The Genetics of Stasis and Punctuation

Annual Reviews
www.annualreviews.org/aronline

Ann. Rev. Genel. 1983.17:11-25
Copyright©1983by AnnualReviewsInc. All rights reserved

THE GENETICS OF STASIS
AND PUNCTUATION

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org John Maynard Smith
by University of Wisconsin - Madison on 08/30/05. For personal use only.
Schoolof Biological Sciences, Universityof Sussex,Falmer,Brighton, BNI9QG,
Sussex, United Kingdom

CONTENTS 11
13
1. THETHEO.R...Y................................................................................. 13
2. CAN STASIS BE EXPLAINEDBY DEVELOPMENTACLONSTRAINTS?......... 14
15
a. ConstraintsLeadingto ContinuouVs ariation............................................ 19
b. ConstraintsLeadingto DiscontinuouVs ariation........................................
3. THEGENETICOSFSPECIEHSYBRID..S................................................. 20
4. THEPATTERONFVARIATIOINN NATUR..E.......................................... 22
5. CANSTASIS BE EXPLAINEDBY LARGEPOPULATIOSNIZE? ................... 22
6. CANSTASIS BE EXPLAINEDBY NORMALIZINSGELECTION?..................
7. CONCLUS..I.O...N...S..........................................................................

1. THE THEORY

Thetheory of punctuatedequilibria ( 16, 24, 25, 60) claimsthat mostevolution-
ary changeoccursin rapid bursts, at the time of lineage splitting (speciation),
andthat suchpunctuationalevents are separatedbylong periodsof stasis during
whichlittle or no morphologicalchangetakes place. Thetheory arose from a
study of the fossil record, andits acceptanceor rejection ultimately will depend
on our interpretation of that record. This reviewdoes not discuss howfar the
theory is correct, but rather asks whatexplanationscan be offered for it. The
question wouldbe a waste of time if the theory werewhollyfalse, of course.
However,it seemsclear that stasis is a real phenomenona, t least in some
lineages at sometimes. It is harder to be sure aboutthe nature of the changes,
whenthey do occur. Thusthe suddenreplacement of one form by another in a

11

0066-4197/83/1215-0011502.00

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org Annual Reviews
by University of Wisconsin - Madison on 08/30/05. For personal use only. www.annualreviews.org/aronline

12 MAYNARDSMITH

particular place maymeanno morethan that the newformevolvedelsewhere; it
does not by itself prove that the newform evolved suddenly, at the time of
speciation.

Thetheory of punctuatedequilibria wasfirst presented(16) as the paleonto-
logical consequenceof Mayr’s(42, 43) theory of allopatric speciation, accord-
ing to whichnewspecies arise whenperipheral populations are isolated from
the mainbodyof the species. Later formulations of the theory hhve taken a
moreradical position, arguing that macroevolutioncan be "decoupled"from
microevolution (60), that the "hopeful monsters" imagined by Goldschmidt
(23) haveplayeda critical role (24, 49), andthat species selection rather
selection at the individual level is the majordriving force of macroevolution
(25, 60).

Thesimplest resolution of this controversyis to note that a changeoccurring
over several thousandgenerationswouldbe very slowto a geneticist but almost
instantaneous to a paleontologist. Perhapsthe best documentedcase of stasis
and punctuation is Williamson’s (72) study of fresh-water mollusks in the
Lake Turkana region of Africa. The data show that morphological change
wasconcentrated into brief periods, but also that, whenchangesdid occur,
they occurred in large populations over manythousands of generations. Com-
menting on these data, Jones (28) remarked that "one man’spunctuation
another man’sgradualism," although this is not the view held by Williamson
(73).

Thereis, however,real disagreementamongthe various interpreters of the
punctuatedequilibria theory. Populationgeneticists can explain the pattern of
evolutioninsofar as it is punctuationalbysayingthat mostof the time selection
is stabilizing, leading to stasis, andoccasionallydirectional, leading to punc-
tuational change.Thealternative viewis that the failure of a species to change
over millions of years mustbe explained in someother way--usuallyeither by
"developmentaclonstraints," or by the inability of large populationsto evolve;
escape fromthese limitations leads to a burst of rapid evolution.

In section 2 1 discuss the nature of developmentalconstraints and conclude
that in the present context the potentially important constraints are those
causing discontinuous variation. Themostdirect wayof deciding whethersuch
constraints have in fact been important in evolution is by studying species
hybrids; this topic is reviewedin section 3. In section 41 ask whetherthe often
discontinuous pattern of variation in nature is caused by developmentalcon-
straints, or whether it is a consequenceof sexual reproduction. Section 5
considers the alternative explanationof stasis proposedbythe punctuationists,
namely, that there is somethingpeculiar about large population size that
inhibits evolutionary change. Finally, section 6 returns to the explanation of
stasis in terms of normalizingselection.

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org Annual Reviews
by University of Wisconsin - Madison on 08/30/05. For personal use only. www.annualreviews.org/aronline

GENETICSOF STASIS ANDPUNCTUATION 13

2. CAN STASIS BE EXPLAINED BY DEVELOPMENTAL
CONSTRAINTS?

Basicto the theory that stasis can be explainedby developmentaclonstraints is
the idea that there are limits to the waysa populationcan change,determinedby
the waysorganismsdevelop. Gastropodshells growby accretion at the margin,
for example,and partly in consequencetheir shapes are restricted to a set
defined by as few as three parameters(50). I argue belowthat for our present
purposethe importantquestion is whetheror not the constraints are continuous.

a. Constraints Leading to Continuous Variation

Physiological constraints maylead to a trade-off betweenone aspect of the
phenotypeand another. For example,there is likely to be a negative correlation
betweenthe speed of limb movemenatnd the force that can be exerted, because
anyincrease in the leverage of a musclewill, other things beingequal, increase
the force and decrease the speed. This places a constraint on the kinds of
animalsthat are possible. Ofcourse, a changein the properties of muscl,e might
makeit possible to increase speed and strength simultaneously, but such
changesare likely to be rarer than merechangesof shape.

In a secondexample,Williams(71) suggests that somegenetic changesthat
increase the efficiency of an animalwhenyounghave an adverse effect on the
same animal whenold. For example, Rose &Charlesworth (54) found pleio-
tropic effects of genesaffecting life history characteristics in Drosophilawith
high fecundity whenyoungassociated with a shorter life span. This is not
surprising, because treatments that prevent’females from laying eggs (even
sterilization by Xrays) prolonglife (31).

Theseexamplesillustrate the rather obviouspoint that the range of pheno-
types possible to a species is constrained. Thecauses vary fromunbreakable
laws of nature, like the law of levers, to h~istorical features of development.
Often the causes are unknownW. hensuch constraints exist, they will present
themselves to a geneticist as cases of pleiotropism. Somegeneticists have
argued that pleiotropism is unimportant as a constraint on evolution (37)
because genetic correlations can be changed(1). In all probability wecan
expect the full range betweeneasily brokencorrelations and almost unbreak-
able ones. Either way, however, such continuous constraints cannot explain
stasis. Theyexplain whyspecies cannot evolve in any conceivable direction,
but not whythey do not changeat all. For example,directional selection in
Drosophilacould increase fecundity at the expenseof longevity or vice versa.
If a species remainsat a particular point on the constraint surface for long
periods, it mustbe becauseof normalizingselection and not becausethere is
nowhereit can go.

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org Annual Reviews
by University of Wisconsin - Madison on 08/30/05. For personal use only. www.annualreviews.org/aronline

14 MAYNARDSMITH

b. Constraints Leading to Discontinuous Variation

There are two reasons whyvariation at the phenotypic level maybe discon-
tinuous. First, it is to be expected on theoretical groundsthat continuous
variation in someparameter at one stage of developmentwill give rise to
discontinuousvariation later; suchbifurcations are characteristic of all complex
dynamicsystems. Asimple illustration is afforded by Turing’s (63) reaction-
diffusion theory of morphogenesis,accordingto whichcontinuousvariation in
reaction and diffusion rates causesno changein the final pattern produceduntil
a threshold is reached beyondwhicha newpattern appears. Oster et al (47)
makethe samepoint in discussing a visco-elastic modelof morphogenesis.
Empirically, manycases are knownin whichcontinuous changeat the genetic
level causes discontinuous phenotypic change(2, 27, 38, 41, 52, 75).

There is a second wayin whichthe nature of the developmentalsystem may
imposediscontinuities on phenotypic variation. Weknowthat different gene
loci are activated in different tissues or in different morphogenetifcields (36).
This makesit possible, for example,for serially homologoupsarts to develop
differently. In both arthropods and vertebrates, differentiation of initially
similar parts has been an important process in evolution. This mustoften have
required the evolution (e.g. by gene duplication) of newelementsin the system
of genetic regulation. Thechangefromthe activation of the samegene in two
morphogenetifcields to the activation of different genesis essentially discon-
tinuous.

Therelevance of discontinuous phenotypicvariation to evolutionary stasis
depends on its cause. In the case of a threshold response to continuous
variation, it couldbe that the threshold is never crossedbecausethe underlying
variable is selectively maintained well awayfrom the threshold, perhaps by
pleiotropic effects. If, on the other hand, newphenotypicvariation requires
newregulating elements, the necessary mutational events maybe very infre-
quent. Either way, new phenotypes on which selection could act arise very
rarely.

It is convenient to use the term macromutationfor any genetic change
leading to a stalking change in phenotype, even if the change is a point
mutation. Suchmacromutationsare likely to be ill-adapted until compensating
changeshave occurred at other loci. This led Fisher (18) to argue that such
mutations are unimportant in evolution. I have never found this argument
entirely convincing(37; see also 49). It is true that a large randomchangein
car engine wouldmakeit less efficient, but organismsare not motorcars.
Developmentis regulated, so that one change will be compensatedfor by
others, withoutwaitingfor further mutation;in a striking example,Sliyper (56)
describes a complexseries of adaptive changesin the backboneand associated
musculature of a goat born without forelegs and therefore forced to adopt a
bipedal gait. Hence,althoughthere is little to recommenGdoldschmidt’s(23)

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org Annual Reviews
by University of Wisconsin - Madison on 08/30/05. For personal use only. www.annualreviews.org/aronline

GENETICSOF STASIS ANDPUNCTUATION

conceptof systemicmutations according to whicha newmorphologyrequires a
complexchange in chromosomalstructure, one cannot reject a priori his
concept of "hopeful monsters," interpreted as the initiation of evolutionary
novelty by macromutations.

Thereis nothingparticularly revolutionaryaboutthis idea. It doesnot require
that the newmutantbe reproductively isolated fromthe ancestral population;
like any other mutant, its establishmentin the population dependson selective
superofity since it could hardly be neutral. Noris it plausible that a new
complexadaptationcould arise in a single mutationalstep. Whatis plausible is
that, if a macromutatioins established, newselective forces will act on other
loci, leading to further rapid evolution.

I therefore see nointrinsic implausibility in the idea that macromutationasre
important in evolution and maymakepossible changes that wouldnot occur
gradually. It is harder to decide whether, and if so in which cases, such
mutationshavein fact beenimportant. Themostdirect approachis to study the
genetics of species hybrids: if twoclosely related species differ morphological-
ly, is the difference causedby manygenes, eachof small effect, or primarilyby
one or a few genes?

3. THE GENETICS OF SPECIES HYBRIDS

The idea that there is somethingqualitatively peculiar about the differences
betweenspecies is not new;it washeld by the early Mendeliansi,n particular by
de Vries and Bateson. This belief led to extensive workon the genetics of
hybridsbetweenspecies and subspecies, particularly in plants. Themainresult
of these studies wasthe conclusionthat species differences are similar in kind
to, althoughgreater in extent than, those betweenindividuals of a species. This
conclusion contributed to the establishment of the "modemsynthesis" in the
1940s.

Thereis onerespect in whichspecies hybridsdo differ: often the hybrids are
of reducedinviability or fertility. I return to this subject in section5. For the
present it is sufficient to maketwopoints. First, everydegreeof isolation can be
observed,frominviability of the F1 to minorinfertility in the F2or backcross
generations; the degree of isolation correlates poorly with morphologicadl if-
ference. Second, breakdowndepends on genes at manyloci; a classic example
is Dobzhansky’s(12) study of hybrids betweenDrosophilapseudoobscuraand
D. persimilis.

Moreimmediatelyrelevant is whethermorphologicaldifferences are typical-
ly polygenic, as Lande(33) and Charlesworthet al (7) argue. Clearly nothing
can be learned unless the F1 hybrids are sufficiently fertile to give F2 or
backcross progeny. If a difference is highly polygenic, then the F2 will be
hardly morevariable than the F1, and there will be no clear character segrega-

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org Annual Reviews
by University of Wisconsin - Madison on 08/30/05. For personal use only. www.annualreviews.org/aronline

16 MAYNARDSMITH

tion in the F2 or backcross. If few genes are involved, the F2 will be more
variable than the FI. Wright(quoted in 6) established a methodof estimating
the numberof loci involved. Lande (35) has extended the methodand has
applied it to six data sets; the twocases (eye size in cavefish, headwidthin
Drosophila)involving morphologicaldifferences betweenwild populations are
discussed below. There is also a semantic issue: how manyloci makea
difference polygenic? Asa roughguide, if five or moreloci of approximately
equal effects on sometrait are segregating, the populationwill respondsmooth-
ly to selection andsingle geneeffects will not be detectable. For this reason, I
suggest that such cases be regarded as polygenic.

Theearly workon plants is summarizedby Stebbins (61, p. 259) as follows:

AlthougthheFIprogenoyfaninterspecificrossareusuallyas muclhike eachotherasare
thedifferenitndividuaolsftheparentaslpeciest,heoffsprinign theF2andlatergenerations
areextremevlyariabled, ueto Mendelisaengregatioonfthegeneticfactorsresponsibfleor
theinterspecifidcifference..s.. Thestrikingfactaboumt anoyf theseprogenieiss notonly
theirvariabilitybutalsothepresencoefvariantswhiclhookas if theyhaveentirely"new"
characteristic.s... In somoeftheseprogeniesu, chas Quercutsy,pescloseto theoriginal
parentscanberecovereindtheF2generatiobnu, tin mosotthersthisis notpossibluenlessa
verylargenumbeorfindividualiss raised.

This testimonyis ambivalentas far as the polygenicnature of differences is
concerned:the difficulty of recoveringthe parental types supports a polygenic
interpretation, but the extremevariability of the F2 points the other way.

Stebbins’s conclusionsare borneout by the original paperson whichthey are
based. Afew examples(not all quotedby Stebbins) follow. Kristofferson (30)
studied species crosses in Malva,a herbaceousperennial. In the cross M. crispa
x M. neglecta, whichinvolved considerable sterility, sometraits gave mono-
factorial segregations---e.g. "crisp" (wavy-edged)and dissected leaves--and
others, such as flower size and carpel number,werepolygenic. In the backcross
to M. neglecta, manyplants wereoutside the range of the original parents and
"someof the plants resembledscarcely Malva."Muntzing(46) studied species
hybrids in deadnettles, Galeopsis Tetrahit x G. bifida. Theparental species
differ strikingly in leaf shape. TheF2 is morevariable than the F1, but less so
than wouldbe expected if the difference wascaused by only one or twoloci.

Clausen(9) studied hybrids betweenecotypes of a spring annual, Layia, in
California. Themaritimeecotype differed fromthe inland formin the absence
of a central stem, horizontal side branches,andlater flowering.For eachtrait,
the full difference could be causedby as few as twoloci, but different pairs of
ecotypesdiffered at different loci. Afinal example(20) tells us little aboutthe
genetic basis of the difference but is interesting in suggestinga developmental
threshold. In Geummontanumthe fruits are wind-dispersed, with a long
feathery plume;in G. rivale the fruits are hookedandare dispersed byanimals.

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org Annual Reviews
by University of Wisconsin - Madison on 08/30/05. For personal use only. www.annualreviews.org/aronline

GENETICSOF STASIS ANDPUNCTUATION 17

In the F1 hybrids, both fruit types are foundon a single plant, suggesting a
threshold. Genetically G. montanumis tetraploid and G. rivale hexaploid, and
it is likely that the genesresponsible for the hookedfruits wereintroducedby
hybridization with another genus.

Turning to animals, the data on Drosophila are disappointing, mainly
because species that can be crossed are morphologicallysimilar. Coyne(10)
has studied hybrids betweenD. melanogasterand D. simulans and betweenD.
simulans and D. mauritiana. The only morphologicaldifference betweenthese
species is in the shape of the posterior process of the malegenital arch. The
formerpair differ by at least four genesubstitutions affecting this characterand
the latter (morerecently diverging)pair byat least five substitutions; in each
case, this is the largest numbeor f substitutions that couldhavebeendetectedby
the methodsemployedI.n the latter pair, geneeffects of different chromosomes
wereapproximatelyadditive; in the formerpair, they wereless than additive.
Muller(45) concludedthat species differences are polygenicbut offered little
evidence. He quotedSpencer’s study ofD. virilis × D. americanahybrids, but
the published paper (58) showsthat the morphologicaldifferences betweenthe
species are too small for any conclusionto be drawn. Patterson & Stone (48)
reviewextensive data on the viability and fertility of species hybrids but say
nothing relevant to our present question. Themostrelevant data are those of
Val (65), analyzed by Templeton(62) and Lande(35), D. sil vestris × D.
heteroneura hybrids. Landeestimates that the dramatic difference in head
shape is caused by genes at 6-9 loci.

Recently, Garcia-Bellido (22) has reviewed the morphologicaldifferences
amongDrosophilaspecies, drawingattention to a series of qualitative differ-
ences, suchas the numbeor r type of rowsof chaetae (innervatedbristles) on the
wingmargin,the presenceof chaetae at particular sites onthe headand thorax,
and the presenceof rowsof special chaetae on the forelegs of males. Heregards
these traits as the stable states of a bifurcating developmentaplrocess because
they are uniform within species, and because (whatever phylogeny one
assumes) at least sometraits have evolved several times independently.
Mutantsexist that producechaeta patterns characteristic of other species (21).

In viewof Garcia-Bellido’s suggestion that differences in chaeta patterns
amongspecies mayrepresent different stable developmentasl tates, particular
interest attaches to anygenetic analysis of such cases. Theonly exampleknown
to meconcernsadditional rowsof chaetae on the forelegs of the male of some
species of HawaiianDrosophila(5). In five closely related species, there are
twoadditionalrowsof bristles (cilia) onthe dorsal surface of the foreleg tibias
that are used to stimulate the female during courtship. In one species, D.
silvestris, somepopulationshave the typical tworows, but in others the males
have a muchlarger numbeor f cilia with a greatly increased variance. Thereare
additional cilia in the marginalrows, anda new,irregular, intermediate row.A

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org Annual Reviews
by University of Wisconsin - Madison on 08/30/05. For personal use only. www.annualreviews.org/aronline

18 MAYNARDSMITH

genetic analysis of the between-populatiodnifferencein D. silvestris is still to
be published, but Bryant & Carson (4) have published a brief account
hybrids (F1, F2, and backcross)betweenD. silvestris with the additional cilia
andD. heteroneura,one of the species with only tworows. Theresults indicate
that the interspecific difference is causedbya small numberof sex-linked and
autosomal genes.

Thelarge variance of the D. silvestris populationswith additional cilia is
interesting. It suggests that the newphenotypehas not yet been stabilized by
selection of further modifyinggenes. Thusthere is no reason to think that a
phenotypewith three rowsof bristles is intrinsically variable as a consequence
of somedeep feature of Drosophiladevelopment.Waddington(67) argues that
phenotypicuniformity wascaused by stabilizing selection, and there is evi-
dence(41, 52, 53) that the chaeta pattern can be madeless variable by artificial
selection.

The pattern of chaetae in Drosophilahas also been altered by artificial
selection (41, 52, 53)~i.e. by introducing a major mutant that disrupts the
typical pattern and then selecting on tlie variation so produced.Sondhi(57) was
able in this wayto produceflies with a bristle characteristic of anotherfamily.

Turning to other animals, Wilkens(70) analyzed the genetic basis of eye
degeneration in two cave populations of fish related to the surface-living
Astyanax(the cave formsare best regarded as subspecies, although placed in
different genus). Thedifference is polygenic;Lande(35) estimates six loci.
contrast, differences in color pattern seemoften to be causedby a few major
loci with modifiers (8, 59, 64). These differences involve pattern as well
pigments and in two cases involve accurate mimicry.

Perhapsthe moststriking differences to be studied in hybrids are those found
by Danforth (11) between the commonpheasant, Phasanius, and the golden
pheasant, Chrysolophus.For example,malesof the latter species have a cape
of some200 feathers, modified in size, shape, color pattern, and detailed
morphologyt,hat can be raised to forman arc aroundthe eye. Onlyone F1 male
was obtained; it almost completely lacked a cape. Backcrossing to Chryso-
lophus again producedonly a single male, whichpossessed a well-developed
cape. In the next backcrossgeneration, consisting of several males, the cape
was uniformand approachedperfection. Danforth concludes that "the occur-
rence of the highly distinctive Chrysolophuscapecould be attributed to... one
mainand only a few reinforcing genes." He draws a similar conclusion for
other features, such as the crest and modifiedwingsand tail.

Danforth’sconclusionis consistent with his evidence, but I find it hard to
believe that such a complexstructure as the Chrysolophuscape, requiring
integration of feather movements,hape, and bandingpattern to formconcentric
rings roundthe eye, couldhavearisen in evolution by a single mutationalstep.
Thedata can equally well be explainedby supposingthat the full cape requires a

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org Annual Reviews
by University of Wisconsin - Madison on 08/30/05. For personal use only. www.annualreviews.org/aronline

GENETICSOF STASIS ANDPUNCTUATION 19

numberof almost dominantgenes that are expressed only in the presence of a
single recessive controlling gene. Thepresence of such a control is mademore
plausible by the fact that the cape developsonly in males. Unfortunately,the
absence of female hybrids makesa full genetic analysis impossible. Wemust
remembeorne point, however;the fact that a trait segregates in a cross as if
causedby a single gene does not prove that it arose in evolution in a single
mutational step, since wemaybe observingthe segregation of a gene control-
ling manyothers. In the sameway,the mutant tetraptera in Drosophila,which
replaces halteres withwings,doesnot provethat halteres arose in a single step.

These and other data on species hybrids lead to the conclusion that some
differences are polygenic (e.g. head width in D. silvestris, genitalia in D.
melanogaster,eye reduction in cave fish) and others involve majorgenes (e.g.
crisp leaves in Malva,horizontal branching and absence of central stem in
Layia ecotypes, color patterns in newtsand butterflies, andperhapssecondary
sexual characteristics in pheasants). This maybe a disappointingconclusionto
those wholike clear-cut answers,but I cannotsee that it is a surprising one. I
remain open-mindedabout the possibility that developmentmayimpose dis-
continuousconstraints on the pattern of phenotypicvariation. If so, mutations
of large phenotypiceffect maysometimesinitiate newevolutionary departures.
It is also possible that, without any need for developmentalbifurcations, a
populationmayadoptnewhabits, either throughlearning and cultural transmis-
sion or from environmentally imposed necessity. If so, newhabits would
imposenewselective forces and henceinitiate punctuational change(37, 68,
69).

4. THE PATTERN OF VARIATION IN NATURE

Whateverconclusions we maydraw from species hybrids, it remains true that
there are discontinuities in the pattern of variation in nature, at least at anyone
time and place, that makeit possible to classify organismsinto species. Why
shouldthis be so? Thereare three possible answers:that the discontinuities in
organismsreflect discontinuities in the environment(i.e. there are discrete
ecological niches); that they are imposedby development;or that they are
consequencesof sexual reproduction. Bateson(3) argued that since environ-
mentalvariation is usually continuousit couldnot accountfor specific distinc-
tions; therefore, "the discontinuity of species results fromthe discontinuityof
variation."

In drawingthis conclusion, Batesonoverlookedthe fact that sexual repro-
ductionitself places a limitation onthe rangeof phenotypicvariation possible
to a single species, becausehybridsbetweenvery different parents are likely to
be of lowfitness. Thusthe possibility arises that discontinuities are a conse-
quenceof sexualreproduction.Thedecisive test of this hypothesisis the pattern

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org Annual Reviews
by University of Wisconsin - Madison on 08/30/05. For personal use only. www.annualreviews.org/aronline

20 MAYNARDSMITH

of variation in apomictictaxa; if the hypothesisis correct, then agamiccom-
plexes shouldfail to showdiscontinuities and shouldresist classification into
species. This seemsto be true for higher plants such as TaraxacumH, ieracium,
and Rubus(26, 61).

The situation is less clear in animals. In sometaxa (e.g. Cnemidophorus
lizards) the uniformity and discreteness of parthenogenetic"species" probably
reflect their recent origin. Howevert,he rotifer order Bdelloidearepresents a
serious problem.It consists of some200species classified into four families,
yet no male has ever been discovered. Clearly parthenogenesis has set no
absolutebarrier to evolutionarychange,but there are indications in the taxono-
micliterature that classification into species is arbitrary. Ruttner-Kolisko(55),
writing of rotifers in general, says "whereas generic categories are well-
defined, variability within genera is very great, so that the establishment of
clear species boundaries in many~nemight almost say all~ases is extreme-
ly difficult. Veryfrequentlythis has led to an excessivesplitting of genera....
Wehave here in the zoological field conditions similar to those in the plant
genera Hieracium and Rubus." Reporting a workshopon rotifer taxonomy,
Dumon(t15) writes: "C. Ricci remarkedthat she found it impbssible to apply
species concepts to Bdelloids. J. Donneradmitted that all species namesin
Bdelloids should be regarded as preliminary." Thusit seemsthat Bdelloids do
not offer anyserious challengeto the viewthat discontinuities betweenspecies
exist only if there is sexual reproduction.Howeveri,t is not clear whetherthe
difficulty of applyingthe species conceptto Bdelloidsis peculiar to that asexual
taxon, or whetherit applies also the cyclically parthenogenetic monogononts;
rotifers are not the easiest animalsto study morphologically.

Thus in plants the data on parthenogenetic taxa support the view that
discontinuities betweenspecies exist only if there is sexual reproduction. The
data on animals neither contradict nor strongly support this conclusion.

5. CAN STASIS BE EXPLAINED BY LARGE
POPULATION SIZE?

Mayr(43) supports his argumentfor the importanceof peripheral isolates
speciation by claiming that large, widespreadpopulations are less able to
evolve. This claim has been taken over by the punctuationists (e.g. 25, 60).
This claimmightbe justified in twoways.First, in termsof selective forces, it
could be arguedthat a widespreadpopulation musthave achieveda satisfactory
fit with its environmenatndtherefore will not be subject to directional selec-
tion. Alternatively,the direction of selection will differ in different parts of the
range, andgeneflowwill prevent evolution. Thefirst alternative explains stasis
in terms of normalizingselection and is discussed in section 6. Thesecondis

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org Annual Reviews
by University of Wisconsin - Madison on 08/30/05. For personal use only. www.annualreviews.org/aronline

GENETICSOF STASIS ANDPUNCTUATION 21

probablyfallacious, since different selective pressures in different places will
lead to speciation without the need for majorgeographical barriers (17).

Howevers,elective justification is not the argumentthe punctuationists have
in mind.Instead, they believe that there is somethingpeculiar aboutthe genetic
structure of large populationsthat makesevolutiondifficult or impossible.This
is mistaken:insofar as the rate of evolution is limited by mutation,the exact
opposite will be the case [40; but see Lande(34) for an estimate that the
mutation rate is sufficient to maintainrapid evolution even in rather small
populations]. However,one evolutionary process mayoccur faster in a small
population than in a large one. This is evolution from one state, say ab, to
another of higher fitness, say AB,whenthe intermediate states, AbandaB, are
of lowfitness. Such"adaptive valleys" can arise either fromepistatic fitness
interactions or fromheterozygoteinferiority. Perhapsthe strongest reasonfor
thinking that such events sometimesoccur is that the chromosomeosf related
species or subspecies often differ structurally in a waythat wouldlower the
fertility of heterozygotes. Lande(32) estimates typical demesizes from the
frequencyof such transitions.

If population structure permits chromosomeevolution, it will presumably
also permitthe crossing of valleys arising fromepistatic interactions between
genes. Howeveri,t does not follow that innovationwill occuronly in peripheral
isolates. Wright(74, 76) imaginesthat a valley mayfirst be crossed by any
demeand that the newgenotypes, once formed, will subsequently spread
through the whole species. Onthe time scale studied by paleontologists,
Wright’s"shifting balance"theory is a modelof phyletic gradualismrather than
of punctuationalevents in peripheral isolates.

Asin the case of macromutationsw, ecannot rule out a priori the evolution-
ary relevanceof the crossingof adaptivevalleys by small local populations,but
it is hard to decide howimportant such events have been in practice. Except,
perhaps, in the case of structural changesin chromosomesth, ere is no obvious
necessity to assumesuch events. In particular, the widespreadoccurrence of
inviability in species hybrids mightsuggest that an adaptive valley has been
crossed, but this is not so. It is easy to construct modelsin whichan infinite
random-matingpopulation wouldevolve fromstate 1 to state 2 and yet hybrids
betweenthe twostates wouldbe of lowfitness. In a haploid this requires at
least three independent loci, but two loci are sufficient in a diploid with
dominance.

Also, it is possible for a population to movefrom one adaptive peak to
another without the need for chanceevents in a small population. Kirkpatfick
(29) showsthat, if there are twofitness peakswith an interveningvalley, a large
population with polygenic variance can shift several standard deviations in a
few tens or hundreds of generations. Sucha shift could be initiated by an

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org Annual Reviews
by University of Wisconsin - Madison on 08/30/05. For personal use only. www.annualreviews.org/aronline

22 MAYNARDSMITH

environmentalchange(altering the fitness surface) or by a mutationally in-
duced changein the variance of the population.

Hencethe presenceof strongepistatic fitness effects in populationhybrids is
not evidence for valley-crossing. Recently "molecular drive" has been pro-
posedas an explanationfor hybridinviability andinfertility (13, 14, 19). The
suggestion is plausible, althoughlarge differences in repetitive DNAhave in
somecases been shownto have little effect either on the phenotype or on
chromosompeairing in meiosis (44, 51).

6. CAN STASIS BE EXPLAINED BY NORMALIZING
SELECTION?

Ageneticist can say little aboutwhetherstasis can be explainedby normalizing
selection. Clearly if selection continues to favor the samephenotypefor many
millions of years, then stasis will be the result. It mayseemimplausiblethat
selection could be as conservative as this. However,we must distinguish
betweenthe physical and the biotic features of the environment.Thephysical
environment has been changing rather rapidly during the past few million
years. Howeverw, henthe climate changes, organismstend not to stay still and
adapt to the change; instead, populations moveso that physical conditions
remainconstant and are replaced by other species at the original site.

It is harder to knowwhatto expect of the biotic environment.If we assume
that physical conditions are constant andthat eachspecies evolvesonly because
its competitors, predators, and parasites are evolving, wecan drawtwopossi-
ble conclusions about the resulting dynamics(39; N. Stenseth, J. Maynard
Smith, in preparation). Oneis that evolution will continue at a steady rate,
including extinction, speciation, and phyletic change, as imaginedin Van
Valen’s(66) "RedQueen"hypothesis. Thealternative is that evolution of all
species slows downand stops, being kicked into motion occasionally by
changesin the physical environment. The choice betweenthese alternatives
does not seempossible on theoretical groundsand so will dependon paleonto-
logical evidence.If the latter is correct, it will explainhighlyvariableevolution
rates in selective terms but will predict that rapid changes should occur
simultaneouslyin interacting species.

7. CONCLUSIONS

Directional selection acting ongenetic variability of a conventionalkind is able
to producechangesat a rate that wouldappear virtually instantaneous in the
fossil record. Thereis, however,a question as to howthe observedvariability
of evolution rates should be explained. Variation in rate maysimply reflect
variation in the intensity of directional selection; punctuationalchangeoccurs

Annual Reviews
www.annualreviews.org/aronline

GENETICS OF STASIS AND PUNCTUATION 23

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org whena population crosses a selective threshold. Alternatively, variation in rate
by University of Wisconsin - Madison on 08/30/05. For personal use only. maybe caused by developmental constraints of a discontinuous kind, arising
either from bifurcations in development or from changes in systems of gene
regulation.

Data on species hybrids show that the genetic basis of species differences is
similar in kind to that of variation within species. Somespecies differences

show polygenic inheritance, and others are caused by one or a few major genes.
The fact that a trait behaves in a cross as if caused by a single gene does not

prove that it arose in evolution by a single mutational step, because wemaybe
observing the segregation of a gene regulating the activity of many others.

Existing organisms, at least in a single place, typically show sharp discon-
tinuities between species. This is not evidence for any discontinuities in the

range of possible organisms imposed by developmental constraints, because
the discontinuities are probably associated with sexual reproduction. The

inviability and infertility of species hybrids are not evidence for the crossing of
adaptive valleys in evolution, because a large population can evolve determi-
nistically from one state to another and yet the hybrids between the two states
can be of low viability.

Literature Cited

1. Atchley,W.R., Rutledge,J. J. 1980. 9. Clausen,J. 1958.Genesystemsregulat-
Geneticcomponenotfs size andshape.I. ing characters of ecological races and
Dynamicosf phenotypevariability and subspecies,lOth Int. Congr.Genet.1:
covariabilty during ontogenyin the 434-43
laboratoryrat. Evolution34:1161-73
10. CoyneJ,. A. 1983.Geneticbasis of dif-
2. BatemanK, . G. 1959.Geneticassimila- ferences in genital morphologyamong
tion of four venation phenocopies.J. three sibling species of Drosophila.
Genet. 56:443-74 Evolution.In press

3. Bateson, W. 1894. Materials for the 11. Danforth, C. H. 1950. Evolution and
Studyof Variation.London:MacMillan. plumagteraits in pheasanthybrids,Pha-
598pp. sanius X Chrysolophus.Evolution 4:
301-15
4. Bryant, P. J., Carson,H. L. 1979.Ge-
netics of an interspecificdifferencein a 12. DobzhanskyT,. 1936.Studieson hybrid
secondarysexual character in Hawaiian sterility. II. Localizatioonf sterilityfac-
Drosophila.Genetics91:515-16 tors in Drosophilapseudoobscurahy-
brids. Genetics21:113-35
5. Carson,H. L., Val,F. C., SimonC, .M.,
Archie, J. W.1982.Morphometreicvi- 13. Dover, G. 1982. Molecular drive: A
dencefor incipient speciationof Dro- cohesivemodeof speciesevolution.Na-
sophila silvestris fromthe island of ture 299:111-17
Hawaii. Evolution36:132-40
14. Dover,G., BrownS, ., Coen,E., Dallas,
6. Castle, W. E. 1921. An improved J., Strachan, T., Trick, M.1982.The
methodof estimatingthe numbeor f ge- dynamicsof genomevolution and spe-
netic factorsconcerneidn casesof blend- cies differentiation. In GenomEevolu-
ing inheritance.Science54:223 tion, ed. G.A. DoverR, .B. Flavell,pp.
343-72. London/NewYork: Academic.
7. Charlesworth,B., Lande,R., Slatkin, 382pp.
M.1982. A neo-Darwiniancommentary
on macroevolution.Evolution 36:474- 15: DumontH, . J. 1979. Workshoopn tax-
98 onomyandbiogeographyP. roc. 2ndInt.
Rotifer Symp., pp. 205-06.TheHague:
8. Clarke, C. A., Sheppard,P. M.1959.
The genetics of somemimeticforms of Junk.
Papilio dardanus,Brown,and Papilio
glaucus, Linn.J. Genet.56:236~a0 16. Eldredge,N., Gould,S. J. 1972.Punctu-
ated equilibria:Analternativeto phyletic

Annual Reviews
www.annualreviews.org/aronline

24 MAYNARD SMITH

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org gradualism. In Modelsin Palaeobiology, 3zt. Lande, R. 1980. Genetic variation and
by University of Wisconsin - Madison on 08/30/05. For personal use only. ed. T. J. M. Schopf, pp. 82-115. San phenotypic evolution during allopatric
Francisco: Freeman, Cooper. speciation. Am. Nat. 116:463-79
17. Endler, J. A. 1977. GeographicVaria-
tion, Speciation and Clines. Princeton: 35. Lande, R. 1981. The minimumnumber
Princeton Univ. Press. 246 pp. of genes contributing to quantitative
18. Fisher, R. A. 1930. The Genetical variation between and within popula-
Theory of Natural Selection. Oxford: tions. Genetics 99:541-53
Clarendon. 291 pp.
19. Flavell, R. B. 1982. Sequenceamplifica- 36. Markert, C. L. 1963. Epigenetic control
tion, deletion and rearrangement: Major of specific protein synthesis in dif-
sources of variation during species di- ferentiating cells. In Cytodifferentiation
vergence. See Ref 14, pp. 301-23 and MacromolecularSynthesis, ed. M.
20. Gajewski, W. 1963. The heredity of Locke, pp. 65-84. NewYork: Academic
seed-dispersing mechanisms in Geum.
37. MaynardSmith, J. 1958. The Theory of
11th Int. Congr. Genet. 2:423-30 Evolution. Harmondsworth: Penguin.
21. Garcia-Bellido, A. 1981. Fromthe gene 320 pp.

to the pattern: Chaetadifferentiation. In 38. Maynard Smith, J. 1960. Continuous,
Cellular Controlsin Differentiation, ed. quantized and modal variation. Proc.
C. W.Lloyd, D. A. Rees, pp. 281-304. Roy. Soc. LondonSer. B 152:397-409
NewYork: Academic. 324 pp.
22. Garcia-Bellido, A. 1983. Comparative 39. MaynardSmith, J. 1976. A commenton
anatomyof cuticular patterns in the genus the red queen. Am. Nat. 110:325-33
Drosophila. In Developmentand Evolu-
40. Maynard Smith, J. 1976. What deter-
tion, ed. B. C. Goodwin,N. J. Holder, mines the rate of evolution? Am. Nat.
C. C. Wylie. London/NewYork: Cam- 110:331-38
bridge Univ. Press. In press "
23. Goldschmidt, R. B. 1940. The Material 41. MaynardSmith, J., Sondhi, K. C. 1960.
Basis of Evolution. NewHaven: Yale Thegenetics of a pattem. Genetics 45:
Univ. Press. 436 pp. 1039-50
24. Gould, S. J. 1980. Is a newand general
theory of evolution emerging?Palaeobi- 42. Mayr,E. 1942. Systematics and the Ori-
ology 6:119-30 gin of Species. NewYork: Columbia
25. Gould,S. J., Eldredge, N. 1977. Punctu- Univ. Press. 334 pp.
ated equilibria: The tempoand modeof
evolution reconsidered. Palaeobiology3: 43. Mayr, E. 1963. Animal Species and
115-51 Evolution. Cambridge, Ma: Harvard
26. Grant, V. 1981. Plant Speciation. New Univ. Press. 797 pp.
York: Columbia Univ. Press. 563 pp.
2nd ed. 44. Miklos, G. L. G. 1982. Sequencing and
27. Gruneberg, H. 1963. The Pathology
of Development. Oxford: Blackwell. manipulating highly repeated DNA.See
Ref. 14, pp. 41-68
267 pp. 45. Muller, H. J. 1940. Bearings ofthe"Dro-
28. Jones, J. S. 1981. Anuncensoredpage of sophila" work on systematics. In The
NewSystematics, ed. J. Huxley, pp.
fossil history. Nature293:427-28 185-268. London: Oxford Univ. Press.
29. Kirkpatrick, M. 1982. Quantumevolu- 583 pp.
46. Muntzing,A. 1930. Outlines to a genetic
tion and punctuated equilibria in con- monographof Galeopsis. Hereditas 13:
tinuous genetic characters. Am.Nat. 199: 185-341
47. Oster, G., Odell, G., Alberch, P. 1980.
833-48 Mechanics, morphogenesis and evolu-
30. Kristofferson, K. B. 1926. Species cross- tion. Lect. Math. Life Sci. 13:165-255
48. Patterson, J. T., Stone, W. S. 1952.
ings in Malva. Hereditas 7:233-354 Evolution in the genus Drosophila. New
31. Lamb,M. J. 1964. Theeffects of radia- York: MacMillan. 610 pp.
49. Rachootin, S. P., Thomson,K. S. 1981.
tion on the longevity of femaleDrosophi- Epigenetics, palaeontology, and evolu-
la subobscura. J. Ins. Physiol. 10:487- tion. Proc. 2nd Int. Congr. Syst. Evol
97 Biol., pp. 181-93 Pittsburgh, PA: Hunt
32. Lande, R. 1979. Effective demesizes Inst. Botan. Doc., Carnegie-Mellon
during long-term evolution estimated Univ. 486 pp.
from rates of chromosomalrearrange- 50. Raup, D. M. 1966. Geometricanalysis of
ments. Evolution 33:234-51 shell coiling: General problems. J.
Paleontol. 40:1178-90
33. Lande, R. 1980. Microevolutionin rela- 51. Rees, H., Jenkins, G., Seal, A. G.,
tion to macroevolution.Palaeobiology6: Hutchinson, J. 1982. Assays of pheno-
233~38 typic effects of changes in DNA
amounts. See Ref. 14, pp. 287-97

Annual Reviews 25
www.annualreviews.org/aronline

GENETICS OF STASIS AND PUNCTUATION

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org 52. Rendel, J. M. 1959. Canalization of the evolution in Heliconius: A defense of
by University of Wisconsin - Madison on 08/30/05. For personal use only. scute phenotype of Drosophila. Evolu- neodarwinism.Ann. Rev. Ecol. Syst. 12:
tion 13:425-39 99-121
65. Val, F. C. 1977. Genetic analysis of the
53. Rendel, I. M., Shddon, B. L., Fiday, morphological differences between two
D. E. 1965. Canalization of development interfertile species of HawaiianDro-
of scutcllar bristles in Drosophilaby con- sophila. Evolution 31:611-29
trol of the scute locus. Genetics 52: 66. VanValen, L. 1973. A newevolutionary
law. Ecol. Theory 1:1-30
113%51 67. Waddington, C. H. 1957. The Strategy
54. Rose, M. R., Charlesworth, B. 1981. of the Genes. London: Allen & Unwin.
262 pp.
Genetics of life history in Drosophila 68. Waddingmn,C. H. 1959. Evolutionary
melanogaster. I. Sib analysis of adult adaptation. In Evolution after Darwin,
females. Genetics 97: 173-86; II. Ex- ed. S. Tax, pp. 381-402. Chicago: Univ.
ploratory selection experiments, pp. ChicagoPress.
187-96 69. Waddington,C. H. 1961. Genetic assim-
55. Ruttner-Kolisko, A. 1963. Theinterrela- ilation. Adv. Gen. 10:257-90
70. Wilkens,H. 1971. Genetic interpretation
tionships of the Rotatoria. In The Lower of regressive evolutionary processes:
Metazoa, pp. 263-72. Berkeley: Univ. Studies on hybrid eyes of two Astyanax
Calif. Press. cave populations (Charaeidae, Plsees).
56. Sliyper, E. J. 1946. Comparative bio- Evolution 25:530-44
logic-anatomical investigations of the 71. Williams, G. C. 1957. Pleiotropy, natu-
Avkeartde.brWalet.c, olAummsnteorfdammaAmfmd.alNsV.aeturhu.rkKd.. ral selection, andthe evolutionof senes-
cence. Evolution 11:398-411
42:1-128 72. Williamson, P. G. 1981. Morphological
57. Sondhi, K. C. 1962. Theevolution of a stasis and developmentaclonstraint: Real
problems for neo-Darwinism. Nature
pattern. Evolution 16:186-91 294:214-15
58. Spencer, W. P. 1940. Subspecies, hy- 73. Williamson,P. G. 1981. Palaeontologic-
al documentationof speciation in ceno-
brids and speciation in Drosophilahydei zoic molluscs from Turkana basin. Na-
and Drosophila virilis. Am. Nat. 74: ture 293:437-43
74. Wright, S. 1931. Evolution in Mendelian
157-79 populations. Genetics 16:97-159
59. Spurway,H. 1953. Genetics of specific 75. Wright, S. 1934. The results of crosses
betweeninbred strains of guinea pigs,
and subspecific differences in European differing in numberof digits. Genetics
newts. Syrup. Soc. Exp. Biol. 7:200-37 19:537-51
60. Stanley, S. H. 1979. Macroevolution. 76. Wright, S. 1970. Randomdrift and the
San Francisco: Freeman, 332 pp, shifting balance theory of evolution. In
61. Stebbins, G. L. 1950. Variation and Mathematical Topics in Population
Evolution in Plants. NewYork: Colum- Genetics, ed. K. Kojima,pp. 1-31. Ber-
bia Univ. Press. 643 pp.
62. Templeton, A. R. 1977. Analysis ofhead lin: Springer. 400 pp.
shape differences betweentwo inteffer-
tile species of Hawaiian Drosophila.
Evolution 31:330-41
63. Turing, A. M. 1952. The chemical basis
of morphogenesis. Philos. Trans. Roy.
Soc. LondonSet. B 237:37-72
64. Turner, J. R. G. 1981. Adaptation and

Annu. Rev. Genet. 1983.17:11-25. Downloaded from arjournals.annualreviews.org
by University of Wisconsin - Madison on 08/30/05. For personal use only.