Introduction

The family Osmylidae has a rich fossil history in comparison with other neuropteran families, going back to the Early Jurassic. These oldest osmylids are represented by four genera, Mesosmylina Bode, 1953, Sogjuta Martynova, 1958, Petrushevskia Martynova, 1958, and probably Tetanoptilon Bode, 1953 from Germany, Mongolia, and Kyrgyzstan (Bode 1953; Martynova 1958; Ponomarenko 1984, 1996; Lambkin 1988; Ansorge 1996). Their venation is quite typical for Osmylidae assuming an earlier divergence of the family from its ancestor.

The diversity of osmylids in the Middle—Upper Jurassic was probably highest than in any other period. The Chinese locality Daohugou being of this age has yielded numerous varied osmylids represented by more than one thousand known specimens (QY and VNM, unpublished data). Some of these are similar to extant taxa of different subfamilies (see e.g., Wang et al. 2009b, 2010), while some are very dissimilar, represented by taxa not crossing into the Cretaceous. For example, there were the large and extremely multi-veined species of the extinct subfamily Saucrosmylinae and of another yet undescribed extinct subfamily, as well as many species of the relict subfamily Gumillinae, some of which being minute species with peculiar and reduced venation (Ren and Yin 2003; Ren and Engel 2007; Wang et al. 2009a; Yang et al. 2010). Among this material, a new osmylid genus was discovered whose venation appears to be most “primitive” and somewhat similar to that of Archeosmylidae. Riek (1976: 788) argued that the Late Permian—Triassic family Archeosmylidae “is apparently directly ancestral to Osmylidae”. The study of this new genus probably supports this supposition.

In the present paper, we describe a new genus and species of Osmylidae, discuss its systematic position and phylogenetic implications, analyze the composition of Archeosmylidae and provide its characteristics.

Institutional abbreviations.—CNUB, Capital Normal University, Beijing, China; PIN, Paleontological Institute of the Russian Academy of Sciences, Moscow, Russia.

Other abbreviations.—Venation abbreviations: 1A–3A, anal veins; C, costa; CuA, anterior cubitus; CuP, posterior cubitus; MA and MP, anterior and posterior branches of media (M); ng, nygma; R1, first branch of radius (R); Rs, radial sector; Rs1, most proximal branch of Rs; Sc, subcosta.

Material and methods

This study is based on one specimen collected near Daohugou Village (Shantou Township, Ningcheng County, Inner Mongolia, China). The insect-bearing beds are considered as belonging to the Jiulongshan Formation dated as Bathonian (Middle Jurassic) (Gao and Ren 2006). The map of the Daohugou locality is given in Wang et al. (2009a: fig. 1). The specimen was examined using an MZ12.5 dissecting microscope a Leica, illustrated with the aid of Adobe Photoshop, and photographed with a Nikon SMZ1000.

Systematic paleontology

Discussion and conclusions

Characters of Archaeosmylidia forewing.—The forewing Archaeosmylidia differs from ones of other hitherto known osmylid genera by its “primitive” appearance. This is exhibited in particular by some venational details more characteristic of other families with generalized venation (e.g., Ithonidae and Nevrorthidae) than Osmylidae: (i) the entire absence of zigzag pattern characteristic of most other osmylids including the oldest, Early Jurassic (i.e., the longitudinal veins in Archaeosmylidia are smooth, not zigzagged); (ii) the pectinate branching of CuA, 1A and 2A are somewhat more oblique than in most other osmylids; (iii) some subcostal veinlets are markedly curved (straight or only slightly curved in other osmylids); (iv) the non-pectinate CuP (strongly pectinate in most other osmylids); (v) the crossveins in the radial to medial spaces are scarce. All these features taken together make the forewing venation of this species “primitive”.

The important forewing characters are considered below.

Subcostal crossveins.—Osmylidae have normally one basal crossvein between Sc and R. The numerous subcostal crossveins as found in Archaeosmylidia occur very rarely in this family. The monotypic Australian subfamily Porisminae, with Porismus strigatus (Burmeister, 1839), is probably the only exception among the extant taxa. Also, there are two fossil genera, Osmylochrysa Jepson, Makarkin, and Coram, 2012 from the English Purbeck having widely spaced subcostal crossveins (Jepson et al. 2012), and an undescribed genus from Daohugou which has many closely spaced crossveins between Sc and R/R1 (Chaofan Shi, VNM, QY, and DR, unpublished data). No subcostal crossveins are detected in Archeosmylidae. We presume that the presence of numerous crossveins between Sc and R/R1 is autapomorphic of these four genera because their venation is strongly dissimilar to each other and they likely represent different lineages.

Fig. 1.

Forewing of osmylid neuropteran insect Archaeosmylidia fusca gen. et sp. nov. (Osmylidae) from the Jiulongshan Formation of Daohugou, China. Holotype CNU-NEU-NN2009103. Photograph (A) and line drawing of venation (B). Scale bars 5 mm. Abbreviations: 1 A, 2A, anal veins; C, costa; CuA, anterior cubitus; CuP, posterior cubitus; MA and MP, anterior and posterior branches of media (M); ng, nygma; R1, first branch of radius (R); Rs, radial sector; Sc, subcosta.

The subcostal/R1 spaces ratio.—In the majority of the osmylid genera the subcostal space is narrow, narrower (sometimes much narrower) than the R1 space (between R1 and Rs). In Archaeosmylidia, these spaces are nearly equal in width, as the subcostal space is dilated as compared with that of other osmylids, and the R1 space is narrow. The ratio of these spaces of Archaeosmylidia is similar to that of Archeosmylidae (Fig. 2).

Distal branches of Rs.—These branches in Archaeosmylidia are nearly straight for most of their length or slightly curved. This condition is plesiomorphic as it occurs in the majority of other families, including ones known from the Permian. The distal branches of Rs of the most extant osmylids are sinuous, more or less expressed.

Radial crossveins.—In the vast majority of the extant and extinct Osmylidae, the crossveins in the radial space are numerous and not arranged in gradate series except one outer. The arrangement of the crossveins in the genera of Mesosmylininae and Protosmylinae is rather similar to that of Archaeosmylidia, i.e., relatively scarce and arranged mainly in few gradate series.

Branching of CuA.—There is much variety in this branching observed across the osmylids. CuA is clearly (sometimes strongly) pectinate in the subfamilies Mesosmylininae, Protosmylinae, Saucrosmylinae (some), Gumillinae (many), Spilosmylinae, Porisminae, Osmylinae (some), and clearly not pectinate (sometimes simple with only marginal fork) in the subfamilies Saucrosmylinae (some), Eidoporisminae, Kempyninae, Stenosmylinae, Osmylinae (some). It should be noted that the pectinate branches of CuA originate at an obtuse angle (apomorphic state), differing from Archeosmylidae (see below). The non-pectinate branching of CuA in Osmylidae is probably secondary.

Branching of CuP.—CuP is pectinate (often strongly) in the vast majority of known genera of Osmylidae; branches of CuP originate at obtuse angle. The only other known subfamily with the non-pectinate branching of CuP is the Middle Jurassic Saucrosmylinae. However, this taxon is now regarded as a distinct family, Saucrosmylidae (Wang et al. 2011).

Systematic position of Archaeosmylidia. —This genus can not be assigned to any of the ten known subfamilies of Osmylidae. It is very unlike Osmylinae, Kempyninae, Stenosmylinae, Porisminae, Eidoporisminae, and Spilosmylinae comprising the vast majority of the genera of the Recent fauna. Also, the fossil Saucrosmylinae and Gumillinae represented in Daohugou are very different from it by many derived features. Archaeosmylidia most resembles the genera of Protosmylinae and Mesosmylininae, but it cannot be assigned to any of them. These subfamilies have quite similar forewings (e.g., relatively small with the crossveins in the radial space are arranged mainly in few gradate series), and may be synonyms. The majority of the oldest osmylid genera (the Early Jurassic Mesosmylina, Sogjuta, and Petrushevskia) probably all belong to Mesosmylininae. Protosmylinae are known from the Middle Jurassic to today (Wang et al. 2010). Until the discovery of Archaeosmylidia, these two subfamilies possessed the most “primitive” wings. The protosmyline and mesosmylinine genera clearly differ from Archaeosmylidia in particular by the pectinate CuP and the zigzag vein pattern. It is obvious that this genus represents some basal group, but we refrain from establishing a new subfamily pending detailed phylogenetic analysis of entire group.

Archeosmylidae.—The forewing venation of Archaeosmylidia is rather similar to that of some species of the poorlyknown Late Permian—Triassic family Archeosmylidae. This family was created by Riek (1953) for the monotypic genus Archeosmylus Riek, 1953 based on A. pectinatus Riek, 1953 from the Late Permian of Australian Belmont, Queensland. Later, three additional species of this genus have been identified from the Carnian (Late Triassic) of Australia (i.e., A. stigmatus Riek, 1955 and A. costalis Riek, 1955 from the Mount Crosby Formation of New South Wales, and an unnamed species from the Blackstone Formation of Denmark Hill, Queensland), and two species from the Rhaetian (Late Triassic)-earliest Jurassic of Gloucestershire, England (A. complesus Whalley, 1988 and A. alysius Whalley, 1988) (Riek 1955, 1956; Whalley 1988; Jarzembowski 1999). Ponomarenko (1996) subsequently transferred these two British species to the genus Prohemerobius Handlirsch, 1906, and therefore to Prohemerobiidae. Although their assignments to Prohemerobius are probably incorrect (judging from the examination of the type species of Prohemerobius), these two species are most probably not congeneric with A. pectinatus; their actual generic and family affinities will be clear only after the re-examination of their types. Actual generic affinities of “Archeosmylus” stigmatus and “A.” costalis are also not clear.

One unnamed species from the Middle Beaufort Series (Late Permian) of Natal (South African Republic) was assigned to Archeosmylidae based on a fragmentary wing (Riek 1976: pl. 6: 6).

The only other genus described in this family is the monotypic genus Babykamenia Ponomarenko and Shcherbakov, 2004 from the terminal Permian—basal Triassic Maltsevo Formation of Babii Kamen' in Siberia (Ponomarenko and Shcherbakov 2004). The single species is represented by a crumpled forewing.

The genus Lithosmylidia Riek, 1955 from the Middle to Upper Triassic of Australia most likely can be assigned to Archeosmylidae as well. It is heterogeneous genus, and each species (L. lineata Riek, 1955, L. barrone Lambkin, 1988, and L. parvula Lambkin, 1988) might be generically distinct. Lambkin (1988: 457) determined family affinities of known species to be “species similar to Osmylidae or Polystoechotidae, but not able to be placed in either” for L. barrone and L. parvula, and “probable Polystoechotidae” (Lambkin 1988: 457) for L. lineata. The latter species is the type of the genus; it was subsequently removed from Polystoechotidae (Archibald and Makarkin 2006) and was not included in Ithonidae sensu lato (Winterton and Makarkin 2010). We believe that all three species may belong to the family Archeosmylidae. This implies from the examination of numerous undescribed wings having similar venation from Middle—Upper Triassic of the Madygen Formation, Kyrgyzstan in Central Asia (Fig. 2) (see Shcherbakov 2008 for characteristics of the locality). The venation of these wings is very similar to that of Lithosmylidia (especially to L. lineata) and Babykamenia; these taxa certainly belong to the same family (compare Fig. 2 with Lambkin 1988: fig. 1, and Ponomarenko and Shcherbakov 2004: fig. 3). However, the association of these Triassic taxa with the Late Permian Archeosmulus pectinatus (the type of Archeosmylidae) is not obvious. The main problem is that the holotype of A. pectinatus is represented by an incomplete forewing. Nevertheless, its preserved venation has all main features in common with that of Lithosmylidia, Babykamenia and undescribed taxa from the Madygen Formation (see Riek 1953: fig. 66; Jell 2004: two upper figures on p. 81).

We consider the following described genera as confamilial and belonging to Archeosmylidae: Archeosmylus (only type species A. pectinatus), Babykamenia (one species, Babykamenia eskovi Ponomarenko and Shcherbakov, 2004), and Lithosmylidia (three species: L. lineata, L. barrone, and L. parvula). The forewing of this family is preliminarily characterized as follows: The costal space is similarly constructed to that of Osmylidae, with mainly simple subcostal veinlets; the subcostal crossveins are not detected; Sc and R1 are fused distally; Sc+R1 is smoothly curved running nearly parallel to the costal margin, entering wing margin well before apex; M is forked distal to the origin of Rs; Cu is divided into CuA and CuP near to the wing base; CuA, CuP and 1A are few-branched, often pectinate, but their branches are very oblique and not as strongly pectinate as in Osmylidae; the crossveins in the radial space are scarce, often very rare. The known temporal range of the family is from the Late Permian to Late Triassic.

Fig. 2.

Forewings of undescribed archeosmylid insects from the Madygen Formation, Middle to Upper Triassic, Kyrgyzstan, Central Asia. A. PIN 2785/4081. B. PIN 2555/2096. C. PIN 2069/3604. Scale bars 5 mm. Abbreviations: 1A, 2A, anal veins; CuA, anterior cubitus; CuP, posterior cubitus.

Archeosmylidae are sometimes considered a synonym of Permithonidae (Whalley 1988; Makarkin and Archibald 2003; Engel and Grimaldi 2008), but its validity (as treated here) is unquestioned. Permithonidae is easily differentiated from Archeosmylidae by R1 (or Sc+R1) being distally straight, not running parallel to the costal margin, M profusely branched, often proximad the origin of Rs, and CuP simple or at most with a marginal shallow fork.

Phylogenetic implications.—Osmylidae and Archeosmylidae (as treated herein) are probably closely related families. Their known temporal ranges are not overlapping: Late Permian to the Late Triassic for Archeosmylidae, and the Early Jurassic to today for Osmylidae. They have generally similar venation, but conspicuously differ by the forewing configuration of CuP and/or CuA. These are strongly pectinate with branches that originate at obtuse angle in the vast majority of the osmylid genera and are never strongly pectinate with branches that originate at acute angle in Archeosmylidae. Differing from most other osmylids, Archaeosmylidia and Archeosmylidae share the few-branched CuP, the absence of zigzag vein pattern, and the scarcity of the crossveins in the radial space. It is quite possible that Archaeosmylidia represents a relict of the Triassic—Early Jurassic osmylids that survived into the Middle—Upper Jurassic fauna retaining these and some other “primitive” features. Osmylidae might have originated in the Triassic from some “archeosmylid-like” ancestor, as predicted by Riek (1976). One contradiction to this hypothesis is that the nygmata are not detected yet in Archeosmylidae, but these are present in all osmylids, including Archaeosmylidia. It is possible, however, that the state of preservation of known archeosmylids is not good enough to preserve the nygmata. The presence of these structures is a plesiomorphic state in the order, and their reformation from the membrane is very unlikely. Therefore, a hypothetic ancestor of Osmylidae had to possess the nygmata.