Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 101 of the 101 references it cites were found in a registry or an index; 67 of the 86 sources its claims cite were checked for support.

Evidence for sponges as sister to all other animals from partitioned phylogenomics with mixture models and recoding

Open access
Redmond, Anthony K., McLysaght, Aoife · 2021 · Nature Communications

Abstract

Resolving the relationships between the major lineages in the animal tree of life is necessary to understand the origin and evolution of key animal traits. Sponges, characterized by their simple body plan, were traditionally considered the sister group of all other animal lineages, implying a gradual increase in animal complexity from unicellularity to complex multicellularity. However, the availability of genomic data has sparked tremendous controversy as some phylogenomic studies support comb jellies taking this position, requiring secondary loss or independent origins of complex traits. Here we show that incorporating site-heterogeneous mixture models and recoding into partitioned phylogenomics alleviates systematic errors that hamper commonly-applied phylogenetic models. Testing on real datasets, we show a great improvement in model-fit that attenuates branching artefacts induced by systematic error. We reanalyse key datasets and show that partitioned phylogenomics does not support comb jellies as sister to other animals at either the supermatrix or partition-specific level.

Overview

CoChat Summary

## TL;DR

This paper argues that once partitioned phylogenomics is analyzed with better-fitting site-heterogeneous mixture models and amino-acid recoding, the apparent support for comb jellies (Ctenophora) as sister to all other animals disappears. Across reanalysed datasets, the results instead support sponges (Porifera) as sister to all other animals, with the authors concluding that prior Ctenophora-sister support was driven by long-branch attraction and model misspecification.

## Key findings

  • In the three test datasets, site-heterogeneous models fit the data much better than standard site-homogeneous models: by L2, 131/133 BEA gene partitions and all 146 gene partitions in both LEAN and LEAP were better fit by site-heterogeneous models; by L4, complex T4 models were best fitting for 83% (BEA), 88% (LEAN), and 90% (LEAP) of partitions.
  • For the BEA dataset, support for the long-branch-attraction topology declined as models improved (UFBOOT: 99% → 96% → 81% from L1 to L3), and at L4 the LBA topology was not recovered; instead Microsporidia were placed within eukaryotes as sister to fungi, with UFBOOT = 94%.
  • In the bilaterian test datasets, better-fitting models reduced LBA: for LEAN, the LBA topology was recovered only at L1 (79%) and the accepted topology rose to 94% at L4; for LEAP, LBA support fell from 100% at L1 to 53% at L4.
  • Reanalysis of the animal datasets showed that the authors’ best-fitting analyses no longer support Ctenophora as sister to all other animals: under recoding plus site-heterogeneous models, REA recovered Placozoa sister weakly (UFBOOT = 52%), while WEA15 and WEA17 recovered Porifera sister with strong support (95% and 99%, respectively).

## Methods

  • The study reanalysed three test datasets designed to expose long-branch attraction (BEA, LEAN, LEAP) and three animal phylogeny datasets (REA, WEA15, WEA17), using partitioned phylogenomics and amino-acid recoding.
  • Datasets were partitioned by gene; for LEAP and LEAN, gene boundaries were manually inferred and verified with BLAST, while BEA already had gene partition information. Analyses were run in IQ-TREE (v1.5.4-omp and v1.6.12).
  • Model comparison used BIC across four analysis levels (L1–L4) and recoded levels (RL1–RL2), incorporating standard site-homogeneous models, multi-matrix models, CAT-like site-heterogeneous mixture models, and SR4 recoding; support was assessed with 1000 ultrafast bootstrap replicates.

## Limitations

  • The paper states that WEA17 appears to have poor ctenophore occupancy and a high level of orthology errors, and notes that these issues may contribute to its stronger signal for Ctenophora sister; it also mentions orthology errors reported previously for REA and WEA15.
  • The authors note that recoding can reduce information and may have detrimental effects on shallower nodes, though they argue this is acceptable when supported by complementary analyses. This is stated in the paper.
  • The study did not analyse the more gene-rich dataset of Simion et al. (2017) because higher analysis levels were estimated to require >1000 GB of RAM; this is a computational limitation stated in the paper.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/s41467-021-22074-7

Cites

101 of 101 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Fig. 1 Overview of context and methodology of this study. a Summary of typical phylogenomic approaches: phylogenomics conventionally relies on a matrix of concatenated ortholog alignments (top) and thus includes different evolutionary patterns that exist both within and between genes (different colours indicate evolutionary heterogeneity). Despite this, basic phylogenomic analyses assume a uniform evolutionary process across the supermatrix (one site-homogeneous model). More sophisticated approaches may allow the evolutionary model to differ between genes (partitioned analysis, left: each gene is uniform, but inter-gene heterogeneity is accommodated – i.e. many site-homogeneous models), or use a single model where sites can vary infinitely and without respect to gene structure (right: infinite site-heterogeneous model). Relative model fit cannot easily be compared between these approaches, and, as illustrated for the animal phylogeny (bottom), they can produce conflicting results. To help resolve which is the correct animal phylogeny, we introduce site-heterogeneous models (allowing intra-gene variation) (b), and amino acid recoding (reducing data heterogeneity) (c), into partitioned phylogenomics, and lay out a strategy of comparing increasingly complex analysis levels with which to test whether this improves model fit and LBA resilience. We use three well characterized datasets with known susceptibility to LBA on which to test these approaches (d), before applying them to three key animal phylogeny datasets that have previously produced conflicting topologies under different modelling strategies (e).
Fig. 1 Overview of context and methodology of this study. a Summary of typical phylogenomic approaches: phylogenomics conventionally relies on a matrix of concatenated ortholog alignments (top) and thus includes different evolutionary patterns that exist both within and between genes (different colours indicate evolutionary heterogeneity). Despite this, basic phylogenomic analyses assume a uniform evolutionary process across the supermatrix (one site-homogeneous model). More sophisticated approaches may allow the evolutionary model to differ between genes (partitioned analysis, left: each gene is uniform, but inter-gene heterogeneity is accommodated – i.e. many site-homogeneous models), or use a single model where sites can vary infinitely and without respect to gene structure (right: infinite site-heterogeneous model). Relative model fit cannot easily be compared between these approaches, and, as illustrated for the animal phylogeny (bottom), they can produce conflicting results. To help resolve which is the correct animal phylogeny, we introduce site-heterogeneous models (allowing intra-gene variation) (b), and amino acid recoding (reducing data heterogeneity) (c), into partitioned phylogenomics, and lay out a strategy of comparing increasingly complex analysis levels with which to test whether this improves model fit and LBA resilience. We use three well characterized datasets with known susceptibility to LBA on which to test these approaches (d), before applying them to three key animal phylogeny datasets that have previously produced conflicting topologies under different modelling strategies (e).CC BY · PubMed Central
Fig. 2 Analyses of test datasets incorporating site-heterogeneous models and SR4 recoding into partitioned phylogenomics. a Percentage of gene partitions best fit by each model tier (indicated by shading) at each analysis level for amino acid (left) and recoded (right) data. b Topologies recovered at different analysis levels for both amino acid (L1-4) and recoded (RL1-2) data in long-branch attraction tests. Ultrafast bootstrap (UFBOOT) percentages for the topology-defining node (indicated with a circle) are show for each analysis level. Scale bars are shown for each tree. For full tree topologies and UFBOOT support values see Supplementary Figs. 2-7.
Fig. 2 Analyses of test datasets incorporating site-heterogeneous models and SR4 recoding into partitioned phylogenomics. a Percentage of gene partitions best fit by each model tier (indicated by shading) at each analysis level for amino acid (left) and recoded (right) data. b Topologies recovered at different analysis levels for both amino acid (L1-4) and recoded (RL1-2) data in long-branch attraction tests. Ultrafast bootstrap (UFBOOT) percentages for the topology-defining node (indicated with a circle) are show for each analysis level. Scale bars are shown for each tree. For full tree topologies and UFBOOT support values see Supplementary Figs. 2-7.CC BY · PubMed Central
Fig. 3 Analyses of test animal phylogeny datasets incorporating site-heterogeneous models and SR4 recoding into partitioned phylogenomics. a Percentage of gene partitions best fit by each model tier at each analysis level for amino acid and recoded data. Results are shown for each of the three datasets: REA, WEA15 and WEA17. b L4 topology recovered for each dataset, with UFBOOT support for Ctenophora sister (defining node indicated with a circle) under L1–L4 and RL1 shown. c Topologies recovered under RL2 for each dataset, with UFBOOT support for Placozoa sister shown for the REA dataset, and for Porifera sister for the WEA15 and WEA17 dataset. Note that for the REA and WEA17 datasets UFBOOT support for some backbone relationships are low under RL2 as discussed in the ‘Partitioned phylogenomics does not support Ctenophora-sister’ results subsection and Supplementary Note 1. For full tree topologies and UFBOOT support values see Supplementary Figs. 17-19 and 21-23.
Fig. 3 Analyses of test animal phylogeny datasets incorporating site-heterogeneous models and SR4 recoding into partitioned phylogenomics. a Percentage of gene partitions best fit by each model tier at each analysis level for amino acid and recoded data. Results are shown for each of the three datasets: REA, WEA15 and WEA17. b L4 topology recovered for each dataset, with UFBOOT support for Ctenophora sister (defining node indicated with a circle) under L1–L4 and RL1 shown. c Topologies recovered under RL2 for each dataset, with UFBOOT support for Placozoa sister shown for the REA dataset, and for Porifera sister for the WEA15 and WEA17 dataset. Note that for the REA and WEA17 datasets UFBOOT support for some backbone relationships are low under RL2 as discussed in the ‘Partitioned phylogenomics does not support Ctenophora-sister’ results subsection and Supplementary Note 1. For full tree topologies and UFBOOT support values see Supplementary Figs. 17-19 and 21-23.CC BY · PubMed Central
Fig. 4 Partition-specific support for Ctenophora or Porifera sister under different analysis levels. a Number of partitions supporting either Ctenophora or Porifera sister from analysis levels L1–L4 at ΔPSlnl (change in partition-specific log-likelihood between trees/hypotheses) thresholds of >0, ≥0.5 and ≥1. b Relative proportion of partitions that support either Ctenophora or Porifera sister from L1–L4 (i.e. partitions that do not support either topology at that threshold level are not included).
Fig. 4 Partition-specific support for Ctenophora or Porifera sister under different analysis levels. a Number of partitions supporting either Ctenophora or Porifera sister from analysis levels L1–L4 at ΔPSlnl (change in partition-specific log-likelihood between trees/hypotheses) thresholds of >0, ≥0.5 and ≥1. b Relative proportion of partitions that support either Ctenophora or Porifera sister from L1–L4 (i.e. partitions that do not support either topology at that threshold level are not included).CC BY · PubMed Central
Fig. 5 Remaining plausible hypotheses of the backbone animal phylogeny following rejection of Ctenophora sister in favour of Porifera sister. Ctenophora is now most likely placed either as sister to Cnidaria (Coelenterata) within the Eumetazoa, as is traditional (a), or as the sister group of all non-sponge animals (Ctenophora second). In this latter scenario it is not yet clear whether Bilateria (b) or Placozoa (c) would fall sister to Cnidaria.
Fig. 5 Remaining plausible hypotheses of the backbone animal phylogeny following rejection of Ctenophora sister in favour of Porifera sister. Ctenophora is now most likely placed either as sister to Cnidaria (Coelenterata) within the Eumetazoa, as is traditional (a), or as the sister group of all non-sponge animals (Ctenophora second). In this latter scenario it is not yet clear whether Bilateria (b) or Placozoa (c) would fall sister to Cnidaria.CC BY · PubMed Central

Verification

Claims the paper makes

29 Valid · 1 Not valid · 6 Not checked · 75 Own results and interpretations

58 more claims · 5 with their cited sources checked

Create a free account to see all 111 claims and to verify the 6 references still open.

See all claims — free account
  • It may state two separate claims as one. 28
  • Probably not a substantial claim. 23

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Citations

Cites

Show all 100 references

Cited by

See all 146

Metadata

Authors
Redmond, A. K., & McLysaght, A.
Affiliations
Trinity College Dublin (IE)
Corresponding author
McLysaght, Aoife
Venue
Nature Communications
Published
2021-03-19
Volume
12
Issue
1
Pages
1783-1783
ISSN
2041-1723 (ISSN-L)
OpenAlex
W3136068442
Cited by
139
References
100
Citations by year
2026: 19 · 2025: 30 · 2024: 28 · 2023: 29 · 2022: 25 · 2021: 7
FWCI
17.42
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • Trinity's Access to Research Output (TARA) (Trinity College Dublin) (submittedVersion)
Funders
H2020 European Research Council (771419)
Topics
  • Life Sciences › Agricultural and Biological Sciences › Animal Science and Zoology › Marine Invertebrate Physiology and Ecology
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Genomics and Phylogenetic Studies
  • Life Sciences › Agricultural and Biological Sciences › Ecology, Evolution, Behavior and Systematics › Protist diversity and phylogeny
Keywords
sponge phylogeny, comb jellies, sister group relationships, recoding, systematic errors, model fit, evolution of multicellularity, animal origins
MeSH terms
Animals, Biological Evolution, Models, Genetic, Phylogeny, Porifera, Species Specificity, Genome, Genomics, Ctenophora

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Evidence for sponges as sister to all other animals from partitioned phylogenomics with mixture models and recoding. https://app.cochat.ai/papers/10.1038/s41467-021-22074-7
The paper
Redmond, A. K., & McLysaght, A. (2021). Evidence for sponges as sister to all other animals from partitioned phylogenomics with mixture models and recoding. Nature Communications. https://doi.org/10.1038/s41467-021-22074-7

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.