This article is reproduced with the permission of New Scientist for exclusive use by Nova users.

Evolution: hacking back the tree of life
13 June 2007 news service
Laura Spinney
Some of our brains are missing

If you want to know how all living things are related, don't bother looking in any textbook that's more than a few years old. Chances are that the tree of life you find there will be wrong. Since they began delving into DNA, biologists have been finding that organisms with features that look alike are often not as closely related as they had thought. These are turbulent times in the world of phylogeny, yet there has been one rule that evolutionary biologists felt they could cling to: the amount of complexity in the living world has always been on the increase. Now even that is in doubt.

While nobody disagrees that there has been a general trend towards complexity - humans are indisputably more complicated than amoebas - recent findings suggest that some of our very early ancestors were far more sophisticated than we have given them credit for. If so, then much of that precocious complexity has been lost by subsequent generations as they evolved into new species. "The whole concept of a gradualist tree, with one thing branching off after another and the last to branch off, the vertebrates, being the most complex, is wrong," says Detlev Arendt, an evolutionary and developmental biologist at the European Molecular Biology Laboratory in Heidelberg, Germany.

The idea of loss in evolution is not new. We know that snakes lost their legs, as did whales, and that our own ancestors lost body hair. However, the latest evidence suggests that the extent of loss might have been seriously underestimated. Some evolutionary biologists now suggest that loss - at every level, from genes and types of cells to whole anatomical features and life stages - is the key to understanding evolution and the relatedness of living things. Proponents of this idea argue that classical phylogeny has been built on rotten foundations, and tinkering with it will not put it right. Instead, they say, we need to rethink the process of evolution itself.

It is not hard to see how the mistake might have happened. In the past, the tree of life was constructed on the basis of similarity of morphological features. The more similar two species looked, the more closely related they were thought to be. But looks can be deceptive. This became abundantly clear more than a decade ago, when molecular biologists began comparing small numbers of genes from various organisms and found that many species were not what they appeared. Hippos, for example, were once thought to be the kissing cousins of pigs, but genetic evidence revealed their closest living relatives to be the cetaceans (whales, dolphins and porpoises).

Without the insights of molecular analysis, traditional morphologists also had no way of knowing whether a particular feature had been lost in a given lineage, or had never been there in the first place. In line with the idea that things evolve towards increasing complexity, they tended to assume the latter, sometimes quite incorrectly. Take the sea squirt. Its larva swims around looking like a tadpole, with a nerve cord along its back, gill slits for feeding and a tail - all classic features of chordates, the large group of animals with backbones that includes us. Then, however, it stands on its head and turns into a sack of jelly, having first digested what it had of a brain. The adult looks suspiciously like a plant. For a long time it was considered to be one of the most primitive chordates because of its simple adult form - about as far from vertebrates as it was possible to get. In between were myriad other groups, including the lancelets - fish-like animals that hang on to their nerve cords into adulthood. Then molecular studies revealed that sea squirts are genetically closer to us than are lancelets, and the tree had to be reshuffled.

In recent years, genetic analysis has forced biologists to consider the possibility that organisms such as the sea squirt might have lost some of the complexity of their ancestors. Yet even now, few recognise the full implications of loss as a key player in evolution. The entire tree of life has been built on the assumption that evolution entails increasing complexity. So, for example, if two groups of animals were considered close because both had a particular prominent feature, then someone discovered a third, intermediate line that lacked that feature but shared many other aspects of the two groups, traditional phylogenists would conclude that the feature had arisen independently in the two outlying groups, by a process known as convergent evolution. They often did not even consider the alternative explanation: that the feature in question had evolved just once in an ancestor of all three groups, and had subsequently been lost in the intermediate one. Now a handful of molecular biologists are considering that possibility.

Instead of simply looking to see whether two species share certain genes, the new approach involves taking the "molecular fingerprint" of different types of cells. It identifies the unique combination of transcription factors - molecules that control which of a cell's genes are turned on and when - that specify the make-up of a cell, including the molecular signals it transmits and receives. If two groups of organisms share the same type of cells, with the same molecular fingerprint, giving rise to similar features in both, then it is extremely unlikely that these features evolved twice. So any intermediate groups of organisms that lack that feature would most likely have lost it during the course of evolution. Only now, with the ability to explore at the molecular level how morphological features have been lost, gained and modified over time, is the true extent of evolutionary loss coming to light.

Arendt's convictions about the vast scale of this loss are based on his molecular fingerprinting studies of a tiny annelid worm called Platynereis dumerilii. It is an unprepossessing animal that lives in tubes stuck to rocks in shallow seas, bathed in a nutritious blanket of algae and reproducing according to the tides and the lunar cycle. "We think that it has always lived in this ecological niche," he says, "and that this might resemble the environment of the common ancestor [of all animals that are symmetrical along the axis from head to tail]". This enormous group, called bilaterians, encompasses all vertebrates and most invertebrates; the descendants of a long-extinct creature known as Urbilateria that lived between half a billion and a billion years ago. No fossils of this species have been found, but as Platynereis is thought to have occupied the same niche as Urbilateria, Arendt suspects it might also have retained some of the mysterious ancestor's features.

Brainy ancestors

What Arendt's group has discovered about the brain of this lowly worm is intriguing. Within the animal kingdom, the simplest and most evolutionarily ancient type of nervous system is a diffuse neural net. Sea anemones and corals, for example, have this system, in which a single type of neuron is distributed throughout the animal. More recently evolved species have a central nervous system (CNS), with specialised sensory and motor neurons clumped together into a nerve cord and brain.

A CNS is found in all vertebrates and some invertebrates, including Platynereis and two of biology's supermodels, the fruit fly Drosophila melanogaster and the nematode worm Caenorhabditis elegans, but there are obvious differences between the vertebrate and invertebrate CNS. Vertebrates have a spinal cord at the back, while invertebrates usually have a chain of neuronal clusters or ganglia, connected like a rope ladder, in their belly. This led morphologists to think that Urbilateria had a diffuse neural net and that centralisation arose separately by convergent evolution in the different lines after they split. Arendt believes they were wrong.

Earlier this year, his group reported that Platynereis neurons share molecular fingerprints with vertebrate neurons during development (Cell, vol 129, p 277). For example, genes known to be important in patterning the vertebrate CNS also divide the worm's nervous system into domains. What's more, domains with corresponding gene expression patterns give rise to the same types of neurons in both. Arendt concludes that Platynereis and vertebrates both inherited their CNS from Urbilateria. The reason they take a different form today, he suggests, is that when early vertebrates began swimming freely, "front" and "back" lost their significance and the animals simply inverted the two. As the rope ladder nervous system became enclosed in the neural tube characteristic of vertebrates, the ancestral mouth was trapped inside. It is still detectable there, Arendt says. Using the molecular fingerprinting technique he has been able to find this obsolete mouth within the vertebrate brain. "Its position is very clear," he says, "It's behind the hypothalamus."

If Arendt is correct, then the ancestral CNS was lost completely in two major animal groups: the echinoderms (starfish, sea urchins and the like) and hemichordates (acorn worms and other worm-like marine organisms). Both of these are sister phyla to chordates (see Diagram), yet their members have lost their brains and instead have diffuse neural nets. The same seems to be true of various molluscs, brachiopods, phoronids and bryozoans that have evolved to be sedentary filter-feeders. "If you just sit around your entire life you don't need much of a sensory integration centre coupled to a locomotor nerve cord," says Arendt.

Not everyone agrees, however. In 2003, Chris Lowe from the University of Chicago and colleagues compared the genes expressed in the development of the acorn worm and vertebrate nervous systems. "We showed that the exact same genes are involved in patterning a nerve net as in patterning a CNS," he says. "So our argument is that you cannot use these genes as really solid markers of a CNS." Given the scarcity of comparative molecular data so far, Lowe thinks it is too early to rule out convergent evolution in annelids and vertebrates.

While controversy continues to rage over convergent evolution versus loss, it has emerged that Urbilateria is not the only very early animal ancestor that was more complex than some of its descendants. David Miller of James Cook University in Queensland, Australia, studies the coral genus Acropora, the main reef-building corals of the Indo-Pacific region. Acropora belongs to the phylum of cnidarians, which are thought to have branched off after Urmetazoa, the common ancestor of all animals, but before Urbilateria. Yet Miller is uncovering surprising genetic complexity in Acropora.

For example, it has a version of a gene that was thought to be exclusive to vertebrates as it is involved in the vertebrate immune system, which works by remembering past threats and adapting its response to them. "All the textbooks tell you that adaptive immunity is a specific characteristic of vertebrates," Miller says, "yet at least one of the proteins is clearly present in our animal."

Miller's findings are intriguing, but more work is needed to pin down the origins of adaptive immunity. The dangers of jumping to conclusions about early evolution followed by loss on the basis of limited genetic information are highlighted by work on body segmentation. Many bilaterians have bodies made up of repeating anatomical units, and the discovery of certain similarities between the developmental genes that determine segmentation in Platynereis and in insects suggested a common origin. Then Elaine Seaver at the University of Hawaii's Kewalo Marine Lab in Honolulu used molecular fingerprinting to show that almost none of the genes involved in insect or vertebrate segmentation are deployed in the same place at the same time in developing annelids. "The evidence is accumulating that segmentation arose at least three times independently," says Seaver's colleague Mark Martindale.

Now that the spectre of loss has been raised, however, proponents of the new model see it everywhere - everywhere, at least, where animals have evolved to occupy niches in which their pre-existing complexity might be superfluous. Last month, Marcus Davis of the University of Chicago and colleagues reported that a species of paddlefish shows patterns of gene expression during development that were previously thought to be exclusive to land-living vertebrates - in other words, those with limbs. This paddlefish is the living species that most closely resembles the bony fish of the Palaeozoic era, which lived more than 250 million years ago. Davis concludes that primitive bony fish may have had something like limbs, which were lost in their descendants (Nature, vol 447, p 473).

Parasitism is another potential driver of simplification. When one living organism colonises another, it may discard features it could not have survived without as a free-living creature - features that gave it mobility and the ability to seek out food, for instance. Max Telford from University College London gives the example of a genus of barnacle called Sacculina. Barnacles are crustaceans that don't look much like crustaceans because they are filter-feeders and sessile, meaning they anchor themselves to a substrate - often boats or piers. Sacculina doesn't look like a barnacle, let alone a crustacean. It parasitises crabs, producing an almost plant-like system of roots which invades the host tissue. It is known to be a barnacle only because it has a barnacle-like larval stage. "So barnacles have lost many crustacean characters because they are sessile, and Sacculina has gone even further because it is parasitic," says Telford.

Another driver of simplification might be miniaturisation. Rotifers are microscopic aquatic organisms with a feeding wheel - tufts of cilia around the mouth that waft food into it - and nothing that could strictly be called a brain. Arendt believes that they represent the larval stage of an animal that, on shrinking to fit its planktonic niche, discarded its adult body plan and developed no further, becoming sexually mature early. "This is one very efficient means of throwing out ancestral complexity and becoming secondarily simple, and I think it happens frequently," he says.

If loss is so common, the challenge now is to distinguish the organisms that were always simple from those that have evolved simplicity. Genetics will be an invaluable tool here, but it will take a lot more analysis and comparison between a wide range of species before a definitive tree of life emerges. The very genetic complexity of Acropora, for example, has led some to question its position in the tree, arguing that it may have evolved later than was thought - that it may in fact be a descendant of Urbilateria that became secondarily simple while retaining genes that were later incorporated into the vertebrate immune system. If evolutionary biologists today are to avoid the mistakes of their predecessors, they need to eliminate precisely that kind of circularity.

"Molecular biology is making real inroads into this, but it has not been easy to reconstruct events that happened over half a billion years ago," Martindale says. Still, the new phylogenists are more resolute than ever. "There can only be one true relationship of animals to one another," he says.

Laura Spinney is a writer based in London and Paris

From issue 2608 of New Scientist magazine, 13 June 2007, page 48-51

For the latest from New Scientiist visit

Academy disclaimer: We cannot guarantee the accuracy of information in external sites.