The first Permian Diaphanopterodea (Insecta, Megasecopteromorpha) from China

A new Diaphanopterodea (Insecta, Palaeoptera, Megasecopteromorpha), Sinoelmoa yangquanensis gen. et sp. nov. , is described based on a single specimen discovered from the Shuiquan Gully locality (Shanxi For mation; Permian, Cisuralian, Asselian; China). A broad comparative analysis of the wing venation of the known members of the diaphanopterodean families Parelmoidae and Elmoidae allowed assigning the new taxon to the former family. This new occurrence represents the first record of a Permian Diaphanopterodea from China, and both the earliest and most oriental record for the Parelmoidae. It sheds new light on the distribution and diversity of these extinct taxa.


Introduction
The insect order Diaphanopterodea, a member of the broader taxon Megasecopteromorpha Béthoux, 2020in Yang et al. (2020), is a particular member of Late Palaeozoic faunas.Compared with other megasecopteromorphan lineages, it is represented by fewer species and fossil specimens.The capacity of these insects to hold wings backwards, along the abdomen, at rest ('neoptery') has long been considered a distinctive feature of the group (Carpenter 1992).Different opinions emerged as to whether the occurrence of this ability, in such ancient insects, should be regarded as the ancestral condition for the entire winged insects, or should be accounted for by convergent acquisitions, within Megasecopteromorpha and within Neoptera (Yang et al. 2020;and references therein).The latter hypothesis is now generally accepted.Also, in the course of their 75 million years of evolution, the group evolved very small forms, with a very reduced wing venation (Carpenter 1992) reminiscent of that of Hymenoptera.
One of the best documented families is the Parelmoidae Rohdendorf, 1962, composed of various Cisuralian (i.e.lower Permian) genera and species from Czech Republic (Obora; Sakmarian), USA (Elmo & Midco; Artinskian), France (Lodève; Kungurian) and Russian Federation (Chekarda; Kungurian).Here, we describe a new, isolated wing discovered from the Shuiquan Gully, Shanxi Formation (China; Asselian), which can be confidently assigned to this family.Our broad comparative analysis suggests that it belongs to a previously unknown genus and species.It constitutes the first record of Permian Diaphanopterodea from China.

Geological setting
The new material (specimen YQZYW 15) was collected from the siltstone layer of the middle Shanxi Fm., in a rock profile at Duanjiabei District, Yanquan, Shanxi Province, North China (Fig. 1A-C).The Shanxi Fm. (also spelled 'Shansi' in some accounts) is a set of deposits of continental-oceanic interaction facies, mainly consisting of coal, mudstone, siltstone, and sandstone, which conformably overlies the Taiyuan Fm. and conformably underlies the Xiashihezi Fm. (Fig. 1D).The upper part of the fossiliferous layer is blackish grey and contains some invertebrates, such as brachiopods and gastropods, while the lower part is grey with abundant plant fossils, including Neuropteris ovata, Pecopteris orientalis, Pecopteris linsiana, Pecopteris sp., Sphenopteris nystroemii, Sphenopteris sp. and Cordaites sp. (Fig. 1D) (Wu 1997).The newly described insect specimen was collected from the lower layer.Based on lithological characters, this layer likely belongs to the Shanxi Fm., which is Asselian in age (Permian, Cisuralian; Shen S. et al. 2020;Shen B. et al. 2022).

Documentation of fossil material
The new material (specimen YQZYW 15) is housed at Yangquan City Planning and Natural Resources Bureau (Geological Specimen Room).It was photographed using a Nikon SMZ25, and a Canon 5DS coupled with a Canon MP-E 65 mm macro lens, under both dry and ethanol conditions.Photographs of this specimen reproduced herein are the result of a combination of photographs taken under both dry (best-preserved side) and ethanol conditions (both sides) ('eth-eth-dry' composite).
New photographs of material housed at the Museum of Comparative Zoology (MCZ; Cambridge, MA, USA) were provided by the MCZ staff.A photograph of the specimen PIN 1700/492 (Paleontological Institute, Academy of Sciences; Moscow, Russia; part of Fig. 2F) was provided by Anastasia Felker.All photographs were optimized using Adobe Photoshop CS6 (Adobe Systems, San Jose, California, USA).
New photographic data on the specimen Ld LAP 365 (Musée of Lodève, France) were also collected.Two reflectance transformation imaging (RTI) files were generated based on photographs taken using a ~30 cm diameter, automated light dome driving a Canon EOS 5DS digital camera, itself coupled to a Canon MP-E 65 mm macro lens.The first file provides an overview of the side Ld LAP 365a, while the second file is focused on the wing base of the side Ld LAP 365b (each based on a set of 42 photographs).Original photographs were optimized using Adobe Photoshop CS6 prior to RTI processing, itself achieved using the RTI builder software (Cultural Heritage Imaging).We provide an online Dryad dataset (Yang et al. 2024, forthcoming) containing these RTI files.Images (other than drawings) reproduced on Fig. 5 were extracted from the overview RTI file, the second item having been obtained using the 'normals visualisation' mode, which assigns a colour code to each pixel according to the orientation of the vector perpendicular to the plane tangent to the object at the corresponding point.This extract was optimized for contrast.
In addition to photographs, hand-drafted drawings were also produced.For the MCZ material such drawings were prepared using an unspecified dissecting microscope equipped with a drawing tube; for the specimen Ld LAP 365, hand-drafted drawings were prepared using a Zeiss SteREO Discovery V8 Stereomicroscope equipped with a pair of W-PL 10×/23 eye pieces, a Plan Apo S 1.× FWD objective, and a drawing tube (Jena, Germany); and for the specimen YQZYW 15, hand-drafted drawings were prepared using a Leica MZ75 equipped with a drawing tube.Final drawings were then prepared using both draft drawings and photographs.For the specimen PIN 1700/492, a drawing was derived from photographs only.

Terminology
We follow the serial insect wing venation groundplan and the associated wing venation nomenclature (Lameere 1922(Lameere , 1923;;Kukalová-Peck 1991) Remarks.Commonly among Megasecopteromorpha, and more particularly in Diaphanopterodea, MA shows some degree of connection with RP.It ranges from a connection via a strong cross-vein, shortly after the origin of MA, to a full fusion with R, then continuing along RP, from which MA diverges at some stage.Because Diaphanopterodea also exhibit oblique crossveins, it can sometimes be difficult to determine whether an oblique structure occurring between RP and MP is the genuine MA or a cross-vein.Unlike previous authors (Carpenter 1963(Carpenter , 1992;;Béthoux and Nel 2003), Prokop and Kukalová-Peck (2017) suggested that a full fusion of MA with R/RP occurred in Diaphanoptera Brongniart, 1893, the type-genus of the family from which the names of the order derives.These authors based their interpretation on vein elevation as observed in the specimen MNHN.F.R51214, holotype of Diaphanoptera munieri Brongniart, 1893.However, our observation reveals that rock compression this particular specimen  experienced (like several other Commentry specimens) makes it impossible to derive a solid inference on the elevation of elements attributable to MA.In contrast, the specimen MNHN.F.R51196 [considered conspecific to the specimen MNHN.F.R51214 by Béthoux and Nel (2003)] is better preserved in that respect.Our observation revealed that the structure regarded as the base of MA by Béthoux and Nel (2003) and as a cross-vein by Prokop and Kukalová-Peck ( 2017) is indeed concave, while MA is convex distal to the point where it diverges from RP.This difference in elevation underlies the argument by Prokop and Kukalová-Peck (2017).However, an elevation shift, applying to both MA and CuA, is a general feature of the broader taxon Rostropalaeoptera (Béthoux 2008).Most decisively, in Eukulojidae, whose wings lack cross-venation, the then undisputable bases of MA and CuA are both concave for a short distance before turning convex, after having approached RP and M, respectively (Béthoux 2008: fig.3).The assumption that MA is fused very early with R/RP in Diaphanoptera is therefore unsubstantiated.
Commented diagnosis.ScP long, ending beyond the first fork of RP (plesiomorphy within Diaphanopterodea); near wing base, shortly after its origin, CuA fused for some distance with, or running closely along, R+M (apomorphy; as currently documented, shared with all Diaphanopterodea except Sinodiaphidae, in which CuA is connected with M by a short cross-vein, and Diaphanopteridae, in which the connection of CuA and M is very brief); first cross-vein in the CuA-CuP area very short and oblique, with CuA displaying a clear inflexion at the point of connection with this cross-vein [also present in Diaphanopteridae, Carrizodiaphanoptera and, to some extent, Elmoidae (Fig. 3); possibly an apomorphy of the entire Diaphanopterodea -except for Sinodiaphidae, in which this state is primarily absent-, and with presumed secondary losses in various families, such as Martynoviidae]; MP branched (plesiomorphy within Diaphanopterodea); CuP with 1-3 distal branches (plesiomorphy within Diaphanopterodea); developed anal area (putative apomorphy for the family).
Remarks.The combination of (i) an overall rich venation, (ii) a long fusion of CuA with R+M (or, CuA running very close to R+M for some distance), and (iii) a cua-cup cross-vein very short, is generally used to identify members of the Parelmoidae.The Pennsylvanian Diaphanopteridae differ from this family only by lacking character state (ii) (see Béthoux and Nel 2003).Given the polarity of several character states listed as diagnostic of the family, it is not excluded that this taxonomic concept might represent a paraphyletic entity, to include other Diaphanopterodea families, such as Elmoidae.Carpenter (1992) considered the six genera and 10 species reported from the Obora locality (Permian, Cisuralian, Sakmarian), and originally assigned to the Elmoidae or Parelmoidae (Kukalová-Peck 1974), as of uncertain familial affinities.Addressing aspects of species delimitation, and the systematics of these species, is made difficult by post-depositional deformations this material endured.Nevertheless, wing venation character states they display tend to indicate a placement to the family Parelmoidae as delimited above, in particular the well-developed anal area.Also, despite deformation, this comparatively large sample allows appreciating variation in character states variability (in other words, how the extent of variability varies within families), likely to differ among (and within) the closely related families Diaphanopteridae, Parelmoidae and Elmoidae (see below and Kukalová-Peck and Sinichenkova 1992).
Notably, a distal fork of CuP is common in the Obora material, with some specimens displaying an early fork, and even a 3-branched CuP.The relation between RP and MA is also very variable across the corresponding species, ranging from a complete lack of fusion to a long one.In contrast, species of the genus Parelmoa Carpenter, 1947 (Fig. 2A-D) show more stable venational features (CuP simple; MA and RP connected by a short crossvein).Among other Parelmoidae, the monotypic genera Pseudelmoa Carpenter 1947 (Fig. 2E) and Permuralia Sinichenkova & Kukalová-Peck, 1997 (Fig. 2F) remain similar to Parelmoa spp. in most of their venational features.Also, Elmoa trisecta Tillyard, 1937 (Elmoidae;Fig. 3) shows rather stable venational features, with a consistent occurrence of a (i) distally forked CuP and (ii) a MA distinct from RP, with a cross-vein connecting the two veins shortly after the origin of the latter.The Obora material is therefore unusual in several respects.Instead of attempting to finely resolve relationships between the corresponding taxa, whose variability in many aspects cannot be properly appreciated, we believe it is more sensible to use the Parelmoidae as a broad taxonomic concept, possibly paraphyletic (i.e., a grade), to include the Obora material.Type material.YQZYW 15, part and counterpart.Etymology.Named after the Yangquan city where the Shuiquan Gully locality is located.
Diagnosis.Area between anterior wing margin and R/RA dark; ScP vanishing in the area between anterior margin and R/RA; MA/MP split opposite the RA/RP split (as opposed to MA/MP split well distal of RA/RP split); CuP forked; AA long (as opposed to AA short), ending on posterior margin beyond wing mid-length.
Description.Positive and negative imprints of a right forewing, distal part missing; dark area between anterior margin and R/RA; near wing base, preserved anterior wing margin very oblique, suggestive of the presence of a short portion of ScA distinct from the anterior wing margin (see Fig. 4A); ScP vanishing in the area between the anterior wing margin and RA, just beyond the brief RP-MA connection; stem of R+M convex, with a distinct inflexion opposite the point of separation of R and M (located about 3.6 mm distal from wing base); RA convex, simple and strong, parallel to anterior wing margin; RP posteriorly pectinate, with 3 simple branches preserved; MA/MP split opposite the RA/RP split; MA diverging anteriorly and then shortly connected with RP; MA simple; MP forked distally; short Cu stem visible; CuA diverging anteriorly from Cu, then close and parallel to R+M stem for some distance, suddenly diverging posteriorly, slightly basal to the R/M split; CuA simple; CuP forked distally; CuA-CuP area narrow until the first cross-vein occurring in this area, which is short and strong, located slightly distal to MA/MP split; anal area very well-developed, with a total of eight terminal branches (anterior-most branch, presumably AA, with 3 terminal branches); cross-veins difficult to observe, evenly distributed over the whole wing, forming two gradate series.
Systematic placement.The presence of a very short cua-cup cross-vein allows assigning the new specimen to the Diaphanopterodea and, within this taxon, allows excluding affinities with the Sinodiaphidae.Furthermore, the derived state 'long fusion of CuA with R+M (or, CuA running very close to R+M for some distance)' allows excluding the new material from the Diaphanopteridae.Then, an assignment to the Parelmoidae as delimited above is straightforward.Nevertheless, a possible assignment to the family Elmoidae was also considered.Two main character states allow distinguishing members of this family (Fig. 3) from the Parelmoidae (Fig. 2), namely (1) an anal area narrow, with a simple AA (Fig. 3A, B, C, E), sometimes with a short AP-like vein (Fig. 3D, F) (as opposed to a well-developed anal area, with numerous long veins, in Parelmoidae); and (2) ScP terminating on RA near the basal third of the wing length (as opposed to distal to the wing mid-length, in Parelmoidae).Even though the termination of ScP is not visible in the new material, it is clearly not directed towards RA at the point where it vanishes, which is already distal to the point where it reaches RA in Elmoidae.The assignment of our new material to the Elmoidae can therefore be excluded.
Within the Parelmoidae, the extent of the AA area is a useful character to consider first.This area is distinctively long in the new material, a state shared with Elmodiapha (see Kukalová-Peck 1974, text-figs 1, 2), from Obora (Czech Republic; Cisuralian, Sakmarian), and Parelmoa (Fig. 2A-D) and Pseudelmoa (Fig. 2E), from the Elmo & Midco localities (Kansas, Oklahoma, USA; Cisuralian, Artinskian).However, the MA/MP split is located well distal of the RA/RP split in these three genera, whereas they are at the same level in new material.The new material further differs from Parelmoa, Pseudelmoa and Elmodiapha by the brief connection of RP and MA (the two veins are distinct in the three genera) and the presence of a forked CuP (this vein is simple in the three genera).Incidentally, in addition to differences in the extent of the AA area, the same characters allow excluding affinities with the genus Permuralia (Fig. 2F; Chekarda, Russian Federation; Cisuralian, Kungurian; and see Kukalová-Peck and Sinichenkova 1992), in which RP and MA are fused for a more or less long distance, and CuP is simple.
Permelmoa magnifica Prokop & Nel, 2011, from the Lodève locality (France; Cisuralian, Kungurian), is also currently assigned to Parelmoidae.To better assess the affinities of the new material we carried out new observations of the material of this species (Fig. 5; and see Yang et al. 2024).In contrast with Prokop and Nel (2011), we observed that (i) RP has four branches (as opposed to three); that (ii) CuA runs close to (or is fused with) R+M for a short distance, and then sharply diverge posteriorly from the R+M stem, just basal of the first split of this stem, a character generally occurring in Parelmoidae (Fig. 2) and Elmoidae (Fig. 3) (and see Carpenter 1943Carpenter , 1947Carpenter , 1992)); that (iii) the first fork of MP is located distally [the basal portion of the stem regarded by Prokop and Nel (2011) as the first posterior stem of MP (long dotted line on Fig. 5A) is herein regarded as a twin, shifted impression of CuA, because (i) the apical part of the wing is preserved as a shifted impression, on a layer different from the rest of the wing (Fig. 5A), (ii) the elevation of this vein portion is inconsistent with an assignment to MP and (iii) such twin imprints have already been documented for Lodève material (Béthoux et al. 2007: p. 185;and O.B. pers. obs.) but also from other fossil localities (e.g., see Béthoux 2015); it may be the consequence of delamination and then shifting of the two epidermic layers composing the wing, or of multiple impressions implying a release from the  sediment, displacement, and second impression of a single wing, a phenomenon yet to be demonstrated experimentally], this being consistent with previous reports on the wing morphology of Parelmoidae and Elmoidae (Figs 2, 3;and see Kukalová-Peck 1974); and (iv) despite a very incomplete preservation, it can be assessed that CuA and CuP, distal to their respective origins, approximate each other before departing, and are therefore most likely connected by a short, oblique cua-cup cross-vein known in Diaphanopteridae, Parelmoidae and Elmoidae (see above).Additionally, we propose homology conjectures alternative to those followed by Prokop and Nel (2011) regarding the MA/ MP split.These authors adopted a traditional interpretation (Fig. 5B) involving a free stem of M splitting into MA and MP near the origin of RP.However, this implies the presence of a very strong, oblique crossvein between RA and RP (* on Fig. 5B), unknown in other Parelmoidea and Elmoidae.A possible alternative interpretation (Fig. 5C) predicts that the MA/MP split occurs at the point where R and M diverge, and that MA runs fused with R for some distance.The 'strong oblique cross-vein' can then be interpreted as the base of RP (* in Fig. 5C).Incidentally, RP and MA are then connected by a strong cross-vein, as is commonly the case in Parelmoidae and Elmoidae.However, the first cross-vein in the MA-MP area is then located in a more basal position than is usually the case in these families, but this can be legitimately related to the more basal position of the MA/MP split.It must be emphasized that a R+MA common stem has already been advocated for Permuralia sharovi (Kukalová-Peck and Sinichenkova 1992) (although a free base of MA is still present) and is admitted for several other members of the order Diaphanopterodea (see Prokop and Kukalová-Peck 2017: text-figs 4, 6;andin Asthenohymenidae Tillyard, 1924 andMartynoviidae Tillyard, 1932).Following this interpretation, and in conjunction with a very long ScP, Permelmoa magnifica stands out as a very unique Parelmoidae.Regardless of the favoured interpretation on the course of MA, the new material differs from Permelmoa magnifica in many respects, including the respective position of the RA/RP and MA/MP split, the extent of ScP, and the extent of the AA area.
In summary, it is legitimate to erect a new genus and species for the new material.

Discussion
Thanks to its good preservation, the material of Sinoelmoa yangquanensis gen.et sp.nov.allows addressing some uncertainty of the course of main veins near the wing base in Diaphanopterodea.Except for the Sinodiaphidae and Diaphanopteridae, the area between the Cu stem and R+M is very narrow in these insects; and, concurrently, the distal free portion of CuA clearly diverge in the close vicinity of the split of R+M (into R and M).Up to now, this situation made it difficult to clearly assess whether the entire stem of Cu, or CuA only, fuses with R+M (and, if so, at which point the (R+M)+CuA fusion takes place).In the newly described specimen a Cu stem independent from R+M is clearly visible; and a simple CuA diverges from it, runs along R+M for some distance, and then diverges abruptly opposite the bending of R+M, just basal of the R/M split (Fig. 4B).This new observation overturns the assumption of the occurrence of a R+M+Cu/CuA early common stem (or, of a CuA/CuP split located at the wing base) in these insects, and instead corroborates previous observations of a more or less brief connection of CuA with R+M (or M alone; see Kukalová-Peck 1974: fig. 10;Kukalová-Peck and Sinichenkova 1992).Furthermore, it suggests that the CuA/CuP split being located opposite the point where CuP patently diverges posteriorly is a general feature of Diaphanopterodea (except for Sinodiaphidae and Diaphanopteridae), and is likely the case in more specialized families in which R+M and Cu cannot be easily distinguished, such as Martynoviidae and Asthenohymenidae.It can be reasonably assumed that it is also the case in the megasecopteran family Protohymenidae, which acquired an 'Astenohymenidae' habitus convergently.
The discovery of Sinoelmoa yangquanensis gen.et sp.nov.has also relevance regarding the age and distribution of the Parelmoidae.Being Asselian in age, it composes the earliest occurrence of the family, but also the most oriental one, along the eastern margin of the Palaeothetys, on the North China Block.This new record suggests that these rather infrequent insects may have had a large distribution, at least along the lower latitudes of the Northern Hemisphere.
of Lodève, Lodève) for allowing the loan of the specimen Ld LAP 365, and to Anne-Laure Decombeix and Mélanie Debiais-Thibaud (UMR AMAP, Monptellier) for managing its transfer to the MNHN; to Anastasia Felker and Nina Dmitrievna (Paleontological Institute, Moscow) for providing photos of the specimen PIN 1700/492; to the Museum of Comparative Zoology and Harvard University for allowing the reproduction of copyrighted material; to A. Lethiers (CR2P) for providing the palaeogeographic map of early Permian (Cisuralian); to the Museum für Naturkunde Berlin for the publication free waiver.This work was supported by the Science and Technology Program of Guangzhou (202102080271), the National Natural Science Foundation of China (No. 42272004, No. 32020103006).

Figure 4 .
Figure 4. Sinoelmoa yangquanensis gen.et sp.nov., holotype, specimen YQZYW 15. A. Overview, drawing and photograph (dryeth-eth composite; dashed line delimiting the area missing on side a) of right forewing.B. Detail of the course of CuA, drawing and photograph (eth-eth-dry composite), as located in A.

Figure 5 .
Figure 5. Wing venation of Permelmoa magnifica Prokop & Nel, 2011 (Parelmoidae Rohdendorf, 1962), holotype specimen Ld LAP 365, right forewing.A. Interpretative drawing under interpretation favoured herein (dotted line, twin imprints of vein sections; and see text), photograph (RTI extract), and normals visualization (RTI extract).B, C. Detail of the radial and median systems, as located in A, under the interpretation followed by Prokop and Nel (2011) (B) and the interpretation favoured herein (C).