ANTARCTIC fish deploy antifreeze proteins to survive in cold water. Tasty viceroy butterflies escape predators by looking like toxic monarchs. Disease-causing bacteria become resistant to antibiotics. Everywhere you look in nature, you can see evidence of natural selection at work in the adaptation of species to their environment. Surprisingly though, natural selection may have little role to play in one of the key steps of evolution – the origin of new species. Instead it would appear that speciation is merely an accident of fate.
So, at least, says Mark Pagel, an evolutionary biologist at the University of Reading, UK. If his controversial claim proves correct, then the broad canvas of life – the profusion of beetles and rodents, the dearth of primates, and so on – may have less to do with the guiding hand of natural selection and more to do with evolutionary accident-proneness.
Of course, there is no question that natural selection plays a key role in evolution. Darwin made a convincing case a century and a half ago in On the Origin of Species, and countless subsequent studies support his ideas. But there is an irony in Darwin’s choice of title: his book did not explore what actually triggers the formation of a new species. Others have since grappled with the problem of how one species becomes two, and with the benefit of genetic insight, which Darwin lacked, you might think they would have cracked it. Not so. Speciation still remains one of the biggest mysteries in evolutionary biology.
Advertisement
Even defining terms is not straightforward. Most biologists see a species as a group of organisms that can breed among themselves but not with other groups. There are plenty of exceptions to that definition – as with almost everything in biology – but it works pretty well most of the time. In particular, it focuses attention on an important feature of speciation: for one species to become two, some subset of the original species must become unable to reproduce with its fellows.
How this happens is the real point of contention. By the middle of the 20th century, biologists had worked out that reproductive isolation sometimes occurs after a few organisms are carried to newly formed lakes or far-off islands. Other speciation events seem to result from , which suddenly leave some individuals unable to mate successfully with their neighbours.
It seems unlikely, though, that such drastic changes alone can account for all or even most new species, and that’s where natural selection comes in. Species exist as more or less separate populations in different areas, and the idea here is that two populations may gradually drift apart, like old friends who no longer take the time to talk, as each adapts to a different set of local conditions. “I think the unexamined view that most people have of speciation is this gradual accumulation by natural selection of a whole lot of changes, until you get a group of individuals that can no longer mate with their old population,” says Pagel.
Until now, no one had found a way to test whether this hunch really does account for the bulk of speciation events, but more than a decade ago Pagel came up with an idea of how to solve this problem. If new species are the sum of a large number of small changes, he reasoned, then this should leave a telltale statistical footprint in their evolutionary lineage.
Whenever a large number of small factors combine to produce an outcome – whether it be a combination of nature and nurture determining an individual’s height, economic forces setting stock prices, or the vagaries of weather dictating daily temperatures – a big enough sample of such outcomes tends to produce the familiar bell-shaped curve that statisticians call a normal distribution. For example, people’s height varies widely, but most heights are clustered around the middle values. So, if speciation is the result of many small evolutionary changes, Pagel realised, then the time interval between successive speciation events – that is, the length of each branch in an evolutionary tree – should also fit a bell-shaped distribution (see diagram). That insight, straightforward as it was, ran into a roadblock, however: there simply weren’t enough good evolutionary trees to get an accurate statistical measure of the branch lengths. So Pagel filed his idea away and got on with other things.
Then, a few years ago, he realised that reliable trees had suddenly become abundant, thanks to cheap and speedy DNA sequencing technology. “For the first time, we have a large tranche of really good phylogenetic trees to test the idea,” he says. So he and his colleagues Chris Venditti and Andrew Meade rolled up their sleeves and got stuck in.
The team gleaned more than 130 DNA-based evolutionary trees from the published literature, ranging widely across plants, animals and fungi. After winnowing the list to exclude those of questionable accuracy, they ended up with a list of 101 trees, including various cats, bumblebees, hawks, roses and the like.
Working with each tree separately, they measured the length between each successive speciation event, essentially chopping the tree into its component twigs at every fork. Then they counted up the number of twigs of each length, and looked to see what pattern this made. If speciation results from natural selection via many small changes, you would expect the branch lengths to fit a bell-shaped curve. This would take the form of either a normal curve if the incremental changes sum up to push the new species over some threshold of incompatibility, or the related lognormal curve if the changes multiply together, compounding one another to reach the threshold more quickly.
To their surprise, neither of these curves fitted the data. The lognormal was best in only 8 per cent of cases, and the normal distribution failed resoundingly, providing the best explanation for not a single evolutionary tree. Instead, Pagel’s team found that in 78 per cent of the trees, the best fit for the branch length distribution was another familiar curve, known as the exponential distribution ().
Happy accidents
Like the bell curve, the exponential has a straightforward explanation – but it is a disquieting one for evolutionary biologists. The exponential is the pattern you get when you are waiting for some single, infrequent event to happen. The time interval between successive telephone calls you receive fits an exponential distribution. So does the length of time it takes a radioactive atom to decay, and the distance between roadkills on a highway.
To Pagel, the implications for speciation are clear: “It isn’t the accumulation of events that causes a speciation, it’s single, rare events falling out of the sky, so to speak. Speciation becomes an arbitrary, happy accident when one of these events happens to you.”
All kinds of rare events could trigger the accident of speciation. Not just physical isolation and major genetic changes, but environmental, genetic and psychological incidents. The uplift of a mountain range that split a species in two could do it. So too could a mutation that made fish breed in surface waters instead of near the bottom, or a change in preference among female lizards for mates with blue spots rather than red ones.
The key point emerging from the statistical evidence, Pagel stresses, is that the trigger for speciation must be some single, sharp kick of fate that is, in an evolutionary sense, unpredictable. “We’re not saying that natural selection is wrong, that Darwin got it wrong,” Pagel adds. Once one species has split into two, natural selection will presumably adapt each to the particular conditions it experiences. The point is that this adaptation follows as a consequence of speciation, rather than contributing as a cause. “I think what our paper points to – and it would be disingenuous for very many other people to say they had ever written about it – is what could be, quite frequently, the utter arbitrariness of speciation. It removes speciation from the gradual tug of natural selection drawing you into a new niche,” he says.
“The trigger for speciation must be some single, sharp kick of fate that is, in an evolutionary sense, unpredictable”
This has implications for one of the most contentious aspects of evolution: whether it is predictable or not. If Pagel is correct, natural selection shapes existing species in a gradual and somewhat predictable way, but the accidental nature of speciation means that the grand sweep of evolutionary change is unpredictable. In that sense his findings seem to fit with the famous metaphor of the late Stephen Jay Gould, who argued that if you were able to rewind history and replay the evolution of life on Earth, it would turn out differently every time.
“If you were able to rewind history and rerun the evolution of life on Earth, it would turn out differently every time”
So far, other evolutionary biologists have been reluctant to accept Pagel’s idea wholeheartedly. Some regard it as interesting but in need of further testing. “The single, rare events model is brilliant as an interpretation – as a potential interpretation,” says Arne Mooers at Simon Fraser University in Vancouver, Canada.
Others suspect Pagel’s analysis has highlighted only part of the story. “It’s telling you about one necessary but not sufficient component of speciation,” says Daniel Rabosky at the University of California, Berkeley. “You have to have two things: something to cause isolation, and something to cause differentiation.” And the latter process – through which the two isolated populations change enough that we recognise them as two distinct species – is likely to involve gradual, adaptive change under the hand of natural selection.
The notion that the formation of a new species has little to do with adaptation sits uncomfortably with fundamental ideas about evolution. A particular stumbling block is what evolutionary biologists call “adaptive radiations”. When ecological opportunities open up – as, for example, when finches from the South American mainland first colonised the Galapagos – species seem to respond by diversifying into a host of new forms, each adapted to a particular niche. These bursts of speciation suggest that organisms need not wait for some rare event to push them into speciating, but instead can be pulled there by natural selection. “I would take it that there is quite a bit more pull than push,” says Leigh Van Valen at the University of Chicago.
But is there? In his analysis, Pagel specifically looked for the signature of this kind of evolutionary exuberance. Bursts of speciation would manifest as trees with lots of branching at irregular intervals; in other words, a highly variable rate of change over time, giving rise to a subtly different curve. “It was the model that, going in, I thought would explain far and away the most trees,” says Pagel.
He was wrong. “When it works, it works remarkably well,” he says. “But it only works in about 6 per cent of cases. It doesn’t seem to be a general way that groups of species fill out their niches.”
This finding has independent support. Luke Harmon at the University of Idaho in Moscow and his colleagues have examined 49 evolutionary trees to see whether there are bursts of evolutionary change early in a group’s history, when unfilled niches might be expected to be most common. There is little evidence for such a pattern, they report in a paper that has been accepted for publication in the journal Evolution.
Why so many rodents?
If speciation really is a happy accident, what does that mean for the way biologists study it? By focusing on the selective pressures that drive two species into different ecological niches, as they currently do, they may learn a lot about adaptation but not much about speciation. “If you really want to understand why there are so many rodents and so few of other kinds of mammals, you should start to look at the catalogue of potential causes of speciation in an animal’s environment, rather than take the view that there are all these niches out there that animals are constantly being drawn into,” Pagel says.
Rodents adapted to cool climates, for example, would be prone to isolation on mountain tops if the climate warmed. That could make them more likely to speciate than mammals adapted to warm temperatures. Likewise, marine animals whose larval stages live on the sea floor might be more likely to split into separate isolated populations and therefore speciate more often than those with free-floating larvae. Indeed, this is exactly what palaeontologist David Jablonski of the University of Chicago has found among marine snails. Similarly, species with narrow habitat requirements or finicky mate-choice rituals may also be prone to accidental splits.
What other possible accidents might there be? No one knows yet. “We’d like people to start compiling the lists of these things that might lead to speciation, and start making predictions about who’s going to have a high rate of speciation and who’s going to have a low rate,” says Pagel. If these lists help us understand the broad sweep of evolutionary history – the rise of mammals, why there are so many species of beetles, or the remarkable success of flowering plants – then we will know Pagel is onto something fundamental.
In the meantime, though, Pagel’s take on speciation may help explain another puzzling feature of the natural world. Over and over again, as biologists sequence the DNA of wild organisms, they find that what appears superficially to be a single species is actually two, several or even many. The forests of Madagascar are home to 16 different species of mouse lemurs, for example, all of which live in similar habitats, do similar things, and even look pretty much alike. These cryptic species complexes are difficult to explain if speciation is the end result of natural selection causing gradual divergence into different niches. But if new species are happy accidents, there need be no ecological difference between them.
Pagel’s own epiphany in this regard came in Tanzania, as he sat at the base of a hardwood tree watching two species of colobus monkeys frolic in the canopy 40 metres overhead. “Apart from the fact that one is black and white and one is red, they do all the same things,” he says. “I can remember thinking that speciation was very arbitrary. And here we are – that’s what our models are telling us.”
