Rooted Phylogenetic Tree & Evolutionary Cladogram
8.1 Reading the diagram #
8.1.1 Parts #
A phylogeny is a hypothesis of relationship drawn as a tree. Tips are the sampled species, populations, or genes. An internal node is a hypothetical ancestor of the tips that descend from it. A branch connects nodes; its length may mean time, amount of change, or nothing beyond connection. The root is the oldest node. An unrooted tree records connectedness and does not say which node is oldest. Sister groups share an immediate common ancestor. A clade is a node plus all its descendants. A polytomy is a node with more than two daughters: usually an unresolved split, sometimes a real simultaneous one. Those readings should not be confused.
Shared evolutionary novelty is the evidence for a group, as the next section defines it. The algorithms that align sequences beforehand are Bioinformatics. Choosing a tree by parsimony, distance, or a likelihood model is the inference step. These notes use the product. A tree is preferred because it accounts for the characters under a stated rule, and a different rule can prefer a different tree.
An outgroup, related to the group under study but outside it, polarises characters. A state shared with the outgroup is treated provisionally as ancestral; a state found only inside is treated as derived. The wrong outgroup mislabels both. Polarity is a hypothesis, tested by whether characters agree.
8.1.2 A tree is not a ladder #
Tips of a tree of living species are contemporary. None is the ancestor of the others. A chimpanzee is not an earlier human. “Primitive”, if used at all, applies to a character that has changed little, not to a whole organism. Every living species has the same time back to the ancestor it shares with any other. Some have changed more along that time. Branch length in a tree scaled to change records that. It is not a rank.
The order of tips across the page is not information. Any node can be rotated and the relationships stay. Placing humans at the right-hand end does not make them the goal of the tree. A relationship should be stated as “A and B are closer to each other than either is to C”, which survives rotation. A fossil tip is also a tip. Putting it on an internal node claims direct ancestry, which is usually stronger than the characters support. Chapter 2’s warning about transitional fossils is this rule applied to bones.
8.2 Groups and characters #
8.2.1 Monophyly, paraphyly, polyphyly #
A monophyletic group contains an ancestor and all of its descendants. It is a clade. A paraphyletic group contains an ancestor and some but not all descendants: what remains after a clade has been pulled out and named separately. A polyphyletic group gathers members by convergence. The ancestor that would unite them did not have the character used to gather them. Birds and mammals together, if united only by endothermy, are polyphyletic. That ancestor was not warm-blooded in either of their ways.
| Group | Ancestor included | All descendants | Example |
|---|---|---|---|
| Monophyletic | Yes | Yes | Birds, as a clade |
| Paraphyletic | Yes | No | “Reptiles” if birds are excluded |
| Polyphyletic | No, for the trait used | No | Bats plus birds, grouped as “winged vertebrates” |
Modern classifications prefer clades because a name then refers to one historical entity. A paraphyletic name refers to a grade, a level of organisation. It can be useful in conversation and misleading as a taxon. “Fish” excluding tetrapods is paraphyletic: a lungfish is closer to a cow than to a salmon. “Invertebrates” is everything that is not a vertebrate. The tree does not have remainders as branches.
8.2.2 Synapomorphy and homoplasy #
A synapomorphy is a shared derived character, inherited from the most recent common ancestor of the group it defines. A symplesiomorphy is a shared ancestral character. It was a synapomorphy at a deeper node, and it does not pick out a subgroup now. Vertebrae group vertebrates against a lancelet. Within vertebrates, having vertebrae does not group mammals with birds apart from sharks. An autapomorphy is derived and unique to one tip. It diagnoses that tip and groups nothing.
Homoplasy is similarity that is not homology: convergence, parallelism, or reversal. Bat wings and bird wings are homoplastic as wings and homologous as forelimbs. The character has to be specified. Characters conflict, and a tree is a choice of which similarities to treat as synapomorphies and which as homoplasy. Parsimony prefers the tree with the fewest extra origins and losses. Likelihood prefers the tree that makes the data probable under an explicit model. Synapomorphy is a status relative to a node, not a stamp on a trait. Grouping whales with fish because both swim treats a homoplasy as if it defined a clade. The ear characters that group whales with artiodactyls survive the rest of the skeleton and the sequences. The swimming does not.
8.2.3 Reptiles, birds, and the preference for clades #
Traditional reptiles — turtles, lizards, snakes, crocodilians — exclude birds. Crocodilians are more closely related to birds than to lizards and snakes. Birds and crocodilians are the living archosaurs, and dinosaurs, including birds, are on the bird side of that split. A group that keeps crocodilians and lizards and drops birds takes an ancestor and leaves out descendants. It is paraphyletic. Scales, and the absence of the novelties birds gained, are not synapomorphies of that remainder.
Classifications that prefer clades either include birds inside a redefined Reptilia or drop Reptilia as a formal group and use clade names. What is not legitimate is to treat the school class as a branch coordinate with birds and mammals. Chapter 2’s Archaeopteryx belongs here: a feathered fossil is not a reptile trying to become a bird. It is evidence that birds sit inside the group traditional language called reptiles. The same correction makes “apes” excluding humans paraphyletic (Chapter 11).
8.3 Molecules, and trees that disagree #
8.3.1 Molecular phylogenies #
Sequences are characters, once sites are aligned so that the comparison is between homologous positions. The gain is volume. The hazards are ordinary. A bad alignment pairs non-homologous sites, and every later conclusion inherits the mistake. Multiple substitutions at one site erase signal in deep divergences. Long branches can group together because each has evolved away from the ancestor, not because they share a clade. Morphology has its own convergences. A published tree is the output of decisions about data and models.
Molecules from living organisms do not, by themselves, describe the anatomy of an extinct node. Fossils place minimum ages on nodes and preserve character combinations no living species retains. A molecular date is a clock estimate (Chapter 4). A fossil date is a minimum if the fossil is securely inside the clade, and it is only as good as the identification and the stratigraphy. Disagreement between them is a problem to resolve, not a choice of the preferred number.
8.3.2 Gene trees are not species trees #
A species tree is the history of populations splitting. A gene tree is the history of copies of one piece of DNA. They differ for several reasons.
Incomplete lineage sorting is the surprise for anyone who has just learned that DNA gives the tree. If a species splits, and one daughter splits again before ancestral polymorphism has been fixed by drift, alleles can be inherited in a pattern that does not match the order of splits. The mismatch is likely when the ancestral population was large and the time between speciations was short. Humans and chimpanzees are sister groups, but a substantial minority of genes group humans with gorillas, or chimpanzees with gorillas, because sorting in that region was incomplete. More genes do not make every gene tree match the species tree. They make the disagreement measurable, and the species tree can be estimated from the distribution of gene trees.
Horizontal gene transfer inserts a gene from another lineage, common in bacteria and occasional in eukaryotes. Duplication and loss produce a similar trap: if one copy of a duplicate is lost in some species and the other copy in others, a tree of the survivors groups by which copy remains. The comparison wanted is of orthologues, copies separated by speciation, not paralogues, copies separated by duplication. Hybridisation (Chapter 7) is a further cause. A history of fusion is a network. Forcing a tree through it is a modelling choice, not a discovery that fusion did not happen.
8.4 Key terms, summary, and self-check #
Key terms. Phylogeny; root; node; clade; sister group; polytomy; outgroup; monophyly; paraphyly; polyphyly; synapomorphy; symplesiomorphy; homoplasy; incomplete lineage sorting; orthologue; paralogue.
Summary. Relationships are nodes. Living tips are not ancestors, and their order on the page is not a ranking. Monophyletic groups contain an ancestor and all its descendants; traditional reptiles excluding birds, and fish excluding tetrapods, do not. Synapomorphies are shared derived characters. Shared ancestral characters and convergences are not. Molecular trees use the same logic and depend on alignment, which Bioinformatics treats as a method. Gene trees disagree with species trees because of incomplete lineage sorting, horizontal transfer, duplication and loss, and hybridisation. A substantial minority of genes around the human–chimpanzee–gorilla node are expected to disagree, and that is a population-genetic consequence, not a sequencing failure.
Self-check.
1. You rotate a node so that humans move from the right-hand end to the middle. What changes about the relationships?
2. Define monophyly, paraphyly, and polyphyly. Which is “reptiles” if birds are left out, and why?
3. Why is “has vertebrae” useless for grouping mammals with sharks against another vertebrate, and useful for grouping vertebrates against a lancelet?
4. Give two mechanisms by which a correctly sequenced gene can still have a tree that differs from the species tree.
5. What is the difference between an orthologue and a paralogue, and why does it matter for a tree?
