--- title: "Distance and Likelihood Trees, and What Branch Support Means" description: "Sequence trees can be fitted to pairwise distance estimates or inferred by maximum likelihood under a substitution model. Ordinary bootstrap support reports how often resampled alignments recover a sp" canonical: https://lightmysky.com/learn/science/distance-and-likelihood-trees-and-what-branch-support-means-mt_mEO9PUEdh6 source: https://lightmysky.com/learn/science/distance-and-likelihood-trees-and-what-branch-support-means-mt_mEO9PUEdh6.md retrieved: 2026-11-12 --- > **Agent view.** This is the Markdown twin of the page, for tools and assistants. > When to use this site, and the call that answers each job: https://lightmysky.com/agent-instructions.md > API description (OpenAPI 3.1): https://lightmysky.com/openapi.json · Authentication: https://lightmysky.com/auth.md > Pricing: https://lightmysky.com/pricing.md · Catalog: https://lightmysky.com/llms.txt · Full catalog: https://lightmysky.com/llms-full.txt > Every machine-readable file on this domain: https://lightmysky.com/.well-known/ai-catalog.json > Ask for Markdown with `Accept: text/markdown`, a `.md` address, or `?mode=agent`. # Distance and Likelihood Trees, and What Branch Support Means Sequence trees can be fitted to pairwise distance estimates or inferred by maximum likelihood under a substitution model. Ordinary bootstrap support reports how often resampled alignments recover a split. Other support measures require their own interpretation. Subject: Science · Area: Genetics & Evolution · Ages 20 to 21 Page: https://lightmysky.com/learn/science/distance-and-likelihood-trees-and-what-branch-support-means-mt_mEO9PUEdh6 ## Ready when they can - Explains how a substitution model accounts for changes that raw sequence differences can hide. - Explains ordinary bootstrap resampling and reads its support percentage as split recovery across replicate trees. - Identifies a weakly supported split in a reported result and proposes a check suited to a stated limitation. ## Lesson: Reading trees without fooling yourself You have compared trees by counting character changes. For sequence trees, compare homologous sites in an alignment. Present-day differences can hide repeated substitutions or reversals: A can change to G and back to A. A substitution model estimates change under assumptions. Models may allow unequal base frequencies and rates among sites; they do not reveal the exact history. *(drawing: A changes to G, then back to A. Comparing the endpoints records no difference.)* Distance methods fit a tree to pairwise dissimilarity estimates from aligned sequences. Distances may be raw or model-corrected, not alignment gaps. Maximum likelihood scores the observed alignment under candidate trees and a substitution model, fitting branch lengths and other parameters. Parsimony minimizes inferred changes. Cutting a branch divides the taxa into two sets: a split. For trees with the same taxa, match splits by these sets, not their position in the drawing. For an ordinary sequence bootstrap, sample alignment columns with replacement. Each sampled column keeps all taxa together. Make each replicate the original length. Columns may repeat or be absent; every column can also appear once. Rebuild a tree from each replicate using the same procedure. A split's bootstrap percentage is the fraction of replicate trees containing it, not the probability that the split is true. **Example.** Suppose a split appears in 38 of 100 ordinary bootstrap trees. Its support is 38 percent. Most replicates do not recover it, so this result does not justify treating that relationship as resolved. Running more replicates reuses the same alignment sites; it adds no sequence evidence. *(drawing: The bars count bootstrap results. They do not draw a tree or assign probabilities to its truth.)* Weak support can reflect limited information, conflicting signals, alignment problems or poor model fit. Choose a response after checking the cause. Additional informative sites may help a short alignment; useful additional taxa may help separate long branches. Check alignment quality and plausible models too. No response guarantees stronger support. Repeated changes can make raw similarity misleading, but unsuitable model assumptions can also mislead. When methods disagree, investigate the data and assumptions rather than trusting a method automatically. Even high bootstrap support can repeat a biased result. Check how support was calculated: other statistics, including posterior probabilities, are not ordinary bootstrap frequencies. **Recap.** Distances summarize sequence differences. Likelihood scores the alignment under a tree and model. Ordinary bootstrap support counts split recovery in resampled alignments; it does not prove a split. *Lesson, questions and study cards written for this dataset.* ## Practice 8 questions on this page, each with its working shown. ## Needs first - [Building a Phylogeny: Homology, Shared Derived Characters and Parsimony](https://lightmysky.com/learn/science/building-a-phylogeny-homology-shared-derived-characters-and-parsimony-mt_GFmiVZ8g-F) ## Optional connections - [Sequence alignment and reading a search result](https://lightmysky.com/learn/science/sequence-alignment-and-reading-a-search-result-mt_jNbEMFqgZ5) ## Opens up - [Models of Sequence Evolution and Maximum Likelihood Trees](https://lightmysky.com/learn/science/models-of-sequence-evolution-and-maximum-likelihood-trees-mt_Ujtg0xul9N)