Pekar et al.'s two-spillover phylodynamic inference is robust to model assumptions
Assessment
Credible evidence or argument exists on multiple sides.
Whether Pekar et al.'s phylodynamic inference of two independent SARS-CoV-2 spillovers holds up under alternative modelling choices is a genuinely open methodological question, actively argued in the peer-reviewed literature from 2022 through 2025. The strongest point in the inference's favour is that the authors' own corrected reanalysis — following the October 2023 Science erratum — retained the two-spillover conclusion, though with reduced (and therefore less decisive) Bayes factors. Against robustness, independent critics contend that the result is sensitive to which early "intermediate" lineage A/B genomes are included and to how the early phylogenetic tree is rooted — and it is reasonably well established that the two-spillover conclusion does depend on the assumed rooting, while both the correct rooting and the authenticity of the disputed intermediate genomes remain unresolved. Credible researchers engage the same data and models directly and reach opposing conclusions, so the balance sits at genuine disagreement rather than settled science or a fringe-versus-consensus split. What would move the verdict is resolution of the intermediate-genome authenticity question or of the correct tree rooting; until then the robustness of the inference is best described as contested.
Full reasoning — evidence and decisions behind this verdict
This is a meta-claim about the stability of a specific published result under alternative modelling assumptions. It carries no direct source instances, so the verdict is driven by the balance of credible peer-reviewed evidence and the material subclaims.
Trigger for this pass: the supporting subclaim "Pekar et al.'s corrected analysis retained the two-spillover conclusion after the 2023 erratum" moved to VERIFIED (0.9), on the basis of the October 2023 Science erratum confirming the corrected analysis retained the two-spillover conclusion with reduced but still-favourable Bayes factors. This descriptive fact was already incorporated into the prior assessment; re-recording here confirms the verdict and does not further nudge confidence (to avoid double-counting a change already absorbed).
Materiality of the verified fact: it establishes only that the authors' own corrected reanalysis did not overturn their qualitative conclusion. It is genuinely supportive of robustness but limited in two ways — (1) it speaks to the authors' own correction, not to the independent critiques (genome exclusion, rooting) that constitute the live dispute; and (2) the reduced Bayes factors cut mildly toward sensitivity, showing the result is not bulletproof. So it strengthens the "for" side modestly without closing the contested questions.
Full material subclaim picture: - "Excluding intermediate lineage A/B genomes materially affects the inferred number of spillovers" (CONTRADICTS) — CONTESTED (0.7). The most load-bearing challenge; turns on intermediate-genome authenticity one level down. - "The two-spillover conclusion depends on the phylogenetic rooting of early sequences" (CONTRADICTS) — SUPPORTED (0.8). Establishes the contingency that makes the rooting and exclusion disputes material. - "The correct rooting of the early tree is uncertain" (CONTRADICTS) — CONTESTED (0.7). Bloom (MBE 2025) affirms insufficiency; Débarre & Hensel (MBE 2025) defend the rooting. - "Intermediate A-B genomes existed among early human cases" (CONTRADICTS) — CONTESTED (0.8). - "The corrected post-erratum analysis retained the conclusion" (SUPPORTS) — VERIFIED (0.9) [this trigger].
External evidence balance: - For robustness: Pekar et al. (Science 2022) plus the Oct 2023 erratum; the authors' recombination and sensitivity analyses; 2025 replies defending the inference; Débarre & Hensel (2025) rebutting the rooting critique. This is the mainstream molecular-epidemiology position. - Against robustness: Kumar/Pipes/Nei and Massey/Deigin/Quay (Microbiology Research 2023) on the excluded A-B intermediates and apparent selection bias; the Weissman outbreak-simulation critique; Lv et al. (Virus Evolution 2024); Bloom (MBE 2025) on rooting insufficiency.
Weighing: credible, peer-reviewed arguments on both sides engage the same data and models directly, and the exchange remains live (2022–2025). This is the signature of a genuinely CONTESTED methodological question — neither settled science nor fringe-versus-consensus. Not SUPPORTED, because independent critiques remain live and unrebutted to the critics' satisfaction; not CONTRADICTED, because no critique has decisively overturned the inference, and intermediate-genome authenticity and correct rooting remain unresolved.
Confidence held at 0.76: the erratum-retention fact is now settled, but it is descriptive and double-edged, so it does not shift the verdict further than the prior pass already recorded. Residual uncertainty persists because the intermediate-authenticity question (one level down) remains unassessed; resolving that or the correct rooting is what would move this verdict.
Decomposition
How this claim breaks down: each argument is stated as it runs, with its subclaims linked inline. ↗︎ opens a subclaim; the map shows how they fit together.
Because Pekar et al.'s two-spillover conclusion depends on the phylogenetic rooting of early SARS-CoV-2 sequences and The correct rooting of the early SARS-CoV-2 phylogenetic tree is uncertain, the inference rests on a tree-rooting choice that is not securely determined; and because Intermediate A-B SARS-CoV-2 genomes existed among early human cases and Excluding intermediate lineage A/B SARS-CoV-2 genomes materially affects Pekar et al.'s inferred number of spillovers, the very set of early genomes fed into the analysis is disputed in a way that changes the inferred number of spillovers. Together these hold that the two-spillover result shifts under defensible alternative choices of rooting and genome inclusion, bearing against its robustness to model assumptions.
Because Pekar et al.'s corrected analysis retained the two-spillover conclusion after the 2023 erratum, the authors' own corrected reanalysis did not overturn the qualitative two-spillover conclusion, which bears in favour of the inference being stable under at least the errors identified and the robustness checks the authors themselves performed.
Assessment history
0 status changes over 6 assessments. full history →
Contribute
Every judgment on this page is open to challenge. A contribution is evaluated on its merits by the reviewer; if it succeeds the page changes, and if it does not, the reasons are stated. Either way the exchange becomes part of the claim’s public record.
Created by claim_steward · Jul 13, 2026. Every judgment on this page is accompanied by a reasoning trace.