Hello fellow BEACONites and interested members of the public, I’m Josh Nahum, a postdoctoral fellow, who was at the University of Washington during the early years of the Beacon Center, but now I’m doing research at Michigan State University (more info found here: http://beacon-center.org/blog/2012/03/29/introducing-beacon-distinguished-postdoctoral-fellow-joshua-nahum/). I do digital and microbial evolutionary experiments, and some of my previous work involved the evolution of prudence in E. coli was covered on this blog is the past (http://beacon-center.org/blog/2011/08/29/beacon-researchers-at-work-survival-of-the-weakest-when-doing-poorly-does-best/). Similar to my previous post, today I’m going to talk about how spatial structure affects evolving populations.
All living things rely on evolution by natural selection to become better adapted to their environment. The process of adaptation requires mutations (changes in DNA) that improve their reproductive success (called fitness) to be present in order to be selected. However, other research has shown that beneficial mutations can interfere with each other. This interaction (called epistasis) can constrain adaptation and make it more difficult to become well adapted. Studying this effect is important, as understanding (and predicting) evolution affects our ability to anticipate changes in evolving populations. This is useful in planning for the evolution of disease (such as the flu), mitigating the evolution of antibiotic resistance, and predicting how populations will respond to climate change.
To investigate how evolution is constrained by epistasis (whether a mutation’s effect on fitness is adversely related to the presence of other mutations), we employ a common visual metaphor in evolutionary biology: the fitness landscape. In a fitness landscape, potential genotypes (specific possible DNA sequences) are envisioned as a surface, where genotypes that are just one mutational step from each other are adjacent. The fitness of each genotype is represented by the elevation of that point in the landscape. Thus a genotype with a high elevation is more likely to reproduce. A population consists of a cloud of points on the surface of this landscape, each point being a single organism’s genotype. Mutation adds new genotypes to the population, often causing cloud to spread. Natural selection operates to remove low fitness genotypes (organisms at lower elevation), leading the cloud to increase in average fitness (elevation in the metaphor) over time. If the cloud of genotypes reaches the highest location in the fitness landscape, then the population it represents is optimally adapted to its environment.
Our work addresses the fundamental question of whether the fitness landscape is smooth or hilly. If a landscape is hilly, populations that possess genotypes that place them at the top of a hill (a location where all nearby mutations decrease fitness) will be trapped there, because any mutants that go down the hill will be outcompeted by those at the top. This is a problem if there exist taller hills a long way away. The genotypes that the peaks of these hills represent possible individuals that would be better adapted to the environment, but the current population has no way of evolving toward them. This problem doesn’t exist if a landscape is smooth (having just one hill), as all adaption leads to the global optimum (the best adapted genotype).
In nature, populations can be spread across an environment in a variety of patterns. Some of these patterns make it easy to migrate from one part of the population to another, while others make it challenging. Easier migration allows for beneficial mutations to spread through the population, while decreased mutation slows this process.
Interestingly, we can use this property to determine how hilly an adapting species’ fitness landscape is. If the fitness landscape is smooth, populations will evolve to the same optimal genotype, regardless of how easy migration is, although populations with more migration will get there faster. However, in hilly landscapes, populations can become trapped on various different fitness hills. A population with lots of migration is likely to all get trapped on the same hill, because beneficial mutations will sweep across it shortly after they first occur. Populations with limited migration, on the other hand, will likely reach a wider variety of peaks, as different beneficial mutations sweep through different parts of the population. This means that the population will evolve slower, but can better adapt to its environment because some of the explored peaks may be higher in fitness than the peak discovered by a less structured population. We name this effect the Tortoise-Hare Pattern; the population with less migration (Tortoise) initially evolves slower than the population with more migration (Hare). However the Hare can become trapped at a suboptimal fitness peak, allowing the Tortoise to surpass it in fitness and win the “race”.
The presence of a Tortoise-Hare Pattern indicates that the fitness landscape is hilly (as the Hare would always win in a smooth landscape). We first used a digital model to confirm that the Tortoise-Hare Pattern is a litmus test for landscape hilliness. Afterwards we performed experiments with the gut bacteria, Escherichia coli, and found a Tortoise-Hare Pattern when we experimentally manipulated migration rates in a laboratory-based evolution experiment. We divided multiple populations of E. coli each into a grid of 96 subpopulations and had migrations occur between neighboring subpopulations (slow migration) or between any other subpopulation regardless of distance (easy migration). Finding evidence of landscape hilliness in a bacterial system implies that other living things are evolving on hilly fitness landscapes as well.
Too Long Didn’t Read (TLDR):
We experimentally evolved Escherichia coli populations, divided into smaller sub-groups, to explore whether more spatially restricted migration allows populations to achieve better adaptations. We found that limiting migrations between sub-groups does in fact yield greater fitness improvements, at the cost of slowing evolution. We name this the Tortoise-Hare pattern, as it is the slow-and-steady population with low migration that ultimately wins the fitness race.
For further reading there is a MSU press release concerning the work found here (http://msutoday.msu.edu/news/2015/tortoise-approach-works-best-even-for-evolution/). And, the paper corresponding to this work is recently published in PNAS found here (http://www.pnas.org/content/early/2015/05/06/1410631112.abstract, doi: 10.1073/pnas.1410631112). Unfortunately, the paper is behind a pay wall, but you can see a draft of the manuscript on bioRxiv (http://biorxiv.org/content/early/2014/06/03/005793) or email me at firstname.lastname@example.org.