In Building Cladograms you grouped species by shared derived traits. Now we build trees from tables of data, of traits or of DNA differences, and we pick between possible trees with a rule called parsimony. Molecular clocks can then date the branch points.
Key idea
From a table of differences, the pair with the fewest differences shares the most recent common ancestor; join them first, then add the next closest. With traits, group by shared derived traits. When several trees are possible, parsimony picks the one that needs the fewest changes. Every tree is a hypothesis.
| Human | Chimpanzee | Gorilla | Orangutan | |
|---|---|---|---|---|
| Human | — | 1 | 2 | 4 |
| Chimpanzee | — | 2 | 4 | |
| Gorilla | — | 4 | ||
| Orangutan | — |
Why it works
Why trust the fewest changes? Each change of a trait or a base is an event. A tree that needs a change to happen twice, independently, in two branches is possible (it is called convergence), but less likely than one where it happened once in a common ancestor. Parsimony is a way of choosing the simplest explanation of the data. It can be wrong, and other methods weigh how likely each kind of change is, which is why scientists test trees with more data.
Worked examples
Worked example: Joining from a table
Use the table above to build the tree.
- Fewest differences: human and chimpanzee (1). Join them.
- Next: gorilla differs from both by 2. Join gorilla to that pair.
- Orangutan differs from all by 4: it branched off first.
Answer: Orangutan first, then gorilla, then human and chimpanzee as closest relatives.
Worked example: Counting changes
Four plants: moss (no tubes, no seeds), fern (tubes, no seeds), pine (tubes, seeds), rose (tubes, seeds). Tree 1: (moss, (fern, (pine, rose))). Tree 2: (fern, (moss, (pine, rose))). How many changes does each need, if all started without tubes or seeds?
- Tree 1: tubes appear once (before fern), seeds once (before pine and rose): 2 changes.
- Tree 2: tubes must appear in fern, and separately in the pine-rose branch (or appear early and be lost in moss): 3 changes.
- Parsimony prefers tree 1.
Answer: Tree 1 needs 2 changes; tree 2 needs 3. Tree 1 is the more parsimonious hypothesis.
Common mistake
Thinking a tree built from data is final. Trees are hypotheses: new genes, new fossils or better methods can move a branch. Also check how you read a tree: closeness is about the most recent common ancestor, not about which tips sit next to each other on the page, since branches can rotate.