8/8/2019 Lewontin - Genetics 1997 - Dobzhansky
Copyright 0 1997 by the Genetics Society of America
PerspectivesAnecdotal, Historical And Critical Commentaries on Genetics
Edited by James F. Crow and W illiam F. Dove
DOBZHANSKYSenetia and the origin of Species: Is It Still Relevant?Richard C. Lewontin
Museum of Comparative Zoology, Harvard University, Cambridge, Massachusetts 02138
T E first edition of THEODOSIUSOBZHANSKYSe-netics and the origzn of Species appeared almost pre-cisely 60 years before this issue of GENETICS.t wouldbe hard to find anyone, even envious authors of othermagna Opera,who woulddisagree WithJEFFREY POWELLS50th anniversary assessment that it is the most m-portant and nfluential book on evolution of the twenti-eth century (POWELL987). It must be remembered,however, that the book was only the concrete form ofthe JesupLectures given the previous year at the invita-tion of L. C. DUNN nd others at Columbia, an invita-tion that signaled the importance that nfluential biolo-gists already placed on DOBZHANSKYS previous0 yearsofwork on genetics and evolution. In an mportantsense, the publication of the book, to be followed byother Jesup Lectures books by MAYR (1942) and SIMP-SON (1944) and eventually STEBBINS1950),was a mani-festo representing a view that was already taking hold.
Nothing would be gained by plagiarizing POWELLSmasterful summary of the schism between the commu-nities of genetics and evolutionary biology at the time,and of the highlights of DOBZHANSKYSntegration ofMendelism and Darwinism. In this respect, two thingsthat were unique to DOBZHANSKYSook need to beemphasized. First, Genetics and the Origin of Speciesseemed to be essentially a treatise in observational biol-ogy, speaking in the language and using the biologicalmaterials of experimentalists and natural historians.Second, DOBZHANSKYSntire schema began with theorigin of variation and culminated with the formationof species, thus seeming to engage DARWINSutlinedirectly. Some years before, FISHER 1930), WRIGHT(1931), and HALDANE (1932) had already completedsyntheses of genetics and evolution at the conceptuallevel, showing how Mendelism served asbasis for evo-lutionary change. But their expositions did not rely, asDOBZHANSKYS did,n a large amount of description ofobservations from nature, and the problem of the ori-gin of species was treated by them only en passant(FISHERpent three and a half pages on it), although
Author email: email@example.comGenetics 147: 351-355 (October, 1997)
WRIGHTSicture of alternative adaptive peaks certainlysidled up to the problem.
One historical viewpoint for an appreciation f DoBz-HANSKYS book is that of the observer in 1937 lookingretrospectively, seeing how DOBZHANSKYSynthesis suc-ceeded in bringing the full apparatus of genetics andof genetic observations in natural populations to bearon the observable facts of speciation and species diver-sity. It is this synthetic element that was so compellingto his readers of the time. But, like any major scientificsynthesis, Genetics an d the Origin of Species was not simplya compelling reorganization of existing knowledge intoa unified structure. The real test of its importance layin its prospective aspect, in the implicit program forevolutionary genetics from 1937 into hefuture. AsPOWELLointed out in his essay, the book, especiallyin later editions (1941,1951),had an important mpactin establishing observational population genetics as ascientific fieldfor investigation. In fact, the entireprob-lematic of evolutionary genetics for the last 60 years,including its detailed formulation at present, flows fromthe organization and content of DOBZHANSKYSesupLectures and the book that embodied them.DOBZHANSKYSrgument, which every graduate stu-dent in Zoology at Columbia in his day was expectedto reproduce on the written qualifying examination forthe Ph.D., was the skeleton of DARWINSheory of theorigin of species. Species are groups of interbreedingorganisms that have beencut off,biologically, fromsharing heredity with other species withwhich heyshare a common ancestry in the remote past. This re-productive isolation is the final step in divergence be-tween geographically separated populations, geograph-ical races, whichwereoriginally keptapar t only bygeography, but which have acquired during their geo-graphical separation sufficient genetic difference toprevent future nterbreeding. But for genetic differ-ences to accumulate between populations, there mustbe genetic variation within populations to begin with.That is, species evolution is a process of the conversionof the variation present between individuals within pop-ulations at a given moment intovariation between pop-
8/8/2019 Lewontin - Genetics 1997 - Dobzhansky
35 2 R. C. Lewontinulations in time and space. This scheme hen placesthe investigation of intrapopulation genetic variationand polymorphism at the very center of the study ofevolutionary dynamics. Even the description of differ-ences between populations is in the form of the statisti-cal description of their polymorphisms rather than bycharacteristic typological differences. This point of viewhas differentiated populationan d evolutionary geneticsfrom all other modes of studying evolution. It is thereason that in a book of 321 pages of text, whose ulti-mate goal is to explain the origin of species, 178 pagesat the beginning areaken up with intrapopulation vari-ation. It is the reason that at present so many popula-tion geneticists are skeptical of simple post hoc optimalityexplanations of species characteristics, for they are pre-disposed to consider the contingency of just the rightkind of genetic variation to make the stories work. Ican call monsters from the vasty deep says that ur-adaptationist OWEN GLENDOWER. Why,o can I andso can any man. But will they come when you do callfor them? replies the doubtful population geneticistHOTSPUR.t is the reason that evolutionary geneticistsuntil recently had so neglected a detailed genetic tudyof the differences thatunderlie species divergence.After all, species differences are simply the final disposi-tion of the standing geneticvariation within species, soit is the nature of that standing variation and of theforces modulating it that is the real stuff of evolutionarygenetics. All else is just developmental and molecularbiology.
The degree to which the first edition of Genetics andthe Origzn of Species was the enunciation of a problematicfor he future, rather han a synthesis of an alreadyadequate body of fact and theory, can be seen in acomparison between the original and later editions.While so much emphasis was placed on the mportanceof intrapopulation genetic variation in the first edition,the actual evidence was pretty thin . Aside from the fewhumanbloodgroups hen known, DOBZHANSKYndEPLINGS1944) survey of inversion polymorphism ingeographical populations of Drosophila pseudoobscura,and a few studies of simple Mendelizing morphologicalpolymorphisms and chromosomal lethals by the Dubi-nin school (see pp. 42-46 in DOBZHANSKY937), virtu-ally nothing was known of he frequencies f Mendeliangenetic variations innaturalpopulations. DOBZHAN-SKYS own famous Genetics of Natural Population seriesbegan to appear only after the Jesup Lectures. By thetime he finished the revised thirdedition in 1951,twenty papers in that series had appeared, comprisinga model for how genetical variation in natural popula-tions could be studied. This included observations oftemporal variation and stability in polymorphism, esti-mates of migration and effective population size, evi-dence for the existence of selective differences in na-ture, and the reation of laboratory model populationsin which selection could bedemonstrated an d esti-
mated. The third revised edition of 1951 now couldrefer to 15 years of data from natural and laboratorypopulations estimating parameters of selection, migra-tion, and breeding structure. oreover, large quantitiesof data were now available on the viability an d fertilityvariation among genomes sampled from natural popu-lations of Drosophila, data made possible by an adapta-tion of MULLERSlB trick for making chromosomeshomozygous. Nor was it DOBZHANSKYSchool alonethat pursued the program, nor Drosophila alone thatwas the object of study.As a consequenceof the medicaldemands created by the Second World War, great ad-vances had been made in immunological genetics, re-sulting in an explosion of information on humanbloodgroups and HLA polymorphisms. The most completemodel for how to study a Mendelian polymorphismwithin and between local populations was LAMOTTES(1951) monograph on thehell color and banding poly-morphisms in Cepaea nem oralis. Amanifesto hadbecomean industry.
The program, while seemingly prosperous, was, how-ever, in deep difficulties. Aside from the occasional,genetically simple morphological or immunologicalpolymorphism, studies of natural geneticvariation weredependenton observations of whole chromosomesrather than single physiological and developmentallydefined loci. Inversion polymorphism, while serving as amodel object of study, could really give no informationabout thegenerality of variation o n which the geneticaltheory of evolution depended. Alternatively, the mea-surement of viability and fertility variation in nature,surely the stuff ofevolutionary change, could be assayedonly at the whole chromosome level, providing no realinformation about how much genic variation existed.DOBZHANSKYad created a field and focused investiga-tion on a problematic that seemed mpossible to clarify.
The response to this conundrum was the introduc-tion of a methodof investigation, protein electrophore-sis, that seemed to cut through the difficulty because it(1) provided a phenotype whose variation waseasilyobservable; (2)didnotdepend on any assumptionson thephysiological or developmental consequences ofthe variation; (3 ) would detect a large fraction of thevariation at a large fraction of loci, locus by locus; and(4) could be applied to any organism irrespective of itsamenability to genetic manipulation (HUBBYnd LEW-ONTIN 1966). While it might be f lattering to the self-esteem of those who introduced the technique o thinkof it as revolutionizing the field, the truth s quite theopposite. The immense popularity that electrophoreticstudies enjoyed for nearly 20 years after their introduc-tion in 1966 was precisely that they seemed to providethe possibility of at last coming to grips with the prob-lematic that had occupied evolutionary genetics since1937. Unfortunately, the main strength of the methodwas its fatal flaw. ts essence was that itallowed theassessment of genetic variation unaffected by the physi-
8/8/2019 Lewontin - Genetics 1997 - Dobzhansky
Perspectives 353ological and developmental consequences of that varia-tion. But, by liberating theobservations from physiologyand development, the method also guaranteed hat,except in the very extraordinary circumstance that al-lelic variation at a single locus had a strong marginaleffect on fitness, no inferences about he forces o perating on the variation could be tested. The Dobzhan-skian problematic was even more frustratingly stymied.Now we could describe genetic variation quite generallybut seemed barred from explaining it!
The impasse was broken, at least in part, by a luckyfact of nature: the lack of a one-to-one correspondencebetween the DNA sequence and he amino acid se-quence of proteins. The degeneracy of the code, theexistence of introns, of transcribed but untranslatedand of nontranscribed DNA, all mean that within thesame small genic region there are classes of nucleotideswith very different relationships to the physiology anddevelopment of the organism. Different patterns of ge-netic variation of these different classes could then pro-vide internal evidence about the cumulative effect ofselection, which should operate differently on the dif-ferent classes, asopposed to orces of mutation, popula-tion structure, and recombination, which should affectall classes equally. Beginning with the original demon-stration by KREITMAN (1983) of the unique power ofDNA sequence studies to detect even very weak naturalselection unambiguously, thecentral problematic ofevolutionary genetics seemed once again to be accessi-ble. More than a dozenyears of population genetics atthe nucleotide level have clearly shown that selectiveconstraints exist for all classes of nucleotides includingso-called silent positions in codons, as well as intronsand flanking sequences. (see, as an example, RICHTERet al. 1997). Moreover, it has been possible to detect,in patterns of haplotypes, traces of migration amongpopulations (RICHTER et al . 1997) and the constraintsimposed on variation by differing amounts of recombi-nation (BEGUNand AQUADRO 992). But, more thanthis, the study of nucleotide variation has allowed evolu-tionary genetics to proceed to the next set of questionsposed by the schema outlined in Genetics and the originof Species.
The existence of genetic variation and the modula-tion of its pattern within a population at any time areonly the beginning of the process of species evolution,according to DOBZHANSKY.t is not sufficient that localpopulations are simply different in gene frequency, forevery population must differ from every other one inthe real world of finite assemblages. Species are notsimplyassemblagesof organisms thatarenot nter-breeding, but are distinct life forms with distinct rela-tions to the environment, making a living in distinctways. Nor can this ecological differentiation commonlybe a process that follows after reproductive isolationhas already occurred, for then we would often observethat partially reproductively isolated populations would
show no adaptive differentiation. Unless the popula-tions have come to occupy different peaks in the a daptive landscape, the local populations or geographicalraces are not likely to be in the preliminary stages inspecies formation. This, then, poses a second set ofproblems for population genetics: to demonstrate thatnatural selection has played a role in population differ-entiation. It is easy enough to show that strains drawnfrom different populations have different norms of re-action and different fertilities and viabilities in differentexperimental circumstances. The first edition of Genet-ics and the Or@n of Species uses precisely such evidenceto demonstrate genetic differences between local popu-lations. It is a very different matter, however, to linkthese divergences to specific genetic differences, toshow that they matter in nature and that theyhavebeen established by some process of adaptive naturalselection.
Because of the evident difficulty of such demonstra-tions, this critical next element in the Dobzhanskianprogram has been the subject of a great deal of talkbut only limited action. The geographical variation ofshell patterns in Cepaea provided the opportunity fora long struggle between the English school, which ex-plained all variation asa consequence of localvariationsin environmental conditions (see, for example, CAINand SHEPPARD1950),nd the French school, whichinterpreted the esults asa consequence of genetic drift(LAMOTTE 1951). During the eyday of electrophoreticstudies, a number f cases of eographical or altitudinalclines in the frequencies of variants were found, andthese were correlated with various environmental vari-ables, usually temperature. The most detailed studieslinking the frequencies of variants with heir enzymatickinetics, physiology, and behavior, while successful indemonstrating such a relationship, are unable to dealwith the basic issue facing all who study natural selec-tion in natural populations, namely, the question ofwhich aspects of the organisms biology account forvariance offitness in nature. That is, it may be that,ceteris paribu s, an increase in egg-laying rate of femaleswould increase the fitness of their genotype, but if fe-males innature lay so few eggs that differences in physi-ological potential are rrelevant, then differential physi-ological fertilitys not a significant component of fitnessvariance. The challenge of studying adaptive variationin nature is that one has to know so much about thebiology of the organism. Thus, it would seem that thesecond phase of the Dobzhanskian project, to show thatgenetic differentiation has occurred by natural selec-tion, seems to evade us. Once again, studies of nucleo-tide Variation have provided a possibility of progress. Byfinding short regions of the genome that aremarkedlydepauperate of nucleotide variation for silent sites andintrons, as compared with other regions in the samegenome, a strong ase can be made for aelective genefixation in the relatively recent past. A striking example
8/8/2019 Lewontin - Genetics 1997 - Dobzhansky
354 R. C. Lewontinis the demonstration by BERRYt al. (1991) of a selectivesweep on the microchromosome of Drosophila melano-gaster. Of course, we d o not know the biological causeof the sweep nor which sites within the region are re-sponsible for it, as opposed to being carried along byhitchhiking. The demonstration of adaptive differentia-tion can be carried even further to look for evidenceof adaptive divergence between species, detected by anexcess of amino acid replacements as compared withsilent divergences between them (MCDONALD andK R E I T ~991).There is, in principle, no limit to howmuch of the genome couldbe investigated in this way,choosing particular genesor gene regions and sequenc-ing them within and between species. We could thenestimate, for any collection of populations or relatedspecies, how much of their differentiation has beendriven by selective sweepswithin populations and selec-tive divergences between species. In this way, the secondphase of DOBZHANSKYSeneral scheme could be real-ized. The only question is whether it is the investigatorsor the granting agencies that would grow tired first.
The continuation of DOBZHANSKYSrogram byse-quencing studies revealsts original limitation. Al-thoughDOBZHANSKY s usually thought of s thefounder of experimental population genetics and his1937 book as the founding document, there is in factno experim ent described there until the last chapteron hybrid sterility, where experimental crosses andbackcrosses between Drosophila pseudoobscura and Dro-sophila persimilis using marked chromosomes are dis-cussed. The entire body of evidence marshaled on thecontrol of natural variation within populations and theconversion of that variation into genetic differences intime and space is from static data. It depends upon hatinferences can be made from the standing patterns ofgenetic variation in nature. In this case, the testing ofhypotheses that we usually associate with experimentsis of a special statistical sort, manifest in the 1937book,in most of DOBZHANSKYSexperimental (observa-tional) papers, and in the present state of molecularpopulation genetics. By using population genetic the-ory, either in explicit mathematical form or more heu-ristically, a prediction is made of what the distributionof genetic variation should look like under some simplemodel, say no selection an d nomigration. T he observedstanding pattern of variation is then compared with thisnull prediction, and some inference is made from theagreement or disagreement between the observed andthe expected. The observations of inversion clines orregional variations in theviability ofstrains when testedin a standard laboratory condition are, in this respect,of the same evidentiary nature as the comparison ofthe standing variation within and between species inthe ratio of silent to replacement nucleotide substitu-tions. In 1951 Lamotte attempted to explain the varia-tion in Cepaea by fitting the distributionof colony genefrequencies to a stationary Wrightian distribution. To-
day, molecular population geneticists fi t the nucleotidepolymorphism an d diversity between populations at ev-eral loci to the predictions of coalescent theory. Ofcourse, one can attempt to show that a genetic differ-ence observed innature has some consequence orphysiology an d selection in a laboratory model, just asDOBZHANSKYhowed that inversions would be subjectto selection in the laboratory unde r some conditions.But the success or failure of such experiments does nottell us what forces have operated historically or arenowoperating in nature. Population genetics, then as now,is an observational and statistical science, not an experi-mental one.As a consequence,while it can offer statisti-cal evidence supporting the past action of one or an-other of the evolutionary forces having operated, it can-not cash these inferencesout in the form of actualbiological mechanisms.
An irony of the intellectual history of Genetics and the%gzn of Species is that DOBZHANSKYame into evolution-ary genetics from the study of morphological diversityin nature and so was able to relate the abstractions ofgenetic theory to the biology of organisms, yet in theend he and the field he founded became captives ofthe abstractions. Despite 40 years of study of the chro-mosomal variation in natural populations of Drosoph-ila, DOBZHANSKYublished no observations from natureon the possible biological mediation of the natural se-lection he had detected. There is, in the entire corpusof 43 papers on the Genetics of Natu ral Populations, nopaper on the ecology and life history of . pseudoobscura.The closest he came was to measure the rate of move-ment of genetically marked, laboratory-raised fliesalong a trap line of attractive banana bait. Nor did hemake any pretense that the demonstration of selectionof chromosomal inversions in laboratory populationcages was meant to reveal the natural biology of thispolymorphism. The purpose of those experiments wasto show that theallelic contents of the inversions could,indeed, under some circumstances make a large differ-ence to their fitness and that, in these circumstances,heterozygotes were more fit than homozygotes, as hebelieved them to be in nature. In fact, although therewas selection in the cages at 25, there was none at 18.
Thus, we see that Genetics and the Origan of Species, likeDOBZHANSKYSubsequent research career,althoughseeming to speak in the anguage of organisms, had theultimate effect not of uniting genetics with the naturalhistorical and physiological biology, but of building ascience that speaks the language of gene frequencies.One of the consequences of that alienation of popula-tion genetics from organismal biology was the failureof the projects of the 1960s to build a unified scienceof population biology ou t of the elements of ecologyand population genetics.
Another consequence of the way in which DOBZHAN-SKY constructed the problem of the origin of specieshas been to emove the problemof the actual speciation
8/8/2019 Lewontin - Genetics 1997 - Dobzhansky
Perspectives 35 5process from the concern of most population geneti-cists. All of the issues of natural selection in relation toadaptation are, for DOBZHANSKY,lready dealt with inthe problem of adaptive population divergence. Thefinal stage of separation of the species becomes a ques-tion of the genetics of reproductive isolation (ORR1997), a problem in neurobiology and developmentalgenetics, of why flies dont like each others looks, orwhy a particular sperm cant fertilize a particular kindof egg, or why somatic or germline development failsin hybrid embryos. This mechanical view of the prob-lem of ultimate species divergence is already containedin DOBZHANSKYSdherence to a particular view of spe-cies. In Genetics and th e Origin of Species he reaffirms hisprevious (1935) definition of a species as that stageof evolutionary process at which the once actually orpotentially interbreeding array of forms becomes segre-gated in two or more separate arrays which are physio-logically incapable of interbreeding (p . 312). By de-fining it in this way, DOBZHANSKYhen created the prob-lematic forhe study of speciation,heeneticelucidation of an aspect of developmental and neurobi-ology. Where are the genes? What are their developmental interactions? What determines when and howthey are read?And this is, indeed, the current problem-atic of speciation studies (COYNE 992), divorced fromthe rest of evolutionary genetics until such time as p o pdation geneticists finally fold developmental biologyinto their considerations.
What is now the classical definition of species leadsto a problem thatDOBZHANSKYcknowledges. What arewe to do with all those asexual organisms where thisdefinition of species is irrelevant? He admits that suchorganisms are not continuously distributed in pheno-typic and genotypic space, but that there areaggrega-tions of more or less distinct biotypes and that, justlike sexually reproducing forms, these biotypes areclustered around some of the adaptive peaks in thefield of gene combinations and that the clusters arearranged in a hierarchical ordern a way which is againanalogous to that encountered n sexual forms. (p.320). But, he says, they are not species. So what arethey, and why do we lavish so much interest on theproblem of reproductive isolation? We are never told,because this is the penultimate paragraph in the book.What DOBZHANSKYas done is to finesse one of themost interestingquestionsn evolutionary biology,which iswhy organisms occupy the phenotypic statespace in the hierarchically clustered pattern thatwe see,sex or no sex. That is, how d o organisms acquire newand quite distinct ways of making a living? This is theantecedent question that makes the problem of repro-ductive isolation relevant for sexual species. Whatever
the forces are that cluster organisms in state space, thatclustering is destroyed by sexual recombination, so anorganism that exploits the advantages of sex has a spe-cial problem that asexual ones do not have. In orderto allow sexual organisms to maintain the clustersagainst the disruptionof sex, they have to develop isolat-ing mechanisms. Those that fail become extinct fromtoo much compromise.DOBZHANSKYSonstruction of the problemof specia-
tion as solely the problem of reproductive isolation wasa piece of scientific synecdoche, substituting the processof reproductive isolation for the speciation process inits entirety. It is a testimony to the influence that eneticsan d the Origm of Species has wielded over 60 years thatwe continue to study the speciation process withoutreference to the world that organisms construct andoccupy.
LITERATURE CITEDBEGUN, .J., and C. F. AQUADRO,992 Levels of naturally occurring
DNA polymorphism correlate with recombination ates in D .melanogaster. Nature 356: 519-520.BERRY, .J. ,J. W. AJIOKAand M. KREITMAN,991 Lack of polymor-
phism o n the Drosophilafourth chromosome resulting from selec-tion. Genetics 129: 1111-1117.
CAIN, A.J., and P. M. SHEPPARD, 950 Selection in the polymorphicland snail Cepaea nemoralis. Heredity 4: 275-294.
COYNE,. A,, 1992 Genetics and speciation. Nature 355: 511-515.DOBZHANSKY,H., 1935 A critique of the species concept in biology.
Philos. Sci. 2: 344-355.DOBZHANSKY,H., 1937, 1941, 1951 Genetics and the Origzn of Species.Ed. 1 , 2, 3. Columbia University Press, New York.DOBZHANSKY,H., and C. EPLING, 944 Contributions to th e Genetics,Taxonomy, and Ecology of Drosophila pseudoobscura and Its Relatives.
Pub. 554, Carnegie Institute of Washington, Washington, DC.FISHER, R. A ,, 1930 The Genetical Themy of Natural Selection.Clarendon Press, Oxford.
HALDANE,. B. S . , 1932 The Causes of Evolution. Harper and Row,New York.
HUBBY,. L., and R.C. LEWONTIN,966 A molecular approach tothe study of genic heterozygosity in natural populations. I. Th enumber of alleles at different loci in Drosophilapseudoobscura.Genetics 54: 577-594.KREITMAN,., 1983 Nucleotide polymorphism at the alcohol dehy-drogenase locus of Drosophila melanogaster. Nature 304 412-417.
LAMOT~E,M ., 1951 Recherches sur la structure gtnetique des popu-lations naturelles de Cepaea nemmalis (L). Bull. Biol. Fr. Belg.,Suppl. 35: 1-238.MAm, E., 1942 Systematics and the origzn of Species. Columbia Univer-sity Press, New York.
MCDONALD,. H., and M. WITMAN, 1991 Adaptive protein evolu-tion at the A dh locus in Drosophila. Nature 351: 652-654.Om , H. A., 1997 DOBZHANSKY,ATESONnd the genetics of specia-tion. Genetics (in press).
POWELL,. R., 1987 In the airTHEoDOs1us DOBZHANSKYSe-netics and the origzn of Species. Genetics 117: 363-366.RICHTER,B.,M.LONG,R..LEWoNTINandE.NITMm, 1997 Nucleo-
tide variation and conservation at the dpp locus, a gene control-ling early development in Drosophila. Genetics 145: 311-323.SIMPSON, . G., 1944 Tempo and Mode in Evolution. Columbia Univer-sity Press, New York.
STEBBINS, . L., 1950 Variation andEvolution in Plants. ColumbiaUniversity Press, New York.WRIGHT,S., 1931 Evolution in Mendelian populations. Genetics 1697-159.