How much of evolution relies on randomness? (Image: Collage: Eugenia Loli; Illustration: Katelyn Daubek)
TAKE 100 newly formed planets of one Earth mass. Place each in the habitable zone of a G-type main sequence star. Set your timer for 4 billion years. What do you get? A hundred planets teeming with life forms quite similar to those on Earth, perhaps even dominated by naked apes? Or would evolution produce very different outcomes every time, if life even got started at all?
Some biologists argue that evolution is a deterministic process, that similar environments will tend to produce similar outcomes. Others, the most famous of whom was Stephen Jay Gould, think its course follows unpredictable twists and turns, and that the same starting point can lead to very different results.
Advertisement
The answer does matter. If the Gould camp is right, the study of evolution is like the study of history: something we can understand only in retrospect. If, however, the vagaries of chance play just a minor role, then biologists can predict the course of evolution to a large extent – and predicting evolution is crucial to stopping tumours becoming drug-resistant, or bacteria shrugging off an antibiotic, or bedbugs becoming immune to pesticides, or viruses killing people who have been vaccinated against them and so on.
So which is it? We might not have 100 Earths and a time machine, but we can look at how evolution has turned out on, say, neighbouring islands, or even rerun it over and over in the lab. These kinds of studies are giving us a better idea of the role of chance.
First things first. Evolution does begin with chance events, in the form of mutations. But it is not a case of anything goes; far from it. Which mutations survive and spread depends on natural selection – the survival of the fittest. Put another way, chance is the creative partner that comes up with all the ideas – some brilliant, others hopeless – while natural selection is the ruthlessly practical one, picking what works.
Many biologists, most notably Richard Dawkins, therefore insist that although mutations may be random, evolution is not. This insistence might make sense when explaining evolution to people who have not grasped the basic concept. But there is an element of chance in evolution, even when natural selection is firmly in the driving seat.
Take the evolution of flu viruses. We can predict with confidence that, over the next few years, the structure of a viral surface protein called haemagglutinin will evolve so that the human immune system can no longer recognise and attack it. What’s more, we can even be fairly sure that the mutations that allow new strains of flu to evade the immune system will happen at one of seven critical sites in the gene coding for haemagglutinin, says Trevor Bedford, an evolutionary biologist at Fred Hutchinson Cancer Research Center in Seattle. In this sense, the evolution of flu is non-random and predictable.
“Probability of a randomlypicked clover having four leaves: about 1 in 10,000”
But it’s a matter of chance which of those seven sites mutate, and how. Predicting the course of flu’s evolution is almost impossible more than a year or two in advance, says Bedford. This is why flu vaccine makers do not always get it right, and why this season’s flu vaccine was largely ineffective.
What’s more, as important as natural selection is, its powers are limited. The fittest do not always survive; instead, the course of evolution is . If it hadn’t been for an asteroid strike, for instance, we mammals might still be scurrying about in mortal fear of dinosaurs. And if a different bird had been blown to the far-off Galapagos Islands a few million years ago, we might talk about Darwin’s crows instead of Darwin’s finches.
We’ve long known about this “founder effect”, but recent studies suggest it may be more important than thought. For example, a handful of little birds were the ancestors of several populations of Berthelot’s pipit on the Selvagem and Madeira island chains, in the Atlantic. There are big variations among them in the shape and size of beaks, legs and wings.
When Lewis Spurgin of the University of East Anglia in Norwich, UK, studied these populations, he expected to find environmental differences that explained this variation, but . Instead, he concluded that the physical differences were not driven by natural selection but were just a result of the small number of founders: accidents of history, in other words ().
Accidental process
The founder effect can even create new species without the need for natural selection. When Daniel Matute, now at the University of North Carolina in Chapel Hill, took a large population of fruit flies and created 1000 founder populations of a single male and female in identical vials in his lab, most populations simply went extinct because of inbreeding. But in three of the surviving populations, the founders produced offspring different enough that they were with the larger parental population – the first step to the creation of a new species.
Effects like these might explain why the islands of Hawaii have such a rich diversity of fruit flies. In fact, a few biologists think speciation is almost always an accidental process, rather than one driven by natural selection (żěè¶ĚĘÓƵ, 13 March 2010, p 30).
Yet more evidence of the limits of natural selection comes from genomes, which are littered with the products of chance. Despite many claims to the contrary, , for instance. This junk has accumulated because natural selection has not been strong enough to remove it, says Michael Lynch, an evolutionary biologist at Indiana University in Bloomington. In small populations, even mutations that are slightly harmful can spread throughout the population simply by chance.
Does this kind of genetic drift really matter? At least sometimes, it does. Joe Thornton of the University of Chicago has been turning back the clock and replaying evolution to see if it could have turned out differently. Think Jurassic Park, except rather than recreate extinct animals, Thornton has recreated ancient proteins. His team began with living vertebrates that each have their own version of the gene encoding the protein that detects the stress hormone cortisol. By comparing the versions, they could work out how it had evolved over hundreds of millions of years, from a protein that could detect another hormone.
“Probability of aroyal flush in poker: 1 in 649,739”
Then Thornton’s team went a lot further. They actually made some of these ancient proteins and tried them out to see what effect each mutation had. Switching to cortisol took five mutations: two to recognise cortisol and three to “forget” the previous hormone.
But when the team made only these five changes, they destabilised the protein and wrecked it. It turns out the transition to cortisol was possible only because two other mutations that stabilise the protein had occurred first. But these “permissive” mutations have no effect by themselves. They must have arisen by chance, not by natural selection ().
“We think of these permissive mutations as opening doors, so that evolution has the opportunity to follow pathways that were inaccessible without the permissive mutations,” says Thornton. And there seems to be only one way the door to the cortisol-binding pathway could have opened. Thornton tested thousands of other mutations, but none did the trick. “There is nothing else in the neighbourhood around the ancestral protein that could have opened that door,” he says.
In Thornton’s view, the course of evolution often – although not always – hinges on such seemingly insignificant chance events. In this way, evolution is a lot like life, he notes: a seemingly inconsequential decision one night to go to one party rather than another might lead to meeting your future partner and thus change the course of your life.
Then again, who we hook up with seldom alters the course of history. Although all these studies suggest that chance plays a bigger role in evolution than generally acknowledged, the big question is how much difference it makes in the long run. The detailed paths taken by evolving populations might depend largely on chance, yet still lead to similar outcomes. There are only so many ways of flying and swimming, for instance, which is why wings and fins have independently evolved on many occasions. If Thornton’s protein hadn’t evolved the ability to bind cortisol, perhaps another protein would have instead.
There are many examples of this kind of convergent evolution. Arctic and Antarctic fish have that work in the same way, for example, while several snake lineages have separately come up with they eat.
In the Greater Antilles in the Caribbean, meanwhile, evolution has effectively been rerun on four islands – and turned out the same way. Each of the islands has long-legged Anolis lizards that run on the ground, short-legged ones that grasp twigs, and lizards with big toepads that stick to leaves. But seem to derive from a single founder population, meaning they independently evolved to fill the same niches.
Does this mean Gould was wrong after all, that in the long run chance does not matter that much? Perhaps the closest we can get to an answer is the Long-Term Experimental Evolution Project, led by Richard Lenski of Michigan State University. On 24 February 1988, Lenski took samples of one kind of E. coli bacteria and used them to found 12 new populations. Every day since then – on weekends and holidays, despite blizzards and grant deadlines – someone has kept them growing by transferring samples to new nutrient medium.
Replaying evolution
In the 27 years that have passed, Lenski’s populations evolved for about 60,000 generations. For comparison, Homo sapiens has gone through perhaps 20,000 generations in its entire existence. All 12 populations have changed in similar ways, evolving larger cells and faster growth rates, showing that sometimes evolution really does unfold in predictable ways.
But even without external events like asteroid strikes, its course was not always predictable. One population evolved into a mix of two lineages, each of which survive because they pursue slightly different strategies. Another suddenly developed, at about the 31,500th generation, the ability to feed on citrate, an additive to the culture medium that E. coli cannot normally use. “They started from the same place and were subjected to exactly the same conditions, and differences still pop up,” says Zachary Blount, who works with Lenski on the project. “The differences arise purely out of the chance that is inherent in the evolutionary process.”
Was the citrate-using mutation a lucky break, or could evolution find it again? Because Lenski’s team freezes a sample of each culture every 500 generations, Blount was able to go back into the archives of this population and literally rerun evolution. When he did so, the only time citrate use evolved was when he began the replay with cells from the 20,000th generation or later.
Clearly, some mutation or mutations must have happened around the 20,000th generation that set the stage for citrate use to evolve much later, just as Thornton’s hormone receptor required permissive mutations before it could switch to recognise a different target. “We still haven’t figured out what that mutation was, which is really frustrating,” says Blount. Until they can, the team will not know whether the permissive mutation offered some other advantage to the bacteria. Even if it did, however, it seems clear that its role in permitting citrate use must have been just a lucky by-product.
So what would we get if we could replay evolution over and over on a planetary scale? One possibility is an awful lot of slime. Nick Lane of University College London thinks that the emergence of complex cells depended on a highly unlikely merger of two kinds of simple cell (żěè¶ĚĘÓƵ, 23 June 2012, p 32). If he’s right, bacteria-like life forms are common on other worlds but rarely give rise to more sophisticated organisms.
No naked apes
But assuming life did get past the slime stage on our worlds, what would it be like? “There is a fairly good chance that such replays would often yield worlds that look broadly like ours in terms of what niches are filled, and what sorts of major traits you see,” says Blount. In other words, you’d still expect to see photosynthesisers and predators, and parasites and decomposers. But the details are likely to differ sharply from one replay to the next, he says. Even if we replayed evolution a hundred times, it’s highly unlikely that we would end up again with a big-brained primate ruling the planet.
But would some other brainy, social animal take over the planet? Maybe. “There’s clearly an adaptive zone in most habitats that involves intelligence,” says David Jablonski, a palaeontologist at the University of Chicago. And it has become clear that many traits we once thought of as uniquely human, from language to tool-making, exist to some extent in many other animals. So although naked apes might not emerge on any of the 100 planets, other smart tool-users might.
It is a question we might even be able to answer one day. Thousands of exoplanets have now been discovered and even though we’ve yet to find any just like ours, all the evidence suggests there are plenty of Earth-like planets close enough that we might not only determine whether they support life, but also learn a little about it. The answer may be in the stars.
Read more: “Chance: How randomness rules our world“
This article appeared in print under the headline “Life chances”
