Acknowledgments
Iamindebtedtomanyindividuals—scientists,historians,andphilosophers—fordiscussionsabout topicscoveredinthisbook,especiallyDavid McCandlish,PabloRazeto-Barry,SahotraSarkar, DavideVecchi,JohnBeatty,JoannaMasel,IngoBrigandt,EricHaag,ArnaudMartin,LeeAltenberg,and WesAndersonthePhilosopher;alsoWallaceArthur, LindleyDarden,Jean-BaptisteGrodwohl,Astrid Haase,PhilippeHunemann,YogiJäger,TomJones, ReesKassen,EugeneKoonin,FrancescaMerlin, KiraMakarova,RobertaMillstein,StuartNewman, JoshPayne,JoshPlotkin,JustinPritchard,Premal Shah,DavidStern,JayStorz,JakeWeissman,and JulianXue.IthankAndreiChabes,AlexCouce,Ron Ellis,DarinRokyta,GloriaRudenko,andPremal Shahforsharingdataorfigures.IthankIanShermanforencouragementandadvice,andCharles BathandthestaffatOxfordforproductionexpertise.Forcamaraderieduringthousandsofhoursof writinginthelocalcoffeeshop,IthankRamyar, Cary,Arcely,Elaine,Lauren,Lucas,Melanie,Nada, Ben,Peter,Valerie,andespeciallyIsabel.
Thisbooksurelyincludessomemistakes,which aremyown.Fortopicsthatarenotinmynarrow areaofexpertise,Ihavecitedrelevantworkwithout alwaysknowingthemostrecentormostrelevant work.Iapologizetothoseauthorswhoserelevant workwasnothighlighted.
Ithankthedevelopersoffreelyavailablesoftwareproductsinvolvedinthecreationofthis book:LATEX(TEXShop,MacTEX,BibTEX),R,RStudio, TextWrangler(BBEdit),andGraphViz.
IamparticularlyindebtedtoDavidMcCandlish. Everyweekforthepasteightyears,DavidandI havehadaregularlyscheduledone-hourconversation,typicallyaboutcollaborativeprojects,butalso aboutthisbookandanyotheraspectsofevolutionarybiology.Oninnumerableoccasions,hehas helpedmetoclarifymythinking,offeredincisive interpretationsofthetheoreticalliterature,andsuggestedhowtoexplaindifficultconcepts.Thisbook ispartlyanexpressionofourongoingdialog.
Listoffigures xi
Listoftablesandboxes xiii
1Introduction:acuriousdisconnect1
1.1Mutationaloriginationasanevolutionarycause1
1.2Whatthisbookisabout5
1.3Whothisbookisfor9
1.4Howtheargumentunfolds11
1.5Synopsis13 2Ordinaryrandomness15
2.1Introduction15
2.2Lackinginforesight16
2.3Uniformityorlackofpattern16
2.4Stochasticorprobabilistic20
2.5Indeterminate21
2.6Subjectivelyunpredictable22
2.7Spontaneous24
2.8Independent(part1)28
2.9Independent(part2)30
2.10Synopsis33
3Practicalrandomness35
3.1Introduction35
3.2Whatgoodisarandomnessassumption?35
3.3Uniformity39
3.4Independence40
3.5Predictability41
3.6Therandomnullhypothesis43
3.7Beyondrandomness:theprincipleofindifference44
3.8Synopsis45
4Evolutionaryrandomness47
4.1Introduction47
4.2Rejectionofpervasivelydirectedmutations47
4.3RejectionofLamarckism48
4.4Independencefromadaptationorevolution52
4.5Independencefromfitnesseffects53
4.6Exceptionsandpossibleexceptionstoindependence56
4.7Conditionalindependenceandrelatedideas60
4.8Mutationandaltereddevelopment63
4.9Synopsis64
5Mutationalmechanismsandevolvability67
5.1Introduction67
5.2Whataspeciallyevolvedmutationsystemlookslike68
5.3Specializedsystemsofgermlinemutationinmicrobes70
5.3.1Multiple-inversionsystems(shufflons)70
5.3.2Diversity-generatingretroelements71
5.3.3CRISPR-CasandpiRNAs71
5.3.4Multiplecassettedonation74
5.3.5Phasevariation75
5.3.6Mating-typeswitching76
5.4Formulatingplausiblescenarios77
5.5Challengesandopportunities82
5.6Conditionalindependenceandspecializedmutationsystems85
5.7Evolvability86
5.8Synopsis89
6Randomnessasirrelevance93
6.1Introduction93
6.2Argumentsfromanalogyandmetaphysics94
6.2.1The“rawmaterials”metaphor94
6.2.2Creativity95
6.2.3Levelsandtypesofcauses96
6.3Directempiricalarguments98
6.4Mechanisticarguments100
6.4.1Creativityarguments100
6.4.2Directionality:the“opposingforces”argument100
6.4.3Initiativeandrate:the“genepool”arguments102
6.5Themethodologicalargument103
6.6Theexplanatoryargument105
6.6.1Darwin’sarchitect106
6.6.2Laterarguments107
6.7Synopsis107
7Theproblemofvariation111
7.1Introduction111
7.2Thepowerofthemorphotron112
7.3Sourcelawsandconsequencelaws113
7.4TheMendelianchallenge116
7.5Thecontemporarychallenge118
7.5.1TheGmatrixaspredictor119
7.5.2Thechallengetogradualism120
7.5.3Thenewgeneticsofadaptation122
7.5.4Evo-devo124
7.5.5Molecularevolution:thecaseofcodonusagebias126
7.5.6Thegenomicchallengetoadaptationism130 7.6Synopsis132
8ClimbingMountProbable135
8.1Introduction135
8.2ClimbingMountProbable136
8.3One-stepadaptivewalksundermutationbias137
8.4Extendedadaptivewalksundermutationbias140
8.5Proteinadaptationundermutationbias143
8.6Origin-fixationdynamics145
8.7Thesushiconveyorandthebuffet146
8.8Whythetheoryofforcesfails149
8.9Thesourcesandformsofbiases152
8.10Understandingdevelopmentalbiasesasevolutionarycauses155
8.11Aninterpretationofstructuralism157
8.12Parallelevolution159
8.13Conditioningonmutationaleffects161 8.14Synopsis162
9Therevoltoftheclay165
9.1Introduction165
9.2Apredictivemodelofproteinsequenceevolution166
9.3Mutation-biasedadaptationinthelab169
9.4CpGmutationalhotspotsandaltitudeadaptation176
9.5Transitionbiasinnaturalparallelisms177
9.6Preferencesforregulatoryorstructuralchanges183
9.7Developmentalbias186
9.8Evaluatingtheargument188
9.8.1Crypticfitnessbiasesactuallyexplainthedata188
9.8.2Theconnectiontotheoryisthin189
9.8.3Selectiondidallthehardwork191
9.8.4Mutationonlyaffectstheboringparts192
9.9Synopsis194
10Movingon197
10.1Introduction197
10.2Summaryashistoricalnarrative197
10.3Asynopsisofkeypoints201
10.4Theobjectsandformsofexplanations203
10.5Theimportanceofverbaltheoriesofcausation206
10.6Discerningtheoriesandtraditions210
10.7Synopsis214
AppendixA:Mutationexemplars217
A.1Areplicationerror217
A.2Error-pronerepairofDNAdamage218
A.3Asymbolicmutationprocessinacomputerprogram220
A.4Human-engineeredmutations221
AppendixB:Countingtheuniverseofmutations223
B.1Preliminaries223
B.2Anecessarysimplification224
B.3Pointmutations224
B.4 Denovo insertions225
B.5Inversions,deletions,andtandemduplications226
B.6Transpositions(translocations)226
B.7Lateralgenetransfers227
B.8Compoundevents228
B.9Summingup228
B.10Recurrences229
AppendixC:Randomnessquotations231
C.1Introduction231
C.2Listofquotations232
AppendixD:Irrelevancequotations237
D.1Introduction237
D.2Listofquotations237
Bibliography 243 Index 267
Listoffigures
1.1Mutationrate,fitness,andfrequencyevolved2
2.1Ratesofmutationspanmanyordersofmagnitude17
2.2Thelengthdistributionofindelsinpseudogenes18
2.3Mutationstemplatedbyanimperfectpalindrome18
2.4Apatternindicatingcontext-dependenceofnucleotidemutations19
2.5Spontaneouscombustion24
2.6TheLuria-Delbrückfluctuationtest25
2.7SwitchingofVSGgenesin Trypanosomabrucei 27
2.8Thechanceconjunctionofnumbersontwodice29
2.9Rainfallplotshowingkataegisinapancreaticcancercellline30
2.10HowdNTPconcentrationsinfluencemutation31
2.11Negativecorrelationofmutationrateandcelldensity33
3.1Predictionisimprovedbyknowledgeofabias42
4.1Dependenceofexperimentaladaptationonmutation53
4.2Nucleotidechangesdiffersystematicallyinfitnesseffects57
4.3Singlet,doubletandtripletreplacementsdifferinfitnesseffects58
4.4Correspondenceofmutationandselectionacrossenvironments60
4.5Independence,conditionalindependence,anddependence61
4.6Dependenceofmutationandfitnessviaintermediates62
5.1SchemeofVDJrecombinationofantibodygenes69
5.2Schemeofaninversionshufflon70
5.3Schemeofdiversity-generatingretroelemens(DGRs)71
5.4HowCRISPR-Cas9confersresistance72
5.5Schemeofcassettedonation74
5.6SchemesofphasevariationbySTRandinversion76
5.7Schemeofyeastmating-typeswitching77
6.1Anillustrationofrawmaterials94
6.2Atheoryofevolutionbyshiftinggenefrequencies105
7.1Themorphotron,animaginarydevice112
7.2Fisher’sgeometricmodel121
7.3Expectedsizesofeffectsfixedinadaptation121
7.4Codonusage,mutationbias,andgeneexpressionlevel127
7.5Aneo-Darwinianmodeloflateralgenetransfer129
8.1Aruggedmountainlandscape136
8.2TheYampolsky-Stoltzfusmodel137
8.3EffectofpopulationsizeintheYampolsky–Stoltzfusmodel137
8.4Effectsofthesizesof s and u intheYampolsky-Stoltzfusmodel138
8.5Chemotaxiswithunbiasedorbiasedtumbling139
8.6SpecialcasesofevolutionarywalksonanNKlandscape141
8.7AdaptivewalksonsimplifiedsmoothandruggedNKlandscapes142
8.8GeneticcodetablewithAT-andGC-richcodons143
8.9AdaptivewalksonasmoothproteinNKlandscape144
8.10AdaptivewalksonaruggedproteinNKlandscape144
8.11Theorigin-fixationmodel145
8.12Thebuffetandthesushiconveyor147
8.13TheYampolsky-Stoltzfusmodelinthebuffetregime148
8.14Population-genetic“forces”operatinginanallele-frequencyspace149
8.15Transition-transversionbiasandGC:ATbias153
8.16Adrunkard’swalkthroughgene-scramblingspace154
8.17AmoleculardevelopmentalbiasinducedbyaGPmap155
8.18AbiasinducedbyaGPmap, via thephenocopyeffect156
8.19Networksofidenticalfolds(phenotypes)insequencespace158
8.20Thechanceofparallelismisasumofsquares160
8.21Thechancesofparallelismwithequivalenceclasses161
9.1Apotteratworkshapingclay166
9.2Observedtendenciesofaminoacidchange166
9.3Asimplifiedmodelofnucleotidemutationrates167
9.4Predictionofobservedtendenciesofaminoacidchange167
9.5Themutationalcomponentofthemodel,consideredalone168
9.6Distributionoffitnesseffectsfortransitionsvs.transversions168
9.7Mutationrate,fitness,andfrequencyevolved170
9.8Cefotaximeresistancewithdifferentmutationspectra170
9.9TwopredictionmodelsappliedtoRokytaetal.(2005)171
9.10Resultsofreplicateone-stepadaptationof4relatedbacteriophages173
9.11Transitionbiasinexperimentaladaptationof λ phage174
9.12ParallelchangesinATPα 1(glycosideresistance)178
9.13Adevelopmentalbiasinhermaphroditespermsize187
9.14FitnessresultsfromSackmanetal.(2017)189
A.1Pathwaysofproof-readingandmismatchrepair217
A.2ForwardandreversewobblepairsforAandC218
A.3Twokindsofpyrimidinephoto-dimers219
A.4Nucleotideexcisionrepairpathway219
B.1Modelfordeletions,inversionsandduplications226
B.2Modelforcountingpossibletranspositions227
Introduction:acuriousdisconnect
Toseewhatisinfrontofone’snoseneedsaconstantstruggle GeorgeOrwell
1.1Mutationaloriginationasan evolutionarycause
MacLeanetal.(2010)evolvedresistancetothe antibioticrifampicininmanyreplicateculturesof thebacterium Pseudomonasaeruginosa.Resistant strainstypicallyhavemutationsinthe rpoB gene,encodingRNApolymerase.For11resistant mutantsin rpoB,MacLeanetal.(2010)measured themutationrate,theeffectonfitness,andthe frequencyevolved.Theinter-relationshipsofthese threevariablesareshowninFig.1.1.
Whatdotheresultsshow?Thefrequencywith whichanoutcomeevolvesisstronglycorrelated withtherateofmutation(middlepanel).Thatis,the highertherateatwhicharesistantvariantemerges bymutation,themorelikelyitsappearanceinreplicatecultures,reachingahighfrequencyineach population.Thiseffectdoesnotoccurbecausemutationratesarecorrelatedwithmutantfitness(right), whichhaslittleeffect(left),giventhatallthemutationsarestronglybeneficial,withasimilarlylarge chanceoffixation.
Thisimmediatelypromptsquestionsaboutthe causes ofthe50-foldrangeinmutationratesevident inFig.1.1—allforsingle-nucleotidemutations.That is,whenfacedwithevidencethatthecourseof evolutionreflectswhatismutationallylikely,oneis naturallyconcernedwiththecausesofthosetendencies.Yet,inthecaseofMacLeanetal.(2010), noattemptwasmadetopredictorunderstandsuch effects.
InChapter9,wereviewadditionalevidence— bothfromexperimentalevolution,andfromnat-
uralevolutionindiversetaxa(includinganimals andplants)—thatbiasesinmutationinfluencethe courseofadaptation.Suchresultssuggestacause–effectrelationshipbetweentherateatwhichvariantsareintroducedbymutation,andtheirchances ofbeingmanifestedasevolutionarychanges.Aswe willsee,thiseffectcanbelarge,inthesensethat aseveral-foldeffectinevolutionisalargeeffect. Tendenciesofmutationmayimposepredictablebiases onevolution.Conversely,patternsinevolutionmay havemutationalexplanations,notmerelyinthe sensethatmutationisnecessaryforchange,but inthesensethatmutationisadifference-maker,a dispositionalfactorthatcausesonekindofchange tohappenmoreoftenthananother.
Evolutionarybiologyisthescienceofevolutionarycausesandeffects.Whatarethetermsthatevolutionarybiologistsusetodescribethiscausalrelationship?Whatisthetheoryforitsoperation?How widespreadareitseffects?Whatnotableevolutionarythinkershaveespouseditsimportance?The inter-relationshipsofthethreevariablesmeasured byMacLeanetal.(2010)—rateofmutationalorigination,fitnesseffect,andfrequencyevolved—seem veryfundamental.Whatisknownabouttheserelationships,generally?
Accordingtostandardaccountsintheevolutionaryliterature,Darwindiscoveredtheprincipleof selection,andby1930,Fisher,Haldane,andWright combinedthisprinciplewithgeneticstoyielda generalframeworkforunderstandingevolutionary causes.Thus,ifweweretoguessattheanswers tothequestionsinthepreviousparagraph,we wouldbeginbysupposingthatHaldane,Fisher,and
Figure1.1 Inter-relationsofmutationrate,fitnesseffect,andfrequencyofevolutionfor11rifampicin-resistantvariants(datafrom MacLeanetal.,2010).
Wrightdiscoveredtheimpactofbiasesinvariation anddevelopedthetheoryforit.
Infact,thisisnotthecaseatall.Aswewill discover,Fisher,Haldane,andWrighteachargued againsttheideathatthecourseofevolution mightreflectinternaltendenciesofvariation,on thegroundsthatmutationratesaretoosmallto haveanappreciableeffect.Classicalpopulationgeneticstheorylacksanaccountofthecausesand consequencesofbiasesintheintroductionprocess, asdotheclassicworksofMayr,Simpson,Stebbins, Dobzhansky,Huxley,andothers.Thekindof causationinwhichabiasinmutationalorigination imposesabiasonevolutionwasformallydescribed justtwodecadesago.
Howcouldsuchabasicideaescaperecognition forsolong?Aswillbecomeclear,theeffectsof theintroduction(origination)processfallwithina blindspotorgapinevolutionarythinking.This blindspotisnotaccidental:itrepresentstheshadow castbyadensenexusofargumentsconstructed deliberatelytopromoteanddefendthebeliefthat variationismerelyasourceofrandomrawmaterialsforselection—supportingahigh-levelviewof selectionasthepotterandvariationastheclay—and todiscouragealternatives,particularlytheideathat evolutionmightexhibitinternaltendenciesdueto thewaythatvariationemerges.
Amaintaskofthisbookistodocumentthisnexus ofargumentsthoroughly,andthentodeconstruct it.Becausethisnexuswasbuiltdeliberately,over decades,itcannotsimplybeexcisedordeflatedwith asingleprecisestroke:itmustbeattackedvigorouslyinallitsmanifestations,likeacancer,until
itspoweroverourimaginationhaswithered.Once thistaskiscompleted,wecanbegintorebuildour understandingoftheroleofvariationinevolution, drawingonsomeavailabletheoryanddata.
TheresultsofMacLeanetal.(2010),andthoseof themorewellknownstudyofparallelismbyRokyta etal.(2005),roughlyfitthebehaviorexpectedfrom origin-fixationmodelsdepictingevolutionasa simpletwo-stepprocess,witharatedetermined bymultiplyingtherateofmutationalintroduction ofnewallelesbytheirprobabilityoffixation.For instance,forthe11mutationsfromMacLeanetal. (2010),thetotalrateofevolutioncanberepresented asasumover11origin-fixationrates
denotingtheparentalgenotypeas i andthe alternativeswith k.Here, μN isarateoforigin (mutationratetimespopulationsize),and π isa probabilityoffixation.Thechancethataparticular mutation j willbeobservedinagivenreplicateisa fractionofthatsum:
Thisprobabilityislinearlydependentonboththe rateofmutationalintroduction,andtheprobability offixation.Thus,forsitesevolvingaccordingto origin-fixationdynamics,the50-foldrangeofrates ofmutationevidentinFig.1.1hasanexpected 50-foldeffectonthechanceofevolving.The11 mutationalchangeshavemeasuredfitnesseffects rangingfrom s = 0.3to s = 0.9,thustheprobability
offixation,usingKimura’sformula π = 1 e 2s 1 e 2Ns , rangeslessthan2-foldfrom0.45to0.83(assuming N = 106 ).Undertheseconditions,thedifferencemakingpowerofmutationisstrongandthatof selectionismodest(withmoredata,oneexpectsto seeacorrelationbetweenfrequencyevolvedand fitness).
Bycontrast,aninfluenceofmutationbiasonthe outcomeofevolutionaryadaptationisnotexpected underthetheorythatadaptationresultsfrom shiftingthefrequenciesofallelesalreadypresentin aninitialpopulation.Forinstance,ifall11variant rpoB allelesarepresentinitially,theoutcomeof evolutionwillbeessentiallydeterministic,withthe fittestalleleprevailingeverytime.Mutationalshifts amongpre-existingallelesmayoccur,butbecause mutationratesaresosmall,theirinfluencewill benegligiblecomparedtothemuchlargershifts causedbyselection.
Thenotionthatnaturalpopulationsmaintainan abundantstockofvariation,sothatselectionnever hastowaitforanewmutation—the“genepool” theory—wasakeyconceptualinnovationofthe ModernSynthesis,providingtherationaletoredefineevolutionasaselection-drivenshiftinthefrequenciesofalleles.
Yet,inthe1960s,assoonasthisshifting-genefrequenciesviewbecameanestablishedorthodoxy, molecularsequencecomparisonspromptedbiochemiststodepictevolutionquitedifferently,asa processoftheacceptanceorrejectionofindividual mutations.ThisledKingandJukes(1969)and KimuraandMaruyama(1969)toproposeoriginfixationmodelsrelatingtherateofevolution directlytotherateofmutationalintroduction(see McCandlishandStoltzfus,2014).Foryears,the significanceofthisdevelopmentwasmutedby thesensethat“molecularevolution”isaspecial case,anisolatedworldofneutralityirrelevantto adaptationandothermajorissues.Yet,eventually, theoreticiansinterestedinadaptationandother majorissuesbegantoremarkontheinadequacy oftreatingevolutionasaprocessofshiftingallele frequencies,arguinginsteadthatthedynamicsof long-termevolutiondependondiscreteeventsby whichmutationintroducesnewtypes(Hartland Taubes,1998;EshelandFeldman,2001;Yedidand Bell,2002).
Thatis,someevolutionarygeneticistshave rejectedtheideathatallofevolution,including long-termandlarge-scalechanges,followsfrom thekindofselection-drivenprocessofshifting allelefrequenciesthatoneseesinexperimental populationsofanimalsandplantsoversmall numbersofgenerations.Thisextrapolationist doctrineisusuallyexpressedbystatingthat macroevolutionfollowsfrommicroevolution(see Dobzhansky,1937,p.12).
Theimplicationsofthisfoundationalshiftare stillemerging,andtheyhavenotpenetratedvery farintoevolutionarythinking.Theorthodoxythat emergedinthemid-twentiethcenturydependsfor itsvalidityonrejectingtheviewthattheoutcomeof evolutionreflectsthetimingandcharacterofindividualmutationalevents,andinstead,embracing theviewthatevolutionisahigher-levelmass-action processofshiftingallelefrequenciesinapopulation. Overtheyears,thispositionhasbeenquietly abandoned,andisnolongerafoundationofevolutionarygenetics. Asaresult,thereisacuriousdisconnectbetweenwhatformalmodelsimplyaboutthepossible dispositionalroleofvariationinevolution,andthe guidanceprovidedbyfamiliarverbaltheoriesthatdepict mutationasmerelytheultimatesourceofrawmaterials forselection,orasaweakforceincapableofinfluencing theoutcomeofevolution. Ofparticularinteresthereis that,whenthedynamicsofevolutiondependonthe dynamicsoftheintroductionprocess,evolutionary changeis biased bymutation,inthesensethat,if therateofmutationfor A → B isgreaterthanfor A → C,thiselevatesthechancesofevolvingfrom A to B relativeto A to C.Thatis,abiasinmutational introductionisapriorbiasonevolution.
Importantly,nonuniformitiesinindividualmutationrates,asinMacLeanetal.(2010), arenotthe onlysourcesofbiasesintheintroductionofvariation Forinstance,inthestudyoflaboratoryadaptation ofbacteriophagesbyRokytaetal.(2005)(seeCh.9), a Met → Ile changeisobservedrepeatedly,dueto mutationsatposition3850.Thegeneticcodedictatesthatthisparticularaminoacidchangecantake placebythreedifferentsingle-nucleotidemutations: ATG → ATT , ATG → ATC,and ATG → ATA.By contrast,otheraminoacidpairstypicallyhaveonly oneortwopossiblemutationalpathsinterconnectingthem(e.g., Met → Val, ATG → GTG).Becauseof
this,the Met → Ile changehasakineticadvantage. Evenifallratesofmutationfromonecodonto anotherarethesame,thisaminoacidchangeis probabilizedbyhavingmultiplemutationalroutes oforigination.Infact,twoofthethreedifferent mutationalpathwaysatposition3580areseenby Rokytaetal.(2005).
Thisisnotabiasinthemechanismofmutation itself,butabiasinhowmutationsareexpressed, formallyidenticaltoakindofdevelopmentalbias suggestedrepeatedlyintheevo-devoliterature (e.g.,Emlen,2000),bywhichsomephenotypes aremorelikelytoarisebymutation-and-altereddevelopmentbecausetheyhavemoremutationally accessiblegenotypes.Molecularandmorphological versionsofthisproposalareunitedbytheconceptof agenotype-phenotype(GP)map.Thegeneticcode istheGPmapthatrelatestripletgenotypestoamino acidphenotypes.DisruptingtheATGgenotype underlyingthemethioninephenotypeismorelikely toresultintheisoleucinephenotype,accessible viathreemutations,thanvaline,accessiblebyone mutation.
Likewise,ifwegeneralizefurtheronthiseffect, wecanunderstandthebasisofargumentsintheliteratureofself-organizationandevolvabilityabout thedifferentialaccessibilityofalternativephenotypesingenotype-space(Fontana,2002).Concepts likegeneticproximityormutationalaccessibilityare relevanttoevolutionpreciselybecausetheevolutionaryexplorationofgeneticspacesissubjectto kineticbiasesduetomutation.
Thatis,whereasthepatternsofmutation-biased adaptationinMacLeanetal.(2010)arespecifically (1)aneffectofasymmetriesinmutationrates,and donotrepresent (2)asymmetriesindevelopmental responsestoperturbation, nor (3)asymmetriesin thedensityandaccessibilityofgenotypenetworks, allthreekindsofasymmetriesarepotentialsourcesof biasesintheintroductionofvariation.If,startingfrom state A,therateofintroductionof B isthreetimes higherthan C,theexpectedevolutionaryconsequencesofthis3-foldbiasarethesamewhether (1) B and C areuniquegeneticalternatives,and thedifferentialeffectreflectsmutationbiasperse, (2) B and C arealternativephenotypicstates,and (aggregatingeffectsovertheentiremutationspectrum) B isthreetimesmoreaccessibleviamutation-
and-altered-development,or(3) B and C arephenotypicallydefinednetworksingenotype-space,and statesin B are3timesmoreaccessiblefromthe parentalnetwork A
Notethat,whenweinvokedthiskindof causationtointerprettheresultsofMacLeanetal. (2010),wedidsowithinadistinctiveexplanatory paradigm.Confrontedwithaspectrumofchanges, eachwithanobservedfrequency(Fig.1.1),our focuswasonexplainingwhyevolutionhasthe tendenciesthatithas.Molecularevolutionistsare familiarwiththisparadigm—asoldasDayhoff’s empiricalmatricesofaminoacidreplacement—in whichevolutionisaprocesswithpropensities,and ourjobasevolutionistsistopredictandexplain thesepropensities.
Thetraditionalapproach,bycontrast,istoestablishaplausiblenarrativefor auniqueevolutionary outcomethatisconsideredasaconclusionorend-point, i.e.,whatSober(1984)calls“equilibriumexplanation.”AsReeveandSherman(1993)putit,thefocus oftheadaptationistresearchprogramisonexplaining“phenotypeexistence,”e.g.,whythezebrahas itsstripes.Theparadigmofanexplanationisto showthattheobservedphenotypeisthemostfit possible,giventhevariousconstraintsatwork.This paradigmhashadaprofoundinfluenceonconceptionsofexplanationandcausation.
ToapplythetraditionalparadigmtoMacLean etal.(2010),onewouldbeginwiththeobservationofasingleinstance,atokenoutcome,asif wehaveonlyasinglerifampicin-resistantstrain, e.g.,theonewitharginineatsite518.Theexistence ofthisaminoacidwouldbecreditedtotherole ofselectionshiftingtheArg518alleledeterministicallyfromalowfrequencytoahighfrequency,e.g., wecouldusethemeasuredselectioncoefficientto modelthedynamicsofallelereplacement,andwe couldtestthisexperimentallybyexamininghow selectionincreasesthefrequencyoftheArg518allele fromlowtohigh.
Bycontrasttothisdeterministicappealtoselection,theinvolvementoftheArg518allelewould bedescribed(inthetraditionalparadigm)byreferencetoamutationthatoccurred bychance atsome timeinthepast.Evolutionwas“contingent”on thepresenceofthisalleleinthegenepoolofthe startingpopulation,sothatselectioncouldraise
itsfrequencydeterministicallyfromlowtohigh.If someotherallele,e.g.,leucine531,providesgreater rifampicinresistance,thenouraccountofArg518 wouldstress“constraints”or“limits”onthe“power ofselection”toachieveperfection(BartonandPartridge,2000).Ourexplanationwouldnotethatthe Arg518allelewaspresentinthegenepool,andthe Leu531allele(forinstance)wasabsent,bychance.
Whenwelookatthetraditionalparadigminthis way,wecanseethatitreliesontheidealizationthat selectiondeterministicallyensuresthatthefittest survives.Asthisidealisticparadigmhasbroken down,someevolutionarythinkershaveturnedto denotingexceptionsbyappealingtochance,contingency,andconstraints(e.g.,Futuyma,2010).
Theseconceptsmakeitpossibletorescuethetraditionalparadigm,butonlysuperficially.Chance, constraints,andcontingencydonotdenotecausal theoriestoaccountforevolutionarydispositions orpreferences.Chanceisnotacauseofanything. Contingencyisnotacausalforce.Constraintsdo notcauseoutcomes,butexplainnonoutcomes,and inthecaseofmutationbiases,theyfaileventobe explanatory(thereisnoconstraintpreventingmutationsthatoccuratlowerrates—theysimplyoccur atlowerrates).Thatis,theseconceptsareexplanatoryratherthancausal:theyrefernottopistonsor levers,buttoexcuses.Theyarenotalternativetheories,butverbalmarkersindicatingdeparturesfrom idealsofdeterminacyandequilibration.Theiruse intheevolutionaryliteraturerepresentstheunstabletransition-statebetweenafailedparadigmand somealternative.
Thatis,theseconceptsfailtosatisfythescientific imperativetoexplorecausaltheoriesinwhich quantitiessuchasmutationratesareusedtomake precisepredictions.However,asdemonstrated earlier,adifferentandmoresuitableexplanatory paradigmexiststhatallowsustotreatvariational causesascauses,ratherthanaslimitstothe imaginaryomnipotenceofselection.
1.2Whatthisbookisabout
Therecognitionthatbiasesintheintroductionof variationareacauseofevolutionarybiashasfarreachingimplicationsforevolutionaryresearchand forevolutionarytheory.Biasesintheintroduction
processhavebiologicalcausesandevolutionary consequences.Theircausesresideinproperties ofthemutation-generatingsystem,propertiesof development,andbroadfeaturesofthearchitecture ofgeneticspaces.The consequences ofbiasesin theintroductionprocessareintheprovinceof populationgenetics.Expandingourunderstanding ofpopulation-geneticcausationtoincludethe consequencesofbiasesintheintroductionof variationprovidesacohesiveandpreviously unrecognizedbasisforaddressingkeyconcernsof scientistsinterestedinneutralevolution,evo-devo, evolvability,andself-organization.
Re-thinkingtheevolutionaryroleofvariationalongtheselinesisamajorchallengefor twenty-first-centuryevolutionarybiology.Abroad approachtothischallengewouldrequireabook muchlargerthanthisone,anditwouldnecessarily leavemanyissuesunresolved,dependenton experimentsthathavenotyetbeendesignedor carriedout.Torespondadequatelytothischallenge willrequiretheworkofmanyscientistsovermany years.
Thisbookfocusesmorenarrowlyonthe developmentandapplicationofbasicprinciples, relyingonmolecularexamplestoestablishkey arguments.Ratherthanfullyansweringthegrand challengeofrethinkingtheroleofvariation,itaims topreparethegroundby(1)illustratingsomeofthe complexityofmutationalprocessesandusingthat informationtoexplainwhybiasesareinevitable, andwhyvariousprecisesensesofrandomness failtoapply,(2)showingthatthemutation-israndomclaimcommoninevolutionarydiscourseis bestunderstoodintermsofahistoricallypopular viewinwhichtheinternaldetailsofmutationand developmentareirrelevanttohowevolutionturns out,and(3)articulatinganalternativeviewbased onexploringpropensitiesofvariationascausesof evolutionarypropensities,andsummarizingthe currenttheoreticalandempiricalsupportforthis connection.
Mutationisoftenmischaracterizedandmisunderstood.Forinstance,theadjective“accidental”or “spontaneous”isoftenapplied,butmutationisnot anaccidentlikeabranchfallingontheroofofyour homeduringastormandleavingahole.Damage isnotmutation.Whencellsleaveunrepairedholes
intheirDNA,thetypicaloutcomeiscellulardeath, notmutation.Instead,mutationislikeabranch fallingonahouseandleavingaholeintheroof, followedbyarepair-botdetectingthedamageand thenshinglingovertheholeusingthewrongcolor ofshingles.Themutationisnotthehole(whichhas beenrepaired),buttheslightdepressionintheroof andthepatchofdifferentlycoloredshingles.
Aswillbecomeapparent,mutationsemergeby complexpathways,oftenstimulatedbydamage,yet carriedoutbyenzymes.Nounifiedtheoryofmutationexists.Tolearnaboutmutationasabiological processistolearnaboutavastassortmentofdifferentprocessesthatcontributetotheemergenceof mutations.However,becauseteachingaboutmutationisnotthemainpurposeofthisbook,nochaptersaredevotedspecificallytomutation.Instead, thisbookapproachesthetopicindirectly,describingmutationalprocessesinordertoconsiderthe implicationsofrandomness,andtoillustratekey concepts(e.g.,mutationspectrum)importantfor conceptualizingmutation.
Thedoctrinethatmutationisrandomisnotwhat itseems.Ordinaryscientificclaimstypicallyare justifiedbyappealto(1)firstprinciples,whichin thiscasemightprohibitnonrandomness,or(2)systematicobservationsorexperiments,whichinthis casemighthaveshownagainandagain,under avarietyofconditions,thatmutationisrandom. Thefirstoptionisruledoutbecausetherandomnessdoctrinelongprecededanydetailedknowledgeofhowmutationworks—indeed,theclaim ismadebyevolutionarybiologists,notbymutationresearchers.Thesecondoptionseemsunlikely becausesourcesthatpromotetherandomnessdoctrinedonotcitealargebodyofsystematicevidence, thatis,onedoes not seestatementslike“mutation hasbeenshowntobecompletelyrandominavarietyofspecies(Smith,1939;Johnson,1944;Jones, 1951).”
“Randomness”hasmanypossibleconnotations, andtheliteratureofevolutiondrawsonthem indiverseways.Thesituationhasbecomeso confusingthat,inthepast25years,multiple authors—includingprofessionalphilosophers— havewadedintotheconfusion,aimingtosortout whatevolutionarybiologistsreallymeantosay.The typicalapproachistoassumethataspecial“evolutionary”meaningofrandomnesscanbefound
thatjustifiesthedoctrineretroactively.Wewill devoteconsiderablespacetoexploringtheseideas, includingthepossibleconditionalindependenceof mutationsandfitnesseffects—conditionalupona commonenvironment.Yet,conditionalindependenceandotherattemptstomakesenseofthe randomnessdoctrineeitherfailtomatchthefacts ofmutation,orfailtomatchwhatevolutionary biologistshavebeensaying.
Othercasesexistinwhichscientistshaveused aconceptformanyyearswithoutaclearconsensusonitsmeaning.Indeed,thehistoryofscience repeatedlyshowsthatconfusionorconflictsover themeaningofkeyconceptsisnormalandmaypersistforgenerations.Scientiststookcenturiestosettle onacommonunderstandingof“heat.”Theconcept of“probability”hasprovenenormouslyusefulin spiteoftwocenturiesofdebateoverwhatitmeans, whetheritreferstosomethingreal,andwhether itcanbederivedfromfirstprinciples.Manypracticallyusefulconcepts,e.g.,entropy,aresubjectto rarefieddisputes.
Yetprobability,heat,andentropyare undoubtedly usefulconcepts:theirusesupportsprecisecalculationsthathelpustounderstandtheworldbetter.Whatisthepurposeoftherandomnessdoctrine?Whatpreciseandusefulcalculationsdoes itsupport?Whatlogicalconclusionsdependon it?Whatadvantagewouldbelostbyrejectingit? Oneoftenfindsevolutionarythinkersurgingcaution,warningusthat“random”isaproblematic word,yetifthesubjectrequirescaution, whytake apositionatall?Thesesameauthorsdonotfeel obligedtostakeoutacarefullywordedposition ontherandomnessofanythingelse.Whydidit becomesoimportanttoassociatemutationwith randomness?
AcloserlookattheclassicSynthesisliterature revealsthatrandomnessattributionsco-occurwith otherargumentsthatminimizetheimportanceof variationinevolution,andthatthesearguments all refertothesamecontrast-case:selection.Mutationis random,selectionisnot;mutationisweak,selection isnot;selectionactsatthe(right)populationlevel, mutationactsatthe(wrong)individuallevel;selectionprovidesdirection,variationmerelyprovides rawmaterials.Inthiscontext,therandomnessdoctrineproclaimsadeepermutation-is-unimportant doctrineinwhichvariationismadesubservientto
selection,supportingtheassignmentofrolesthat onefindsinthetopicarticleon“naturalselection” (Ridley,2002)inthe OxfordEncyclopediaofEvolutionaryBiology:
Inevolutionbynaturalselection,theprocessesgeneratingvariationarenotthesameastheprocessthat directsevolutionarychange...Whatmattersisthatthe mutationsareundirectedrelativetothedirectionof evolution.Evolutionisdirectedbytheselectiveprocess (p.799).
Inthisway,therandomnessdoctrineprovidesa metaphysicalguaranteeoftheclassicdichotomyof selectionandvariationasthepotterandtheclay, thatis,itdifferentiatesselection,thesourceoforder, shape,anddirection,frommutation—notthesource ofthosethings,becauseit’s“random.”Patternsand interestingfeaturesandotherorderlyoutcomes inevolutionmaybesafelyattributedtoselection, becausemutationisrandom.Noordinarydefinition ofrandomnesscleanlydistinguishesonebiological processfromanother,thusaspecial“evolutionary” definitionisdevelopedtofillthisgap. Thenatureof therequired“evolutionary”definitionisnowobvious: itmustapplytomutationbutnottoselection,e.g., therandomnessclaimisinterpretedtomeanthat mutation,unlikeselection,doesnotinvariably increasefitness(Section4.2;seeEble,1999).
Thefinalaimofthisbookistoarticulatean alternativeview,andtomakeacaseforitsimportance.Evolutionischange.Thegoalofevolutionary research,inmyopinion,istounderstandchange, andparticularlytounderstandwhysometypes ofchangeshappenmoreoftenthanothers.The earlygeneticists,whofirstcontemplatedthe roleofmutationinMendelianpopulations,saw mutationasadifference-maker,asapotentially importantsourceofinitiative,creativity,direction, anddiscontinuityinevolution(Stoltzfusand Cable,2014).Thispositionwasrejectedbythe architectsoftheModernSynthesis,whoargued thatmutationismerelyarandomsourceofraw materials,andfurtherclaimedthattheviewheldby earlygeneticists—inwhichthecourseofevolution mayreflectthetimingandcharacterofindividual mutations—wasincompatiblewiththegeneticsof naturalpopulations,whichalwayshaveabundant standingvariationtofuelevolution.
Aswewilldiscover,theshifting-gene-frequencies theoryunderlyingthesweepingclaimsmadeby thearchitectsoftheModernSynthesis(seeBox6.2) doesnotcorrespondtotheopen-endedframework thatsomescientistsimaginetoday,whentheyrefer toamid-century“Synthesis.”TojustifyaneoDarwiniandivisionoflaborbetweenvariation (sourceofrawmaterials)andselection(source ofcreativity,direction,etc),theshifting-genefrequenciestheoryofevolutionpositsa“buffet” viewofpopulationgenetics,designedprecisely tocircumscribetheroleofmutation,inwhich everythingselectionneedstoadapttocurrent circumstancesisalreadypresentinthegenepool, incarefullymaintainedabundance.Contemporary thinkingstillreliesontermsandconceptsfrom thisnarrowtheory,includinga“forces”theorythat seemstoruleoutadispositionalroleofmutation (onthegroundsofbeinga“weakforce”easily overcomebyselection),eventhoughsucharole isperfectlycompatiblewithabroaderconception ofevolutionarygenetics.
ThesuccessfulpromotionofthistheorybyErnst Mayrandhiscohortofinfluencersleftablindspot inthedevelopmentofevolutionarythinkingabout mutation.Whereonemightexpecttofindadescriptionofthepropensitiesofmutation(variation),and thecausalprincipleslinkingthemtopropensitiesof evolution—whatwewillcallthe sourcelaws and consequencelaws ofvariation—oneinsteadfinds adenial thatanysuchprinciplesexist.Theoreticalandempiricalresultshavebeenencroachingonthisblindspot fordecades,withlittleovertrecognition.
Inthisbook,weconsidersomesimpletheoretical resultsthatarefundamentaltounderstandingthe roleofmutationanddevelopmentinevolution,but whicharenotexplainedintextbooks.Wediscover thattheintroductionofvariantformsbymutationand-altered-developmentisapredictablecauseof nonrandomnessinevolutionarychange.Wefind thatthiskindofcausecanbemodeledusingprinciplesofpopulationgenetics,anditseffectscanbe studiedusingconventionalscientificmethods. Patternsofchangecanhavemutationalcauses,notsimply inthesensethatmutationismateriallynecessaryfor evolution,butinthesensethattendenciesofmutation canbedifference-makers,causingonekindofthingto happenmoreoftenthananother.
Box1.1TheoryA andtheoryC
Theconceptofatheoryplaysanimportantroleinevolutionarydiscourse.Theterm“theory”hasbeenusedin thesenseofagrandconjectureforhundredsofyears, andcontinuestobeusedthiswayinevolutionarybiology, e.g.,Kimura’s“NeutralTheory”istheconjecturethatmost changesatthemolecularlevelresultfromrandomfixation ofselectivelyneutralalleles;the“ExonTheoryofGenes” (Gilbert,1987)proposesthatgenesevolvedfromcombining exon-minigenes;theendosymbiotictheoryholdsthatmitochondriaandchloroplastsarosefromprimordialbacterial endosymbionts.
However,theword“theory”doesnotalwaysmeanthis. Populationgeneticstheory,forinstance,isnottheconjecture thatpopulationshavegenetics,norismusictheorythe conjecturethatmusicexists.Instead,thesearebodiesof abstractprinciples.Indeed,scientificwritingsusethesame termforboth(1)theoryC (concrete,conjectural),agrand conjectureormajorhypothesistoaccountforsomesetof observedphenomena,asinthe“continentaldrifttheory”or “Lamarck’stheoryofevolution,”and(2)theoryA (abstract, analytical),abodyofabstractprinciplesrelevanttosomediscipline,methodology,orproblemarea,asin“musictheory,” “quantumfieldtheory,”or“populationgeneticstheory.”
Usuallyonemustrelyoncontexttodeterminewhich meaningapplies.Forinstance,awhitepaperon“TheRole ofTheoryinAdvancing21st CenturyBiology”(National AcademyofSciences,2007)emphasizesthedevelopmentof formalismsratherthanconjectures,andsaysthat“auseful waytodefinetheory[notetheabstractnoun]inbiologyis asacollectionofmodels,”clearlyareferencetotheoryA (the reportalsoreferstoafewtheoriesC ).
Thetwotypesoftheoryareevaluatedindifferentways. ThestandardoftruthforatheoryC isverisimilitude—how welldoesitmatchtheactualworld?AtheoryC takes risks,andcanberefutedbyfacts.BecauseatheoryC isa conjecture—notnecessarilyacorrectone—itstilliscalled a“theory,”evenbythosewhodoubtitsverisimilitude.By contrast,toevaluateastatementintheoryA ,onedoesnot consideranyfactsabouttheworld,butonlywhetherthe statementiscorrectlyderivedfromitsassumptions.Oncea pieceoftheoryA isvalid(correctlyderived),itremainsvalid forever,evenifitreliesonimaginarythingssuchasinfinite populations.
Toderiveexpectationsforsomepossibilities,priortohavinganyreasontopreferoneoveranother,onerequiressome
theoryA .Fisher(1930b)wrotethat“Nopracticalbiologist interestedinsexualreproductionwouldbeledtowork outthedetailedconsequencesexperiencedbyorganisms havingthreeormoresexes;yetwhatelseshouldhedoif hewishestounderstandwhythesexesare,infact,always two?”(infact,thesexesarenotalwaystwo,butthisisnot relevanthere).Thecollectionofallthemodelsfordifferent numbersofmatingtypeswouldbepartofthetheoryA of sexes.AconcretetheoryC ofsexeswouldbesomethingquite different,e.g.,itmightproposeacausalexplanationforthe actualhistoricphenomenonoftheoriginandmaintenance ofsexualreproductioninanimals.
AlthoughthedistinctionbetweentheoryA andtheoryC isnotalwaysclear,applyingthedistinctionremainsausefulexercise.Forinstance,thetheoryofkinselection(“kin selectiontheory”or“inclusivefitnesstheory”)isfrequently describedasasetof“tools”(Michod,1982),implying theoryA .Yet,thecontextfortheuseofthesetoolsisthe broadconjectureofHamiltonthatkinselectioniscrucial toaccountfortheevolutionofsocialbehaviorinanimals. MostofthedisputesaboutkinselectiontheoryaretheoryA disputesamongmathematiciansandphilosophersabout suchthingsaswhethertheassumptionsunderlyingcertain formulationsofHamilton’srulearecorrectlydescribed,or whetherkinselectiontheoryA isequivalentwithgroupselectiontheoryA (BirchandOkasha,2014).
Obviously,thereisaconnectionbetweenthetwotypes oftheories,inthatabstractprinciplesoftheoryA ,rendered concretewithobservedorconjecturedvalues,canprovide thebasisofatheoryC .InthecaseofKimura’sNeutral Theory(Kimura,1983),thetheoryC —theconjecturethat mostchangesatthemolecularlevelresultfromtherandom fixationofselectivelyneutralalleles—andtheoryA were developedsomewhatseparately.Thedefinitionofeffectivelyneutralalleles(perpetuallymisunderstoodbycritics) andtheprobabilityoffixationunderpuredrifthadbeen knownfordecades(seeWright,1931;Fisher,1930b,Ch.IV; Haldane,1932,Appendix).KimuracombinedmostlypreexistingtheoryA (includingplausibilityargumentsbasedon the“cost”ofselection)withtheconcreteassertionthat thevaluesofcertainquantities(relatingtopopulationsizes andmutanteffects)weresuchthat,forDNAandprotein sequences,neutralevolutionbymutationandrandomfixationwouldbefarmorecommonthananyonehadimagined previously.
OpponentsoftheNeutralTheory,whodenythetruth ofthetheoryC ,arenonethelessquitehappytomakeuse ofitstheoreticalA infrastructureineffortstorejectneutral models,asinthereviewbyKreitman(1996).Thatis,valid theoryA isrequiredtocarryout modustollens reasoning, inwhich X isrejectedbasedonitsimplication X ⇒ Y andtheobservationthat Y isabsent.Inthecaseofneutral models, X = neutrality ,and Y issomeexpectationabout ratesorpatterns.Torejectneutralitybasedonevidence requiresacorrectlyderivedmodelofanimaginaryabstract
1.3Whothisbookisfor
Intheprocessofuntanglingtheconceptualmess attheintersectionofmutation,randomness,and evolution,thisbookaddressesvarioustopics thataretimelyandofcentralimportanceto evolutionarybiology.Nevertheless,thewaythat thetopicsareframedmaybeunfamiliar,evento professionalscientists.Iwouldnotexpectreaders tohaveaclearsenseofwhetherornottheymight beinterestedinreadingthisbook,basedsolelyon thetitle,orbasedonaphrasesuchas“biasesin theintroductionofvariation,”orevenbasedonthe earlierreferencetoaconceptualmess.Therefore,it mayhelptoexplainhowthemainargumentsrelate toseveralissuesthataremorefamiliar.
First,inthepasttwodecades,interesthas emergedintakingamoredetailedlookattheroleof variationinevolution,includinganinterestinthe “arrivalofthefittest”ratherthanthe“survivalofthe fittest,”apparentindiscussionsofdevelopmental constraints,evolvability,robustness,facilitated variation,andsoon.Mostofthepreviousinterest inthisissuehasbeenfocusedondevelopment ofvisiblephenotypesofmacroscopicorganisms, whereasmybackground,training,andinclinations aremuchmoremolecularthanorganismal.This bookpresentsanontraditionalviewoftheroleof variationthatisfirmlygroundedintheoryandin empiricaldata,inwhichtendenciesofevolutionare relatedtotendenciesofvariation.Itbuildsafoundationofbasicconcepts,whichitthenusestoaddress, andsometimesresolve,long-standingproblems.
worldinwhichneutralityistrue,i.e.,thetheoryA statement X ⇒ Y mustbeabstractlytrue.TheparadoxinKreitman’s title“Theneutraltheoryisdead.Longlivetheneutraltheory” isperfectlyresolvedbythefactthatitrefersfirsttotheoryC , andthentotheoryA
Intheremainderofthisbook,thedistinctionbetween theoryA andtheoryC isnotmadeexplicitexceptinafew cases,sothatthereadermayapplythedistinctionandassess itsutility,withoutsufferingtheannoyanceofbeingforced todoso.
Forinstance,fordecades,evenasevo-devohas gainedinpopularity,othershaveappealedtopopulationgenetics—widelyregardedasthelanguage ofcausationinevolution—toarguethatevo-devo hasnotcontributedanynewprinciplesorcausesto evolutionarythinking.Intheabsenceofageneralcauseclaim,attemptstojustifyevo-devoturned tofuzzyclaimsaboutalternative“narratives”and “explanatoryparadigms.”Inthisbookwediscover thatclassicargumentsaboutcausationusedagainst evo-devo,suchasMayr’s“proximatecause”objection,Dobzhansky’s“wronglevel”argument,orthe weak-pressureobjection(e.g.,ReeveandSherman, 1993),areinadequate,andthisisnotamatterof paradigmsorreductionism,butamatterofmaking thewrongassumptionsaboutpopulationgenetics, andfailingtorecognizetheintroductionprocessas anevolutionarycausewithdistinctiveimplications. Theabilityofgenerativeprocessestoimposeabias ontheoutcomeofevolutionarychangeisthefirstordercauseimplicitlyatworkinvariousevo-devo ideasabouthigher-ordereffects.
Second,thisbookistailor-madeforthosein molecularevolution,microbiology,andcomparativegenomicswhofeelthatevolutionarybiology hasbeenstubbornlyresistanttothelessonsofthe molecularera,whichhasonlypartiallyshiftedour viewsofevolutionwhenitshouldhavetransformed them.Inthisregard,severalhighlyoriginalbooks haveappearedrecently,includingNei(2013) (mutation,notselection,drivesevolution),Lynch (2007b)(nonadaptivemechanismsmakegenomes complex),Shapiro(2011)(engineering,notaccident,
providesinnovation),andKoonin(2011)(after Darwinism,thingsgetcomplicated).
Thepresentbooksharessomethingwitheachof these,thoughtheoverlapincontentissmall.Like Lynch,Iamconcernedtojustifyanovelposition onthecausesofobservedevolutionarypatterns intermsoftheoreticalimplicationsofpopulation genetics.LikeShapiro,Koonin,andNei,Ibelieve thatevolutionarythinkingisdeeplyshapedby vestigesofaneo-Darwinianview—selectionand variationasthepotterandtheclay—thatisbroadly incompatiblewiththefactsofmolecularevolution. Morethananyoftheseauthors,Ibelievethat progressdependsonconceptualandcultural reform,toincludere-evaluatingcoreconcepts(“raw materials,”“forces”),developingnewconceptsand metaphorstoguideourthinking,changingwhat istaughttostudents,andreformingthedistorted historiographyofourfield.
Third,everyscientist,philosopher,orlay-person interestedinevolutionhassurelyencountered theideathat,beginninginthe1980s,ahighleveldisputehasbeensimmering,regardingthe adequacyofamid-twentieth-centuryorthodoxy formerlyknownasthe“ModernSynthesis.”In recentyears,thisdisputehastakentheformof acallforan“ExtendedEvolutionarySynthesis.” Thisbookexplainsandthoroughlydocuments thecommitmentofthearchitectsoftheModern Synthesistoamistakenconceptionoftherole ofvariationthatremainsdeeplyembeddedin evolutionarythinking,e.g.,inthe“rawmaterials” doctrine,the“mutationisrandom”doctrine,and the“forces”theory.
Thesehistoriccommitmentshavebeenobscured byasubsequentprocessofnormality-driftthat hasperpetuallyredefinedthe“Synthesis”(and,to someextent,“neo-Darwinism”),creatingafalse impressionofpermanence.Wewillseeexactly howthemeaningsofkeytermshavechanged, e.g.,“rawmaterials”usedtobeananalogyto rawmaterials,andnowismerelyasynonymfor “variation,”usedevenformutationssuchasgene duplications thatareinnowayanalogoustoraw materials.Originally,theterm“adaptation”didnot implicatetheluckymutantview(theadventitious fixationofabeneficialmutant),whichwascalled “pre-adaptation”andassociatedwithpre-Synthesis
geneticists.Todaytheoreticianswhomakemodels ofthisprocesscallit“Darwinianadaptation”or “Darwinianevolution,”whileDarwinandFisher rolloverintheirgraves.Inmolecularevolution andevolutionarygenetics,scientistsfrequently assumeaformofmutation-limiteddynamics thatdirectlycontradictsthetheoryofshifting genefrequenciesunderlyingtheoriginalModern Synthesis.
Thishasanimportantimplicationforhighleveldebatesonevolutionarytheory.Mostof thesedebatesseemtoassumethatevolutionary biologymusthaveaGrandUnifyingTheory ofEvolutioncoveringallofbiology.Ashapeshifting“Synthesis”isthenputforward,typically defined,notasgenuinescientifictheory,butasa traditionconsistingofpeopleandtheirchanging distributionofbeliefs:thisputsanimpossible burdenonwould-bereformers,whomustcomeup withanalternativethatis(1)equallycomprehensive,andyet(2)obviouslydistinctfromtheshapeshifting“Synthesis.”Thisdebatehasmuddiedthe watersandpreventedreformfordecades.Inreality, theactualhistoricModernSynthesisceasedtobea validuniversaltheorysometimeinthe1970s:itwas notusefulforunderstandinglong-termsequence divergence,nordiditstheoryofrecombinationfueledchangeinthediploidsexual“genepool” applyusefullytotheasexualprokaryoticorganisms thathavedominatedtheplanetthroughmostofits history.EvolutionarybiologyhasnothadaGrand UnifiedTheoryfornearly50years.Apparently,one isnotneeded.
Finally,manyscientificreadersmaybedrawn tothisbook,duespecificallytotheirinterestin the“directedmutation”or“adaptivemutation” controversythateruptedin1988basedonexperimentsbyCairns,Foster,Hall,andShapiro,and theirskepticismofthe“mutationisrandom”claim. Inbrief,Cairnsandotherssuggestedthatcells haveevolvedwaysof(probabilistically)generating situation-appropriatemutations,anargumentthat setstraditionalthinkingagainstitself,pittinga hopefulbeliefintheadaptationofmutational mechanismsagainstthesimplifyingexplanatory paradigminwhichselectionreceivesallthecredit forthehardworkofevolution,onthegroundsthat mutationmerelysuppliesrandomrawmaterials.
Thesustainedhostilereactiontothisidea indicates thewillingnessofevolutioniststosacrifice abeliefinthepervasivenessofadaptationsoasto preservetheneo-Darwinianexplanatorydichotomy inwhichvariationplaysastrictlypassivematerial role.
Readerswho(likemyself)followedthedirected mutationscontroversywillbeinterestedinseveral relevantpoints.First,CRISPR-Casisunarguably anevolvedsystemthatgeneratessituationappropriatemutations,andwhichcontradicts conventionalwisdom(e.g.,Luria–Delbrückrandomness).Second,manysystemsforgenerating situation-appropriatemutationshavebeenreported inmicrobialpathogens,thoughtheyrarelyreceive attentionfromevolutionarythinkers,e.g.,diversitygeneratingretro-elements,elaboratecassetteswitchingschemes,phasevariationsystems,and multiple-inversion“shufflons.”Suchsystemsoften playacriticalroleinimmuneevasionorhostphagearmsraces.Third,thesesystems donot generallyexhibitthekindofnonrandomnessin whichthechancesofmutationsareinfluencedby theirincipientfitnesseffects,aswasthecasefor somemodelsproposedfor“directedmutations” (e.g.,Cairns’sgeneralizedreversetranscription model).Understandingtheoriginandmaintenance ofthesespecializedmutation-generatingsystems isamajorchallengeforscientistsinterestedinthe evolutionofevolvability.
Thus,therandommutationdoctrinebreaks downintwodistinctways,basedontwodistinct meaningsofthedoctrine:independencefrom fitness,andexplanatoryirrelevance.Microbial pathogenshavespecializedsystemsofmutation thatenhancethechancesofmutationsuseful forimmuneevasion,contrarytothedoctrine thatmutationsoccurstatisticallyindependently offitness.However,thesecontradictionswill probablybeviewedasexceptions,andthey willnotwinanyargumentsagainstthebroader doctrineofchancevariation,whichreallymeans thatvariationcannotbeadispositionalfactor inevolution.Thisversionoftherandomness doctrinealsobreaksdown,muchmorebroadly, notjustinmicrobes.Aswewillsee,thisbreakdowncanbedocumentedbyreferencetoordinary mutationbiasessuchastransition-transversion
bias,withouteverconsideringspecializedmutation systems.
1.4Howtheargumentunfolds
Thefocusofthisbookshiftssignificantlythrough itsthreemainparts.Thefirstpart,comprising Chapter2,Chapter3,andAppendicesAand B,addresseshowwellthebiologicalprocessof mutationisdescribedbysomeoftheordinary meaningsof“chance”or“randomness”inscience: lackofpurposeorforesight,uniformity(homogeneity),stochasticity,indeterminacy,unpredictability, spontaneity,andindependence(chance).
Throughoutthispart,Irefertofourpathways ofmutationdescribedinAppendixA,which explainstheprocessbywhicha T → C mutation (1)arisesfromanerrorduringgenomereplication, (2)arisesfromerror-pronerepairofdamage,(3) emergessymbolicallyinacomputer-generated MonteCarlosimulation,and(4)isengineered inageneusinghumantechnology.Nearlyall readerswillbenefitfromreadingAppendixA: theexamplesareprovidedspecificallytobuilda foundationforexplainingandevaluatingconcepts ofrandomness.Computer-simulatedmutations,for instance,arefullydeterministicandpredictable (thusnot“random”inthosesenses),butthey canbeuniformandtheiruseinasimulation canrepresent“chance."Human-engineeredmutationstypicallyarenotfullydeterministic,but canbehighlypredictable,nonuniform,and nonindependent.
SomebasicconceptsofrandomnessarereconsideredinamorepracticalcontextinChapter3. Thischapterintroducestheideathatsomeways ofthinkingaboutmutationareuseful,evenifthey areonlyapproximatelycorrect.Approximations comeatacost,andthusthepracticaluseofan approximation,e.g.,theassumptionthatmutation isuniformwhenitreallyisnot,isamatterofweighingcostsandbenefits.Chapter3alsointroducesthe ideathattheapplicationofprobabilisticreasoning toproblemsofmutationmaybeunderstoodas anextensionoflogicthatdoesnotrelyonany conceptof“randomness.”Inthiscontext,references to“chance”or“randomness”assomethingthat existsinthephysicalworld,ratherthaninour
