Carl Zimmer reports on an experiment from Richard Lenski's lab, reconstructing the events that led one lineage of bacteria to win out over another. Lenski has kept a population of evolving E. coli bacteria for upwards of 50,000 generations, freezing a sample of the bacteria every 500 generations. This allows them to thaw out past moments in the population, like reviving a fossil record of their current bacteria. They use these to run repeated trials of evolution, in one case seeing whether the ultimate success of one bacterial lineage was foreordained by its evolutionary pathway in contrast to its competitor lineage:
The eventual winners still consistently beat out the eventual losers, the researchers found. On average, they ended up growing 2.1 percent faster than their rivals. Their success, in other words, was not the result of good fortune. They were better prepared to make the most of beneficial mutations.
It's an example where the evolutionary history was path-dependent, but that the path itself is multiple steps deep. Early steps constrain the later ones, and may release adaptability or limit it. Simple in theory, but showing it in a replicable experimental context is pretty sweet.