Chapter 20: Population Genetics & Evolutionary Forces

Welcome to Last Minute Lecture.

This free chapter overview is designed to help students review and understand key concepts.

These summaries supplement not replaced the original textbook and may not be redistributed or resold.

For complete coverage, always consult the official text.

Welcome to the Deep Dive.

Today we're taking a deep dive into population genetics.

It's the study of genes in groups, you know, tracking how variation gets passed on and how the genetic makeup of a whole population can change over time.

Basically evolution in action.

Exactly.

And to set the scene, we've got this incredible sort of natural genetic experiment.

Think HMS Bounty Mutineers.

Ah, the Pitcairn Island story,

classic.

Right.

1790, they set Blyadrift, these nine British guys and some Polynesian men and women, a baby.

They end up in Pitcairn Island, totally uninhabited, completely isolated.

What a bop from the world.

Totally.

So this tiny group, they were just starting a colony, they were founding a whole new genetic lineage.

Everything depended on their small numbers, who they mated with, who survived.

And that really frames what population genetics is all about, doesn't it?

How so?

Well, the whole field is fundamentally about allele frequencies.

It's a theory of allele frequencies.

We're trying to understand how common different gene versions are.

Like the LM and LN alleles for blood types.

Precisely.

How common they are, how we measure that, predict it, and crucially, what forces push those frequencies around.

So the big question for you listening is, how do we go from just counting individuals in a sample to predicting the genetic future of an entire population?

That's the journey.

Okay, so let's start at the beginning.

Estimating those frequencies.

We can't test everyone, right?

No, definitely not.

You take a representative sample, the M blood types are a perfect example.

You've got two alleles, LM and LN.

Giving three genotypes, LM, LM, LM, LN, LN, LN.

Right.

So you count how many people in your sample have each genotype.

From that, it's pretty straightforward math to figure out the frequency of the LM allele, we call that P, and the LN allele, that's Q.

And since there are only two options here, P plus Q has to equal one, right?

Exactly.

That's our baseline, the allele frequencies in the sample population.

Okay, so we have P and Q.

Now what?

How do we predict?

Ah, now we get to the Hardy -Weinberg principle, developed back in 1908, independently by Hardy, a mathematician, and Weinberg, a physician.

What's the core idea?

It makes a really powerful prediction, but it relies on one key assumption,

random mating.

Meaning individuals choose mates without any regard for their genotype for that specific gene.

Precisely.

If that holds true, it's like all the sperm and eggs from the population get tossed into one giant pool, and the next generation is formed by just randomly pulling out pairs.

Like a genetic lottery.

Kinda, yeah.

And if that's happening, the principle predicts the genotype frequencies in the next generation will be two P of odor for one homozygote type.

Like LM, LM?

Two P odors for the other homozygote, LN, LN.

Okay.

And two P odors for the heterozygotes, LN, LN.

And if those proportions,

two P odors, two P odors, two P odors, two P odors, two LLs, two LLs for same generation after generation.

Then the population is in Hardy -Weinberg equilibrium for that gene.

It's stable.

It's kind of the null hypothesis.

No evolution is happening at that specific gene locus.

Can we test this?

Like with the MN and both types?

Absolutely.

People did.

They took the calculated P and Q values for LM and LN, plugged them into two P2O or two P2O odors, and calculated the expected number of people with each genotype.

And how did it compare to the actual observed numbers?

It was incredibly close.

Almost a perfect match.

Wow.

So that suggests people really are mating randomly regarding MN blood type.

It strongly suggests that, yeah, it's a neutral trait in that sense.

Okay.

That's cool for checking assumptions, but what about using it predictively?

Like for genetic diseases, say, phenylketonuria, PKU, it's rare, recessive.

Can HLU tell us how many people are carriers, even if they don't have the disease?

Yes.

And this is a really important application.

We run the HW principle in reverse.

We usually know the frequency of individuals with the recessive disease.

That frequency is $2.

Because you need two copies of the recessive allele Q to have the condition.

Right.

So if we know two tallers and say one in 10 ,000 for PKU, we just take the square root to find $2, the frequency of the mutant allele itself.

Okay.

So square root of 0 .0000 is 0 .01.

So Q equals 0 .01.

Exactly.

And since we know P plus Q is 1, P must be 0 .99.

Now we can calculate the carrier frequency, which is 2PP01.

So 2 times 0 .99 times 0 .01.

That's 0 .0198.

Which is almost 2%.

So even though the disease is very rare, one in 10 ,000, about one in 50 people are carriers.

HW lets us uncover that hidden genetic load.

That's incredibly useful.

Does this work for genes on sex chromosomes too, like X -linked color blindness?

It does with a slight modification.

Since males only have 1x, the frequency of affected males directly equals the allele frequency Q.

For females, the standard two tallers, two tallers, two tallers, two tallers, two tallers applies.

You can also extend it to genes with multiple alleles, like ABO blood types, using a trinomial expansion instead of binomial.

Okay.

So HW describes this perfect stable state, but you said it's often a baseline.

What knocks it off balance?

What are the exceptions that actually drive genetic change, you know, evolution.

Right.

HW equilibrium relies on several assumptions.

Random mating, large population size, no mutation, no migration, and equal survival rates.

When any of these are violated, the equilibrium breaks down.

The main drivers are non -random mating, unequal survival, population subdivision, and migration.

Let's take those one by one.

Non -random mating.

What's the most common type?

Consanguineous mating, or inbreeding, mating between relatives.

How does that affect HW?

It messes with the random pooling assumption.

Inbreeding doesn't actually change the overall allele frequencies, P and Q, in the population, at least not immediately, but it drastically changes the genotype frequencies.

Compared to the HW prediction, two PAQs, it reduces the proportion of heterozygotes and increases the proportion of both types of homozygotes, key of two and two past two.

And we can measure how much inbreeding is happening.

Yes, using something called the inbreeding coefficient, F.

It ranges from zero, completely random mating, to one, complete self -fertilization, basically.

The higher the F, the more heterozygotes decrease and homozygotes increase.

And that's why rare recessive disorders sometimes pop up more often in isolated groups or families with a history of intermarriage.

Exactly.

Inbreeding brings those hidden recessive alleles together in homozygotes more often than random mating would.

Okay, what's next?

Unequal survival.

This ties into selection.

If the different genotypes produced at fertilization don't all survive to adulthood at the same rate, then the genotype frequencies you observe in adults won't match the initial K2 loader to P2P2 prediction.

So if, say, homozygotes have a lower survival rate.

Then you'd find fewer individuals in the adult population than HW predicts.

Sometimes you might even see an excess of heterozygotes if they happen to have a survival advantage.

We see this in lab studies with Drosophila, for instance.

Makes sense.

Then there's population subdivision.

You mentioned fish in separate lakes.

Right.

The HW principle assumes a single, large, randomly mating population, what we call panmixia.

But real populations are often structured.

Maybe they live in different patches, like those fish, or maybe there are social barriers to mating.

So even if each little subgroup is in HW equilibrium internally?

When you lump them all together and calculate allele frequencies for the total population, it won't be in HW equilibrium.

You'll typically find fewer heterozygotes overall than predicted for the combined group.

It's called the Walland effect.

Because the allele frequencies differ between the subgroups.

Okay.

And the last one, migration.

Or gene flow.

If individuals move from one population to another, and these populations have different allele frequencies, it obviously changes the frequencies in the recipient population.

Like if population A has P .5 and population B has P .0 .8, and some Bs move to A.

Exactly.

The frequency P in population A will increase.

But there's a subtler effect too.

Imagine two whole populations merging instantly.

Let's say half the new combined population came from Pop I, P .0 .5, and half from Pop II, P .0 .8.

Okay, the new average P would be .65.

Right.

But if you look at the genotype frequencies immediately after merging, just adding the individuals together, they won't fit the HW prediction for P .0 .65.

You'll have an excess of homozygotes and a deficit of heterozygotes, compared to the new equilibrium expectation.

2P dollars, a rich .652 with 3 .4225, et cetera, et cetera.

Ah, because you just mixed existing genotypes, you haven't had random mating within the mixed group yet.

Precisely.

But here's the key thing about migration's effect on HW structure.

It only takes one generation of random mating within that newly merged population to restore Hardy -Weinberg equilibrium based on the new allele frequency.

Heavey is 0 .65.

So migration changes allele frequencies and causes a temporary HW upset, but random mating quickly sorts out the genotype proportions again.

Correct.

Unlike inbreeding, which maintains altered genotype frequencies as long as it continues.

Okay, so we've seen things that disrupt the HW balance, sometimes temporarily.

But let's get to the big driver of long -term directional change, natural selection.

This links back to unequal survival, right?

Absolutely.

Naturally, natural selection is all about differential survival and reproduction based on fitness.

Fitness, symbolized W, is basically a measure of an organism's ability to survive and pass on its genes.

And the average fitness of a population tells us something.

Yeah, if the average fitness is one doll, the population size is stable.

If it's greater than one, it's growing.

Less than one, it's declining.

But comparing fitness between genotypes is key for selection.

Right.

We often use relative fitness.

We assign the most successful genotype, a relative fitness of one, and measure the others relative to that.

The difference between one and the fitness of a less successful genotype is called the selection coefficient.

So as measures how strongly selection is acting against a particular genotype.

Exactly.

Let's take that insect example again.

Dark color A is dominant, light A is recessive.

Maybe in a dark forest, the light moths A are easily spotted by predator.

But they have lower survival.

Let's say their relative fitness, W, is 0 .9 compared to the dark moths, AA and AA, whose fitness is 1.

The selection coefficient against AA is then S equals 1, 0 .9 equals 0 .1.

They are 10 % less likely to survive and reproduce.

And this as value drives the change in allele frequency.

Yes.

We can calculate exactly how much the frequency of the A allele Q will decrease in one generation due to the selection pressure.

If you start with PQ of 0 .5, that 10 % selection against a poral will cause Q to drop from 0 .5 to about 0 .487 in a single generation.

It's a systematic force.

Now you mentioned something crucial earlier, how selection acts differently on dominant versus recessive alleles.

Yes.

This is really important.

Selection is very effective at removing a harmful dominant allele.

Why?

Because every individual carrying it, AA or AA, expresses the trait and is potentially exposed to selection.

It can't hide.

Makes sense.

But selection is much less effective against a harmful recessive allele, especially when it's rare.

Why is that?

Because most copies of a rare recessive allele are found in heterozygotes, AA.

They carry the allele, but they don't express the harmful trait, assuming it's fully recessive.

So selection can't see the allele in them.

The allele effectively hides in the heterozygotes.

So it can linger in the population at low frequencies for a very long time, even if it's really bad for the homozygotes.

Exactly.

It's very hard for selection to eliminate it completely.

Is there a real -world example that hammers this home?

The classic textbook case is the peppered moth, Biston d 'Achilleria in England.

Before the Industrial Revolution, most moths were light -colored, camouflaged on lichen -covered trees.

A dark form caused by a dominant allele was very rare.

Okay.

Then came industrial pollution.

Right.

Soot killed the lichens and blackened the tree bark.

Suddenly, the light moths stood out against the dricks' background and birds gobbled them up.

The rare dark form was now camouflaged.

So the selection pressure completely flipped.

Dramatically, the relative fitness of the light form plummeted.

Selection favored the dominant dark allele.

And because it was dominant, selection acted fast.

The frequency of the dark form went from less than 1 % to over 90 % in some industrial areas in just about 50 years.

Wow.

That's evolution happening practically in front of people's eyes.

It really is.

And then later, when pollution controls cleaned up the air and the trees became lighter again, the selection pressure reversed again.

Yep.

The light form started to increase in frequency again.

It's a powerful demonstration of selection intensity and how it tracks environmental change.

Okay.

So selection is this powerful, systematic force.

But evolution isn't just systematic, is it?

There's chance involved, too.

Absolutely.

And that brings us to the second major evolutionary force.

Random genetic drift.

Drift.

This is about random fluctuations.

It's purely about chance events and survival and reproduction, and especially in the transmission of alleles from one generation to the next.

Think about it.

Even if an A parent has two offspring, there's no guarantee one will get A and one will get A.

It's like flipping a coin for each offspring.

So allele frequencies can change just by random luck of the draw, especially when numbers are small.

Precisely.

The smaller the population size, n, the bigger the impact of these random sampling errors.

In a huge population, random fluctuations tend to cancel each other out, and drift is negligible.

But in a small population, drift can be the dominant force changing allele frequencies.

How do we measure the effect of drift?

One key way is by looking at genetic variability, often measured by heterozygosity, H, the proportion of heterozygotes.

Drift systematically reduces heterozygosity over time.

It removes variation.

Yes.

Each generation, heterozygosity is expected to decrease by a factor of T1 to n.

It's a small amount each time if n is large, but it's relentless and cumulative.

Over many generations, drift leads to the random loss of alleles and the fixation of others.

Fixation meaning an allele reaches a frequency of 100 % P1 or Q1.

Correct.

Eventually, purely by chance, one allele will become fixed and all other alleles at that locus will be lost from the population.

Let's bring back Pitcairn Island.

That started as a very small population.

Exactly.

A perfect scenario for drift.

Let's say their average effective population size over the first 10 generations or so is maybe around n20.

That's really small.

So, using that formula, how much heterozygosity would they be expected to lose?

Well, the proportion remaining after tit generations is 6th areas, 1, 12th, h2r.

So after 10 generations with n20, that factor, 1, 140, 10, is about 0 .776.

They'd be expected to have lost almost a quarter, around 22 -23%, of their original heterozygosity purely due to random drift.

Just because of the small numbers.

That shows how vulnerable small populations are to losing genetic diversity.

Definitely.

And if an allele provides no selective advantage or disadvantage, if its selectively neutral drift is the main game,

the probability that any given neutral allele will be the one that eventually drifts to fixation is simply its current frequency.

So if a neutral allele has a frequency of p, 0 .2, it has a 20 % chance of eventually taking over the population and an 80 % chance of being lost.

Possibly by chance, yes.

Okay, we've got selection pushing frequencies systematically and drift pushing them randomly.

But things can also reach a stable point, right?

Not just the no change of hw, but a different kind of stability.

Yes, that's dynamic equilibrium.

This isn't the static, no -evolution state of Hardy -Weinberg.

It's a situation where opposing evolutionary forces are acting, but they balance each other out, leading to a stable allele frequency that's somewhere between 0 and 1.

What's a major example of this?

The classic is balancing selection, specifically heterozygote advantage or over dominance.

This is when the heterozygote AA is actually fitter than both homozygotes, AA and AA.

Exactly.

So selection is working to eliminate the A allele through the A homozygotes, and it's working to eliminate the A allele through the AA homozygotes.

But because the A heterozygote is the most fit, selection actually preserves both alleles in the population.

It maintains diversity.

Yes, it leads to a stable balanced polymorphism where both alleles persist.

The textbook example involves the sickle cell allele, HBBS, in regions with malaria.

Okay, walk us through that.

The HBBS allele, when homozygous, HBBS, HBBS, causes severe sickle cell disease, which is often lethal, so there's strong selection against this allele.

Allele is close to one.

Right.

But in areas where malaria is common, heterozygotes, HBBA, HBBS, who carry one normal allele, HBBA, and one sickle cell allele, have a significant survival advantage because they are resistant to malaria.

The normal homozygotes, HBBA, are fully susceptible to malaria.

So the heterozygote is fitter than both homozygotes in that environment.

Correct.

The HBBS homozygote suffers from sickle cell disease.

The HBBA homozygote suffers from malaria.

The HBBS heterozygote is relatively protected from both.

A tough trade -off, but the heterozygote wins out.

Which means selection actively maintains the harmful HBBS allele in the population at surprisingly high frequencies, like 10 % or more in parts of West Africa, because of the protection it offers heterozygotes against malaria.

We can even calculate the expected equilibrium frequency based on the selection coefficients against each homozygote.

Amazing.

What other kinds of balance are there?

Another key one is mutation selection balance.

We know selection tends to remove deleterious harmful alleles, but mutation is constantly reintroducing them, albeit at a low rate.

So mutation creates them, selection removes them.

And eventually they reach an equilibrium frequency where the rate of creation by mutation equals the rate of removal by selection.

For a harmful recessive allele, we can calculate this equilibrium frequency Q based on the mutation rate and the selection coefficient S against the homozygote.

The formula comes out as QLA exceses.

So even nasty recessive lethals, where S1 are maintained at a low frequency, determined by how often mutation creates them.

Exactly.

Typically around Q0 .0001 or so, based on common mutation rates.

And one more balance.

You mentioned drift earlier.

Right.

Mutation -drift balance.

This applies mainly to neutral alleles where selection isn't a major factor.

Drift constantly removes variation, reduces heterozygosity H, while mutation constantly introduces new variation, increases H.

So they fight each other.

Pretty much.

An equilibrium level of heterozygosity is reached that depends on both the population size and the mutation rate.

The formula is approximately 1H equals 4NEU, 4NEU plus $1.

What does that tell us?

It shows that in very large populations, large N, heterozygosity will be high because mutation introduces variation faster than drift can remove it.

But in very small populations, small N, drift dominates and heterozygosity will be low, regardless of the mutation rate.

Like the cheetahs, famous for having very low genetic diversity.

Precisely.

Their history likely involves population bottlenecks, leading to small N allowing drift to strip away much of their variation, and mutation hasn't had time or sufficient population size to replenish it effectively.

Wow.

Okay, that's quite a tour.

We went from the ideal static Hardy -Weinberg baseline through things like inbreeding and migration that mess with genotype frequencies,

then into the big engines of allele frequency change,

systematic selection, and the peppered moths, and random genetic drift, especially powerful in small populations like Pitcairn, and finally ending up with these dynamic equilibria where opposing forces like selection, mutation, and drift hold each other in check.

It gives you a framework for understanding nearly all genetic variation within and between populations.

So what's the one big takeaway you want people to remember from all this?

I think it's the concept of the genetically effective population size, often written knee.

All those calculations involving NEC, like for drift or mutation -drift balance, they don't really use the census population size, the actual head count.

Why not?

Because not everyone contributes equally to the next generation.

Maybe mating isn't perfectly random.

Maybe some individuals have way more offspring than others.

Maybe the sex ratio is skewed.

The effective population size, knee, is the size of an idealized population, the kind assumed by HW with random mating, et cetera, that would experience the same amount of genetic drift as the real population.

And knee is usually smaller than the actual number of individuals.

Often much smaller.

For humans, for example, while there are billions of us, estimates of our long -term effective population size based on genetic diversity are maybe only around 30 ,000 to 40 ,000.

It reflects bottlenecks and constraints in our evolutionary past.

It's a much better measure of a population's genetic resilience and evolutionary trajectory than just counting heads.

That's a fascinating point.

OK, a final thought for you, our listener.

We saw how selection dramatically reshaped the peppered moth population in response to environmental change.

Today, human populations are migrating and mixing like never before, which, as we learned, creates temporary Hardy -Weinberg disequilibria, but also changes allele frequencies through gene flow.

Thinking long -term about the human genetic landscape, which forced systematic selection adapting us to new pressures,

or the ongoing mixing and migrating do you think will have the more profound impact on our genetic future?

Something to ponder.

Definitely something to think about.

Thank you for diving deep with us today.

We'll catch you on the next one.