Cladistics: An Introduction

Cladograms are the heart of paleontology in these opening years of the Third Millennium. The theory and practice of these diagrams is a subject which would fill many pages. In fact it does fill many pages on this site; and the questions get very technical indeed. Here, we present only a basic introduction.

Cladograms are simply diagrams which show how species, or groups of species, are interrelated. They look like this:

Notice that we don't have to put in every group. Presumably there are other tetrapods besides dinosaurs, monkeys and you. Nor do we have to name every group.  For example, there is some taxon that unites you and monkeys to the exclusion of dinosaurs (e.g., Primates or Mammalia), but it is represented simply by  the '+' in the diagram. It's the relative position that is important.

This cladogram can be written several ways. The most inclusive group always goes at the top, by convention. However, we could describe precisely the same set of relationships like this:

In both diagrams, to go from the first vertebtrate to you, we have to split off sharks, go through Tetrapoda, split off dinosaurs and split off monkeys --  in exactly that order. So, we are presenting the same evolutionary sequence, however it is displayed. The path between taxa counts, not where they fall on the page.

For reasons we will not get into here, each group is normally expected to diverge into precisely two others. However, we are often unsure (and unwilling to guess) exactly what the sequence was. Thus, you will frequently encounter:

Absent some unexpected miracle of genetics, you do not represent an evolutionary advance over Tansen or Mozart.  (As to Chopin -- perhaps the less said the better.)  The diagram simply indicates that, for phylogenetic purposes, you, Mozart, Chopin and Tansen are all the same distance from monkeys, dinosaurs, etc. and that all of you are more closely related to each other than to any of the other animals on the list (...with the possible exception of Chopin).

So why do we use cladograms? Cladograms do not necessarily require a "cladistic" view of the world. On the other hand, the cladogram does focus on phylogeny -- how different groups relate to one another. In that sense, cladograms differ a good deal from the usual Linnaean block diagrams. Let's look at a very simple example, a high-level description of the ornithischian dinosaurs in a Linnaean format:

Note that every group has a "rank:"  order, family, genus, species, and innumerable gradations in between. Much breath and paper has been wasted arguing the appropriate rank of a group. What does a rank indicate?  It cannot be a measure of internal similarity. There is no possible scientific answer to questions such as: "are hypsilophodonts more like each other than ceratopsians are like other ceratopsians?"  Even if the question had an answer, it would tell us nothing. All sauropods, for example, are very much alike. Yet no one would argue that they are a mere family. Nor is rank a measure of diversity. Many families have only one member. Others have a hundred or more.

Perhaps a more significant problem is that the scheme doesn't tell us about the evolution or relationships of these groups. The current best guess, for example, is that all ornithischians are descended from "fabrosaurs" (the reason for the quotation marks will be explained later). The hypsilophodonts and heterodontosaurs are independently derived from the ornithopod stem. The iguanodonts may or may not be specialized hypsilophodonts, but the hadrosaurs are definitely of the Iguanodontia. The Linnaean diagram may be good systematics, that is, a reasonable classification scheme. But it tells us nothing about phylogeny. It would be equally valid, and less misleading, to sort them in alphabetical order or by the zodiacal sign of the date of first publication.

Another way of stating this problem is that the Linnaean formalism doesn't tell us whether these taxa are natural groups or just a man-made assortment of similar organisms. Linnaean taxa are defined, if at all, by a list of characteristics. This is sometimes referred to as an "apomorphy-based definition" [1]. What if another, unrelated, organism were found with the same characteristics, or (as is more usually the case) it turns out that a Linnaean taxon contains organisms that have arrived at the same condition by convergent evolution?  Conversely, if, for example, hadrosaurs are descended from iguanodonts, where do we draw the line?  Ouranosaurus has characteristics which are somewhat intermediate between Iguanodon and "advanced" hadrosaurs. In the Linnaean scheme, we can call it an iguanodont or a hadrosaur, depending on exactly what characteristics we use to define these mutually exclusive terms. But, in doing so, we are arbitrarily putting things in boxes we ourselves have built and labeled. We are not discussing natural groups, but human constructs. Nor are we making testable scientific statements.

The cladistic view of the same group (with some additions) may be represented by the cladogram:

Here, there are no ranks. The phylogenetic relationships are open and obvious. They may not be correct. But we can see what they are posited to be and can challenge the cladogram with evidence. It is testable, unlike a Linnaean scheme. The reason it is testable is that each taxon should although even scientists are sometimes sloppy about this) have an explicit phylogenetic definition. The Ornithischia are "all dinosaurs more closely related to Triceratops than to birds." The phrase "more closely related to" is used in the following special sense:  a dinosaur is more "closely related to" Triceratops than to birds if its last common ancestor with Triceratops is more recent than its last common ancestor with birds. For example, consider Ouranosaurus. Its last common ancestor with Triceratops was some primitive cerapod Cerapoda being the name of the "intersection" just to the left of "Ornithopoda" in the cladogram). The last common ancestor with birds would be much further up on the diagram.

There are four important things to note about this definition: 

1) We didn't make up this group. We applied a man-made name, but it is a group which was produced by nature and defined by the actual course of evolution -- even though we may not have full knowledge of what that course was.

2) We don't know a priori who belongs in each category, or what characteristics the members of each group might have. Instead we use the tools of science to determine the answers to those questions. Taxa may be characterized by having particular physical attributes, but they are never defined by reference to these attributes. The attributes ("characters") are things we discover about the taxon.

3) There are no boxes. The Ornithischia are all dinosaurs closer to Triceratops then to birds. Not just some dinosaurs who look more like our mental preconception of what an ornithischian "ought" to be like.

4) Ornithischia is an example of a stem group: "all organisms closer to x than to y."  A second kind of group used in cladistics is the "crown group."  Usage has made this term a bit ambiguous. In its most general sense, a crown group is any group defined in the form: "the last common ancestor of x and y and all the descendants of that ancestor."  For example, Dinosauria is defined as the last common ancestor of Triceratops and birds and all of its descendants. Cladistics, at its most elegant, describes phylogeny as a series of triads consisting of a crown group and two complementary stem groups:

As in the diagram above, stem groups are usually described by the shorthand form "x > y". So, Ornithischia  =  "Triceratops > birds."  Crown groups are indicated "x + y."  Dinosauria = "Triceratops + birds."  Note that the anchor taxa (Triceratops and birds) can be changed without necessarily changing the definition. It doesn't matter what species of bird we might select, for example, if we are correct in believing that all birds have a single common ancestor.

In some cases, of course, those assumptions may be wrong. We should be careful in picking anchor taxa. However, it is much easier to remember Craniata as "hags + hagfish," rather than "Myxine + Ausktribosphenos."  Unless something is very wrong with our picture of vertebrate evolution, the two are logically equivalent definitions.

However, it is critical to remember the "all" part of the definitions. Birds are dinosaurs because Dinosauria includes all of the descendants of the last common ancestor of birds and Triceratops -- including all birds. Birds are not "descended from" dinosaurs. Many people, scientists included, are frustrated by this rule. However, it is only by rigorous application of this (and other) cladistic rules that we can be sure that we are doing science and not just making arbitrary boxes. A natural group is called a clade, and only clades are valid taxa for purposes of cladistic analysis. The ultimate reasons for this are well beyond the scope of this essay. For the moment, it will be enough to remember that the use of taxa other than clades invites confusion and ambiguity when we are generally using cladistic principles.

That is the reason that "fabrosaurs" are found only in quotation marks on this page. "Fabrosauridae" is not a clade. It is simply a collection of primitive ornithischians. The best data available today suggest that their last common ancestor was also the ancestor of some (and perhaps all) other ornithischians. Since "Fabrosauridae" does not contain all of the descendants of the ancestral "fabrosaur" it is a paraphyletic group, rather than a clade. Conversely, a group which does not contain its own last common ancestor is referred to as polyphyletic.

But how do we go about making these determinations? That is a much longer and more difficult story. In fact, it is the whole subject of the science, theory, practice and art of cladistics. Aspects of this topic are discussed at many different points in Palaeos. A more systematic introduction to the subject may be added here in due course.

For the moment, this section contains a master cladogram of the Vertebrata -- a very high level cladogram containing only about 60 of the larger taxa. This is followed by a series of intermediate-level cladograms (incomplete as of  this date) which cover the vertebrates in large chunks. For the most detailed level, see the individual units. ATW020430, last modified ATW041013.

[1] That's not quite right. An apomorphy is a characteristic unique to a natural group -- a group descended from a single ancestor who had this character. The characters listed in a Linnaean definition need not, and often do not, meet this rigorous requirement. By the way, remember that, taken by itself, an apomorphy is inherently useless when it comes to inferring relationships to organisms lacking the apomorphy. The group which shares a derived character has a synapomorphy, and we may infer that they are more closely related to each other than to others. So, for example, Mozart, Tansen, and (presumably) you enjoy good music and form a sort of musical clade. That doesn't mean that Chopin was more closely related to musical monkey than to Mozart. That's possible, of course, and perhaps even likely, but we would need to identify other specific characteristics which Chopin shared with monkeys to the exclusion of Mozart and other tetrapods -- probably a long Liszt. ATW041013, revised ATW070828.