Molecular Basis of Evolution 2: Genetic Drift and Selection

Learning Objectives

Deepen understanding of genetic drift.
Explore the effects of different types of natural selection on allele frequencies.

Selection on dominant and recessive alleles.
Underdominance and overdominance.

Understand how populations diverge over time under drift and selection.

Under what conditions do populations diverge from one another?
When is inter-population variability minimized and maximized?

Introduction

For this workshop, you will be using a simulator of a finite population experiencing genetic drift and natural selection.

Read the ‘About the Model’ section on the simulator page to learn more about how it works. Then acquaint yourself with the interface by running a few simulations with various combinations of parameter values.

Exploring Drift

In the last workshop you explored genetic drift by running a single simulation at a time. Using this simulator, we can examine the patterns that emerge when simulate multiple independent populations experiencing drift at the same time.

Run simulations using each of the following parameter combinations, each with 100 replicate simulations:

Small: N = 200, simulation length = 2000, initial red = 50%, no selection.
Medium: N = 2,000, simulation length = 2000, initial red = 50%, no selection.
Large: N = 20,000, simulation length = 2000, initial red = 50%, no selection.
Very large: N = 20,000,000, simulation length = 2000, initial red = 50%, no selection.

For each simulation save the output plot. Don’t forget to give them descriptive filenames so you know which plot goes with which simulation run!

What differences did you notice in the plots of the simulations?
Create a document to save your work from this workshop and paste your figures into your document. Write a descriptive caption that includes the simulation parameters and enough information for someone unfamiliar with the simulator to understand the figure.

One of the predictions of the Fisher-Wright model is that over time, drift will cause the allele frequencies to follow different patterns in different populations. The model output s is the standard deviation of the values of p at the end of the simulation. This gives us an estimate of the amount of variability among populations at the end of the simulation. To get an intuitive feeling for how this works, try several simulations where you vary the population size from small to very large.

How does the population size affect the value of s?
How would you state your findings in plain, non-technical English?

Next choose a population size of 2000 and run several simulations varying the simulation length from 1,000 to 20,000 generations.

Examine the plots for each simulation and describe qualitatively what happens to the divergence among different populations as you increase the number of generations.
What happens to the value of s and the number of allele fixations as you increase the number of generations?
How many generations do you need to simulate before you start to observe populations going to fixation?

Exploring Selection with Dominant and Recessive Alleles

Natural selection can result in a favorable allele increasing in frequency in the population.

The simulator’s selection parameters allow you to reduce the relative fitness of one or more genotypes. Which of the genotypes should you choose to select against in each of the following scenarios?

Selection for a dominant allele.
Selection for a recessive allele.
Underdominance.
Overdominance.

In which of the above scenarios would you expect a balanced polymorphism to remain in the population? Why?

With a population size of 2000 and a simulation length of 1000 generations, try out some simulations with different combinations of selection against different genotypes. Do you notice any patterns?

Now, let’s contrast what happens when we have selection for a dominant and a recessive allele. First, let’s assume that the red allele is favored and that it is dominant. In other words, any genotype with at least one copy of the red allele has higher relative fitness.

Which genotype or genotypes should be selected against if the favorable red allele is dominant?

Let’s see what happens with a dominant allele that is selected for. Choose a simulation with a population of 200,000, a simulation length of 10,000 generations, and initial R frequency of 50%. First run the simulation without selection. What do you observe?

Dominant: Next, apply a selection coefficient of 0.001 against the genotype or genotypes you identified. You should paste copies of the simulation figures along with a list of parameter settings for each of the simulations below so that you can make comparisons. Also record the values of p-bar and s.

How did the values of p bar and s change between the two simulations? Interpret these values in plain English.
Did you observe any fixations in either simulation?

Recessive: Now let’s compare what happens when the favored allele is recessive.

Which genotype or genotypes should be selected against if the favorable red allele is recessive?

Apply a selection coefficient of 0.001 against the genotype or genotypes you identified.

Describe the differences you observe in the plots of the dominant and recessive simulations.
Did you observe any fixations in either simulation?
How were the values of p-bar and s different? Try to explain these differences in terms of the visual intuition gained from looking at the plots.
On which type of allele (recessive or dominant) did selection act more quickly?

In the last two simulations you started with the red allele at high frequency, now let’s contrast that with what happens to recessive and dominant alleles when they begin at a low frequency. For the next two simulations, set the initial red frequency to 2% and run the simulations for 20,000 generations.

Dominant: run the simulation with selection coefficients set to 0.001 for the genotype or genotypes appropriate for a dominant favorable allele.

Recessive: now re-run the simulation with the selection coefficients applied to the appropriate genotypes for selection on a recessive allele.

What was different between the dominant and recessive scenarios? Consider both the individual populations (red traces) and the average behavior (black traces).
What was surprising you when you compare these to the simulations in which we started with R at high frequency?
Comparing selection on recessive and dominant alleles, which resulted in more inter-population variability at the end of the simulations? Was this different when you started with R at high and low frequencies?

When starting at high frequency, the recessive simulations had less inter-population variability and more populations approached fixation, while the ‘noise’ remained in the dominant simulation, even while p-bar approached 1.

Finally, it looks like there might be some interesting behavior that wasn’t quite captured in the short recessive simulation starting at low frequency. For your final simulation, re-run the recessive selection simulation but extend the number of generations to 200,000. Note that this simulation might take a minute or more to run, be patient!

Describe the patterns you see.
If a mutation results in a favorable allele, it will start out at a low frequency in the population. Discuss how the likelihood of the new favorable mutation reaching high frequency is different if it is recessive or dominant. In which scenario is there more uncertainty?
Paste a figure of a simulation of a dominant favored allele and a figure of a recessive favored allele. Write a brief description of each and be sure to highlight any differences you notice. Hint: to make the patterns clearer you might want to simulate very large populations (10,000,000) and relatively weak selection (in the range of 0.001). Run your simulations for enough time steps to be able to observe the long-term pattern. I suggest around 10,000 generations.

Bonus: What happens to the rate of fixation when you have incomplete dominance (when the heterozygote has intermediate fitness)? How do the rates of fixation of dominant, recessive, and incompletely dominant alleles compare?

Exploring Overdominance and Underdominance

We’ve explored what happens with selection on dominant and recessive alleles, but what happens when selection acts only on the heterozygote?

Recall that underdominance means the heterozygote is selected against, while overdominance (also known as heterozygote advantage) means that the heterozygote is more fit than either homozygote.

Let’s start with underdominance.

Which genotype or genotypes should you select against to simulate underdominance?

Play around with some underdominance simulations. Vary the simulation length, the population size, and the strength of selection against the heterozygote.

What do you notice compared to simulations without selection? Do populations go to fixation at the same rate with underdominance as compared to no selection?

After you have developed an intuitive understanding of what happens to populations with underdominance with different parameter settings, paste a figure of a representative simulation or simulations into this document and write a brief description of what is shown.

Now explore overdominance with some simulations.

Which genotype or genotypes should you select against to simulate overdominance?
Try large and small populations and various simulation lengths.
Try different selection strengths.

Choose a representative figure from one of your simulations of overdominance and paste it into the document with a caption and description.

Compare the behavior of simulations with overdominance to those with only drift.

Which scenario results in rapid divergence of populations, i.e. fixation to one or the other allele?
Which scenario results in a gradual divergence of populations?
Which scenario results in a stable polymorphism, i.e. both alleles remain in the population?