So what if we could replay evolution and see how different things would be? Evolutionary logic dictates that things could potentially be quite different since, many times, little random events can have important long term consequences. People that have worked designing evolutionary algorithms, like genetic algorithms (the only acceptable combination of design and evolution in a single sentence that I can think of) already know that multiple runs of the same simulation configuration can lead to different results. People working on evolutionary algorithms don’t normally allow for open ended evolution, meaning that they normally have an idea of the outcome they would like to get, which severely limits the number of possible evolutionary paths. Despite that, my experience with evolutionary algorithms tells me that different executions will produce different results through different evolutionary paths.
But one thing is theory or computing, and a different one is a colony of bacteria as, say, E. coli. Now, that would be an interesting experiment…if it wasn’t because it has already been done! ScienceNews_ reports about this experiment performed at Michigan State University’s Lenski’s Lab.
The work has been described at length in PNAS this week.
Lenski and colleagues collected samples of the bacteria of, initially, identical E.coli in 12 colonies, every 500 generations for 20 years (!!!). This represents 40000 generations of bacteria. Not entirely unexpectedly each colony followed a different path, in one case significantly deviating from the original type. In the 12th colony, bacteria eventually evolved the ability to use a nutrient that the regular kind can’t absorb. Observing this, the researchers took a number of samples of earlier generations and the more remote the starting point, the smaller the likelihood of the colony to re-evolve that ability. At some point, going back to restart evolution stopped yielding the bacteria capable of absorbing new nutrients which hinted the evolutionary period in which the crucial genetic innovation was introduced.
Of course it is not that every colony went in an entirely different way and some of the dynamics were rather predictable. Such as that bacterial colonies got predictably better at absorbing their usual nutrients and dividing faster. So given a non harsh environment and enough colonies I guess is not difficult to foresee some trends.
David, thanks for pointing out this wonderful paper. And I think it rather assuring that evolution is not predictable.
I agree, it’s not predictable…neither random. It’s clearly exciting!
What a beautiful experiment. You basically enter the lab with your son and tell him: “My son, one day, all this will be yours”. And you do hope that it will be, or the experiment can’t continue.
Today’s scientists often don’t have the patience for an experiment that takes 20 years as in this paper. And how do you get research funding today for something that isn’t finished before 2028?
Hey David, it is a nice experiment (but a tough project to sell to a grad student). The length of time reminds me of the thirty something year lizard evolution paper earlier this year in PNAS here – Herrel et al 105, 4792 – hope that link works, never tried to do it before! Just to show that even vertebrates can change given time and the right circumstances.
On another issue you raise the idea of predicting what would happen if we were to rerun the evolutionary clock – since so much of what we see is presumable stochastic. Dawkins discusses this a little in the last chapter of The Ancestors Tale (and probably elsewhere), albeit again at a larger scale. He suggests that if you were to start again (or, I guess, for that matter look on another similar planet) certain characteristics (such as the evolution of eyes or flight) are likely to occur since these have evolved independently many times, while other characteristics might simply not occur.
Massimo: I guess you don’t need your own son to continue your work, you can always adopt a number of graduate students. Hopefully more than one to decrease the odds of the guy/gal leaving everything to become a buddhist monkl
Martin: I agree with you. I don’t think I am aware of all potential sources of funding out there but I doubt that there is a body so trusting and enlightened as to generously fund a lab for 20 years expecting little before the end of the project.
Simon:Thanks for the link to the PNAS paper, will have to wait until I am back to the office (no access to PNAS articles from home). Of course if you have to study something more complicated than E.coli even 20 years look like not enough time!
The Ancestors Tale is one of the few books from Dawkins that I haven’t read (yet). About the stochasticity in evolution, I’d imagine that it could have an effect similar to that in weather forecasting. Short term predictions are possible but long term predictions are unreliable due to the amount of small stochastic events.
Yes but….., while it is not possible to predict details of where they will strike it can reasonably be said that there will be tornadoes over the mid-West next year (as well as rain in Dundee and hurricanes near Florida!) So while it might not be possible to say that a bipedal flying animal with feathers will evolve, it could be suggested that there is a high probability of flying animals of some sort.
Hi Simon, thanks for your comment.
I take your point and agree that even some long term outcomes can be predicted although the precise timing in that case would be out of the question.
A better knowledge of genotype to relevant-phenotypes mapping would be a start. This way it’d be possible to know how many paths to a particular phenotype there are and what is the likelihood of a phenotype to emerge. This coupled with a better understanding of the microenvironment, that selects for the phenotypes, should be enough to make predictions about what phenotypes and in what proportions would be reasonable to expect in a progressing tumour…
My goodness, the creationists didn’t like this paper! Check out the dialogue between Lenski and the owner of conservapedia.com regarding the paper and requests for the mutated strains: http://rationalwiki.com/wiki/Lenski_affair
Hi Nick,
Sorry for taking so long to reply. Great link, very entertaining. This Lenski wrote some good replies!