[M.D. Holloway]

After 10,000 generations, it became apparent that not all the flasks were on the same trajectory. Though the cells in all the flasks became larger, each population differed in the maximum size the cells reached. The populations also differed in how much fitter they were than their ancestors, when the researchers grew them in direct competition. The “experiment demonstrates the crucial role of chance events (historical accidents) in adaptive evolution,” Lenski and his colleague Michael Travisano wrote in a 1994 paper.

The experiment has progressed, and several of the flasks now contain mutater strains, bacteria that have defects in their DNA replication system. Such defects make mistakes more likely to happen every time those bacterial strains copy their DNA to divide. Sometimes a mistake can have lethal consequences, damaging a gene crucial for survival. But other times coloring a bit outside the lines creates opportunity for advancement.

Even within a given flask, some bacteria take slightly different paths. One flask now contains two separate strains — one that evolved to make large colonies when grown on petri dishes, and one that makes small colonies. The large- and small-colony strains have coexisted for more than 12,000 generations. The large-colony producers are much better at using glucose so they grow quickly, but they make by-products that the small-colony producers can eat. Each of the populations, both large and small, have improved their ability to use glucose over the generations. By at least one measure, the two populations could constitute separate species, Lenski and his colleagues proposed in 2005 in the Journal of Molecular Evolution.

Still, it seems that Conway Morris was basically right: Though the details were different, replaying evolution in a dozen flasks produced very similar outcomes in each. But then something completely unexpected happened.

After about 31,500 generations, glucose-eating bacteria in one of the flasks suddenly developed the ability to eat a chemical called citrate, something no other E. coli do, the researchers reported last June in the Proceedings of the National Academy of Sciences (SN Online: 6/2/08).

“They’ve been eating the main course for thousands of generations,” Lenski says. “They didn’t realize that there was a dessert tray around the corner.”

The switch was clearly a radical change of destination for the bacteria. The inability to eat citrate is a biochemical hallmark of the E. coli species, so by some definitions, the citrate eaters in that flask are no longer E. coli, but a different species.

But a single change did not a citrate eater make. The researchers found that the bacteria went through a series of steps before evolving the ability to use citrate. One initial mutation happened at least 11,000 generations before the citrate eaters appeared. Lenski and his crew don’t yet know which specific DNA changes led to citrate use, but the researchers have enough evidence to say that the ability to use citrate is dependent, or contingent, upon those earlier changes. And even the bacteria that have undergone those initial changes are still not guaranteed to find the dessert cart. Blount tested 40 trillion bacteria from earlier generations to see if any could evolve the ability to eat citrate. Fewer than one in a trillion could.

The profound difference between the citrate eaters and the other 11 strains, as well as the dependence of the citrate change on earlier mutations, seems to suggest that Gould was also right: Replaying evolution will result in some surprise endings. “The long-term evolution experiment with E. coli provides some of the best evidence for both Conway