Adaptive molecular convergence is pervasive across deep time and largely decoupled from phenotypic convergence

We find adaptive molecular convergence to be pervasive across medusozoan evolution over deep time. However, this convergence is not predictably associated with phenotypic or phylogenetic similarity, and instead reflects broader, lineage-specific ecological pressures. Our results suggest that adaptive molecular convergence is widespread but usually difficult to relate to specific phenotypes or selective agents. Here, we discuss the implications of this pattern for three related questions: the drivers of molecular convergence over deep time, the interpretation of genomic signatures of adaptation, and the predictability of evolutionary outcomes. First, we argue that molecular convergence over deep time is not primarily structured by shared genomic constraints, but instead reflects selection across complex ecological dimensions. The identities and functions of convergent genes are not associated with phylogenetic distance in Medusozoa, indicating that constraints, which should decay with genetic divergence (22–25), do not determine which genes repeatedly experience convergent selection. We do not exclude that constraints influence molecular convergence (1,29), but our data suggest that genome-wide patterns of adaptive convergence are mainly driven by selection on diverse, often unobserved, ecological factors. As a result, genes that interface with the external environment show elevated rates of convergence overall, and individual events of molecular convergence occur idiosyncratically across lineages. This pattern is consistent with highly multidimensional selective regimes (6, 30). Species 8 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint therefore experience many uncorrelated and lineage-specific selective pressures that may overlap among taxa, such that the underlying selective agents are often unknown. Second, this perspective reshapes interpretations of genome-wide molecular convergence and helps reconcile seemingly incongruent patterns in the literature. On the one hand, many studies have identified repeated use of specific genes or pathways in association with convergent phenotypes or ecological transitions (2, 11). On the other hand, genome-wide analyses have often failed to detect elevated molecular convergence in lineages with repeatedly evolved traits (31–36), a pattern frequently attributed to a pervasive background of neutral convergence (4, 37); but see (33, 34). We show that, in addition to a neutral background (9), there is also a widespread background of adaptive convergence that is not restricted to specific traits. As a result, molecular convergence is not necessarily enriched in focal lineages, even when phenotypes have evolved repeatedly, because trait-specific signals are diluted within this adaptive background. This perspective also reframes patterns such as the reported decline in gene reuse over time (2). Although the decline is real, it may not reflect decreasing reuse of genes underlying specific adaptations. Our results suggest that a similar pattern arises even from comparisons among randomly chosen taxa. An important direction for future studies will be to explicitly quantify axes of ecological variation, which could account for some of the variables driving convergent selection (21, 30). We used amino acid substitutions as a signature of convergent selection because this approach is uniquely tractable across long timescales and large gene families. However, further work is needed to understand patterns of convergence across time at other genetic and molecular levels, which can be uncoupled from one another (38). Regulatory evolution, for instance, may be more important than protein sequence changes for many convergent phenotypes, particularly morphological traits (39, 40). Therefore, integrating multiple signatures of convergence across coding sequences, regulatory regions, and gene expression may provide a more complete view of the genomic basis of repeated adaptation (41). However, such analyses remain limited in taxonomic and temporal scope by the availability of high-quality genomes, and further theoretical advances are also needed to better associate regulatory and expression changes with positive selection [e.g., (42)]. Finally, our results reveal a tension between two perspectives on evolutionary repeatability. The pervasive nature of adaptive molecular convergence indicates that protein evolution is unexpectedly repeatable even among highly divergent lineages. At the same time, the lack of consistent map- 9 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint ping between molecular convergence and phenotype suggests that this repeatability is not easily interpretable or predictive. This is highlighted by the observation that homologs of known eye- related genes consistently experienced convergent selection in species without eyes. More generally, genomic signatures of selection do not directly identify the ecological causes of selection, and pro- cesses such as pleiotropy, epistasis, and functional divergence can produce convergent substitutions with distinct causes and consequences. Similar principles apply at other levels of organization, including morphology: similar phenotypes can arise from different selective pressures and similar selective pressures can produce different phenotypes (38). Overall, ecology appears to broadly shape patterns of convergent evolution across deep time, but linking genetic changes to specific traits or selective regimes will require a deeper understanding of genotype–phenotype relationships. 10 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint F i l i f e r a III IV F i l i f e r a III IV Siphonophorae Siphonophorae Capitata Capitata L e p t o t h e c a t a L e p t o t h e c a t a AplanulataAplanulata Trachylina Trachylina SCYPHOZOA SCYPHOZOA STAURO . STAURO . CUBO. CUBO. H Y D R O Z O A H Y D R O Z O A F i l i f e r a III IV F i l i f e r a III IV Siphonophorae Siphonophorae Capitata Capitata L e p t o t h e c a t a L e p t o t h e c a t a AplanulataAplanulata Trachylina Trachylina SCYPHOZOA SCYPHOZOA STAURO . STAURO . CUBO. CUBO. H Y D R O Z O A H Y D R O Z O A Medusa Losses Upright Colony Origins Posterior Prior Clade Ages MyA C Eye Origins 100300500700 A B D Staurozoa Aplanulata Filifera III IV Trachylina Leptothecata Capitata Cubozo a Scyphozoa Anthozoa Hydrozoa Medusozoa Cnidari a Siphono. No Medusa Medusa Wihtout Eye Medusa With Eye Solitary Encrusting Upright Pelagic No Polyp 100 300 500 684 Ma 100 300 500 684 Ma Figure 1: Convergent character transitions have occurred dozens of times across 680 My of medusozoan evolution. (A) Joint ancestral state reconstructions (ASR) of medusa stage and eye presence. ASRs were inferred using corHMM under the best-fit model by AIC (data S4). Time- calibrated phylogeny was inferred from an MCMCtree analysis of 68 taxa chosen to evenly sample major clades; these times were transferred to a 561-taxon phylogeny for ASR. Filifera I-II are not highlighted here for space but can be found in the full trees on Dryad. “No medusa” includes medusoids, sporosacs, and absent gonophores. Medusa and eye presence were plotted together because these characters are correlated (eyes only occur on medusae; see Methods). Full species trees with taxon labels and branch support values can be found on the Dryad and Github; see fig. S16 for ASR of eyes only and fig. S17 for ASR of medusa stage with detailed character categories. (B) ASR of colony architecture. (C) Ages of key character transitions. Points and error bars show the mean and 95% CI of node ages calculated from 500 chronograms sampled from the MCMCtree posterior distribution. (D) Estimated ages of key clades. Points and error bars show the mean and 95% highest-posterior density intervals for the prior (fossil calibrations without sequence data) and posterior. 11 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint 0.1 0.2 0.0 0.1 0.2 0200400600 Divergence time (Myr) ZFEL -1 0 1 Standardized effect size Eye Div. time Medusa Colony Div. time x col. Div. time Protein dist. Div time x tx Dist. x tx C Genomic predictors Transcriptome size Phenotypic predictors 3-way interaction Div. time x med. Adaptive Molecular Convergence A Div time x dist. Adaptive Molecular Convergence B Div. time x eye 6 4 2 0 Observed - Null Figure 2: Adaptive molecular convergence declines with divergence time and exceeds null expectations. (A) Adaptive molecular convergence (AMC) among medusozoan lineages as a function of divergence time. AMC was quantified as the proportion of gene families exhibiting convergent branch pairs. AMC values were transformed with a Yeo-Johnson transformation ( 𝜆 = −0.565); the regression is plotted here back-transformed onto the original scale. Points representing pairwise species comparisons ( 𝑛 = 3570) were binned for visualization purposes, with darker shading indicating higher observation density. The solid line shows the phylogenetic regression between AMC and divergence time, with the shaded ribbon indicating the model confidence interval. (B) Observed AMC compared with expectations under Zero-Force Evolutionary Law (ZFEL) simulations. Boxplots show the distribution of observed convergence (data in Panel A) binned into 50 My intervals. Individual points were plotted for intervals with < 10 data points. Dashed lines show median convergence expected from ZFEL simulations, which model the decline in similarity among independently evolving systems without correlated selection. The multiple lines show different realizations of the null with varying rates of loss. All scenarios are set to converge on the same steady-state value, which equals the convergence in the final time point of observed data (∼ 4%; see Methods). Below, bars show mean differences between observed and null values within each bin. Asterisks indicate 𝑝 < 0.05 (Bonferroni-corrected within each null model). The most conservative null distribution (dark brown) was constructed such that its lower 0.025 quantile would include the observed value at our oldest time point; this explains why null values slightly exceed observed values in the oldest bin (see fig. S14 and Supplementary Text). (C) Predictors of adaptive molecular convergence. Top, standardized regression coefficients from a model evaluating the effects of divergence time, protein sequence distance, transcriptome size, and their interactions on AMC. Confidence intervals are smaller than points. Bottom, standardized regression coefficients for a model testing whether eyes, medusa stage, and colony form predict AMC. Horizontal bars indicate confidence intervals and shaded circles indicate significance (𝑝 < 0.05). 12 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint B C D A cellular processes nucleotide metabolism genetic information processing immune system Enrichment significance (-log10 P-value) All housekeeping Count of GO terms Phylogenetic covariance 0 100 200 300 Phylogenetic covariance 0 1 2 3 4 Number of species pairs sharing GO term 0 200 400 600 800 0 10 100 1000 C-score GO semantic similarity 0 5 10 15 20 25 bLANK 0 0.5 1.0 All environment-facing xenobiotics biosynth. secondary metabolites sensory system environmental info. processing P=0.05 Set size 100 500 1000 1500 2000 Most terms occur in ≤2 species pairs 0 100 200 300 r = 0.0011 2 r = 1 × 10 2 -5 Figure 3: Adaptive molecular convergence lacks phylogenetic structure but is associated with environment-facing genes. (A) Identities of convergent genes are not associated with phylogenetic distance. C-scores represent the overlap of gene families that independently experienced conver- gence in two pairs of species, with each unit increase representing one standard deviation above the expected overlap from a null hypergeometric distribution. Each point shows the C-score of a randomly-chosen species quartet (n=236,070) plotted against the phylogenetic covariance of that quartet, with darker shading indicating higher observation density. Higher covariance values indi- cate more shared phylogenetic history. Dashed line indicates average value and solid line indicates a linear regression. (B) GO annotations of convergent genes are not associated with phylogenetic distance. Each point represents the mean semantic similarity of the lists of GO terms enriched among convergent genes between two pairs of species (n=4,755,188 quartets) plotted against phy- logenetic covariance. Semantic similarities were calculated as the mean of the maximum similarity between each GO term and members of the other list. Dashed line indicates average value and solid line indicates a linear regression. (C) Number of species pairs with enrichment of each GO term (log scale on X axis). Most GO terms are only found in one or two species pairs. (D) Gene set enrichment analysis of “environment-facing” and “housekeeping” genes. Genes were categorized based on KEGG annotations and grouped into the sub-categories shown here and in table S4. Dashed line indicates raw p-value = 0.05, and point sizes indicate the number of annotated genes in that category. 13 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint

and Notes 1. B. A. Fraser, J. R. Whiting, What Can Be Learned by Scanning the Genome for Molec- ular Convergence in Wild Populations? 1476 (1), 23–42, doi:10.1111/nyas.14177, https: //nyaspubs.onlinelibrary.wiley.com/doi/full/10.1111/nyas.14177. 2. M. Bohut ´ınsk´a, C. L. Peichel, Divergence Time Shapes Gene Reuse during Repeated Adap- tation 39 (4), 396–407, doi:10.1016/j.tree.2023.11.007, https://www.cell.com/trends/ ecology-evolution/abstract/S0169-5347(23)00325-7. 3. A. R. Macdonald, M. E. James, J. D. Mitchell, B. R. Holland, From Trees to Traits: A Review of Advances in PhyloG2P Methods and Future Directions 17 (9), evaf150, doi:10.1093/gbe/ evaf150, https://doi.org/10.1093/gbe/evaf150. 4. J. B. Allard, S. Kumar, The Genetic Foundations of Convergent Traits pp. 1–16, doi:10.1038/ s41576-026-00933-7, https://www.nature.com/articles/s41576-026-00933-7. 5. S. D. Smith, M. W. Pennell, C. W. Dunn, S. V. Edwards, Phylogenetics Is the New Genetics (for Most of Biodiversity) 35 (5), 415–425, doi:10.1016/j.tree.2020.01.005, https://www. cell.com/trends/ecology-evolution/abstract/S0169-5347(20)30007-0. 6. A. D. C. MacColl, The Ecological Causes of Evolution 26 (10), 514–522, doi:10.1016/ j.tree.2011.06.009, https://www.cell.com/trends/ecology-evolution/abstract/ S0169-5347(11)00175-3. 7. G. L. Conte, M. E. Arnegard, C. L. Peichel, D. Schluter, The Probability of Genetic Parallelism and Convergence in Natural Populations279 (1749), 5039–5047, doi:10.1098/rspb.2012.2146, https://royalsocietypublishing.org/doi/full/10.1098/rspb.2012.2146. 8. R. A. Goldstein, S. T. Pollard, S. D. Shah, D. D. Pollock, Nonadaptive Amino Acid Convergence Rates Decrease over Time 32 (6), 1373–1381, doi:10.1093/molbev/msv041, https://doi. org/10.1093/molbev/msv041. 14 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint 9. Z. Zou, J. Zhang, Are Convergent and Parallel Amino Acid Substitutions in Protein Evolution More Prevalent Than Neutral Expectations? 32 (8), 2085–2096, doi:10.1093/molbev/msv091, https://doi.org/10.1093/molbev/msv091. 10. M. L. Smith, M. W. Hahn, New Approaches for Inferring Phylogenies in the Presence of Paralogs 37 (2), 174–187, doi:10.1016/j.tig.2020.08.012, https://www.sciencedirect. com/science/article/pii/S0168952520302122. 11. T. H. Oakley, Building, Maintaining, and (Re-)Deploying Genetic Toolkits during Convergent Evolution 64 (5), 1505–1512, doi:10.1093/icb/icae114, https://doi.org/10.1093/icb/ icae114. 12. K. Fukushima, D. D. Pollock, Detecting Macroevolutionary Genotype–Phenotype Associ- ations Using Error-Corrected Rates of Protein Convergence 7 (1), 155–170, doi:10.1038/ s41559-022-01932-7, https://www.nature.com/articles/s41559-022-01932-7. 13. S. A. S. Anderson, S. Kaushik, D. R. Matute, The Comparative Analysis of Lineage-Pair Traits p. syaf061, doi:10.1093/sysbio/syaf061, https://doi.org/10.1093/sysbio/syaf061. 14. N. Picciani, et al., Prolific Origination of Eyes in Cnidaria with Co-option of Non-visual Opsins 28 (15), 2413–2419.e4, doi:10.1016/j.cub.2018.05.055, https://www.cell.com/ current-biology/abstract/S0960-9822(18)30691-2. 15. M. P. Miglietta, C. W. Cunningham, EVOLUTION OF LIFE CYCLE, COLONY MOR- PHOLOGY, AND HOST SPECIFICITY IN THE FAMILY HYDRACTINIIDAE (HYDRO- ZOA, CNIDARIA) 66 (12), 3876–3901, doi:10.1111/j.1558-5646.2012.01717.x, https: //doi.org/10.1111/j.1558-5646.2012.01717.x. 16. P. Cartwright, A. M. Nawrocki, Character Evolution in Hydrozoa (Phylum Cnidaria) 50 (3), 456–472, doi:10.1093/icb/icq089, https://doi.org/10.1093/icb/icq089. 17. S. Birch, N. Picciani, T. Oakley, D. Plachetzki, Cnidarians: Diversity and Evolution of Cnidarian Visual Systems, in Distributed Vision: From Simple Sensors to Sophisticated Combination Eyes, E. Buschbeck, M. Bok, Eds. (Springer International Publishing), pp. 21–47, doi:10. 1007/978-3-031-23216-9 2, https://doi.org/10.1007/978-3-031-23216-9_2. 15 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint 18. P. Cartwright, M. K. Travert, S. M. Sanders, The Evolution and Development of Coloniality in Hydrozoans 336 (3), 293–299, doi:10.1002/jez.b.22996, https://onlinelibrary.wiley. com/doi/abs/10.1002/jez.b.22996. 19. D. W. McShea, R. N. Brandon,Biology’s First Law: The Tendency for Diversity and Complexity to Increase in Evolutionary Systems (University of Chicago Press). 20. H. A. Orr, The Genetic Theory of Adaptation: A Brief History 6 (2), 119–127, doi:10.1038/ nrg1523, https://www.nature.com/articles/nrg1523. 21. S. Chaturvedi, et al., Climatic Similarity and Genomic Background Shape the Extent of Parallel Adaptation in Timema Stick Insects 6 (12), 1952–1964, doi:10.1038/s41559-022-01909-6, https://www.nature.com/articles/s41559-022-01909-6. 22. J. Tischler, B. Lehner, A. G. Fraser, Evolutionary Plasticity of Genetic Interaction Networks 40 (4), 390–391, doi:10.1038/ng.114, https://www.nature.com/articles/ng.114. 23. E. V. Koonin, Evolution of Genome Architecture 41 (2), 298–306, doi:10. 1016/j.biocel.2008.09.015, https://www.sciencedirect.com/science/article/pii/ S1357272508003907. 24. T. Gabald ´on, E. V. Koonin, Functional and Evolutionary Implications of Gene Orthology 14 (5), 360–366, doi:10.1038/nrg3456. 25. M. E. Goldberg, K. Harris, Mutational Signatures of Replication Timing and Epigenetic Mod- ification Persist through the Global Divergence of Mutation Spectra across the Great Ape Phylogeny 14 (1), evab104, doi:10.1093/gbe/evab104, https://doi.org/10.1093/gbe/ evab104. 26. J. B. Losos, Phylogenetic Niche Conservatism, Phylogenetic Signal and the Relationship be- tween Phylogenetic Relatedness and Ecological Similarity among Species11 (10), 995–1003, doi:10.1111/j.1461-0248.2008.01229.x, https://onlinelibrary.wiley.com/doi/abs/ 10.1111/j.1461-0248.2008.01229.x. 16 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint 27. T. D. Price, et al., Determinants of Northerly Range Limits along the Himalayan Bird Di- versity Gradient doi:10.1086/661926, https://www.journals.uchicago.edu/doi/10. 1086/661926. 28. S. Yeaman, A. C. Gerstein, K. A. Hodgins, M. C. Whitlock, Quantifying How Con- straints Limit the Diversity of Viable Routes to Adaptation 14 (10), e1007717, doi: 10.1371/journal.pgen.1007717, https://journals.plos.org/plosgenetics/article? id=10.1371/journal.pgen.1007717. 29. E. V. Koonin, Y. I. Wolf, Constraints and Plasticity in Genome and Molecular-Phenome Evo- lution 11 (7), 487, doi:10.1038/nrg2810, https://pmc.ncbi.nlm.nih.gov/articles/ PMC3273317/. 30. Y. E. Stuart, et al., Contrasting Effects of Environment and Genetics Generate a Continuum of Parallel Evolution 1 (6), 0158, doi:10.1038/s41559-017-0158, https://www.nature.com/ articles/s41559-017-0158. 31. G. W. Thomas, M. W. Hahn, Determining the Null Model for Detecting Adaptive Convergence from Genomic Data: A Case Study Using Echolocating Mammals 32 (5), 1232–1236, doi: 10.1093/molbev/msv013, https://doi.org/10.1093/molbev/msv013. 32. A. D. Foote, et al., Convergent Evolution of the Genomes of Marine Mammals47 (3), 272–275, doi:10.1038/ng.3198, https://www.nature.com/articles/ng.3198. 33. O. Chabrol, M. Royer-Carenzi, P. Pontarotti, G. Didier, Detecting the Molecular Basis of Phenotypic Convergence 9 (11), 2170–2180, doi:10.1111/2041-210X.13071, https: //onlinelibrary.wiley.com/doi/abs/10.1111/2041-210X.13071. 34. Z. Mer ´enyi, et al., Unmatched Level of Molecular Convergence among Deeply Divergent Complex Multicellular Fungi 37 (8), 2228–2240, doi:10.1093/molbev/msaa077, https:// doi.org/10.1093/molbev/msaa077. 35. B. Lu, H. Jin, J. Fu, Molecular Convergent and Parallel Evolution among Four High-Elevation Anuran Species from the Tibetan Region 21 (1), 839, doi:10.1186/s12864-020-07269-4, https://doi.org/10.1186/s12864-020-07269-4. 17 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint 36. R. B. Corbett-Detig, S. L. Russell, R. Nielsen, J. Losos, Phenotypic Convergence Is Not Mirrored at the Protein Level in a Lizard Adaptive Radiation37 (6), 1604–1614, doi:10.1093/ molbev/msaa028, https://academic.oup.com/mbe/article/37/6/1604/5728642. 37. Z. He, et al., Convergent Adaptation of the Genomes of Woody Plants at the Land–Sea Interface 7 (6), 978–993, doi:10.1093/nsr/nwaa027, https://doi.org/10.1093/nsr/nwaa027. 38. J. B. Losos, CONVERGENCE, ADAPTATION, AND CONSTRAINT 65 (7), 1827–1840, doi:10.1111/j.1558-5646.2011.01289.x, https://doi.org/10.1111/j.1558-5646. 2011.01289.x. 39. J. G. Roscito, et al., Phenotype Loss Is Associated with Widespread Divergence of the Gene Regulatory Landscape in Evolution 9 (1), 4737, doi:10.1038/s41467-018-07122-z, https: //www.nature.com/articles/s41467-018-07122-z. 40. T. B. Sackton, et al., Convergent Regulatory Evolution and Loss of Flight in Paleognathous Birds 364 (6435), 74–78, doi:10.1126/science.aat7244, https://www.science.org/doi/ 10.1126/science.aat7244. 41. E. Osipova, et al., Convergent and Lineage-Specific Genomic Changes Shape Adaptations in Sugar-Consuming Birds 391 (6788), eadt1522, doi:10.1126/science.adt1522, https://www. science.org/doi/10.1126/science.adt1522. 42. H. Yan, et al., PhyloAcc-GT: A Bayesian Method for Inferring Patterns of Substitution Rate Shifts on Targeted Lineages Accounting for Gene Tree Discordance 40 (9), msad195, doi: 10.1093/molbev/msad195, https://doi.org/10.1093/molbev/msad195. 43. E. S. Chang, et al., Genomic Insights into the Evolutionary Origin of Myxozoa within Cnidaria 112 (48), 14912–14917, doi:10.1073/pnas.1511468112, https://www.pnas.org/ doi/abs/10.1073/pnas.1511468112. 44. N. Picciani, et al., Comparative Analysis of Convergent Jellyfish Eyes Reveals Extensive Differences in Expression of Vision-Related Genes 15 (7), e71784, doi:10.1002/ece3.71784, https://onlinelibrary.wiley.com/doi/abs/10.1002/ece3.71784. 18 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint 45. A. Macias-Mu ˜noz, R. Varney, E. Katcher, M. Everhart, T. H. Oakley, Genome Assembly of Bougainvillia Cf. Muscus (Cnidaria: Hydrozoa) 15 (7), jkaf110, doi:10.1093/g3journal/ jkaf110. 46. A. M. Bolger, M. Lohse, B. Usadel, Trimmomatic: A Flexible Trimmer for Illumina Se- quence Data 30 (15), 2114–2120, doi:10.1093/bioinformatics/btu170, https://academic. oup.com/bioinformatics/article/30/15/2114/2390096. 47. L. Song, L. Florea, Rcorrector: Efficient and Accurate Error Correction for Illumina RNA-Seq Reads 4 (1), doi:10.1186/s13742-015-0089-y. 48. E. Kopylova, L. No´e, H. Touzet, SortMeRNA: Fast and Accurate Filtering of Ribosomal RNAs in Metatranscriptomic Data 28 (24), 3211–3217, doi:10.1093/bioinformatics/bts611. 49. M. G. Grabherr, et al., Trinity: Reconstructing a Full-Length Transcriptome without a Genome from RNA-Seq Data 29 (7), 644–652, doi:10.1038/nbt.1883. 50. W. Li, A. Godzik, Cd-Hit: A Fast Program for Clustering and Comparing Large Sets of Protein or Nucleotide Sequences 22 (13), 1658–1659, doi:10.1093/bioinformatics/btl158. 51. B. Buchfink, C. Xie, D. H. Huson, Fast and Sensitive Protein Alignment Using DIA- MOND 12 (1), 59–60, doi:10.1038/nmeth.3176, https://www.nature.com/articles/ nmeth.3176. 52. F. A. Sim ˜ao, R. M. Waterhouse, P. Ioannidis, E. V. Kriventseva, E. M. Zdobnov, BUSCO: Assessing Genome Assembly and Annotation Completeness with Single-Copy Orthologs 31 (19), 3210–3212, doi:10.1093/bioinformatics/btv351, https://doi.org/10. 1093/bioinformatics/btv351. 53. C. P. Cantalapiedra, A. Hern ´andez-Plaza, I. Letunic, P. Bork, J. Huerta-Cepas, eggNOG- mapper v2: Functional Annotation, Orthology Assignments, and Domain Prediction at the Metagenomic Scale 38 (12), 5825–5829, doi:10.1093/molbev/msab293. 54. L. Gabriel, et al., BRAKER3: Fully Automated Genome Annotation Using RNA-seq and Protein Evidence with GeneMark-ETP, AUGUSTUS and TSEBRA p. 2023.06.10.544449, doi:10.1101/2023.06.10.544449. 19 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint 55. D. M. Emms, S. Kelly, OrthoFinder: Phylogenetic Orthology Inference for Comparative Genomics 20, 238, doi:10.1186/s13059-019-1832-y, https://pmc.ncbi.nlm.nih.gov/ articles/PMC6857279/. 56. S. Whelan, I. Irisarri, F. Burki, PREQUAL: Detecting Non-Homologous Characters in Sets of Unaligned Homologous Sequences 34 (22), 3929–3930, doi:10.1093/bioinformatics/bty448, https://doi.org/10.1093/bioinformatics/bty448. 57. S. Capella-Guti ´errez, J. M. Silla-Mart ´ınez, T. Gabald ´on, trimAl: A Tool for Automated Alignment Trimming in Large-Scale Phylogenetic Analyses 25 (15), 1972–1973, doi: 10.1093/bioinformatics/btp348, https://doi.org/10.1093/bioinformatics/btp348. 58. T. K. F. Wong, et al., IQ-TREE 3: Phylogenomic Inference Software Using Complex Evolu- tionary Models https://ecoevorxiv.org/repository/view/8916/. 59. U. Mai, S. Mirarab, TreeShrink: Fast and Accurate Detection of Outlier Long Branches in Collections of Phylogenetic Trees 19, 272, doi:10.1186/s12864-018-4620-2, https://www. ncbi.nlm.nih.gov/pmc/articles/PMC5998883/. 60. C. Zhang, R. Nielsen, S. Mirarab, ASTER: A Package for Large-Scale Phylogenomic Re- constructions 42 (8), msaf172, doi:10.1093/molbev/msaf172, https://doi.org/10.1093/ molbev/msaf172. 61. L. Lecl `ere, P. Schuchert, C. Cruaud, A. Couloux, M. Manuel, Molecular Phylogenetics of Thecata (Hydrozoa, Cnidaria) Reveals Long-Term Maintenance of Life History Traits despite High Frequency of Recent Character Changes 58 (5), 509–526, doi:10.1093/sysbio/syp044. 62. E. Kayal, et al., Phylogenetic Analysis of Higher-Level Relationships within Hydroidolina (Cnidaria: Hydrozoa) Using Mitochondrial Genome Data and Insight into Their Mitochondrial Transcription 3, e1403, doi:10.7717/peerj.1403, https://peerj.com/articles/1403. 63. M. L. Borowiec, et al., Evaluating UCE Data Adequacy and Integrating Uncertainty in a Comprehensive Phylogeny of Ants p. syaf001, doi:10.1093/sysbio/syaf001, https://doi. org/10.1093/sysbio/syaf001. 20 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint 64. Z. Yang, B. Rannala, Bayesian Estimation of Species Divergence Times Under a Molecular Clock Using Multiple Fossil Calibrations with Soft Bounds 23 (1), 212–226, doi:10.1093/ molbev/msj024, https://doi.org/10.1093/molbev/msj024. 65. M. Panchaksaram, L. Freitas, p. u. family=Reis, given=Mario, Bayesian Selection of Relaxed- Clock Models: Distinguishing between Independent and Autocorrelated Rates74 (2), 323–334, doi:10.1093/sysbio/syae066, https://doi.org/10.1093/sysbio/syae066. 66. J. M. Eastman, L. J. Harmon, D. C. Tank, Congruification: Support for Time Scaling Large Phy- logenetic Trees 4 (7), 688–691, doi:10.1111/2041-210X.12051, https://onlinelibrary. wiley.com/doi/abs/10.1111/2041-210X.12051. 67. S. A. Smith, B. C. O’Meara, treePL: Divergence Time Estimation Using Penalized Likelihood for Large Phylogenies28 (20), 2689–2690, doi:10.1093/bioinformatics/bts492,https://doi. org/10.1093/bioinformatics/bts492. 68. K. J. L. Maurin, An Empirical Guide for Producing a Dated Phylogeny with treePL in a Max- imum Likelihood Framework, doi:10.48550/arXiv.2008.07054, http://arxiv.org/abs/ 2008.07054. 69. L. S. Miranda, A. G. Collins, Eyes in Staurozoa (Cnidaria): A Review 7, e6693, doi:10.7717/ peerj.6693, https://pmc.ncbi.nlm.nih.gov/articles/PMC6448553/. 70. J. D. Boyko, J. M. Beaulieu, Generalized Hidden Markov Models for Phylogenetic Compar- ative Datasets 12 (3), 468–478, doi:10.1111/2041-210X.13534, https://onlinelibrary. wiley.com/doi/abs/10.1111/2041-210X.13534. 71. C. S. McFadden, et al., Phylogenomics, Origin, and Diversification of Anthozoans (Phy- lum Cnidaria) 70 (4), 635–647, doi:10.1093/sysbio/syaa103, https://doi.org/10.1093/ sysbio/syaa103. 72. J. D. Boyko, J. M. Beaulieu, Reducing the Biases in False Correlations Between Discrete Char- acters 72 (2), 476–488, doi:10.1093/sysbio/syac066, https://doi.org/10.1093/sysbio/ syac066. 21 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint 73. M. Suyama, D. Torrents, P. Bork, PAL2NAL: Robust Conversion of Protein Sequence Align- ments into the Corresponding Codon Alignments 34, W609–W612, doi:10.1093/nar/gkl315, https://doi.org/10.1093/nar/gkl315. 74. I.-K. Yeo, R. A. Johnson, A New Family of Power Transformations to Improve Normality or Symmetry 87 (4), 954–959, https://www.jstor.org/stable/2673623. 75. W. M. Fitch, An Improved Method of Testing for Evolutionary Homology 16 (1), 9–16, doi: 10.1016/s0022-2836(66)80258-9. 76. D. A. Levin, Y. Peres, Markov Chains and Mixing Times, Second Edition . 77. A. Alexa, J. Rahnenfuhrer, Gene Set Enrichment Analysis with topGO . 78. F. Supek, M. Bo ˇsnjak, N. ˇSkunca, T. ˇSmuc, REVIGO Summarizes and Visualizes Long Lists of Gene Ontology Terms 6 (7), e21800, doi:10.1371/journal.pone.0021800, https: //journals.plos.org/plosone/article?id=10.1371/journal.pone.0021800. 79. A. B. Kamran, H. Naveed, GOntoSim: A Semantic Similarity Measure Based on LCA and Com- mon Descendants 12 (1), 3818, doi:10.1038/s41598-022-07624-3, https://www.nature. com/articles/s41598-022-07624-3. 80. D. I. Speiser, et al., Using Phylogenetically-Informed Annotation (PIA) to Search for Light- Interacting Genes in Transcriptomes from Non-Model Organisms 15 (1), 350, doi:10.1186/ s12859-014-0350-x. 81. A. M. Tarrant, C. A. Berger, Cnidarian Circadian Clocks Model How Animals Find Predictabil- ity in a Complex World 65 (3), 688–700, doi:10.1093/icb/icaf038, https://doi.org/10. 1093/icb/icaf038. 82. S. Chen, Z. Zou, Detecting Convergence of Amino Acid Physicochemical Properties Underlying the Organismal Adaptive Convergent Evolution25 (8), e70052, doi:10.1111/1755-0998.70052. 83. C. Venditti, A. Meade, M. Pagel, Detecting the Node-Density Artifact in Phylogeny Recon- struction 55 (4), 637–643, doi:10.1080/10635150600865567,https://academic.oup.com/ sysbio/article/55/4/637/1679618. 22 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint 84. Y. I. Wolf, P. S. Novichkov, G. P. Karev, E. V. Koonin, D. J. Lipman, The Universal Distribution of Evolutionary Rates of Genes and Distinct Characteristics of Eukaryotic Genes of Different Apparent Ages 106 (18), 7273–7280, doi:10.1073/pnas.0901808106, https://www.pnas. org/doi/abs/10.1073/pnas.0901808106. Acknowledgments We would like to thank F. Diskin for helping with the medusa morphology literature search; J. Wolfe, P. Nosil, K. Peichel, J. Feder, Z. Gompert, members of the Oakley lab, and the EEMB Evolution

group for insightful remarks on the manuscript; A. Collins for providing Scolionema samples; and the Smithsonian Tropical Research Institute (STRI), Bocas del Toro Research Station (BRS), R. Collin, and members of the Benthic Cnidaria class at BRS for assistance with sample collection. Funding: This work was supported by the National Science Foundation (grants DEB-2153773 to T.H.O.; DEB-2153774 to P.C.; DEB-2153775 to M.M.; and OISE-1828949 to R. Collin, which sup- ported sample collection). The Center for Scientific Computing (CSC) is supported by the California NanoSystems Institute and the Materials Research Science and Engineering Center (supported by NSF grant DMR-1720256) at UC Santa Barbara. Use was made of computational facilities pur- chased with funds from the National Science Foundation (CNS-1725797) and administered by the CSC. Author contributions: Conceptualization: C.A.B., M.M., P.C., T.H.O; Resources: R.M.V., M.M., P.C.; Investigation: C.A.B., M.I.S., R.M.V., S.C.A; Formal analysis: C.A.B, S.C.A.; Visualization: C.A.B., T.H.O.; Funding acquisition: M.M., P.C., T.H.O.; Supervision: M.M., P.C., T.H.O.; Writ- ing—original draft: C.A.B., T.H.O.; Writing—review & editing: all authors. Competing interests: There are no competing interests to declare. Data and materials availability: Sequence data generated for this study are available at NCBI under Bioproject PRJNA1455705, and accessions of previously published data can be found 23 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint in Data S1. Transcriptome assemblies, alignments, gene and species trees, CSUBST output, and other large data files will be available on Dryad upon publication. Scripts and smaller data files are available at https://github.com/ucsb-oakley-lab/Medusozoa_project and will also be archived on Dryad. Supplementary materials

Supplementary Text Figs. S1 to S18 Tables S1 to S6

(43-84) Data S1 to S5 24 .CC-BY-NC 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted April 24, 2026. ; https://doi.org/10.64898/2026.04.23.718300doi: bioRxiv preprint