Two new species of Sinosmylites Hong (Neuroptera, Berothidae) from the Middle Jurassic of China, with notes on Mesoberothidae

Abstract Two new species of the genus Sinosmylites Hong are described from the Middle Jurassic locality at Daohugou (Inner Mongolia, China): Sinosmylites fumosus sp. n. and Sinosmylites rasnitsyni sp. n. This is the oldest known occurrence of the family Berothidae. The berothid affinity of this genus is confirmed by examination of the hind wing venation characteristic of the family. The Late Triassic family Mesoberothidae may represent an early group of Berothidae.


Introduction
Today, the Berothidae (including Rhachiberothinae) is a small neuropteran family comprised of about 100 species discontinuously distributed mainly across tropical and warm-temperate regions of the world (Aspöck 1986;Aspöck and Nemeschkal 1998). Their larvae are thought to be associated with termite nests, where they feed on termites (e.g., Tauber et al. 2003); however, this habit is only documented in the North American genus Lomamyia Banks (Johnson and Hagen 1981;Faulkner 1992).
Currently, 34 fossil berothid species have been described from various localities (listed in Table 1; others are described but unnamed, only illustrated or represented by larvae). The family was previously only known tentatively from the Jurassic: Grimaldi and Engel (2005) and Engel and Grimaldi (2008) considered the Jurassic / Early Cretaceous Mesithonidae as possible primitive Berothidae or 'stem-group berothids'. Formerly, berothids were believed to be a very ancient family, even when it almost entirely lacked a known fossil record. Tillyard (1932) stated, characteristically: "on the totality of characters, it would now appear as if the Berothidae are the oldest existing family of Planipennia [=Neuroptera]." (p. 29). Here, we describe two new species of the genus Sinosmylites Hong, 1983 from the Middle Jurassic of China, which we confidently assign to the family Berothidae based on characters of the hind wing. We compare the forewing venation of this genus and those of the Late Triassic Mesoberothidae, and by their similarities confirm the great antiquity of the berothid lineage.

Material and methods
This study is based on three specimens collected from Daohugou Village (Shantou Township, Ningcheng County, Inner Mongolia, China) and housed in the Key Laboratory of Insect Evolution and Environmental Changes, College of Life Sciences, Capital Normal University, Beijing, China (CNUB; Dong Ren, curator). These insectbearing beds are here considered as belonging to the Jiulongshan Formation and are dated Bathonian, Middle Jurassic (Gao and Ren 2006).
Specimens were examined using a Leica MZ12.5 dissecting microscope; line drawings were prepared with CorelDraw 12 graphics software with the aid of Adobe Photoshop; photographed by a Nikon SMZ1000 stereomicroscope.
Venational terminology principally follows Comstock (1918) as modified by Oswald (1993) and Archibald and Makarkin (2006). Berothid wings possess crossvenation basically similar to that of Hemerobiidae, arranged in four (incomplete) gradate series in both families. Consequently, we adopt Oswald's (1993) designation of crossveins: principal crossveins are designated by the longitudinal veins which they connect and numbered by the gradate series to which they belong in sequence from the wing base, e.g., 1a1-a2, the crossvein connecting 1A and 2A in the first gradate series; 2m-cu, the crossvein connecting M and Cu in the second gradate series; 2icu, the intracubital crossvein (i.e., between CuA and CuP) in the second gradate series; and 4rs1-rs2, the crossvein between Rs1 and Rs2 in the fourth gradate series. Terminology of wing spaces mainly follows Oswald (1993). Abbreviations used in the text and figures are as the follows: 1A-3A, first to third anal veins; CuA, CuP, anterior and posterior branches of the cubital vein (Cu); MA, MP, anterior and posterior branches of the medial vein (M); R1, anterior branch of the radial vein (R); Rs1, most proximal branch of the radial sector (Rs); Rs2, branch of the radial sector located distal to Rs1; Rs3, branch of the radial sector located distal to Rs2; Sc, subcostal vein.

Family Berothidae Handlirsch, 1906
Genus Sinosmylites Hong, 1983, sit. n. http://species-id.net/wiki/Sinosmylites Sinosmylites Hong, 1983: 94, 198 [Osmylitidae]; Ren et al. 1995 Comments. The venation of these two new species is very similar to that of Sinosmylites pectinatus. The latter species is represented by a nearly complete forewing (Hong 1983). Unfortunately, however, it is quite poorly figured, and its type is now lost. Nevertheless, all main features of S. fumosus sp. n. and S. rasnitsyni sp. n. forewings agree well with those confirmed of S. pectinatus, i.e., similar size (length 5.5 mm in S. pectinatus; about 6.5 mm in S. fumosus sp. n.; 6.7 mm in S. rasnitsyni sp. n.), coloration (a single, more or less fuscous color), and the venational character states are as provided in the generic diagnosis. The few differences between the type species and the two new species (e.g., the presence of two 'inner' gradate series, and the CuP twice forked in S. pectinatus) are at most specific.
Hind wing poorly preserved, approximately 6.5 mm long, 2.6 mm wide. Costal space narrow, distally only slightly dilated. Subcostal veinlets simple, rather closely spaced. Sc distally fused with R1 far from wing apex; Sc+R1 with 13 simple veinlets. Subcostal space relatively narrow; no crossveins detected. R1 space broad, dilated basally; two crossveins before fusion of Sc, R1, one after. Rs originating far from wing base, with eight branches, each forked distally 1-3 times except Rs1 which deeply forked four times. Fork of M not detected. MA once forked distally; MP dichotomously branched distally. CuA long, almost parallel to hind margin, its branches poorly preserved. CuA space relatively broad. CuP fragmentary preserved, quite short. Anal veins not preserved. Crossveins posterior to stem of Rs not detected except one distal between MP, CuA (4m-cu).
Type locality and horizon. Daohugou Village, Shantou township, Ningcheng county, Inner Mongolia, China. Jiulongshan Formation, Middle Jurassic. Description. Forewing with broad-rounded apex, about 6.0 mm long (as preserved, estimated complete length about 6.5 mm), 2.6 mm wide. Costal space mod- erately broad, most dilated at proximal 1/5 of wing length. Subcostal veinlets simple, regularly arranged, less closely spaced than in previous species. Sc distally fused with R1 far from wing apex; Sc+R1 with nine veinlets (eight simple, one forked). Subcostal space broad, with two basal crossveins. R1 space nearly as wide as subcostal space; six crossveins before fusion of Sc and R1, one after. Rs with nine pectinate, regularly spaced branches; four proximal-most branches with 2-4 terminal forks, other branches once forked. Rs1 originating at some distance from origin of Rs. M not fused basally; forked much distal to origin of Rs1. MA, MP almost parallel, distally with one (simple) , two (one simple) branches respectively. Cu divided into CuA, CuP proximal to origin of Rs. CuA pectinate, with 7 branches; proximal-most branch once forked. CuP once deeply forked. Anal veins incompletely preserved; 1A with single marginal fork; 2A with two marginal short branches; 3A very incomplete, with single fork preserved. Four gradate series of crossveins posterior to stem of Rs, all incomplete. First series consists of three crossveins: 1r-m (located at origin of Rs), 1m-cu, 1a1-a2 (longer than previous); Second series includes two crossveins: 2m-cu (connecting MP, CuA), 2icu (connecting CuA, anterior branch of CuP). Third ('inner') series with six crossveins (3rs-rs7, 3rs5-rs4 to 3rs2-rs1; two between Rs3, Rs2). Fourth ('outer') series with five crossveins (from 4rs2-rs1 to 4m-cu; two between Rs1, MA). Wing one color, fuscous. Veins mainly dark brown as preserved. Description. Hind wing approximately 6.5 mm long, 2.7 mm wide. Humeral lobe not extended; frenulum poorly-developed consisting of few bristles. Costal space narrow, dilated towards apex. Subcostal veinlets simple, more closely-spaced apically. Sc distally fused with R1 far from wing apex; Sc+R1 with seven long veinlets (one forked). Subcostal space relatively broad, with one basal crossveins. R1 space nearly as wide as subcostal space; four crossveins before fusion of Sc, R1. Rs with seven pectinate, regularly spaced branches; one branch deeply forked. Rs1 originating at some distance from origin of Rs. Proximal crossvein m-r long, connecting Rs1 near its origin with M. M forked distal to origin of Rs1. MA, MP almost parallel, distally with few branches. CuA long, slightly incurved, in general parallel to hind margin, with nine forkes branches originated at angle >45 degrees, one simple branch. CuP short, with two branched preserved. 1A-3A not preserved. Crossvein between CuA, 1A (or CuP). One crossvein between Rs, Rs6 in 'inner' gradate series (possibly anomalous). Six crossveins (from Rs4 to CuA) in 'outer' gradate series preserved. Wing one color, fuscous. Veins appear mainly dark brown.
Comments. The venation of this hind wing is typical for Berothidae. In particular, the configuration of CuA is characteristic of this family; although this also occurs in the Nevrorthidae, nevrorthid venation is otherwise dissimilar. Also, the basal crossvein between R and M systems is straight, perpendicular to connecting veins; this is characteristic of all Berothidae except Rhachiberothinae. In the vast majority of extant Berothidae, the complete CuP is lost, but the basal or/and distal parts there are often present. CuP is entirely lost in some genera, both fossil (e.g., Microberotha macculloughi Archibald and Makarkin, 2004) and extant (e.g., Cyrenoberotha MacLeod and Adams, 1968, Berlekrumyia Aspöck and Aspöck, 1988. Therefore, it is hard to determine which vein is preserved in this hind wing, the distal part of CuP or 1A (see Fig. 5B, labeled CuP), as the proximal portion of the wing is not preserved. We tentatively consider this vein to be CuP.
This specimen is tentatively assigned to Sinosmylites. The hind wings of Sinosmylites rasnitsyni sp. n., the only species of the genus in which these are known, are quite poorly preserved and their venation does not enable its generic character states to be determined with confidence. However, this hind wing shares similar size, coloration, and venation (e.g., Sc and R1 are fused; Sc+R1 with many branches; several crossveins between R1 and Rs; the same configuration of the Rs branches) with the forewings of Sinosmylites species. Therefore, this generic affinity is most likely.

Discussion
As the type species of the genus Sinosmylites is represented by a single forewing, its family affinity cannot be confidently determined. Its venation is more or less similar to that of such families as Berothidae, Sisyridae, Mesoberothidae, Archeosmylidae, and Prohemerobiidae, all except the first two are extinct and poorly understood. Their forewing venations show superficial similarities in the structure of the costal space (narrowed basally), Sc fused with R1 (only convergent distally but not fused in Prohemerobiidae), similar (in general) branching of Rs, M, and Cu, and their sparse cross-venation. One feature occurs rarely (if at all) in these taxa, i.e., the strongly pectinate CuA, complicating family determination. Therefore, based solely on the forewing, it may be theoretically associated with most of these families, at least provisionally. Fortunately, one of two new species described herein bears its hind wing, although poorly preserved. Its structure indicates that the berothid affinity of Sinosmylites is most probable, as its general venation does not conflict with that of Berothidae, and the presence of the long CuA running nearly parallel to the hind margin characteristic of the family. Moreover, the berothid affinity of a better preserved hind wing ("Sinosmylites sp.") is doubtless, as all of its character states are characteristic only of the Berothidae.
The forewing venation of Sinosmylites differs rather greatly from that of the vast majority of extant (advanced) genera of Berothidae. Particularly, M is forked distinctly more distally than in most berothid genera (including Cretaceous genera: see Engel and Grimaldi 2008), and CuP is strongly pectinate as rarely occurs in Berothidae. It appears most closely related to an Early Cretaceous genus from the Purbeck Limestone Group, which is represented by two nearly complete forewings (Jepson et al. submitted). These two genera share common venational character states (including a pectinate CuA and distal forking of M), but the Purbeck genus is much smaller (forewing length 3.7-3.8 mm) and in general has simpler venation. Sinosmylites is quite similar also to Banoberotha enigmatica Whalley, 1980 from the Early Cretaceous of Lebanese amber by the very similar outline of the costal space, simple veinlets, and M forked much distally to the origin of Rs, but otherwise their venation is different.
The hind wing venation of "Sinosmylites sp." appears amazingly modern. Even if our generic attribution turns out to be incorrect, its berothid family affinity is doubtless.
Triassic berothid-like taxa have been treated as belonging to the family Mesoberothidae. This taxon was created by Riek (1955) as the family Proberothidae for two genera from the Late Triassic Mount Crosby Formation in Australia, Proberotha Riek, 1955 andProberothella Riek, 1955. The name Proberothidae was later replaced with Mesoberothidae by Carpenter (1991) as its type genus turned out to be a junior homonym of Proberotha Krüger, 1923. The family has never been revised, with Riek's two genera its only members (Jell 2004). It has remained, as Carpenter (1992: p. 349) noted, a "little-known family".
Mesoberotha is represented by a single forewing specimen of M. superba Riek, 1955, whose venation is similar to that of Sinosmylites (especially S. fumatus sp. n., known from a better preserved forewing), i.e., the costal space is similarly constructed, narrowed basally; Sc and R1 are fused; the subcostal space is relatively broad; few crossveins are present, most of which are arranged in a gradate series of crossveins; M, Cu, CuA and CuP are branched in a similar manner; 1A with only marginal (shallow) branches. As mentioned above, this genus cannot be assigned to a particular family with confidence based only on characters of the forewing. However, the similarity of the venation between Sinosmylites and Mesoberotha is so distinct as to strongly suggest that Mesoberotha may belong to Berothidae, and that Mesoberothidae is, therefore, a synonym of Berothidae. However, it is necessary to find and examine the Mesoberotha hind wing in order to test this hypothesis. Riek (1955) believed that the family Mesoberothidae "seems to be directly ancestral to the Berothidae." (p. 674). Similarly, Mesoberothidae was referred by Engel and Grimaldi (2008) to the epifamily Mantispoidae (or 'dilarid clade') of the superfamily Hemerobioidea, which also contains Berothidae, Rhachiberothidae, Mantispidae, and Dilaridae. Haring and Aspöck (2004) stated that "the Berothidae are the sister group of the ithonid clade + (Mantispidae + (Chrysopidae + Hemerobiidae))" (p. 427). The oldest known taxon of the latter group is the mantispid genus Liassochrysa Ansorge and Schlüter, 1990 from the Early Toarcian of Dobbertin (Germany) (Wedmann and Makarkin 2007), and, therefore, the Berothidae must be at least of the same Early Jurassic age. Estimates of divergences times based on molecular analysis indicate that the clade consisting of Berothidae and Mantispidae arose during the Early Triassic (Winterton et al. 2010). Consequently, the Late Triassic Mesoberothidae may well represent an early group of Berothidae.
'Archeosmylus' ?costalis Riek, 1955 from the same Australian locality as Mesoberotha probably also belongs to the Mesoberothidae. Its venation differs not sufficiently from that of Mesoberotha superba, and it may possibly belong to this genus. The family affinity of 'Archeosmylus' stigmatus Riek, 1955 is not yet clear.