Global stability is more difficult to determine than local stability.
In the next few lectures, we will see that some of the equations we have been studying may be solved directly, providing a general solution for how the system behaves over time. This general solution can be used to determine the global stability properties of equilibria.
Other methods exist to determine global behavior of the dynamics.
For one-variable discrete models, a graphical method is particularly useful:
Method:
We will demonstrate the method using the diploid selection model.
For the diploid selection model, there are four cases that we must consider.
Case 1: Directional selection favoring A (WAA>WAa>Waa)
The only equilibria are =0 and =1 (the polymorphic equilibrium is not valid in this case).
Near =0, f'()=WAa/Waa and is greater than one.
Near =1, f'()=WAa/WAA and is less than one.
Consequently, the function p[t+1] versus p[t] will look like:
Since this is always above the diagonal, the A allele frequency increases over time.
=1 is globally stable.
Case 2: Directional selection favoring a (WAA < WAa < Waa)
The only equilibria are =0 and =1 (the polymorphic equilibrium is not valid in this case).
Near =0, f'()=WAa/Waa and is less than one.
Near =1, f'()=WAa/WAA and is greater than one.
Consequently, the function p[t+1] versus p[t] will look like:
Since this is always below the diagonal, the A allele frequency decreases over time.
=0 is globally stable.
Case 3: Heterozygote Advantage (WAA < WAa > Waa)
There are now three valid equilibria.
Near =0, f'()=WAa/Waa and is greater than one.
Near the polymorphic equilibrium, f'() is less than one.
Near =1, f'()=WAa/WAA and is greater than one.
Consequently, the function p[t+1] versus p[t] will look like:
This function is above the diagonal when p[t] is less than and the A allele rises in frequency.
This function is below the diagonal when p[t] is greater than and the A allele decreases in frequency.
The polymorphic equilibrium is globally stable.
Case 4: Heterozygote Disadvantage (WAA > WAa < Waa)
There are again three valid equilibria.
Near =0, f'()=WAa/Waa and is less than one.
Near the polymorphic equilibrium, f'() is greater than one.
Near =1, f'()=WAa/WAA and is less than one.
Consequently, the function p[t+1] versus p[t] will look like:
This function is below the diagonal when p[t] is less than and the A allele decreases in frequency.
This function is above the diagonal when p[t] is greater than and the A allele increases in frequency.
There is no single global equilibrium point.
The population will move towards =0 if it starts below the equilibrium and towards =1 if it starts above it.