Genetic Differentiation Within andBetween DrosophilaGenetic Differentiation Within andBetween Species ofthe Drosophila ... The pattern of genetic variation ... Our techniques of starch-gel

Proc. Nat. Acad. Sci. USAVol. 71, No. 3, pp. 999-1003, March 1974

Genetic Differentiation Within and Between Species of the Drosophilawillistoni Group*

(natural selection/evolution/adaptation/isozymes/Caribbean islands)

FRANCISCO J. AYALA AND MARTIN L. TRACEY

Department of Genetics, University of California, Davis, Calif. 95616

Communicated by Theodosius Dobzhansky, October 2, 1973

ABSTRACT We describe allelic variation at 28 loci insix Caribbean populations of four sympatric species ofDrosophila. Within any one species the allelic frequenciesare very similar from population to population, althoughthere is evidence of local as well as regional genetic dif-ferentiation. The genetic distance is greater betweenpopulations from different islands than between popula-tions of the same island. When the allelic frequencies arecompared between different species, a remarkable patternappears. In any pair of species nearly half of the loci haveessentially identical allelic frequencies, while nearly theother half of the loci have different alleles and in differentfrequencies. The loci with nearly identical allelic fre-quencies are different when different pairs of species arecompared. The patterns of allelic variation within andbetween species are inconsistent with the hypothesisthat the variation is adaptively neutral. Migration ormutation cannot explain the patterns of genetic variation,either. Balancing natural selection is the main processmaintaining protein polymorphisms in natural popula-tions.

Four species of the Drosophila willistoni group are sympatricin the Greater Antilles. Three species are siblings, D. willistoniSturtevant, D. tropicalis Burla and de Cunha, and D. equi-noxialis caribbensis Ayala. The fourth, D. nebulosa Sturtevant,is a close relative of the other three, but can be easily distin-guished from them by external morphology. We have studiedallelic variation in 28 genes coding for enzymes in samples ofthese species collected in two localities in Hispaniola, three inPuerto Rico, and one in St. Kitts, one of the Lesser Antilles.We report our results concerning the amount of genetic varia-tion within local populations, between local populations in thesame island, between populations of different islands, andbetween different species. The pattern of genetic variationwithin and between species is inconsistent with the notionthat the variation is adaptively neutral.

MATERIALS AND METHODS

Flies of the D. willistoni group were collected over bananabaits and naturally occurring decaying fruits in late Februaryand early March, 1972, in the following localities. DominicanRepublic: Santiago, about 10 km NE of that city, on the foot-hills of the Cordillera Septentional; Santo Domingo, about25 km NW of that city, on the southeastern slopes of theCordillera Central; Puerto Rico: Mayagfiez, in the westernend of the island on the grounds of the U.S.D.A. ExperimentalStation; Barranquitas, 2-5 km from that town, in a smallforest (less than 0.01 kM2) within the grounds of the home ofDr. Roberto Aponte, M.D., on the Cordillera Central;Yunque, in the rain forest known by that name, near the

place called La Mina, Sierra de Luquillo, near the easterntip of the island. St. Kitts: on a small gallery forest above thetown of Canyon, on the southeastern slopes of Mt. Misery.D. willistoni was collected in all six localities; D. tropicalis andD. e. caribbensis in all but St. Kitts; D. nebulosa in all exceptYunque.Our techniques of starch-gel electrophoresis and enzyme

assay are described elsewhere (1). The 28 gene loci and theenzymes they code for are as follows: esterases (EC 3.1.1.2),six loci (Est-2, Est-S, Est-4, Est-5, Est-6, Est-7); TPN+-de-pendent malate dehydrogenase (EC 1.1.1.40), two loci (Me-i,Me-2); adenylate kinase (EC 2.7.4.3), two loci (Adk-1, Adk-2);hexokinase (EC 2.7.1.1), three loci (Hk-1, Hk-2, Hk-3); andone locus for each of the following: leucine aminopeptidase(EC 3.4.1.1) (Lap-5), alkaline phosphatase (EC 3.1.3.1)(Aph-1), acid phosphatase (EC 3.1.3.2) (Acph-1), aldolase(EC 4.1.2.13) (Ald), alcohol dehydrogenase (EC 1.1.1.1)(Adh), malate dehydrogenase (EC 1.1.1.37) (Mdh-2), a-glyc-erophosphate dehydrogenase (EC 1.1.9.5) (aGpdh), isoci-trate dehydrogenase (EC 1.1.1.41) (Idh), glyceraldehyde-3-phosphate dehydrogenase (EC 1.1.1.8) (G3pdh), octanol de-hydrogenase (EC 1.1.1.1) (Odh-1), xanthine dehydrogenase(EC 1.2.3.2) (Xdh), aldehyde oxidase (EC 1.2.3.1) (Ao-1),tetrazolium oxidase (To), triose phosphate isomerase (EC1.2.1.9) (Tpi-2), and phosphoglucomutase (EC 2.7.5.1)(Pgm-1).Wild males were used directly for electrophoresis. Individual

females were placed in separate culture bottles; one F1 femalewas run from each culture for a given enzyme.Numbers are used to refer to the alleles. At each locus, one

allele is taken as standard, and named 100. The other allelesare designated by adding or subtracting from 100 the dif-ference in anodal migration, expressed in millimeters, be-tween the enzyme coded by each allele and the standard.

RESULTS AND DISCUSSION

The allelic frequencies at each of 27 loci are shown in Table 1for the six populations of D. willistoni. The sample size is thenumber of wild genomes studied. The proportion of heterozy-gotes given is the value expected according to the Hardy-Weinberg principle; the observed and expected frequenciesgenerally agree quite well.As it has been reported earlier in natural populations of

D. willistoni and other species of the group (1-5), the patternsof the allelic frequencies are very similar in all localities.Generally, at any one locus the same allele is the most fre-quent in every population, and the same alleles appear atintermediate and low frequencies. Nevertheless, at some locithere are significant differences between localities in the allelic

* This is paper no. 9 in a series: "Enzyme variability in theDrosophila willistoni group." Paper no. 5 is ref. 3.

999

Dow

nloa

ded

by g

uest

on

June

25,

202

0

1000 Genetics: Ayala and Tracey Proc. Nat. Acad. Sci. USA 71 (1974)

TABLE 1. Allelic variation at 27 loci in natural populations TABLE 1. Allelic variation at 27 loci in natural populationsof Drosophila willistoni of Drosophila willistoni

Santo Barran- Santo Barran-Gene Alleles Santiago Domingo Mayag,3ez quitas Yunque St. Kitts Gene Alleles Santiago Domingo Mayasuez quOtas Yunque St. Kitts

Lap-5 Sample size 320 33694 .000 .00096 .000 .00698 .044 .057

100 .597 .494103 .338 .429105 .022 .015

Heterozygotes .527 .569

Est-2 Sample size 320 33298 .000 .006100 .181 .175102 .813 .807104 .006 .009106 .000 .003


Est-3 Sample size 320 33698 .025 .012100 .963 .985102 .013 .003


Est-4 Sample size 322 33698 .003 .000102 .991 .997104 .006 .003


Est-5 Sample size 318 33690 .000 .00095 .000 .000

100 1.000 1. 000105 .000 .000


Est-7 Sample size 158 16892 .000 .00694 .000 .00095 .006 .00096 .025 .01898 .120 .119

100 .525 .530102 .291 .274105 .032 .054107 .000 .000120 .000 .000


Aph-l Sample size 80 17098 .000 .000

100 .950 .953102 .025 .012104 .025 .035


Acph-l Sample size 322 32288 .000 .00094 .006 .01997 .003 .000100 .988 .978104 .003 .003106 .000 .000


Ald Sample size 198 15696 .000 .00098 .000 .013100 .677 .647102 .303 .333104 .000 .006105 .020 .000


Adh Sample size 316 33690 .000 .003

100 .997 .99.4106 .003 .003


Mdh-2 Sample size 198 16686 .000 .00688 .000 .00094 .298 .404100 .702 .584104 .000 .000106 .000 .006112 .000 .000


aPpdh Sample size 322 33288 .000 .00094 .003 .003100 .997 .997106 .000 .000

Reteroxygotes .006 .006

Idh Sample size 198 16692 .005 .00096 .010 .000

100 .985 .994104 .000 .006108 .000 .000


G3pdh Sample size 198 16696 .010 .00098 .005 .018

100 .944 .958102 .005 .024105 .035 .000


458 624 498 364.000 .000 .004 .000.004 .005 .000 .003.039 .032 .014 .008.664 .736 .801 .810.282 .223 .177 .170.011 .005 .004 .008.478 .408 .327 .314

454 636 502 372.004 .003 .024 .032.110 .129 .129 .132.881 .855 .837 .833.004 .013 .010 .003.000 .000 .000 .000.212 .252 .283 .287

372 628 424 328.013 .013 .026 .015.984 .976 .972 .985.003 .011 .002 .000.032 .047 .055 .030

456 634 466 372.007 .000 .002 .000.985 .992 .991 1.000.009 .008 .006 .000.030 .016 .017 .000

458 650 500 372.000 .002 .000 .000.002 .002 .002 .000.998 .994 .992 1.000.000 .003 .006 .000.004 .012 .016 .000

229 313 251 186.000 .000 .004 .000.000 .000 .004 .005.009 .000 .008 .000.022 .013 .012 .016.118 .096 .139 .145.576 .562 .530 .575.236 .256 .243 .220.035 .061 .048 .032.004 .013 .008 .005.000 .000 .004 .000.596 .605 .638 .598

186 144 168 20.032 .014 .012 .000.871 .917 .869 1.000.032 .069 .107 .000.065 .000 .012 .000.235 .155 .232 .000

456 684 618 372.002 .003 .000 .000.011 .015 .011 .000.000 .003 .000 .000.987 .978 .982 .989.000 .000 .000 .003.000 .001 .006 .008.026 .043 .035 .021

136 374 426 218.000 .003 .000 .000.000 .011 .005 .000.713 .834 .660 .651.279 .150 .336 .349.007 .003 .000 .000.000 .000 .000 .000.4.3 .282 .452 .454

422 662 506 364.000 .002 .002 .000.993 .995 .998 1.000.007 .003 .000 .000.014 .009 .004 .000

226 470 192 198.000 .002 .000 .000.000 .004 .000 .000.217 .200 .214 .000.774 .794 .781 1.000.000 .000 .005 .000.004 .000 .000 .000.004 .000 .000 .000.353 .330 .344 .000

458 682 510 370.000 .000 .000 .003.002 .000 .000 .000.998 .997 1.000 .992.000 .003 .000 .005.004 .006 .000 .016

212 482 436 220.000 .000 .000 .000.005 .008 .002 .000.995 .992 .991 .995.000 .000 .005 .005.000 .000 .002 .000.009 .016 .018 .009

228 342 436 176.026 .029 .030 .0 34.000 .000 .000 .000.934 .956 .947 .938.000 .000 .000 .000.039 .015 .023 .028.125 .085 .101 .119

Odh-l Sample size 188 16686 .000 .00094 .032 .03096 .000 .000

100 .952 .934104 .011 .024108 .005 .012


Me-i Sample size 188 16492 .000 .00096 .005 .012

100 .979 .976104 .016 .012


Me-2 Sample size 198 14692 .010 .00096 .035 .021100 .753 .637104 .202 .342108 .000 .000


ILh Sample size 320 24696 .000 .00497 .006 .00098 .125 .11499 .088 .089

100 .484 .585101 .269 .203102 .025 .004103 .003 .000104 .000 .000


Ao-i Sample size 70 6295 .000 .03296 .000 .01698 .100 .081

100 .886 .774102 .014 .097103 .000 .000105 .000 .000

lSeterozygotes .205 .383

To Sample size 260 19686 .000 .01094 .000 .000

100 1.000 .980108 .000 .010


Tpi-2 Sample size 198 16692 .000 .00094 .000 .000

100 1.000 1.000106 .000 .000


Pgm-1 Sample size 198 16480 .000 .00096 .025 .024100 .960 .976104 .015 .000108 .000 .000


Adk-l Sample size 198 11894 .015 .034100 .212 .314106 .636 .593112 .126 .051118 .010 .008


Adk-2 Sample size 198 16496 .000 .006100 .985 .988104 .010 .006108 .005 .000


Hk-i Sample size 198 16296 .025 .019100 .864 .914104 .111 .068108 .000 .000


Hk-2 Sample size 198 16092 .000 .00096 .000 .000100 .934 .538104 .020 .012108 .045 .450


Hk-3 Sample size 198 16492 .005 .00696 .005 .037

100 .985 .927104 .005 .030108 .000 .000


110 372.000 .000.000 .000.009 .000.973 .978.009 .016.009 .005.053 .042

228 480.009 .000.018 .006.969 .985.004 .008.060 .029

182 346.000 .003.016 .020.868 .801.110 .173.005 .003.234 .329

342 592.006 .000.003 .007.079 .140.094 .074.404 .515.365 .253.044 .010.006 .000.000 .000.687 .645

- 54- .000- .000- .093- .833- .000- .019- .056- .294

294 508.000 .006.000 .000

1.000 .994.000 .000.000 .012

228 480.000 ..002.000 .002

1.000 .996.000 .000.000 .008

228 482.000 .002.004 .002.961 .981.035 .015.000 .000.076 .037

202 316.000 .009.277 .266.609 .639.114 .079.000 .006.539 .514

228 472.009 .009.965 .983.018 .004.009 .004.068 .034

228 480.004 .008.930 .927.061 .060.004 .004.132 .137

228 472.000 .000.009 .004.956 .936.026 .019.009 .040.085 .121

228 478.000 .002.004 .004.982 .979.013 .006.000 .008.035 .041

334 112.003 .000.006 .000.003 .000.961 .991.023 .009.003 .000.076 .018

434 218.000 .000.005 .005.991 .991.005 .005.018 .018

428 210.000 .000.042 .014.743 .762.210 .219.005 .005.402 .371

502 318.002 .000.004 .022.092 .126.034 .050.446 .374.406 .418.014 .009.000 .000.002 .000.626 .666

457 292.009 .003.002 .000.989 .997.000 .000.022 .007

428 220.000 .000.000 .000.998 .995.002 .005.005 .009

434 218.000 .000.014 .000.965 .963.021 .032.000 .005.067 .071

404 220.000 .000.309 .177.574 .736.114 .086.002 .000.562 .419

422 218.005 .000.943 .954.043 .046.009 .000.109 .088

420 220.000 .009.979 .964.021 .023.000 .005.042 .071

430 220.002 .000.014 .009.895 .891.063 .082.026 .018.194 .199

430 220.002 .000.009 .005.967 .968.021 .027.000 .000.064 .062

frequencies, for instance, at the loci Lap-5, Ald, Mdh-2, Xdh,and Hk-2 (Table 1).

Comparison of the data in Table 1 with published results(1, 2, 5) shows that differentiation occurs also between regions

Dow

nloa

ded

by g

uest

on

June

25,

202

0

Proc. Nat. Acad. Sci. USA 71 (1974)

TABLE 2. Allelic variation at 28 loci in natural populationsof four species of the Drosophila willistoni group

Genetic Differentiation 1001

TABLE 2. Allelic variation at 28 loci in natural populationsof four species of the Drosophila willistoni group (Cont'd.)

Gene Alleles D. willistoni D. tropicalis D. e. caribbensis D. nebulosa Gene Alleles D. willistoni D. tropicalis D. e. caribbensis D. nebulosa

Lap-5 Sample size98

100103105107109

Heterozygotes

Est-2 Sample size9698

100102

Heterozygotes


100102

Heterozygotes


Heterozygotes


Heterozygotes


Heterozygotep

Est-

Acph

Aid

Adh

Mdh-2

aGpdh

Idh

G3ydh

Odh-1

Me-I

2600.031.698.258.010.000.000

.437+.043

2616.000.011.138.848

. 277+.016

2408.000.017.977.006

.044+.007

2586.002.000.992

.015+.004

2634.997.002

.005+.003

Not assayed

526.002.008.120.848.021.000

.254+.017

544.976.002.013.000

.058+.013

506.000.000.972.000

.058_.007

544.004.987.009

.021+.006

No activity

Not assayed

-7 Sample size 1305 27296 .017 .01198 .121 .121

100 .552 .449102 .251 .324105 .045 .074107 .006 .011

Heterozygotes .615+.007 .669+.011

-1 Sample size 768 180100 .909 .056102 .049 .900104 .029 .044

Heterozygotes .135+.037 .192+.071

-1 Sample size 2774 54288 .001 .00994 .011 .980

100 .983 .009106 .003 .000108 .000 .000

Heterozygotes .032+.004 .043+.007

Sample size 1508 350100 .708 .686102 .282 .291

Heterozygotes .420+.029 .404+.055

Sample size 2606 438100 .996 .998110 .000 .000

Heterozygotes .008+.002 .003+.003

Sample size 1450 35296 .001 .99494 .214 .00398 .000 .000

100 .781 .003106 .001 .000

Heterozygotes .324+.069 .007+.007

Sample size 2674 546100 .997 .993

Heterozygotes .006+.002 .014+.009

Sample size 1714 338100 .992 .997104 .002 .003

Heterozygotes .016+.003 .006+.006

Sample size 1546 30096 .024 .017

100 .947 .940105 .023 .023

Heterozygotes .103+.007 .140+.043

Sample size 1282 258100 .965 .953104 .017 .027

Heterozygotes .068+.016 .071+.043

Sample size 1712 24094 .000 .979

100 .984 .008104 .008 .000

Heterozygotes .036+.007 .049+.031

1564.000.000.000.000.757.230

.373+.007

1326.006.045.049.873

.326+.060

736.000.014.976.010

.107+.058

1430.986.007.001

.072+.026

1550.366.620

.493+.007

288.698.090.134

.496+.037

No activity

650.028.945.025

.105+.026

1610.563.017.145.226.012

.507+.074

694.729.255

.406+.043

1462.997.000

.032+.027

934.000.000.000.012.986

.023+.014

1708.999

.006+. 003

970.997.001

.020+.014

754.786.182.000

.108+.013

786.989.008

.034+.015

942.000.000.997

.013+.009

214.028.000.762.051.014.019

.420+.021

308.068.860.010.000

.209+.058

140.964.000.007.007

.086+.002

316.987.003.000

.017+.009

No activity

314.809.010.140

.300+.025

161.000.000.006.006.006.932

.134_.052

16.000.125.875.219

312.000.000.000.199.782

.402+.056

176.886.102

.221_.065

308.003.987

.031+.009

176.000.000.994.000.006.000

314.987

.014+.014

172.919.058

.183+.043

Not assayed

1521.000.000.000

176.000.028.966

.068+.020

Me-2 Sample size96

100104106108

Heterozygotes

xdh Sample size939596979899

100101102

Heterozygotes

Ao-l Sample size98

100102105

Heterozygotes

To Sample size100120

Heterozygotes

Tpi-2 Sample size100105106

Heterozygotes

Pgm-1 Sample size9698

100104

Heterozygotes

Adk-l Sample size100106112124

Heterozygotes

Adk-2 Sample size100104

Heterozygotes

Hk-l Sample size96

100104

Heterozygotes

k-2 Sample size100104108

Heterozygotes

Hk-3 Sample size100104

Heterozygotes

Sowm rare alleles are omitted from the table. Populations with sample sizesmaller than 20 have not been used to estimate the average frequency ofheterozygotes in the species.

at some loci of D. willistoni. At the Lap-5 locus, allele 103 isthe most frequent, followed in frequency by allele 100, in allcontinental populations of D. willistoni studied (1, 2). In thepresent samples from the Caribbean islands, as well as in sixother islands from the Lesser Antilles (2), allele 100 is themost frequent, and allele 103 the second most common.Another example of notable regional differentiation exists atthe ilidh-2 locus. Allele 94 is very rare in every populationpreviously studied, including those from the Lesser Antilles;its frequency in 7584 wild genomes previously sampled is0.011, while the frequency of allele 100 is 0.981 (1, 6). Thefrequencies of M71dh-294 and Mlfdh-2100 in St. Kitts agree withthe frequencies found in previous studies. In Puerto Rico andthe Dominican Republic, however, Mdh-294 has a frequencyabout 20 or more times greater (between 0.200 in Barran-quitas and 0.404 in Santo Domingo) than it has in all otherpopulations studied.

1510.027.765.203.000.003

.368+.033

2320.000.000.002.007.114.069.468.322.017

.648+.014

186.091.833.038.016

.294+.051

2007.994.000

.014+.006

1720.998.000.001

.004+.002

1724.010.000.969.020

.063+.007

1458.263.628.099.000

.519_.021

1702.968.022

.059_.014

1708.009.936.053

.131+.029

1708.885.039.069

.205+.063

1718.971.016

.062_.016

184.005.179.087.000.723

.322+.096

512.387.523.020.006.000.000.000.000.000

.553_.029

106.038.849.057.028

.234+.099

4201.000.000.000

356.997.000.000

.004+.004

352.009.000.099.807

.322_.044

344.340.573.081.000

.543+.018

354.986.006

.029+.012

352.009.969.023

.057+.007

352.835.043.114

.243+.068

352.980.014

.049+.018

900.020.876.099.000.003

.223+.023

1626.000.000.000.000:007.030.476.438.049

.574+.024

456.009.195.020.706

.195+.028

1313.998.000

.003+.001

968.001.000.996

.025+.012

964.002.000.058.933

.203+.064

770.140.742.110.000

.418+.036

968.001.979

.078+.046

966.982.008.000

.046+.011

968.937.014.043

.164_.057

960.978.011

.047_.010

166.000.000.012.940.024

.121+.045

294.000.058.711.167.034.000.000.000.000

.484+.063

Not assayed

145.000.986

.019+.019

176.000.989.000

.015+.015

174.810.155.006.000

.327+.065

174.000.138.644.161

.540+.011

174.006.971

.050_.020

174.000.138.862

.237+.041

176.898.006.091

.163+.075

1761.000.000.000

Dow

nloa

ded

by g

uest

on

June

25,

202

0

1002 Genetics: Ayala and Tracey

50

40

0-i

zLLU

a.

30 F

201-

0

50

40

0

-J

z

LU

w

0LJ

301

20

10

I ,,i , r0. 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

GENETIC SIMILARITY

I I

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

GENETIC SIMILARITYFIG. 1 (top). Histogram showing the distribution of loci with

respect to genetic similarity, I, when pairs of sibling species are

compared.FIG. 2 (bottom). Histogram showing the distribution of loci

with respect to genetic similarity, I, when pairs of nonsiblingspecies are compared.

The patterns of genetic variation in the other three species,D. tropicalis, D. e. caribbensis, and D. nebulosa, are similar tothose found in D. willistoni. At each locus, the same allelesare found with roughly the same frequencies in all local popu-lations of a given species, although local and regional dif-ferentiation occur at some loci. As a measure of genetic dif-ferentiation, we have used the statistics I ("genetic similar-ity") and D ("genetic distance") as defined by Nei (7). Themean values of I between local populations are: D. willistoni,0.952 i 0.006; D. tropicalis, 0.949 +t 0.010; D. e. caribbensis,0.954 0.011; and D. nebulosa, 0.991 +4 0.002. In every spe-

cies, populations from the same island are genetically more

similar than populations from different islands, but differencesbetween the mean Is are statistically significant only in D.willistoni, I within islands = 0.980 + 0.007, I between is-lands = 0.942 + 0.005. A summary of genetic variation inthe four species is given in Table 3.

Table 2 gives the allelic frequencies in the four species for28 loci when the data for all local populations are pooledwithin each species. Some rare alleles have been omitted. Aremarkable pattern appears. Generally, when two species are

compared at any one locus, the configuration of allelic fre-quencies is either essentially identical (I > 0.95) or very dif-ferent (I < 0.05). The species that are similar vary fromlocus to locus. At six loci (four weakly polymorphic, aiGpdh,Idh, Odh-1, and Hk-3, and two moderately polymorphic,Ald and Hk-2), the four species are genetically very similar.

TABLE 3. Summary of genetic variation in naturalpopulations of four species of Drosophila

D. willistoni D. tropicalis D. e. caribbeasits D. nebulosa

Number of 27 26 27 25loci studied

Genes sampled 18.50 i 121 370 + 24 1048 + 73 203 i 15per locus

Polymorphic lociper population:

(1) 0.680 0.079 0.678 i 0.072 0.674 0.075 0.687 0.081(2) 0.4454±0.087 0.438 0.092 0.470 0.086 0.510 0.022

Heterozygous loci 0.182 0.039 0.168 0.038 0.189 0.036 0.170 0.033per individual

A locus is considered polymorphic (1) when the frequency ofthe second most common allele > 0.01; (2) when the frequency ofthe most common allele <0.95. The proportion of polymorphicloci per population and of heterozygous loci per individual havebeen calculated by estimating the appropriate parameter for eachlocus, and averaging these values over all loci.

At four loci (Lap-5, Est-5, Acph-1, and Mdh-2, all moderatelyor highly polymorphic in one or more species), no two specieshave similar patterns of allelic frequencies. At the other 18loci, one pair, two pairs, or a triplet of species are very similar,while the others are quite different.The pattern of variation of allelic frequencies is most sig-

nificant when any two species are compared through all locistudied in both species. Consider the pair, D. tropicalis andD. e. caribbensis. At 11 loci (Est-3, Aph-1, Ald, Adh, aGpdh,Idh, Odh-1, To, Pgm-1, Hk-2, and Hk-3) these two specieshave a genetic similarity, I, greater than 0.95, and at one

more locus (Adk-1) I > 0.90. On the other hand, at 10 loci(Lap-5, Est-2, Est-4, Acph-1, M1-dh-2, Mle-i, Xdh, Tpi-2, Adk-2,and Hk-1) these two species have very different configurationsof allelic frequencies, I < 0.05. Only three loci (G3pdh, JMle-2,and Ao-1) fall between these two extremes. A similar pictureemerges when either D. tropicalis or D. e. caribbensis is com-

pared with D. willistoni. When any of the three sibling speciesis compared with D. nebulosa most loci also have extremevalues of I, although the lprol)ortion of loci at which I < 0.05is greater and that at which I < 0.95 is smaller than whensibling species are compared. The distribution of loci fallingwithin each interval of I is shown in Fig. 1 for comparisonsbetween siblings, and in Fig. 2 for comparisons between non-

sibling species. The actual I values for each comparison ofpairs are given in Table 4. The mean I for comparisons be-tween sibling species is 0.546 + 0.040 (D = 0.610 0.071);and for comparisons between nonsiblings it is 0.334 0.048(D = 1.115 0.135).What processes are responsible for the observed slatterns

of allelic frequencies? It has been suggested that most or allgenetically controlled lrotein lpolymorlphisms are adaptivelyneutral (8, 9). This theory predicts that genetically isolatedpol)ulations should have at a given locus different alleles and.in different frequencies. If the protein variation is adaptivelyneutral, the occurrence of the same alleles in similar frequen-cies in two different l)ol)ulations could be explained by

TABLE 4. Genetic similarity, 1,in comparisons between pairs of species

D. tropicalis D. e. caribbensis D. nebulosa

D. willistoni 0.623 0.527 0.287D. tropicalis 0.489 0.286D. e. caribbensis 0.429


Dow

nloa

ded

by g

uest

on

June

25,

202

0


assuming that gene flow exists between the pol)ulations (9)or, alternatively, by assuming that the populations becamegenetically isolated only recently and there has been insuffi-cient time for their divergence by random drift processes.These two alternatives can be discarded in light of the presentevidence. The lack of gene flow between these species is wellauthenticated (10). If there were sufficient gene flow betweentwo species to keep the allelic frequencies similar at a givenlocus, the allelic frequencies should be equally similar at allother neutral loci as well. Yet, in each pair of species the con-figuration of allelic frequencies is very similar at nearly halfof the loci, and very different at nearly the other half.

Similarities between pairs of species at many loci cannot beexplained, either, as due to lack of sufficient number of gen-erations for divergence of allelic frequencies. Allelic frequencieshave become in fact completely different at about half theloci, on theaverage, between any pair of species.A conceivable suggestion is that some loci have become

different in a relatively short time due to natural selection,while adaptively neutral loci have remained quite similarbecause not enough time has passed since the species becamereproductively isolated. However, most loci that have similarallelic frequencies in one pair of species have very differentallelic frequencies in some other pair. Only 6 out of 28 locistudied have similar allelic frequencies in all four species, andfour of them are essentially monomorphic. This unlikely sug-gestion would, in any case, require that strong natural selec-tion operate at 22, or 79%, of the loci surveyed.The opposite alternative is also conceivable; namely, that

similar configurations of allelic frequencies are maintained bynatural selection at some loci, while other loci have becomedifferentiated as the result of random drift of neutral allelesover many generations. All four species differ substantiallyfrom each other in the allelic frequencies at only four loci:Lap-5, Est-5, Acph-1, and Mfdh-2; at all other loci, there isat least one pair of species with very similar allelic frequencies.But even those four loci would have to be excluded as beingadaptively neutral when other lopulations of these and closelyrelated species are taken into consideration. The allelic fre-quencies are essentially identical: (1) at the Lap-5 locus incontinental populations of D. willistoni and of D. tropicalis(5); (2) at the Est-5 locus in continental populations of D.willistoni and of D. equinoxialis (1, 3); (3) at the Acph-1locus in populations of D. nebulosa, D. paulistorum, and D.insularis (the latter two are siblings of D. willistoni); and (4)at the Mdh-2 in continental populations of D. equinoxialisand D. paulistorum.Another conceivable explanation is that alleles are adap-

tively equivalent, but their frequencies are similar in dif-ferent species owing to mutation pressure. Mutations tocommon alleles would occur more frequently than mutationsto rare alleles. Allozymes with a given electrophoretic mo-bility do not necessarily have identical aminoacid sequences.That is, electrophoretically detectable alleles need not besingle alleles, but rather they may be classes of alleles codingfor different polypeptides with identical electrophoreticmobilities. Therefore, the frequency of an electrophoretic"allele" might represent the number of possible differentalleles within that mobility class. If the number of alleleswithin each class is large, genetically isolated populationscould nevertheless have similar frequency distributions of themobilitv classes, even though they might be adaptivelyequivalent. The distribution of allelic frequencies in the four

species rules out this possibility, since all four species havesimilar allelic frequencies at only 6 of the 28 loci.

Figs. 1 and 2 show the distribution of loci according to theirgenetic similarity (I) values in comparisons between l)airs ofspecies. Both figures show U-shaped distributions, with mostloci (83% in either figure) in the 0-0.05 or 0.95-1.00 classes.If allelic variation at these loci were adaptively neutral theirdistribution should be approximately normal around themean I value (11). Consider two poI)ulations that, at a giventime, had identical allelic frequencies at all loci but have beenevolving independently for a given length of time. Assumealso that at each locus there are several allelic states adap-tively equivalent. The probability of a given allele to increaseor decrease in frequency by a given amount each generationwould be the same for all alleles at all loci since it would bea function of the effective size of the population. If there weremany alleles at each locus the allelic frequencies in the twopopulations would continuously diverge at every locus. Thatis, the value of I would gradually decrease at a rate that hasthe same probability at each locus. After a given length oftime, the values of I for all loci would have a normal distribu-tion around their mean value. Clearly, the distributions inFigs. 1 and 2 are not normal at all. A conceivable escape fromthis argument is to suggest that at some loci natural selectionkeeps allelic frequencies similar, while the rest of the loci areevolving at random and the mean I for these loci is alproach-ing zero. However, this explanation is unsatisfactory since dif-ferent sets of loci fall in the 0.95-1 class when different pairsof species are compared. Thus the distribution of I given byFigs. 1 and 2 is incompatible with the hypothesis that allelicvariation at the loci studied is adaptively neutral. It shouldbe lointed out that if the allelic variation is adaptively neu-tral, the distribution of I should also be normal even whenmany alleles exist within each mobility class.The Ieattern of allelic variation in different local populations

of a given sl)ecies of the D. willistoni group has been shownelsewhere (1-3, 5) to be incoml)atible with the hypothesisthat l)rotein l)olymorl)hisms are adaptively neutral. Thisconclusion is further confirmed by the l)attern of allelic varia-tion in different species of the group. Balancing natural selec-tions (heterotic or frequency dependent) are the most likelyprocesses to account for the patterns of genetic variationobserved in natural l)o)ulations. Needless to say, allelicfrequencies determined by natural selection are also affectedby random drift, particularly in small populations.We thank -Miss Lorraine G. Barr for excellent technical

assistance. Prof. Th. Dobzhansky read the manuscript andprovided valuable advice and criticism.1. Ayala, F. J., Powell, J. R., Tracey, AI. L., -Mourdo, C. A. &

PWrez-Salas, S. (1972) Genetics 70, 113-139.2. Ayala, F. J., Powell, J. R. & Dobzhansky, Th. (1971) Proc.

Nat. Acad. Sci. USA 68, 2480-2483.3. Ayala, F. J., Powell, J. R. & Tracey, Al. L. (1972) Genet.

Res. 20, 19-42.4. Richmond, 1?. C. (1972) Genietics 70, 87-112.5. Ayala, F. J. (1972) Proc. Sixth Berkeley Symp. Math.

Stat. Prob. V, 211-236.6. Ayala, F. J. & Anderson, W. W. (1973) Nature New Biol.

241, 274-276.7. Nei, AI. (1972) Amer. Natur. 106, 283-292.8. King, J. L. & Jukes, T. H. (1969) Science 164, 788-798.9. Kimura, Ml. & Ohta, T. (1971) Nature 229, 467-469.

10. Spassky, B., Richmond, R. C., Perez-Salas, S., Pavlovsky,O., M\ourdo, C. A., Hunter, A. S., Hoeningsberg, H.,Dobzhansky, T. & Ayala, F. J. (1971) Evolution 25, 129-143.

11. Ayala, F. J. & Gilpin, M. E., manuscript in preparation.

Genetic Differentiation 1003

Dow

nloa

ded

by g

uest

on

June

25,

202

0

Documents

Genetic Differentiation Within andBetween DrosophilaGenetic Differentiation Within andBetween Species ofthe Drosophila ... The pattern of genetic variation ... Our techniques of starch-gel