Chapter 28: Complex and Quantitative Traits

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, when you think about traits like everyday things, your height, maybe the shape of your nose, even how fast you metabolize your morning coffee.

They're not just simple on -off switches, are they?

You're not just tall or short.

There's a whole range.

Exactly.

It's not like having attached earlobes or not.

These things are continuous.

They're complex.

And in genetics, we call them complex traits or sometimes quantitative traits.

Quantitative, meaning you can measure them, put a number on them.

Precisely.

Height in inches or centimeters, weight in pounds or kilos.

That quantitative bit is really important because it signals we need different tools to study them than simple Mendelian traits.

And that's really what we're diving into today, isn't it?

How do we get a handle on these traits that are shaped by, well, a whole bunch of genes and by the environment around us?

Right.

It's a mix.

And understanding that mix is huge, whether you're talking about improving crops in agriculture or trying to figure out the roots of human diseases like diabetes or heart disease.

So for this deep dive, we're leaning heavily on a fantastic resource, Robert Brooker's Genetics, Analysis and Principles.

We want to give you a solid grasp of this complex topic, drawing straight from the experts.

It's a great text.

Really lays out the foundations clearly.

So let's get into it.

OK, let's start with the basics.

What does complex really mean in this genetic context?

Well, fundamentally, it means the trait is influenced by multiple genes.

We often call that polygenic, poly meaning many.

And critically, it's also significantly shaped by environmental factors.

It's that combination.

So not just one gene, one outcome.

Rarely for these kinds of traits.

And quantitative, like we said, just means it can be described numerically.

Height, weight, metabolic rate, even things you can count, like the number of bristles on a fruit flies back.

And these are just everywhere.

Once you start looking for them, you mentioned height, weight.

What else?

Oh, loads of things.

In humans, besides the obvious physical traits, think about predisposition to many common diseases, heart disease,

many cancers, diabetes.

These are classic complex traits.

Your risk isn't determined by one disease gene, usually, but by the interplay of many genes and your lifestyle, your environment.

That makes sense.

And in agriculture, this must be absolutely central.

Hugely.

Farmers are constantly selecting for quantitative traits.

Think about milk yield in cows, the weight of pigs,

fruit size in plants, disease resistance.

These are all measurable, complex traits that directly impact food production.

Even things like how fast a cheetah can run or how strong a tree branch is.

Evolutionary adaptations often involve quantitative traits, speed, strength, camouflage patterns.

These are traits that vary along a continuum and have been shaped by natural selection over long periods.

They allow species to adapt.

So understanding these isn't just, you know, an academic exercise.

It has real world importance.

Absolutely.

Better crops, understanding human health risks, figuring out how evolution actually works.

Yeah.

It all comes back to quantitative genetics.

You also mentioned these traits fall into a few categories, continuous, moristic.

So continuous traits are the ones that show a whole range of phenotypes, like height.

There aren't distinct steps.

Moristic traits are ones you can count in whole numbers, like the number of petals on a flower or scales on a fish.

Still quantitative because you're counting.

And the third one sounded particularly interesting.

Threshold traits.

Yes.

Threshold traits.

These are really relevant for diseases.

Think of them as being polygenic, influenced by many genes.

But they only appear if an individual crosses a certain threshold.

Meaning?

Meaning you need to accumulate a certain number of contributing factors, both the genetic risk alleles and potentially environmental triggers before the trait.

Often a disease like type 2 diabetes or schizophrenia actually manifests.

Below that threshold, you might carry some risk factors, but not show the condition.

Okay.

So if these traits exist on a spectrum, how do scientists actually describe this variation?

You can't just make neat little categories.

You can't.

So what we do is create a frequency distribution.

You measure the trait in a population, say height, in a group of students.

Then you divide the range of measurements into convenient intervals, like maybe one -inch increments.

And then you just count how many individuals fall into each little bin.

Exactly.

You plot the number of individuals against the measured trait value for each bin.

And very often, especially with large sample sizes.

You get that bell curve shape, right?

Yeah, the famous normal distribution.

It's the symmetrical bell -shaped curve.

It pops up a lot because it's the natural result when a trait is influenced by many small independent factors, which is precisely the case for most complex polygenic traits plus environmental influences.

Okay.

So describing the variation makes sense, but analyzing it.

If we can't use those simple 3 .1 Mendelian ratios, what do we use?

This sounds like where statistics comes in.

You nailed it.

The math gets a bit more involved.

We need statistical tools.

This really goes back to pioneers like Francis Galton and Carl Pearson around the turn of the 20th century.

They basically founded the field of biometrics, applying stats to biology.

And some of their basic tools are still fundamental today, I assume, like the mean.

Absolutely.

The mean is just the average.

Sum up all the measurements and divide by the number of individuals.

If your corn ears average 15 centimeters long, that's your mean.

Simple starting point.

But the average doesn't tell you about the spread, does it?

Some ears might be tiny, some huge.

Exactly.

For that, we need variance.

It measures how spread out the data points are around that mean.

It's calculated based on the square differences between each measurement and the mean.

Square differences, okay.

The key thing about variance, and this is super important in genetics, is that it's often additive.

So the total phenotypic variance you see, Vp, can often be broken down, roughly, into the variance caused by genetic differences, Vg, and the variance caused by environmental differences, Ve.

So Vp, Avg, plus Ve.

Ah, so you can start to tease apart the nature and nurture components statistically.

That's the idea.

It's a powerful concept for understanding the relative contributions.

And variances in squared units, which can be a bit weird to think about.

Is there something more intuitive?

Yes.

The standard deviation, Sd, it's simply the square root of the variance.

So it brings the measure of spread back into the original units, like centimeters or kilograms.

Makes it easier to visualize.

Right.

And it relates directly to the normal distribution.

About 68 % of individuals fall within one standard deviation of the mean, 95 % within two, and almost everyone, 99 .7 % within three.

So if you know the mean in Sd, you have a really good picture of the distribution.

OK,

so mean, variance, standard deviation help us describe one trait.

But what if we want to see if two traits are related?

Like, do taller people tend to weigh more?

Good question.

That's where we use tools to look at the relationship between two variables.

First there's covariance, co, x, y.

It tells you how two traits vary together.

Do they both increase together, or does one go up while the other goes down?

OK.

And from covariance, we can calculate the correlation coefficient R.

This is a really useful single number, always between minus one and plus one.

And that tells you the strength of the relationship.

Exactly.

And the direction.

A positive R means as one trait increases, the other tends to increase, like maybe height and weight.

A negative R means as one goes up, the other goes down,

maybe like hours spent studying and number of mistakes on a test.

And zero.

Zero R means no consistent linear relationship between the two traits.

So if you found, say, a correlation of plus point four between rainfall and crop yield.

That would suggest a moderate positive relationship.

More rain tends to mean higher yield, but it's not a perfect lockstep connection.

Got it.

But here's the really important bit I always hear.

Correlation doesn't equal causation.

Right.

Serious point.

Absolutely vital.

Just because two things are correlated, even strongly correlated, it does not prove that one causes the other.

There might be a third factor influencing both, or it could be coincidence, or the causal relationship could be reversed.

So if parents and kids are correlated for, say, musical ability, we can't just jump to it's all genetic.

Exactly.

They share genes.

Yes.

But they also share environments, maybe musical upbringing, access to instruments.

Correlation shows an association is real, statistically speaking, but it doesn't tell you why it exists.

Never forget that.

OK, that's a really important caution.

So we have the stats.

Let's dive deeper into the genes themselves.

Polygenic inheritance.

Right.

This just means traits controlled by multiple genes.

And the classic foundational work here was by Herman Nilsen Ellis back in 1909 studying wheat.

Ah, yes.

The wheat hull color experiment.

This one's fascinating.

It really is.

He crossed true breeding red wheat with true breeding white wheat.

The first generation, the F1, were all intermediate in color, kind of light red.

OK, that sounds like incomplete dominance, maybe.

That's what you might think initially.

But then he self -fertilized those F1 plants.

And the next generation, the F2, wasn't just three types.

He saw a whole spectrum.

But specifically, he could classify them into five categories, from dark red all the way to white, in a very specific ratio, 1 .4 .6 .4 .1.

Whoa.

That's not a simple Mendelian ratio.

Not at all.

Nilsen Ellis brilliantly figured out that this ratio could be explained if two different genes controlled the color, and the alleles for redness had an additive effect.

Additive, meaning each red allele just adds a bit more red pigment.

Precisely.

So if you have zero red alleles, all white alleles, you get white.

One red allele gives light red, two give medium red, three give darker red, and four red alleles, two from each gene, give the darkest red.

That model perfectly explained the 1 .4 .6 .4 .1 ratio.

So the more red doses you have, the redder the wheat.

You got it.

And you can imagine if you have three genes involved, or four, or ten,

plus some environmental influence on pigment intensity.

Those distinct categories just blur together into a smooth curve like the normal distribution we talked about.

Exactly.

That's why for most quantitative traits, we don't see simple ratios.

The effects of many genes, each adding a small amount plus environmental noise, creates that continuous variation.

Trying to draw punnet squares becomes basically impossible.

OK, so if there are potentially dozens or hundreds of genes involved, how on earth do scientists find them?

It sounds like looking for needles in a haystack.

It can be.

But we have techniques.

One major approach is mapping quantitative trait loci or QTLs.

QTLs, loci just means locations on chromosomes, right?

Right.

A QTL is a specific region on a chromosome that contains one or more genes affecting the quantitative trait you're interested in.

QTL mapping is about finding those regions.

How does that work?

Is it like genetic detective work?

Kind of, yeah.

The basic strategy involves finding identifiable landmarks on the chromosomes.

We call these molecular markers.

Think of them like signposts along the DNA.

These markers, like RFLPs or SNPs, vary between individuals.

OK, so you have these genetic signposts.

Then you take two strains that are very different for the trait you care about, say tomatoes with big fruit versus tomatoes with small fruit, and they also need to differ in their marker patterns.

You cross them.

And then look at the offspring.

Yes.

You look at the offspring, often through several generations or specific crosses, like back You measure the trait, like fruit size, in each offspring, and you also determine which markers they inherited from each parent.

And you're looking for...

You're looking for markers that consistently get inherited along with the trait value.

If offspring with, say, marker variant A at a specific location nearly always have larger fruit, while those with marker B have smaller fruit, it strongly suggests that a gene affecting fruit size, a QTL, is located near that marker on the chromosome.

Because genes and markers that are close together tend to get inherited together.

Linkage.

Exactly.

Linkage is the key principle.

A great real -world example was Andrew Patterson's work in 1988.

His team used this QTL mapping approach in tomatoes.

What did they find?

They successfully identified several QTLs responsible for variation in important traits, like fruit weight, the amount of soluble solid, which affects taste, and even the acidity, pH of the tomatoes.

It was a landmark study showing this really works for finding genes underlying complex agricultural traits.

Okay, this is making sense.

We can describe the variation, we know multiple genes are involved, and we can even hunt for where those genes are.

This leads us to maybe the biggest question.

How much is genes and how much is environment?

The whole heritability issue.

Ah yes, heritability.

Probably one of the most important and often misunderstood concepts in quantitative genetics.

Let's try to get it right then.

We start with the total variation we see in a population for a trait, the phenotypic variance VP.

Right, and we want to partition that VP.

The simplest view is that it's made up of variants due to genetic differences, VG, and variants due to environmental differences, VE.

So VP can be more complex interactions.

Yes, it definitely can be.

There's genotype by environment interaction,

VGXE.

This is when the effect of the environment depends on the genotype.

For example, one plant variety might grow best in sunny conditions while another does better in shade.

The environment, sunshade, has a different effect depending on the genes.

Okay, so the environment's impact isn't uniform across all genetic types.

Exactly.

And there's also genotype environment association, VGE, where certain genotypes tend to be found in specific environments.

Think about dairy cows bred for high milk production being primarily raised on farms with rich feed.

The genotype and environment are associated.

For simplicity and introductory explanations, we often focus on VP equals VG plus VE, but it's crucial to remember these interactions and associations exist and can complicate things.

Got it.

So, assuming we can roughly use VP equals VG plus VE, how do we estimate those VG and VE components?

One common experimental approach is to use organisms where you can control the genetics.

For instance, you can study highly inbred strains where individuals are virtually genetically identical like clones.

So if they're genetically identical, any variation you see must be environmental, right?

Precisely.

If you raise an inbred mouse strain in a standard environment and measure their weight variance, VP, that variance equals the environmental variance, VE, because VG is essentially zero in that group.

Then, you take a genetically heterogeneous group of mice,

normal outbred mice with lots of genetic variation, and raise them in the exact same controlled environment.

You measure their total phenotypic variance, VP,

so if you already estimated VE from the inbred group, you can calculate VG by subtraction.

VG equals VP, heterogeneous VE.

Clever.

So you isolate the environmental part first, then figure out the genetic part.

It's a neat experimental design.

Now, once you have estimates for VG and VP, you can calculate heritability.

Which tells us what proportion of the total variation is due to genes.

Basically, yes.

But with some very important caveats.

Heritability is defined as the proportion of the total phenotypic variance, VP, that is due to genetic variance, VG.

So A trait equals VG, VP.

Okay, a number between zero and one.

Right.

If A trait is one, all the variation in that population is due to genetic differences.

If it's zero, all the variation is environmental.

Most complex traits fall somewhere in between.

What the caveat?

Huge G caveats.

Heritability is not about an individual.

It doesn't tell you your height is 80 % genetic.

It's a population measure.

It describes the sources of variation within a specific group of individuals in a specific environment.

Change the group or change the environment and the heritability estimate can change.

That's really important.

It's not a fixed biological constant.

Not at all.

It's a context -dependent statistic about variance in a group.

And you mentioned there are two kinds, broad sense and narrow sense.

Yes.

Broad sense heritability, HBWAR, uses the total genetic variance, VG, in the numerator.

HBWAR equals VG, VP.

This includes all types of genetic effects, additive, dominance, epistasis, gene interactions.

But often, especially in animal and plant breeding, we're more interested in narrow sense heritability, HNANI.

This only considers the additive genetic variance, VA, in the numerator.

HNWAR equals VA, VP.

Why focus just on the additive part?

Because additive effects are the ones that reliably get passed down from parent to offspring and cause offspring to resemble their parents.

Dominance and epistatic effects depend on specific combinations of alleles that get shuffled during reproduction.

So narrow sense heritability is much better for predicting how a trait will respond to selection in a breeding program.

Ah, predictability.

That makes sense for breeders.

How do they estimate this narrow sense heritability?

One way is by looking at the correlation between relatives.

You measure the trait in related individuals, say parents and offspring, or siblings.

You calculate the observed correlation coefficient, ROBS.

The deck we talked about earlier with covariance.

Exactly.

Then you compare that observed correlation to the expected correlation based purely on their known genetic relationship.

For example, a parent and child share exactly half their genes, so their expected correlation due to additive genetics is 0 .5.

Okay.

Narrow sense heritability can then be estimated as the ratio.

HN40P equals ROBS -REx.

So if the observed correlation for, say, egg weight between mother hens and their daughters, ROBS is 0 .3, then HN -DONLESS equals 0 .30 .5, you're at 0 .60.

Meaning 60 % of the variation in egg weight in that flock is due to additive genetic differences.

In that specific flock and environment, yes.

It suggests selection for egg weight would likely be effective.

Is there a good human example of estimating heritability this way?

Oh, a classic one.

The Sarah Holt study of human fingerprint ridge counts back in the mid -20th century.

It's a fascinating trait because it's set early in development and doesn't change.

Fingerprints seem pretty unique.

How does she quantify them?

She developed a method to count the number of dermal ridges on all 10 fingers, giving a single quantitative score for each person.

And she compared relatives.

Yes.

She looked at correlations between identical twins who share 100 % of their genes, fraternal twins who share about 50%, like regular siblings, siblings, parents, and children, and unrelated individuals.

What did she find?

The results were striking.

The correlation for identical twins was incredibly high, around 0 .95.

For fraternal twins, it was much lower, around 0 .49, close to the 0 .5 expected for siblings.

Parent -child correlation was also around 0 .48.

Wow, that's a big difference between identical and fraternal twins.

Exactly.

Since both types of twins typically share very similar environments growing up, that huge difference strongly points towards genetics being the major driver of variation in ridge count.

The overall Narrow Sense Heritability Estimate was calculated to be about 0 .97.

Almost entirely genetic variation determining the differences between people's fingerprint patterns in that population.

It's one of the highest heritability estimates for any human trait, a really compelling case for the power of genetics in shaping a complex trait.

Okay, so we understand variation,

polygenic inheritance, QTLs, heritability.

How has humanity used this understanding to actively change traits over time?

This must be where selective breeding comes in.

Absolutely.

Selective breeding or artificial selection is basically humans taking the wheel of evolution for a specific species.

We choose which individuals get to reproduce based on traits we find desirable.

And people have been doing this for ages, right?

Long before they knew about genes.

Millennia.

It even inspired Charles Darwin.

He was fascinated by pigeon fanciers and how they created wildly different breeds from a common ancestor just by selecting for specific features over generations.

The only difference from natural selection is human choice dictates reproductive success, not survival or mating success in the wild.

And the results can be pretty staggering.

You mentioned pigeons, dogs are another amazing example.

Oh, doys are incredible.

All domesticated dogs descended from the gray wolf.

Yet look at the diversity.

Tiny bulldogs, massive Great Danes.

It's purely the result of humans selecting for different traits like size, temperament, coat type.

We now know specific genes like the AGF -1 gene play a big role in size differences and selection acted on alleles of those genes.

And plants too.

I remember seeing that diagram of wild mustard.

Brassica oleracea, yes.

It's the wild ancestor of an amazing variety of vegetables.

By selecting for large terminal buds, we got cabbage.

Selecting for lateral buds gave us Brussels sprouts, flower clusters,

broccoli and cauliflower,

leaves, kale, swollen stems,

kohlrabi, all from selectively emphasizing different quantitative traits in the same original plant.

That's incredible.

So the basic mechanism is just pick the best parents.

Pick the parents that show the traits you want, let them breed and hope their offspring inherit those advantageous alleles.

Over generations, the frequency of those alleles increases in the population and the trait changes.

Is there a long -term experiment showing this clearly?

The Illinois corn oil content experiment is a classic.

Started way back in 1896.

They selected two lines of corn.

One for high oil content in the kernels and one for low oil content.

And they just kept selecting the highest oil plants for one line and the lowest for the other year after year.

For over a century, though the key results were clear after about 77 generations.

The starting population had around 4 -6 % oil content.

The high line eventually reached over 18 % all, while the low line dropped to less than 1%.

A massive divergence driven purely by selection.

But does selection work forever?

Can you just keep selecting for higher and higher oil content indefinitely?

No.

Eventually, you usually hit a selection limit or a plateau.

Why does it stop?

Two main reasons.

One, you might simply run out of genetic variation.

If selection has made the population essentially homozygous for all the desirable alleles affecting the trait, there's no more genetic fuel for selection to act upon.

Everyone has the best genes already.

Makes sense.

What's the other reason?

Sometimes, the alleles you're selecting for might have negative side effects on overall fitness, especially when they reach high frequencies or are selected to extreme levels.

For instance, selecting for extremely rapid growth might lead to skeletal problems or reduce fertility, counteracting the benefits of the trait you wanted.

Natural selection might start working against your artificial selection.

So there's a limit.

Now, you mentioned heritability helps predict the response to selection.

Can selection experiments also tell you something about heritability?

Yes, they can.

This gives us what's called realized heritability.

It's based on the actual response observed after a round of selection.

How does that work?

We use the Breeders' Equation.

R is HN -MUTTER -S -R is the response to selection.

How much the average trade value changed in the offspring compared to the original population's average.

S is the selection differential.

How different the parents you selected were from the original population average.

So R is what you got.

S is what you asked for by choosing the parents.

Exactly.

And HN -NOTO, narrow sense heritability, is the link between them.

If you measure R and S, you can calculate the realized heritability.

HN -NOTO -WGULS -R -S.

It tells you how much of the selection pressure actually translated into a response in the next generation.

So breeders can use this to estimate heritability directly from their breeding programs.

Yep.

And they can also use it predictively.

If they know the narrow sense heritability of a trait, maybe from previous experiments or relative studies, and they decide how strongly they want to select their S, they can predict the response R they're likely to get in the next generation.

Very practical.

Very cool.

Okay.

One last fascinating phenomenon related to breeding,

heterosis, also called hybrid vigor.

That's right.

Heterosis is when you cross two different inbred strains.

And the resulting hybrid offspring are actually more vigorous, bigger, healthier, higher yielding than either of the parental inbred strains.

So the hybrid is somehow better than both parents.

That seems counterintuitive.

It does.

But it's a very real and widely used phenomenon, especially in agriculture.

Think about hybrid corn or hybrid rice.

They often have significant yield advantages, sometimes 15, 20 % or more over the inbred lines.

Why does this happen?

What's the genetic explanation?

There are two main hypotheses and both likely contribute depending on the trait and species.

The first is the dominance hypothesis.

Inbred lines, because they're highly homozygous, tend to accumulate slightly harmful recessive alleles.

Let's say line A is homozygous A -B -B -C -C -D -D.

And line B is A -A -B -B -C -C -D -D, where lowercase letters are slightly detrimental recessives.

Each line suffers a bit from its homozygous recessives.

But when you cross them, the hybrid offspring are A -A -B -B -C -C -D -D.

They're heterozygous for all these genes.

The detrimental effects of the recessive alleles inherited from one parent are masked by the functional dominant alleles from the other parent.

So the hybrid avoids the negative effects plaguing both parent lines.

Ah, heterozygosity masks the bad stuff.

Makes sense.

What's the other idea?

The overdominance hypothesis.

This suggests that for certain genes, the heterozygote, say A1A2, is simply intrinsically more vigorous or performs better than either homozygote A1A1 or A2A2.

Being heterozygous itself confers an advantage.

So not just masking bad alleles, but actually getting a bonus from having two different alleles.

Exactly.

Evidence exists for both hypotheses.

Dominance seems well supported in studies of rhizoterozis, while some studies on corn yield point towards overdominance, playing a significant role for certain QTLs.

It's likely often a combination of both effects contributing to that hybrid figure.

Wow.

Okay, so putting this all together, this deep dive has really shown how these complex quantitative traits are just fundamental to understanding biology.

There's this fascinating interplay of multiple genes and the environment.

And we have tools to describe them statistically, to map the genes involved using QTL analysis, to estimate the relative contributions of genetics and environment through heritability.

And even to actively shape these traits through selective breeding, harnessing concepts like realized heritability and hybrid vigor, it touches everything from agriculture to evolution to human health.

It really does.

It highlights how variation, driven by genetics and modulated by the environment, is the raw material for so much of the biological diversity we see.

So let's leave our listeners with a final thought.

We've talked about height, crop yield, even fingerprint ridges.

But what about incredibly complex human traits, things like intelligence or aspects of personality?

How might these concepts we've discussed, additive genetic variance, those tricky genotype environment interactions, maybe even selection limits apply to understanding the variation we see in those characteristics?

What does thinking about genetics in this quantitative way tell us about the possibilities and maybe more importantly, the inherent limitations when we consider influencing such deeply complex human traits?

That's a really profound question to ponder, isn't it?

It forces us to grapple with the sheer complexity involved and the ethical considerations that inevitably arise.

It reminds us that even traits we experience every day are the result of incredibly intricate biological processes we're still working to fully understand.

Lots to think about there.

Thank you for being part of the Deep Dive family.

We look forward to delving into more fascinating topics with you next time.