This series of posts is intended as a basic introduction to the science of evolution for non-specialists. You can see the introduction to this series here. In this post we explore the “Long Term Evolution Experiment” – perhaps one of the best experiments for comparing the effects of contingency and convergence in a controlled environment.
In the last post in this series, we explored two competing interpretations of earth’s evolutionary history as championed by two eminent paleontologists. As we saw, Simon Conway Morris favors a view that evolution is largely repeatable (i.e. convergent) – and thus not dominated by chance. The late Stephen J. Gould, in contrast, famously argued that “replaying the tape of life” would lead to dramatically different results, given his view that chance (i.e. contingency) has the upper hand in evolution.
Of course, the most rigorous way to test Gould’s conjecture (and Conway Morris’ objection to it) would be to do the experiment: rewind the “tape” to the Cambrian period, allow it to go forward multiple times, and note the pattern that emerges over numerous trials. As fascinating as such an experiment would be, this is (of course) not even remotely feasible. Scientists have to settle for far more humble approaches in hopes of addressing these issues.
The LTEE: biosphere in a bottle
One of the more ambitious experiments to date that sheds light on the roles of contingency and convergence in evolution is the “Long Term Evolution Experiment” (LTEE). This experiment, which began in 1988 and continues to this day, is remarkably simple – it tracks the evolution of twelve bacteria populations and compares them over time. Housed in the laboratory of microbiologist Richard Lenski, the LTEE began with twelve cultures of the bacterium E. coli derived from a single cell. From this identical starting point, the twelve cultures have been grown separately, effectively playing twelve “tapes” from an identical starting point.
The mechanics of the experiment are simple: each day, the cultures are provided with new nutrient broth to grow in. The next day, a fraction of the previous day’s culture is transferred to new broth – and so on. Every so often, samples of the cultures are frozen in a way that puts the bacteria into long-term stasis and stored – providing something analogous to a fossil record for each strain, but with the advantage that every “fossil” can be revived and studied. The twelve strains are given the exact same conditions (the same broth, the same temperature, et cetera) and the experiment has gone on, day in and day out, since 1988.
Lenski has described his early expectations for the LTEE (PDF):
When I began this experiment, I thought big differences among the 12 lines would soon be apparent. The random occurrence of mutations meant that some populations would get lucky by generating a beneficial mutation (and one that survived the daily dilutions) sooner than others. And just as in a game, different early moves—mutations—might open some doors while closing others. Some populations might get stuck with beneficial mutations that ultimately led nowhere, while others would follow paths that had long-term potential.
In other words, Lenski expected contingency to be the main player, shaping the trajectories of the lines early on in ways that would lead to their significant divergence. Much like Stephen J. Gould’s famous thought experiment, Lenski expected that these twelve simultaneous “tapes of life” would play out in markedly different ways, with chance as the predominant force shaping their evolution.
The experimental results, however, did not match Lenski’s initial prediction. Though many of the precise mutations were unique to a given bacterial line (and thus contingent, as one would expect for random mutations), the overall evolution of the twelve lines was strikingly similar. As Lenski would note, it was this overall pattern ofconvergence that stood out against the backdrop of contingency:
To my surprise, evolution was pretty repeatable. All 12 populations improved quickly early on, then more slowly as the generations ticked by. Despite substantial fitness gains compared to the common ancestor, the performance of the evolved lines relative to each other hardly diverged. As we looked for other changes—and the “we” grew as outstanding students and collaborators put their brains and hands to work on this experiment—the generations flew by. We observed changes in the size and shape of the bacterial cells, in their food preferences, and in their genes. Although the lineages certainly diverged in many details, I was struck by the parallel trajectories of their evolution, with similar changes in so many phenotypic traits and even gene sequences that we examined.
Contingency strikes back
At this point in the experiment, it certainly seemed like convergence was the order of the day – the twelve lines were divergent in the details of their mutations, yes, but the overall pattern, driven by selection, was that of boringly repetitive convergence. In their identical environments, the lines were shaped to very similar outcomes despite the effects of chance. The tapes had been replayed, and the outcomes were predominantly convergent. Case closed.
Or was it?
It was at this point in the LTEE that something dramatic would happen within just one of the twelve cultures – something so shocking that Lenski and his colleagues thought one culture had been accidently contaminated. Suddenly, one culture was using a food source it had never been able to use before: citrate.
Now E. coli cannot, in the presence of oxygen, normally use the compound citrate as a food source. The nutrient broth used in the LTEE has a significant amount of citrate in it, interestingly enough – but it is there solely as an inexpensive means to buffer the pH of the solution. The broth recipe is an old one, hailing back to a time when microbiologists did things more simply and cheaply. Since the E. coli lines of the LTEE are grown with oxygen present, the citrate in the broth was useless to them – until, by chance, one of the twelve cultures hit upon a way to use it. To return to Lenski’s description,
For 15 years, billions of mutations were tested in every population, but none produced a cell that could exploit this opening. It was as though the bacteria ate dinner and went straight to bed, without realizing a dessert was there waiting for them. But in 2003, a mutant tasted the forbidden fruit. And it was good, very good. The descendants of that mutant rose to dominance owing to their access to that second course. At first, I thought this flask had been contaminated by some other species that consumed citrate. However, DNA tests showed the citrate-eating cells were descendants of the E. coli ancestor used to start the experiment.
Later work would reveal that this striking change was highly contingent in nature, requiring numerous mutations to assemble over tens of thousands of generations. After years of convergence, the most significant change observed in the LTEE would be shown to be a chance, one-off event unlikely to be easily repeated in the other lines – at least in the near future.
Of course, “in the near future” is the key caveat. Will some of the other 11 lines someday find a way to exploit this resource in their environment? Only time will tell. Until then, both Conway Morris and Gould can find support for their views in the LTEE, and the larger question of convergence and contingency remains open – as one would expect for a frontier area.
In the next (and final) section of this series, we will return to the larger question raised by the Gould / Conway Morris debate: does evolution, as a science, preclude holding a purposeful view of life on earth?