Phylogenetic Root Choice
Choosing the correct root for a phylogenetic tree is a critical step that determines the direction of evolution and the interpretation of ancestral states. If you are building trees from molecular sequences, especially for ecological or evolutionary studies, this guide provides a structured approach to selecting and validating a root. Root choice is not arbitrary: a misplaced root can invert relationships and mislead conclusions.
At a Glance
| Concept | Description |
|---|---|
| Root | The inferred common ancestor of all taxa in the tree, defines the direction of evolution. |
| Outgroup | A taxon known to be outside the ingroup, used to polarize character changes. |
| Midpoint rooting | Places the root at the point where the two most divergent lineages meet, assumes a molecular clock. |
| Unrooted tree | A tree topology without an explicit ancestor, all nodes represent common ancestors without direction. |
| Ingroup | The set of taxa of primary interest, the outgroup is chosen relative to them. |
For a deeper introduction to phylogenetic concepts, see the NCBI Bookshelf for general molecular evolution references.
Decision Criteria for Rooting Methods
Outgroup Rooting (Preferred When Possible)
Select an outgroup that is closely related to the ingroup but clearly outside it. The outgroup must share enough homology to align reliably, yet be sufficiently divergent to be placed unambiguously as sister to the ingroup. Practical criteria for a good outgroup:
- Phylogenetic position: Known from independent evidence (e.g., morphology, earlier phylogenies) to be outside the ingroup.
- Sequence similarity: Long enough alignment with minimal gaps.
- Single representative: Multiple outgroup taxa can be used but may cause long‑branch attraction if they are too divergent.
Resources like the EMBL‑EBI Training offer exercises on selecting outgroups for molecular datasets.
Midpoint Rooting (When No Reliable Outgroup Exists)
Midpoint rooting places the root at the node where the longest path between any two tips is bisected. This method assumes a roughly constant evolutionary rate (molecular clock). Use it when:
- No external outgroup is available.
- The tree is ultrametric or you have performed a clock calibration.
- You are working with a dataset of closely related species where rate variation is minimal.
Midpoint rooting is a simple default but can be misleading if substitution rates differ substantially among lineages. Test for rate heterogeneity before relying on it.
Bayesian or Likelihood Root Proposals
Some model‑based approaches (e.g., MrBayes, BEAST) allow the root position to be sampled during MCMC. This is useful when you are unsure about the outgroup and want to incorporate uncertainty. However, these methods require careful prior specification. The study in PLoS Comput Biol demonstrates how prior sensitivity affects tests of temporal signal, a key concern when rooting phylogenies with molecular dates.
Rooting in Phylogenetic Networks
When the true history includes reticulation, root choice becomes more complex. For semidirected networks (where edge direction is ambiguous), a new dissimilarity measure can help infer the root without assuming a strictly bifurcating tree. See IEEE Trans Comput Biol Bioinform for a technical solution to rooting in the presence of horizontal gene transfer or hybridization.
Practical Workflow for Root Choice
Step 1: Assemble Your Dataset
Collect sequences for your ingroup and potential outgroup taxa. For root‑associated fungal studies, primers designed for specific groups (e.g., Mucoromycotinian arbuscular mycorrhizal fungi) can target both ingroup and outgroup taxa. The work in Environ Microbiome describes such primers.
Step 2: Align and Estimate an Unrooted Tree
Use alignment software (MAFFT, MUSCLE) and a phylogeny estimator (RAxML‑NG, IQ‑TREE, MrBayes). Your output is an unrooted tree. At this stage, do not assume a root. The Galaxy Training Network provides hands‑on tutorials for this pipeline.
Step 3: Choose a Rooting Strategy Based on Data
- If a reliable outgroup is available: Add it to the alignment, rerun tree estimation, and root the tree on the branch leading to the outgroup. Confirm that the outgroup does not fall inside the ingroup (which indicates a poor outgroup choice).
- If no outgroup exists: Use midpoint rooting. Check for rate variation using a relative rate test or by inspecting branch lengths.
- If you have temporal data (e.g., ancient DNA, known divergence times): Use Bayesian rooting with a relaxed clock model. The approach described in PLoS Comput Biol addresses model sensitivity in this context.
Step 4: Validate the Root
Examine whether the root makes biological sense. For example, in a tree of orchid mycorrhizal fungi, the root should not place a generalist fungus as sister to a set of unrelated orchids if independent evidence suggests otherwise. The study on orchid mycorrhizas in the Ryukyu archipelago PubMed provides an example where root placement clarifies symbiotic specialization.
Step 5: Document and Report
Report the rooting method used, the outgroup (including accession numbers), and any tests that supported the choice. Provide the unrooted and rooted trees as supplementary files.
Common Mistakes
- Using a distant or misidentified outgroup. A highly divergent outgroup attracts long branches and can cause the tree to be incorrectly rooted. Check that the outgroup is not a member of the ingroup by running a separate phylogenetic analysis with the outgroup included and verifying its placement.
- Assuming midpoint rooting is always safe. Rate heterogeneity among lineages will produce a biased root. Always test for a molecular clock using a likelihood ratio test or Bayesian model comparison.
- Ignoring sequence platform effects. The choice of sequencing platform can affect the composition of root‑associated microbial communities and, consequently, the phylogenetic signal. Read about these effects in J Microbiol.
- Rooting a network as if it were a tree. When your data contain recombination or hybridization, forcing a rooted tree may lead to false conclusions. Use network‑aware methods.
Limits and Uncertainty in Root Interpretation
A root is always an inference, never an absolute fact. Uncertainty arises from:
- Missing taxa: Without a proper outgroup, many root positions may be nearly equally likely.
- Model misspecification: Substitution models that ignore rate variation across sites or lineages can shift the inferred root.
- Horizontal gene transfer: In microbial phylogenies, gene trees may have different roots than the species tree.
- Temporal calibration: Root dating requires fossil or biogeographic constraints, and these carry their own uncertainties.
Interpret your results within the bounds of the data. If two different rooting methods give conflicting answers, present both and discuss the evidence for each. Do not overstate confidence.
Frequently Asked Questions
Q: Can I root a tree using a single sequence as an outgroup? A: Yes, but it is risky. A single outgroup can be subject to long‑branch attraction. Use at least two outgroup taxa from different lineages if possible, and perform a jackknife or bootstrap to see if the root is stable.
Q: What do I do if my outgroup falls inside the ingroup? A: This indicates the outgroup is actually part of the ingroup or the alignment is corrupt. Check for contamination or misidentification. Redefine the ingroup or choose a different outgroup based on independent evidence.
Q: Is midpoint rooting appropriate for trees of bacteria? A: Only if the bacterial lineages have relatively constant substitution rates, which is rare. For bacteria, an outgroup from a sister phylum is usually needed.
Q: How do I assess the statistical support for a root? A: For outgroup rooting, the bootstrap support on the branch separating outgroup and ingroup indicates root confidence. For Bayesian methods, the posterior probability of the root partition can be examined. Do not rely solely on the tree drawing.
References and Further Reading
- NCBI Bookshelf , Free textbooks on molecular evolution and phylogenetics.
- EMBL‑EBI Training , Online courses on sequence analysis and tree building.
- Galaxy Training Network , Practical workflows for phylogenomics and rooting.
- Bioconductor , R packages for phylogenetic tree manipulation and root testing.
- Generalist core orchid mycorrhizal fungi structure symbiotic associations (PubMed) , Example of root placement in fungal symbiosis studies.
- Effects of sequencing platforms on root mycorrhizal communities (J Microbiol, PubMed) , Relevance of sequencing technology to phylogenetic inference.
- A Dissimilarity Measure for Semidirected Networks (IEEE Trans Comput Biol Bioinform, PubMed) , Advanced rooting in networks with reticulation.
- Bayesian tests of temporal signal (PLoS Comput Biol, PubMed) , Model sensitivity and root dating.
- New fungal primers for Mucoromycotinian AMF (Environ Microbiome, PubMed) , Primer design for targeted root studies.