This is a lesson I did for Grade 12 IB students. The lecture part is about one hour long, including a lot of student participation, and it was followed by a 30 minute lab where they had to complete a morphological character matrix for seven animal phyla (not included here).
The goal of the lesson is to show how cladistics is connected to classification and evolution. These are links that must be made and understood by students, but are not very clear in the official syllabus structure.
Note that my lessons always include a lot of planned and improvised drawing on the board; I will note where this is necessary, but keep in mind that every slide can be supplemented by student interaction or board drawing.
We can say that classification is one of the key mental traits that humans possess. We classify objects all the time, according to size, function, colour. One month after we’re born, we start classifying behaviours and people – mommy, daddy, behaviours done by mommy, behaviours done by daddy, friend, enemy, toy, good, bad.
It should come as no surprise, then, that classification of nature is one of the oldest sciences. The history of biological classification can be said to have started with Artistotle in the 3rd century BC. He grouped animals by habitat (aquatic or terrestrial) and by whether they had blood (which basically ended up as vertebrates and invertebrates). He also came up with a system called the scala naturae, by which animal species are viewed as unchanging and occupying a certain hierarchy (humans were at the top, of course). Shortly after Aristotle, Theophrastus did the same work for plants.
The next big step in classification came with the binomial system developed by Carl von Linné (also called Linnaeus). He gave every species a name consisting of two parts: a generic name and a specific epithet. This way, new species can be described and added to a catalogue of life, with no duplications. We still follow Linnaeus’s binomial system today, although we’ve added a ton of rules and the system is not as seamless as it once was.
The next big step came with Darwin and Wallace and the discovery that species can change and evolve, and that they are all somehow linked on an evolutionary tree of life, but it took the work of Willi Hennig, pictured on the slide, to bring it all into the modern perspective. Everything we will learn in this lesson is based directly on his work.
Willi Hennig was a German entomologist, and he invented the modern fields of cladistics and phylogenetic systematics. In his landmark 1950 book, Grundzüge einer Theorie der phylogenetischen Systematik, he asked three questions that are still as relevant today as they were back then:
- What is a phylogenetic relationship? In other words, what does it mean to say that two or more organisms are related and share a common ancestor?
- How is a phylogenetic relationship established? I.e. how do we find out that organisms share a last common ancestor?
- How is knowledge of a phylogenetic relatiosnhip expressed so that misunderstandings are excluded? I.e. what is the best way to visualise this information?
We’ll start with the third questions.
This is a very fancy phylogenetic tree, but it holds only three key types of information.
Ask students to try to name them.
- Group names, like vertebrates, invertebrates, plants, etc.
- The relationships between all the groups.
- A timescale showing when each group and relationship originated, but only because this is a very fancy tree.
Names and relationships are what cladistics is all about, and the tree diagram (this is a circular tree) is the most efficient way of visualising the information.
Species are all related together, and while we do not believe in a scala naturae anymore, the concept of hierarchies is very practical. A beetle is an insect, which is an arthropod, which is an animal, which is a eukaryote. In practice, we use many more ranks – subspecies and variety, suborder and subphylum, superfamily, tribe, infraclass, and so on, but these eight, from species to domain, are the essential ones.
Names give objects an identity, and we give names a meaning. They are the only way to individually look at the millions of different species on Earth.
However, this is only true if we use rigorously defined scientific names. Take this example of Carabus granulatus, a very common European carabid beetle. We’ll go up the classification with it and see what information we can glean from the names and their definitions.
It’s a eukaryote, so we know its cells have a nucleus and mitochondria. Within the eukaryotes, it’s in the kingdom Metazoa, meaning it’s multicellular and has an extracellular matrix. It’s in the phylum Arthropoda, meaning it has a segmented body, segmented limbs, and complex eyes. It’s in the class Insecta, meaning one pair of antennae, two pairs of wings, and three pairs of legs. It’s in the order Coleoptera, meaning forewings hardened into an elytra and flight enabled by hind wings. It’s in the family Carabidae, meaning its antennae have up to 9 segments. It’s in the genus Carabus, which are defined by a similar characteristic only found in that genus, and Carabus granulatus has a combination of characteristics unique to it.
All this information was contained within this classification and the rank names and the associated definitions. By contrast, the common name doesn’t even allow you to specify which of the hundreds of thousands of beetle species you’re talking about – the best you can do is “ground beetle living in Europe”.
This is why in systematics, we only ever use scientific names. Common names might be useful for small and conspicuous groups, like birds and mammals, but as soon as you want reliability, you will need to switch to using scientific names. In the very worst cases, you have common names like “daddy long legs”, which can refer to any species of opilionid, pholcid, or crane fly. None of these are really closely related to each other, and each contains hundreds of species.
A common argument for using common names is that they’re much easier to memorise and can be more fun, but that’s just silly. Here are some fun scientific names that will stick in anybody’s head (as long as they have a sense of humour). There are no excuses for using common names.
The classification ranks are universal, and can be used for all animals and plants. On the left is another animal and on the right a pine tree. Plants use a somewhat different classification, hence why there is no mention of “gymnosperm” here.
And of course, we can’ forget the unicellular non-eukaryotes, the Bacteria and the Archaea. Here is E. coli, probably the most famous bacterium, and Pyrococcus furiosus, the most significant archaean owing to its use in PCR.Bacteria and Archaea bring up the next important factor about scientific names and classifications: they are all hypotheses, and thus are liable to change. A name may change its definition, or a new relationship might be discovered necessitating the creation of a new name and dumping an older one.
Archaea is a great example. Anybody who was in high school 10 years ago probably did not learn about a three domain system of life. I certainly didn’t. What we learned was a five kingdom system. The modern three domains of life come from the work of Carl Woese and his students and colleagues, who sequenced the ribosomal RNA of a bunch of organisms in the 1980s. They then compared how similar these RNA sequences were.
This is best drawn on the board with circles and lines.
What they found was that their organisms clustered into three distinct groups. One was the group of Bacteria. Another was the group of Eukaryota. And the third group was a novelty, and they termed them the Archaea.
Additionally, they found a very clear relationship: the Archaea sequences were much more similar to the Eukaryote sequences than they are to Bacteria sequences, so Archaea and Bacteria must be more closely related.
Just like that, they found that all life on Earth can be split into these three groups, each of which has a clear boundary.
In cladistics and phylogenetics, such a case of clearly defined groups is the ultimate goal. In many cases, we don’t yet have this. For an extreme example, take a look at the situation in the arthropods, where just about every single combination of relationship has been proposed with ample evidence and given a name.
Bacteria, Archaea, and Eukaryota are termed monophyletic groups. This means they each share a last common ancestor, and the group includes all of that ancestor’s descendants. In the diagram above, this is group I.
Monophyletic groups are the only acceptable type of grouping in phylogenetics.
There are two other types of group that can come up, and they are both wrong: the paraphyletic group and the polyphyletic group.
A paraphyletic group is one that has a last common ancestor, but that leaves out some of that ancestor’s descendants. Think of a family tree that has your grandparents, your parents, your sibling, but not you. That is paraphyletic.
Polyphyletic means you chose several groups that share no relation and put them together. So a polyphyletic group would be a family tree that just includes you and everyone in your class. No relation, and no genetic justification for grouping you together.
There are two very famous and well-known examples of a paraphyletic group and a polyphyletic group. Ask students to try and guess, they will not be able.
The famous paraphyletic group is the “Repitilia”, the reptiles. Ask students to name reptiles. Reptiles are snakes, crocodiles, lizards, and dinosaurs.
But what about birds? Birds are dinosaurs, but we never talk about them when we talk about reptiles. This means that Reptilia is paraphyletic. The correct name is Sauropsida, as this includes the last common ancestor of all these animals.
The word reptile was coined before we knew about the ancestry of birds, and has since become so deeply ingrained in our vocabulary that we can’t get rid of it.
The example of polyphyly is also a big embarrassment: Invertebrata. I’m an invertebrate palaeontologist. My field’s name is invalid. Invertebrates are all the animals, minus the Chordata, even though the Chordata evovled from invertebrates. The rest of the invertebrates are only related together in that they are animals. We basically selected all the animal phyla except one and grouped them as Invertebrata. Nonsensical.
Again, the reason for this name is old and cultural: back in the old days of classification, vertebrates were given special status, and almost all other animals were invertebrate “worms”. Eventually, the field of zoology became split between those who studied vertebrate animals, and those who studied anything else, i.e. anything that wasn’t vertebrate. And so “invertebrate zoology” was born, not out of any evolutionary consideration (this was before the days of evolution).
Ask students to guess which of these is the monophyletic. The correct answer is Carnivora, but most will answer fish, some will choose wasp.
If you answered fish, you are wrong because you and I are also fish – we are descendants of fish that evolved to walk on land and breathe air. But when we say “fish”, we only mean those lobe-finned vertebrates that remained in the oceans.
This can get especially confusing when discussing whales and dolphins. Every smartass will tell you that whales are mammals and not fish, but they ignore the fact that mammals are fish themselves.
The monophyletic group is the Carnivora. It’s confusing because “carnivore” is an ecological term that encompasses many unrelated organisms; but Carnivora is a specific term with a specific definition that groups together the last common ancestor of the bears, dogs, and cats, and all of its descendants. As the tree shows, this is a grouping with 100% statistical support.
This is best drawn on the board.
“Wasps” is another biologically ambiguous term. All wasps are in the insect order Hymenoptera. At the base of the Hymenoptera is a group called the Symphyta, the wood-boring wasps. Then you have the three big groups: the Parasitica (parasitic wasps), the Vespoidea (“true” wasps), and the Apocrita (bees). Within the Vespoidea, you additionally have the Formicidae family (ants).
Im other words, when you say “wasp” you are purposely excluding bees and ants while including animals that are not as related. Hymenoptera is a mouthful, but it’s the only accurate way of referring to all wasps, which must include ants and bees. If you want to refer to only big yellow stinging wasps, you should say vespoid.
With the theoretical background out of the way, let’s get to the most basic skill needed for understanding cladistics and phylogenetics: how to read a tree. The following are a series of small exercises in which the right answer is decided by properly analysing the tree.
First up, you have to find the last common ancestor of sponges and mushrooms, according to this diagram. The correct answer is d. The way to find it is to trace back from the mushroom or the sponge branch, and find the first node where their lines intesect.
In this diagram, a represents the last common ancestor of mushrooms. b is the last common ancestor of plants and opisthokonts (the grouping of animals and fungi), which could also be construed as the last common ancestor of all eukaryotes. c is a hypothetical point in the lineage leading up to the opisthokonts. This is where a branch leading up to a transitional organism between plants and opisthokonts would be rooted. e is the last common ancestor of sponges and mice, or the last common ancestor of all animals.
The answer is false. According to this diagram, the last common ancestor of crocodiles and birds is found at the node where crocodiles split off. The last common ancestor of crocodiles and lizards is further back, requiring you to go two nodes away rather than just one.
The answer here is c. All the trees show a grouping of E+D, then C, then B. Only in c are B and C grouped together in a clade.
Now that we know how to read a tree, let’s look at how they’re made. Phylogenetic trees are mathematical constructs, built out of masses of raw data analysed and sorted by numerous algorithms. We will not discuss these algorithms because they sometimes can get close to black magic for non-mathematicians. Suffice it to say that as a scientist, part of your job will be to choose the correct algorithm for your data and your research question.
The data itself comes mainly from two sources: molecules and morphology.
Molecules used are most often proteins. In the past, the amino acid composition of common proteins such as haemoglobin were used. With modern technology, it is now no longer a problem to use genes, RNA (usually from a ribosome), or even the entire genome of species (mitochondrial or nuclear) – what was once prohibitively expensive is now routine work you learn to do in the first year of university.
The data we need from genones is the series of A T G C, and how that series is different between our research species.
Note that there are other, more specialised molecular data that can be used, for example chemoreceptor molecules when comparing closely related organisms that use chemoreception. It all depends on your research question. When building a general tree of life though, only genetic molecules provide us with a complete dataset, so they are by far the most commonly used.
The reason why molecules are a great source of data isn’t only because there are plenty of them. It’s because of the way they evolve. Molecular evolution is a nearly neutral process – it is rarely affected by natural selection. When mutations happen, they are most often benevolent and have no effect to be selected for or against. When they do have an effect, it is most often negative and leads to death or at least non-reproduction of the individual, meaning the mutation disappears at once from the population. Only rarely is a mutation beneficial and gets selected for.
This is important for us because we need the amount of sequence changes to be representative of how long ago the species diverged from each other. Natural selection distorts the steady rhythm of neutral mutations that show this.
Of course, things are not that simple. It is obvious that different molecules will evolve at different rates, and be more or less affected by mutations. A mutation in Histone IV, a very ancient molecule responsible for packaging DNA in the chromosome, will very likely be detrimental, whereas a mutation that affects a few random proteins will likely not affect much and will be kept.
This means that we have to be careful with our choice of genes and proteins. If we eat to reconstruct the tree of all life, using Histone IV is a much wiser choice than the fast-evolving fibrinopepetides, while the opposite is true if you want to reconstruct the tree of mammals.
Taxa also evolve at different rates at the genetic level, as can be seen above. The reasons are mostly taxon-specific, and your job as the phylogeneticist is to account for any idiosyncrasies in your dataset. Genetic and molecular data cannot just be plugged into a program and expected to produce an automatic good result.
The other category of data used in phylogenetics is morphological data. These are phenotypic al characteristics – organs, hairs, physiologies, nerves, anything that can be seen and touched. In contrast with molecular data, morphological data provide the only tangible historical information on organismal biology.
However, that is a double-edged sword. Compiling morphological data involves a lot of subjectivity. What characters should be taken into account? Should we only consider presence and absence, or should we make subcategories? Should characters be given importance or evolvability scores to account for their varying complexities?
Again, these are questions with no universal answer, and where your job as a phylogeneticist involves many philosophical considerations. Even more questions come up when you eat to bring in fossil data into your dataset, since fossils preserve a highly biased set of characters. There is a very real danger in underestimating the effect of how we get and categorise morphological data. For example, analysing only the skulls of dinosaurs yields a very different phylogeny than an analysis of their skeleton, as shown in the diagram above.
At this point, it is important to stress again that phylogenetics is an attempt at recovering the historical relationship between organisms, not an attempt at seeing what is most similar to what.
This distinction must be at the forefront of any analysis, especially a morphological one. The biggest danger lurks in convergent evolution.
Take the four animals on the slide.going clockwise from top left, we have:
- A praying mantis (insect)
- A mantis shrimp (crustacean)
- Sanctacaris (stem-group arthropod)
- A mantispid (insect)
What unites these animals is that they are all descended from the same last common ancestor at the root of the Arthropoda phylum, not that they all share the same kind of raptorial appendage at the front. That’s just an adaptation that they all just happened to evolve convergently, independently of each other and their ancestry.
So as the scientist coding these animals’s characters to reconstruct arthropod relationships, you are faced with several ways to face this obvious issue:
- You can add so many other arthropods that this characteristic shared by only four animals is drowned out. This is the best way.
- You can add so many other characters that this single shared one is drowned out. This is another good way, as long as you are careful in your choice of characters.
- You can not include the character in your analysis. This is generally a bad idea, since you are making an a priori decision about what is important I’m the evolutionary history of a taxon. In this particular example, you cold justify it by the obviousness of the convergence of raptorial appendages, but in most cases, convergence is what needs to be shown through the analysis and convergence is not at all obvious, especially when you are examining lower systematic levels (families, etc.).
So before you even begin with a phylogeny, you really have to consider your potential data sources and how they can contribute to your research question, because each dataset analysed independently will most likely yield different results, as the trees of animals above show. If you are studying the history of populations, don’t bother looking past genetics. If you want to know how the last common ancestor looked like, you will need to bring in the fossil record. Morphology will give you information on whole-organism biologies and ecologies that genetics cannot rival (in most cases!).
There is no single method to solve every problem.
Finally, we’ve been talking about characters and data all this time, but what does it look like in practice?
It’s basically a giant matrix. The columns are characters, the rows are taxa. You can build one in Excel, in Notepad, or in any of the handfuls of dedicated phylogenetic programs. Depicted above is Mesquite, the most user-frienfly of the bunch.
In a typical morphological matrix, you only need 1s and 0s. 1 denotes presence of a character, 0 it’s absence. You can have multi-state characters too, in case a character comes in many shapes.
A molecular matrix is similarly multi-state, with ATGC as the four possible states (in case of a genetic matrix) or the amino acids in case of a protein matrix.