Generating Evolutionary Trees (S)

Loading...

From the course by Duke University

Introduction to Genetics and Evolution

537 ratings

Duke University

Introduction to Genetics and Evolution

537 ratings

Introduction to Genetics and Evolution is a college-level class being offered simultaneously to new students at Duke University. The course gives interested people a very basic overview of some principles behind these very fundamental areas of biology. We often hear about new "genome sequences," commercial kits that can tell you about your ancestry (including pre-human) from your DNA or disease predispositions, debates about the truth of evolution, why animals behave the way they do, and how people found "genetic evidence for natural selection." This course provides the basic biology you need to understand all of these issues better, tries to clarify some misconceptions, and tries to prepare students for future, more advanced coursework in Biology (and especially evolutionary genetics). No prior coursework is assumed.

From the lesson

Speciation and Phylogenetics

This module gets into the nitty gritty of what causes the formation of new species, and how evolutionary relationships between species are inferred.

Reading Evolutionary Trees (G)18:49

Generating Evolutionary Trees (S)17:46

Phylogenetic Comparative Method (S)8:18

Meet the Instructors

Dr. Mohamed Noor
Earl D. McLean Professor and Chair,
Biology

Hello and welcome back to Introduction to Genetics and Evolution.

In this set of videos, we are looking at evolutionary trees and

the study of phylogenetics, or the relationships among organisms.

In the preceding video, we talked about how to read an evolutionary tree.

In this video we're gonna look at generating evolutionary trees.

And I'll give you the opportunity to generate your own evolutionary tree

using some DNA sequence data by the end of this video.

So let's get started.

What do we need to generate an evolutionary tree.

Well we need to look at a set of traits and

decide on similarity among the organisms being studied.

So this is a set of four organisms that I showed you in the preceding video.

So what traits should we use?

Well I pictured just a random set of four traits and

you'll see that the answer isn't totally straight forward of what we do.

That we can look at the exterior and

say hairy versus having feathers versus smooth.

Well how would this categorize these?

Well this one is hairy, the one's got feathers,

this one's smooth, this one's hairy.

So we should say the spider and the guinea big are more closely related.

That is a little bit funny.

Let's look at the others.

Which ones have a back bone and which does not?

Having a back bone would be that these two have a backbone, these two do not.

That might be okay.

What about something else?

What about number of legs?

Well here we have eight, here we have six, here we have two, here we have four.

Well that's not useful at all.

And the last thing is what about presence or absence of wings?

Well that was the example I showed you in the preceding video,

that these two have wings, and those two did not.

So you can see, it's not clear what to do when you're looking at a single character.

And you can, in fact, be misled when looking at single characters.

So, the way this is addressed in practice is that we try to study

the weight of evidence, especially by looking at multiple characters, all right.

Then we're looking at the number of characteristics

that favor one set of potential evolutionary relationships versus another.

And I'm gonna talk about this first.

Then we're gonna come back to looking at new versus old characters,

and I'll just make a quick comment at the end about some more in-depth models.

So, with respect to similarity,

Here's the same set of animals that we looked at before.

And here I've laid out a bunch of different characteristics, and

these are ones that are potentially useful.

What do we see when we look at this?

Well, Taxon two which is the ladybug here and

Taxon three being the wood duck, they share the fact that they both have wings.

Okay.

In contrast, the spider and guinea pig do not.

For all the other characters, you'll notice that Taxon three and

four are similar.

These are both, these both have an endoskeleton,

their skeleton's inside their body, whereas these both have an exoskeleton.

These two are warm-blooded, whereas these are cold-blooded, but

that's a little bit of a cheat that I'm not going to talk about in this video.

These two have closed circulatory systems, these two have open.

These two have simple eyes, whereas these two have complex eyes.

And these two have a backbone whereas these do not.

So, on balance, when you look at the weight of all the evidence here.

You see that there is a strong relationship between Taxon three and

four as being similar, and one and

two being different from them, even though this category of wings throws that off.

But I have to emphasize that similarity is not enough.

You can't just make a good evolutionary tree based on similarity alone.

Instead, we end up having to look at new versus old characters.

Characters that evolved fairly recently versus ones that may have been retained

since the most recent common ancestor.

Now let me show you why this is true.

Similarity alone can be misleading when you're looking at old or

reputatively ancestral characters,

characters that have been around potentially since the root of the tree.

Now let's imagine your trait is body color, and

let's say the two body colors that we have here are black and red.

And let's say the ancestor was black, and let's say the color of the branch that I

put on here is the phenotype or the appearance that was present at the time.

And we have a new mutations in red that arose right here in the lineage

leading to C and D, okay.

If you were to go by similarity alone, C and

D are very closely related to each other because they are both red.

And then, A, B, E and F are all closely related, well is that true?

Look at the tree for a second, is that true?

The answer is partially.

It is true that C and D are closely related, as you can see over here, but A,

B, E and F are not particularly closely related.

You see that, in fact, B is more similar to C than B is to A.

Because the most recent common ancestor of B and C is right here,

whereas B and A is further back.

So this works for new or derived traits, traits that have arisen

during the evolutionary time of the tree that you're looking at.

This now assumes that changes are infrequent, and again,

this tends to work just for the newly evolved or derived trait.

Now, how can we leverage this to actually study evolutionary relationships?

The fact this only works for

the derived traits in identifying which are most closely related.

So this leads to what's referred to as the parsimony method for tree-building.

Basically what we're gonna do is we're gonna compare the traits

to what we believe the ancestor would have had in terms of the trait.

Now typically speaking the ancestor not usually still around,

we don't usually get to look at them.

So what people do in practice is they look at what's referred to as an out group.

So, for example let's say you were looking at phylogeny of insects.

Let's say these were all different insects.

The out group you might use might be for example, a spider.

So something that's similar but clearly not an insect and so it's just outside.

Using the spider, and looking at all these insects, you would then try to identify

which forms for each trait are ancestral, or basically have been around

since that common ancestor and might, for example, be present in spiders as well.

Or if they're derived, a trait that has risen

within the evolutionary history of insects that you're trying to study.

So what you're gonna do then is we're going to construct a tree

using the minimum number of changes from the ancestral form to derived form.

Right the minimum number of possible changes, you'll be using for actually

looking at relationships, you'll be using what we refer to as shared derived traits.

The technical term for this is synapomorphies, but

don't worry about that.

You're looking for shared derived traits.

In this tree right here, the ancestral trait is black, right?

Black body color.

The derived trait is red.

So, which individuals share the derived trait?

The derived trait is red, the individuals that share it are Taxon C and Taxon D.

So the assumption here is that the simplest tree

is correct in terms of these derived characters.

I'm looking specifically at those.

So that places C and D as being close relatives.

This tree also uses the minimum number of changes because it only requires

one change.

So what would have happened instead?

So we have A-B-C-D-E-F and you see there's the common ancestor.

What would happen if instead C was more closely related to F, for example?

Well, let's take a look at what that tree would have to look like.

Well there it is.

Now C is most closely related to F, D is more closely related to E.

I just changed the tree around.

This in contrast to the proceeding one requires two changes, right?

We have to have a mutation from the ancestral form of black to red,

in the lineage leading to C.

And mutation from black to red in the lineage leading to D.

This requires two mutations now.

Why is that?

What if the mutation was way back over here somewhere.

Well with mutation, what happened way back here, E and F would also be red.

But instead we just have D and C being red, so this now requires two mutations.

So this is clearly not the most parsimonious tree.

Now I won't say this tree is not possible.

We don't necessarily know the simplest answers always correct.

But it's not the simplest possible tree.

Now the important thing to keep in mind here is

all of this is showing the fact that C and

D have the shared derived character is that they are more closely related.

It actually says nothing about the rest of the evolutionary tree.

So what about A, B, E and F?

Well, we can move them someplace else.

Now I put E and F over here, and A and B over here.

This is actually just as parsimonious as the first tree I showed you.

So all you're able to infer from looking at all this is C and D

are close relatives and you don't know the relative relationships of A, B, E and F.

So ultimately, making a tree using the parsimony method requires that you have to

use multiple characters.

Or if you are using, say,

DNA sequence you'll use multiple nucleotide positions within a DNA sequence

to place all the species you're studying onto the tree, okay.

Cuz you can only look at those shared derived characters.

So let's do an example.

So this is a pretty straightforward example where we have an out group

species, right, and we have its set of characters for a particular trait.

We're going to use the out group species to infer which

form of each trait is ancestral or derived, okay.

We have four species, let's go ahead and draw a tree, based on what we've got here.

We already know that the out group is more distantly related,

that is just defined as such.

When we look at these different traits,

what do we see in terms of these shared derived characters?

That is the most important thing we are looking for, shared derived characters.

For trait number four, the derived character is only found in species A.

So it's not shared.

It's not a shared derived character,

it's just a derived character found in species A.

That's actually not useful for phylogenic inference at all.

Let's draw a line through that.

We'll actually come back to that in just a second, but it's not useful for

phylogenic inference.

Trait number three, what do we see?

The derived character is two legs, cuz the ancestral form might be four.

And we see is that all four species share that.

That's actually not particularly useful either for phylogenic inference.

So we might say right here there's a transition in trait three,

from having four legs to two legs.

We don't actually know it's right there,

could be the link is going to the out group too.

But just for ease of argument, let's just put it right there.

So that doesn't really help us a whole lot.

What about trait two?

Well here, the out group has four arms.

Species D also has four arms but these three have two arms.

So we have a derived form, two arms, and we have it shared by species A, B and C.

So given it's shared by A, B and

C that would infer that D is more distantly related.

It does not share the derived character.

We have a transition point right here for

trait two, going from four arms to two arms in taxon A, B, and C.

For the last one here,

our out group has four eyes, species D has four eyes, and species B has four eyes.

But A and C share the derived character.

So in that case we can say B comes out over here.

And A and C are the closest relatives.

So we can place in here that extra transition for trait one right there.

So this shows all the transitions so.

Here's the point in time right there on the evolutionary tree where

trait one transitions from four eyes to two eyes.

Here's where trait two transitioned from four arms to two arms.

Here's where trait three from four legs to two legs, possibly.

It could be on this other branch.

And then, I mentioned trait four over here.

We can go ahead and place it on there.

That would be right there.

It doesn't particularly help evolutionary inference,

but we can place it on the tree nonetheless.

So, what would happen if we add a character

that doesn't agree with the other characters.

Hm.

Well here's trait five.

When we look at this we have the out group having no wings and A and

D share the derived form.

A and D were not sister species as you saw before.

Let's look at the possible alternative trees here and

let's place all the characters on there and see how many transitions it takes.

The left version of this has the derived version of trait five

arising only a single time.

Here it is, trait five arises once and it clusters in A and B.

Or I'm sorry, A and D, I said that wrong.

So what does this do to the other things?

Well, because of this if we assume that this one is right and the others

are wrong, then traits one and two have to have arisen twice somewhere in the tree.

It can't have just arisen once,

because if we assume trait five is correctly reflecting evolutionary history,

then traits one and two have to have arisen twice.

In contrast, the tree on the right assumes that first tree that we saw, and

this is assuming that trait five arose twice.

So here's trait five once, and here's trait five a second time, and

that's what makes A and D similar to each other in that context.

But here we have traits one, two, three, and four all arising only a single time.

So, how many changes do each of these require?

Let me clear all these circles so you can see a little bit better.

Well over here we have one, two, three, four, five, six,

seven changes, right, to make this evolutionary tree.

In contrast, this one over here requires only six changes.

So which of these is more parsimonious?

The answer is the one on the right is slightly more

parsimonious because it only involves six changes, rather than seven changes.

Now the way I did this, there are other ways you could actually make the same tree

if you have changes going, for example, to one form and then back.

But no matter how you do it, this tree is slightly more parsimonious because

there's no way to make the trees that would involve fewer than six changes.

So let me give one for you to try.

Here's one for you to try.

This now, I've changed a little bit so

now this involves DNA sequences instead of wings or no wings, four legs or two legs.

But the principle is exactly the same.

Now you might be slightly panicked.

What! How do I do this?

So let me break it down for you.

First thing you do is you just look at what are the sites

that are variable, right?

Well these are the sites that are variable on the tree, or on the sequence.

So we have traits, this is equivalent to the same trait one, two,

three, four, and five.

So that's pretty straight forward and what you do is you're comparing the out group

to the four species and you're seeing which of them share the derived form.

And then from that I want you to make an evolutionary tree, so

why don't you pause the video for just a second.

Try to construct a tree and then I'll give you three choices and

then I want you to do the in video quiz to try each of the three choices.

See which choice it is.

Okay, so go ahead and stop the video for a second and construct a tree.

Hopefully you've all ready stopped the video, I'll go ahead and

show you the three choices now and

then you have to vote which of the choices it is in the in video quiz.

Here are the three possible choices.

I hope that wasn't too difficult.

So, again, what are we trying to do?

We're looking at the shared derived forms.

So, I've highlighted here the derived forms for each of these things.

And as you can see, trait two is not useful because the out group has C.

Everybody else has C except for species B.

So this does not have a shared derived form so we can just ignore this one.

For the others, trait five is also not particularly useful because all it has is

the out group difference from the three non-out group species.

So,we don't really need this either.

The ones that are more useful are traits one, three and four.

So who shares the derived forms?

Well for trait one, A and D share the derived form so

we can draw something that clusters A and D, right.

For trait four, again A and D show the derived form so again A and D.

So if we're mark on the tree, this is where trait one changes and

this is also where trait four changes.

And then what happens with trait three?

Well that clusters B and C together.

So this would be the evolutionary tree that we'd expect.

That B and C are clustered together by trait three, A and

D are clustered together by traits one and four.

And theoretically if we were to put the other things on there, this is where trait

five would have transitioned or potentially the out group lineage.

Right before trait, right before species B,

you'd have the trait two one if you were to place it all on there.

Essentially B and C are sisters, A and D are sisters.

The question to ask is, is the answer right?

Well, the real answer is that we don't know.

Parsimony doesn't always result in a tree

that accurately reflects what happened over evolutionary history.

Because you can have similarity in some traits that doesn't reflect ancestry,

but reflects a convergence, that you had two mutations that happened to go to

A at the same site, that made them become more similar.

That's often referred to as convergence,

and those traits are often referred to as homo plastic.

You basically can't have these shared derived traits evolving more than once,

and it's something that there's no way around.

Now, the way people have tried to address this is to use various model-based

methods, such as maximum likelihood methods or basing methods, and

those often do better, but they're still not always correct.

They're not 100% correct, and different genes may give you different trees.

The next video will talk about an application of phylogenetic trees for

testing evolutionary hypotheses.

Under the assumption that your tree is correct, or

at least very likely to be correct.

Thank you.

I hope you enjoyed this video.

Explore our Catalog

Join for free and get personalized recommendations, updates and offers.

Coursera provides universal access to the world’s best education, partnering with top universities and organizations to offer courses online.

© 2018 Coursera Inc. All rights reserved.

Download on the App Store

Get it on Google Play