Published online 11 October 2004 | Nature | doi:10.1038/news041011-2


Molecular clock tied to fossil record

Evolutionary trees may finally provide answers everyone can agree on.

Studies of DNA and the fossil record often give wildly differing estimates of when different groups of organisms diverged.Studies of DNA and the fossil record often give wildly differing estimates of when different groups of organisms diverged.

A way to date prehistoric events using molecules from living creatures is finally becoming precise enough to be useful. A team of scientists has improved on a 'molecular clock' system that can fix a rough date for the last common ancestor of two separate species.

Determining when two branches of living things parted company is not an easy task. For more recent events, or for bigger animals, there might be a fossil record: a set of bones that represent a last common ancestor or first separate species. These can then be dated by the rocks around them or by carbon dating.

But what if there is no fossil record? Several decades ago it was first proposed that if DNA accumulates mutations at a constant rate, then you should be able to measure the differences between the DNA from two present-day species and extrapolate back to a time when the DNA was identical - to when one species became two.

There were problems, though. The rate of change was first calculated for vertebrates, using fossil vertebrates to calibrate the scale, but it then turned out that evolution progresses at a different rate in different groups of organisms, so the vertebrate rate gave wacky dates when applied to anything else. And it was not clear how constant the rate of mutation was over time for any group.

Worse, dates given by the molecular clock consistently disagreed with the fossil record, tending to give estimates that were much older, by as much as several hundred million years.

Setting the clock

Emmanuel Douzery, a molecular phylogenist at the University of Montpellier in France, and his team have now created a 'relaxed' molecular clock that allows for different rates of mutation in different groups of species.

They used 36 diverse living species to create an evolutionary tree that included all the major groups of organisms, then tied it to the fossil record at six points. That is, for six ancient creatures, the researchers made sure that the dates stayed within a range determined by conventional dating of the fossils. The rest of the tree would have to fit with these six knowns.

Then the researchers looked at more than 100 proteins in each of the 36 living species. These proteins are all essential molecules for life and their amino-acid sequences remain remarkably unchanged during evolution. But, over long periods of time, small mutations in the organisms' DNA make the proteins' sequences drift apart. The researchers used the differences between the species to estimate how fast the mutation rate is in each group.

Finally, they used a computer model to fit the different mutation rates to their tree, together with the dates from the six fossils.

The resulting family tree, published in the Proceedings of the National Academy of Sciences1 took months to assemble, and overall it fits well with the fossil record. Here and there, species show up in the tree before they do in the fossil record. But this makes sense, Douzery says, because a fossil may have formed quite a bit later than the first appearance of that species.

Debashish Bhattacharya, who works on comparative genomics at the University of Iowa, in Iowa City, has also produced a 'relaxed molecular clock', as have several others, although Douzery's tree is the most exhaustive so far. Bhattacharya is sceptical about Douzery's dates for one mysterious red alga - the tree says that it appeared after the fossil - but pleased with the work overall. "The real strength of the analysis is they have a large data set," he says.

Bhattacharya believes that molecular clocks will soon be a useful tool for researchers. "The work is all going in the same direction. I do think it is going to crystallize in a couple more papers," he says. 

  • References

    1. Douzery E. J. P., et al. PNAS, (2004).