LITERATURE CITED

Herman J and Celarier E., Eds. (1996) TOMS version 7.UV-Erythemal exposure: 1978–1993. Data developed

by NASA Goddard Space Flight Center Ozone Process-ing Team. CD-ROM USA_NASA_916_OPT_007.

Relethford JH (1997) Hemispheric difference in humanskin color. Am. J. Phys. Anthropol. 104:449–458.

Reply

John H. Relethford1*and Richard L. McKenzie2

1Department of Anthropology, State Universityof New York College at Oneonta, Oneonta, NewYork 138202National Institute of Water and AtmosphericResearch, NIWA Lauder, Central Otago, NewZealand

In my original article (Relethford, 1997), Iprovided evidence of a hemispheric differ-ence in human skin color. Although theevolutionary reasons for this difference arenot known, it was noted that there is evi-dence for a hemispheric difference in ultra-violet (UV) radiation. Several reasons, in-cluding a hemispheric difference in theearth–sun distance, were noted. Chaplinand Jablonski’s comment (1998) does notdispute the hemispheric difference in skincolor but argues against a hemispheric differ-ence in UV radiation. Because their argu-ment concerns itself with data from the fieldof atmospheric science, I invited RichardMcKenzie, author of many of the cited pa-pers on UV radiation, to participate in thisreply.

Model calculations of clear-sky UV radia-tion indicate that based on recent climatolo-gies of ozone measured from satellites, thereare significant hemispheric differences inthe latitudinal decreases in UV radiation.Latitude for latitude, the calculated dosesare larger in the Southern Hemisphere. Inannually averaged doses, the hemisphericdifferences in erythemally weighted (or ‘‘sun-burning’’) UV radiation (McKinlay and Dif-fey, 1987) are generally less than 10%. How-ever, for doses received during peak months(summer), the differences exceed 15% atlatitudes greater than 25°. The hemisphericdifference in both mean and maximum UVradiation is shown in Figure 1.

The calculated hemispheric differencesarise from the lower column amounts ofozone in the Southern Hemisphere and be-cause the time of closest approach betweenthe earth and sun coincides with the South-ern Hemisphere summer. Because of thestrong solar zenith angle dependence of UVtransmission, most of the annual dose of UVradiation arrives in the summer. These cal-culations ignore the effects of clouds andaerosols, both of which can have importanteffects on UV radiation.

Because of instrument intercalibration dif-ficulties, there have been few measurementsthat directly quantify the magnitude of thesegeographical differences in UV radiation.The few measurements that have been madewith cross-calibrated instruments (e.g., Seck-meyer et al., 1995 and references therein)have shown even larger hemispheric differ-ences (40–60% in erythemal UV radiation),both for clear-sky conditions and for all-weather conditions. The larger differencesare attributable in part to differences intropospheric pollution. However, there areimportant sampling issues that need to beconsidered because there can be large geo-graphical differences in cloud cover and pol-lution. It is unlikely that the measurementsthat have been made were at sites that wereaccurately representative of the whole hemi-sphere.

Recently, satellite data have been used toestimate global patterns of UV radiationdoses, including cloud effects. Chaplin andJablonski’s analysis, in which they findsmaller hemispheric differences, seems tocontradict the results based on clear-skymodel calculations and the direct measure-ments. However, Chaplin and Jablonski pre-

*Correspondence to: Dr. John H. Relethford, Department ofAnthropology, SUNY College at Oneonta, Oneonta, NY 13820-4015. Tel: (607) 436-2017. Fax: (607) 436-2653. E-mail:relethjh@oneonta.edu

Received 3 July 1998; accepted 5 July 1998.

223NOTES AND COMMENTS

sented their results in terms of hemisphericdifferences (area weighted) rather than con-sidering latitude by latitude. Because 50% ofthe globe lies within 30° of the equator,where the UV differences are smallest (Fig.1), this averaging tends to minimise thedifferences discussed by Relethford (1997).We accept that when only land masses areconsidered, the habitable land mass of theSouthern Hemisphere is concentrated nearerto the equator. However, this is irrelevant tothe arguments put forward by Relethford,unless one assumes that the possibility ofmigration patterns over a wider range oflatitudes can have important moderatingeffects on skin color. Other possible reasonsfor the smaller differences (compared withthose for clear skies) include (1) increasedeffects on UV of cloud cover in the South, or(2) greater mean elevations above mean sealevel in the North.

However, limitations of the satellite re-trievals should also be kept in mind. First,the method is relatively insensitive to extinc-tions by pollutants in the lower troposphere(e.g., aerosols and ozone) that are moreprevalent in the Northern Hemisphere. How-ever, in the context of the evolution of skincolor, these differences are not importantbecause they probably developed only re-

cently (e.g., since the Industrial Revolution).Second, very little work has been done so farto validate the satellite estimations of UVradiation, particularly with regard to hemi-spheric differences. The first CD ROM is-sued by NASA of UV doses estimated fromthe TOMS instruments contained errors andhas since been revised (Herman et al., 1998).However, the error should not have causedsignificant systematic differences in derivedUV radiation between the hemispheres, so itwould not have altered the conclusions ofChaplin and Jablonski. Clearly, more mea-surements are required at further sites bothto validate the satellite estimations and toquantify the latitudinal differences in UVradiation. We further note that if maximumUV levels rather than annual doses arerelated to skin color, then the clear-skycalculations may be more appropriate.

Chaplin and Jablonski have raised someinteresting points. Further work needs to bedone to unequivocally establish the causal-ity between latitudinal differences in UVand skin color. Future studies of skin colorshould include regions at larger latitudes,where hemispheric differences in UV radia-tion become more pronounced.

LITERATURE CITED

Chaplin GC and Jablonski NG (1998) Comment on‘‘Hemispheric Difference in Human Skin Color.’’ Am.J. Phys. Anthropol. 107:221–223.

Herman, JR, McKenzie RL, Diaz SB, Kerr JB, Madron-ich S, and Seckmeyer G (1998) Ultraviolet radiation atthe earth’s surface. In DL Albritton, PJ Aucamp, GMegie, and RT Watson (ed): UNEP/WMO ScientificAssessment of the Ozone Layer. Global Ozone Re-search and Monitoring Project, Geneva (in press).

Madronich S (1993) The atmosphere and UV-B radia-tion at ground level. In AR Young, LO Bjorn, J Moanand W Nultsch (eds.): Environmental UV Photobiol-ogy. New York: Plenum, pp. 1–39.

McKinlay AF and Diffey BL (1987) A reference actionspectrum for ultra-violet induced erythema in humanskin. In WF Passchier and BFM Bosnajakovic (eds):Human Exposure to Ultraviolet Radiation: Risks andRegulations. Amsterdam: Elsevier, pp. 83–87.

Relethford JH (1997) Hemispheric difference in humanskin color. Am. J. Phys. Anthropol. 104:449–457.

Seckmeyer G, Mayer B, Bernhard G, McKenzie RL,Johnston PV, Kotkamp M, Booth CR, Lucas T, Mes-techkina T, Roy CR, Gies HP, and Tomlinson D (1995)Geographical differences in the UV measured byintercompared spectroradiometers. Geophys. Res. Lett.22:1889–1892.

Fig. 1. Hemispheric differences in clear-sky dose oferythemal UV radiation. Monthly daily dose of erythe-mal UV was calculated for clear-sky conditions using aclimatology of ozone measurements from the TOMSinstrument from 1979 to 1992. Calculations were fromthe ‘‘tuv’’ radiative transfer code described by Madronich(1993). For each 5° interval, the annual average andmaximum doses were then computed. These data wereconverted to ratios in this graph to show the higherlevels of UV radiation in the Southern Hemisphere.

224 J.H. RELETHFORD AND R.L. MCKENZIE