How to read a phylogenetic tree
Reading a phylogenetic tree is mostly about three things the diagram encodes: the branching order, the branch lengths, and the support at each node. Get those and the rest follows.
Branching order, who is related to whom
Two taxa are more closely related when they share a more recent common ancestor. Trace back from each toward the root; the pair whose paths meet at the nearer node are the closer relatives. Relatedness is read through the nodes, never from how near the tip labels sit on the page.
Branch lengths, how much change
This is where a phylogenetic tree carries more than a cladogram. If the tips are ragged, ending at uneven positions, the branch lengths are meaningful, and there should be a scale bar. A long branch means a lot of change along that lineage; a short branch means little.
The scale bar is the key. It might read "0.05", meaning a branch as long as the bar represents 0.05 substitutions per site. Measure a branch against the bar to read its length. If the tips are all flush and there is no scale bar, the lengths are cosmetic, you're looking at a cladogram, and only the branching order is being claimed.
| If the tips are… | and there is… | then it's a… | and branch length means… |
|---|---|---|---|
| flush (aligned) | no scale bar | cladogram | nothing |
| ragged | a scale bar | phylogram | amount of change |
| flush | a time axis | chronogram | time |
Support values, how much to trust each branch
The numbers sitting at internal nodes are support values. They answer "how confident are we that this group is real?", and they come in two flavours:
- Bootstrap (0–100). The data are resampled many times and the tree rebuilt each time; the value is the percentage of those trees containing this exact group. Above ~70% is usually treated as decent support, above ~95% as strong.
- Posterior probability (0–1). From Bayesian analysis; 0.95 and up is strong.
A node with low support is a grouping the data can't firmly justify, treat that part of the tree as uncertain, however confident the drawing looks.
Rooted vs unrooted
A rooted tree has a defined starting point and a direction of time: you can point to the common ancestor of everything and read descent outward. An unrooted tree shows only how taxa cluster, with no "first" split, neighbour-joining, for instance, produces one until you root it, usually on an outgroup.
On a rooted tree, the root is the deepest node; everything descends from it. If a tree is drawn with a trifurcation (three branches meeting) at its base, that's the tell-tale sign it's unrooted.
Clades and monophyly
A clade is a node plus every descendant of that node, a complete branch you could snip off. Cladistically valid groups are clades (monophyletic). Groups that leave out some descendants (paraphyletic, like "reptiles" without birds) or lump unrelated lineages on a shared trait (polyphyletic) are not. The cladogram reading guide has the full breakdown of these terms, which apply identically here.
Build yours in the browser, no signup, no install.
Open the cladogram makerCommon misreadings
- Reading adjacent tips as relatives. Only the nodes encode relatedness.
- Assuming right-hand taxa are "more evolved". Every living tip has the same amount of history behind it.
- Reading branch length as time. In a phylogram it's amount of change; only a chronogram shows time.
- Ignoring low support. A crisp-looking branch with 52% bootstrap is barely supported.
- Trusting a scale-bar-free tree's lengths. No scale bar means the lengths mean nothing.