Biology 162
Discussion section
Week 10
PHYLOGENY
Introduction
The principle aim of phylogenetic systematics (or cladistics) is to discover the ‘tree of life.’
This amounts to finding natural groups within the multitude of species present on Earth (present
and past). The mere statement of this objective leads to three questions:
1. What are natural groups?
2. Of what biological interest is the discovery of natural groups?
3. How do we go about determining the natural groups to which organisms belong?
In this exercise, we will dwell mostly on Question 3, but 1 and 2 are worth briefly discussing.
You will also have the opportunity to find natural groups among a collection of buttons.
What are natural groups?
The ancient Greeks posited that all things were defined by certain properties reflective of
their true ‘essence’. Items sharing essential properties were grouped together. This allowed a
hierarchical organization based on common essential properties. Groups were nested within
larger, more inclusive groups. This Aristotelian concept of ‘natural groups’ applies to all things,
biological or otherwise. For example, my Ford Ranger XLT can be classified as follows:
all motor vehicles
all Ford vehicles
all Ford Rangers
all Ford Ranger XLTs
my Ford Ranger XLT
This example illustrates a major pointnatural groups are subjective constructs based on
which characters are deemed essential. A Ford Ranger is a pickup truck. Is ‘purpose’ a more
essential quality than ‘brand’? What about the distinction between two-wheeled vehicles and
four-wheeled vehicles? Shouldn’t the classification scheme put pickups and cars in a group
separate from motorcycles? Note also that any of these groups could be subdivided further. I
might choose to include my Ranger in a group composed of all XLTs made in the same plant on
the same day if I thought those were important considerations.
Finally, remember that natural groups are characterized by shared properties. Imagine a
group composed of all vehicles except Ford Rangers. Would that be a natural group? No, since
that group would be composed of vehicles lacking the qualities of Rangersa group defined by
what it is not. The shorthand for natural and unnatural groups is X groups and not-X groups,
respectively.
Of what biological interest are natural groups?
As Aristotle sought the natural order of all things, he also worked to classify biological
diversity. The practice of taxonomy predates widespread acceptance of biological evolution. It is
interesting to mention in this context that in 1859 (the same year in which Darwin published The
Origin of Species) the great naturalist Louis Agassiz built Harvard’s Museum of Comparative
Zoology immediately adjacent to The Divinity Schoolhe saw the role of taxonomy as one of
classifying the products of God’s creation. Since Darwin, however, the outlook of taxonomists
has shifted: rather than serving as a mere tally of the Creator’s handiwork, biological
classification should reflect the history of evolution. In other words, natural groups should
embody the evolutionary relationships between organisms.
In a biological classification, the things that need classifying are species. Species can be
grouped according to their shared essential properties. Consider the following classification:
all vertebrates
all vertebrates with jaws
all jawed vertebrates with four axial limbs (i.e., tetrapods)
all tetrapods with an amniotic membrane surrounding their embryos
all amniotes with mammary glands and hair
all mammals with a marsupium
the Kangaroo
This biological classification is exactly analogous to the vehicle classification given above.
The Kangaroo is included among a group that includes all other marsupials, and that group, in
turn, is nested within a larger group composed of all mammals. These groups are defined by
essential properties, but the shared characters that unite them are not arbitrary. For example, all
mammals have mammary glands and hair. According to Aristotle’s way of thinking, these
animals all belong in the same group because they share those essential properties. But, why do
they share those properties? Because of evolution, because of descent with modification. At one
time, hundreds of millions of years ago, there was the first mammal, a single species presumably
bearing mammary glands and hair. It is from that ancestral species that all of the present
diversity of mammals evolved. You have inherited the hair on your head from your parents who
inherited it from their parents, etc. all the way back to that first species of mammal. Mammals
are a natural group sharing the essential properties of hair and mammary glands because those
species all evolved from a common ancestor with those properties.
How do we determine natural groups in biology?
The methods used in phylogenetic systematics to discover the pattern of evolution are based
on the assumption of descent with modification. Traits (characters) are passed on to descendants
unless modified or lost. Evolutionary innovations will define natural groups in biology. Two
closely related species should share more characters than distantly related organisms. This is
perhaps best understood by working through an example.
Figure 1- A cladogram for four species.
The cladogram in Figure 1 shows four species and their evolutionary relationships. Species C
is more closely related to D than it is to either B or A—C and D share a more recent common
ancestor with each other than either does with B or A. The common ancestor of C and D is
represented by the junction of their branches. Thus, C and D form a natural group—a group
composed of an ancestor and all of its descendant species. Do A and B form a natural group?
Where is their common ancestor?
The example in Figure 1 is only concerned with a single essential propertycolor. A and B
are white, C and D are black. The blow-up shows a schematic of individual relationships within
the population ancestral to B, C, and D. Each square or circle (males and females, respectively)
represents an individual; generations run horizontally, lines connecting individuals indicate
either mating or offspring-parent relationships. Although time and population size have been
compressed, the diagram illustrates a speciation event. At the beginning of this sequence of
generations, there was a single population—a single, randomly mating collection of individuals.
Where the cladogram branches, this single population has divided into two (a number of
different mechanisms could be responsible). The result is two new lineages, each with their own
evolutionary trajectory distinct from that of their common ancestor. The same model could be
applied to any branch-point on a cladogram.
So, how does this explain why C and D are black? Consider again the ancestor of B, C, and
D—it is white. This is because there is some gene, or constellation of genes, causing individuals
to express the color white. After the speciation event shown in Figure 1, a mutant black form
appeared in one of the lineages, spread throughout the population, and became fixed (i.e., all
individuals in the population ancestral to C and D are black). Although C and D have since
diverged from their common ancestor, they are both black because they share a common black
ancestor. B is still white because the ‘black gene’ arose after B and C/D split.
In looking at this one character, color, we have determined that it has two states, white and
black, and that white is ancestral and black is derived. That is, the original color was white, and it
was modified to be black. As was already stated above, C and D form a natural group, but now
we can also see that this group is defined by the essential property (i.e., shared, derived
character), black. Natural biological groups are defined only by shared derived characters.
Possession of the ancestral character is due solely to lacking the derived character. To say that A
and B are “white” is the same as saying they are “not black”. How does this relate are X groups
and not-X groups?
For these same four species, we can look at even more characters. Figure 2 shows the
distribution of ten characters on the same cladogram. Each numbered tick along a branch
indicates a derived character shared by all the species above it: characters 1 and 2 are shared by
B, C, and D, etc. Because we know the phylogeny of these four species, we know that for each
character the ancestral state is to not have the character, and the derived state is to express the
character (Is this always the case? Think about animals that live in caves...). The cladogram
diagrammatically shows the natural groups nested within A-D and the shared derived characters
that define them. We can also express this information in tabular form (Table 1).
Figure 2- A cladogram showing the origin of various characters among four species.
Species
A
B
C
D
1
no
yes
yes
yes
2
no
yes
yes
yes
3
no
yes
no
no
4
no
yes
no
no
5
no
no
yes
yes
Table 1- The distribution of characters among four species.
6
no
no
yes
yes
7
no
no
yes
no
8
no
no
yes
no
9
no
no
no
yes
10
no
no
no
yes
Now let’s work the process from the perspective of not knowing the phylogeny of B, C, and
D (we’ll ignore A for the time being). This is the situation that biologists face when trying to
determine the evolutionary history of a species. Although we don’t know how they are related,
we can determine the distribution of the ten different characters among them by examination.
This gives us Table 1. We can use these characters to test hypotheses of relationship.
While we are first examining specimens of B, C, and D, we will not know which character
states are ancestral or which are derived. For example, C and D both share character 5 whereas B
lacks it. There are two alternative hypotheses here: (1) lacking 5 is the ancestral state and having
5 is a derived character defining the group C/D, or (2) having 5 is the ancestral state and lacking
5 represents a loss of that character in B. Since we don’t want to use ancestral characters to
define our groups, it is important to resolve this problem. This is where species A comes in—we
can use it as an outgroup. Simply put, an outgroup is a reference species that is known or
suspected to be outside of the group of species in question—the “ingroup” species (i.e., B, C, and
D) share a more recent common ancestor with each other than any does with the outgroup. If the
outgroup is properly chosen, it can be used to determine the polarity of character change: for any
character, the ancestral state is the one observed in the outgroup. Since A lacks 5, possession of
that character state is the derived condition. After doing this for all of our characters, we will
have the a data set just like that seen in Table 1.
There are three possible phylogenies for B, C, and D. These cladograms are shown in Figure
3 with the ten characters traced on them. All three of these cladograms have B, C, and D as a
natural group defined by characters 1 and 2, but they differ in which two of the three species are
more closely related. How do we choose among these hypotheses?
Figure 3- Three possible cladograms.
We are looking for the arrangement of species and characters that best corroborates our
earlier hypotheses of character transformation. In the first cladogram, C and D form a natural
group, and this hypothesis is corroborated by shared derived characters 5 and 6. But what about
the second cladogram? Here B and D form a natural group, but it is not defined by any of the
characters in our data matrix. Moreover, that cladogram hypothesizes that both characters 5 and
6 arose independently in species C and D. That is, C and D do not both exhibit 5 and 6 because
they are derived from a common ancestor but because the same characters evolved more than
once. This is called convergent evolution or convergence. For example, birds, bats, and
pterodactyls all have wings derived from their forelimbs, but is that a shared derived character
among those taxa?
The best-corroborated hypothesis is the one that minimizes convergence, the cladogram that
requires the least ‘amount’ of evolution. For our purposes, we can ‘quantify’ evolution by
counting the number of steps each cladogram requires. When we trace the characters on the
cladograms, which one has the fewest number of steps? Since the second and third cladograms
both require characters 5 and 6 to each evolve twice, they should be longer (12 steps each).
Therefore, the first cladogram is the shortest (10 steps). This method of choosing among
alternative phylogenetic hypotheses is known as parsimony analysis.
Before I give a practical example of parsimony analysis using real taxa, two further points
should be emphasized. Firstly, as the number of taxa in your analysis increases, so does the
number of alternative hypotheses—as few as ten species requires analysis of over 2 million
cladograms! Therefore, it more often more practical to use this method to choose among
previously stated hypotheses of phylogeny. Secondly, you will notice that many of the characters
we employed for this study were uninformative: 3, 4, 7, 8, 9, and 10 did nothing to contribute to
our parsimony analysis. Any derived character that occurs in only one species or in all species
considered is uninformative and serves only to complicate the counting and tracing steps in the
procedure.
A Practical Example: Do fishes comprise a natural group?
Let’s consider an example using some familiar organisms—do fishes comprise a natural
group? What we are really asking here is, “Do fish species share a more recent common ancestor
with each other than they do with other vertebrates?” For the time being, we can hypothesize that
the answer to that question is, “Yes.” To test that hypothesis, we need five things: (1) an ingroup,
(2) an outgroup, (3) a data matrix of characters, (4) a cladogram with fishes as a natural group,
and (5) at least one alternative cladogram (i.e., one without fishes as a natural group). Our
choices at each of these steps will affect the outcome of the analysis, so a 6th component is also
necessary: a basic knowledge of the species and characters we are studying.
Our ingroup is comprised of those species whose relationships we wish to determine. In this
case, we are interested in fishes and their relationships to other vertebrates. Thus, we want
representatives of the major groups of fishes: lamprey, shark, tuna, and lungfish. We will also
include members of some other vertebrate groups as well: frog, gecko (a lizard), kangaroo, and
human. Our outgroup should be a close relative of vertebrates, but a species that is clearly not a
vertebrate. For our purposes, a sea star will serve nicely.
The characters we choose to examine are of utmost importance. They should all be
informative characters that do not occur in the outgroup and that might define natural groups
within the ingroup. For this example we will consider six characters: (1) presence of gills in the
adult, (2) presence of jaws, (3) presence of a bony skeleton, (4) presence of four limbs arranged
axially, (5) possession of an amniotic egg (i.e., the embryo is surrounded by an amniotic
membrane), and (6) presence of mammary glands. The distribution of these characters among the
eight ingroup and one outgroup species is shown in Table 2. Because the outgroup lacks all six
characters, it is reasonable for us to hypothesize, at this time, that the ancestral condition for each
is the absence of the character.
Species
sea star
lamprey
shark
tuna
lungfish
frog
gecko
mouse
human
adult
gills?
no
yes
yes
yes
yes
no
no
no
no
jaws?
no
no
yes
yes
yes
yes
yes
yes
yes
bony
skeleton?
no
no
no
yes
yes
yes
yes
yes
yes
four axial
limbs?
no
no
no
no
yes
yes
yes
yes
yes
amniotic
egg?
no
no
no
no
no
no
yes
yes
yes
mammar
y glands?
no
no
no
no
no
no
no
yes
yes
Table 2. The distribution of characters among the sea star and representative vertebrates.
Figure 4- Two hypotheses for the evolution of fishes.
The cladogram illustrated in Figure 4A depicts fishes as a natural group. After tracing our six
characters onto this cladogram, we can see that the group ‘fishes’ is hypothesized to be defined
by a single derived character: adult gills. It is also evident, however, that this hypothesis requires
convergent evolution of jaws, a bony skeleton, and the tetrapod limb arrangement. The overall
length of this cladogram is nine steps. Can you think of a cladogram that is fewer steps but that
maintains fishes as a natural group?
Look at the alternative cladogram shown in Figure 4B. Here fishes are not a natural group.
This hypothesis also depicts a nested hierarchy of each of the six characters and, at the same
time, a arranges our species into nested, natural groups defined by shared derived characters.
This arrangement, however, is not in complete agreement with our earlier hypotheses of
character evolution. It requires that the absence of adult gills in the frog, gecko, kangaroo, and
human to be a shared derived character rather than retention of the ancestral condition. The
overall length of this cladogram is seven steps.
These two opposing hypothesis paint different pictures of vertebrate evolution. How do we
choose among them? The best-corroborated hypothesis is the one that minimizes convergent
evolution, the one that is parsimonious (i.e., requires the fewest steps). The best-corroborated
hypothesis should also agree with other lines of evidence, such as embryology.
The cladogram in Figure 4B is the shorter of the two (in fact, it is the shortest tree of the
135,135 possible arrangements of these species). Figure 4B requires no convergence among
jaws, bony skeleton, or tetrapod limbs but does suggest that the frog, gecko, kangaroo, and
human have lost their adult gills. That is called a character reversal. Can you see how a character
reversal (as shown in Figure 4B) is different from retention of the ancestral condition (as in
Figure 4A)? Frogs, geckos, kangaroos, and humans all do have gills but at earlier stages in their
life cycle—the frog has gills as a tadpole, and the other three pass through a gilled stage as
embryos. Further supporting the cladogram in Figure 4B is the developmental reality that jaws
are derived from gills.
This analysis supports the cladogram in Figure 4B over that in 4A. In fact, the hypothesis that
fishes are not a natural group is the one accepted by most vertebrate biologists today. As the
cladogram shows, vertebrates with four axial limbs (which includes lungfish!) evolved from a
fish ancestor. So, fishes, by definition, are not a natural group since not all the descendants of
their common ancestor are included. Can you think of any other vertebrate groups that are
potentially unnatural? From what sorts of animals did mammals and birds evolve? What
characters could you use to test your answer?
Button Phylogeny
Today, you will work together to investigate “the evolution of buttons”. Each group will
receive a set of buttons to examine thoroughly. Your goal is to determine the evolutionary
relationships between those buttons using the phylogenetic methods described in the previous
section. First, spread all the buttons out and look for similar characteristics.
Which buttons do you think comprise a natural group?
Now, phrase your guess in the form of a phylogenetic hypothesis:
To test your hypothesis, you need five things:
(1) Ingroup. What is your ingroup?
(2) Outgroup. Your outgroup should be a close relative of the buttons, but clearly not a button.
Today, your outgroup will be a Penny.
(3) Characters.
Your next step is to choose informative button characters. Remember, you should choose
characters that might define natural groups and do not occur in the outgroup. It is also important
that you choose characters that are discrete and not continuous. Discrete characters are present in
distinct forms instead of along a gradient. Examples of discrete characters are wings (because
they are either present or absent) and number of legs (because there are either 2, 4, 6, etc., but not
5_). We refer to the different forms of discrete characters as character states.
What are some informative button characters? What are the character states?
Decide as a class which button characters you will use to test your hypotheses. Fill in
those characters across the top row of the Character Matrix (next page). Then, fill in the
character state for each button.
After filling in the Character Matrix, use the Shared Derived Character Matrix to help
determine which buttons are most similar to each other. Fill in the matrix by counting the
number of derived character states shared by two buttons. Derived character states are traits not
found in the ancestor. In this case, we have used the Penny (the outgroup) to represent the
ancestor of the buttons, so none of its traits are derived, they are all ancestral.
Based on your matrices, which shared derived characters might define natural groups?
(4) Cladogram.
Now that you have filled in the matrices, draw a cladogram (A) that would support your
original hypothesis about which buttons comprise a natural group. Remember, a natural group is
defined by a shared derived character, and contains all the descendents of a common ancestor.
The bar at the base of the cladogram is where you can list the ancestral character states present in
the Penny. As you draw your cladogram, you should place a bar between each branch to indicate
what derived character(s) the buttons above that point share.
How many steps are in your cladogram? Does your cladogram require any convergent evolution?
(5) Alternative Cladogram.
You should also draw at least one more cladogram as an alternative hypothesis. Using the
patterns of shared derived characters from your matrices, draw a cladogram (B) depicting natural
groups different from your original hypothesis. Again, fill in the derived characters between the
branches along the cladogram.
How many steps are in this cladogram? Does this cladogram require any convergent evolution?
Character Matrix
Characters
Buttons
Penny
(Outgroup)
1
2
3
4
5
6
7
8
9
10
Shared Derived Character Matrix
Buttons
Penny
(Outgroup)
1
2
3
4
5
6
7
8
9
10
Penny
(Outgroup)
X
0
0
0
0
0
0
0
0
0
0
1
0
X
2
0
3
0
4
0
5
0
6
0
7
0
8
0
9
0
10
0
X
X
X
X
X
X
X
X
X
Cladograms – Two hypotheses for “the evolution of buttons”
A
B
Questions
1. Did either hypothesis contain a character reversal?
2. Which hypothesis is more parsimonious? How do you know?
3. Which hypothesis is better supported? Based on what evidence?
4. In your best-supported cladogram, which buttons comprise natural groups?
What characters define those natural groups?
5. In your best-supported hypothesis, did any characters turn out to be uninformative? Why?
6. Which cladogram in the class is best-supported? Based on what evidence?
7. What are other alternative hypotheses you could test?