Recalibration of the insect evolutionary time scale using Monte San Giorgio fossils suggests survival of key lineages through the End-Permian Extinction

Abstract

Insects are a highly diverse group of organisms and constitute more than half of all known animal species. They have evolved an extraordinary range of traits, from flight and complete metamorphosis to complex polyphenisms and advanced eusociality. Although the rich insect fossil record has helped to chart the appearance of many phenotypic innovations, data are scarce for a number of key periods. One such period is that following the End-Permian Extinction, recognized as the most catastrophic of all extinction events. We recently discovered several 240-million-year-old insect fossils in the Mount San Giorgio Lagerstätte (Switzerland–Italy) that are remarkable for their state of preservation (including internal organs and soft tissues), and because they extend the records of their respective taxa by up to 200 million years. By using these fossils as calibrations in a phylogenomic dating analysis, we present a revised time scale for insect evolution. Our date estimates for several major lineages, including the hyperdiverse crown groups of Lepidoptera, Hemiptera: Heteroptera and Diptera, are substantially older than their currently accepted post-Permian origins. We found that major evolutionary innovations, including flight and metamorphosis, appeared considerably earlier than previously thought. These results have numerous implications for understanding the evolution of insects and their resilience in the face of extreme events such as the End-Permian Extinction.

1. Introduction

Insects have diversified over the past approximately 400 million years (Myr) into an estimated 5.5 million species (ranging from 2.6 to 7.8 million species) [1] that occupy nearly all continental habitats. Insects were among the first animals to colonize land following the establishment of plants, evolving an array of traits including flight, complete metamorphosis and advanced sociality [2]. Among the hypotheses put forward to explain the extreme diversity of insects, two that have received support are the great antiquity of the group and a relatively low extinction rate [3,4].

Compared with other animal groups, insects appear to have been relatively unaffected by major extinction events [2], with possible exceptions being the catastrophic End-Permian Extinction (EPE; 251.9 Ma) [5–7] and the Cretaceous–Palaeogene mass extinction (66 Ma) [8]. In the case of the EPE, two distinct extinction events during an approximately 60 000-year period caused the disappearance of 80–96% of marine species and 70% of terrestrial vertebrate species [9–11]. Extant insect orders are generally represented in fossils post-Permian, whereas numerous extinct groups are only found in deposits from the Permian or earlier [5]. For example, a number of major insect groups (e.g. Palaeodictyoptera, Megasecoptera and Archaeorthoptera) appear to have gone extinct at the end of the Permian, or soon afterwards. A similar pattern emerged from a recent phylogenomic study on insects [12], which estimated that a number of major groups, including Diptera and Lepidoptera, emerged after the EPE.

Reconstructions of insect diversity through evolutionary time are dependent on the presence of representative entomofauna across all stratigraphic stages since the origin of the group, estimated to have occurred during the late Silurian [2,13]. One period that is depauperate of insect fossils is that immediately following the EPE [2,14]. The relatively small number of fossils from this period presents a hindrance to our understanding of how insects responded to this mass extinction event.

Here, we investigate the impact of eight exceptionally well-preserved fossil insects from the Monte San Giorgio (MSG) Lagerstätte (Italian–Swiss border; 240–239 Ma; electronic supplementary material, figure S1) on estimation of the insect evolutionary time scale. Although the MSG has been recognized for its important vertebrate assemblage [15], its entomofauna has received relatively little attention. Nonetheless, its importance has recently been highlighted by the discovery of the oldest representatives of some extant groups, including a jumping bristletail (Archaeognatha: Machilidae) [16,17] and a lace bug (Hemiptera: Tingidae) [18].

We performed a phylogenomic dating analysis of a sequence alignment of 220 615 amino acids generated by Misof et al. [12]. This dataset has comprehensive coverage of insect ordinal diversity. Although there have been further intra-ordinal phylogenomic studies of insects over the past five years (e.g. [19]), we analysed the dataset of Misof et al. [12] because their estimate of the evolutionary time scale is the most widely recognized.

We used eight key MSG fossils (figure 1; electronic supplementary material, table S1 and text S1) plus the fossils used by Misof et al. [12] to calibrate the molecular clock for our dating analysis (electronic supplementary material, table S2). Each of these fossils was chosen according to strict criteria [20]. Briefly, the eight MSG fossils that we used for additional calibrations were (figure 1): (i) the bristletail Gigamachilis triassicus [16], the oldest fossil of crown Machilidae; (ii) the mayfly Tintorina meridensis [21]; (iii) the oldest lace bug fossil, Archetingis ladinica [18]; (iv) the adephagan beetle Praedodromeus sangiorgiensis [22]; (v) the stonefly MCSN8462 [22]; (vi) the webspinner MCSN8457 [22], approximately 130 Myr older than the second-oldest confirmed webspinner, Sorellembia estherae [23]; (vii) the polyphagan (not Staphyliniformia) beetle MCSN8464 [22]; and (viii) a reticulated beetle elytron of Notocupes sp. [21] (detailed information for each fossil is provided in electronic supplementary material, text S1). In addition to these fossils, due to their exceptional features, we report here an unusually large holometabolous larva MCSN8531 (figure 1h) and a specimen with a grasshopper-like body plan MCSN8457 (figure 1i,j); these fossils were not used for calibrations in our dating analysis because of their uncertain phylogenetic positions.

Figure 1.

Macro photographs of the MSG fossil insects. (a) Gigamachilis triassicus (Archaeognatha: Machilidae). (b) Tintorina meridensis (Ephemeroptera: Tintorinidae). (c) Archetingis ladinica (Hemiptera: Tingidae). (d) Praedodromeus sangiorgiensis (Coleoptera: Trachipachidae). (e) Completely phosphatized stonefly nymph (Plecoptera), with a total length of 9.38 mm. (f) Phosphatized webspinner (Embioptera) of 18.3 mm in length, preserved in ventral view; fore basitarsi swollen, three segmented tarsi, hindlegs with enlarged femora; muscles and tendons are visible within the legs. (g) Coleoptera Polyphaga (not Staphyliniformia) of 2.9 mm in length preserved in dorsal/lateral view. (h) Phosphatized holometabolous larva (MCSN8531) of approximately 4 cm in length (not complete), not assigned to an order. (i) Phosphatized specimens with grasshopper-like body plan; within the abdomen, at its end, a group of tubules is visible, possibly identifiable as the Malpighian tubules; in (j) a magnification of these structures, scale bar, 200 µm. Details of image acquisition are reported in electronic supplementary material, text S1. (Online version in colour.)

The MSG insect fossils are remarkable because of their preservation of soft tissues and internal organs, which are typically absent in invertebrate compression fossils. For example, in the stonefly nymph MCSN8462 (figure 1e), there is clear preservation of part of the cerebrum, the suboesophageal ganglion, the meso- and metathoracic wing pads, and part of the alimentary canal (possibly the midgut). In the winged male webspinner MCSN8457 (figure 1f), phosphatized enlarged basal fore-tarsomers (diagnostic of Embioptera) and leg muscle bundles are preserved. These features, in association with the impression of the folded wings, suggest that the insect died suddenly in its burrow with a rapid establishment of anoxic conditions, allowing the phosphatization of organic matter and fossilization. Within the abdomen of the grasshopper-like fossil MCSN8457 (currently unassigned; figure 1i,j), Malpighian tubules are evident, representing the first case, to our knowledge, of preservation of this organ in compression fossils.

2. Material and methods

(a). Selection of fossils used for calibrations

To calibrate our molecular date estimates, we included 35 fossils previously used by Misof et al. [12] and selected according to the criteria proposed by Parham et al. [20] (electronic supplementary material, table S2). In addition, we used eight MSG fossils for calibration (figure 1; electronic supplementary material, table S1 and text S1). Descriptions of diagnostic characters and taxonomic assignments of the MSG fossils were provided in previous publications and a PhD thesis [16–18,21,22,24]; in addition, a brief description is reported in electronic supplementary material, text S1. With the exception of Notocupes sp. [21] and the stonefly nymph [22], the MSG fossils satisfy the criteria proposed by Parham et al. [20] for reliable calibrations. Museum identifiers, taxonomic assignments and references of the eight MSG fossils are provided in electronic supplementary material, table S1. The specimens included as calibrations are or will be deposited at Museo Cantonale di Storia Naturale in Lugano (CH).

(b). Genomic dataset

In order to evaluate the impacts of the MSG fossil entomofauna on phylogenomic estimation of the insect evolutionary time scale, we chose to focus on the genomic dataset recently published by Misof et al. [12]. The transcriptomic data of 141 arthropod species consisting of 1478 single-copy nuclear genes, from which we obtained the dataset analysed in this study (220 615 aligned amino acids), are available from the Dryad Digital Repository (https://doi.org/10.5061/dryad.3c0f1 [25]).

The genomic dataset includes representatives of every major insect order, as well as springtails and proturans. Ten taxa were used as outgroups: seven crustaceans, two myriapods and the tick Ixodes. Here, we briefly repeat the methods used by Misof et al. [12] to prepare the data for analysis. A detailed description is provided in electronic supplementary material, text S1.

(c). Estimation of divergence dates

We described our calibration priors as uniform priors with soft minimum bounds, reflecting the uncertainty in fossil evidence in molecular dating [26]. The minimum age of each MSG fossil corresponds to the age of the fossil deposit (i.e. 239.51 ± 0.15 Ma [27]). The maximum age of each calibration was set to either 580 Ma, which is the approximate estimated age of the origin of arthropods based on the oldest Ediacaran fossils; or to 450 Ma, when there was sufficient terrestrial food to support Hexapoda [28]. The former maximum age limit was applied to each of the oldest calibrated nodes, representing the split between two outgroup lineages and the split between the outgroup and ingroup lineages. The latter, younger age limit was used for every other calibration within the insect tree. Information on the nodes at which minimum age constraints were informed by the MSG fossils are reported in electronic supplementary material, text S1, table S2 and figure S3.

We ran five sets of phylogenomic dating analyses to compare the effects of the MSG fossils on estimation of the insect evolutionary time scale (details provided in electronic supplementary material, text S1). The tree topology was fixed for all of these analyses, having been estimated using maximum likelihood by Misof et al. [12]. To infer the evolutionary divergence times in the tree, we used the Bayesian dating approach in MCMCTREE [29], which is able to use an approximate likelihood calculation to reduce computational burden [30,31]. For all of our phylogenomic dating analyses, we used a Dirichlet-gamma prior for the mean substitution rate, with α = 2 and β = 20, which represents a diffuse prior distribution. We used a uniform prior for the relative node times. The posterior distribution of node times was estimated using Markov chain Monte Carlo (MCMC) sampling. The first 100 000 MCMC steps were discarded as burn-in before we drew samples every 50 steps over 2 million steps. For each of the five analyses, we combined all samples from the respective 85 meta-partitions before calculating the combined 95% credibility intervals for each node in the tree to produce a combined estimate of the evolutionary time scale.

3. Results and discussion

(a). Impact of Monte San Giorgio fossils and uniform calibration priors on estimated divergence times

We performed five separate phylogenomic dating analyses in MCMCTREE to examine the effects of adding eight MSG fossils to a set of 37 fossils previously used in a phylogenomic analysis of insect evolution [12] (electronic supplementary material, table S3). These analyses were performed using the same dataset and fixed tree topology (figure 1 of Misof et al. [12]), but with five different sets of fossil calibration priors. A comparison of date estimates from three of these five analyses at 13 key nodes is shown in figure 2, with a full comparison for all 140 node times provided in electronic supplementary material, tables S3 and S4.

Figure 2.

Comparison of age estimates for 13 key nodes obtained from three analyses employing different fossil calibration priors. For each lineage, vertical bars show the 95% credibility intervals of the age estimates; circles denote median values, and the red triangle corresponds to the age of MSG fossils. The purple dashed line corresponds to the Permo-Triassic boundary. The main comparisons are between the results of Analysis 1 (grey bars, replicating the calibration scheme used by Misof et al. [12]), Analysis 3 (yellow bars, replicating the calibration scheme used to produce fig. 1B in Tong et al. [32]) and Analysis 4 (red bars, based on the addition of the MSG fossil calibrations). We also include estimates for some nodes obtained by Wheat & Wahlberg [33] (purple bars), Ronquist et al. [34] (green bars) and Wahlberg et al. [35] (blue bars); for some of these, squares denote estimated mean values. The periods and epochs from the International Chronostratigraphic Chart v 2019/05 [36] are shown on the left. Hexa, Hexapoda; Ecto, Ectognatha; Pter, Pterygota; Holo, Holometabola; Arch, Archaeognatha; Poly, Polyneoptera incl. Zoraptera; Zora+Derm, Zoraptera + Dermaptera; Phth, Phthiraptera; Orth, Orthoptera; Hemi: Hete, Hemiptera: Heteroptera; Hymenopt, Hymenoptera; Lepidopt, Lepidoptera; Dipt, Diptera; Mississip., Mississippian; N, Neogene; Pennsylv., Pennsylvanian; Q , Quaternary; Silur, Silurian; Terreneuv., Terreneuvian. (Online version in colour.)

We first replicated the analyses of Misof et al. [12], who used a total of 37 fossil calibrations. Lognormal priors were used for 20 of these calibrations, and uniform priors for the remaining 17 calibrations (Analysis 1). The results of these analyses were similar to those previously reported; in terms of median posterior estimates, there was a mean difference of 8.16 Myr from the values across the 140 nodes reported in electronic supplementary material, table S25 of Misof et al. [12] (see electronic supplementary material, tables S3 and S4 and grey bars in figure 2). Next, we added the eight MSG fossils with uniform calibration priors, retaining the original lognormal and uniform priors for the fossil calibrations used by Misof et al. [12] (Analysis 2; 43 fossils used in total; note that two of the MSG fossil calibrations replaced those used by Misof et al. [12]). This set of calibrations led to notable increases in the median posterior ages of some nodes (electronic supplementary material, table S4), with an average increase of 13.4 Myr across the 140 internal nodes in the tree compared with Analysis 1.

The use of restrictive lognormal priors can lead to underestimation of node ages [26,32], so we performed two further analyses with uniform calibration priors only. The first analysis employed only the fossils used by Misof et al. [12] (Analysis 3; 37 fossil calibrations), and is equivalent to the analysis reported in fig. 1B of Tong et al. [32] (see yellow bars in figure 2). The second analysis involved the addition of the MSG fossils (Analysis 4; 43 fossil calibrations; see red bars in figure 2). Median posterior node ages across the tree increased by an average of 42.9 Myr (Analysis 3) and 50.5 Myr (Analysis 4) compared with Analysis 1 (in which 20 lognormal calibration priors were used; electronic supplementary material, table S3; figure 2). We discuss the results from Analysis 4 in detail in the following section. Finally, we performed an analysis using both sets of fossil calibrations (43 taxa) and an autocorrelated relaxed-clock model (Analysis 5). We did this analysis to allow comparison with the results of Analysis 4, which employed an uncorrelated relaxed-clock model. Analyses 4 and 5 produced similar estimates of node ages (average difference of 4.3 Myr in median posterior node ages; electronic supplementary material, tables S3 and S4). Each of the eight MSG fossils used for calibration was found to have a strong influence on the date estimate for its respective node, with the dates being pushed back in time past the EPE in Analyses 2, 4 and 5 compared with Analyses 1 and 3 (electronic supplementary material, figure S2).

(b). The evolution of early-branching insect lineages and key insect innovations

The time scale of evolution estimated in our analysis that used uniform priors for 43 fossil calibrations (Analysis 4) is shown in figure 3, electronic supplementary material, figure S3. The median posterior ages of a total of 63 out of 140 internal nodes, 32 of which represent the last common ancestors of major hexapod crown groups, increased by more than 50 Myr (maximum 147 Myr) compared with those estimated in Analysis 1 (which were similar to those reported in Misof et al. [12]). The median posterior ages of 96 internal nodes (including 49 major lineages) increased by more than 25 Myr compared with the node-age estimates from Analysis 1 (electronic supplementary material, table S4).

Figure 3.

Time-calibrated phylogenetic tree of insect evolutionary relationships. Tips show the names of taxa at ordinal and sub-ordinal levels. Bars on nodes correspond to the 95% credibility intervals of node-age estimates, with red indicating the nodes for which the median age is pushed back by more than 50 Myr. Green stars indicate the nodes calibrated with MSG fossils; yellow circles indicate nodes calibrated with other fossils. Two key evolutionary innovations in insects (wings and complete metamorphosis) are shown at the relevant nodes. Red triangles indicate Hexapoda and Insecta. Below the tree the following are shown: (i) bars with periods and epochs from the International Chronostratigraphic Chart v. 2019/05 [36]; (ii) scale axis of the tree expressed in millions of years; (iii) origin of plant groups; (iv) bar with warmer (red) and cooler (blue/dark blue) periods [37]; and (v) plot of O₂ concentration versus time according to [38]. Abbreviations as in figure 2. (Online version in colour.)

The increases in median ages between Analyses 1 and 4 were generally matched by similar increases in the lower and upper bounds of the 95% CIs of each analysis (average increases of 34.6 and 62.6 Myr for the lower and upper bounds, respectively; electronic supplementary material, table S4). In the sections below, we focus on differences in median posterior node ages between the two analyses and those of other relevant studies. However, the 95% CIs for node ages across the tree were generally very wide (means of 101.6 and 129.6 Myr across all nodes for Analyses 1 and 4, respectively; electronic supplementary material, table S3), such that there was typically some overlap between the date estimates from the two analyses.

Our analysis supports an origin of crown-group hexapods in the Cambrian to Ordovician (approx. 504 Ma, 95% CI 469–540 Ma). This is earlier than previous estimates in the Silurian [33] but agrees with the results of other studies [12,32,39,40] (figure 2). We estimated that the last common ancestor of extant insects (Ectognatha) appeared in the Ordovician 465 Ma (95% CI 439–493 Ma), contemporary with the origin of land plants [41–44] (figure 3; electronic supplementary material, table S3). This result extends the median estimate of the age of crown Ectognatha by approximately 20 Myr compared with some previous estimates [12,33] (figure 2; electronic supplementary material, table S3), but is consistent with the results of other studies [39,45].

Crown Collembola was estimated to have originated in the Carboniferous to Triassic (285 Ma, 95% CI 209–365 Ma), with the median posterior age being approximately 30 Myr earlier than a previous estimate [12] (electronic supplementary material, table S3). This raises the possibility of an origin of the group before the EPE. The last common ancestor of extant lineages of monocondylous insects (Archaeognatha) was estimated to have occurred during the Devonian to Triassic (290 Ma, 95% CI 227–388 Ma; figures 2 and 3), with a median posterior age approximately 140 Myr earlier than previously reported [12] (electronic supplementary material, table S3; figures 2 and 3). We estimated an origin of crown Palaeoptera, including damselflies, dragonflies and mayflies, in the Silurian to Carboniferous approximately 396 Ma (95% CI 327–439 Ma), with a median posterior age in the Early Devonian rather than in the Late Devonian to Carboniferous [12,33] (electronic supplementary material, table S3). This median posterior estimate is in agreement with those from other studies [32,39].

Our analyses show that several innovations critical to the success of insects, including the appearance of wings and complete metamorphosis, evolved earlier than previously proposed. We estimated that crown-winged lineages (Pterygota) emerged during the Silurian (434 Ma, 95% CI 413–449 Ma). This median posterior estimate is at least approximately 30 Myr earlier than some estimates (e.g. [12,46]; electronic supplementary material, table S3) but in agreement with others (e.g. [39,45,47]). Our results are consistent with the thermoregulatory hypothesis of insect winglet evolution and compatible with the evolution of insect flight during a period of high atmospheric pO₂ [48], which is estimated to have occurred in the Early to Middle Silurian [38]. The appearance in the ectothermic protopterygote of small winglets is believed to have led to an increase in body temperature [49], which might have conferred an adaptive advantage under the cool climate conditions prevalent during the Late Ordovician to Early Silurian (figure 3). On the basis of our estimates, the ability to fold wings (crown-group Neoptera) evolved within approximately 10 Myr after their appearance (421 Ma, 95% CI 399–441 Ma).

We estimated the origin of insects exhibiting complete metamorphosis (crown Holometabola) to have occurred during the Devonian (approx. 389 Ma, 95% CI 359–419 Ma), with a median posterior age more than 40 Myr earlier than proposed in the majority of previous estimates (e.g. [12,33,50]; electronic supplementary material, table S3), but in agreement with a recent time scale inferred using transcriptome data [39]. The broader spectrum of available ecological niches provided by communities of vascular and macrophyllous plants from the Late Silurian to Early Devonian [43], as well as a global transition from hyperoxic to hypoxic conditions [38], might have been associated with the evolution of complete metamorphosis and the development of semaphoronts able to use different habitats and food resources.

The ancestor of crown polyneopterans, a group that includes, among others, ground lice, stoneflies, crickets, leaf insects, cockroaches and termites, was estimated here to have appeared during the Devonian (389 Ma, 95% CI 347–428 Ma), with a median posterior age approximately 90 Myr earlier than previous estimates [12,33] (electronic supplementary material, table S3). Similarly, we found that the ancestor of crown Zoraptera + Dermaptera, representing the sister lineage to the rest of the polyneopterans, occurred 345 Ma (95% CI 245–414 Ma), approximately 170 Myr earlier than previous estimates ([12]; electronic supplementary material, table S3, figures 2 and 3). The last two node ages are in agreement with the results of Tong et al. [32], who included ‘roachoid’ fossils from the Late Carboniferous in one of their analyses. During the Ordovician to Late Devonian, the establishment of complex ecosystems dominated by woodland-like vegetation, fungi and arthropods increased the availability of food resources and opened new ecological niches [6]. This might have promoted the appearance and diversification of crown polyneopterans, in agreement with the hypothesis of an early origin of these lineages [51].

Parasitism is a key trait among insects. We estimated that the last common ancestor of mosquitoes (based on representatives of Anopheles and Aedes) occurred 131 Ma (95% CI 62–229 Ma). This raises the possibility of coevolution of these insects with live-bearing mammals, whose last common ancestor is thought to have occurred in the Jurassic [52], or with a non-mammalian host. Our results indicate that crown parasitic lice (order Phthiraptera) appeared in the Mesozoic (approx. 129 Ma, 95% CI 54.5–213 Ma). This is inconsistent with the hypothesis that they diversified after the emergence of avian and mammalian hosts [12,53], but supports the alternative hypothesis that they evolved on feathered theropod dinosaurs [54].

(c). Multiple major insect groups may have survived the End-Permian Extinction

Our finding that a large number of crown lineages potentially arose before the EPE (figures 2 and 3) suggests that the most catastrophic extinction event in Earth's history might not have reduced insect diversity to the same degree that it did in other animal groups. A similar finding was recently reported for plants [55], another major group of organisms inhabiting continental ecosystems, suggesting that these ecosystems were not as dramatically affected by the EPE as previously thought. The extensive species diversity of phytophagous insect groups, including Orthoptera, Lepidoptera, Coleoptera, Heteroptera, Hymenoptera and Diptera, was previously thought to have been associated with the diversification of angiosperms [56–58]. However, our results and those of others [19,59–62] raise the possibility that these groups appeared prior to the EPE, and that their initial diversification occurred in association with gymnosperms [63] rather than with flowering plants, which are thought to have diversified in the Mesozoic [64,65].

Our median posterior estimate for the age of crown Hemiptera was in the Devonian to Carboniferous (358 Ma, 95% CI 317–401), rather than in the middle Carboniferous to Permian as previously estimated [12,32]. This is in accordance with a recent analysis of a comprehensive dataset comprising 2395 protein-coding genes from 193 hemipteroid taxa [62]. Our date estimate for crown Heteroptera (277 Ma, 95% CI 241–318 Ma) is consistent with the attribution of Paraknightia magnifica Evans 1943, a highly debated fossil from the Late Permian, to this group [66]. Although predation has been postulated as a plesiomorphic state of heteropterans [67,68], anatomical features suggest that Archetingis ladinica was phytophagous, possibly feeding on gymnosperms or horsetails. Therefore, it is possible that the ancestor of true bugs was phytophagous rather than predatory [67,68]. Similar results were also obtained for crown Coleoptera, which we dated at 283 Ma (95% CI 257–305 Ma) in the Permian, with the last common ancestor of Coleoptera and Strepsiptera occurring in the late Carboniferous to Early Permian (301 Ma, 95% CI 279–313 Ma). These findings are consistent with other studies that placed the base of crown Coleoptera in the Permian [2,12,32,50,69–71].

Our analysis placed the crown age of true flies (Diptera) in the late Carboniferous to early Triassic (approx. 282 Ma, 95% CI 233–333 Ma), in agreement with the results of certain previous studies [32,72] but not of others [12,50]. On the basis of our estimates, the last common ancestor of extant Lepidoptera occurred approximately 271 Ma (95% CI 208–331 Ma). This is slightly earlier than the first glossatan fossilized wing scales, recently discovered in an Upper Triassic deposit [73], but well before the first fossils reliably identified as members of the Lepidoptera dated to the Early Jurassic [74–77]. Therefore, suctorial mouthparts for feeding on gymnosperm pollination drops might have evolved earlier than generally thought. Molecular estimates of the age of crown Lepidoptera range from the Early Cretaceous [12,50] to the Late Triassic [35] and the Middle Permian (this study; [32]), suggesting the potential benefits of further analyses of a more comprehensive dataset, both in terms of taxa and molecular data. The crown group of butterflies (here included as members of the genera Polyommatus and Parides) was estimated to have appeared in the Cretaceous to Palaeogene (82.8 Ma, 95% CI 36.1–134 Ma), with a median posterior age approximately 45 Myr after the earliest appearance of eudicots [35,64,78] and almost in agreement with a recent study based on a data set comprising more than 6000 nucleotides [79].

4. Conclusion

Our phylogenomic dating analysis, calibrated using new MSG fossils from the Middle Triassic, provides a revised time scale for the appearance of key insect innovations, shifting the median estimates and 95% CIs for several nodes to substantially earlier than reported by previous studies. Our results raise the possibility that the EPE had a lower impact on insect evolution than previously thought. Our results indicate that terrestrial or amphibiotic protopterygotes evolved approximately 435 Ma, and that the appearance of complete insect metamorphosis followed the origin of vascular plants (approx. 390 Ma). A number of hyperdiverse crown groups such as Lepidoptera, Hymenoptera, Diptera, Sternorrhyncha and Heteroptera may have evolved prior to the EPE and were not strictly associated with the evolutionary diversification of angiosperms. We tentatively propose that the EPE had only a limited impact on insect evolution and on the faunal turnover postulated to have occurred following this tremendous upheaval of Earth's biodiversity. Our results provide an updated time scale for comparative analyses of insect evolution.

Data accessibility

The transcriptomic data of 141 arthropod species consisting of 1478 single-copy nuclear genes, from which we obtained the dataset analysed in this study (220 615 aligned amino acids), are available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.3c0f1 [25]. Further information about the MSG fossils used for calibration is available in electronic supplementary material, tables S1 and S2.

Authors' contributions

M.M. and N.L. conceived the study; A.T. led and participated in fossil excavations; M.M, G.M, L.S. and A.T. analysed fossil specimens; M.M., K.J.T., G.M. and S.Y.W.H. performed the analyses; G.M., M.M., S.Y.W.H. and N.L. conceived and made graphs and figures; M.M. and N.L. wrote the manuscript. All authors discussed the results and commented on the manuscript.

Competing interests

We declare we have no competing interests

Funding

The study was partially supported by the Linnean Society of London and the Systematics Association (Systematics Research Fund 2016 assigned to M.M.). S.Y.W.H. and N.L. were supported by ARC Future Fellowships (FT160100167 and FT160100463).