A Surprise for Evolution in Giant Tree of Life

Researchers build the world’s largest evolutionary tree and conclude that species arise because of chance mutations — not natural selection.

Katie Scott for Quanta Magazine


Honeycreepers, small birds inhabiting the Hawaiian Islands, have a rich assortment of beak shapes. Some species have long, thin beaks suited to plucking insects from leaves. Others possess thick beaks good for cracking open tough seeds. According to the classic view of evolution, natural selection drove the development of these different species. Each variant adapted to suit a different ecological niche. But Blair Hedges, a biologist at Temple University in Philadelphia, has proposed a provocative alternative: Adaptation had little to do with it. It was simply a matter of chance and time.

This controversial proposal stems from efforts by Hedges and collaborators to build the world’s most comprehensive tree of life — a chart plotting the connections among 50,000 species of Earth’s vast menagerie. Their analysis suggests that speciation is essentially random. No matter what the life form — plant or animal, insect or mammal — it takes about 2 million years for a new species to form. Random genetic events, not natural selection, play the main role in speciation.

Evolutionary biologists find the research effort intriguing, particularly in its size and scope, but they are also somewhat skeptical of the provocative ideas that have emerged. “It’s a huge tour de force,” said Arne Mooers, a biologist at Simon Fraser University in British Columbia. “There are lots of interesting claims — the devil will be in the details.”

Courtesy of Blair Hedges

Blair Hedges, a biologist at Temple University in Philadelphia, challenges the idea natural selection drives the development of new species.

To build the tree, Hedges, his Temple colleague Sudhir Kumar, and their collaborators compiled data from nearly 2,300 published studies, gleaning from each the time when two species diverged from a common ancestor. They used those data to construct a map of relationships among different species, known as a “timetree.” To form a branch, the researchers started with the two species within a closely related taxonomic group that have the most recent common ancestor. Then they added the next closest species, and so on. (In a family tree, that’s akin to starting with siblings, then adding in first cousins and second cousins.) Bringing all those branches together results in a comprehensive timetree of life.

“It’s an astonishingly large exercise they’ve done,” said Michael Benton, a paleontologist at the University of Bristol in England. “Many people in the field would be unable to do that.”

It will take some time for scientists to sort through the technical details of the paper, which was published in April in the journal Molecular Biology and Evolution. And while some scientists have been complimentary, others immediately challenged the results, questioning both the accuracy of the tree and the conclusions that Hedges has drawn. “I am very skeptical about inferring patterns of speciation from such a broad overview of the tree of life,” said Chris Jiggins, a biologist at the University of Cambridge in England.

Species Enigma

One reason scientists are skeptical is that Hedges’ clocklike pattern conflicts with the traditional picture of how evolution unfolds. “The classic view of evolution is that it happens in fits and starts,” Benton said. A change in the environment, such as a rise in temperatures after an ice age, might spark a burst of speciation as organisms adapt to their new surroundings. Alternatively, a single remarkable adaptation such as flight in the ancestors of birds or hair in mammals might trigger a massive expansion of animals with those characteristics.

Hedges argues that while such bursts do occur, the vast majority of speciation is more prosaic and evenly timed. To start, two populations become separated, driven apart by geography or other factors. New species emerge every 2 million years, on average, in a metronomic rhythm tapped out by the random nature of genetic mutations. He likens the process to radioactive decay. It’s impossible to predict when an individual radioactive nucleus will decay, but a clump of many atoms will decay at a highly predictable rate known as the material’s half-life. Similarly, mutations strike the genome randomly, but over a long enough time the accumulation of mutations follows a pattern. “There is a kind of speciation clock ticking along,” Hedges said.

Emily Fuhrman for Quanta Magazine

In this interactive infographic, three lines trace how frequently organisms speciate over time. Across the natural world — in vertebrates, arthropods and plants — new species most often arise after a little more than 2 million years. This consistency suggests to some researchers that random genetic mutations rather than natural selection drive the development of new species.

Consider Hawaii’s honeycreepers. The speciation clock started once the birds migrated to a new island and began to accumulate random mutations. The vast majority of these mutations were neutral, having no effect on the birds’ appearance or behavior. Occasionally a beneficial mutation appeared, such as one that made the beak longer and its bearer a more efficient hunter. According to the traditional model of speciation, the adaptations eventually made the two populations too different to interbreed even if they were to come back into contact. In this scenario, adaptations drive the creation of a new species.

But Hedges contends that speciation and adaptation are two distinct processes, each proceeding along its own path. (A team led by Mark Pagel, an evolutionary biologist at the University of Reading in England, has made a similar proposal, though for different reasons.) According to Hedges’ model, after about 2 million years the two groups of birds accrued so many random genetic differences that they became incompatible. It wasn’t adaptive mutations that made it impossible for the birds to intermingle, but rather the accumulation of enough mutations overall, most of them neutral ones. Geographic isolation provided the necessary spark for speciation, but simple time drove the process to its conclusion.

Tangled Trees

The seed for Hedges’ tree sprouted 17 years ago, when he and Kumar began assembling a database of species divergence — the time when two species split from a common ancestor. But it wasn’t until recently that the researchers developed a method capable of synthesizing the collection of data into one grand tree of life.

Hedges convened a small conference on biological diversity at Temple last month, where he and others presented their work. Many in the audience were eager to apply techniques that Hedges and his team had developed, particularly an efficient method for identifying dates on trees. “It became a running joke that we had all spent years of computer time building our own dated trees when we could have done it a million times faster using their approach,” Mooers said.

Is the Earth Full?

Only so many people can safely fit on an airplane — once the seats are filled, there’s no more room. Is this also the case for the planet? Can Earth’s ecological niches be filled to capacity? A number of studies published over the past few years suggest that the answer is yes. They found that the rate at which new species are forming is slowing down, implying that there has not been room for new variants to expand and take root. But Hedges’ team saw no evidence for this. “We found that overall, the rate has been constant for almost the whole history of life,” Hedges said. In addition, past patterns may not have much bearing on the future. In recent years, extinction rates have skyrocketed to 1,000 to 10,000 times the natural level, far surpassing the rate at which new species evolve. Even if the world is full, human activity may be opening new niches.

But there was an undercurrent of doubt beneath the excitement. Building phylogenetic trees, let alone giant ones, is a process fraught with disagreement. While scientists have been constructing them for two decades, it’s only in the last five years that they have been able to build large trees with more than a thousand species. The structure of a tree can vary depending on the data that go into it and the methods used to assemble it. To construct a comprehensive tree based on thousands of individual studies, Hedges’ team had to figure out how to deal with trees built using very different methods. “Any little biases in the studies going in can percolate up and be amplified,” Mooers said — a problem for any study encompassing thousands of species.

What’s more, many of the dates in these trees are uncertain or conflicting. Estimates for when two species split might range from 3 million to 8 million years ago, and the scientists have to decide which figure to use. “A lot of objective decisions have to be made,” Hedges said. “You have to reconcile differences and somehow put them together in one single consensus.”

Hedges’ tree also excludes groups for which no genetic data are available, a category that includes many organisms that have gone extinct. Paleobiologists say this choice can skew the outcome. For example, excluding extinct species from the analysis can make it look as though the number of species is increasing even when it’s not, said Charles Marshall, a paleobiologist at the University of California, Berkeley.

“The deep patterns will require a lot of technical evaluation,” Mooers said. “You can disagree with one-half of the paper and still find a lot to discuss and agree with.” Already, some researchers are questioning the basic results. Many of the geological ages of the nodes don’t match what scientists have gleaned from the fossil record or from smaller trees, Marshall said.

In addition, some scientists disagree with one of the main conclusions of the analysis — that a wide variety of species seem to evolve on a time scale of roughly 2 million years. Even Hedges was surprised by the finding, given the enormous differences between, say, a wasp and a whale. One hundred generations of insects might pass in the lifetime of a mammal, so one might expect the former to develop new species more quickly. Hedges concluded that they don’t. “There are huge biological and ecological differences in these groups,” he said. “And yet speciation time is very similar.”

This finding contradicts previous studies, which showed large differences in how quickly new species develop. “Our work shows that rates of speciation are indeed constant within groups of organisms, such as within bumblebees, thistles or dogs,” said Pagel, “but that speciation rates can vary among groups.”

Hedges maintains that constant speciation times are driven by the relatively constant pace of genetic mutations across the natural world. But it’s possible that Hedges’ consistent speciation rates result from averaging different rates across a range of organisms, Benton said. “To suggest a kind of regular pattern is one people will find quite challenging,” he said. “I see where they get it from, but I think there are other possible explanations.”

While researchers debate their findings, Hedges’ group is already working on an even bigger tree — one with perhaps half a million of the roughly 1.5 million known species. “That will give us a more accurate picture of the evolution of life and how biodiversity will change in the future from human activities,” Hedges said.

This article was reprinted on Wired.com.

View Reader Comments (28)

Leave a Comment

Reader CommentsLeave a Comment

  • Darwin had said that speciation is essentially random. Hedges should have read “On the Origin of Species” carefully,

  • I don’t agree. Whether a mutation is good, neutral or bad is relative to the organism’s environment. Assume a native population will evolve to some idealized state. In that state assuming no environmental change a mutation can only be neutral or bad, with the mutations that would be considered good in the remote environment (such as a longer beak) appearing at the same rate as in the remote environment but with no competitive advantage for passing on the trait, and without that advantage the trait will rarely if ever be observed. In the remote environment the good mutation will provide an advantage over the neutral mutations, but also over the vast majority of organisms with no mutation. The random element of speciation may provide a maximum speed at which it can occur, but natural selection increases the speed at which it actually does occur.

  • Variation and selection. Are you telling me the study publishers have just discovered that there’s no selection unless there’s random variation? The modern evolutionary synthesis dates back to the 30’s and 40’s. Good to see everything we knew a lifetime ago becomes new again after another lifetime. And that foolishness remains a constant.

  • So if I understand correctly, the crux of the controversy is this:

    > “According to the traditional model of speciation, the adaptations eventually made the two populations too different to interbreed even if they were to come back into contact. In this scenario, adaptations drive the creation of a new species. But Hedges contends that speciation and adaptation are two distinct processes, each proceeding along its own path. […]According to Hedges’ model, after about 2 million years the two groups of birds accrued so many random genetic differences that they became incompatible. It wasn’t adaptive mutations that made it impossible for the birds to intermingle, but rather the accumulation of enough mutations overall, most of them neutral ones. Geographic isolation provided the necessary spark for speciation, but simple time drove the process to its conclusion.”

    Which boils down to two points:

    * Speciation and adaptation are separate processes
    * Given that a population becomes isolated, speciation will happen at constant rate, regardless of adaptations, and take about 2 million years.

    I’m tempted to agree with the first.

    If speciation is defined as “when two separated populations are no longer capable of producing offspring”, then the incompatibility is more likely to exist at the genetic level before it exists at the physiological level. Furthermore it has to be widespread enough within the populations to be the “norm”. So then the question is whether adaptation has a strong effect on how fast genetic diversification happens. If this influence is small, then concluding that speciation and adaptation are separate processes sounds reasonable.

    Also, there are clear examples of wildly different adaptations that don’t mean incompatibility: we still consider the many dog breeds to belong to the the same species (although if we have a big and a small breed, there might be a physiological incompatibility). Then there’s the story of the fast adaptations of the stickleback fish.

    Perhaps the real issue is that the definition of “species” isn’t precise enough or universally agreed upon? Kind of like how “planets” used to be an ambiguous term in astronomy?

    As for the second point, I guess that’s a matter of figuring out what the various driving forces influencing speciation, and determining their “weight” in the process.

  • German Sterin – Darwin did NOT say that speciation is ‘random.’ Natural selection is exactly the opposite of randomness. You need to read ‘Origin’ more carefully.

    Evolution (including speciation) is far from random. Individual mutations may occur ‘randomly’ (btw, that term is used too loosely), but, each propagated mutation constrains the sustainability (probability of selection) of subsequent mutations. That is to say, mutations represent commitments to ever more specific paths of differentiation.

    If speciation were simply the accumulation of ‘random’ mutations without selection, there would be blind (cave) fish in open water and sighted fish in lightless caves. Hedges notion is not biology – it’s more akin to the kind of speculation found in anthropology.

    The convergent evidence for natural selection of adaptive mutations is so overwhelming that the principle is indisputable.

  • The only thing new in this study is the claim that the speciation rate is constant, which in itself is not surprising given that the models of genetic change have always shown a relatively constant mutation rate for parts of the genome which are not critical. The question seems to be the variation about this rate as a function of species type.

  • Never let it be said that this author understated the importance of his findings. In my book humility is a trait of a good scientist. Even Newton said he “stood on the shoulders of giants”

  • No news here, folks. Move along. *Of course* species arise from random mutation. Whether or not a species *survives* (and for how long it survives) is where natural selection comes into play.

  • “New species emerge every 2 million years, on average, in a metronomic rhythm tapped out by the random nature of genetic mutations. He likens the process to radioactive decay. ” It’s quite remarkable since it suggests the rate of mutation is independent of the rate of reproduction. It suggests that the mutation of the seed is what matters… which then brings us to Chromosome 2 for homo sapiens… Yes, mus musculus is a great example of a species with an acrocentric centromere and multiple karotypes, but I’d imagine reproduction rate would have some bearing on how often such a merging of chromosomes would be both stable and reproducible.

  • The rate of mutation is not necessarily dependent on the lifespan of the individual. Mutations could occur during replication of the DNA in the cells in the individual that pass along the DNA between the fertilized egg and the sperm and egg cells for the next generation. Mutations can also occur in eggs cells that developed early in the lifespan and just wait, all the while being subjected to ambient radiation, retro-viruses, and other effects on the DNA when it is not involved with reproduction.

  • Shortly after its appearance I posted the following comment on the Hedges paper in PubMed Commons. It draws attention to the late nineteenth century view of Darwin’s research associate that adaptation was decoupled from speciation. This can be interpreted in the modern era as the non-adaptive ticking of the GC% clock that generates GC% differences sufficient to impair meiotic pairing (hybrid sterility):

    ADAPTATION DECOUPLED FROM SPECIATION. This fine new paper presents an impressive synthesis of phylogenetic data aiming to “explore how it bears on evolutionary hypotheses and mechanisms of speciation and diversification.” In keeping with the results of Venditti et al. (2010) that are cited, the major conclusion is that “if adaptation is largely decoupled from speciation, we should not expect it to be a driver of speciation.” Indeed, “Cases where the phenotype has changed little (e.g. cryptic species) … are interpreted here as evidence of uncoupling.” There is reference to geographic isolation as “the major model,” but it is noted that “time constraints should be similar with ecological speciation, and other models exist.”

    One of these “other models” is considered by Venditti et al. (2008 Biologist 55, 140-146), who note:

    “There is a growing appreciation amongst evolutionary biologists that rapid reproductive isolation is more common than previously thought and is often associated with what is known as sympatric speciation, or speciation between populations which share the same geographic range.”

    The idea of a non-geographic decoupling of adaptation from speciation was advanced by Darwin’s research associate George Romanes in 1886. As with Venditti et al. (2010), the present results nicely support Romanes, whose work is the major focus of my speciation text (The Origin of Species, Revisited, McGill-Queen’s University Press, 2001). There is further elaboration both in our biography of the geneticist William Bateson (Treasure Your Exceptions, Springer, New York, 2008) and in my textbook Evolutionary Bioinformatics (Springer, New York, 2011).

  • A naive analysis suggests the 2 million year rule right be plausible.

    After all, larger animals present a greater surface area to accumulate genetic mutations through radiation damage. They also live longer allowing more mutations to accumulate per generation.

    So it may well be that the level of genetic diversity acquired through mutation is constant over time regardless of animal lifespan: smaller, shorter lived animals require more generations to acquire the same level of genetic diversity through mutations than larger ones.

    Note the emphasis on “through mutation” – I would guess “rate of natural selection” would depend on rate of environmental change and (inversely) lifespan (so if climate changes too fast for lifespan the species dies out).

    The idea that speciation is purely dependent on mutation makes sense too- after all separated colonies aren’t subject to speciation pressure through natural selection one way or the other.

    I think the interesting aspect of this work- if it turns out to be correct!- is that it highlights a subtle aspect of the interplay between natural selection, random mutation and speciation.

    Very interesting article by the way, thanks for that.

  • Well for cryin’ out loud…”natural selection VS chance mutations”.
    When natural selection occurs, it is “chance mutations” that are being selected in or out. If animal life remained genetically static and no mutation occurred, then natural selection would either select everything away into the void of nothingness or evolution would just be a circular process in which life endlessly ate its own tail.

  • Re: Germán Sterin, Hominid: Darwin’s two great insights were 1) variation is random, 2) as Steven Podvoll wrote, survival (and its obverse, extinction) is the means by which natural selection works.

    With regard to 1: random variation does not necessarily imply that speciation itself is random (as it requires a functional set of variations). Strict adaptionists tend to assume that for variation to reach the level of creating new species, they must be better adapted than their predecessors, but I am not aware of any proposed mechanism for speciation (e..g physical separation) that requires it, and this study is offering evidence that it is not so.

    With regard to 2: both the creation of new environmental niches and extinction provide an opportunity for some species to prosper, but it does not bring them in to existence.

    Darwin was well aware of the role of extinction in evolution (e.g. his wedge metaphor), and that it would be controversial in a society that believed in a benevolent creator. This study does not contradict either of Darwin’s great insights; it sheds some light on how these forces have played out in practice.

  • With all due respect, what “Darwin said” ( in any great detail) is as irrelevant to modern evolutionary biology as what Marx said is to modern economics.  These broad and suggestive philosophical ideas are now 150 years old, and so thoroughly modified in practice from their original form that their originators would not recongnize them. Becoming involved in a scriptural reading of either is to miss the point of science altogether. IT MOVES ON. The point at issue here is a long-standing debate going back not quite 100 years – and that is the force of mutation in population genetics and evolution. As Ronald Fisher said back in 1930 “evolution and natural selection are not synonymous.” Natural selection may work, but the neutral theory (Kimura) and Lynch’s notions of the evolution of genome architecture also give persuasive accounts of the reasons for change, including speciation.This tree of life is an interesting development in the debate, but does not and will not “answer the question” in any definitive way.

  • Anyone else notice how the patterns look like fractals in the troughs and peaks of the three trees of life? I think this should be studied on an information theory level, I would be interested to see what it reveals if anything?

  • Like jbspry, I find it very difficult to understand how causality can be ignored, one must have a mutation before one can have a selection, surely.
    Without having spent a lot of time on the subject, I always just assumed that everyone already agreed that mutations first, selection follows….and adaptation is merely “little mutations”, some advantageous, others not.
    So what is it I am missing that Blair Hedges is being “provocative” in proposing that adaptation had little to do with speciation?

  • I don’t understand all the hoopla. If you start with the fact that mutations are random: if a mutation gives the mutated group a slightly better chance of surviving one season than the original group, what happens over 1,000,000 seasons?

    Lets run the numbers to see how this works out. In very simple terms, and for the moment ignoring the always present problem of limited resources, if the chance of a group surviving one season is X, then the chance of that group to survive 1,000,000 seasons is X raised to 1,000,000. If we normalize the chance of survival for the initial group to be 100%, or 1, 1 raised to 1,000,000 is still 1.

    If a chance mutation gives the second group a 0.001% greater chance of surviving for a season, then in simplistic terms the chance that they have survived for 1,000,000 seasons is 1.00001 raised to one million or 22,025. 22,025:1 is a huge difference in the chance of survival over 1,000,000 seasons.

    Since resources are always limited, with such a huge difference in the chance for survival over the long run will result in the modified group slowly crowding out the original group. Its not that the unmodified group will necessarily be killed off, rather they would have moved into a different niche with less competition for survival.

    Keep in mind that evolution is not some kind of effect that drives the survival of a creature. It is just a term we use to describe the history of what actually took place. The same is true about the statement “Survival of the fittest”. It is simply a statement made when looking back to see what has already happened. It is not a plan for competition of future generations.

    There is no directed plan or scheme to survive in any particular environmental niche; there is just the random chance that having a beneficial mutation gives the mutated group a slightly greater chance to survive in the long run in that niche. It all boils down to random mutations causing a slight increase in survival exponentiated over an enormous number of years.

  • I still don’t get it.
    Since every recorded mutation involves the corruption or loss of genetic info, speciation has to be the result of loss of info to be anything but say a chihuahua. Where as the original (mutt)dog could
    produce a variety of offspring depending on surroundings.
    There is however evidence that some speciation (mutations) are not random but pre programmed responses
    to severe environmental changes.

  • The issue here is one of cause and effect. The observations may be validated but that does not say what is causing this. Maybe the chart is simply measuring the average and range of geologically-induced environmental changes in ecosystems. We know that climactic and environmental conditions change over time and there is no reason to think that these changes cannot provide the driving mechanism to improve reproductive fitness to particular random mutations. Remember, mutations that do not lead to speciation (because they do not improve fitness) do not appear in the tree. Like we say, its the winners who get to write history.

  • I find it a little bizarre that so much effort could go into a project like this and yet it completely ignores symbiosis, hybridisation, and other processes that result in new species. Every macro-organism is involved in symbiosis at some level. Be it our gut bacteria or mycorrhizal fungi. It’s decades since Lynn Margulis pointed out the role of symbiosis in the emergence of new species, and since then a great deal of research has been published on the idea. She was also critical of the representation of evolution as a tree.

    When we know for a fact that humans themselves hybridised, what is the point of a phylogenetic tree? Are we still in denial about our genetic heritage? The “tree” takes no account of recombination, our heritage is not one of binary splits leading to unique individuals. Yes, there is divergence, but there is also convergence. Something like 20% of plants and 10% of fishes regularly hybridise and produce viable offspring. Clearly hominids were active hybridisers as well.

    The idea that a tree structure can represent the process of evolution is outdated and inaccurate. Evolution has a network structure. And it’s not just me saying it.



  • To assert that “species arise because of chance mutations — not natural selection” seems meaningless to me! Natural selection operates on chance mutations to select what survives. Both are independently necessary. Like saying ‘the vehicle moves because of the wheels – not the engine’.

  • Broadly correct. But there are many examples of “sibling species” (especes jumelles, Geschwisterarten) that appear anatomically and physiologically identical, but are reproductively isolated from each other, so meeting the definition as species. Thus species “wheels” seem to operate before the “engine” of natural selection gets into the act.

  • One piece of the argument which seems inexplicably neglected is the concept of genetic drift. Evolution is commonly defined by biologists from various disciplines as simply a change in allele frequencies over time. The argument regarding the honeycreepers and the situation in which their small, isolated populations diverged without the changes necessarily being driven by selection is actually a great example of the process of genetic drift. This phenomenon is one of the situations that anybody who didn’t sleep through their entire undergrad genetics course should be rather familiar with and recognize as having the ability to drive changes in allele frequency over time. I have not yet had a chance to read the actual paper, but the argument presented in the article above is incorrectly implying that a finding of evolution in the absence of clear natural selection is somehow a problem for our current views on speciation.

  • This has been known in population genomics for years! From the 4 main “motors” in speciation (natural selection, genetic drift, mutation, migration) natural selection is the fastest but weakest in terms of changing a population’s heterozygocity (and consequently raising new species).
    The presence of a selective challenge can make new species in a few generations but once that challenge has been surpassed the probability of getting new species comes back to it’s regular pace.
    In the long run, genetic drift and mutations are more than enough to increase biodiversity. That’s why, evolution “is a ladder” rather than a slope. Most of the comments are centered in discussing semantics instead of looking at what truly rules biodiversity (the probability of gaining heterozygocity)

  • I think we can progress a lot better and faster in evolution biology science if vast majority of science folks stop treating Darwin as some unquestionable god, including the comments here.
    Anyway, in the graph above there also appears to be a convergence of Arthropods and Vertebrates at near consistent intervals of 5 million years.

  • Emily, why do you always use the word develop in place of evolve? Species do not develop, they evolve. This is not the first article where you do this, and I urge you to get the nomenclature right.

  • Scientists have been building phylogenetic trees for 2 decades? What? Phylogenetic trees have been built since 1901 when British zoologist Peter Mitchell did it for birds (Schuh, 2000; Folinsbee, 2007), and then by British-Australian zoologist and geologist Robert Tillyard for insects (Tillyard, 1921), and W. Zimmermann for plants in 1943 (Schuh, 2000). The method was elaborated by German zoologist Willi Hennig in 1950 in the classic work, Grundzuge einer Theorie der phylogenetischen Systematik (''characteristics of a theory of phylogenetic systematics''), which was translated in '66 as Phylogenetic Systematics by D.D. Davis and Rainer Zangerl. And Cladistics Magazine has been around since '85.

Comments are closed.