Philippa Kaur

Department of Anthropology, East Carolina University, Greenville, NC, USARyan SchachtDepartment of Environmental Science, Policy and Management and Museum of Vertebrate Zoology, University of California, Berkeley, CA, 94720, USASteven R. BeissingerDepartment of Ecology and Evolution, University of Lausanne, 1015, Lausanne, SwitzerlandClaus WedekindEcology & Evolution, Research School of Biology, The Australian National University, Acton, Canberra, 2601, AustraliaMichael D. JennionsMARBEC Univ Montpellier, CNRS, Ifremer, IRD, Montpellier, FranceBenjamin GeffroyELKH-PE Evolutionary Ecology Research Group, University of Pannonia, 8210, Veszprém, HungaryAndrás LikerBehavioural Ecology Research Group, Center for Natural Sciences, University of Pannonia, 8210, Veszprém, HungaryAndrás LikerBehavioral Ecology and Sociobiology Unit, German Primate Center, Leibniz Institute of Primate Biology, 37077, Göttingen, GermanyPeter M. KappelerDepartment of Sociobiology/Anthropology, University of Göttingen, 37077, Göttingen, GermanyPeter M. KappelerGroningen Institute for Evolutionary Life Sciences, University of Groningen, 9747 AG, Groningen, The NetherlandsFranz J. WeissingDepartment of Anthropology, University of Utah, Salt Lake City, UT, USAKaren L. KramerInstitute of Global Health, University College London, London, UKTherese HeskethCentre for Global Health, Zhejiang University School of Medicine, Hangzhou, P.R. ChinaTherese HeskethIHPE Univ Perpignan Via Domitia, CNRS, Ifremer, Univ Montpellier, Perpignan, FranceJérôme BoissierStockholm University Demography Unit, Sociology Department, Stockholm University, 106 91, Stockholm, SwedenCaroline UgglaKem C. Gardner Policy Institute, David Eccles School of Business, University of Utah, Salt Lake City, UT, USAMike HollingshausMilner Centre for Evolution, University of Bath, Bath, BA2 7AY, UKTamás SzékelyELKH-DE Reproductive Strategies Research Group, Department of Zoology and Human Biology, University of Debrecen, H-4032, Debrecen, HungaryTamás Székely More

Predictors of reservoir statusOur analyses include all known rodent reservoirs for zoonotic pathogens (282 species). These reservoirs harbour a total of 95 known zoonotic pathogens (34 viruses, 26 bacteria, 17 helminths, 12 protozoa and six fungi) employing all known modes of transmission (43 vector-borne, 32 close-contact, 28 non-close contact, and 13 using multiple transmission modes) (Supplementary Data 2). Compared to presumed non-reservoirs (species currently not known to harbour any zoonotic pathogens), we observed that reservoir rodents are strikingly synanthropic (Figs. 2, 3a, Table 1). Despite potential geographic biases, and the general possibility that synanthropic species are better studied compared to non-synanthropic species (see Sampling bias and Supplementary Figs. 1, 2), synanthropy emerged as a defining characteristic of nearly all (95%) currently known rodent reservoirs. Of the 155 synanthropic species, only six are considered as truly synanthropic, i.e., predominately, if not exclusively, occurring in or near human dwellings, while the remaining species only occasionally show synanthropic behaviour (Supplementary Data 1).Fig. 2: Predictors of reservoir status.Final structural equation model linking reservoir status of rodent species (n = 269) with their synanthropy and hunting status, population fluctuations (s-index, log-transformed), and adult body mass, controlling for their occurrence in a range of habitats and the number of studies available per species. One-sided (directional) arrows represent a causal influence originating from the variable at the base of the arrow, with the width of the arrow and associated value representing the standardised strength of the relationship. The small double-sided arrows and numbers next to each response (endogenous) variable represent the error variance.Full size imageFig. 3: Characteristics of reservoir and synanthropic rodents.a Reservoir rodents are predominately synanthropic (n = 436 with n (non-reservoir) = 154, n (reservoir) = 282). b Synanthropic rodents display high population fluctuations (high s-index) (n = 269) and c, occur in multiple artificial habitats (n = 269) (Tables 1–3). In a, estimated probability and 95% confidence intervals are shown and in b–c, estimated probability is shown and shaded areas show 95 % confidence intervals.Full size imageTable 1 Summary of best-fit generalized linear mixed effects model for reservoir status (n = 436)Full size tableCompared to non-reservoirs, we also found that rodent reservoirs are disproportionately exploited by humans (hunted for meat and fur). Seventy-two of the regularly hunted rodent species (n = 83) are reservoirs (87%), and hunted rodent species harbour on average five times the number of zoonotic pathogens than non-hunted species (Table 2).Table 2 Summary of rodent characteristics divided by rodent group with respect to hunting, reservoir status, and synanthropic behaviourFull size tableWe explored causal pathways using a structural equation model (SEM) linking synanthropy, reservoir status, and their hypothesized predictors. The final model, which we established a priori, had 17 free parameters and 21 degrees of freedom (n = 269). The model fit, based on the SRMR (standardized root mean squared residual) and the RMSEA (root mean squared error of approximation) indicated a good fit (see Methods). From the initially formulated full model, the pathways linking reservoir status to population fluctuations (s-index, Methods), occurrence in grasslands, number of artificial habitats a species occurs in, and number of studies found per species were not significant and thus removed from the final model (Supplementary Fig. 3). Similarly, pathways linking synanthropy and occurrence in grasslands were not significant and also removed. All reported coefficients for pathways are standardized to facilitate comparisons among the different relationships. The relationships and coefficients below all refer to those in the final model.The focal variable in the model was reservoir status, which was strongly and positively associated with synanthropy and had the highest estimated pathway coefficient (standardised estimate = 0.58, 95% CI 0.49–0.66, Fig. 2). Controlling for synanthropy, species were more likely to be a reservoir with increasing adult weight (0.13, 0.04–0.22). Species that occur in savanna were less likely to be reservoirs (−0.13, −0.22 to −0.04), while hunted species were more likely to be reservoirs (Fig. 2, 0.20, 0.11–0.30).Synanthropy was influenced by four habitat variables: a species was more likely to be synanthropic if it occurs in a higher number of artificial habitats (0.17, 0.04–0.31), and occurs in urban areas (0.14, 0.01–0.27), deserts (0.12, 0.01–0.23), or forests (0.13, 0.02–0.24). Notably, species with higher s-index, and thus larger population fluctuations, were more likely to be synanthropic (0.12, 0.01–0.22), and the s-index itself decreased as adult weight increased (−0.16, −0.27 to −0.04). Finally, hunted species were characterized by higher adult bodyweight (0.35, 0.25–0.44) (Fig. 2).The number of studies per species was positively associated with both a species’ synanthropic behaviour (0.29, 0.19–0.39) and its reservoir status (0.09, 0.00– 0.19), albeit with weaker evidence for the latter effect (p = 0.054) (Fig. 2),The confirmatory generalized linear mixed effects models (GLMMs) (Tables 1, 3), which control for correlation among species within the same family, showed that our SEM results were robust. Indeed, synanthropy was a significant predictor of reservoir status. These models underscore synanthropy as the most important predictor of reservoir status in our analysis (Table 1, Figs. 2–3).Table 3 Summary of best-fit generalized linear mixed effects model for synanthropic status (n = 269)Full size tablePopulation fluctuations affect transmission riskOur newly compiled data on the magnitude of population fluctuations enabled comparative investigations beyond theoretically straightforward predictions that transmission risk increases with reservoir abundance for density-dependent systems. We show that while strong population fluctuations (measured as the s-index) are found frequently in both reservoir and non-reservoir rodents (Table 2), synanthropic rodents exhibit much larger population fluctuations compared to non-synanthropic rodents (Table 2, Figs. 2–3). This pattern was apparent despite broad confidence intervals in the relationship between the s-index and the probability of being synanthropic (Fig. 3b, Tables 2, 3). Taken together, our results suggest that larger population fluctuations in reservoir species increase zoonotic transmission risk via synanthropic behaviours of rodents, thereby increasing the likelihood of zoonotic spillover infection to humans.Habitat generalism and habitat transformation increase transmission riskWe also find that reservoir species thrive in human-created (artificial) habitats (Fig. 3a, c, Tables 2–3), which reflects a general flexibility in their use of diverse habitat types compared to non-reservoir species (Fig. 4a, Table 2). In addition, the number of zoonotic pathogens harboured by a rodent species increased with habitat breadth (r436 = 0.34, p  More

Keeling, P. J. & Burki, F. Progress towards the tree of eukaryotes. Curr. Biol. 29, R808–R817 (2019).Article 
CAS 

Google Scholar 
Gawryluk, R. M. R. et al. Non-photosynthetic predators are sister to red algae. Nature 572, 240–243 (2019).Article 
CAS 

Google Scholar 
Janouškovec, J. et al. A new lineage of eukaryotes illuminates early mitochondrial genome reduction. Curr. Biol. 27, 3717–3724 (2017).Article 

Google Scholar 
Lax, G. et al. Hemimastigophora is a novel supra-kingdom-level lineage of eukaryotes. Nature 564, 410–414 (2018).Article 
ADS 
CAS 

Google Scholar 
Oren, A. Prokaryote diversity and taxonomy: current status and future challenges. Philos. Trans. R. Soc. Lond. B 359, 623–638 (2004).Article 
CAS 

Google Scholar 
Shu, W. S. & Huang, L. N. Microbial diversity in extreme environments. Nat. Rev. Microbiol. 20, 219–235 (2022).Article 
CAS 

Google Scholar 
Massana, R., del Campo, J., Sieracki, M. E., Audic, S. & Logares, R. Exploring the uncultured microeukaryote majority in the oceans: reevaluation of ribogroups within stramenopiles. ISME J. 8, 854–866 (2014).Article 

Google Scholar 
de Vargas, C. et al. Eukaryotic plankton diversity in the sunlit ocean. Science 348, 1261605 (2015).Article 

Google Scholar 
Flegontova, O. et al. Extreme diversity of diplonemid eukaryotes in the ocean. Curr. Biol. 26, 3060–3065 (2016).Article 
CAS 

Google Scholar 
Ahlering, M. A. & Carrel, J. E. Predators are rare even when they are small. Oikos 95, 471–475 (2001).Article 

Google Scholar 
Hehenberger, E. et al. Novel predators reshape holozoan phylogeny and reveal the presence of a two-component signaling system in the ancestor of animals. Curr. Biol. 27, 2043–2050 (2017).Article 
CAS 

Google Scholar 
Tikhonenkov, D. V. et al. Description of Colponema vietnamica sp. n. and Acavomonas peruviana n. gen. n. sp., two new alveolate phyla (Colponemidia nom. nov. and Acavomonidia nom. nov.) and their contributions to reconstructing the ancestral state of alveolates and eukaryotes. PLoS ONE 9, e95467 (2014).Article 
ADS 

Google Scholar 
Tikhonenkov, D. V. et al. New lineage of microbial predators adds complexity to reconstructing the evolutionary origin of animals. Curr. Biol. 30, 4500–4509 (2020).Article 
CAS 

Google Scholar 
Mylnikov, A. P. & Tikhonenkov, D. V. The new alveolate carnivorous flagellate Colponema marisrubri sp. n. (Colponemida, Alveolata) from the Red Sea. Zool. Zh. 88, 1163–1169 (2009).
Google Scholar 
Strassert, J. F. H., Irisarri, I., Williams, T. A. & Burki, F. A molecular timescale for eukaryote evolution with implications for the origin of red algal-derived plastids. Nat. Commun. 12, 1879 (2021).Article 
ADS 
CAS 

Google Scholar 
Rodriguez-Ezpeleta, N. et al. Detecting and overcoming systematic errors in genome-scale phylogenies. Syst. Biol. 56, 389–399 (2007).Article 
CAS 

Google Scholar 
Strassert, J. F. H., Jamy, M., Mylnikov, A. P., Tikhonenkov, D. V. & Burki, F. New phylogenomic analysis of the enigmatic phylum Telonemia further resolves the eukaryote tree of life. Mol. Biol. Evol. 36, 757–765 (2019).Article 
CAS 

Google Scholar 
Lanfear, R., Kokko, H. & Eyre-Walker, A. Population size and the rate of evolution. Trends Ecol. Evol. 29, 33–41 (2014).Article 

Google Scholar 
Bahler, M. & Rhoads, A. Calmodulin signaling via the IQ motif. FEBS Lett. 513, 107–113 (2002).Article 
CAS 

Google Scholar 
Schaffer, D. E., Iyer, L. M., Burroughs, A. M. & Aravind, L. Functional innovation in the evolution of the calcium-dependent system of the eukaryotic endoplasmic reticulum. Front. Genet. 11, 34 (2020).Article 

Google Scholar 
Morita-Yamamuro, C. et al. The Arabidopsis gene CAD1 controls programmed cell death in the plant immune system and encodes a protein containing a MACPF domain. Plant Cell Physiol. 46, 902–912 (2005).Article 
CAS 

Google Scholar 
Rosado, C. J. et al. The MACPF/CDC family of pore-forming toxins. Cell. Microbiol. 10, 1765–1774 (2008).Article 
CAS 

Google Scholar 
Ishino, T., Chinzei, Y. & Yuda, M. A Plasmodium sporozoite protein with a membrane attack complex domain is required for breaching the liver sinusoidal cell layer prior to hepatocyte infection. Cell. Microbiol. 7, 199–208 (2005).Article 
CAS 

Google Scholar 
Satoh, H., Oshiro, N., Iwanaga, S., Namikoshi, M. & Nagai, H. Characterization of PsTX-60B, a new membrane-attack complex/perforin (MACPF) family toxin, from the venomous sea anemone Phyllodiscus semoni. Toxicon 49, 1208–1210 (2007).Article 
CAS 

Google Scholar 
Tikhonenkov, D. V., Mazei, Y. A. & Embulaeva, E. A. Degradation succession of heterotrophic flagellate communities in microcosms. Zh. Obs. Biol. 69, 57–64 (2008).CAS 

Google Scholar 
Tikhonenkov, D. V. et al. On the origin of TSAR: morphology, diversity and phylogeny of Telonemia. Open Biol. 12, 210325 (2022).Article 
CAS 

Google Scholar 
Picelli, S. et al. Full-length RNA-seq from single cells using Smart-seq2. Nat. Protoc. 9, 171–181 (2014).Article 
CAS 

Google Scholar 
Keeling, P. J., Poulson, N. & McFadden, G. I. Phylogenetic diversity of parabasalian symbionts from termites, including the phylogenetic position of Pseudotrypanosoma and Trichonympha. J. Eukaryot. Microbiol. 45, 643–650 (1998).Article 
CAS 

Google Scholar 
Medlin, L., Elwood, H. J., Stickel, S. & Sogin, M. L. The characterization of enzymatically amplified eukaryotic 16S-like rRNA-coding regions. Gene 71, 491–499 (1988).Article 
CAS 

Google Scholar 
Tikhonenkov, D. V., Janouškovec, J., Keeling, P. J. & Mylnikov, A. P. The morphology, ultrastructure and SSU rRNA gene sequence of a new freshwater flagellate, Neobodo borokensis n. sp. (Kinetoplastea, Excavata). J. Eukaryot. Microbiol. 63, 220–232 (2016).Article 
CAS 

Google Scholar 
Andrews, S. FastQC: a quality control tool for high throughput sequence data (Babraham Bioinformatics, 2010); https://www.bioinformatics.babraham.ac.uk/projects/fastqc/.Zhang, J., Kobert, K., Flouri, T. & Stamatakis, A. PEAR: a fast and accurate Illumina Paired-End reAd mergeR. Bioinformatics 30, 614–620 (2013).Article 

Google Scholar 
Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30, 2114–2120 (2014).Article 
CAS 

Google Scholar 
Grabherr, M. G. et al. Full-length transcriptome assembly from RNA-seq data without a reference genome. Nat. Biotechnol. 29, 644–652 (2011).Article 
CAS 

Google Scholar 
Altschul, S. F., Gish, W., Miller, W., Myers, E. W. & Lipman, D. J. Basic local alignment search tool. J. Mol. Biol. 215, 403–410 (1990).Article 
CAS 

Google Scholar 
Laetsch, D. R. & Blaxter, M. L. BlobTools: interrogation of genome assemblies. F1000Research 6, 1287 (2017).Article 

Google Scholar 
Haas, B. J. et al. Denovo transcript sequence reconstruction from RNA-seq using the Trinity platform for reference generation and analysis. Nat. Protoc. 8, 1494–1512 (2013).Article 
CAS 

Google Scholar 
Li, W. & Godzik, A. Cd-hit: a fast program for clustering and comparing large sets of protein or nucleotide sequences. Bioinformatics 22, 1658–1659 (2006).Article 
CAS 

Google Scholar 
Buchfink, B., Xie, C. & Huson, D. H. Fast and sensitive protein alignment using DIAMOND. Nat. Methods 12, 59–60 (2015).Article 
CAS 

Google Scholar 
Shen, W. & Ren, H. TaxonKit: a practical and efficient NCBI taxonomy toolkit. J. Genet. Genomics 48, 844–850 (2021).Richter, D. J. et al. EukProt: a database of genome-scale predicted proteins across the diversity of eukaryotes. Peer Community Journal 2, e56 (2022).Simao, F. A., Waterhouse, R. M., Ioannidis, P., Kriventseva, E. V. & Zdobnov, E. M. BUSCO: assessing genome assembly and annotation completeness with single-copy orthologs. Bioinformatics 31, 3210–3212 (2015).Article 
CAS 

Google Scholar 
Kanehisa, M., Furumichi, M., Sato, Y., Ishiguro-Watanabe, M. & Tanabe, M. KEGG: integrating viruses and cellular organisms. Nucleic Acids Res. 49, D545–D551 (2021).Article 
CAS 

Google Scholar 
Moriya, Y., Itoh, M., Okuda, S., Yoshizawa, A. C. & Kanehisa, M. KAAS: an automatic genome annotation and pathway reconstruction server. Nucleic Acids Res. 35, W182–W185 (2007).Article 

Google Scholar 
Burki, F. The eukaryotic tree of life from a global phylogenomic perspective. Cold Spring Harb. Perspect. Biol. 6, a016147 (2014).Article 

Google Scholar 
Waskom, M. et al. mwaskom/Seaborn: v0.8.1 (September 2017). Zenodo https://doi.org/10.5281/zenodo.883859 (2017).Eddy, S. R. Accelerated profile HMM searches. PLoS Comput. Biol. 7, e1002195 (2011).Article 
ADS 
MathSciNet 
CAS 

Google Scholar 
Finn, R. D. et al. The Pfam protein families database: towards a more sustainable future. Nucleic Acids Res. 44, D279–D285 (2016).Article 
CAS 

Google Scholar 
Letunic, I. & Bork, P. 20 years of the SMART protein domain annotation resource. Nucleic Acids Res. 46, D493–D496 (2018).Article 
CAS 

Google Scholar 
Almagro Armenteros, J. J. et al. SignalP 5.0 improves signal peptide predictions using deep neural networks. Nat. Biotechnol. 37, 420–423 (2019).Article 
CAS 

Google Scholar 
Katoh, K. & Standley, D. M. MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol. Biol. Evol. 30, 772–780 (2013).Article 
CAS 

Google Scholar 
Burns, J. A., Pittis, A. A. & Kim, E. Gene-based predictive models of trophic modes suggest Asgard archaea are not phagocytotic. Nat. Ecol. Evol. 2, 697–704 (2018).Article 

Google Scholar 
Emms, D. M. & Kelly, S. OrthoFinder: phylogenetic orthology inference for comparative genomics. Genome Biol. 20, 238 (2019).Article 

Google Scholar 
Altschul, S. F. et al. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25, 3389–3402 (1997).Article 
CAS 

Google Scholar 
Hall, T. A. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Symp. Ser. 41, 95–98 (1999).CAS 

Google Scholar 
Minh, B. Q. et al. IQ-TREE 2: new models and efficient methods for phylogenetic inference in the genomic era. Mol. Biol. Evol. 37, 1530–1534 (2020).Article 
CAS 

Google Scholar 
Whelan, S., Irisarri, I. & Burki, F. PREQUAL: detecting non-homologous characters in sets of unaligned homologous sequences. Bioinformatics 34, 3929–3930 (2018).CAS 

Google Scholar 
Capella-Gutierrez, S., Silla-Martinez, J. M. & Gabaldon, T. trimAl: a tool for automated alignment trimming in large-scale phylogenetic analyses. Bioinformatics 25, 1972–1973 (2009).Article 
CAS 

Google Scholar 
Roure, B., Rodriguez-Ezpeleta, N. & Philippe, H. SCaFoS: a tool for selection, concatenation and fusion of sequences for phylogenomics. BMC Evol. Biol. 7, S2 (2007).Article 

Google Scholar 
Lartillot, N., Rodrigue, N., Stubbs, D. & Richer, J. PhyloBayes MPI: phylogenetic reconstruction with infinite mixtures of profiles in a parallel environment. Syst. Biol. 62, 611–615 (2013).Article 
CAS 

Google Scholar 
Dayhoff, M., Schwartz, R. & Orcutt, B. in Atlas of Protein Sequence and Structure (ed. Dayhoff, M.) 345–352 (National Biomedical Research Foundation, 1978).Susko, E. & Roger, A. J. On reduced amino acid alphabets for phylogenetic inference. Mol. Biol. Evol. 24, 2139–2150 (2007).Article 
CAS 

Google Scholar 
Lartillot, N. & Philippe, H. A Bayesian mixture model for across-site heterogeneities in the amino-acid replacement process. Mol. Biol. Evol. 21, 1095–1109 (2004).Article 
CAS 

Google Scholar 
Quang le, S., Gascuel, O. & Lartillot, N. Empirical profile mixture models for phylogenetic reconstruction. Bioinformatics 24, 2317–2323 (2008).Article 

Google Scholar 
Wang, H. C., Minh, B. Q., Susko, E. & Roger, A. J. Modeling site heterogeneity with posterior mean site frequency profiles accelerates accurate phylogenomic estimation. Syst. Biol. 67, 216–235 (2018).Article 
CAS 

Google Scholar 
Kück, P. & Struck, T. H. BaCoCa—a heuristic software tool for the parallel assessment of sequence biases in hundreds of gene and taxon partitions. Mol. Phylogenet. Evol. 70, 94–98 (2014).Article 

Google Scholar 
Shimodaira, H. An approximately unbiased test of phylogenetic tree selection. Syst. Biol. 51, 492–508 (2002).Article 

Google Scholar 
Kumar, S., Stecher, G. & Tamura, K. MEGA7: molecular evolutionary genetics analysis version 7.0 for bigger datasets. Mol. Biol. Evol. 33, 1870–1874 (2016).Article 
CAS 

Google Scholar 
Bankevich, A. et al. SPAdes: a new genome assembly algorithm and its applications to single-cell sequencing. J. Comput. Biol. 19, 455–477 (2012).Article 
MathSciNet 
CAS 

Google Scholar 
Dierckxsens, N., Mardulyn, P. & Smits, G. NOVOPlasty: de novo assembly of organelle genomes from whole genome data. Nucleic Acids Res. 45, e18 (2017).
Google Scholar 
Kuznetsov, A. & Bollin, C. J. in Multiple Sequence Alignment (ed. Katoh, K.) 261–295 (Springer, 2021).Lohse, M., Drechsel, O., Kahlau, S. & Bock, R. OrganellarGenomeDRAW—a suite of tools for generating physical maps of plastid and mitochondrial genomes and visualizing expression data sets. Nucleic Acids Res. 41, W575–W581 (2013).Article 

Google Scholar 
Johnson, P. Z., Kasprzak, W. K., Shapiro, B. A. & Simon, A. E. RNA2Drawer: geometrically strict drawing of nucleic acid structures with graphical structure editing and highlighting of complementary subsequences. RNA Biol. 16, 1667–1671 (2019).Article 

Google Scholar 
Burger, G., Gray, M. W., Forget, L. & Lang, B. F. Strikingly bacteria-like and gene-rich mitochondrial genomes throughout jakobid protists. Genome Biol. Evol. 5, 418–438 (2013).Article 

Google Scholar 
Criscuolo, A. & Gribaldo, S. BMGE (Block Mapping and Gathering with Entropy): a new software for selection of phylogenetic informative regions from multiple sequence alignments. BMC Evol. Biol. 10, 210 (2010).Article 

Google Scholar 
Zhang, D. et al. PhyloSuite: an integrated and scalable desktop platform for streamlined molecular sequence data management and evolutionary phylogenetics studies. Mol. Ecol. Resour. 20, 348–355 (2020).Article 

Google Scholar 
Nguyen, L.-T., Schmidt, H. A., von Haeseler, A. & Minh, B. Q. IQ-TREE: a fast and effective stochastic algorithm for estimating maximum-likelihood phylogenies. Mol. Biol. Evol. 32, 268–274 (2015).Article 
CAS 

Google Scholar 
Ibarbalz, F. M. et al. Global trends in marine plankton diversity across kingdoms of life. Cell 179, 1084–1097 (2019).Article 
CAS 

Google Scholar 
Massana, R. et al. Marine protist diversity in European coastal waters and sediments as revealed by high-throughput sequencing. Environ. Microbiol. 17, 4035–4049 (2015).Article 
CAS 

Google Scholar 
Gendron, E. M. S., Darcy, J. L., Hell, K. & Schmidt, S. K. Structure of bacterial and eukaryote communities reflect in situ controls on community assembly in a high-alpine lake. J. Microbiol. 57, 852–864 (2019).Article 
CAS 

Google Scholar 
Minerovic, A. D. et al. 18S-V9 DNA metabarcoding detects the effect of water-quality impairment. Ecol. Indic. 113, 106225 (2020).Article 
CAS 

Google Scholar 
Pearman, J. K. et al. Cross-shelf investigation of coral reef cryptic benthic organisms reveals diversity patterns of the hidden majority. Sci. Rep. 8, 8090 (2018).Article 
ADS 
CAS 

Google Scholar 
Rodas, A. M. et al. Eukaryotic plankton communities across reef environments in Bocas del Toro Archipelago, Panamá. Coral Reefs 39, 1453–1467 (2020).Article 

Google Scholar 
Schoenle, A. et al. High and specific diversity of protists in the deep-sea basins dominated by diplonemids, kinetoplastids, ciliates and foraminiferans. Commun. Biol. 4, 501 (2021).Article 
CAS 

Google Scholar 
Schulhof, M. A. et al. Sierra Nevada mountain lake microbial communities are structured by temperature, resources and geographic location. Mol. Ecol. 29, 2080–2093 (2020).Article 
CAS 

Google Scholar 
Yi, Z. et al. High-throughput sequencing of microbial eukaryotes in Lake Baikal reveals ecologically differentiated communities and novel evolutionary radiations. FEMS Microbiol. Ecol. 93, fix073 (2017). More

Aagaard-Tillery KM, Grove K, Bishop J, Ke X, Fu Q, McKnight R et al. (2008) Developmental origins of disease and determinants of chromatin structure: maternal diet modifies the primate fetal epigenome. J Mol Endocrinol 41:91–102Article 
CAS 

Google Scholar 
Alekseev V, Lampert W (2001) Maternal control of resting – egg production in Daphnia. Nature 414:899–901Article 
CAS 

Google Scholar 
Anderson OS, Sant KE, Dolinoy DC (2012) Nutrition and epigenetics: an interplay of dietary methyl donors, one-carbon metabolism and DNA methylation. J Nutritional Biochem 23:853–859Article 
CAS 

Google Scholar 
Agrawal AA, Laforsch C, Tollrian R (1999) Transgenerational induction of defenses in animals and plants. Nature 401:60–63Article 
CAS 

Google Scholar 
Asselman J, De Coninck DIM, Vandegehuchte MB, Jansen M, Decaestecker E, De Meester L, Busshe JV, Vanhaecke L, Janssen CR, De Schamphelaere KAC (2015) Global cytosine methylation in Daphnia magna depends on genotype, environment, and their interaction. Environ Toxicol 34:5
Google Scholar 
Bernardo J (1996) Maternal effects in animal ecology. Am Zool 36:83–105Article 

Google Scholar 
Berger SL (2007) The complex language of chromatin regulation during transcription. Nature 447:407–412Article 
CAS 

Google Scholar 
Bewick AJ, Vogel KJ, Moore AJ, Schmitz RJ (2017) Evolution of DNA methylation across insects. Mol Biol Evol 34:654–665CAS 

Google Scholar 
Bird A (2007) Perceptions of epigenetics. Nature 447:396–398Article 
CAS 

Google Scholar 
Boersma M (1995) The allocation of resources to reproduction in Daphnia galeata: against the odds? Ecology 76(4):121–1261Article 

Google Scholar 
Boersma M (1997) Offspring size in Daphnia: does it pay to be overweight? Hydrobiologia 360:79–88Article 

Google Scholar 
Boycott AE, Diver C (1923) On the inheritance of the sinistrality in Limnea peregra. Proc R Soc Lond B 95:207–213Article 

Google Scholar 
Brett MT (1993) Resource quality effects on Daphnia longispina offspring fitness. J Plankton Res 15(4):403–412Article 

Google Scholar 
Burns CW (1995) Effects of crowding and different food levels on growth and reproductive investment of Daphnia. Oecologia 101:234–244Article 

Google Scholar 
Cameron NM, Shahrokh D, Del Corpo A, Dhir SK, Szyf M, Champagne FA, Meaney MJ (2008) Epigenetic programming of phenotypic variations in reproductive strategies in the rat through maternal care. J Neuroendocrinol 20:795–801Article 
CAS 

Google Scholar 
Champagne FA (2012) Epigenetics and developmental plasticity across species. Dev Psychobiol 55:33–41Article 

Google Scholar 
Chan SY, Vasilopoulou E, Kilby MD (2009) The role of the placenta in thyroid hormone delivery to the fetus. Nat Clin Pract Endocrinol Metab 5:45–54Article 
CAS 

Google Scholar 
Chong S, Whitelaw E (2004) Epigenetic germline inheritance. Curr Opin Genet Dev 14(6):692–696Article 
CAS 

Google Scholar 
Clark J, Garbutt JS, McNally L, Little TJ (2017) Disease spread in age structured populations with maternal age effects. Ecol Lett 20:445–451Article 

Google Scholar 
Colbourne JK, Herbert PDN, Taylor DJ (1997) Evolutionary origins of phenotypic diversity. In: Givnish TJ, Systma KJ (eds) Daphnia in molecular evolution and adaptive radiation. Cambridge University Press. p 163–188Colbourne JK, Pfrender ME, Gilbert D, Thomas WK, Tucker A, Oakley TH et al. (2011) The ecoresponsive genome of Daphnia pulex. Science 331(6017):555–561Article 
CAS 

Google Scholar 
Desmarais KH (1997) Keeping Daphnia out of the surface film with cetyl alcohol. J Plankton Res 19(1):149–154Article 

Google Scholar 
Dorts J, Falisse E, Schoofs E, Flamion E, Kestermont P, Silvestre F (2016) DNA methyltransferases and stress-related genes expression in zebrafish larvae after exposure to heat and copper during reprogramming of DNA methylation. Sci Rep 6:34254Article 
CAS 

Google Scholar 
Ducker GS, Rabinowitz JD (2016) One-carbon metabolism in health and disease. Cell Metab 25:27–42. https://doi.org/10.1016/j.cmet.2016.08Dudycha JL, Brandon CS, Deitz KC (2012) Population genomics of resource exploitation: insights from gene expression profiles of two Daphnia ecotypes fed alternate resources. Ecol Evol 2:329–340Dzialowski EM, Reed WL, Sotherland PR (2009) Effects of egg size on double-crested cormorant (Phalacrocorax auritus) egg composition and hatchling phenotype. Comp Biochem Physiol A Mol Integr Physiol 152:262–267Article 

Google Scholar 
Frost PC, Ebert D, Larson JH, Marcus MA, Wagner ND, Zalewski A (2010) Transgenerational effects of poor elemental food quality on Daphnia magna. Oecologia 162(4):865–872Article 

Google Scholar 
Gabsi F, Glazier DS, Hammers-Wirtz M, Ratte HT, Preuss TG (2014) How to interactive maternal traits and environmental factors determine offspring size in Daphnia magna?. Ann Limnol 50:9–18Article 

Google Scholar 
Garbutt JS, Little TJ (2016) Bigger is better: changes in body size explain a maternal effect of food on offspring disease resistance. Ecol Evolution 7:1403–1409Article 

Google Scholar 
Gibney ER, Nolan CM (2010) Epigenetics and gene expression. Heredity 105:4–13Article 
CAS 

Google Scholar 
Gillis MK, Walsh MR (2019) Individual variation in plasticity dulls transgenerational responses to stress. Funct Ecol 33:1993–2002Glazier DS (1992) Effects of food, genotype, and maternal size and age on offspring investment in Daphnia magna. Ecology 73(3):910–926Article 

Google Scholar 
Gliwicz ZM, Guisande C (1992) Family planning in Daphnia: resistance to starvation in offspring born to mothers grown at different food levels. Oceologia 91:463–467Article 

Google Scholar 
Goos JM, Swain CJ, Munch SB, Walsh MR (2018) Maternal diet and age alter direct and indirect relationships between lifer-history traits across multiple generations. Funct Ecol 33:491–502Article 

Google Scholar 
Goulden CE, Horning LL (1980) Population oscillations and energy reserves in planktonic cladocera and their consequences to competition. Proc Natl Acad Sci USA 77:1716–1720Article 
CAS 

Google Scholar 
Groothuis TG, Schwabl H (2008) Hormone-mediated maternal effects in birds: mechanisms matter but what do we know of them? Philos Trans R Soc Lond B Biol Sci 363:1647–1661Article 
CAS 

Google Scholar 
Guisande C, Gliwicz ZM (1992) Egg size and clutch size in two Daphnia species at different food levels. J Plankton Res 14(7):997–1007Article 

Google Scholar 
Hearn J, Chow FW-N, Barton H, Tung M, Wilson P, Blaxter M et al. (2018) Daphnia magna microRNAs respond to nutritional stress and ageing but are not transgenerational. Mol Ecol 27:1402–1412Article 
CAS 

Google Scholar 
Hearn J, Pearson M, Blaxter M, Wilson PJ, Little TJ (2019) Genome-wide methylation is modified by caloric restriction in Daphnia magna. BCM Genetics 20:197Hearn J, Plenderleith F, Little TJ (2021) DNA methylation differs extensively between strains of the same geographical origin and changes with age in Daphnia magna. Epigenetics Chromatin 14:4. https://doi.org/10.1186/s13072-020-00379-zHead JA (2014) Patterns of DNA methylation in animals: an ecotoxicological perspective. Integr Comp Biol 54:77–86Article 
CAS 

Google Scholar 
Hebert PDN (1981) Obligate asexuality in Daphnia. Am Nate 117:784–789Article 

Google Scholar 
Herman JJ, Sultan SE (2016) DNA methylation mediates genetic variation for adaptive transgenerational plasticity. Proc Biol Sci 283(1838):20160988. https://doi.org/10.1098/rspb.2016.0988Article 
CAS 

Google Scholar 
Hiruta C, Nishida C, Tochinai S (2010) Abortive meiosis in the oogenesis of parthenogenetic Daphnia pulex. Chromosome Res 18:833–840Article 
CAS 

Google Scholar 
Ho DH, Burggren WW (2010) Epigenetics and transgenerational transfer: a physiological perspective. J Exp Biol 213:3–16Article 
CAS 

Google Scholar 
Ho DH (2008) Morphological and physiological developmental consequences of parental effects in the chicken embryo (Gallus gallus domesticus) and the zebrafish larva (Danio rerio). Diss: University of North TexasInnes DJ, Fox CJ, Winsor GL (2000) Avoiding the cost of males in obligately asexual Daphnia pulex (Leydig). Proc: Biol Sci 267(1447):991–997CAS 

Google Scholar 
Jeremias G, Barbosa J, Marques SM, De Schamphelaere KAC, Van Nieuwerburgh F, Deforce D, Gonçalves FJM, Pereira JL, Asselman J (2018) Transgenerational inheritance of dna hypomethylation in Daphnia magna in response to salinity stress. Environ Sci Technol 52(17):10114–10123Article 
CAS 

Google Scholar 
Jian X, Yang W, Zhao S, Liang H, Zhao Y, Chen L et al. (2013) Maternal effects of inducible tolerance against the toxic cyanobacterium Microcystis aeruginosa in the grazer Daphnia carinata. Environ Pollut 178:142–146Article 

Google Scholar 
Keating KI (1985) The influence of vitamin-B12 deficiency on the reproduction of Daphnia-Pulex Leydig (Cladocera). J Crustacean Biol 5:30–136Article 

Google Scholar 
Kleiven OT, Larsson P, Hobaek A (1992) Sexual reproduction in Daphnia magna requires three stimulie. Oikos 65:197–206Article 

Google Scholar 
Kusari F, O’Doherty AM, Hodges NJ, Wojewodzic MW (2017) Bi-directional effects of vitamin B12 and methotrexate on Daphnia magna fitness and genomic methylation. Sci Rep 7:11872Article 

Google Scholar 
Kvist J, Athanasio CG, Solari OS, Brown JB, Colbourne JK, Pfrender ME, Mirbahai L (2018) Pattern of DNA methylation in Daphnia: evolutionary perspective. Genome Biol Evolution 10(8):1988–2007Article 
CAS 

Google Scholar 
Lamka GF, Harder AM, Sundaram M, Schwartz TS, Christie MR, DeWoody JA, Willoughby JR (2022) Epigenetics in ecology, evolution, and conservation. Front Ecol Evol 10. https://doi.org/10.3389/fevo.2022.871791LaMontagne JM, McCauley E (2001) Maternal effects in Daphnia: what mothers are telling their offspring and do they listen. Ecol Lett 4:64–71Article 

Google Scholar 
Li Q, Jiang X (2014) Offspring tolerance to toxic Microcystis aeruginosa in Daphnia pulex shaped by maternal food availability and age. Fundam Appl Limnol 185:315–319Article 

Google Scholar 
Mastorci F, Vicentini M, Viltart O, Manghi M, Graiani G, Quaini Fet al. (2009) Long-term effects of prenatal stress: changes in adult cardiovascular regulation and sensitivity to stress. Neurosci Biobehav Rev 33:191–203Article 

Google Scholar 
Mkee D, Ebert D (1996) The interactive effects of temperature, food level and maternal phenotype on offspring size in Daphnia magna. Oecologia 107(2):189–196Article 

Google Scholar 
Mousseau TA, Fox CW (1998) The adaptive significance of maternal effects. Trends Ecol Evol 13:403–407Article 
CAS 

Google Scholar 
Nguyen ND, Matsuura T, Kato Y, Watanabe H (2020) Caloric restriction upregulates the expression of DNMT3.1, lacking the conserved catalytic domain, in Daphnia magna. Genesis 58:12Article 

Google Scholar 
Nguyen ND, Matsuura T, Kato Y, Watanabe H (2021) DNMT3.1 controls trade-offs between growth, reproduction, and life span under starved conditions in Daphnia magna. Sci Rep 11:7326Article 
CAS 

Google Scholar 
Nusslein-Volhard C, Frohnhofer HG, Lehmann R (1987) Determination of anteroposterior polarity in Drosophila. Science 238:1675–1681Article 
CAS 

Google Scholar 
Pieters BJ, Liess M (2006) Maternal nutritional state determines the sensitivity of Daphnia magna offspring to short-term fenvalerate exposure. Aquat Toxicol 76:286–277Article 

Google Scholar 
R Core Team (2021) R: a language and environmental for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/Richards EJ (2006) Inherited epigenetic variation – revisiting soft inheritance. Nat Rev Genet 7:395–401Article 
CAS 

Google Scholar 
Stollewerk A (2010) The water flea Daphnia – a new model system for ecology and evolution? J Biol 9(2):21Article 

Google Scholar 
Sturtevant AH (1923) Inheritance of direction of coiling in Limnea. Science 58:269Article 
CAS 

Google Scholar 
Suter MA, Chen A, Burdine MS, Choudhury M, Harris RA, Lane RH et al. (2012) A maternal high-fat diet modulates fetal SIRT1 histone and protein deacetylase activity in nonhuman primates. FASEB J 26:5106–5114Article 
CAS 

Google Scholar 
Tessier AJ, Consolatti NL (1989) Variation in offspring size in Daphnia and consequences for individual fitness. Oikos 56:269–276Article 

Google Scholar 
Tessier AJ, Consolatti NL (1991) Resource quantity and offspring quality in Daphnia. Ecology 72(2):468–478Article 

Google Scholar 
Trerotola M, Relli V, Simeone P, Alberti S (2015) Epigenetic inheritance and the missing heritability. Hum Genomics 9(1):17. https://doi.org/10.1186/s40246-015-0041-3Article 
CAS 

Google Scholar 
Trijau M, Asselman J, Armant O, Adam-Guillermin C, De Schamphelaere KAC, Alonzo F (2018) Transgenerational DNA methylation changes in Daphnia magna exposed to chronic γ irradiation. Environ Sci Technol 52(7):4331–4339Article 
CAS 

Google Scholar 
Urabe J, Sterner RW (2001) Contrasting effects of different types of resource depletion on life-history traits in Daphnia. Funct Ecol 15:165–174Article 

Google Scholar 
Vandegehuchte MB, Kyndt T, Vanholme B, Haegeman A, Gheysen G, Janssen CR (2009a) Occurrence of DNA methylation in Daphnia magna and influence of multigeneration Cd exposure. Environ Int 35(4):700–706Article 
CAS 

Google Scholar 
Vandegehuchte MB, Lemiere F, Janssen CR (2009b) Quantitative DNA-methylation in Daphnia magna and effects of multigeneration Zn exposure. Comp Biochem Physiol C Toxicol Pharmacol 150:343–348Article 
CAS 

Google Scholar 
Vandegehuchte MB, Janssen CR (2011) Epigenetics and its implications for ecotoxicology. Ecotoxicology 20:607–624Article 
CAS 

Google Scholar 
Vandegehuchte MB, Janssen CR (2014) Epigenetics in an ecotoxicological context. Mutat Res Genet Toxicol Environ Mutagen 764–765:36–45Article 

Google Scholar 
Walsh MR, La Pierre KJ, Post DM (2014) Phytoplankton composition modifies predator-driven life history evolution in Daphnia. Evol Ecol 28:397–411Article 

Google Scholar 
Walsh MR, Cooley F, Biles K, Munch SB (2015) Predator-induced phenotypic plasticity within- and across generations: a challenge for theory? Proc R Soc B Biol Sci 282:20142205Article 

Google Scholar 
Wickham H (2016) ggplot2: elegant graphics for data analysis. Springer-Verlag New York. ISBN 978-3-319-24277-4. https://ggplot2.tidyverse.orgWolf JB, Wade MJ (2009) What are maternal effects (and what are they not)? Philos Trans R Soc Lond B Biol Sci 364(1520):1107–1115Article 

Google Scholar 
Zaffagnini F (1987) Reproduction in Daphnia. Mem Ist Ital Idrobiol 45:245–284
Google Scholar 
Zhao S, Fernald RD (2005) Comprehensive algorithm for quantitative real-time polymerase chain reaction. J Comput Biol 12(8):1045–1062Article 

Google Scholar  More

Listen to the latest from the world of science, with Benjamin Thompson.
Your browser does not support the audio element.

Download MP3

In this episode:00:45 World’s oldest DNA shows that mastodons roamed ancient GreenlandDNA recovered from ancient permafrost has been used to reconstruct what an ecosystem might have looked like two million years ago. Their work suggests that Northern Greenland was much warmer than the frozen desert it is today, with a rich ecosystem of plants and animals.Research Article: Kjær et al.Nature Video: The world’s oldest DNA: Extinct beasts of ancient Greenland08:21 Research HighlightsWhy low levels of ‘good’ cholesterol don’t predict heart disease risk in Black people, and how firework displays affect the flights of geese.Research Highlight: ‘Good’ cholesterol readings can lead to bad results for Black peopleResearch Highlight: New Year’s fireworks chase wild geese high into the sky10:31 Modelling the potential emissions of plasticsWhile the global demand for plastics is growing, the manufacturing and disposal of these ubiquitous materials is responsible for significant CO2 emissions each year. This week, a team have modelled how CO2 emissions could vary in the context of different strategies for mitigating climate change. They reveal how under specific conditions the industry could potentially become a carbon sink.Research Article: Stegmann et al.News and Views: Plastics can be a carbon sink but only under stringent conditionsSubscribe to Nature Briefing, an unmissable daily round-up of science news, opinion and analysis free in your inbox every weekday.Never miss an episode. Subscribe to the Nature Podcast on Apple Podcasts, Google Podcasts, Spotify or your favourite podcast app. An RSS feed for Nature Podcast is available too. More

Fernández-Martínez, M. et al. Nutrient availability as the key regulator of global forest carbon balance. Nat. Clim. Chang. 4, 471–476 (2014).Article 
ADS 

Google Scholar 
Wright, S. J. Plant responses to nutrient addition experiments conducted in tropical forests. Ecol. Monogr. 89, e01382 (2019).Article 

Google Scholar 
Levy-Varon, J. H. et al. Tropical carbon sink accelerated by symbiotic dinitrogen fixation. Nat. Commun. 10, 5637 (2019).Article 
ADS 
CAS 

Google Scholar 
Batterman, S. A. et al. Key role of symbiotic dinitrogen fixation in tropical forest secondary succession. Nature 502, 224–227 (2013).Article 
ADS 
CAS 

Google Scholar 
Ter Steege, H. et al. Continental-scale patterns of canopy tree composition and function across Amazonia. Nature 443, 444–447 (2006).Article 
ADS 

Google Scholar 
Hedin, L. O., Brookshire, E. N. J., Menge, D. N. L. & Barron, A. R. The nitrogen paradox in tropical forest ecosystems. Annu. Rev. Ecol. Evol. Syst. 40, 613–635 (2009).Article 

Google Scholar 
Menge, D. N. L. et al. Patterns of nitrogen-fixing tree abundance in forests across Asia and America. J. Ecol. 107, 2598–2610 (2019).Article 
CAS 

Google Scholar 
Matson, W. J.Jr Herbivory in relation to plant nitrogen content. Annu. Rev. Ecol. Syst. 11, 119–161 (1980).Article 

Google Scholar 
Coley, P. D., Bateman, M. L. & Kusar, T. A. The effects of plant quality on caterpillar growth and defense against natural enemies. Oikos 115, 219–228 (2006).Article 

Google Scholar 
Wieder, W. R., Cleveland, C. C., Smith, W. K. & Todd-Brown, K. Future productivity and carbon storage limited by terrestrial nutrient availability. Nat. Geosci. 8, 441–444 (2015).Article 
ADS 
CAS 

Google Scholar 
Barron, A. R., Purves, D. W. & Hedin, L. O. Facultative nitrogen fixation by canopy legumes in a lowland tropical forest. Oecologia 165, 511–520 (2011).Article 
ADS 

Google Scholar 
McCulloch, L. A. & Porder, S. Light fuels while nitrogen suppresses symbiotic nitrogen fixation hotspots in neotropical canopy gap seedlings. New Phytol. 231, 1734–1745 (2021).Article 
CAS 

Google Scholar 
Brookshire, E. N. J. et al. Symbiotic N fixation is sufficient to support net aboveground biomass accumulation in a humid tropical forest. Sci Rep. 9, 7571 (2019).Article 
ADS 
CAS 

Google Scholar 
Gei, M. et al. Legume abundance along successional and rainfall gradients in Neotropical forests. Nat. Ecol. Evol. 2, 1104–1111 (2018).Article 

Google Scholar 
Vance, C. P. in Nitrogen-fixing Leguminous Symbioses. Nitrogen Fixation: Origins, Applications, and Research Progress, Vol. 7 (eds Dilworth, M. J. et al.) (Springer, 2008).Vitousek, P. M. & Howarth, R. W. Nitrogen limitation on land and in the sea: how can it occur? Biogeochemistry 13, 87–115 (1991).Article 

Google Scholar 
Menge, D. N. L., Levin, S. A. & Hedin, L. O. Evolutionary tradeoffs can select against nitrogen fixation and thereby maintain nitrogen limitation. Proc. Natl Acad. Sci. USA 105, 1573–1578 (2008).Article 
ADS 
CAS 

Google Scholar 
Sheffer, E., Batterman, S. A., Levin, S. A. & Hedin, L. O. Biome-scale nitrogen fixation strategies selected by climatic constraints on nitrogen cycle. Nat. Plants 1, 15182 (2015).Article 
CAS 

Google Scholar 
Vitousek, P. M. & Field, C. B. Ecosystem constraints to symbiotic nitrogen fixers: a simple model and its implications. Biogeochemistry 46, 179–202 (1999).Article 
CAS 

Google Scholar 
Coley, P. D. & Barone, J. A. Herbivory and plant defenses in tropical forests. Annu. Rev. Ecol. Syst. 27, 305–335 (1996).Article 

Google Scholar 
Fyllas, N. M. et al. Basin-wide variations in foliar properties of Amazonian forest: phylogeny, soils and climate. Biogeosciences 6, 2677–2708 (2009).Article 
ADS 

Google Scholar 
Batterman, S. A. et al. Phosphatase activity and nitrogen fixation reflect species differences, not nutrient trading or nutrient balance, across tropical rainforest trees. Ecol. Lett. 21, 1486–1495 (2018).Article 

Google Scholar 
Menge, D. N. L., Wolf, A. A. & Funk, J. L. Diversity of nitrogen fixation strategies in Mediterranean legumes. Nat. Plants 1, 15064 (2015).Article 
CAS 

Google Scholar 
Ritchie, M. E. & Tilman, D. Responses of legumes to herbivores and nutrients during succession on a nitrogen-poor soil. Ecol. Soc. Am. 76, 2648–2655 (1995).
Google Scholar 
Taylor, B. N. & Ostrowsky, L. R. Nitrogen-fixing and non-fixing trees differ in leaf chemistry and defence but not herbivory in a lowland Costa Rican rain forest. J. Trop. Ecol. 35, 270–279 (2019).Article 

Google Scholar 
Endara, M.-J. et al. Coevolutionary arms race versus host defense chase in a tropical herbivore–plant system. Proc. Natl Acad. Sci. USA 114, E7499–E7505 (2017).Article 
CAS 

Google Scholar 
Kursar, T. A. & Coley, P. D. Convergence in defense syndromes of young leaves in tropical rainforests. Biochem. Syst. Ecol. 31, 929–949 (2003).Article 
CAS 

Google Scholar 
Kursar, T. A. et al. The evolution of antiherbivore defenses and their contribution to species coexistence in the tropical tree genus Inga. Proc. Natl Acad. Sci. USA 106, 18073–18078 (2009).Article 
ADS 
CAS 

Google Scholar 
Taylor, B. N. & Menge, D. N. L. Light regulates tropical symbiotic nitrogen fixation more strongly than soil nitrogen. Nat. Plants 4, 655–661 (2018).Article 
CAS 

Google Scholar 
Adams, M., Turnbull, T., Sprent, J. & Buchmann, N. Legumes are different: leaf nitrogen, photosynthesis, and water use efficiency. Proc. Natl Acad. Sci. USA 113, 4098–4103 (2016).Article 
ADS 
CAS 

Google Scholar 
Coley, P. D. Effects of plant growth rate and leaf lifetime on the amount and type of anti-herbivore defense. Oecologia 74, 531–536 (1988).Article 
ADS 
CAS 

Google Scholar 
Batterman, S. A., Wurzburger, N. & Hedin, L. O. Nitrogen and phosphorus interact to control tropical symbiotic N2 fixation: a test in Inga punctata. J. Ecol. 101, 1400–1408 (2013).Article 
CAS 

Google Scholar 
Eichhorn, M. P., Nilus, R., Compton, S. G., Hartley, S. E. & Burslem, D. F. R. P. Herbivory of tropical rain forest tree seedlings correlates with future mortality. Ecology 91, 1092–1101 (2010).Article 

Google Scholar 
Wink, M. Evolution of secondary metabolites in legumes (Fabaceae). South African J. Bot. 89, 164–175 (2013).Article 
CAS 

Google Scholar 
Currano, E. D. & Jacobs, B. F. Bug-bitten leaves from the early Miocene of Ethiopia elucidate the impacts of plant nutrient concentrations and climate on insect herbivore communities. Glob. Planet. Change 207, 103655 (2021).Article 

Google Scholar 
Wieder, W. R., Cleveland, C. C., Lawrence, D. M. & Bonan, G. B. Effects of model structural uncertainty on carbon cycle projections: biological nitrogen fixation as a case study. Environ. Res. Lett. 10, 044016 (2015).Article 
ADS 

Google Scholar 
Sprent, J. I. Legume Nodulation: A Global Perspective (John Wiley, 2009).Leigh, E. G. Jr Tropical Forest Ecology: A View from Barro Colorado Island (Oxford Univ. Press, 1999).Comita, L. S., Muller-Landau, H. C., Aguilar, S. & Hubbell, S. P. Asymmetric density dependence shapes species abundances in a tropical tree community. Science 329, 330–332 (2010).Article 
ADS 
CAS 

Google Scholar 
Queenborough, S. A., Metz, M. R., Valencia, R. & Wright, S. J. Demographic consequences of chromatic leaf defence in tropical tree communities: do red young leaves increase growth and survival? Ann. Bot. 112, 677–684 (2013).Article 

Google Scholar 
Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9, 671 (2012).Article 
CAS 

Google Scholar 
Pasquini, S. C. & Santiago, L. S. Nutrients limit photosynthesis in seedlings of a lowland tropical forest tree species. Oecologia 168, 311–319 (2012).Article 
ADS 
CAS 

Google Scholar 
Collalti, A. & Prentice, I. C. Is NPP proportional to GPP? Waring’s hypothesis 20 years on. Tree Physiol. 39, 1473–1483 (2019).Article 
CAS 

Google Scholar 
Westbrook, J. W. et al. What makes a leaf tough? Patterns of correlated evolution between leaf toughness traits and demographic rates among 197 shade-tolerant woody species in a Neotropical forest. Am. Nat. 177, 800–811 (2011).Article 

Google Scholar 
Wright, S. J. et al. Functional traits and the growth–mortality trade‐off in tropical trees. Ecology 91, 3664–3674 (2010).Article 

Google Scholar 
Kitajima, K. et al. How cellulose-based leaf toughness and lamina density contribute to long leaf lifespans of shade-tolerant species. New Phytol. 195, 640–652 (2012).Article 

Google Scholar 
Kitajima, K., Wright, S. J. & Westbrook, J. W. Leaf cellulose density as the key determinant of inter- and intra-specific variation in leaf fracture toughness in a species-rich tropical forest. Interface Focus https://doi.org/10.1098/rsfs.2015.0100 (2016).Sedio, B. E., Echeverri, J. C. R., Boya, C. A. & Wright, S. J. Sources of variation in foliar secondary chemistry in a tropical forest tree community. Ecology 98, 616–623 (2017).Article 

Google Scholar 
Brooks, M. E. et al. glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. R J. 9, 378–400 (2017).Article 

Google Scholar 
Smithson, M. & Verkuilen, J. A better lemon squeezer? Maximum-likelihood regression with beta-distributed dependent variables. Psychol. Methods 11, 54–71 (2006).Article 

Google Scholar 
Murphy, S. J., Xu, K. & Comita, L. S. Tree seedling richness, but not neighborhood composition, influences insect herbivory in a temperate deciduous forest community. Ecol. Evol. 6, 6310–6319 (2016).Article 

Google Scholar 
Bates, D., Mächler, M., Bolker, B. & Walker, S. Fitting linear mixed-effects models using lme4. Preprint at https://arxiv.org/abs/1406.5823 (2014).Moles, A. T. & Westoby, M. Do small leaves expand faster than large leaves, and do shorter expansion times reduce herbivore damage? Oikos 90, 517–524 (2000).Article 

Google Scholar 
Bürkner, P. C. brms: An R package for Bayesian multilevel models using Stan. J. Stat. Softw. https://doi.org/10.18637/jss.v080.i01 (2017). More

Fossil record of insectsWe compiled all species-level fossil occurrences of insects using https://paleobiodb.org/ (PBDB) as a starting point (downloaded October 12, 2021). The dataset obtained from PBDB contained initially 5808 occurrences for a period ranging from the Asselian to the Rhaetian. The dataset was cleaned of synonyms, outdated combinations, nomina dubia, and other erroneous and doubtful records, based on revisions provided in the literature and/or on the expertise of the authors. After correction, including data addition from the literature, our dataset was composed of 3636 species (1784 genera, and 418 families) for 17,250 occurrences resulting from an in-depth study and curation of the entire bibliography of fossil insects, spanning from the Asselian (lowermost Permian) to the Rhaetian (uppermost Triassic). Although most of the taxa included in the datasets are nominal taxa (published and named), a few unnamed taxa (genera or species) that are considered separate from others were also included, although not formally named in the literature or not published yet. These unpublished taxa are identifiable by the notation ‘fam. nov.’ or ‘gen. nov.’ following their names.Occurrences used here are specimens originating from a given stratigraphic horizon assigned to a given taxon. The age of each occurrence is based on data from PBDB, corrected with a more precise age (generally stage, sometimes substage), and the age of each time bin boundaries relies on the stratigraphic framework proposed in the International Chronostratigraphic Chart (updated to correspond with the ICS 2022/0295). Similarly, the ages of some species assigned to the wrong stage were corrected. In fact, some species from the French Permian deposit of Lodève were initially considered to be of Artinskian age in PBDB but most species from this deposit originate from the Merifons member, which is of Kungurian age96.Our data compilation allows a robust integration of data before and after our period of interest (i.e. the lower Permian and all geologic stages after the Carnian) to encompass occurrences of genera that may survive until the Late Triassic and to generate a sufficient background for the model to correctly estimate the extinction events around the P/T boundary. Since we used different datasets, the differences between genus-level or family-level occurrence numbers are explained by the systematic placement of some specimens that can only be placed confidently in a family but not in a genus (Supplementary Table 1). Tentative species identifications originally placed with uncertainty (reported as ‘aff.’ or ‘?’) were always included at a higher taxonomic level. Uncertain generic attributions were integrated as occurrences at the family level (e.g. a fossil initially considered Tupus? is recorded as an occurrence of Meganeuridae). Our total dataset was subdivided into smaller datasets, which represent orders or other subclades of insects (e.g. Mecoptera, Holometabola and Polyneoptera). Note that all the ichnospecies—a species name assigned to trace fossils (e.g. resting trace, nest and leaf damage)— and insect eggs (e.g. Clavapartus latus, Furcapartus exilis and Monilipartus tenuis) were not included in the analyses97. To prevent potential issues regarding the diversification estimates for clades with poor delineation, we refrained from analysing several orders that serve as taxonomic ‘wastebaskets’ (e.g. Grylloblattodea). These groups are poorly defined, likely polyphyletic or paraphyletic, and not supported by apomorphic characters—e.g. the monophyly of the ‘Grylloblattodea’ (Grylloblattida Walker, 1914 plus numerous fossil families and genera of uncertain affinities) is not supported by any synapomorphy, nor the relationships within this group. The occurrences assigned to these orders were rather included in analyses conducted at a higher taxonomic level (at the Polyneoptera level in the case of the ‘Grylloblattodea’). The detail of the composition of all the datasets is given in Supplementary Table 14, and each dataset is available in Supplementary Data 1.Studying extinction should, when possible, rely on species-level diversity to better circumscribe extinction events at this taxonomic rank, which is primarily affected by extinction98,99,100. However, in palaeoentomology, species-level occurrence data may contain less information than genus-level data, mainly because species are most of the time only known from one deposit, resulting in reduced life span, and are also sometimes poorly defined. Insects are also less prone to long-lasting genera or species than other lineages, maybe because of the relatively short time between generations (allowing for rapid evolution) or because morphological characters are better preserved or more diagnostic than in other lineages (i.e. wing venation), allowing easier differentiation. Another argument for the use of genus-level datasets is the possibility to add occurrences represented by fossils that cannot be assigned at the species level because of poor preservation or an insufficient number of specimens/available characters. By extension, the genus life span provides clues as to survivor taxa and times of origination during periods of post-extinction or recovery. A genus encompassing extinction events indicates that at least one species of this genus crossed the extinction. To get the best signal and infer a robust pattern of insect dynamics around the P/T events, we have chosen to analyse our dataset at different taxonomic ranks (e.g. genus, family and order levels) to extract as much evidence as possible.To further support our choice to work at these different levels, most recent works aiming to decipher the diversification and extinction in insect lineages have worked using a combination of analyses21,22,26; this also applies to non-insect clades51,101,102. This multi-level approach should maximise our understanding of the Permo–Triassic events.Assessing optimal parameters and preliminary testsPrior to choosing the settings for the final analyses (see detail in Dynamics of origination and extinction), a series of tests were carried out to better evaluate the convergence of our analyses. First, we analysed our genus-level dataset with PyRate36 running for 10 million generations and sampling every 10,000 generations, on ten randomly replicated datasets using the reversible-jump Markov Chain Monte Carlo (RJMCMC) model37 and the parameters of PyRate set by default. As the convergence was too low, new settings were used, notably increasing the number of generations to 50 million generations and monitoring the MCMC mixing and effective sample size (ESS) each 10 million generations. We modified the minimal interval between two shifts (-min_dt option, testing 0.5, 1.5 and 2), and found no major difference in diversification patterns between our tests. We have opted for 50 million generations with a predefined time frame set for bins corresponding to the Permian and Triassic stages, and a minimum interval between two shifts of two Ma. These parameters allow for maintaining a short bin frame and high convergence values while correctly identifying periods of diversification and extinction. For each analysis, ten datasets were generated using the extract.ages function to randomly resample the age of fossil occurrences within their respective temporal ranges (i.e. resampled ages are randomly drawn between the minimum and the maximum ages of the geological stratum). We monitored chain mixing and ESS by examining the log files in Tracer 1.7.1103 after excluding the first 10% of the samples as a burn-in period. The parameters are considered convergent when their ESS are greater than 200.Dynamics of origination and extinctionWe carried out the analyses of the fossil datasets based on the Bayesian framework implemented in the programme PyRate36. We analysed the fossil datasets under two models: the birth–death model with constrained shifts (BDCS38) and the RJMCMC (-A 4 option37). These models allow for a simultaneous estimate for each taxon: (1) the parameters of the preservation process (Supplementary Fig. 17), (2) the times of origination (Ts) and extinction (Te) of each taxon, (3) the origination and extinction rates and their variation through time for each stage and (4) the number and magnitude of shifts in origination and extinction rates.All analyses were set with the best-fit preservation process after comparing (-PPmodeltest option) the homogeneous Poisson process (-mHPP option), the non-homogeneous Poisson process (default option), and the time-variable Poisson process (-qShift option). The preservation process infers the individual origination and extinction times of each taxon based on all fossil occurrences and on an estimated preservation rate, denoted q, expressed as expected occurrences per taxon per Ma. The time-variable Poisson process assumes that preservation rates are constant within a predefined time frame but may vary over time (here, set for bins corresponding to stages). This model is thus appropriate when rates over time are heterogeneous.We ran PyRate for 50 million MCMC generations and a sampling every 50,000 generations for the BDCS and RJMCMC models with time bins corresponding to Permian and Triassic stages (-fixShift option). All analyses were set with a time-variable Poisson process (-qShift option) of preservation and accounted for varying preservation rates across taxa using the Gamma model (-mG option), that is, with gamma-distributed rate heterogeneity with four rate categories36. As explained above, the minimal interval between two shifts (-min_dt option) was modified and a value of 2 was used. The default prior to the vector of preservation rates is a single gamma distribution with shape = 1.5 and rate = 1.5. We reduced the subjectivity of this parameter, and favoured a better adequation to the data, allowing PyRate to estimate the rate parameter of the prior from the data by setting the rate parameter to 0 (-pP option). Therefore, PyRate assigns a vague exponential hyper-prior to the rate and samples the rate along with all other model parameters. Similarly, because our dataset does not encompass the entire fossil record of insects, we assumed that a possible edge effect may interfere with our analyses, with a strong diversification during the lowermost Permian and, conversely a strong extinction during the uppermost Triassic. Because the RJMCMC and BDCS algorithms look for rate shifts, we constrained the algorithm to only search for shifts (-edgeShift option) within the following time range 295.0 to 204.5 Ma. We monitored chain mixing and ESS by examining the log files in Tracer 1.7.1103 after excluding the first 10% of the samples as a burn-in period. The parameters are considered convergent when their ESS are greater than 200.We then combined the posterior estimates of the origination and extinction rates across all replicates to generate rates through-time plots (origination, extinction, and net diversification). Shifts of diversification were considered significant when log Bayes factors were >6 in the RJMCMC model, while we considered shifts to be significant in the BDCS model when mean rates in a time bin did not overlap with the 95% credibility interval (CI) of the rates of adjacent time bins.We replicated all the analyses on ten randomly generated datasets of each clade and calculated estimates of the Ts and the Te as the average of the posterior samples from each replicate. Thus, we obtained ten posterior estimates of the Ts and Te for all taxa and we used these values to estimate the past diversity dynamics by calculating the number of living taxa at each time point. For all the subsequent analyses, we used the estimated Ts and Te of all taxa to test whether or not the origination and the extinction rate dynamics were correlated with particular abiotic factors, as suggested by the drastic changes in environmental conditions known during the Permo–Triassic. We used proxies for abiotic factors, such as global continental fragmentation or the dynamic of major clades of plants, and for biotic factors via species interaction within and between ecological guilds. This approach avoids re-modelling preservation and re-estimating times of origination and extinction, which reduces drastically the computational burden, while still allowing to account for the preservation process and the uncertainties associated with fossil ages. Similarly, the times of origination and extinction used in all the subsequent analyses were obtained while accounting for the heterogeneity of preservation, origination and extinction rates. To discuss the magnitude of the periods of extinction and diversification, we compared the magnitude of these events to the background origination and extinction rates (i.e. not during extinction or diversification peaks).The PyRate approach has proven to be robust following a series of tests and simulations that reflect commonly observed biases when modelling past diversity dynamics31,38. These simulations were based on datasets simulated under a range of potential biases (i.e. violations of the sampling assumptions, variable preservation rates, and incomplete taxon sampling) and reflecting the limitations of the fossil record. Simulation results showed that PyRate is able to correctly estimate the dynamics of origination and extinction rates, including sudden rate changes and mass extinction, even if the preservation levels are low (down to 1–3 fossil occurrences per species on average), the taxon sampling is partial (up to 80% missing) or if the datasets have a high proportion of singletons (exceeding 30% of the taxa in some cases). The strongest bias in birth–death rate estimates is caused by incomplete data (i.e. missing lineages) altering the distribution of taxa; a pervasive effect often mentioned for phylogeny-based models104,105,106. However, in the case of PyRate, the simulations confirm the absence of consistent biases due to an incomplete fossil record36. Finally, the recently implemented RJMCMC model was shown to be very accurate for estimating origination and extinction rates (i.e. more accurate than the BDCS model, the boundary-crossing and three-time methods) and is able to recover sudden extinction events regardless of the biases in the fossil dataset37.The severity of extinctions and survivorsFor each event—the Roadian–Wordian, the LPME, and the Ladinian–Carnian—we quantified the percentage of extinctions and survivors at the genus level. We used the Te and Ts from our RJMCMC analysis and computed the mean for the Te (Tem) and for the Ts (Tsm) of each genus. We then filtered our dataset to keep only the genera with a Tsm older than the upper boundary of the focal event, i.e., we only kept the genera that appeared before the end of the event. Then, we discarded the genera that have disappeared before the lower boundary of the focal event, i.e. Tem older that the lower boundary of the event. The remaining genera, which corresponds to all the genera (total) present during the crisis (Ttgen), can be classified into two categories, ‘survivor genera’ (Sgen), i.e. those that survived the crisis, and those that died: ‘extinct genera’ (Egen). The survivors have a Tem younger than the upper boundary of the focal event, while the ‘extinct genera’ died out during the event and have a Tem between the lower and upper boundaries of the event of interest. To obtain the percentage of survivors, we used the following formula: (Sgen/Ttgen) × 100. Similarly, the percentage of extinction is calculated as: (Egen/Ttgen) × 100.Age-dependent extinction modelWe assessed the effect of taxon age on the extinction probability by fitting the age-dependent extinction (ADE; -ADE 1 option) model50. This model estimates the probability for a lineage to become extinct as a function of its age, also named longevity, which is the elapsed time since its origination. It is recommended to run the ADE model over time windows with roughly constant origination and extinction rates, as convergence is difficult—but not impossible—to reach in extinction or diversification contexts50. We ran PyRate for 50 million MCMC generations with a sampling every 50,000 generations, with a time-variable Poisson process of preservation (-qShift option), while accounting for varying preservation rates across taxa using the Gamma model (-mG option). We replicated the analyses on ten randomised datasets and combined the posterior estimates across all replicates. We estimated the shape (Φ) and scale (Ψ) parameters of the Weibull distribution, and the taxon longevity in a million years. According to ref. 50, there is no evidence of age-dependent extinction rates if Φ = 1. However, the extinction rate is higher for young species and decreases with species age if Φ  1. Although ADE models are prone to high error rates when origination and extinction rates increase or decrease through time, simulations with PyRate have shown that fossil-based inferences are robust50. We investigated the effect of ADE during three different periods (-filter option) as follows: (1) between 264.28 Ma and 255 Ma (pre-decline), (2) between 254.5 Ma and 251.5 Ma (decline) and (3) between 234 Ma and 212 Ma (post-crisis). We monitored chain mixing and ESS by examining the log files in Tracer 1.7.1103 and considered the convergence of parameters sufficient when their ESS were greater than 200.Selection of abiotic and biotic variablesTo test correlations of insect diversification with environmental changes, we examined the link between a series of environmental variables and origination/extinction rates over a period encompassing the GEE, the LPME and the CPE but also for each extinction event. We focused on the role of nine variables, also called proxies, which have been demonstrated or assumed to be linked to extinctions and changes in insect diversity26,67.The variations in the atmospheric CO2 and O2 concentrations are thought to be correlated with the diversification of several insect lineages, including the charismatic giant Meganeuridae65,66,67. Because the increase of O2 concentration has likely driven the diversification of some insects, its diminution may have resulted in the extinction or decline of some lineages. Therefore, we investigated the potential correlation of the variations of this variable with insect dynamics using data from ref. 55. We extracted the data, with 1-million-year time intervals, spanning the Permo–Triassic.Similarly, the modification of CO2 concentration, notably its increase, is known to promote speciation in some modern insect groups107. Therefore, a similar effect may have occurred during the Permian and Triassic but remains to be tested. We based our analyses on the dataset of ref. 108. We used their cleaned dataset and extracted all verified values for the Permo–Triassic interval. Because the initial data (i.e. independent estimates) were made in various locations for the same age, different values of the CO2 concentration are provided. We incorporated all these values in our analysis, allowing PyRate to search for a correlation for each value of the CO2 concentration. We obtain a final correlation independent of the sampling location, in line with our large-scale analysis.The continental fragmentation, as approximated by plate tectonic change over time, has recently been proposed as a driver of Plecoptera dynamics26. Because the period studied encompasses a major geological event, the fragmentation of the supercontinent Pangea, we investigated the effect of continental fragmentation on insect diversification dynamics. We retrieved the index of continental fragmentation developed by ref. 69 using paleogeographic reconstructions for 1-million-year time intervals. This index approaches 1 when all plates are disjoined (complete plate fragmentation) and approaches 0 when the continental aggregation is maximal.Climate change (variations in warming and cooling periods) is a probable driver of diversification changes over the history of insects21,109. Temperature is likely directly linked with insect dynamics109 but also with their food sources, notably plants110. Because it was demonstrated that modification of temperature impacted floral assemblages110, we tested the correlation between temperature variations and the diversification dynamic of insects. Major trends in global climate change through time are typically estimated from relative proportions of different oxygen isotopes (δ18O) in samples of benthic foraminiferan shells111. We used the data from ref. 112, converted to absolute temperatures following the methodology described in Condamine et al.113 (see their section Global temperature variations through time). The resulting temperature data reflects planetary-scale climatic trends, with time intervals inferior to 1-million-year, which can be expected to have led to temporally coordinated diversification changes in several clades rather than local or seasonal fluctuations.The fluctuation in relative diversity of gymnosperms, non-Polypodiales ferns, Polypodiales ferns, spore-plants, and later the rise of angiosperms has likely driven the diversification of numerous insects57,60,61,114. Close interactions between insects and plants are well-recorded during the Permian and Triassic57,60,61. In fact, herbivorous insects are known to experience high selection pressure from bottom-up forces, resulting from interactions with their hosts or feeding plants30,72. Therefore, it appears crucial to investigate the effect of these modifications on the insects’ past dynamics. We used the data from ref. 38 for the different plant lineages (all with 1-million-year time intervals). All the datasets for these variables are available in the publications cited aside from each variable or in Supplementary Data 1.Multivariate birth–death modelWe used the multivariate birth–death (MBD) model to assess to what extent biotic and abiotic factors can explain temporal variation in origination and extinction rates55. The model is described in ref. 55, where origination and extinction rates can change through time in relation to environmental variables so that origination and extinction rates depend on the temporal variations of each factor. The strength and sign (positive or negative) of the correlations are jointly estimated for each variable. The sign of the correlation parameters indicates the sign of the resulting correlation. When their value is estimated around zero, no correlation is estimated. An MCMC algorithm combined with a horseshoe prior, controlling for over-parameterisation and for the potential effects of multiple testing, jointly estimates the baseline origination (λ0) and extinction (µ0) rates and all correlation parameters (Gλ and Gµ)55. The horseshoe prior is used to discriminate which correlation parameters should be treated as noise (shrunk around 0) and which represent a true signal (i.e. significantly different from 0). In the MBD model, a correlation parameter is estimated to quantify independently the role of each variable on the origination and the extinction.We ran the MBD model using 20 (for short intervals) or 50 million MCMC generations and sampling every 20,000 or 50,000 to approximate the posterior distribution of all parameters (λ0, µ0, nine Gλ, nine Gµ and the shrinkage weights of each correlation parameter, ωG). The MBD analyses used the Ts and the Te derived from our previous analyses under the RJMCMC model. The results of the MBD analyses were summarised by calculating the posterior mean and 95% CI of all correlation parameters and the mean of the respective shrinkage weights (across ten replicates), as well as the mean and 95% CI of the baseline origination and extinction rates. We carried out six analyses, over: (1) the Permo–Triassic (between 298.9 and 201.3 Ma); (2) the Roadian–Wordian (R/W) boundary (between 270 and 265 Ma), (3) the LPME (between 254.5 and 250 Ma), (4) the Ladinian–Carnian (L/C) boundary (between 240 to 234 Ma), (5) the Permian period (between 298.9 and 251.902 Ma) and (6) the Triassic period (between 251.902 and 201.3 Ma). We monitored chain mixing and ESS by examining the log files in Tracer 1.7.1103 and considered the convergence of parameters sufficient when their ESS were greater than 200.Multiple clade diversity-dependence modelTo assess the potential effect of diversity-dependence on the diversity dynamics of three or four insect guilds, we used the multiple clade diversity-dependence (MCDD) model in which origination and extinction rates are correlated with the diversity trajectory of other clades31. This model postulates that competitive interactions linked with an increase in diversity results in decreasing origination rates and/or increasing extinction rates. The MCDD model allows for testing diversity-dependence between genera of a given clade or between genera of distinct clades sharing a similar ecology.We estimated the past diversity dynamics for three (i.e. herbivores, predators, and a guild composed of generalists + detritivores/fungivores dubbed ‘others’) or four insect groups or guilds (i.e. herbivores, predators, generalists and detritivores/fungivores) by calculating the number of living species at every point in time based on the times of origination (Ts) and extinction (Te) estimated under the RJMCMC model (see above) (Supplementary Figs. 19–24). We defined our four insect groups with a cautious approach i.e. insect genera, families or orders for which nothing is known about the ecology or about the ecology of their close relatives were not considered for the analysis. For example, no diet was assigned to Diptera, Mecoptera or Glosselytrodea. The ecology of the Triassic Diptera and Permo–Triassic Mecoptera is difficult to establish because extant Diptera and Mecoptera have a wide diversity of ecology. Fossil Mecoptera are also putatively involved in numerous interactions with plants (species with elongated mouthparts), suggesting a placement in the herbivore group, while other species were likely predators. Therefore, we cannot decide to which group each species belongs. Similarly, nothing is known about the body and mouthparts of the Glosselytrodea, most of the time described based on isolated wings; we did not assign the order to any group. The definition and delineation of insect clades have also challenged the placement of several orders (e.g. ‘Grylloblattodea’) in one of our four groups. The order ‘Grylloblattodea’ is poorly delineated and mostly serves as a taxonomic ‘wastebasket’ to which it is impossible to assign a particular ecology. Finally, genera, species, or families not placed in a higher clade (e.g. Meshemipteron, Perielytridae) were not included in the analysis. Oppositely, the guilds ‘herbivores’ and ‘predators’ are well defined, and their ecology is evidenced by the morphology of their representatives and the principle of actualism. For example, the ecology of Meganeurites gracilipes (Meganeuridae) has been deeply studied, and its enlarged compound eyes, its sturdy mandibles with acute teeth, its tarsi and tibiae bearing strong spines, and the presence of a pronounced thoracic skewness are specialisations today found in dragonflies that capture their prey while in flight115. All Odonatoptera are well-known predator insects. The raptorial forelegs of the representatives of the order Titanoptera and their mouthparts with strong mandibles are linked with predatory habits81. The Palaeodictyopteroidea were herbivorous insects with long, beak-like, piercing mouthparts, and probably a sucking organ81,82. Most Hemiptera are confidently considered herbivorous insects by comparison with their extant representatives. For example, the Cicadomorpha or Sternorrhyncha are known to feed on plants and their fossil representatives likely possessed the same ecology because of similar morphologies116. Some hemipteran families (e.g. Nabidae) are predators and we cautiously distinguished herbivorous and carnivorous taxa among Hemiptera. The detail of the ecological assignations for the 1009 genera included in our analyses can be found in Supplementary Data 1 (Table MCCD).We calculated ten diversity trajectories from the ten replicated analyses under the RJMCMC model. The estimation of past species diversity might be biased by low preservation rates or taxonomic uncertainties. However, such trajectory curves are likely to provide a reasonably accurate representation of the past diversity changes in the studied clades, notably because the preservation during the Permian and Triassic period is relatively good for insects (i.e. no gaps).Our MCDD analyses comprise all the insect genera spanning from the lowermost Permian to the uppermost Triassic and were run and repeated on ten replicates (using the Te and Ts estimated under the RJMCMC model) with 50 million MCMC generations and a sampling frequency of 50,000. For each of the four insect groups, we computed the median and the 95% CI of the baseline origination and extinction rates (λi and µi), the within-group diversity-dependence parameters gλi and gµi, and the between-groups diversity dependence parameters gλij and gµij. The mean of the sampled diversity dependence parameters (e.g. gλij) was used as a measure of the intensity of the negative (if positive) or positive interactions (if negative) between each pair of groups. The interactions were considered significant when their median was different from 0 and the 95% CI did not overlap with 0. We monitored chain mixing and ESS by examining the log files in Tracer 1.7.1103 and considered the convergence of parameters sufficient when their ESS were greater than 200.We cross-validated the result of the MCDD model using the MBD model. The MBD model can be used to run a multiple clade diversity-dependence analysis by providing the diversity trajectories of insect guilds as a continuous variable. These data are directly generated by PyRate using the lineages-through-time generated by the RJMCMC analyses (-ltt option). We ran the MBD model using 50 million MCMC generations and sampling every 50,000 to approximate the posterior distribution of all parameters (λ0, µ0, four Gλ, four Gµ and the shrinkage weights of each correlation parameter, ωG). We carried out three analyses, over the period encompassing the three extinction events (between 275 and 230 Ma): (1) for herbivores; (2) for predators; and (3) for ‘others’. For each analysis, the lineages-through-time data of the two other guilds are used as continuous variables to investigate a diversity dependence effect. We monitored chain mixing and ESS by examining the log files in Tracer 1.7.1103 and considered the convergence of parameters sufficient when their ESS were greater than 200.Reporting summaryFurther information on research design is available in the Nature Portfolio Reporting Summary linked to this article. More

SamplesWe examined 1,022 avian species (~ 10% recorded species) in this study, representing 418 genera, from 91 families (37% recorded families) and 29 orders (73% of all orders). Specimens were from the skin collection of the World Museum Liverpool, Tring Natural History Museum, Manchester Museum and Wollaton Hall Museum, all situated in the United Kingdom. All work was carried out in accordance with ethical regulations at Manchester Metropolitan University and with the permission of all aforementioned museums. Only the best-preserved adult specimens (no signs of cut off feathers or holes in the skin near the beak) were chosen for this study to ensure accurate measurements of bristle length, shape and presence, which should not be affected by the process of skin removal and specimen conservation. Species were randomly chosen, without targeting our sampling towards species known a priori to have bristles. Where possible, two specimens per species were measured (occurring in 82% of all species examined). Specimens of each sex were measured when present; however, this was not always possible since labelling was often inaccurate or missing. In total, the sample included 508 males, 412 females and 374 individuals of unknown sex. Both sexes were examined in 274 species and there was no difference whatsoever between the presence of bristles on male or female species (n = 97 with bristles present and n = 180 with bristles absent for both males and females). Length (Mann–Whitney U test, W = 37,962, N = 552, P = 0.94) and shape (Chi-square test, χ2 = 0, N = 552, df = 3, P = 1) of rictal bristles also did not significantly differ between males and females. Therefore, rictal bristles are likely to be sexually monomorphic and data for males and females was pooled for further analyses. Overall, rictal bristles were absent in 64% of species examined (n = 656) and just over a third of species (n = 366) had bristles present.Bristle descriptionsFacial bristles were initially identified by sight and touch in each specimen. Bristles were recorded as either present or absent from the upper rictal, lorial, lower rictal, narial and interramal regions (Fig. 1a). We use the term ‘rictal bristle’ here for bristles on both the upper rictal and/or the lorial region, since there was no clear differentiation and morphological differences between the bristles found in these regions forming a continuum of bristles above the edge of the beak. When present, rictal bristle shape was recorded as: (i) unbranched rictal bristles, (ii) rictal bristles with barbs only at the base (“Base”) and (iii) branched rictal bristles (“Branched”), i.e. barbs and barbules present along the bristle rachis (Fig. 1b). The three longest rictal bristles were measured on both sides of the head of each specimen using digital callipers, and these lengths were averaged to provide a mean length of rictal bristles per species. In species lacking rictal bristles, a length of “0” and a shape category of “Absent” was recorded.Ancestral reconstruction of facial bristle presenceFollowing Felice et al.19, a single consensus phylogenetic tree was generated from the Hackett posterior distribution of trees from Birdtree.org20 with a sample size of 10,000 post burn-in, using the TreeAnnotator utility in BEAST software21 with a burn-in of 0. Maximum Clade Credibility (MCC) with the option “-heights ca” was selected as the method of reconstruction. The common ancestor trees option (-heights ca) builds a consensus tree by summarising clade ages across all posterior trees. Both the consensus tree and posterior distribution of 10,000 trees were imported into RStudio v. 1.2.5 for R22,23 and pruned so that only species present in the dataset of this study remained in the phylogeny. Taxon names were modified where necessary to match those from the Birdtree.org (http://birdtree.org) species record. Negative terminal branches in our consensus tree were slightly lengthened to be positive using ‘edge.length[tree$edge.length  More