Ancestral state reconstruction (ASR) is a phylogenetic toolkit for inferring the evolutionary history of discrete or continuous traits across a clade. A central practical challenge is model choice: whether to allow trait transitions in one direction only (irreversible model) or in both directions (reversible model), and how that choice shapes inferred origin counts and the apparent lability of a character.
A vivid illustration comes from the evolution of haplodiploidy in Acari (mites and ticks). Under a one-rate irreversible model — permitting only transitions from diplodiploidy to haplodiploidy — ASR recovers a mean of 12.9 independent origins of haplodiploidy. Relaxing the constraint to a two-rate model that also allows reversions yields a lower mean of 7.9 origins, because some internal nodes are reinterpreted as secondary losses rather than independent gains. Despite the apparent numerical support for reversibility, the evidence for actual transitions back to diplodiploidy is limited: the signal appears to be driven largely by a single deeply-nested haplodiploid taxon (Parasitidae sp.), and removing it shifts model preference back toward irreversibility (Blackmon et al. 2015, Finding 3).
This example highlights two recurring lessons in ASR practice: (1) origin counts are sensitive to which transitions the model permits, and (2) a handful of phylogenetically anomalous tips can disproportionately influence inferences about reversibility.
Ancestral State Reconstruction
Current understanding
Ancestral state reconstruction (ASR) is a phylogenetic toolkit for inferring the evolutionary history of discrete or continuous traits across a clade. A central practical challenge is model choice: whether to allow trait transitions in one direction only (irreversible model) or in both directions (reversible model), and how that choice shapes inferred origin counts and the apparent lability of a character.
A vivid illustration comes from the evolution of haplodiploidy in Acari (mites and ticks). Under a one-rate irreversible model — permitting only transitions from diplodiploidy to haplodiploidy — ASR recovers a mean of 12.9 independent origins of haplodiploidy. Relaxing the constraint to a two-rate model that also allows reversions yields a lower mean of 7.9 origins, because some internal nodes are reinterpreted as secondary losses rather than independent gains. Despite the apparent numerical support for reversibility, the evidence for actual transitions back to diplodiploidy is limited: the signal appears to be driven largely by a single deeply-nested haplodiploid taxon (Parasitidae sp.), and removing it shifts model preference back toward irreversibility (Blackmon et al. 2015, Finding 3).
This example highlights two recurring lessons in ASR practice: (1) origin counts are sensitive to which transitions the model permits, and (2) a handful of phylogenetically anomalous tips can disproportionately influence inferences about reversibility.
Supporting evidence
- Model dependence of origin counts in Acari. The estimated number of independent origins of haplodiploidy ranges from 7.9 (two-rate model) to 12.9 (one-rate model), with only limited support for the hypothesis that reversions from haplodiploidy back to diplodiploidy are possible. (Blackmon et al. 2015, Finding 3)
Contradictions / open disagreements
The two models applied to Acari haplodiploidy yield divergent origin counts (7.9 vs. 12.9), reflecting a genuine tension between parsimony-style irreversible models and more parameter-rich reversible ones. The apparent support for reversibility collapses when a single influential tip is removed, raising the question of whether the two-rate model is capturing biology or overfitting sparse data. Broader taxon sampling and independent data on ploidy in contentious families would help resolve this ambiguity.
Tealc’s citation-neighborhood suggestions
- Pagel (1994) on discrete character evolution methods (Discrete/BayesTraits) — directly relevant to rate-model choice in ASR.
- Maddison & Maddison (Mesquite) parsimony-based ASR approaches — useful contrast to likelihood methods used in the Acari study.
- Beaulieu & O’Meara on hidden-rates models, which can address apparent irreversibility driven by rate heterogeneity rather than true directionality.