Distance and Likelihood Trees, and What Branch Support Means · seed 1 · A4, ink-friendly. The answer key prints on its own page for grown-ups.
En anglais

Reading trees without fooling yourself

Science · Genetics & Evolution · ages 20-21
Name ______________________   Date ____________

  1. Which comparison correctly distinguishes distance methods from maximum likelihood?

    • Distance minimizes changes; likelihood counts recovered splits
    • Distance fits pairwise estimates; likelihood scores the alignment
    • Distance estimates fossil ages; likelihood assigns truth probabilities

  2. Why can raw differences between aligned sequences underestimate the number of substitutions?

    • Repeated substitutions can leave matching end states
    • Bootstrap resampling deletes all matching alignment sites
    • Each observed mismatch records every earlier substitution

  3. How is a replicate alignment made in an ordinary sequence bootstrap?

    • Sample alignment columns without replacement to the input length
    • Sample alignment columns with replacement to the original length
    • Shuffle each species' bases separately and retain the resulting matches

  4. An alignment contains slow-changing and fast-changing sites. Which feature can a substitution model include to represent this?

    • Different substitution rates among alignment sites
    • One substitution rate determined by the number of sampled species
    • No repeated substitutions wherever the observed bases match

  5. A paper reports that a split appears in 42 of 100 ordinary bootstrap trees. Your conclusion requires that split to be resolved. What should you do?

    • Treat the published split as resolved
    • Interpret 42 percent as the probability that the split is true
    • Report uncertainty and avoid relying on a resolved split

  6. A weakly supported split comes from a short, carefully checked alignment. Few sites distinguish the competing relationships. Which response targets this limitation?

    • Collect more reliably aligned, informative sites
    • Run extra bootstrap replicates to gain new sites
    • Redraw the branch arrangement to make the existing split look certain

  7. High ordinary bootstrap support rules out systematic error caused by a poorly fitting substitution model.

    Circle one:   True   False

  8. A raw-distance tree groups two fast-evolving lineages, but a model-based analysis separates them. Before further checks, which conclusion is justified?

    • The model-based grouping has been proved historically true
    • The raw-distance grouping is correct because matching bases cannot mislead
    • The disagreement calls for checking data and model assumptions
LightMySky · lightmysky.comW1-mt_mEO9PUEdh6-s1

Answer key

For grown-ups. Fold this page away before handing over the rest.

Reading trees without fooling yourself W1-mt_mEO9PUEdh6-s1

  1. Distance fits pairwise estimates; likelihood scores the alignment · Distance methods fit a tree to pairwise dissimilarity estimates. Maximum likelihood scores the observed alignment under trees and a model.
  2. Repeated substitutions can leave matching end states · Repeated changes can return a site to its starting base. Matching end states need not mean that no substitutions occurred.
  3. Sample alignment columns with replacement to the original length · Sample whole alignment columns with replacement. Columns may repeat or be absent, while the replicate retains the original number of columns.
  4. Different substitution rates among alignment sites · Rate variation among sites represents differences between slow and fast positions. It does not require discarding or manually downweighting every fast site.
  5. Report uncertainty and avoid relying on a resolved split · The split is absent from most replicate trees. Report its weak recovery rather than treating the required relationship as resolved.
  6. Collect more reliably aligned, informative sites · Additional informative sites target the stated shortage of evidence. More replicates reuse existing sites, and redrawing adds no evidence.
  7. False · Bootstrap analysis repeats the analysis procedure on resampled data. A biased procedure can repeatedly recover the same misleading split.
  8. The disagreement calls for checking data and model assumptions · Repeated changes can mislead raw similarity, while unsuitable assumptions can mislead model-based inference. Neither method wins automatically.
Worksheet · LightMySky