Reading trees, common ancestors, clades and sister groups

Begin with the lineage, not the row of labels

A phylogenetic tree is a hypothesis about the history of the entities named at its tips. A tip can represent a species, a sampled population or a gene copy. Read its legend: a gene tree and a species tree do not automatically describe the same events. An internal node represents an inferred connection in that history, rather than a photograph of a known ancestral individual.

In our first model, root R connects two ancestral lineages, U and V. U gives rise to A and B; V gives rise to C and D. Following A and B backwards meets at U. Following A and C meets farther back at R. A and B therefore share a more recent common ancestor in this rooted history. A and B are sister tips; the entire A–B group and C–D group are sister groups.

A clade includes an ancestral lineage and all its descendants. U with A and B is one clade in the model. A and C alone are not: their shared ancestor R also has B and D as descendants. Groups can be nested, so being in one clade does not exclude membership in a larger clade.

Rotating the branches attached to an internal node changes their drawing order while preserving every connection. Our second layout puts D next to A on the page, but that does not make them sisters. Nor is one living tip the ancestor of another merely because it is drawn on the left. The tree has no scale of evolutionary progress, usefulness or intelligence.

A root adds a direction to the connections

A rooted tree locates an ancestral starting point for the represented history. An unrooted tree retains connections and splits but does not specify that starting point. In the unrooted quartet, one internal edge separates A/B from C/D. That split is readable without a root; a direction of descent is not.

Place a root on the internal edge and the rooted result has sister groups A/B and C/D. Place it on the edge leading to A and A becomes sister to the remaining B/C/D group. C/D still forms a clade, but A/B no longer does. Root placement therefore matters when identifying ancestors and clades.

An appropriate outgroup can help locate a root: it is a comparison lineage whose relationship places it outside the focal group. Calling an organism an outgroup is an evidence-based assumption to test, not a consequence of its looking simple or old. Rooting methods also have limitations when evolution varies among lineages or deep relationships are hard to resolve.

Ask what branch length measures

Topology means the pattern of connections. In a cladogram, drawn lengths may only organise the picture. A phylogram can instead scale lengths to inferred evolutionary change, commonly expected substitutions per sequence site. A chronogram scales them to elapsed time. These are different quantities: equal time does not require equal molecular change.

To find the path distance on a length-scaled tree, add the values along the unique path between tips. Do not count how many labels lie between them. Estimated substitutions can exceed the number of visible differences because a site may change more than once; the observed sequence records the final state, not every intervening event.

An unresolved node with more than two descendant branches is a polytomy. It can report insufficient resolution, and in some contexts a genuinely rapid or simultaneous divergence hypothesis. The drawing alone does not decide which explanation applies. Check the evidence and the analysis rather than silently turning every unresolved connection into a particular binary history.

Worked example

Check your understanding

1. A rotated diagram places D beside A. Which connection would have to change before A and D became sister tips?

   Show reasoning

   They would have to descend directly from the same immediate ancestral node. In the supplied tree A descends from U with B, while D descends from V with C. Label order alone changes none of those connections.

2. Can the unrooted A/B versus C/D split establish that A/B is a clade?

   Show reasoning

   It establishes the split, but not that clade without a root. Rooting the internal edge makes A/B a clade; rooting A’s terminal edge does not. The same unrooted connections permit both placements.

3. What is the C-to-D molecular path distance, and does it give their divergence date?

   Show reasoning

   It is 0.1 + 0.5 = 0.6 expected substitutions per site. It does not supply a date. Dating requires a time model and suitable calibration, with their uncertainties; the topology alone gives no calendar scale.

What to carry forward

Trace ancestry from rooted connections and read every branch value with its stated meaning and units.