1The parts of a diagram

• Tips (ends of branches): the species or groups being compared, usually living ones.
• Branches: lineages through time.
• Node: a branching point; the most recent common ancestor of everything that branches from it.
• Root: the common ancestor of every lineage on the diagram.
• Clade: an ancestor and all of its descendants. Any node plus everything branching from it is a clade.
• Sister groups: the two lineages that branch from one node; each is the other's closest relative on the tree.
• Out-group: the lineage that branched off first, least closely related to the rest (the in-group).

Relatedness is read from the most recent common ancestor: two species are more closely related the more recently their lineages joined at a node, not by how close their names sit at the tips.

Branches can be rotated around any node, like a mobile, without changing the relationships, so the order of the tips means nothing in itself.

2Trees with time, cladograms without

The CED draws a clear line.

A molecular clock relies on the observation that, over long periods, some DNA sequences accumulate changes at a roughly steady rate. If the rate is known from a pair of species whose split is dated by fossils, the number of differences between another pair gives an estimate of when they split.

Worked example (invented). Fossils show that two lizard lineages split 12 million years ago, and they differ at 48 positions in a gene. Two other lizards differ at 20 positions in the same gene.

The answer assumes a constant rate, which is an approximation; that is why molecular dates come with uncertainty and are checked against fossils.

3Which traits are useful: shared derived characters

A character is a heritable feature (presence of feathers, number of digits, a base at one position in a gene). It can be in its ancestral state, as in the common ancestor, or a derived state, changed in some lineage. Derived states can be gains (a new structure) or losses (snakes losing legs is a derived trait of snakes).

Only shared derived characters group species. If a change happened once in an ancestor, all its descendants inherit it, so the species that share it form a clade. A shared ancestral character is no help: all mammals have a backbone, but so do fish, so a backbone does not show that mammals are a group.

The out-group tells you which state is ancestral. Whatever state the out-group has is assumed to be the ancestral one; a different state found in some in-group species is derived.

4Building a cladogram from a trait table

An invented table for five insect species (1 = derived trait present, 0 = absent). Species V lacks all four traits, so it is taken as the out-group.

The method:

1. Find the out-group: the species with the fewest derived traits, V. It branches off first.
2. Rank the traits by how many species share them. Trait 1 is shared by four species, so it arose earliest, on the branch leading to W, X, Y and Z.
3. Work inwards. Trait 2 (X, Y, Z) arose next; trait 3 (Y, Z) after that.
4. Traits in only one species (trait 4, only Z) go on that species' own branch. They do not group anything.

Figure 1 shows the result. Each orange mark is the point where a derived trait arose; every species above that mark has the trait. Y and Z share the most recent common ancestor (the marked node), so they are each other's closest relatives. W is more closely related to Z than V is, because W and Z share an ancestor (the node just after trait 1) that V does not.

From sequence differences. The same logic works with molecular data. Given the number of differences between each pair of species, join the pair with the fewest differences first; they are sister species. Then attach the next closest species to that pair, and so on, with the most distant species as the out-group.

5Why molecular data usually win

The CED states that molecular data typically give more accurate and reliable evidence than morphology. The reasons:

• There are far more characters. One gene has hundreds or thousands of bases, each a character; a skeleton has a few dozen measurable features.
• They are less often fooled by convergence. Similar body shapes can evolve separately under similar selective pressures (topic 7.10), making unrelated species look related. Long DNA sequences rarely converge by chance in the same way.
• They can compare anything. Bacteria, fungi and whales can all be compared at the same gene; their bodies have almost nothing to compare.
• They can be counted. Differences are numbers, which can feed a molecular clock.

6Trees are hypotheses, and they change

A tree makes testable predictions: if Y and Z are sister species, a newly sequenced gene should also show them as closest. When new data disagree consistently, the tree is redrawn. A real example: whales were long placed apart from other mammals because their bodies are so changed for life in water; DNA evidence showed that their closest living relatives are hippopotamuses, and fossils of early whales with ankle bones like those of hoofed mammals supported this. Trees and cladograms illustrate the speciation events that have happened, and like any scientific model they are revised as evidence comes in.