Nature © Macmillan Publishers Ltd 1997

letters to nature

NATURE | VOL 389 | 9 OCTOBER 1997 593

resulting eruption can be expected to be more violent and dischargemore SO2 into the atmosphere than would be observed without themixing event.

Juvenile products of recent activity at Volcan Popocatepetl,Mexico, exhibit both primary igneous anhydrite28 and evidencefor magma mixing29. Eruptive events have been relatively small sofar. Nevertheless, Popocatepetl has been generating SO2 at sustainedhigh rates30 which are difficult to account for from conventionalsources. It is far too early to draw conclusions, but evidence hintsthat the same sort of process proposed for the June 1991 activity atPinatubo may be at work today at Popocatepetl. M

Received 29 January; accepted 11 August 1997.

1. Scott, W. E. et al. in Fire and Mud: Eruptions and Lahars of Mount Pinatubo, Philippines (eds Newhall,C. G. & Punongbayan, R. S.) 545–570 (Philippine Inst. Volcanol. & Seismol./Univ. Washington Press,Quezon City/Seattle, 1996).

2. Bluth, G. J. S., Doiron, S. D., Schnetzler, C. C. & Krueger, A. J. Global tracking of the SO2 clouds fromthe June, 1991 Mount Pinatubo eruptions. Geophys. Res. Lett. 19, 151–154 (1992).

3. Gerlach, T. M., Westrich, H. R. & Symonds, R. B. in Fire and Mud: Eruptions and Lahars of MountPintatuba, Philippines (eds Newhall, C. G. & Punongbayan, R. S.) 415–434 (Philippine Inst. Volcanol.& Seismol./Univ. Washington Press, Quezon City/Seattle, 1996).

4. Brasseur, G. & Granier, C. Mount Pinatubo aerosols, chlorofluorocarbons, and ozone depletion.Science 257, 1239–1242 (1992).

5. Fournelle, J., Carmody, R. & Daag, A. S. in Fire and Mud: Eruptions and Lahars of Mount Pinatubo,Philippines (eds Newhall, C. G. & Punongbayan, R. S.) 845–864 (Philippine Inst. Volcanol. &Seismol./Univ. Washington Press, Quezon City/Seattle, 1996).

6. Imai, A., Listanco, E. L. & Fujii, T. Petrologic and sulfur isotopic significance of highly oxidized andsulfur-rich magma of Mt. Pinatubo, Philippines. Geology 21, 699–702 (1993).

7. Bernard, A., Demaiffe, D., Mattielli, N. & Punanongbayan, R. S. Anhydrite-bearing pumices fromMount Pinatubo: further evidence for the existence of sulphur-rich silicic magmas. Nature 354, 139–140 (1991).

8. Westrich, H. R. & Gerlach, T. M. Magmatic gas source for the stratospheric SO2 cloud from the June15, 1991, eruption of Mount Pinatubo. Geology 20, 867–870 (1992).

9. Hattori, K. High-sulfur magma, a product of fluid discharge from underlying mafic magma: Evidencefrom Mount Pinatubo, Philippines. Geology 21, 1083–1086 (1993).

10. Evans, B. W. & Scaillet, B. The redox state of Pinatubo dacite and the ilmenite-hematite solvus. Am.Mineral. 82, 625–629 (1997).

11. Pallister, J. S., Hoblitt, R. P. & Reyes, A. G. A basalt trigger for the 1991 eruptions of Pinatubo volcano?Nature 356, 426–428 (1992).

12. Pallister, J. S., Hoblitt, R. P., Meeker, G. P., Knight, R. J. & Siems, D. F. in Fire and Mud: Eruptions andLahars of Mount Pinatubo, Philippines (eds Newhall, C. G. & Punongbayan, R. S.) 687–732 (PhilippineInst. Volcanol. & Seismol./Univ. Washington Press, Quezon City/Seattle, 1996).

13. Hattori, K. in Fire and Mud: Eruptions and Lahars of Mount Pinatubo, Philippines (eds Newhall, C. G. &Punongbayan, R. S.) 807–824 (Philippine Inst. Volcanol. & Seismol./Univ. Washington Press, QuezonCity/Seattle, 1996).

14. Carmichael, I. S. E. The redox states of basic and silicic magmas: a reflection of their source regions?Contrib. Mineral. Petrol. 106, 129–141 (1991).

15. Carroll, M. R. & Rutherford, M. J. Sulfide and sulfate saturation in hydrous silicate melts. J. Geophys.Res. 90, C601–C612 (1985).

16. Katsura, T. & Nagashima, S. Solubility of sulfur in some magmas at 1 atmosphere. Geochim.Cosmochim. Acta 38, 517–531 (1974).

17. Fincham, C. J. B. & Richardson, F. D. The behavior of sulphur in silicate and aluminate melts. Proc. R.Soc. Lond. A 223, 40–62 (1954).

18. Nagashima, S. & Katsura, T. The solubility of sulfur in Na2O-SiO2 melts under various oxygen partialpressures at 1100 8C, 1200 8C and 1300 8C. Bull. Chem. Soc. Jpn. 46, 3099–3103 (1973).

19. Wendlandt, R. F. Oxygen diffusion in basalt and andesite melts: experimental results and discussion ofchemical versus tracer diffusion. Contrib. Mineral. Petrol. 108, 463–471 (1991).

20. Kress, V. C. Thermochemistry of sulfide liquids I: The system O–S–Fe at 1 bar. Contrib. Mineral.Petrol. 127, 176–186 (1997).

21. Kress, V. C. & Carmichael, I. S. E. The compressibility of silicate liquids containing Fe2O3 and the effectof composition, temperature, oxygen fugacity and pressure on their redox states. Cotnrib. Mineral.Petrol. 108, 82–92 (1991).

22. Wallace, P. J. & Gerlach, T. M. Magmatic vapor source for sulfur dioxide released during volcaniceruptions: Evidence from Mount Pinatubo. Science 265, 497–499 (1994).

23. Luhr, J. F. Experimental phase relations of water- and sulfur-saturated arc magmas and the 1982eruptions of El Chichon Volcano. J. Petrol. 5, 1071–1114 (1990).

24. Carroll, M. R. & Rutherford, M. J. The stability of igneous anhydrite: experimental results andimplications for sulfur behavior in the 1982 El Chichon trachyandesite and other evolved magmas. J.Petrol. 28, 781–801 (1987).

25. Mori, J., Eberhart-Phillips, D. & Harlow, D. H. in Fire and Mud: Eruptions and Lahars of MountPinatubo, Philippines (eds Newhall, C. G. & Punongbayan, R. S.) 371–382 (Philippine Inst. Volcanol.& Seismol/Univ. Washington Press, Quezon City/Seattle, 1996).

26. Snyder, D. & Tait, S. Magma mixing by convective entrainment. Nature 379, 529–531 (1996).27. Rutherford, M. J. & Devine, J. D. in Fire and Mud: Eruptions and Lahars of Mount Pinatubo, Philippines

(eds Newhall, C. G. & Punongbayan, R. S.) 751–766 (Philippine Inst. Volcanol. & Seismol./Univ.Washington Press, Quezon City/Seattle, 1996).

28. Del Pozzo, A. L. M., Espinasa, R. & Butron, M. A. Popocatepetl’s 1994–1996 activity: eruptionproducts. (abstr.) Eos 77, F809 (1996).

29. Athenosopolis, P. & Larocque, A. C. L. Recent eruptions from Volcan Popocatepetl in context of pasteruptions. (abstr.) 77, F809 (1996).

30. Delgado, H. & Cardenas-Gonzalez, L. (abstr.) Passive degassing at Volcan Popocatepetl (Mexico): 2.6× 106 tons of SO2 released in 617 days of activity. IAVCEI General Assembly Abstracts 49 (Gobierno delEstado de Jalisco, Guadalajara, Jalisco, Mexico, 1997).

31. Whitney, J. A. Fugacities of sulfurous gasses in pyrrhotite-bearing silicic magmas. Am. Mineral. 69,69–78 (1984).

32. Metrich, N. & Clocchiatti, R. Sulfur abundance and its speciation in oxidized alkaline melts. Geochim.Cosmochim. Acta 60, 4151–4160 (1996).

33. Buchanan, D. L. & Nolan, J. Solubility of sulfur and sulfide immiscibility in synthetic tholeiitic meltsand their relevance ot Bushveld-Complex rocks. Can. Mineral. 17, 483–494 (1979).

34. Peach, C. L. & Mathez, E. A. Sulfide melt-silicate melt distribution coefficients for nickel and iron andimplications for the distribution of other chalcophile elements. Geochim. Cosmochim. Acta 57, 3013–3021 (1993).

35. Haughton, D. R., Roeder, P. L. & Skinner, B. J. Solubility of sulfur in mafic magmas. Econ. Geol. 4, 451–467 (1974).

Acknowledgements. I thank M. R. Carroll, B. W. Evans, M. S. Ghiorso, K. Hattori, N. Metrich, B. O.Mysen, B. K. Nelson, C. G. Newhall, J. S. Pallister, D. A. Snyder and P. J. Wallace for criticisms, suggestions,information and reviews.

Correspondence and requests for materials should be addressed to the author (e-mail: kress@u.washington.edu).

Fitness lossandgermlinemutations inbarnswallowsbreeding inChernobylHans Ellegren*, Gabriella Lindgren*, Craig R. Primmer*& Anders Pape Møller†

* Department of Animal Breeding and Genetics, Swedish University ofAgricultural Sciences, Box 597, S-751 24 Uppsala, Sweden† Laboratoire d’Ecologie, CNRS URA 258, Universite Pierre et Marie Curie,Bat. A, 7eme etage, 7, quai St Bernard, Case 237, F-75252 Paris Cedex 5, France. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

The severe nuclear accident at Chernobyl in 1986 resulted in theworst reported accidental exposure of radioactive material to free-living organisms1. Short-term effects on human populationsinhabiting polluted areas include increased incidence of thyroidcancer2, infant leukaemia3, and congenital malformations innewborns4. Two recent studies5,6 have reported, although withsome controversy7,8, that germline mutation rates were increasedin humans and voles living close to Chernobyl, but little is knownabout the viability of the organisms affected9. Here we report anincreased frequency of partial albinism, a morphological aberra-tion associated with a loss of fitness, among barn swallows,Hirundo rustica, breeding close to Chernobyl. Heritability esti-mates indicate that mutations causing albinism were at leastpartly of germline origin. Furthermore, evidence for an increasedgermline mutation rate was obtained from segregation analysis attwo hypervariable microsatellite loci, indicating that mutationevents in barn swallows from Chernobyl were two- to tenfoldhigher than in birds from control areas in Ukraine and Italy.

The phenotypic consequences of mutation range from alteredbehaviour and physiology to aberrant morphology10. Mutationsaffecting external morphology in birds often lead to partial albinismcaused by recessive genes, resulting in partial loss of plumagepigmentation11. Such aberrant coloration is presumed to be detri-mental in terms of costs of increased risks for predation and reducedmating success11. Partial albinism in birds appears as spots of whitefeathers that completely lack pigments. Barn swallows are sociallymonogamous, aerially insectivorous passerine birds of mass ,20 g,with an annual survival rate of about 28% (ref. 12). Like most otherpasserines, barn swallows start to breed at the age of one year.European populations migrate across the Sahara desert, winteringin southern Africa. Adult and nestling barn swallows capturedaround Chernobyl, Ukraine (the radioactively contaminatedarea), Kanev (an uncontaminated control area in Ukraine), andMilan, Italy (an uncontaminated control area outside the Ukraine)in 1991 and 1996, and museum specimens from before 1986, wereinspected for the presence of albinistic feathers in the red front andthe blue head, neck, back and tail. There was a significantly higherfrequency of partial albinism in Chernobyl than in Kanev swallowsin 1991 and 1996 (Fisher exact test; 1991, P ¼ 0:0000016; 1996,P ¼ 0:046), but not before 1986 (P ¼ 1:00; Table 1). Furthermore,there was a significant difference in frequency of partial albinismbetween swallows from Chernobyl and Italy in 1996 (Fisher exact

Nature © Macmillan Publishers Ltd 1997

letters to nature

594 NATURE | VOL 389 | 9 OCTOBER 1997

test; P ¼ 0:000772), but not before 1986 (pre-1986, P ¼ 1:00;Table 1). These results indicate that adult barn swallows had anincreased frequency of partial albinism in Chernobyl after 1986, buta similar temporal change was not observed in the control area inUkraine or in Italy.

Partial albinism appears to be associated with a reduced prob-ability of survival. None of 13 partially albinistic adult barn swallowsin a Danish study population followed during the years 1985 to 1996survived from one year to the next12, whereas 28.0% of 2,110 non-albinistic adults survived (Fisher exact test, P ¼ 0:0284). Survivalrates based on recaptures are reliable for the barn swallow in thiscase because almost all surviving adults are captured annually owingto an intensive capture effort, and breeding dispersal of adults isextremely limited12. A second piece of evidence also suggests afitness reduction among barn swallows in Chernobyl. The size ofthe breeding population assessed from the number of nests witheggs or nestlings during the first clutch (a reliable estimate of the sizeof the breeding population12) decreased from 292 pairs in 1991 to76 pairs in 1996 in the nine villages near Chernobyl (a reductionof 74.0%), but changed rather less, from 202 pairs in 1991 to 162pairs in 1996, in the six villages in the control area in Ukraine (areduction of 19.8%). Population size decreased in all nine villagesnear Chernobyl between 1991 and 1996, whereas only two of sixvillages near Kanev had a decreasing population (Fisher exact test,P ¼ 0:022). Moreover, analyses of developmental instability offeather characters of adult barn swallows around Chernobyl andKanev in 1991 have similarly indicated that the level of fluctuatingasymmetry and the frequency of developmental anomaly wasincreased considerably in the radioactively contaminated areaafter 1986 (ref. 13).

Partial albinism in birds can be caused by either somatic orgermline mutations11. We compared the parent–offspring resem-blance in partial albinism for the sample of birds from Chernobyl in1991 and 1996. Adult barn swallows develop their plumage in the

African winter quarters, whereas nestlings develop their plumage inthe breeding areas. The similarity of the dichotomously classifiedphenotypes across generations can be used to obtain an estimate ofheritability14. Whereas 12.9% of the offspring with normal-pheno-type parents had albinistic feathers (N ¼ 62 families), 84.6% ofthose with partially albinistic parents had such phenotypes (N ¼ 13families). The resemblance was estimated as rðfÞ ¼ 0:62, whichdiffers significantly from zero (x2 ¼ 25:5, d:f : ¼ 1, P , 0:001; ref.15). This result indicates that the partially albinistic phenotypeswere at least partly determined by germline mutations.

Assessment of the germline mutation rate often proves to beextremely difficult because of the very low frequency of spontaneousmutations seen at most genomic loci. However, two hypervariable(heterozygosity .95%) tetranucleotide microsatellite repeat loci,HrU6 and HrU9, associated with mutation rates of as high as 0.5–3.5% per meiosis16 (C.R.P., A.P.M. and H.E., unpublished data) havebeen isolated from the barn swallow genome17,18. We used thesemarkers to genotype family material from Chernobyl, Kanev andItaly (Fig. 1, Table 2). For HrU6 there was a significantly highermutation rate in Chernobyl than in Kanev (x2 ¼ 3:95, d:f : ¼ 1,P ¼ 0:047) and Italy (x2 ¼ 20:1, d:f : ¼ 1, P , 0:001). The rathersmall sample sizes from Chernobyl and Kanev make a quantitativeestimate of the difference in mutation rate between contaminatedand uncontaminated areas only approximate, but it may be ashigh as an order of magnitude. For HrU9, the mutation rateshowed a significant two- to threefold, increase in Chernobylcompared to Italy (x2 ¼ 5:00, d:f : ¼ 1, P ¼ 0:025), whereasthere was no clear difference in the mutation rate between Cher-nobyl and Kanev (x2 ¼ 0:04, d:f : ¼ 1, P ¼ 0:91). Finally, thecombined mutation rate for HrU6 and HrU9 in Chernobyl wassignificantly higher than in Italy (x2 ¼ 16:1, d:f : ¼ 1, P , 0:001)but not Kanev (x2 ¼ 1:50, d:f : ¼ 1, P ¼ 0:22). Because micro-satellite mutation rates may be higher in male than in femalegerm line19, and may intrinsically also be dependent on the repeat

Figure 1 Examples of microsatellite germline mutations for barn swallow loci in

the Chernobyl population. a, HrU6; b, HrU9. Lane 1, the father; lane 2, the mother;

other lanes show offspring. Mutant alleles are arrowed. Note that the offspring to

the left in b is mutant for both its father’s and mother’s allele.

Table 1 Frequency of adult barn swallows with partial albinism

Year Area

Chernobyl Kanev Milan

Albinistic (%) N Albinistic (%) N Albinistic (%) N.............................................................................................................................................................................

Pre-1986 0.0 17 0.0 24 0.0 661991 15.2 99 0.0 1411996 13.3 75 1.9 51 1.7 180.............................................................................................................................................................................The birds are from Chernobyl, Ukraine (radioactively contaminated area in 1986), Kanev(uncontaminated control area in Ukraine), and Milan, Italy (uncontaminated control areaoutside Ukraine). Pre-1986 samples of adult barn swallows were obtained from museumcollections. Samples from 1991 and 1996 were from mistnet catches of birds at breedingsites.

Figure 2 Distribution of changes in the number of repeat units between mutant

and progenitor alleles at the barn swallow microsatellite loci HrU6 and HrU9.

Populations shown are from Chernobyl (white bars) and Italy (black bars).

Nature © Macmillan Publishers Ltd 1997

letters to nature

NATURE | VOL 389 | 9 OCTOBER 1997 595

length of individual alleles16, these factors need to match betweentwo populations to allow proper comparisons of mutation rates.However, there was no indication that the increased mutation ratefor HrU6 and HrU9 in Chernobyl could be explained by populationdifferences in the sex or allele size distributions; the sex ratio(percentage males) among the meioses scored in the Chernobylgroup (39.4%) was actually lower than in the Kanev (43.6%;x2 ¼ 0:48, d:f : ¼ 1, P ¼ 0:48) and Italian (48.3%; x2 ¼ 4:06,d:f : ¼ 1, P ¼ 0:044) populations. Furthermore, there was no sig-nificant difference in the allele frequency distributions betweenChernobyl and Kanev for HrU6 (Kolmogorov–Smirnov,P ¼ 0:80) or for HrU9 (P ¼ 0:82), or between Chernobyl andItaly for HrU6 (P ¼ 0:41) or for HrU9 (P ¼ 0:54).

The relatively small sample sizes from Chernobyl and Kanev mayexplain why we did not observe a significant difference in themutation rate at HrU9 between the two areas. However, only twoof four human minisatellite systems (two single-locus and twomulti-locus probes) had a significantly increased mutation rate in apopulation from Mogilev, Belarus (a radioactively contaminatedarea) compared to a control population from United Kingdom6. Itis not yet clear which other factors may influence microsatelliteand minisatellite instability in response to mutagens. Microsatellitemutant alleles typically differ by just one (mainly) or a few repeatunits in size from the parental allele16,19, a situation often explainedby slipped-strand mispairing during replication20. Mutant allelesfrom Chernobyl followed a similar distribution of differences inthe number of repeat units compared with the parental allele(Fig. 2), and was not significantly different from that observedfor mutations in the Italian16 study population (Kolmogorov–Smirnov, P ¼ 0:34). If slippage was also the main mechanism forthe observed Chernobyl mutations, it is not likely that this couldbe associated directly with radiation damage, as ionizationgenerally induces double-strand breaks21. Another possibility isdamage in repair-systems genes, and in yeast the slippage rate ofsimple repeats increases dramatically in strains deficient inDNA mismatch repair22. Mutations in mismatch repair genesassociated with colorectal cancer and other carcinomas inhuman and mouse also lead to an increased instability of micro-satellites23–25.

Thus both genetic and morphological data suggest that barnswallows breeding in the Chernobyl area have suffered elevatedlevels of germline mutation rates at both expressed (yieldingalbinism) and non-coding (microsatellite) loci from the 1986release of radioactive material from the nuclear power plant. Thisseems to have resulted in a loss of fitness among individuals in thebreeding population, and may also be associated with a significantdecline in population size between 1986 and 1996. We have noreason to believe that the barn swallow should be uniquely sensitiveto radiation among animal species breeding close to Chernobyl or inother radioactively contaminated areas. M. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Methods

Fieldwork. A.P.M. captured barn swallows at farms near Chernobyl (518179 N,308139 E), Ukraine, at a distance of 25–50 km from the nuclear power plant,and around Kanev (498429 N, 318259 E), Ukraine, 100 km southeast of Kiev,

outside the contamination zone, on 10–29 June 1991 and 16–24 June 1996.N. Saino captured barn swallows at farms near Milan, Italy, during April–July1996. The Chernobyl area had an atmospheric level of radiation of 300–500 mR,which is considerably above the background level (A. A. Tokar, personalcommunication). Adult barn swallows were captured in mist nets, measured,inspected for partial albinism, ringed and provided with an individualcombination of colour on their white breast and belly feathers. A blood sampleof 100 ml was collected from the brachial vein of each individual. Adults wereassigned to their nests using observations of birds feeding nestlings, and bloodsamples of these nestlings were subsequently collected and their plumageinspected for the presence of partial albinism. A.P.M. inspected museumspecimens of adult barn swallows from the breeding season in the Natural-History Museums in Kiev, Ukraine and Milan, Florence and Rome, Italy, for thepresence of partially albinistic feathers in the plumage. Ukrainian museumsamples were divided into birds collected from an area around and northwest ofChernobyl, within the subsequently most heavily contaminated areas,including the area from which field data were sampled in 1991 and 1996, andbirds from an area southeast of Kiev (that includes Kanev), well outside thecontaminated area13. There should be no reason to expect the age distributionamong adults to differ between shot birds and breeding birds trapped by us, sothe origin and age of pre-1986 Chernobyl birds should match the 1991 and1996 samples. Moreover, because Ukrainian museum samples consisted ofadult birds killed in the field using a shotgun, there is no possibility that thesamples were biased with respect to partial albinism. Partial albinism generallyinvolves less than 10 small, white body feathers, and as such is visible only atclose range or when the bird is held in the hand.DNA work. Genomic DNA was prepared by Chelex extraction from about 2 mlof whole blood. Of the DNA preparations, 0.1–0.5% were used as template in10 ml PCR reactions with HrU6 or HrU9 containing 1 pmol of each primer (oneend-labelled with [g-32P]ATP), 1.5 mM MgCl2, 50 mM KCl, 10 mM Tris,200 mM dNTP and 0.5 U Taq polymerase (Dynazyme). After initial denatura-tion at 90 8C for 3 min, 30 cycles of 94 8C for 30 s, 63 8C for 30 s and 72 8C for80 s were run. PCR products were separated in 6% polyacrylamide gels andvisualized by autoradiography. Known size standards were included at regularintervals on the gels. Because the allele size range of HrU6 and HrU9 spansseveral hundred base pairs, extended electrophoresis was performed to achieveappropriate separation of long alleles. Identification of mismatching alleles wasfacilitated by the fact that HrU6 and HrU9 are both tetranucleotide repeats, andso most alleles differ by at least four base pairs in size. An offspring allele notcompatible with the alleles of either of its presumed parents could indicateeither a mutation event (relevant both for maternal and paternal transmission)or false parentage (extra-pair paternity, EPP). Distinguishing between thesetwo events is critical to this study because EPP is not uncommon in barnswallows26. To identify true genetical relationships accurately, an extendedpaternity analysis was therefore conducted by genotyping a set of eightadditional, polymorphic microsatellites HrU2-5, HrU7-8 (ref. 17), HrU10(ref. 18) and Escm6 (ref. 27). In 10 cases for which one or two alleles ateither HrU6 or HrU9 did not match with the presumed mother and/or father,but that of the other locus did, the segregation at all additional eight markerswas compatible with a correct maternity and paternity assignment, stronglyindicating that the mismatch at HrU6 or HrU9 was due to mutation rather thanEPP. According to observed allele frequency distributions, the mean probabilitythat a random individual would match at least 9 of our 10 loci is as low as0.0005 (95% confidence interval of 0.001–0.0001). In the sample of 1996Ukrainian offspring, we would have expected to find only 0.04 birds where anallelic mismatch at one of the 10 loci would be due to EPP rather thanmutation. We consider this possible effect negligible, and conclude that 10microsatellite mutations were present in the sample. For 24 offspring morethan one marker did not match with the presumed father (one case with fourmarkers excluding the associated male, five cases with five markers, nine caseswith six markers, four cases with seven markers and five cases with eightmarkers), and these offspring were inferred as instances of EPP. The frequencyof such instances in the Ukrainian material (30%) was similar to that in ourstudy populations in Italy26 and Denmark17.

Received 3 June; accepted 4 August 1997.

1. Anspaugh, L. R., Catlin, R. J. & Goldman, M. The global impact of the Chernobyl reactor accident.Science 242, 1513–1519 (1988).

Table 2 Frequency of germline mutations at two microsatellite loci

Area Locus

HrU6 HrU9 HrU6 + HrU9

Mutations (%) N Mutations (%) N Mutations (%) N.............................................................................................................................................................................

Chernobyl 5.6 72 9.1 66 7.2 138Kanev 0.0 69 8.0 62 3.8 131Milan 0.5 1,065 3.6 937 2.0 2,002.............................................................................................................................................................................Adult barn swallows are from Chernobyl, Ukraine (radioactively contaminated area), Kanev(uncontaminated control area in Ukraine) and Milan, Italy (uncontaminated control areaoutside Ukraine). N, number of meioses examined.

Nature © Macmillan Publishers Ltd 1997

letters to nature

596 NATURE | VOL 389 | 9 OCTOBER 1997

2. Kazakov, V. S., Demidchik, E. P. & Astakhova, L. N. Thyroid cancer after Chernobyl. Nature 359, 21(1992).

3. Petridou, E. et al. Infact leukaemia after in utero exposure to radiation from Chernobyl. Nature 382,352–353 (1996).

4. Lazjuk, G. I., Kirillova, I. A., Dubrova, Y. E. & Novikova, I. V. in The Chernobyl Papers Vol. 1, Doses tothe Soviet Population and Early Health Effects Studies (eds Mervin, S. E. & Balanov, M. I.) 385–397(Research Enterprises, Richland, WA, 1993).

5. Baker, R. J. et al. High levels of genetic change in rodents of Chernobyl. Nature 380, 707–708 (1996).6. Dubrova, Y. E. et al. Human minisatellite mutation rate after the Chernobyl accident. Nature 380,

683–686 (1996).7. Dubrova, Y. E. et al. Effects of radiation on children (reply). Nature 383, 226 (1996).8. Satoh, C. & Kodaira, M. Effects of radiation on children. Nature 383, 226 (1996).9. Hillis, D. M. Life in the hot zone around Chernobyl. Nature 380, 665–666 (1996).10. Auerbach, C. Mutation Research (Chapman & Hall, London, 1976).11. Dyck, J. in A Dictionary of Birds (eds Campbell, B. & Lack, E.) 472–474 (Poyser, Calton, 1985).12. Møller, A. P. Sexual Selection and the Barn Swallow (Oxford Univ. Press, 1994).13. Møller, A. P. Morphology and sexual selection in the barn swallow Hirundo rustica in Chernobyl,

Ukraine. Proc. R. Soc. Lond. B 252, 51–57 (1993).14. Falconer, D. S. Introduction to Quantitative Genetics 2nd edn (Longman, New York, 1989).15. Siegel, S. & Castellan, N. J. Jr Nonparametric Statistics for the Behavioral Sciences 2nd edn (McGraw-

Hill, New York, 1988).16. Primmer, C. R., Ellegren, H., Saino, N. & Møller, A. P. Directional evolution in germline microsatellite

mutations. Nature Genet. 13, 391–393 (1996).17. Primmer, C. R., Møller, A. P. & Ellegren, H. Resolving genetic relationships with microsatellite

markers: a parentage testing system for the swallow Hirundo rustica. Mol. Ecol. 4, 493–498 (1995).18. Primmer, C. R., Møller, A. P. & Ellegren, H. New microsatellite markers from the pied flycatcher

Ficedula hypoleuca and barn swallow Hirundo rustica genomes. Hereditas 124, 281–283 (1996).19. Weber, J. L. & Wong, C. Mutation of human short tandem repeats. Hum. Mol. Genet. 2, 1123–1128

(1993).20. Levinson, G. & Gutman, G. A. Slipped-strand mis-pairing: a major mechanism for DNA sequence

evolution. Mol. Biol. Evol. 4, 203–221 (1987).21. Ward, J. F. The yield of DNA double-strand breaks produced intracellularly by ionizing radiation: a

review. Int. J. Radiat. Biol. 57, 1141–1150 (1990).22. Strand, M., Prolla, T. A., Liskay, R. M. & Petes, T. D. Destabilization of tracts of simple repetitive DNA

in yeast by mutations affecting DNA mismatch repair. Nature 365, 274–276 (1993).23. Parsons, R. et al. Hypermutability and mismatch repair deficiency in RER+ tumor cells. Cell 75, 1227–

1236 (1993).24. Fishel, R. et al. Binding of mismatched microsatellite DNA sequences by the human MSH2 protein.

Science 266, 1403–1405 (1994).25. Reitmair, A. H. et al. MSH2 deficient mice are viable and susceptible to lymphoid tumours. Nature

Genet. 11, 64–70 (1995).26. Saino, N., Primmer, C. R., Ellegren, H. & Møller, A. P. An experimental study of paternity and tail

ornamentation in the barn swallow (Hirundo rustica). Evolution 51, 562–570 (1997).27. Hanotte, O. et al. Isolation and characterization of microsatellite loci in a passerine bird: the reed

bunting Emberiza schoeniclus. Mol. Ecol. 3, 529–530 (1994).

Acknowledgements. We thank A. A. Tokar for logistic support in Ukraine; the curators of museumcollections in Kiev, Ukraine, and Milan, Florence and Rome, Italy for access to specimens; and N. Sainofor providing phenotypic data from his barn swallow population in 1996. A.P.M. was supported by theDanish Natural Science Research Council, H.E. by the Swedish Research Councils for Natural Sciences,and for Agriculture and Forestry and G.L. by the Elis Wides Fund (Swedish Ornithological Society).

Correspondence and requests for materials should be addressed to H.E. or A.P.M. (respective e-mailaddresses: hans.ellegren@bmc.uu.se or anders.moller@snv.jussieu.fr).

Howthebrain learnstoseeobjectsand faces inanimpoverishedcontextR. J. Dolan*†, G. R. Fink*, E. Rolls‡, M. Booth‡,A. Holmes*, R. S. J. Frackowiak* & K. J. Friston*

* Wellcome Department of Cognitive Neurology, Institute of Neurology,Queen Square, London W1N 3BG, UK† Academic Department of Psychiatry, Royal Free Hospital School of Medicine,London NW3, UK‡ Department of Experimental Psychology, University of Oxford,Oxford OX1 3UD, UK

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

A degraded image of an object or face, which appears meaninglesswhen seen for the first time, is easily recognizable after viewing anundegraded version of the same image1. The neural mechanismsby which this form of rapid perceptual learning facilitates percep-tion are not well understood. Psychological theory suggests theinvolvement of systems for processing stimulus attributes, spatialattention and feature binding2, as well as those involved in visualimagery3. Here we investigate where and how this rapid percep-tual learning is expressed in the human brain by using functionalneuroimaging to measure brain activity during exposure to

degraded images before and after exposure to the correspondingundegraded versions (Fig. 1). Perceptual learning of faces orobjects enhanced the activity of inferior temporal regionsknown to be involved in face and object recognition respec-tively4–6. In addition, both face and object learning led toincreased activity in medial and lateral parietal regions thathave been implicated in attention7 and visual imagery8. Weobserved a strong coupling between the temporal face area andthe medial parietal cortex when, and only when, faces wereperceived. This suggests that perceptual learning involves directinteractions between areas involved in face recognition andthose involved in spatial attention, feature binding and memoryrecall.

The experimental design involved four conditions: two-tone(black and white) images of objects before (Ob) and after (Oa)exposure to the associated grey-scale images of the objects; two-toneimages of faces before (Fb) and after (Fa) exposure to the associatedgrey-scale images of the faces. A factorial experimental design,crossing pre- and post-grey-scale exposure with object or faceperception, ensured that the same visual input was maintainedacross all levels of stimulus presentation, with the condition-specificneural responses being measured with positron emission tomogra-phy (PET) indices of local perfusion. Thus, the factors in theexperiment are exposure (two levels: before and after learning)and stimulus (two levels: object and face). Each combination oflevels was repeated 3 times on 8 subjects (study 1; n ¼ 96) and 6subjects (study 2; n ¼ 72) as described below.

In study 1, we first examined the main effect of exposure using thecontrast ðOa þ FaÞ 2 ðOb þ FbÞ to identify regions activated by thecombined detection of objects and faces in study 1. This comparisonrevealed extensive bilateral activation in the medial parietal region(P , 0:05, corrected), corresponding to the precuneus, and extend-ing laterally to involve the posterior inferior parietal cortex(P , 0:05, corrected; Fig. 2).

We next examined the main effect of stimulus to localizecategory-specific activations. Comparing object with face condi-tions, using the contrast ðOb þ OaÞ 2 ðFb þ FaÞ, showed activa-tions in the left inferior temporal region (P , 0:05, corrected) (Fig.3a). The opposite contrast ðFb þ FaÞ 2 ðOb þ OaÞ revealed activa-tion in the right superior temporal region (P , 0:05, corrected)extending into the right inferior temporal region (P , 0:001,uncorrected) in association with the face conditions (Fig. 3b).Finally, to identify regions whose object-specific responses wereenhanced by exposure-dependent perception, we tested for theconjunction of object-specific responses (that is, the main effectof objects relative to faces, which equals ðOb þ OaÞ 2 ðFb þ FaÞ)and significantly increased perception-dependent activations (thatis, category 3 exposure interaction ¼ ðOb þ OaÞ 2 ðFb þ FaÞ). Inthis analysis, learning-dependent effects due to object perceptionwere seen in the left fusiform region (z ¼ 4:4; P , 0:05 corrected)(Fig. 3c). The equivalent conjunction analysis for learning-depen-dent effects due to faces demonstrated a similar effect in the rightfusiform gyrus (Fig. 3d) (z ¼ 3:22; P , 0:001, uncorrected).

The experimental design interposes an explicit learning phasebetween first and second stimulus presentation and includes alearning component due to repeated exposure to the same stimuli.To remove the influence of this order-effect learning, we contrastedthe patterns of activation associated with explicit learning (study 1)with those elicited in a second experiment (study 2) by interposingnon-associated grey-scale images. These contrasts involved agroup 3 effects interaction where the effects comprised those asso-ciated with category-independent learning (combined post-expo-sure minus pre-exposure) and those specific to learning in the facesand objects conditions (post-exposure minus pre-exposure for facesminus objects and vice versa). The comparisons reach significanceat an uncorrected level and are reported descriptively. Thegroup 3 exposure interaction (category-independent learning)