Genetic Drift Workshop
Conservation Biology
Last week we observed the dynamics of genetic drift in different populations sizes and bottlenecks. This week we will observe how drift interacts with migration, mutation, and selection.
Drift and Migration
Island migration
Last week we saw that the speed of drift is related to the size of a population. Migration opposes drift by introducing new alleles into a population over time, or increasing the size of the meta-population by allowing nearby populations to exchange migrant alleles.
Use your browser to open the simulation here:
From the default settings of the simulations click run to re-familiarize yourself with the drift trajectories of the 5 populations. Then click the migration box once so that it reads island underneath the box. There is now a number 'rate' that you can change. That number is the fraction of alleles that each of the five populations exchanges with its neighbors. Set the rate to 0.01 initial, representing 1% exchange of alleles each generation. Decrease the population size to 25 to make the simulation faster. Run the simulation.
Does the time to fixation/loss increase or decrease for individual populations?
Now increase the migration rate to 0.05. How does the rate of fixation/loss change now?
Experiment with different rates and population sizes. Do you think a more viable conservation strategy to battle allele loss due to drift would be to establish a single population of 100 or 4 populations of 50 that exchange 5% of their alleles per generation? Test your hypothesis with a simulation of each and describe the results.
Source/Sink Migration
If you click the migration box a second time, the mode changes to source/sink mode in which alleles enter the five populations from an outside source of infinite size at a given rate. Enable this and try a few different parameters.
Will fixation of alleles ever occur?
Selection and Fitness
The fitness of each genotype can be set as a fraction or multiple of 1. This selection coefficient is the chance that genotype has of reproducing where 1 means that an individual with that genotype is most likely to produce 1 offspring. Set A2A2 to have a fitness of 0.9 and leave the other homozygote and the heterozygote at 1. Set the population size to 500. Run the simulation.
At first there is a rapid decrease in the frequency of the A2 allele, but then it levels off and drift seems to dominate the fate of the allele. When A2 is at low frequency does it exist mostly in A1A2 heterozygotes or A2A2 homozygotes?
What selection coefficient is the one affecting the frequency of allele A2 when it is at low frequency, 0.9 or 1?
Why does drift dominate the trajectory of the A2 allele when it is at low frequency?
Overdominanceoccurs when the heterozygote is more fit than either homoxygote. The dynamics that occur in this situation maintain allelic diversity in a population. Set A1A1 to also have a selection coefficient of 0.9 so both homozygotes have selection coefficients of 0.9 and the heterozygote is set to 1.0. Run the simulation with different population sizes and fill out the table.
Population size / Time to loss / Time to fixationAt what population size does selection seem to matter less than drift?
At that population level you just found, reduce the fitness of the homozygotes until the strength selection again overcomes drift again. What is the homozygote fitness at this point?
A1A1 =
A2A2 =
What you have observed is the interplay between drift and selection. If you are curious you can do these exercises using directional selection, or under-dominance where the heterozygote is less fit than either homozygote.
Mutation and Drift
Drift removes variation from a population that is originally generated by mutation. We will now add mutation to the basic simulation and see how things change. With default settings, change both the A1 => A2 and A1 <= A2 mutation rates to 0.01. This means that each generation 10% of alleles will mutate to the other version. Run the simulation. Does fixation happen more or less quickly than without mutation?
This is not a biologically realistic mutation rate. In reality most species would have a mutation rate of 1e-6 to 1e-8. That is, every generation each allele has a 1 in 10,000,000 or so chance of mutating. Set the mutation rates to a more realistic number like 1e-6 and rerun the simulation.
In a conservation context, can mutation be relied on to counteract drift's removal of variation from a small population?