Evolutionary rescue

Can evolution save us from the brink of collapse?

Andy Gonzalez thinks so. Gonzalez, an ecologist from McGill University, gave an entertaining seminar to the department last Thursday on the subject. His research group works on the causes and consequences of biodiversity loss, using mathematical models and controlled experiments to investigate how environmental change might affect populations.

Gonzalez teamed up with McGill’s Graham Bell, who is known for using simple systems like yeast and algae to tinker with the evolutionary process through experimental evolution. Gonzalez describes their third collaborator on this project as “painful to work with, but once things were up and running [he was] amazing.” He was, in fact, a robot.

Robotic liquid handling system, more accurately.

The robot allowed Gonzalez and Bell to design an experiment that would have been too big and cruel to impose on graduate students, and test evolutionary rescue theory using populations of baker’s yeast1. This is the same species used for fermentation in baking and brewing.

Evolutionary rescue, or the idea that adaptation can help a species recover from environmental catastrophe, is not new. The theory is well-established: given a large enough population, with a decent supply of genetic variation to fuel natural selection, adaptive recovery should be possible. The hard part is sorting out the details. Although we have a few examples of evolutionary rescue occurring in microbes (think of the evolution of antibiotic resistance in disease-causing bacteria), most of the organisms we have looked at so far fail to adapt in response to anthropogenic change.

By diluting, incubating and transporting the experimental “populations” – wells of liquid containing the yeast cells – the robot was able to keep track of hundreds of replicate populations over many weeks. It performed all the necessary lab duties right on schedule, allowing precise control of the environmental conditions. It even measured the size of the evolving populations objectively, using a spectrophotometer to record the optical density of yeast cells in solution.

The robot was not perfect. Mistakes happened. Gonzalez watched it pick up a set of his yeast samples and throw them into the garbage, “very slowly and deliberately – and it was behind bulletproof glass!”

Ultimately, though, Bell and Gonzelez were able to do something no one else had done before: test the theory behind evolutionary rescue, while controlling for the level of environmental stress and population size. They used high salt concentration to create an environment that would be maladaptive for the yeast cells at first. The patterns they saw in response were exactly what theory predicted. The yeast populations followed a classic U-shaped curve, with a geometric decrease in population size down to a threshold where extinction will occur unless evolutionary rescue takes over (about 10-30 individuals per population). This was followed by a rapid geometric increase when recovery occurred.

Gonzelez and Bell used this experiment to calculate the probability of recovery depending on the level of environmental stress. The results may be broadly applicable. To ensure a 50% chance that a species will recover from stress, you need an initial population size of about 500 individuals.

The robot helped Bell and Gonzalez sort out the details of this “race between demography and adaptive evolution”1,2. Their work shows that rescue, where evolution wins, is possible. In their experiment, the genetic variation for salt tolerance may have already been present in the initial populations. It is not clear whether this is also true of many real world organisms we would like to save. According to Gonzalez, robotic technology makes the kind of high-throughput experimental evolution that we need to address these questions possible.

More recently, the McGill robot has been working with yeast to test models of population range shifts under environmental stress. The traditional hypothesis is that a population will contract towards the centre where it is most abundant, shrinking at the margins where the organism is less adapted to the conditions. Gonzalez is testing the intriguing idea that populations might contract towards the edges instead, collapsing in the centre3. This work is ongoing, but the implications could be critical in the face of global climate change. Edge populations, once thought to be marginal, could have the greatest potential response to selection. They might be the key to promoting evolutionary rescue in the field.

What about rescuing ourselves? Bell and Gonzalez have validated models of evolutionary rescue, but a new paper in Nature has an even bolder proposition: that ecological models might save us from our own greed.

Robert May, an Oxford ecologist, and economist Andrew Haldane claim that mathematical models originally intended to describe biological phenomena might help explain what caused the global financial collapse of 2007-20084. Although they acknowledge major differences between living communities and financial systems – such as the fact that “financial ecosystems” promote the “survival of the fattest rather than the fittest” – Haldane and May argue that their ecosystem model can inform policy intended to minimize risk. A hybrid, their model combines aspects of predator-prey food webs with epidemiological theory of disease transmission. The authors demonstrate how this could be used to test the robustness of the financial system to collapse under different scenarios.

And the potential insights? For one, Haldane and May argue that the way modern banks are connected – with bigger ones disproportionately linked to smaller ones – could promote the rapid transmission of financial shock-waves.

Skeptics like Neil Johnson argue that it would be dangerous to infer global policy from Haldane and May’s simplified models5. Economist Thomas Lux is more enthusiastic in his commentary, but he seems to misunderstand one of Haldane and May’s biological analogies. The Lehman Brothers default of 2008 was not like a ” ‘super-spreader’ disease”, as Lux has it5. The company itself was a super-spreader node, transmitting damage by its sheer size and connectivity4. From Haldane and May:

There has been a spectacular rise in the size and concentration of the financial system over the past two decades, with the rapid emergence of ‘super-spreader institutions’ too big, connected or important to fail.

The issue here is the structure of the host network. It is a precarious scenario for the spread of any infectious disease.

Caution may be warranted. At the same time, I find it hard to see why this is necessarily any less defensible than many of the classical models in economics. From Thomas Malthus to game theory, biology has borrowed much from economic theory. Maybe it is time for evolution to aid a new kind of recovery.


  1. Bell, G. and Gonzalez, A. 2009. Ecology Letters 12: 942-948.
  2. Maynard Smith, J. 1989. Philosophical Transactions of the Royal Society Series B 325: 241-252.
  3. Channell, R. and Lomolino, M. V. 2000. Nature 403: 84-86.
  4. Haldane, A. G. and May, R. M. 2011. Nature 469: 351-355.
  5. Johnson, N. and Lux, T. 2011. Nature 469: 302-303.