I think the importance of “path dependence” is highly overlooked in science. Anytime someone invokes an equation that is based on equilibrium or steady-states (which is often), they are implicitly saying that we can ignore all of the system’s past history. We just wait until everything settles down at some fixed state, which we can determine more easily based on balancing some first derivatives. However, many real situations have multiple equilibria, and the final state of the system depends very strongly on previous trajectory. A recent paper in Nature provides a clear example regarding antibiotic resistance in which bacteria colonies that are exposed to an drug gradually are much more likely to survive than if the drug is introduced suddenly. The explanation is that the trait of resistance at high concentrations of the antibiotic may require multiple mutations, but having exactly single mutation is NOT adaptive at either zero or high concentrations. [Since mutations that confer resistance are costly, they provide negative fitness if no antibiotic is around. And if a lot of antibiotic is around, having just one mutation usually isn’t enough] As a result, bacterial with just one mutation will only be selected for at intermediate concentrations. But, of course, you have to have one before you can have two. A simplified chart with the arrows pointing to states with higher fitness would be:
Low Concentration: Zero Mutations <— One Mutation <—- Two Mutations
Medium Concentration: Zero Mutations —> One Mutation <—- Two Mutations
High Concentration: Zero Mutations —> One Mutation —-> Two Mutations
So there is a “barrier” to having a single mutation that can only be overcome if the system spends time at intermediate concentrations. [There is a strong analogy here with the free energy landscape vs reaction coordinate in chemistry] This allows for the first, “potentiating mutation,” which will be essential when the antibiotic is increased to the high concentration. Since evolution cannot “look ahead,” it can only select for mutations that provide fitness in the current landscape. This is what prevents the first mutation from occurring before the antibiotic is applied (unless we want to include the possibility of random “genetic drift,” but this is would be much less frequent). The bottom line is that our common practice of putting antibiotics into everything – like soap and cleaning sprays – is worse than we thought, since it creates the prefect environment to “train” bacteria to become resistant. Also, this shows why the schedule of administering antibiotic pills is so important. If the concentration in the body is too low for a long time, if can also lead to the evolution of resistant populations.
The current experiment made use of the ability to freeze generations of bacteria for later DNA sequencing so that the complete evolutionary history could be traced. The researchers also used genetic engineering to create bacteria with specific combinations of mutations so that the relative fitness could be quantified.
In the Ancestor’s Tale, Richard Dawkins raises the possibility of using the traits (and DNA) of living species to learn about the past environments its ancestors lived in, since some to the adaptations will still be reflected in the progeny.