Chapter 24: Quantitative Genetics

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.

You know, it's actually a pretty staggering number to try and wrap your head around.

Oh, it really is.

By the year 2050, the United Nations projects that the global human population is going to hit 9 .8 billion.

Yeah, 9 .8 billion.

Right.

Which means 9 .8 billion mouths to feed on a planet that's already dealing with, you know, dwindling petroleum, shifting climates, and this really desperate escalating need for biofuels.

So the question is, how on earth are we going to squeeze that many calories out of the exact same amount of farmland?

Well, I mean, it is arguably the defining agricultural challenge of our lifetime.

And the answer, or well, at least the most promising part of the answer, it doesn't actually lie in, you know, discovering new contents to farm.

The solution is totally microscopic.

It's hiding right there inside the DNA of the crops we already grow.

Yeah.

So let's look at one of the most vital crops on the planet, which is corn.

If you want a corn plant to produce more oil, which we absolutely need for both food and biofuels, you can't just go in and flip a single genetic switch.

No, not at all.

It's wildly complex.

It's influenced by dozens of genes plus environmental factors.

But researchers didn't just, you know, threw their hands up.

They used incredibly precise molecular techniques to map this specific region on corn chromosome 6.

And they eventually traced the oil producing trait all the way down to a single gene.

I think it's known as DG8T12.

Yeah, DG8T12, exactly.

And what they found inside that gene is just this perfect illustration of what we're unpacking today.

So in the high oil producing strains of corn,

there's actually an insertion of a single extra codon, just a tiny little piece of DNA that adds one specific amino acid, Phelamine, to the enzyme that the gene produces.

Just one amino acid.

Just one.

And that single addition just totally supercharges the corn's oil production.

And the wild part to me is that this extra codon, it's actually present in the wild relatives of corn, right?

Yep, it is.

So it looks like it was just accidentally lost during, like, centuries of human domestication.

And now geneticists can use this knowledge to literally just put it back.

Exactly.

We can reverse the mistake.

So welcome to the deep dive.

Our mission today is to decode one of the messiest, but honestly most fascinating fields of biology, which is quantitative genetics.

We're diving straight into Chapter 24 of genetics, a conceptual approach.

It's a great chapter.

It really is.

We are going to explore how geneticists analyze these complex characteristics that just don't fit into neat little Mendel -style boxes.

We're moving way past the simple binary inheritance we usually learn about, and we're digging into the complex math that explains human height, disease risk, and those crop yields we just talked about.

Right, because to understand how we map something as complex as corn oil, we really have to start with why it behaves so differently from the traits early geneticists study.

Right, the Mendel peas.

Exactly, Mendel's peas.

We were talking about the massive transition from discontinuous traits to continuous traits.

So discontinuous traits are the classics, right?

Plants are either tall or dwarf.

A pea is either yellow or green.

It's one or the other.

Right.

They fall into very distinct, non -overlapping categories, qualitative traits.

But continuous traits, which are the actual focus of quantitative genetics in this chapter, they vary along a smooth, unbroken scale.

So, you know, human height, blood pressure, the weight of a seed.

Any individual's phenotype has to be described by a precise measurement, not just a broad category.

Yes, exactly.

So let's construct a bit of a mental model for you, just to see how multiple genes create that sliding scale.

Okay, let's do it.

Imagine a hypothetical plant, and its height is determined by just three distinct genes, or loci.

Let's call them A, B, and C.

Easy enough.

Right.

And every plant starts with a baseline genetic height of 10 cm.

Now, every time it inherits a plus allele, so an A +, B +, or C +, that allele acts like a tiny little factory,

it produces a dose of a growth hormone that adds exactly 1 cm to the plant's height.

Okay, and what about the other allele?

A minus allele adds nothing.

It just gives zero extra height.

Okay, so if a plant inherits all minus alleles across those three loci, it just stays at its baseline of exactly 10 cm.

Yes.

And since there are three loci, and each locus has two alleles, there are actually 27 possible genotype combinations that a plant could inherit.

Wow, okay.

But, and this is kind of the crucial aha moment of quantitative genetics,

how many actual heights or phenotypes do those 27 genotypes produce?

Well, I mean, if you just add them up, it's only seven, right?

The plants can only be 10, 11, 12, 13, 14, 15, or 16 cm tall.

Because the effects are purely additive.

So, a plant that is 12 cm tall might have a plus allele at the A and B loci, or it might just have two plus alleles at the C locus.

Oh, right, because they look exactly the same from the outside.

Exactly.

You can no longer just look at a 12 cm plant and reverse engineer its genotype.

The diagnostic window is basically frosted over.

And we haven't even factored in the chaos of the environment yet.

Like, if one of those 12 cm plants gets slightly more sunlight or maybe a bit more water, it might grow to 12 .5 cm.

Right, it pushes the boundary.

Yeah, and meanwhile, a plant that's genetically coded to be 13 cm might hit a dry patch and get stunted down to 12 .5.

So suddenly, those seven distinct height categories completely blur together into one continuous overlapping bell curve.

That's it, exactly.

And when a characteristic is influenced by many genes, which is called polygenic and influenced by the environment,

we call it multifactorial.

OK, but hang on, I want to push back on this a little bit.

I'm trying to map this continuous curve onto real life.

A sliding scale makes total sense for height or, you know, weight.

But what about diseases?

You don't have like half a heart attack or a quarter of a genetic disorder.

You either have it or you don't.

So does this polygenic math still apply when the outcome looks completely binary?

That is a really fantastic distinction to make.

And the answer is yes, the math absolutely still applies because there are hidden quantitative traits.

So what you're describing is called a threshold characteristic.

It looks binary on the surface.

You either have the disease or you don't.

But the underlying biological mechanism is actually a continuous quantitative mix of all those many genes and environmental stressors.

Oh, I see.

That underlying risk varies on a sliding scale.

It's only when that cumulative risk crosses a very specific biological threshold that the disease actually physically manifests.

So the susceptibility is continuous, even if the symptom itself acts like a light switch.

That is incredibly fascinating.

But OK, if continuous traits are this blurry overlapping mess of multiple genes and environments, how did early biologists ever figure out that they were still following the basic rules of inheritance?

It wasn't easy.

Right.

Because if I were a scientist in the early 1900s looking at a field of plants that are just every conceivable height, I might just assume the whole concept of structured Mendelian inheritance was totally wrong.

Well, you wouldn't be alone.

Many scientists assumed exactly that.

There was a massive decades long debate in biology over this.

Really?

Oh, yeah.

On one side, you had the biometricians who looked at continuous traits and argued that inheritance was just a fluid blending process like mixing paint.

And the other side.

The Wendelians.

They insisted that inheritance relied on distinct, indivisible genetic particles.

It wasn't until a mathematician named Ronald Fisher came along around 1918 that the math was really reconciled.

But the physical proof,

the really elegant demonstration that won the day, came from a scientist named Terman Nielsen -Ellel.

Right.

The wheat kernels.

I want to break this experiment down because it is essentially a magic trick of biology.

You're a lilius.

So Nielsen -Ellel took a pure homozygous wheat plant that grew white kernels and he crossed it with a pure wheat plant that grew purple kernels, which is essentially very dark red.

Now, if inheritance was just simple dominance, the offspring should have been all purple or all white, right?

Right.

But instead, the first generation, the F1s, they were all this intermediate red.

Which, if you're a biometrician, that looked exactly like blending inheritance.

They thought the genetic material had just mixed together like red and white paint making pink.

But then came the F2 generation.

Yes.

The revelation happened when Nielsen -Ellel took those intermediate red F1 plants and interbred them.

And they didn't just produce white and purple.

And they didn't just stay blended red.

They shattered into five distinct phenotypes in a highly specific ratio.

Let's hear the ratio.

Okay.

It was 116 purple, 416 dark red, 616 red, 416 light red, and 116 white.

And that ratio is the smoking gun of modern genetics.

Because that ratio of 1 to 4 to 6 to 4 to 1, it's the exact mathematical fingerprint that you'd expect from independent assortment across two separate gene loci, provided their effects are purely additive.

Nielsen -Ellel proved that the continuous colors weren't blended paint.

They were distinct genetic particles playing strictly by Mendel's established rules.

There were simply more of them interacting at once.

Which gives us this incredibly cool, practical takeaway formula for the listener.

If you cross two purebred extremes, you can estimate the total number of genes affecting a trait just by looking at that F2 generation.

You just find the proportion of offspring that look exactly like one of the original extreme parents.

And the formula from the chapter is 1 over 4 to the power of N, where N is the number of loci.

So in that weak cross, 1 out of 16 plants looked like the original pure purple parent.

And since 116 is 14 squared and equals 2, two distinct genes were controlling the color.

But you know, Nielsen -Ellel was kind of playing on easy mode there.

He really was.

He found a trait with only two genes, strictly additive effects, and almost zero environmental interference.

But for the vast majority of traits like human height or crop yield, we're dealing with dozens of genes and massive amounts of environmental noise.

So true.

And because we usually can't determine genotypes directly from these messy continuous phenotypes, we have to rely on a really heavy statistical toolkit.

Yeah, let's open that toolkit.

Because when you graph a complex trait, it almost always forms a normal distribution, you know, a symmetrical bill curve.

And to describe that curve, geneticists rely heavily on variance.

Which measures how spread out the measurements are from the center, right?

Exactly.

I always think of it like two dart boards.

So on both boards, the darts are perfectly centered around the bullseye.

The average, the mean is identical.

Okay, visualizing that.

But on the first board, the darts are clustered super tightly right around that center.

On the second board, the darts are scattered all the way out to the edges of the wood.

So that second board has a much higher variance.

And we care about variance because understanding why the darts are spread out like, is it a bad player or is it a gust of wind?

That is the entire goal of quantitative genetics.

That's a perfect analogy.

And to figure out the cause of that spread, we also look at how traits move together.

So we use correlation to measure the strength and direction of an association.

Right.

Like if arm length reliably increases as overall height increases, that's a positive correlation.

Exactly.

But knowing things are associated isn't quite enough to understand inheritance.

For that, we need regression.

Ah, regression.

It's basically the predictive engine of genetics.

It really is.

It lets us calculate how much one variable changes when another one does.

It is literally how we predict the traits of offspring based solely on their parents, even if we are totally blind to the underlying DNA.

And this brings us to a brilliantly simple experiment by Edward East, who used tobacco plants to completely untangle that variance we were talking about.

This is a great one.

Right.

He wanted to prove exactly how much of a plant's physical variation was genetic and how much was environmental.

So he crossed two purebred homozygous tobacco plants, one with short 40 .5 millimeter flowers and one with long 93 .3 millimeter flowers.

And the genius of this setup is that because those parent plants were highly inbred and homozygous, there was essentially zero genetic variation within each parent's strain.

They were clones.

Exactly.

So if you looked at a field of the short flower parents and you saw tiny differences in their lengths, like some were 40 millimeters, some were 41, you knew with 100 % certainty that the variance was purely environmental.

It was just differences in soil and sunlight.

But then East crossed those plants, created an F1 generation and bred those together to make an F2 generation.

And suddenly the variance in that F2 generation just exploded.

Right.

The flowers were all over the place in size.

Because for the first time, the hidden genes were actively recombining and shuffling.

That massive F2 spread proved definitively that total variation in a population is a mix of genetic differences stacking on top of environmental differences.

Which perfectly tees up the central master equation of quantitative genetics, because we need to know exactly how much of a trait is nature and how much is nurture.

We do.

So the formula for phenotypic variance, or VP, is VG plus VE plus VGE,

meaning total variance equals genetic variance plus environmental variance plus gene environment interaction.

Let's visualize this for a second.

Imagine the total phenotypic variance, VP,

is the overall sound quality of an orchestra playing a symphony.

Okay, I like that.

VG, the genetic variance, is the sheet music and the quality of the instruments themselves.

VE, the environmental variance, is the acoustics of the concert hall.

Is it humid?

Is it dry?

Is there an annoying echo?

And VGE, the gene environment interaction, that's the really counterintuitive part.

Yeah, it trips people up.

That would be like a specific antique cello that sounds incredibly rich and beautiful in a humid concert hall, but it sounds totally flat and terrible in a dry hall.

The gene isn't just acting on its own, its actual output fundamentally changes depending on the environment it sits in.

Exactly.

Now let's drill deeper into VG, the genetic variance itself.

It's actually made up of three distinct subcomponents.

Right, tearing it apart even further.

First is additive variance, VA.

This is each individual musician playing their part correctly, dependably adding to the total sound.

Second is dominance variance, VD.

This is when alleles interact at the exact same locus, like the first chair violinist playing so aggressively they completely mask the second chair.

And the third.

Third is gene interaction variance, VI, which is epistasis.

This is when genes at totally different loci interfere with each other.

It'd be like the trombone player actively reaching over and knocking down the flutist's music stand.

Which brings us to the concept of heritability.

And we have two types here.

Broad sense heritability is all the genetic variance, so the entire orchestra's instruments divided by the total variance.

But the holy grail for geneticists is narrow sense heritability.

And that is only the additive genetic variance, just the independent non -interfering musicians divided by the total variance.

But why is narrow sense so much more important?

The reason narrow sense heritability is so critical is that additive variance is the only thing that reliably passes from parent to offspring.

Dominance and epistasis, the masking and the interfering, those get broken up and completely reshuffled during reproduction.

Ah, because of meiosis and recombination.

Exactly.

Additive effects are the dependable, predictable building blocks.

Think of narrow sense heritability like a soup recipe.

Only the core additive ingredients are going to survive the generational game of telephone and predictably dictate how the next batch actually tastes.

And we can actually calculate this narrow sense heritability using regression, like we mentioned earlier.

We can.

If you plot the average phenotype of parents on a graph's x -axis and the average phenotype of their offspring on the y -axis, you get a line.

And the slope of that regression line literally is the narrow sense heritability.

Yep, it's that direct.

If the slope is a perfect one -to -one angle, the trait's differences are entirely due to predictable additive genetics.

If the slope is a flat horizontal line, the differences are entirely environmental.

It's just a brilliantly elegant piece of math.

It is.

But OK, I have to jump in here with some skepticism because heritability is probably the most misunderstood concept in modern science.

Oh, absolutely.

If I read a news article that says human height is 80 % heritable, does that mean 80 % of my specific height is due to my genes and 20 % is due to my diet?

I am so glad you brought that up because we need to completely dismantle that misconception.

Heritability tells us absolutely nothing about an individual.

Nothing?

Nothing.

It only applies to populations.

It means 80 % of the variation in height among the people in a specific population is due to genetic differences.

Going back to your dartboard analogy, it's about explaining the total spread of all the darts, not analyzing why a single dart landed where it did.

That is a crucial distinction.

And a high heritability doesn't mean a trait is immune to the environment, right?

I mean, during World War II in Europe, the average human height dropped significantly due to severe malnutrition, even though height is a highly heritable trait.

Precisely.

Furthermore,

there is no such thing as a universal heritability for any characteristic.

It is strictly tied to a specific population in a specific environment.

Makes sense.

If a population has incredible nutrition and universal health care, the environmental variance shrinks to near zero.

Mathematically, that makes the genetic heritability look massive.

But in a population dealing with famine, environmental variance skyrockets, and the genetic heritability of height for that exact same group drops significantly.

And most importantly, heritability tells us absolutely nothing about why two entirely different populations have different averages.

If you plant two fields of genetically similar corn and one gets terrible soil and grows poorly, the difference between the two fields is entirely environmental, even if the heritability of height within each field is basically 100%.

So once we successfully calculate that a trait is highly heritable within a single population, how do we physically go in and find those genes?

And more importantly, how do we use them to evolve a population like, you know, trying to get more oil out of our corn?

Well, historically, we mapped quantitative trait LOCIE, or QTLs, by meticulously crossing different strains over decades.

We'd look for genetic markers that always inherited alongside the trait.

But today, the entire field has shifted to massive data analytics with genome -wide association studies.

Right, GWOs.

GWOs completely changed the game.

Yeah, because instead of taking years to breed carefully controlled crosses,

geneticists can now look at entire naturally interbreeding populations all at once.

They just scan the DNA of hundreds of thousands of individuals looking for single nucleotide polymorphisms, SNPs.

Just tiny single letter changes in the genetic code.

Exactly.

And then computers look for statistical associations between those SNPs and a trait, whether it's, you know, disease -risked in humans or meat -yield in livestock.

And once we identify those genes, we can actively change the population through selection.

And we can predict exactly how fast a population will evolve using a remarkably simple formula from the chapter.

Oh, right.

R equals H squared times S.

Exactly.

So essentially, narrow -sense heritability acts like a speed limit for evolution.

Okay, let's break that formula down for you.

The response to selection R equals R H squared multiplied by the selection differential, S.

And S is just the difference between the specific parents we choose to breed and the general population they originally came from.

Let's walk through the math on this because it's really satisfying.

Okay, imagine a population of fruit flies that have an average of 35 .3 bristles on their abdomens.

We want to breed hairier flies.

So we intentionally select only the hairiest parent from that group who happened to have an average of 40 .6 bristles.

So first, we find the selection differential, the S.

We take the selected parents at 40 .6, subtract the general population average at 35 .3, and we get an S of 5 .3.

Next, we need the heritability speed limit.

From previous regression data, we know the narrow -sense heritability for bristle number in this specific environment is 0 .52.

So we just multiply them.

0 .52 times our differential of 5 .3 gives us a persisted response, R, of 2 .8.

There you go.

That means the very next generation of fruit flies will have, on average, 2 .8 more bristles than the original population.

Right.

And if heritability was lower, that response number would drop.

Progress would be a crawl no matter how selectively we bred them.

It is literally a mathematical formula for predicting evolution.

But nature always has a complication, right?

Oh, always.

Yeah.

We have to watch out for pleiotropy.

Pleiotropy.

This is when one single gene influences multiple, seemingly unrelated traits.

Selecting for one trait almost always pulls on invisible strings that alter something else.

Like an unintended side effect.

Exactly.

For example, in the dairy industry, selecting cattle strictly for higher milk yield actually decreases the percentage of butterfat in the milk, because the underlying genetics are inversely linked.

You pull one string, and the whole web moves.

It is incredible how interconnected it all is.

I mean, from Gregor Mendel's simple binary P's to the massive statistical equations governing how an entire species adapts.

And that interconnected web is exactly what we should be thinking about as we move forward.

Think about GWAS and how easily we can now scan massive populations for SMPs linked to complex traits like yield or behavior or disease resistance.

If pleiotropy guarantees that pulling one genetic string always moves another,

what of the unseen consequences when we aggressively artificially select for a single desirable trait in our agriculture or our livestock?

Are we solving one crisis only to inadvertently breed a totally unforeseen vulnerability into our global food supply?

That is a deeply provocative thought to leave you with.

We've gone from the looming challenge of feeding 9 .8 billion people, mapped the missing phenylnene in corn oil, and waded into the deep end of continuous traits, standard deviations, and narrow sense heritability.

You now have the fundamental statistical toolkit to understand how traits are passed down in the real messy world.

Thank you for taking this journey into quantitative genetics with us, and a warm thank you from the Last Minute Lecture Team for tuning in.

Keep questioning the data, and we'll see you next time.