Distance and Likelihood Trees, and What Branch Support Means
Sequence-based trees are built either from pairwise distances or by asking which tree makes the observed data most probable under a model of change. Support values report how often a branch survives resampling, which is a statement about the data rather than about the truth of the branch.
What a learner can do afterwards
- Explains what a model of sequence change adds beyond counting differences.
- Reads a support value correctly as repeatability, not as probability of being right.
- Identifies a poorly supported node in a published tree and says what would improve it.
1 · Read
Counting raw differences between sequences misses hidden change. Some sites were hit twice and some changes reversed, so two sequences can look close while much change happened. A model of sequence change corrects for these hidden events, adjusting for unequal base frequencies and sites that evolve at different speeds.
Distance and likelihood methods use that correction differently. Distance methods compress sequences into pairwise gaps and then cluster them. Likelihood asks which tree makes the observed data most probable under the model. Parsimony, the older baseline, simply minimizes total steps without any model at all. Convergent and reversed changes create false similarity that only model based correction can discount.
Imagine a published tree with 42 percent support on one branch. That number means resampling recovers that split 42 times in 100, which is weak. Treat the relationship as undecided and rest no bold claim on it. The honest fix is more informative characters, denser sampling of species to break up long branches, or a model that fits the data better.
Read every support value as repeatability, never as the chance the branch is true. Strong conclusions need strong nodes, and strong nodes need evidence behind them. When you redraw the conclusion, keep only what the well supported nodes allow.
Models correct raw counts for hidden change, and support values report how repeatably the data recover a branch, not how likely it is true.
2 · Watch
Take it off screen
Where it sits
8 questions wait behind this lesson, each with its answer explained. Every answer feeds the sky: stars light as they are learned, and dim when it is time to come back.