SUPPLEMENTARY INFORMATION

Transcription

1 doi: /nature12629 Localities and horizons - The Avion outcrop in department of Pas-de-Calais (France) is of Moscovian age, more precisely dated to Westphalian C/D or equivalent to the Bolsovian/Asturian. The fossil insects were found in Terril N 7 that contains rocks from the slag heap of the coal mines 3 and 4 of Liévin, Bolsovian (Westphalian C, Mya, faisceaux de Ernestine ) /Asturian (Westphalian D, Mya, veines Arago, Dusouich, Marthe ) (Bruno Vallois pers. comm.). The sediment of the layers with these tiny insects is extremely fine, explaining their exceptional preservation. - The Commentry outcrop is located in the department of Allier (France) and is of Gzhelian (Stephanian) age. - The Quarry Piesberg (near Osnabrück, Lower Saxony, Germany) is of Asturian (Westphalian D) age 1. Systematic Paleontology: Extended descriptions and comments Standard wing venation terminology was employed throughout the descriptions and as it has been applied to representatives of Hymenoptera 2, Coleoptera 3, and Paraneoptera 4-7. We propose four new monotypic genera and families based on adult material, with the same diagnoses for the genera and families. However, the genus Melabolarva, corresponding to a holometabolan larva, is here considered without family and order attribution (see below). Since all taxa are monotypic, the included species and genera are automatically herein designated as the type species and type genus for their associated generic and familial names, respectively. Clade Holometabola Stem-group to Hymenopterida Family Avioxyelidae Nel, Engel, Garrouste et Coty, fam. nov. Avioxyela gallica Nel, Engel, Garrouste et Coty, gen. et sp. nov. Diagnosis. Forewing characters only. ScP zigzagged and short, with crossveins between it and C and R; a very large pterostigmal cell between branches of RA (autapomorphy); RA1+2 perpendicular to RA3+4 and to anterior wing margin; RP emerging from RP+MA at right angle; RP, MA and MP simple more or less parallel, with few large rectangular cells between them; a very small triangular cell just distal of the pterostigmal cell and a triangular radial cell near wing apex (synapomorphies with the hymenopteridan lineage); CuA long, completely independent of median vein, long CuP with several posterior branches and one cell between it and posterior wing margin (Fig. 1a, Supplemental Figs 1-2). Material. Holotype specimen Avion n 3 (part and counterpart of two wings), P. Roques coll., provisionally stored in the collection of Entomological Laboratory, MNHN, Paris, France; to be deposited in the Musée Géologique Pierre Vetter, Decazeville, France. Westphalian C/D (Bolsovian/Asturian), Avion outcrop, department of Pas-de-Calais, France. Etymology. The genus-group name is a combination of the type locality Avion and the primitive wasp genus Xyela. The name is feminine. The specific epithet refers to the ancient Roman name for France. Avioxyela is type genus for Avioxyelidae and A. gallica is type species for Avioxyela. Description. Two very small forewings, fragmentary, one with anterior portion and fragments of posterior portion preserved and the second with posterior portion preserved partially. These two wing fragments are partly overlapping. One is in better condition than the other that seems to have 1

2 been in great part deformed (Supplemental Fig. 2). Nevertheless ScP, stem of R with basal parts of RA, RP, M, and nearly complete CuA and CuP are visible under scanning electron microscopy (see Supplemental Figs. 2 d-e). The attribution of these two fragments to the two forewings of the same animal is based on the presence of the identical basal part of all veins. Length of first wing fragment ca. 5 mm, 1.7 mm wide; second wing fragment ca. 6 mm long, width 1.2 mm (strongly deformed); probable total length of forewing in life 6 mm; ScP short, ending on anterior wing margin in basal third of wing, well before apex of RA, rather strongly zigzagged, with three crossveins between it and radial stem and one between it and C; area between C and ScP broad, 0.17 mm wide; area between ScP and radial stem broad, 0.25 mm wide; a large and broad pterostigmal cell between RA and anterior wing margin, 1.1 mm long and 0.25 mm wide, basally limited by a long and straight vein called RA1+2 7, perpendicular to second branch of RA (RA3+4) and to anterior wing margin; a very small triangular cell just distal of pterostigmal cell; RA3+4 simple, ending obliquely on anterior wing margin, RP+MA emerging from RA just distad level of ScP apex, straight, with a crossvein 1ra-rp between it and RA; RP simple, separating from MA at right angle and reaching second crossvein 2ra-rp between it and RA, only three large cells between RP(+MA) and RA, including an acute radial cell along anterior wing margin; MA simple and straight; MP simple and straight with a crossvein ma-mp perpendicular to MA and MP and aligned with basal free part of RP; stem of Cu at least 0.35 mm long; divided into long and straight CuA and CuP; three veinlets perpendicular to MP and CuA defining four long rectangular cells, plus perhaps one more in a very proximal position; CuA long and straight, CuP very long and straight, with at least six posterior sigmoidal branches ending on posterior wing margin or (for the two basal ones) on anal vein; at least one anal vein, defining two large cells with CuP; anal vein with 3-4 distal branches ending on posterior wing margin. Discussion. The state of preservation of these two wings hinders their study and interpretation. The discovery after scanning electron microscopy of identical patterns in the two wings permitted exclusion that these were fragments of a single wing or wings from different individuals. Some Mazon Creek Herdinidae (an enigmatic family of Polyneoptera) 8 exhibit some wing venational traits superficially similar to those of Avioxyela, i.e., a similar shape to apical parts of ScP and RA, but the latter strongly differs from herdinids in the very regular pattern of the large rectangular cells in the radial and median areas, while the Herdinidae have an irregular network of small cells. In addition, the median and cubital veins of Avioxyela are without strong branches while the Herdinidae have numerous branches for these veins. Superficial similarities with some polyneopteran Grylloblattodea (Belmophenopterum pectinatum of the Sylvaphlebiidae) 9 can also be imagined if we hypothesize that the wings of Avioxyela are incomplete with the costal area missing, and if what we interpret below as RA and ScP could be in fact RP and a network of cells organised into an intercalary zigzagged vein between RA and RP. Either way, the supposed vein MA of Avioxyela is not organised as in Polyneoptera (except in Dictyoptera) because it is fused with what could be the stem of RP, and the supposed vein MP would be fused with supposed CuA, not with MA, unlike all Polyneoptera. Furthermore, nothing in these wings supports the hypothesis of a missing anterior wing margin and accordingly any attempt to attribute the wings to Polyneoptera seems improbable. The presence of only one median vein (MP) distally free (emerging from a common stem with R) is a character of Dictyoptera, Holometabola, and Paraneoptera (MA is not visible as a vein independent of R, it is completely fused with it). The character CuA independent of median vein excludes affinities with the Paraneoptera. Veins RP+MA, MP, CuA and CuP neutral is an apomorphy of the Holometabola as CuP is strongly concave in the Dictyoptera and CuA is 2

3 strongly convex in the Paraneoptera. Characters putatively synapomorphic with the superorder Hymenopterida are as follows: large pterostigma cell between branches of RA basally limited by a branch RA1+2 perpendicular to RA3+4 and to anterior wing margin (a structure somewhat similar only occurs in the Neuropterida Raphidiidae but these show considerably more branches of RP+MA); RP+MA divided in two parallel veins only (MA and RP) (character uniquely present in Hymenopterida); RP emerging from RP+MA at right angle (a character similar occurs in few Neuropterida, but with many more distal branches of RP and crossveins); very few large rectangular cells between the veins posteriorly emerging from the radial stem (character uniquely present in Hymenopterida); MP diverging from R+M (a character clearly present in Hymenoptera, even in the Triassic representatives 10,11, but also present in some Neuropterida); presence of a short ScP (some Permian Antliophora (e.g. Permopanorpa, Lithopanorpa) also show a shortened ScP but they lack the presence of several crossveins between apical part of ScP and RA, they also differ from Avioxyela and the Hymenopterida in the presence of many more branches of RP+MA, MP, CuA, and CuP 12 ); a very small triangular cell just distal of the pterostigmal cell and a triangular radial cell near wing apex having both a shape very similar to those of a Xyelidae (e.g. Angarixyela 13 ); MP and CuA simple (characters uniquely present in the Hymenopterida); only few large rectangular cells between main veins (character uniquely present in the Hymenopterida). Notable synapomorphies in the wing venation of crown-group Hymenoptera, such as the hind wings distinctly reduced relative to the forewings and the presence of hamuli on the anterior margin of the hind wings, are not preserved but were likely not present in this early progenitor of the Hymenoptera. The forewing of Avioxyela differs from those of the crown-hymenoptera in MP distally independent of RP+MA and CuA completely independent of median vein (only in the Xyelidae do small remnants of a free basal part of CuA occur 2,4 ), plesiomorphies which are expected for a basal, mecopteroid-like representative of the stem leading to Hymenoptera, as these veins are otherwise independent in the Neuropteida and in Panorpida. The common stem Cu separating into CuA and CuP is a plesiomorphic character present in many Polyneoptera (Plecoptera, Dictyoptera, etc.) and in Holometabola. It is especially clear in Neuropteroidea and in Panorpida. The cubitoanal area of Avioxyela strongly differs from those of crown-hymenoptera in the absence of the reduction of CuA but also CuP. By contrast, those of neuropteroids are strikingly similar to that of Avioxyela in the presence of cells delimited by CuP and the anal vein and also in the presence of several posterior branches of CuP 14. In Panorpida, CuP is also a long vein but with less posterior branches than in neuropteroids 15. As the Hymenopterida are currently considered as the sister group of all other Holometabola 16, the presence of the cubito-anal area organised in a similar way as in Panorpida and Neuropteroidea would putatively be plesiomorphic, while this part of the forewing is obviously reduced in the crown-hymenoptera. Another plesiomorphic character in Avioxyela is the zigzagged ScP with the anterior crossveins (but still present in some Hymenoptera, i.e., the Mesozoic Gigantoxyela quadrifurcata 17 ). The very large pterostigmal cell between RA1+2 and RA3+4 is an autapomorphy of Avioxyela, but it is not sclerotized as in the hymenopteran pterostigma. Despite of the absence of some important synapomorphies of the Hymenopterida, the most parsimonious solution is to attribute Avioxyela to the stem group of this superorder because of the presence of some synapomorphies that are not shared by the other clades of Holometabola. Clade Holometabola Stem-group to Coleopterida 3

4 Family Stephanastidae Kirejtshuk et Nel, fam. nov. Stephanastus polinae Kirejtshuk et Nel, gen. et sp. nov. Material. Holotype MNHN.A.49011, stored in the collection of the Palaeoentomological Laboratory, MNHN, Paris, France. Gzhelian (Late Stephanian), Commentry, Allier, France. Part of a body, pronotum with unclear remains of other prothoracic sclerites, distal part of right profemur and protibia with base of protarsus, base of left leg (procoxa, protrochanter and base of profemur), left median and posterior legs, two forewings partly overlapping, abdomen oriented obliquely downwards and exposed below from under forewings. One of the two forewings is preserved as a part while the second is a counterpart so that the veins appear with inverted convexity. Etymology. The genus-group name is a reference to the Stephanian stage (originally named after St. Stephen) and fellow citizen in Greek ( astos : astos or aste belongs to the astu (settlement) and likewise the polites or politis). The name is masculine. The species epithet is a matronym honoring Polina G. Kirejtshuk, daughter of one of the authors of the manuscript. Stephanastus is type genus for Stephanastidae and S. polinae is type species for Stephanastus. Diagnosis. Forewing Sc simple; RP and M simple; concave M and convex CuA with a common stem separated from R (synapomorphies of the superorder Coleopterida); very long CuP reaching distal eighth of forewing; anal veins very short at base of forewing (autapomorphies); forewings sclerotized and longer than abdomen, with very fine and dense iso-diametral cells forming regular rows generating intercalary veins along anterior edge, more diffuse near posterior margin; no adsutural line (sub-marginal anal vein along the whole posterior edge), medial edge of forewings not forming straight adjoining margins, apices not meeting apically; very large area between C and R; coxae not projecting, trochanters apparently short; lateral prothoracic carina absent (no prohypomera). Comparison. The pronotum of Stephanastus polinae forms a monolithic sclerite covering the dorsum and much of the lateral surfaces, as in many dictyopterans. Stephanastus has a considerable resemblance to Moravocoleus perditus and Oborocoleus rohdendorfi, both described from the Lower Permian of Obora, and interestingly also preserved in a seemingly characteristic laterally-compressed position. They all also have a similar structure to the prothorax 18. Nevertheless, both of the Obora fossils have somewhat wider anal areas to their forewings, which are also overlapping, making it difficult to compare the venation of these three species without re-examination of Kukalovà s species. Both of Kukalovà s species were described as belonging to the Protelytroptera, a putative early progenitor of the earwigs (Dermaptera), and were subsequently transferred to Coleoptera along with other tshekardocoleids 19, although these attributions were recently placed in doubt 3. Stephanastus has also a superficial similarity to some representatives of Glosselytrodea with M and CuA of their sometimes tegminous forewings fused sub-basally and with a body rather high but not dorsoventrally compressed; however, the latter group are distinct from Stephanastus in the RP(+MA) with more than three branches, the two veins running parallel in the mid section of the wing to the wing apex being anteriorly the convex MA and posteriorly the concave MP (in contrast to the concave MP and the convex CuA in Glosselytrodea), the shorter CuA and CuP (the latter scarcely extending beyond the midlength of the forewing), the simple concave M and convex R in the sclerotized forewings with dense isodiametric cells forming intercalary pseudoveins, and the short coxae and trochanters 20. Description (Figs. 1 b-d and Supplemental Fig. 3). Body (without head) length 28.0 mm, height at pro-mesothorax 4.5 mm; forewing length 25.5 mm, width 7.5 mm; length of mesotibia 4.8 mm. Head missing; thorax presenting clearly outlined pronotum (visible laterally) with lateral edge 4

5 subparallel to dorsal plane, probable propleuron (? + prosternum) appearing as a subparallel stripe; pronotum with heavily sclerotized and smooth stripe along whole posterior edge before which there is a strip with some very fine, distinct and dense furrows reaching lateral edge of pronotum; remainder of pronotal surface with very coarse irregular cellular sculpture formed by clear sharp ridges and surface of other prothoracic sclerites apparently with similar cells but somewhat smaller and with their marginal ridges somewhat weaker; remainder of thoracic sclerites and legs uniformly finely and densely punctured; fore and mid legs rather narrow and long; profemur with subparallel edges; protibia somewhat narrower and apparently longer than profemur, with subparallel edges and oblique apex; basal protarsomeres apparently narrow and simple; meso- and metathoraces exposed from under forewings without traces of outlines of sclerites and with remains of fine and dense sculpture of small punctures; probable mesotrochanter apparently rather short; mesofemur with slightly convex edges with groove for reception of protibia; mesotibia comparable in length with mesofemur, with nearly subparallel-sided edges and arcuate apex; mesotarsus narrow at least at base with remains of lobes; metatibia narrow and slightly curved; remains of metatarsus somewhat shorter than metatibia, very narrow (or with traces of lobes); abdomen turned downwards and with apex not reaching apices of forewings, its main part apparently rather soft and finely sculptured, with apical segment (or segments) heavily sclerotized. Forewing sclerotized and covered with small cells, aligned in pseudoveins especially visible in area between anterior margin and RA and in radial area; nearly all main veins simple and more or less straight, except for CuA forked near wing apex; areas between anterior margin and Sc and between Sc and R very broad, respectively 1.0 mm and 1.7 mm wide; Sc apically vanishing in network of cells; RA and RP separating 10.5 mm from wing base; RA going to wing apex and closely approximating anterior wing margin; two branches of RP both emerging directly from radial stem, and both simple, vanishing in network of cells near wing apex; M and CuA with a common stem well separated from R, separating 3.2 mm from wing base; RP and M weakly concave veins; strongly convex CuA forked close to wing apex; M and CuA approximate, separated by 0.6 mm; CuP simple, with a broad area between it and CuA, interspace 1.9 mm wide; area between CuP and posterior wing margin 1.7 mm wide; anal area very narrow and reduced, four (or five) longitudinal furrows correspond to anal veins (putatively) very narrowly separated from each other and from CuP and not extending behind anterior fourth. Discussion. Stephanastus shares with the Coleopterida and Umenocoleus Chen et T an, 1973 (an enigmatic Mesozoic taxon described as Coleoptera, later moved to Protelytroptera, also associated with various elytriform roaches in the Blattaria, but recently considered as belonging to the stem- Coleopterida 3 : strongly sclerotized forewing (but this character alone is not sufficient because sclerotized forewings are also present in some Dictyoptera, Glosselytrodea, Hypoperlida Perielytridae 21, and Hemiptera); forewing with a network of small cells very similar to those of beetles (especially visible in some Palaeozoic Tshekardocoleidae 3 ); Sc simple; RA convex and simple; RP simple (putative synapomorphy), more concave than RA; M less convex than R and CuA, simple, and basally fused with CuA (putative synapomorphy), but distally separating from it; M+CuA completely separated from radial stem; and CuA nearly reaching forewing apex (putative synapomorphy). The characters listed above exclude Stephanastus from all other clades that could have representatives with sclerotized forewings. The main differences with beetles and Umenocoleus are as follows and putatively support a sister-group relationship with the (Coleoptera + Strepsiptera): prothorax forms a monolithic sclerite covering dorsum and most part of lateral surfaces, no lateral pronotal carina (no evident prohypomera, plesiomorphy compared to the situation in Coleoptera), CuA with two very short branches near apex instead of being simple 5

6 (possible autapomorphy); CuP very long, instead of being shorter than CuA (plesiomorphy, character also present in Neuropterida); anal area and anal veins strongly reduced (autapomorphy), but not forming an adsutural line (submarginal anal vein) along whole posterior forewing edge (plesiomorphy). A further plesiomorphy shared with the Tshekardocoleidae is the very broad costal and subcostal areas, which is not so broad in many of the more derived beetle lineages. The legs of Stephanastus apparently have a short trochanter and non-projecting coxae. These features are shared with Palaeozoic Coleopterida (probable symplesiomorphies); however, the femora and tibiae are markedly longer than those in Permian beetles. One very important character distinguishing Stephanastus from Coleoptera is the structure of the prothorax. The new genus seems to maintain the initial structure of a prothorax with pronotum having steeply sloping sides and without division into a dorsal section with downcurved lateral sections (prohypomera, apomorphy of Coleoptera). Stephanastus may be considered as a representative of the stem-group to Coleopterida (= Coleoptera + Strepsiptera). It demonstrates for the first time the presence of progenitors of this hyperdiverse lineage in the Pennsylvanian, something already suspected given the recent discovery of Tshekardocoleidae in the Lowermost Permian of Germany 3. Having summarized the current knowledge on Coleoptera and relatives, it is hypothesized that the superorder Coleopterida comprised the subclades (Stephanastus + (Coleoptera + Strepsiptera)). Clade Holometabola Order Incertae sedis Familia incerta sedis Metabolarva bella Kirejtshuk, Prokin, Nel et Wappler, gen. et sp. nov. Material. Holotype MAS Pal 588, coll. A. Leipner , stored in Museum am Schölerberg (MAS), Osnabrück. Quarry in the Piesberg near Osnabrück, Lower Saxony, Germany, Clay stone layer above the Zweibänke coal seam, Osnabrück Formation, Pennsylvanian (Moscovian: Westphalian D). This fossil is the print and counterprint of a body, yellowish to bright yellow on blackish coal, in a somewhat dorsolateral position turning anteriorly to head which is visible dorsally, thoracic segments apparently more sclerotized (with more bright yellow coloration) and more distinctly defined, abdominal segment apparently membranous and with unclear separation, except clearly isolated apical segment. Diagnosis. Body mostly membranous; head transversely suboval with apparently short antennae (if they were not partly lost during fossilization), frontal and epistomal sutures, apparently traced isolated clypeus and labrum, also apparently traced large mandibles; prothorax with paired wide and heavily sclerotized plates narrowly separated along the middle, shorter than meso- and metathoraces; thoracic legs at least four-segmented; abdomen membranous, apparently with (?) eight segments with limits unclear, gradually enlarging distally and with a stigmatal plate bearing a pair of spiracular orifices; distal preapical abdominal segments bearing in total four pairs of thick appendages forked into two (anterior and posterior) processes (or posterior one apparently presented by short tubular base bearing a very thick and short conical sensilla), these appendages without any trace of both stigma or trachea. Etymology. The name of the new genus is formed from the term metabolism and Latin larva (immature active stage of ontogenesis). The name is feminine. The species name bella is in reference to its nice state of preservation. Metabolarva bella is type species of the genus Metabolarva. 6

7 Description (Figs. 2 a-b and Supplemental Fig. 4). Body 9.1 mm long; head 0.8 mm long, 1.4 mm wide; prothorax 0.7 mm long, 2.4 mm wide; mesothorax 1.0 mm long; metathorax 1.0 mm long, 2.2 mm wide; abdomen 5.8 mm long; head transversely oval, with clear frontal and transverse epistomal sutures, an unclear large subtrapezoidal clypeus, apparently short subsemi-circular labrum, unclear short antennae and apparently large subtriangular mandibles partly visible; body segments looking like transversely waving; transverse pronotum subcylindrical (subquadrangular from above), with a pair of paramedian blackish sclerites in anterior 2/3, narrowly separated in the middle; meso- and metathoracic segments subconical, thickening posteriorly and with subuniform sclerotization; prothoracic and one visible mesothoracic legs, apparently with four exposed heavily sclerotized segments; metathoracic legs not clearly visible, but in place, where it could be expected, there is a reddish unclear spot on blackish substrate; abdomen with segmentation not clearly expressed, probably with eight segments somewhat traceable in membranous crimps (folds); distal preapical segments apparently thicker than previous (proximal) ones; four apical segments with thick lateroventral soft appendages apparently forked into two processes (or posterior one apparently with a short tubular base bearing a very thick and short sensilla), these appendages without any trace of stigmata or tracheae; apical segment with a probable oval plate visible dorsally, bearing a pair of spiracles. Comparison. This larva is characterised by a rather large, soft and membranous body providing evidence that it was at least older than the first instar, although it does not show any trace of wing pads. Taking into consideration the peculiarities described below (and also the lack of wing pads), it can be defined as belonging to the eruciform oligopod type which occurs among some holometabolan groups (Trichoptera, Coleoptera, Hymenoptera). The specimen examined is somewhat similar to hexapod nymphs of Diplopoda and Pauropoda, however it has a large separate clypeus, but no additional legless segment before the head (collum) and no large antennae. In addition, this animal differs from the Middle Devonian-Early Permian lineage Arthropleuridea in their membranous body, not dorsoventrally compressed, and without lateral flat lobes; and also head with large clypeus and large antennae. Finally, it differs also from the Silurian-Triassic Euthycarcinoidea in the membranous and not sub-flattened body, absence of two pairs of antennae, absence of paired legs on wide thoracic segments, as well as absence of narrow abdominal segments and narrowly acuminate abdominal apex. The present fossil has many similarities with beetle larvae, but the more or less expressed transverse epistomal sulcus suggests that its epicranium could have had a movable clypeus which is not characteristic of Coleoptera but is present in Megaloptera, Raphidioptera, and Trichoptera. In the following traits this larva differs from those of the modern holometabolan orders: - Neuroptera: in the lack of an intercalary segment (collar) between head and pronotum; lack of curved mandibles that are not prognathous; and from the eruciform larvae of Ithonidae and Mantispidae in the probable absence of spiracles on most abdominal segments; - Mecoptera: from nannochoristid-like (Nannochoristidae) larvae in the presence of articulated legs; from the eruciform larvae of Panorpidae; from Choristidae and Bittacidae in the absence of abdominal pseudolegs; also from Panorpidae and Boreidae in the absence of spiracles on most abdominal segments; - Raphidioptera: in the paired sclerotized pronotal sclerites, a smaller number of abdominal segments, and probable absence of spiracles on most abdominal segments; - Megaloptera: in the subuniform sclerotization of meso- and metanota, absence of processes on the ultimate abdominal segment, and probable lack of spiracles on most abdominal segments; 7

8 - Coleoptera: in the presumed metapneustic respiratory system, which is absent in members with an eruciform body; - Strepsiptera: in the presence of thoracic legs (from parasitic forms) and also in the weakly sclerotized body, presence of the antennae and absence of long processes at abdominal apex (from the triungulin form); - Trichoptera: in the absence of tracheal gills and pseudolegs (= prolegs) on the ultimate abdominal segment; - Lepidoptera: in the absence of pseudolegs (= prolegs) and spiracles on most abdominal segments; - Hymenoptera: in the absence of abdominal pseudolegs (= prolegs); - Diptera: in the well developed thoracic legs. Discussion. The taxon Srokalarva bertei Kukalová-Peck, 1997 from the Moscovian of Mazon Creek was purported to be a holometabolan larva. It was never formally described and has to be considered as a nomen nudum until a formal description is published. No photograph of this fossil has ever been published and it is known only from interpretative drawings 6,22. Moreover, the material upon which it was established has never been made available for further study. Apart from its three pairs of thoracic legs, this fossil was drawn as also having leglets with claws on abdominal segments 2-6 but not on the more distal segments. It is possible to obtain by genetic manipulation a beetle larva with articulated abdominal legs on segments This is congruent with the hypothesis of a Pennsylvanian holometabolan larva with abdominal legs 24. Nevertheless, Srokalarva also looks like a myriapod as behind its metathorax is figured a segment different from those thoracic and abdominal. This intermediate segment could correspond to that of a myriapod, and the legless segments could be alternating with the legged ones 20. A revision of Srokalarva is urgently needed. The second described evidence putatively of a Pennsylvanian holometabolan larva is the ichnofossil (gall) Pteridotorichnos. It remains unassignable to any known extant lineage of Holometabola despite the superficial resemblance to sawfly larval damage 25. The important peculiarity of Metabolarva bella is the possible possession of a metapneustic respiratory system with spiracular orifices only situated at the anal end of the body. Among modern holometabolans such a respiratory system occurs in Diptera, Coleoptera, and Hymenoptera. Among modern Coleoptera the metapneustic respiratory system, with a single pair of functional spiracles on abdominal segment 8, is known in Scirtidae; genus Araeopidius (Ptilodactilidae); psephenid Eurbiinae; and early and intermediate instars of Noteridae; Amphizoidae; Dytiscidae, and chrysomelid Donaciinae 26 and probably in early and intermediate instars of the fossil family Coptoclavidae. However, all coleopteran larvae with the metapneustic respiratory system are not eruciform. Among Diptera, larvae with a metapneustic respiratory system inhabit aquatic environments or liquid and wet substrates (viz. Tipuloidea; Culicidae; Psychodidae; some Dolichopodidae; Ephydridae; Stratiomyidae; Syrphidae; Muscidae; Sarcophagidae; Scathophagidae, etc. and also dipteran larvae of the first instars which change their metapneustic system on the peri-, hemi- and amphipneustic ones in the further instars 27 ). The parasitic hymenopterans with first-instar larvae possessing a pair of functioning terminal spiracles (metapneustic respiratory system) have their final-instar larvae with nine pairs of spiracles present, on segments 3-11 or Such circumstances and the distribution of this respiratory system make it possible to infer that the appearance of spiracles in different groups could have some ontogenetic or morphogenetic predisposition. Ponomarenko 29 inferred that the holometabolan ancestor had an eruciform larva making it possible to generate separate ecological niches for adults and larvae 30. The metapneustic type of respiratory 8

9 system of Metabolarva bella would, if true, lend credence to the idea that it inhabited a moist substrate. According to Ponomarenko 19,29 the oldest beetles and, perhaps holometabolans in general, could be associated with wet and strongly decayed wood infested by fungi. The first holometabolans may have developed in holes of trunks and in stumps filled with water and different organic remains. Such habitats can explain the following: (1) the presence of thoracic legs in M. bella for moving in ephemeral localities; (2) the presence of metapneustic type of respiratory system for development in substrate with a lower general concentration of oxygen; and (3) the presence of abdominal processes without tracheae quite characteristic of inhabitants of moist substrates. Weak sclerotization of the body with transverse metachronal undulations could be regarded as somewhat analogous to the moving callosities in buprestids, cerambycids, and other beetle larvae living in wood. Alternatively, the ancestral holometabolan larvae could have developed in the male cones of gymnosperms, similar to the larvae of the most basal modern hymenopterans 19,20. The present finding is in accord with both explanations and cannot distinguish between the alternative hypotheses. Clade Paraneoptera Order Hemiptera Clade Euhemiptera Family Aviorrhynchidae Nel, Bourgoin, Engel et Szwedo, fam. nov. Aviorrhyncha magnifica Nel, Bourgoin, Engel et Szwedo, gen. nov., sp. nov. Diagnosis. Forewing with a cua-cup vein, long concave PC, very short concave ScP fused with R near wing base and re-emerging distally, a fringed ambient vein (synapomorphies with Paraneoptera or Euhemiptera); a series of veinlets between radius and anterior ambient vein; areola postica with long simple branches; simple CuP and AA1+2 fused near wing base and separating again distally. Material. Holotype specimen Avion n 2 (part and counterpart of a forewing), P. Roques coll., provisionally stored in the collection of Entomological Laboratory, MNHN, Paris, France; to be deposited in the Musée Géologique Pierre Vetter, Decazeville, France. Westphalian C/D (Bolsovian/Asturian), Avion outcrop, department of Pas-de-Calais, France. Etymology. The generic name refers to the type locality Avion, while ~rrhyncha refers to the common suffix of related suborders (Sternorrhyncha, Auchenorrhyncha, etc.). The name is feminine. The species epithet refers to the wonderful state of preservation of the holotype. Aviorrhyncha is type genus for Aviorrhynchidae and A. magnifica is type species for Aviorrhyncha. Description (Figs. 2 c-e, Supplemental Figs. 5 and 6a) Forewing short but relatively broad, preserved part 7.0 mm long, 3.4 mm wide, wing ca. 7.5 mm long; with anal area missing in large part and anteroapical section folded, several parallel folds mimicking concave branches of neutral or convex veins present in posterior half of wing; wing margin developed as classical crosshatched ambient vein, also visible, but narrower in posterior part of wing, extending to level of apex of AA1+2 (= vein E sensu Hamilton) 5,6 ; a concave vein CP ending on costal margin just before level of probable flexion zone, with forking near apex; one crossvein between anterior wing margin and CP and six between CP and R; a strongly concave independent ScP extremely short at wing base, 0.5 mm long, fused with R and re-emerging 2.5 mm distally as a sigmoidal veinlet between R and costal margin, more concavely marked than other veinlets in this area; basal part of CuA 0.3 mm long, distally fused with R+M; stem (ScP)+R+M+CuA short, ca. 0.5 mm long, forked into convex R(+ScP) and more concave M+CuA basad veinlet cua-cup; R nearly straight almost to its apex, 9

10 forked twice in apical portion, then with three terminals (RA and RP); seven long sigmoidal veinlets between R and costal margin, third one being more pronounced than others (re-emerging ScP); distal common vein M+CuA longer than common stem ScP+R+M+CuA, forked at level of apex of weakly concave basal flexion line ; short stem of M more concave than stem of CuA, with anterior branch M1+2 forked into M1 (two-branched) and M2 at level of distal flexion zone and one time again distally, 9-10 parallel and more or less sigmoidal crossveins in area between R and M; posterior branch M3+4 of M forking late into M3 and M4; three crossveins between M1+2 and M3+4, four between M3+4 and CuA; convex stem CuA forked into CuA1 and CuA2 slightly after distal flexion zone (wing is ripped at level of this fork); a faint neutral sigmoidal veinlet cuacup between stem CuP and M+CuA, not connected to CuP and with very small branches near CuA; three or four other straight crossveins just distally; CuP narrower and less convex than surrounding veins M+CuA and preserved anal vein; CuP distally almost straight, forked 1.2 mm distal of cua-cup, into two posterior branches CuP and AA1+2 both reaching posterior wing margin; anterior branch (CuP) apparently more concave than posterior branch (AA1+2); a fragment of distinctly more convex anal vein AA3+4 posterior to CuP with four preserved crossveins between them. Discussion. This fossil corresponds to a forewing based on the broad costal area with numerous transverse veinlets and the general shape suggesting that the anal area is not expanded, as is usually the case in hind wings. The fossil has the paraneopteran synapomorphy in wing venation, i.e., CuA basally fused with R and M, these veins being separated distally, plus the presence of a faint crossvein cua-cup between CuP and CuA 4. An attribution to the Thripida is excluded given the complex forewing venation, and exclusion from the Psocodea is supported by the ScP short and completely independent of R, the presence of numerous crossveins between main veins, the presence of a basal and distal flexion lines, and CuA with four apical branches. In particular, the Psocodea have a ScP at least distally, if not completely, fused with RA, and re-emerging near its apex as a short vein. Aviorrhyncha shares with some of the Euhemiptera, particularly Fulgoromorpha, the following forewing characters: presence of an ambient vein (synapomorphy); a well developed, concave CP 7 (synapomorphy, vein absent in Psocodea, Thripida, and Sternorrhyncha); a concave ScP fused with R very close to wing base and re-emerging distally as a veinlet between R and anterior wing margin (proposed character of Euhemiptera), so that there are two concave veins (CP and ScP) close to wing base in area between R and anterior wing margin; a convex vein AA1+2 at least partly fused with the concave CuP 5-6, as seen in extant Fulgoromorpha (Supplemental Fig. 6b) (proposed character of Euhemiptera). The presence in Aviorrhyncha of a CuP+AA1+2 with a deep distal fork supports the hypothesis that this vein is a composite at least in Euhemiptera, due to the fusion between CuP and AA1+2 5,6 (see Supplemental Fig. 6 b). The forewing crossvein cua-cup could be also interpreted as the mid-section of a zizag in the course of vein CuA, and in this scenario the vein CuP+AA1+2 would be the distal part of CuA, but this interpretation is unlikely given that cua-cup is neutral, distinctly weaker than M+CuA and CuP+AA1+2, branching on these veins as a secondary crossvein and not as a branching of M+CuA, and with very small and faint branches near M+CuA. The preferred interpretation of this vein as a cua-cup implies that CuA is not zigzagged and unlike the situation most common in Euhemiptera, although some modern representatives of this clade have CuA not zigzagged. The Serpukhovian Protoprosbole straeleni (also based on an isolated forewing) and Aviorrhyncha have similar venations in the forewing (ambient vein and CP present, similar patterns of RA, RP, M with two subdivisions and four apical branches, CuA forked, CuP with distal branches, 10

11 presence of a basal flexion line in the same position between M+CuA and ScP+R and a distal flexion line crossing the wing, a strongly convex anal vein AA1+2 fused with CuP at some distance from the wing base), suggesting also a position within the Euhemiptera. The lack of further information about the anal area in Aviorrhyncha being reserved, the main differences between Aviorrhyncha and Protoprosbole are CuA and CuP both with three apical branches in the latter. Protoprosbole was originally considered as belonging to the Hemiptera and as a possible earliest step in evolution of Paraneoptera by Sharov 31, but later as Hypoperlida (a paraphyletic group without synapomorphies) by Rasnitsyn 20 who added that the Protoprosbolidae should be synonymized with the Ampelipteridae Haupt or Aenigmatodidae or Mixotermitidae. The group was recently revised and returned to the Hemiptera 4. Clade Paraneoptera Order Psocodea Family Westphalopsocidae Azar, Nel, Engel et Bourgoin, fam. nov. Westphalopsocus pumilio Azar, Nel, Engel et Bourgoin, gen. et sp. nov. Diagnosis. Forewing narrow elongate with parallel margins, with a cua-cup crossvein; two anal veins fused apically; CuA forming an areola postica; ScP distally fused with RA and re-emerging near its apex; a pterostigmal cell (characters present in Psocodea) between apices of ScP and RA; a series of crossveins between ScP and C (autapomorphy); anterior branch of RP emitting four short anterior branches; three posterior median branches; CuP simple; two anal veins coalescing and fusing apically for a distance. Material. Holotype specimen Avion n 4 (part of a forewing, without counterpart), P. Roques coll., provisionally stored in the collection of Entomological Laboratory, MNHN, Paris, France; to be deposited in the Musée Géologique Pierre Vetter, Decazeville, France. Westphalian C/D (Bolsovian/Asturian), Avion outcrop, department of Pas-de-Calais, France. The veins are fossilised as concretions so that their convexity on the fossil does not correspond exactly to their original convexity. It is clearly visible for ScP that has a normal concavity in its basal part but looks convex distally. Etymology. The generic name refers to the Westphalian stage and Psocus, common stem of many generic names in the Psocodea. The name is masculine. The species epithet refers to the very small size of the species. Westphalopsocus is type genus for Westphalopsocidae and W. pumilio is type species for Westphalopsocus. Description (Fig. 2 f and Supplemental Fig. 7). Part of a very small and narrow elongate forewing with extreme base missing, preserved portion 3.1 mm long, 0.9 mm wide; wing estimated ca. 3.6 mm long in life; no ambient vein; ScP long, basally separated from R, fused with it 1.3 mm basal to wing apex and distally re-emerging from RA 0.1 mm distal of their fusion to end on anterior wing margin; cell between RA and ScP along anterior wing margin deeper than other cells along anterior wing margin, corresponding to a sclerotized pterostigma; area between ScP and C 0.2 mm wide, with five very thin curved crossveins in its distal part; radial stem straight, with RA and RP separating 1.9 mm from wing apex; RA ending on anterior margin 0.2 mm from wing apex; area between ScP and R 0.1 mm wide, free of crossveins; RP distally divided into two main branches, anterior one being subdivided into four short veinlets and posterior one being forked and ending at wing apex; three crossveins between RA and RP in distal part of wing; M and CuA basally fused, closely parallel to R and separating into M and CuA 2.2 mm from wing apex; M with three posterior branches; two crossveins between RP and M at distal half of wing; CuA distally forked, defining an areola postica; veinlet cua-cup 0.3 mm long, between CuP and M+CuA; CuP slightly 11

12 curved and simple; two anal veins clearly visible, coalescing and fusing apically into an inverted Y-vein, i.e., distally fused into a single vein ending on posterior wing margin; a third anal vein apparently present between anal stem and CuP. Discussion. Westphalopsocus can be attributed to the Paraneoptera rather than to Archaeorthoptera 32 or to the poorly defined, paraphyletic and probably polyphyletic Hypoperlida 4,20, owing to the presence of a cua-cup vein between CuP and M+CuA (apomorphy); presence of an aerola postica and of anal veins fusing apically and thus forming an inverted Yvein. These two last specialized characters are convergent among certain Psocodea but not all, such as Zygopsocidae (Zygopsocus); Amphientomidae (Electrentomum, Paraelectrentomum); Manicapsocidae (Nothoentomum); and many Hemiptera, viz. Paleontinidae (Suljuktaja and Suljuktocossus); Probolopseidae (Mundus); Dictyopharidae (Netutela); Boreoscytidae (Archescytinopsis); and Protopsyllididae (Permopsyllidium, Clavopsyllidium, and Psyllidella). ScP completely separated from R until their distal fusion near their apices is a character only present in Psocodea, even if it is absent in some of them in relation to reduction of the wings. This character is always absent in Hemiptera. The sclerotized pterostigma is a character proper to the Psocodea but also to some Hemiptera. The presence of several very thin and curved crossveins in the area between C and ScP is reminiscent of the pterostigmal structure of the psocodean Neurostigma (Neurostigmatidae) 33. Several crossveins of RA in the distal portion of the costal area are proper to some Psocodea like Neurosema and Calopsocus (Calopsocidae). The elongate shape of the forewing with anterior and posterior margins parallel is a character present in many Psocodea but not in Hemiptera. Most noteworthy, the possible presence of an anal vein between the anal stem and CuP would be indicative of the Euhemiptera, and is absent in Psocodea. Westphalopsocus can be definitively attributed to the Paraneoptera, and possibly to the Psocodea, based on this unique combination of features even though the fossil exhibits no single, unreversed psocodean synapomorphy as it is preserved (something which might be expected in an early stem group). Westphalopsocus shows superficial similarities with the archaeorthopteran Ampeliptera limburgica in that ScP is distally fused with R and the presence of two veins fused in their distal portion in the anal area. Crucial differences are the CuA with numerous branches in the latter and the presence of only two veins in the anal area and of a simple CuP in the former 32,34. On the affinities of Miomoptera, Glosselytrodea, Anthracoptilidae, and Tococladidae The Pennsylvanian to Permian Miomoptera have been considered as possible Holometabola or as stem-group Paraneoptera, but are likely a polyphyletic group and unrelated to the aforementioned lineages 4. The Permian to Jurassic Glosselytrodea are currently of uncertain position although likely belonging among the Polyneoptera 13. The Palaeozoic Anthracoptilidae Handlirsch, 1922 (= Strephocladidae Martynov, 1938) were considered as Holometabola 35. After a new examination of the type specimens of Anthracoptilus Lameere, 1917 and Strephocladus Scudder, 1885, Guan, Prokop and Nel (in prep.) alter these conclusions and consider that this family is related to a clade (Paoliida + Dictyoptera) on the basis of the synapomorphy presence of a strongly convex main stem of CuA emitting concave anterior branches 36. This is partly in agreement with earlier authors who considered strephocladids as dictyopterans 37. The Palaeozoic Tococladidae were considered as members of the Holometabola 35 but they have a wing venation typical of the Archaeorthoptera, as confirmed by Prokop, Roques and Nel (submitted) and Béthoux

13 1. Dunlop, J. A., Brauckmann, C. & Steur, H. A Late Carboniferous fossil scorpion from the Piesberg, near Osnabrück, Germany. Fossil Record 11, (2008). 2. Kukalová-Peck, J. & Lawrence, J. F. Relationships among coleopteran suborders and major endoneopteran lineages: evidence from hind wing characters Eur. J. Entomol. 101, (2004). 3. Kirejtshuk, A.G., Poschmann, M., Prokop, J., Garrouste, R. & Nel, A. Evolution of the elytral venation and structural adaptations in the oldest Palaeozoic beetles (Insecta: Coleoptera: Tshekardocoleidae). J. Syst. Palaeontol., in press (2013). 4. Nel, A., Prokop, J., Nel, P., Grandcolas, P., Huang, Di-ying, Roques, P., Guilbert, E., Dostál, O. & Szwedo, J. Traits and evolution of wing venation pattern in paraneopteran insects. J. Morphol. 273, (2012). 5. Hamilton, K. G. A. The insect wing, Part 3. Venation of the orders. J. Kansas Entomol. Soc., 45, (1972). 6. Kukalová-Peck, J. in The insects of Australia, a textbook for students and research workers, Vol. 1 (ed Naumann, I. D.) (Melbourne University Press, 1991). 7. Dworakowska, I. Main veins of the wings of Auchenorrhyncha. Entomol. Abh. 52, (1988). 8. Kukalová-Peck, J. in Richardson s Guide to the Fossil Fauna of Mazon Creek (eds Shabica, C. W. & Hay, A. A.) (Northeastern Illinois University 1997). 9. Rasnitsyn, A. P. & Aristov, D. S. Two new insects from the Upper Permian (Tatarian) of Belmont (NSW, Australia) (Insecta: Hypoperlida: Anthracoptilidae = Strephocladidae; Grylloblattidae: Phenopteridae). Paleontol. J. 38, S158 S163 (2004). 10. Riek, E. F. Fossil insects from the Triassic beds at Mt. Crosby, Queensland. Austral. J. Zool. 3, (1955). 11. Rasnitsyn, A. P. [New Triassic Hymenoptera of Central Asia.] Paleont. Zhurn. 1964, (1964). 12. Willmann, R. Evolution und Phylogenetisches System der Mecoptera (Insecta: Holometabola). Abhand. Senckenber. Naturf. Ges. 544, (1989). 13. Grimaldi, D. A. & Engel, M. S. Evolution of the Insects. (Cambridge University Press, 2005). 14. Shi, Chao-fan, Béthoux, O., Shih, Chung-Kun & Ren, Dong. Guyiling jianboni gen. et sp. n., an antlion-like lacewing, illuminating homologies and transformations in Neuroptera wing venation. Syst. Entomol. 37, (2012). 15. Kukalová-Peck, J. & Willmann, R. Lower Permian mecopteroid-like insects from Central Europe (Endopterygota). Can. J. Earth Sci. 27, (1990). 16. Yeates, D. K., Cameron, S. L. & Trautwein, M. A view from the edge of the forest: recent progress in understanding the relationships of the insect orders. Aust. J. Entomol. 51, (2012). 17. Gao, Tai-ping, Ren, Dong & Shih, Chung-Kun. Abrotoxyela gen. nov. (Insecta, Hymenoptera, Xyelidae) from the Middle Jurassic of Inner Mongolia, China. Zootaxa 2094, (2009). 18. Kukalová, J. On the systematic position of the supposed Permian beetles Tshecardocoleidae, with a description of a new collection from Moravia. Sborn. Geol. Ved, Paleontol. 11, (1969). 19. Ponomarenko A. G. Istoricheskoe razvitie zhestkokrylykh-arkhostemat [Historical development of the Coleoptera - Archostemata.] Trudy Paleontol. Inst. Akad. SSSR 125, (in Russian) (1969). 13