438 NOTES AND COMMENTS

1983. Organizationof ribosomal DNA in yellow plicationto Viciafaba 5.8S ribosomal RNA. Nucl. lupine(Lupinus luteus) and sequenceof the 5.8S Acids Res. 8:1259-1272. RNA gene. FEBS Lett. 152:241-246. TRECo,D., E. BROWNELL, AND N. ARNHEIM. 1982. REEDER, R. H. 1984. Enhancersand ribosomal gene The ribosomalgene nontranscribed spacer, pp. 101- spacers. Cell 38:349-351. 126. In H. Busch and L. Rothblum(eds.), The Cell ROGERS,S. O., S. HONDA,AND A. J. BENDICH. 1986. Nucleus, Vol. XII. Academic Press, N.Y. Variation in the ribosomal RNA genes among in- TURNER, M. E.,J. C. STEPHENS, AND W. W. ANDERSON. dividuals of Viciafaba. Plant Molec. Biol. 6:339- 1982. Homozygosityand patch structurein plant 345. populations as a result of nearest-neighborpolli- SAGHAI-MAROOF,M. A., K. M. SOLIMAN,R. A. JOR- nation. Proc. Nat. Acad. Sci. USA 79:203-207. GENSEN,AND R. W. ALLARD. 1984. Ribosomal WRIGHT, S. 1943. An analysis of local variabilityof spacer lengthpolymorphisms in barley:Mendelian flowercolor in Linanthusparryae. Genetics 28:139- inheritance,chromosomal location, and population 156. dynamics.Proc. Nat. Acad. Sci. USA 81:8014-8018. YAKURA, K., A. KATO, AND S. TANIFUJI. 1983. Struc- ScHAAL, B. A. 1974. Isolation by distance in Liatris tural organizationof ribosomal DNA in four Tril- cylindracea.Nature 252:703. lium and Paris verticillata.Plant Cell Phys- SOKAL, R. R., AND F. J. ROHLF. 1981. Biometry. iol. 24:1231-1240. Freeman, San Francisco, CA. 1984. Lengthheterogeneity of thelarge spacer SYTSMA, K. J., AND B. A. SCHAAL. 1985. Phyloge- of Viciafaba rDNA is due to the differingnumber netics of the Lisianthiusskinneri (Gentianaceae) of a 325 bp repetitivesequence elements. Molec. species complex in Panama utilizingDNA restric- Gen. Genet. 193:400-405. tion fragmentanalysis. Evolution 39:594-608. TANAKA,Y., T. A. DYER, AND G. G. BROWNLEE. 1980. An improveddirect RNA sequence method:Its ap- CorrespondingEditor: D. R. Cavener

Evolution,41(2), 1987, pp. 438-441

REPEATED GEOGRAPHIC VARIATION AT THREE ENZYME LOCI IN THE AMPHIPOD PLATENSIS

JOHN H. McDoNALD Departmentof Ecology and Evolution,State UniversityofNew Yorkat StonyBrook, StonyBrook, NY 11794

Received September8, 1986. Accepted November 10, 1986

A correlationof allelefrequency with an environ- notransferase(ALAT, E.C. 2.6.1.2), glucose-6-phos- mentalfactor is one of the mostcommon forms of phateisomerase (GPI, E.C. 5.3.1.9),and mannose-6- evidencefor natural selection on allozymeloci. How- phosphateisomerase (MPI, E.C. 5.3.1.8). ever,an apparentassociation of allelefrequency with Horizontalstarch gels containing 10.5% starch were a singleenvironmental gradient, such as a latitudinal usedfor electrophoresis. The pH 8.6 Tris-borate-EDTA clinecorrelated with temperature, may also result from buffersystem of Nichols and Ruddle(1973) was used randomdrift in isolatedpopulations followed by mi- forall threeenzymes. The stains used are those of Shaw gration.One wayto testthe drift hypothesis is to ex- andPrasad (1970) for GPI, Nicholsand Ruddle (1973) amineareas of repeatedenvironmental variation for forMPI, andthe fluorescent glutamate pyruvate trans- repeatedpatterns of genetic differentiation, since ran- aminasestain of Harrisand Hopkinson(1976) for dom processeswould not producea consistentasso- ALAT. Eachamphipod was ground whole in an equal ciationof a particularallele with a particularenviron- volumeof 0.05 M Tris-HCl,pH 8. Amphipodswere ment.Examples of such repeated patterns include six capturedby hand from under rocks and seaweed in the enzymeloci in Drosophila melanogaster,in whichlat- upperintertidal area of beaches.The averagesample itudinalclines of allele frequency are present on three size was 49 individuals. continents(Oakeshott et al., 1981,1982, 1983, 1984). Eachlocus had two common alleles. There were also At six loci in Fundulus heteroclitus,latitudinal clines two rareGpi alleles,with mean frequenciesover all arepresent in both Chesapeake Bay and on theAtlantic locationsof 0.031 and 0.011, and one rareMpi allele coast(Powers et al., 1986).In Mytilusedulis, clines at witha meanfrequency of 0.0005. The commonallo- theLap locusare present in Long Island Sound (Koehn zymewhich migrated furthest toward the anode was et al., 1976) and twosmaller estuaries (Boyer, 1974). designated100. HereI reportrepeated differentiation between different For each locus,several pairs of locations were sur- habitatsin the amphipod Platorchestiaplatensis at loci veyedinitially. The locationswithin each pair differed codingfor three polymorphic enzymes: alanine ami- fromeach otherin a subjectiveassessment of habitat NOTES AND COMMENTS 439

CONNECTICUT 1 b Im 2e*,. le k Ns NEW YORK

3 m~~~~~~3

A6 Nevv l seYlRA6 4m4m~~~~~~~~4 X

0 30k~~~~~~~~~~~

so~~9 .. 8x 7x

20p~~~~~~~~~4 0 30 km ~ ~ ~ ~ ~ ~ ~ ~~ 1k

FIG. 1. Locationsof P. platensis samples.Both insets use thesame scale. Letters indicate the type of beach: e, estuanine,m, manine,b, bay;o, ocean;p, protected;x, exposed. type.A hypothesiswas formulatedto explainthe re- gestedthat the Gpi'00 allele might be selectedagainst sultsof this initial survey, and this new hypothesis was on theshores of shelteredbodies of water, where the thentested with a newset of pairs of locations. A sign climateis warmerin summerand cooler in winterthan test(Sokal and Rohlf,1981 pp. 449-450) was used on ocean beaches.To testthis hypothesis, eight bay- becausethe null hypothesisis thatany one pair of oceanpairs were surveyed (locations 5-12). Bay sam- locationsmight differ in allele frequency due to random ples werecollected on the shoresof bays and large drift,but the direction of difference would be random. estuaries,as faras possiblefrom the ocean, while ocean A one-tailedtest was used because a specificaltemative sampleswere collectedon or near AtlanticOcean hypothesis,predicting which allele would be associated beaches.At sevenof these eight pairs of beaches,the witheach habitat,was formulatedbefore the second Gpi100allele was indeed less common at the bay site setof samples was collected.Because subjective, qual- (Fig.2b), which is significantlydifferent from the null itativehabitat classifications were used, it is important expectation(one-tailed sign test, P = 0.035). to emphasizethat the pairs of locationswere chosen At theAlat locus, the initial survey consisted of the and theirhabitat types were determined before the al- fourestuarine-marine pairs and the first five bay-ocean lozymedata were collected. pairs.At thatpoint the Alat'00 allele was less common The initialsurvey of variation at theGpi locus con- at theestuarine and baybeaches in sevenof the nine sistedof fourestuarine-marine pairs (locations1-4, pairs(Fig. 2c). The moststriking exception was Del- Fig. 1).At each pair, an estuarinesample was collected awareBay (sites9o and 9b). The beachwithin Dela- at a beachon theestuarine portion of a river,and a wareBay, like other bay beaches, is probablywarmer marinesample was collectedon a beach outsidethe in summer,cooler in winter,and lowerin salinitythan mouthof the river. A consistentgeographic pattern of the nearbyocean beach. However,unlike other bay Gpi allele frequencieswas not evidentwithin these beaches,where the P. platensishabitat is generallymud pairsof locations (Fig. 2a). However,the beach most andfine sand, the Delaware Bay beach consists of coarse exposedto the moderatingclimatic influence of the sand,gravel, and largerocks. In thisrespect the Del- AtlanticOcean, site 2m, had a higherfrequency of the awareBay beachis similarto P. platensishabitat on Gpi'00allele thanbeaches inside Long Island Sound ocean beaches.This suggestedthat the Alat'00 allele (site4m) and Fishers Island Sound (site I m). Thissug- maybe selectedagainst in protectedareas, which gen- 440 NOTES AND COMMENTS

.35 _ .30 - a b .25- .20- ~.15- .10- .05- 0 1 2 3 4 5 6 7 8 9 10 11 12

.50 c d .40-

.30-

.20-

.10

0_

1 2 3 4 5 6 7 8 9 13 14 15 16 17 18 19 20 .90 e f .80- .70- .60 .50 .40 .30- 1 2 3 4 5 6 7 8 9 13 14 15 16 17 18 19 20 location location FIG. 2. Allelefrequencies atthe locations shown in Figure 1. For each pair of locations, the solid bar represents themarine, ocean, or exposedbeach; and thestriped bar representsthe estuarine, bay, or protectedbeach. a) Gpi,initial survey. b) Gpi,bay-ocean pairs. * Indicatesthat the frequency of Gpi'00is 0. c) Alat,initial survey. d) Alat,protected-exposed pairs. e) Mpi,initial survey. f) Mpi, protected-exposedpairs. erallyhave muddiersediment than beaches exposed againston protectedbeaches was testedat the eight to morewave action.To testthis hypothesis, eight smallbays on LongIsland Sound, and at sevenof these protected-exposedpairs on the northshore of Long theMpi'00 allele was less commonon the protected Islandwere surveyed (locations 13-20, Fig. 1). Exposed beach(Fig. 2f; P = 0.035,one-tailed sign test). beacheson LongIsland Sound consist of coarsesand One explanationfor the unusual Alat and Mpi allele and gravel,while protected beaches on thesmall bays frequenciesat thesite inside Delaware Bay could be are muchmuddier. At sevenof theeight pairs of lo- inadvertentsampling of a secondspecies. The amphi- cations,the Alat'00 allele was lesscommon in thepro- pods fromsites 9e and 9p wererun for an additional tectedsite (Fig. 2d), a significantdifference (one-tailed fiveenzyme systems (aspartate aminotransferase, sign test,P = 0.035). malatedehydrogenase, peptidase, phosphoglucomu- The patternat theMpi locuswas similarto thatat tase,and phosphogluconatedehydrogenase), yielding Alat.At six of the first nine pairs of locations, the Mpi'?' eightisozymes. The twopopulations were monomor- allelewas lesscommon at theestuarine and baysites, phicfor the same allele at all eightloci; because cryptic againwith the striking exception of Delaware Bay (Fig. speciesin othergenera of amphipods have been easily 2e). The hypothesisthat the Mpi'00 allele is selected distinguishedusing allozymes (Bulnheim and Scholl, NOTES AND COMMENTS 441

1981;Siegismund et al., 1985;McDonald, unpubl.), it semi-terrestrialamphipod Orchestia traskiana in is unlikelythat more than one specieswas present. an expanded habitaton Santa Cruz Island, pp. 395- Small-scalegeographic variation in allele frequencies 401. In D. M. Power (ed.), The CaliforniaIslands: has been foundin severalamphipod species. Four Proceedings of a MultidisciplinarySymposium. speciesof Gammarusdiffer in allelefrequency at sev- Santa Barbara Museum of Natural History,Santa eral loci, includingGpi, Alat, and Mpi, betweenthe Barbara, CA. estuarineBaltic Sea andthe more marine Kattegat (Sie- HARRIS,H., AND D. A. HOPKINSON. 1976. Handbook gismund,1985; Siegismundet al., 1985). Traskor- of Enzyme Electrophoresisin Human Genetics. El- chestiatraskiana differs in Mpi and amylaseallele fre- sevier,N.Y. quencybetween two exposed beaches and nearby KOEHN,R. K., R. MILKMAN, AND J.B. MITTON. 1976. brackishponds (Busath,1980). Megalorchestiacali- Population geneticsof marine pelecypods. IV. Se- fornianadiffers in Gpi allelefrequency at one of two lection,migration and geneticdifferentiation in the smallestuaries (McDonald, 1985). Amphipodscarry blue mussel Mytilusedulis. Evolution 30:2-32. theiryoung in a broodpouch, unlike most marine or- McDONALD, J. H. 1985. Size-relatedand geographic ganisms,which have planktonic larvae. Because of the variationat two enzymeloci in Megalorchestiacal- resultinglimited dispersal abilities of amphipods and iforniana (: ). Heredity 54: thepatchy distribution of each species'habitat, all of 359-366. theseexamples of differentiation could be theresult of NICHOLS, E. A., AND F. H. RUDDLE. 1973. A review randomdrift. Both the young and adults of P. platensis of enzymepolymorphism, linkage and electropho- arepoor swimmers, like other talitrid amphipods (Vo- reticconditions for mouse and somatic cell hybrids gel,1985), further increasing the likelihood of random in starchgels. J. Histochem. Cytochem.21:1066- driftin isolated populations. However, the consistently 1081. repeateddifferences in allele frequencyfound at the OAKESHOTT, J.G., G. K. CHAMBERS,J. B. GIBSON, AND Alat, Gpi, and Mpi loci in P. platensissuggest that D. A. WILLCOCKS.1981. Latitudinalrelationships differentialselection is affectingeach of theseloci in of esterase-6 and phosphoglucomutasegene fre- thisspecies, either directly or through a locus in linkage quencies in Drosophila melanogaster.Heredity 47: disequilibrium. 385-396. At thispoint, the environmental factors which may OAKESHOTT, J.G., G. K. CHAMBERS, J.B. GIBSON, W. be responsiblefor the selection at each locusare un- F. EANES, AND D. A. WILLCOCKS. 1983. Geograph- certain.Because many of the environmentalfactors ic variation in G6pd and Pgd allele frequenciesin whichvary among beaches are correlatedwith each Drosophila melanogaster.Heredity 50:67-72. other,the subjective criteria used to classifybeaches OAKESHOTT, J.G., J.B. GIBSON, P. R. ANDERSON, W. inthis study need not be ofselective significance. When R. KNIBB, D. G. ANDERSON, AND G. K. CHAMBERS. comparedwith a beachexposed to themoderating in- 1982. Alcohol dehydrogenase and glycerol-3- fluenceof theocean or Long IslandSound, a bay or phosphatedehydrogenase clines in Drosophila mel- protectedbeach is likelyto be warmerin summer, anogasteron differentcontinents. Evolution 36:86- colderin winter,more variable in temperaturediur- 96. nally,and to have lowerand morevariable salinity. OAKESHOTT, J. G., S. W. McKECHNIE, AND G. K. The coarsesand and gravelof exposedbeaches may CHAMBERS. 1984. Population genetics of the met- provideless protectionfrom fluctuating temperature abolically related Adh, Gpdh and Tpi polymor- and salinitythan the finersediments of protected phisms in Drosophilamelanogaster. 1. Geographic beaches;however, amphipods burrowed in finesedi- variationin Gpdh and Tpi allele frequenciesin dif- mentsmay be at increasedrisk of low oxygenstress. ferentcontinents. Genetica 63:21-30. P. platensiseat drift algae and grasses, which may differ POWERS, D. A., I. ROPSON, D. C. BROWN, R. VAN in relativeabundance between habitats. There are BENEDEN, R. CASHON, L. I. GONZALEZ-VILLASENOR, probablyother, less obvious environmentalfactors AND J. A. DIMICHELE. 1986. Genetic variation in whichalso differamong beaches and mightcause se- Fundulusheteroclitus: Geographic distribution. lectionat enzymeloci. Amer. Zool. 26:131-144. SHAW, C. R., AND R. PRASAD. 1970. Starchgel elec- ACKNOWLEDGMENTS trophoresisof enzymes-A compilationof recipes. Biochem. Genet. 4:297-320. I thank J. W. Ajioka and M. W. Pumell for able SIEGISMUND, H. R. 1985. Genetic studies of assistancein the field,and S. Ferson and R. K. Koehn Gam- marus.II. Geographical variation at polymorphic forconstructive comments on the manuscript.This is enzymeloci in Gammarus salinus and Gammarus contributionnumber 615 in Ecology and Evolution at oceanicus. Hereditas 102:15-23. the State Universityof New York, StonyBrook, NY. SIEGISMUND, H. R., V. SIMONSEN, AND S. KOLDING. 1985. Genetic studies of Gammarus. I. Genetic LITERATURE CITED differentiationof local populations. Hereditas 102: BOYER, J. F. 1974. Clinaland size-dependentvari- 1-13. ation at the LAP locus in Mytilusedulis. Biol. Bull. SOKAL, R. R., AND F. J. ROHLF. 1981. Biometry. 147:535-549. Freeman, San Francisco, CA. BULNHEIM, H.-P.,AND A. SCHOLL. 1981. Electropho- VOGEL, F. 1985. Das Schwimmen der Talitridae retic approach to the biochemical systematicsof (Crustacea, Amphipoda): Funktionsmorphologie, gammarids.Helgoliinder Meeresunters. 34:391-400. Phanomenologie und Energetik. Helgoliinder BUSATH, A. L., 1980. Geneticdifferentiation of the Meeresunters.39:303-339.