CHAPTER 1

CHANCE AND NECESSITY INEVOLUTION

Richard E. Lenski

Introduction

We humans have long recognized the profound tension that existsin our world between chance and necessity, between things thatseem to happen by accident and those that seem inevitable oreven purposeful. Democritus said that "everything existing in theuniverse is the fruit of chance and necessity." The aphorism that"necessity is the mother of invention" finds its counterpoint inMark Twain's quip that "necessity is the mother of taking chances."Even in our most goal-directed endeavors, we see the tensionbetween accident and purpose, as Louis Pasteur did in saying that"chance favors only the prepared mind."

My intention in writing this chapter is not to sort out the tanglednuances of the words chance and necessity. Nonetheless, itmight be helpful to illustrate some of these nuances before proceeding.Chance often invokes some instantaneous disturbance,such as a cosmic ray striking a chromosome and causing a particularmutation. Chance is also sometimes used with reference tocontingent effects of prior historical events, such as how the courseof life on Earth might have unfolded differently had some asteroidnot caused a certain mass extinction. Yet, the cosmic ray may havefollowed a path set by the laws of physics, and the historical influencesmight have been inevitable in their time. What the notionof chance captures is the sense of unpredictability and the absence of controlexerted by the affected system over its own eventual fate.

Necessity is fraught with even more divergent meanings. Necessity isoften used to describe outcomes that are inevitable given the action ofphysical laws, such as the motion of one billiard ball that has been struckby another ball at a particular angle and momentum. Necessity can alsorefer to a purposeful course of action, one that must be followed in orderto achieve some desired end, such as striking one ball with a cue so thatit hits another ball at the angle and momentum that is required to movethe second ball in a particular way. And in an evolutionary context, necessityprovides a shorthand term to describe adaptive solutions, produced bynatural selection, that allow organisms to cope with the various challengesthey face in their environments. The hand-eye coordination that enablesthe pool player to strike a ball precisely as intended might be an adaptationthat was necessary for survival during some part of the history of ourspecies. (Ayala [1999], Pennock [1999], and Ruse [2003] discuss importantsimilarities and differences between designs produced by the deliberateactions of conscious agents and those that result from natural selection.)

The Roles of Chance and Necessity in Evolutionary Thought

The tension between chance and necessity is perhaps more central to evolutionarybiology than to any other science. Physics certainly encompassesthe determinism of classical mechanics and the randomness of quantummechanics, but these forces play out at such different scales that the difficultylies in linking these two realms rather than in disentangling theireffects. By contrast, the tension between chance and necessity enters intocurrent evolutionary thought at two levels that are both central to ourunderstanding of the biological world in which we live.

At one level, we have the historical narrative of life on Earth that is theprimary focus of paleontological and much comparative research. It is agreat struggle, of course, to sort out what happened and when, especiallyacross the vast reaches of time. Nonetheless, things really did happen andat certain moments. Thus, there is only one true history that occurred,although we will never be able to reconstruct it in its entirety. But justas students of human history are fond of asking how things might haveunfolded differently if some past event were altered, so too evolutionarybiologists are fascinated by similar questions. Stephen Jay Gould, in WonderfulLife (1989), offered the thought experiment of "replaying life's tape"to evoke these what-if questions in the context of evolution. What if differentanimal phyla had survived the Cambrian than those that did? Whatif an asteroid had not hit Earth at the end of the Cretaceous? Or whatif the asteroid had been half the size, or twice the size, of the one thatactually hit? For that matter, what if it had hit just one hour sooner orlater? What difference would these accidental circumstances have made tothe subsequent evolution of life, including our own coming into being?For Gould, the quirks of history and the immensity of alternative pathsled him to infer the "awesome improbability of human evolution"—notonly in the narrow sense of our particular species but more generally inthe sense of any species that can wonder and reason about its own origins.Most evolutionists accept Gould's conclusion in the narrow sense,but others have argued against his more general conclusion. Simon ConwayMorris (2003) presents myriad examples of parallel and convergentevolution, whereby multiple lineages have independently evolved similaradaptations to similar challenges (such as eyes to detect light). He thenuses this repeatability to argue that any general features of organisms thatare of great adaptive value (and that are genetically accessible) would havearisen, sooner or later, and human-like intelligence is unlikely to be anexception.

The second level of interplay between chance and necessity lies at theheart of the mechanistic basis of Darwinian evolution itself. Natural selection,of course, provides the directing force by which organisms acquirefeatures that fit them to their environments. Those individuals that havecertain phenotypic features are more successful in the struggle for survivaland reproductive success than others that have different features. If thephenotypic differences are heritable, then those features that enhance performancewill tend to be amplified in later generations, giving the appearanceof direction, design, and purpose. Heritable differences betweenorganisms are encoded in their genomes, and these differences are producedby recombination and mutation. Sexual recombination scramblesthe existing differences between two parental genomes, while mutationprovides the ultimate source of this genetic variation. It is at the level ofmutation that Darwinian evolution is, in essence, random and accidental.

Let me be clear with respect to what evolutionary theory means—anddoes not mean—when we say that mutations are random and occur bychance. We do not mean that mutations occur at the same rate throughouta genome; in fact, some DNA sequences are more mutable than others.Nor do we mean that the environment plays no role in causing mutations;it does, as witnessed by mutagenic agents. Nor do we mean that organismscan exert no control whatsoever over the mutational process; in fact, organismsfrom bacteria to humans possess exquisite molecular machinery forproofreading their DNA and correcting errors during replication. Whatis important, however, is that mutations are random insofar as organismscannot direct the production of particular mutations in response to theirparticular needs. (Humans, through the tools of genetic engineering, areon the cusp, for better or worse, of directing some of our own mutations.)Thus, mutations are genetic accidents, and they do not provide the design-likedirectionality given by natural selection. However, in their scatter-shotway, mutations provide the heritable variation that is needed for selectionto proceed. Because more mutations are deleterious than are beneficial,much of natural selection consists of eliminating deleterious mutations.But some mutations produce useful features, and these have fueled theadaptation of organisms to their environments.

Charles Darwin is justifiably renowned for presenting a coherent bodyof evidence to support the general proposition of organic evolution andespecially for discovering the principle of natural selection. But he waslargely ignorant of hereditary mechanisms, including what we now callmutation. Even so, there was an important aspect of his reasoning thatI think is not nearly as well recognized as it should be. That is, Darwinwas remarkably clear in distinguishing between what he did understand—hownatural selection could improve fitness across generations—and whathe could not understand—the source of the variation on which selectionacted. His chapter on "Laws of variation" (1859, 170) concludes as follows:"Whatever the cause may be of each slight difference in the offspring fromtheir parents—and a cause for each must exist—it is the steady accumulation,through Natural Selection, of such differences, when beneficial tothe individual, that give rise to all the more important modifications ofstructure, by which the innumerable beings on the face of this earth areenabled to struggle with each other, and the best adapted to survive." Atthe outset of this same chapter (131), Darwin describes variation as being"due to chance" but adds, "This, of course, is a wholly incorrect expression,but it serves to acknowledge plainly our ignorance of the cause ofeach particular variation." Thus, a key to Darwin's success was his abilityto separate what he understood from what he did not. (George Zebrowski[2000], a science-fiction author, beautifully captured the fundamentalstrength and limitation of science in a maxim he attributed to the cosmologistHermann Bondi: "The power of science comes from being able tosay something, without having to say everything." Darwin was able to saysomething powerful and profound about the consequences of heritablevariation, even while he humbly and forthrightly admitted his ignoranceabout the underlying causes of that variation.)

For several decades after the rediscovery of Mendel's findings on particulateinheritance, it was widely accepted that mutations were randomevents in the sense that I discussed above. Indeed, many mutations weredemonstrably harmful to the organisms that carried them, and so it madelittle sense to think of them as somehow directed toward producing adaptation.However, it was difficult to test this assumption formally becausemost populations of experimental organisms, such as fruit flies, had substantialstanding variation, thus making it almost impossible to distinguishnew mutations from rare variants already present. Things were evenmore confused for those who worked with bacteria, where it was impossibleto see individual mutants or demonstrate their existence except byimposing selection for some new phenotype. When such selection wasimposed and the bacteria acquired a new phenotype, one could not tellif selection had caused the phenotypic conversion of the entire populationor, alternatively, if selection had allowed some rare mutant type totake over the population. One microbiologist of that era expressed thediscord as follows (Lewis 1934, 636): "The subject of bacterial variationand heredity has reached an almost hopeless state of confusion. Almostevery possible view has been set forth, and there seems no reason to hopethat any uniform consensus of opinion may be reached in the near future.There are many advocates of the Lamarckian mode of bacterial inheritance,while others hold to the view that it is essentially Darwinian." Awhile later, Julian Huxley (1942, 131—132) explicitly excluded bacteria fromthe then-modern evolutionary synthesis by saying, "They have no genes inthe sense of accurately quantized portions of hereditary substance...."

Ironically, just one year after Huxley excluded bacteria from the emergingevolutionary synthesis, the biologist Salvador Luria and the physicist-turned-biologistMax Delbrück published one of the great experiments ofall time, which demonstrated that bacterial mutations do, in fact, occurat random (Luria and Delbrück 1943). Without going into the details oftheir subtle and elegant experiment, they showed that mutations that conferredresistance on bacteria to lethal infections by viruses had occurredin generations prior to the bacteria's exposure to the viruses; hence, themutations could not have been caused by that exposure, and they musthave arisen spontaneously without regard to their utility. Further experimentsperformed by Joshua Lederberg and Esther Lederberg (1952) supportedthe same conclusion, and they did so in a way that made a strikingvisual impression on anyone who remained skeptical of the quantitativereasoning necessary to interpret the experiment of Luria and Delbrück.With these experiments, mutation and selection were firmly established asthe biological processes that correspond, respectively, to chance and necessity.(Again, by saying that mutations are due to chance, one does notimply that mutations lack physical causes. A certain mutation might havebeen caused by a cosmic ray hitting a particular site on a chromosome.But such physical events are beyond the control of the affected organism,in the same way that a gambler does not control the outcome of a throwof the dice, even though dice also obey ordinary physical laws.)

Putting the Powers of Chance and Necessity to the Test

So far in this chapter, I have touched on some important lines of biologicalthought on the ideas of chance and necessity and their evolutionarysignificance, ranging from experimental research focused on the origins ofmutations to paleontological and comparative perspectives on the potentialmacroevolutionary consequences of chance and necessity. I will nowsummarize some of my group's research in this area, which attempts tobridge perspectives and time scales by bringing the macroevolutionaryframework on contingency versus repeatability down to an experimentalscale. These experiments allow us to watch phenotypic and genomic evolutionacross thousands of generations. Also, the experiments involve replicatepopulations that begin with the same ancestor and evolve in identicalenvironments, such that we can characterize both parallel and divergentchanges. And the system can be preserved at intermediate stages, enablingus to rewind and restart the evolutionary tape in order to place hypothesesthat invoke historical contingency into the same framework as those thatinvoke adaptation. With these motivations, I will now discuss an experimentwith the bacterium Escherichia coli that has been underway in mylaboratory for almost two decades.

E. coli has a number of features that make it well suited for experimentsto investigate evolutionary dynamics and outcomes. This species is easy topropagate and enumerate; one can control and manipulate environmentalfactors; its generations are rapid and population sizes are large; and itreproduces asexually by binary fission. Moreover, one can preserve andlater revive ancestral genotypes as well as those from intermediate timesin an experiment. This last feature, coupled with suitable genetic markers,allows us to measure the extent of adaptation by allowing derived genotypesto compete against their own ancestors. Several decades of intensiveresearch on the physiology and genetics of E. coli provide a wealthof information on the inner workings of its cells, while various molecularbiological tools permit precise genetic analysis and manipulation.

In the long-term experiment, twelve populations were founded fromsingle cells of the same ancestral strain, and the populations have now beenpropagated for more than 40,000 bacterial generations in identical environments(Lenski 2004). The environment consists of a simple mediumwith glucose as the sole source of carbon and energy available to the cells.Every day, each population is diluted one hundred-fold into fresh medium,where it grows to several hundred million cells before depleting the glucoseand awaiting the next transfer. Because each population began as asingle haploid cell, there was no variation either within or between populationsat the outset (except a neutral genetic marker embedded withinthe design of the experiment). Therefore, all of the variation required foradaptation and divergence had to arise de novo by mutation, so that thisexperiment encompasses the origin as well as the fate of genetic novelties.Given the population size and knowledge of mutation rates, it is likelythat each population has had more than a billion mutations appear, evenafter taking into account the bottleneck effect during the daily transfers.And given the fact that the genome of E. coli is about five million base-pairs,it then follows that almost all mutations have been tried many timesover in each population. However, the fact that each one-step move hasbeen tried repeatedly does not imply that most genotypes have existed,as only a tiny "corner" of the immense genotypic space is ever probed insuch an experiment. Moreover, most mutations are lost to genetic drift ornatural selection, and I estimate that only tens or hundreds of mutationshave been substituted in a typical population (Lenski 2004).