Experimental evolution meets genomics
Richard Lenski and colleagues have been monitoring evolution of the bacterium Escherichia coli in his laboratory for 40,000 generations. Their latest paper, "Genome evolution and adaptation in a long-term experiment with Escherichia coli" was recently published in Nature.
One nice thing about E. coli is that they can freeze samples of their evolving populations every few thousand generations, for later analysis. So they were able to compare the fitness of different generations by competing each against a thawed ancestor. They also found the complete DNA sequence for many of these strains....
This wouldn't have been possible when their evolution experiment began. It would have been impossibly expensive even a few years ago, but DNA sequencing has been getting cheaper, as Richard Dawkins predicted in his essay, "Son of Moore's Law."
They found that their bacterial populations accumulated genetic changes at a fairly constant rate over the first 20,000 generations. This is what you would expect, if they were randomly accumulating "neutral" mutations, with no effect on fitness. But random neutral mutations would include "synonymous" mutations that change DNA sequence without changing the corresponding protein, whereas they found mostly protein-changing mutations. Those should have some real effect and apparently a positive one.
Over the same period, fitness increased relative to the ancestral strain. But the increase in fitness showed a different pattern from total changes in DNA sequence. While DNA sequence changes accumulated at a fairly constant rate, fitness increased very rapidly over the first 1000 generations or so. Since then, fitness has continued to increase, but much more slowly. This later increase could be roughly linear, but there's enough noise in their data that it's hard to be sure. One possible explanation for this pattern is that the early genetic changes had wide-ranging effects, even if the DNA-level changes were small. This is what you would expect if the changes involved regulatory systems, for example. Those early changes were clearly positive, on balance, but they may have had some negative side effects. The slow-but-steady improvements since then may involve a large number of genes, each with a small beneficial effect, possibly reducing some of those negative side-effects.
Fellow microbial evolutionary biologist Paul Rainey has a commentary on the paper in the same issue. Rainey himself has published a very interesting paper even more recently, which Will Ratcliff has promised to write about.