoneking, H. Harpending et al. 1991. African populations and the evolution of human mitochondrial DNA. Science 253:1503-1507.

Weiner, J.S., N.C. Sebag-Montefiore, and J.N. Peterson. 1963. A note on the skin-colour of Aguarana Indians of Peru. Hum. Biol. 35:470-473.

Wolpoff, M.H., and R. Caspari. 1997. Race and Human Evolution. New York, NY: Simon and Schuster.


1 Department of Anthropology, State University of New York College at Oneonta, Oneonta, NY 13820

Human Biology, October 2000, v. 72, no. 5, pp. 773-780.

Copyright Wayne State University Press Oct 2000

Provided by ProQuest Information and Learning Company. All rights Reserved

You May Also Like

La Alta Alpujarra Oriental (Southeast Spain)

Spontaneous abortion pattern in an isolated Mediterranean population: La Alta Alpujarra Oriental (Southeast Spain) Luna, F Abstract …

Lack of Arg972 Polymorphism in the IRS1 Gene in Parakana Brazilian Indians

Lack of Arg972 Polymorphism in the IRS1 Gene in Parakana Brazilian Indians Bezerra, Rosangela M N Abstract Several polymorphisms in …

Multivariate analysis of the Italian population

Factors affecting the sex ratio in humans: Multivariate analysis of the Italian population Ulizzi, L K. ULIZZI(1) AND L.A. ZONTA(2)<…

Physiological variations in the pigmentation of newborn infants

Physiological variations in the pigmentation of newborn infants Grande, R Studies of pigmentation are abundant for specific populati…