(= Pterygota Lang, 1888)

(a) Introductory remarks. The concept of higher level structure and phylogeny of the subclass is essentially new though based in part on that in Rohdendorf & Rasnitsyn (1980). The list of synapomorphies relies additionally on data from Matsuda (1965, 1970, 1976), Kristensen (1975, 1981, 1998), Boudreaux (1979), Hennig (1981), Grodnitsky (1999).

(b) Definition. Very small (ca. 0.3 mm body length) to very large (0.7 m wing span) insects, winged or secondarily wingless, of extremely diverse appearance and habits, thus with characters difficult to be outlined. When wingless and with well developed paracercus, differing from both bristletails (except the Palaeozoic neotenic Dasyleptus [2.1]), and silverfish in being aquatic. The Carboniferous Ramsdelepidion Kukalová-Peck and Carbotriplura Kluge are unknown to be either silverfish or immature winged insects [2.1a]).

(c) Synapomorphies. Head is hypognathous, with mouthparts directed downward and with mandibles moving in plane subparallel to that of occipital foramen (a flight adaptation aiming to bring eyes forward and to save mouthparts in collisions of flying insect with obstacles). Anterior and dorsal tentorial arms are interconnected and join corpotentorium which is formed by interconnected posterior arms. Lateral cervicalia present. Meso- and metathorax modified, each with a pair of wings of pleural (limb) origin [1c], ontogenetically developing from paranota (seemingly lateral extention of tergum). Wing articulating apparatus permits rocking movement at flight and folding movement toward roof-like position at rest. Groundplan flight is functionally four-winged, in-phase, anteromotoric.

Wing membrane is supported by simple or branching veins forming several main vein systems of characteristic structure and position, as follows (Fig. 67). Costal vein (C) lines wing margin (sometimes, most probably secondarily, runs submarginally, anterobasally, and bears short veinlets there), simple (not branching), and convex (running atop of membrane elevation and more strong on upper wing surface comparing lower one). Subcostal vein (SC) is concave (running along depressure in membrane and more strong on lower wing surface), with fore and apic veinlets, apical ones merging with both C and R well before wing apex. Radial vein (R) is strong, convex, forming main longitudinal axis of wing, branching both forward (toward costal wing margin) and backward, with fore branches subapical, not numerous, and with strong, single, neutral or, rather, weakly concave rear branch called radial sector (RS), which starts submedially in fore wing, sub-basally in hind one (apicad of RS base, main radial stock is labeled either R, or R₁; main RS branches are sometimes relabelled as R₂ through R₅). Median vein (M) is equally abundantly branching and neutral or weakly concave, with main branches often called M_1-4 and with sub-basal, convex, weakly branching M₅: this causes M appearance of inverted R (apomorphically M₅ usually merges with CuA and so makes it convex). Cubital vein (Cu) dichotomised sub-basally into anterior CuA and posterior CuP, both concave (unless and until CuA merges with M₅). Anal veins (A) convex, moderately or weakly branching, originating independently or, in part, jointly from common sclerotised base, except free 1A which is sometimes renamed as postcubital (PCu) or empusal (E) vein, with respective re-labelling of following veins as 1A, 2A, etc. (this is justified morphologically but causes taxonomic confusion and therefore is not followed here). Anal area is slightly enlarged in hind wing but plesiomorphically is not rocking down there at wing folding except for small, veinless, posterobasal portion. Groundplan wing bears two concave folds of aerodynamic importance, associated with M and CuP, and the convex posterobasal one which is involved in wing folding at rest.

Groundplan articulating apparatus of the wing can be restored tentatively only as synapomorphy of living Scarabaeones and Gryllones, being unknown in the most primitive Carboniferous groups. Dorsal apparatus consists basally of three sclerites ascending from lowered tergum margin toward wing, viz. anteriormost tegula which is loosely connected with wing fore margin structures, 1st axillary (1Ax) with wide basal articulation and two distal heads, and 3rd axillary (3Ax) with pointed basal articulation to hind notal process. Articulating zone placed distally of 1Ax varies in construction indicating primary appearance of articulatory sclerites as weakly individualised thickening in unevenly sclerotised region. Coupled with the lack of necessary palaeontological data, this makes safest to propose only preliminary hypothesis of multiple independent individualisation of articulating sclerites (basisubcostal, basiradial, median ones, etc.) of varying pattern and shape. Usual pattern is as follows: 1Ax fore head contacts (directly or indirectly) with bases of anterior veins (SC and/or R). Medial part of 1Ax is articulated with 2Ax which in turn connects to medial plates and further to M and Cu. Hind head of 1Ax is articulated with 3Ax which in turn is connected with base of anal veins. Well muscularised 3Ax is responsible for wing folding and for its movement in horizontal plane

Of the above, only 2Ax presents on the ventral surface: its lower layer rests on the head of pleural wing process which forms the main wing support at flight. Other ventral articulating sclerites are represented by the anterior basalare (of pleural origin) and posterior subalare (of tergal and particularly postnotal origin), both muscularised and working jointly as wing depressors, and individually as wing pronator and supinator, respectively. Tergum acts as the wing depressor and levator by lifting wing base when buckles by contraction of dorsal medial longitudinal muscles, and sinks it being depressed by tergo-sternal, tergo-pedal and some tergo-pleural muscles.

The alternative hypothesis by Kukalová-Peck (1978, 1983, 1991) of the intra-wing metamery claims that the wing and its thoracic tergum are formed by 8 uniform metameres each composed of two branched veins, the convex anterior and the concave posterior ones, with the respective wing membrane area, tergal subsclerite, and four articulating sclerites inbetween. The hypothesis is not accepted here because, the first, no developmental data support the internal metamery of any insect segment (Trueman's (1990) hypothesis is of no support claiming that the wing consists of only two and not 8 metameres, and not concerning tergum). The second, the implied plesiomorphy of the regular wing fluting contradicts to both palaeontological and morphological observations (Rasnitsyn 1981, 1998a, and below).

Groundplan structure of the pterothoracic segment can be equally reconstructed as synapomorphy of more advanced Scarabaeones and Gryllones, again because of insufficient knowledge of the Carboniferous fossils. Tergum consists externally of large anterior notum and narrow posterior postnotum, both extending internally as phragmata. Each phragma belongs to both neighbor segments and bears attachments of respective dorsal longitudinal muscles (indirect wing depressors). Groundplan subdivision of notum is debateable (Brodsky 1991). The most common structures (Fig. 68) are two anterolateral and one posteromedial convexities, the latter is commonly but not always correctly described as scutellum (unless to abandon the notion of scutellum as defined by the presence of intrascutellar muscle t13 and absence of tergo-laterophragmal one, t12). Other common, possibly goundplan notal structures are notauli or parapsidal sutures (although commonly used in non-hymenopterous insects, the term parapsidal sutures has been proposed originally for a different structure of aculeate wasps (e.g. Gibson 1985) and has gained wide acceptance there, so the term parapsidal line/suture is worth to be abandoned in favour of the notaulus) which delimit the anterolateral convexities medially, V-shaped sulcus delimiting scutum before it and scutellum behind it, and medial longitudinal sulcus (derivable from moulting break line) which divide scutum (not scutellum) into right and left halves. The oldest fossils, especially among Dictyoneuridea, even being seemingly adult, often show notum similar to immatures in lack of clear sutures except the moulting line. Unless these fossils were really subimago rather than imago, or unless this is due to their paedomorphy, the difference could imply independent gain of the above structure of the notum in Gryllones and various Scarabaeones).

Archetypic pleuron is firmly connected with sternum below, and consists of two subcircular sclerites, anapleuron and katapleuron, which are fused along paracoxal suture (Fig. 69). Both pleural subsclerites are intersected by subvertical pleural sulcus bearing lateral coxal articulation below, wing articulation above, and pleural apodeme at its crossing with paracoxal suture. Pleural parts separated by pleural sulcus and paracoxal sutures called anepisternum (fore upper quarter), katepisternum (fore lower one), an- and katepimeron (hind fore and lower quarters, respectively). Anepisternum is further intersected by anapleural cleft into anepisternum in narrow sense (above) and postepisternum (below). Posterodorsal part of anepisternum s.str. is delimited as basalar sclerite. Triangular trochantine is separated from katepisternum hind margin to provide anterior coxal articulation.

Groundplan sternum (Fig. 69) consists of fore basisternum (preceded by variously separated and not always easily identifiable presternal sclerites). It is followed by the furcasternum which bears two lateral furcal apodemes, and further by spinisternum with single medial apodeme.

Groundplan adult leg lacks styli on basal segments and has tarsus 5-segmented with pretarsal dactylus lost (probably modified into arolium). Evidence of styli and genual segments (Kukalová-Peck 1983, 1991 etc.) needs reconsideration (Rasnitsyn & Novokshonov 1997).

Pregenital abdominal segments lack delimited coxites and eversible sacs, with simple or annulated limb rudiments persisting in many immatures, very rarely in adults (Rasnitsyn & Novokshonov 1997). Female genital segments (8th and 9th, Fig. 70) with gonapophyses working as ovipositor blades, coxites modified and called valvifer 1st and 2nd, respectively, and 8th segment lacking stylus (gonangulum retained). Male genitalia are plesiomorphic (segment 9th with penis, gonapophyses and styli retained, groundplan stylus not modified). Adult paracercus is lost (supposing it being secondarily retained by some adult mayflies from nymphal stage).

True copulation present, with direct sperm transfer from male to female gonopore. Embryo with amniotic cavity closed and each amnion and serosa continuous (pore lost). Development plesiomorphically archemetabolous, i.e. gradual, with growth and associated moulting continuing in flying insect, and with early immature larva-like, with tarsus entire, possibly even fused with tibia, and with rudiments of imaginal structures (wings, genitalia) appearing as external buds during later (nymphal) section of development (Figs. 71, 387, 388).

Groundplan bionomy is supposed as tree-dwelling at all developmental stages, with eggs laid into plant tissue cut by female gonapophyses, and with active stages feeding on sporangia content of gymnosperm trees. Jumps from branch to branch were practicised to shorten way from one terminal group of sporangia to another, and also to escape predatory chelicerates and myriapods. Both wings and immature winglets were used to make jumps more distant and precise due to body attitude control in the air. For further details see Rasnitsyn (1976, 1980, 1981, 1998a) who discuss alternative scenarios as well. The recent skimming hypothesis of the insect flight origin is considered below [2.2e].

Additional synapomorphes of winged insect are presented by Boudreaux (1979).

(d) Range. Latest Early Carboniferous (Namurian A) until the present, worldwide.

(e) System and phylogeny. The higher level structure of the subclass of winged insect is still an area of acute debates (cf. Kristensen 1998, Kukalová-Peck 1998, Rasnitsyn 1998a). The present account starts from the following observations which are considered to be relatively well grounded. The present day winged insects (and this is essentially true for the Mesozoic and Cainzoic Eras as well) form four clear-cut and most probably monophyletic groups. These are the insects with complete metamorphosis (Holometabola, Endopterygota, or Oligoneoptera), psocopteroid-hemipteroid assemblage (Paraneoptera), orthopteroids in the widest sense (Polyneoptera), and mayflies plus dragonflies (Hydropalaeoptera or Subulicornes). They appear below under the typified names Scarabaeiformes, Cimiciformes, Gryllones, and Libelluliformes, respectively. Their synapomorphies are listed in respective chapters [2.2.1.3c, 2.2.1.2c, 2.2.2c, and 2.2.1.1c]: in short, Scarabaeiformes can be characterised by their complete metamorphosis, Cimiciformes - by the mouthparts with detached, rod- or stylet-like lacinia, Gryllones - by the wings folded flat on their abdomen at rest and with a wider hind wing area rocked down in the rest position along the line running anterior of 2A. Characteristic of Libelluliformes are the wings permanently held in near flight position (either outstretched or raised over abdomen), with RS starting free from near wing base, and with highly pronounced fluting of the wing blade, so as the convex and concave veins alternate within each RS, MA, MP, and CuA vein systems. In fact, many cladistically oriented students tend to consider mayflies as the sister group of all other winged insects including dragonflies. However, the similarity of mayfly and dragonfly wings, and particularly of their most plesiomorphic Carboniferous representatives, Syntonopterida (see Figs. 85, 86) and Eogeropteridae, (see Figs. 104, 105), is so deep, and their putative synapomorphies are so unique, that the monophyly of Libelluliformes appears to be the best grounded hypothesis.

The picture changes considerably when we deepen into the Palaeozoic. The only new, large and more or less clear-cut group found there is the palaeodictyopteroid assemblage (Protorhynchota: here Dictyoneuridea), which is characterised by the piercing beak and the wing blade fluted in the way different from that in Libelluliformes. The entire vein systems RS, MA, MP, and CuA alternate there in being either convex or concave [2.2.1.2.3].

Besides the palaeodictyopteroids, in the Palaeozoic we can see a wealth of extinct groups of different rank. Most of them run smoothly to various well known orders. Some others merit creating orders of their own, and nevertheless their affinities appear more or less apparent. Palaeomanteida (miomopterans) probably represent the stem group of holometabolous insects [2.2.1.3.1], Jurinida (glosselytrodeans) are a neuropteroid side branch [2.2.1.3.3.4], Hypoperlida occupy a position in the roots of both psocidean-rhynchotan and palaeodictyopteran assemblages [2.2.1.2.2], Caloneurida and Blattinipseida occur still more basally in the roots of the whole clade of Cimiciformes + Scarabaeiformes [2.2.1.2.1]. Numerous are the groups of obscure affinities as well. This particularly concerns the Carboniferous insect assemblage, in part because that time the basic insect subclades were not clear-cut yet (cf. the concept of taxa maturation, Rasnitsyn 1996). The main reason, however, is that the Carboniferous stage is still the least known in the insect history, because the majority of descriptions were made in the 19th and early 20th centuries and have never been revised since that. The present attempt to redescribe this material at least in part (Rasnitsyn in preparation) has made possible to reach some new results in understanding of the gross pterygote system and phylogeny.

It was mentioned above that the five main groups of the pterygotes become three while deepening into Palaeozoic, because at the level of the less advanced orders Palaeomanteida, Hypoperlida, Diaphanopterida, Blattinopseida, Caloneurida, the holometabolous, paraneopteran ('psoco-rhynchotan') and palaeodictyopteran clades become very close to each other. This large clade is probably synapomorphic in the cryptosterny: their pterothoracic sterna are charactertistically invaginated along the midventral line called discrimen (Fig. 69), with the furcal arms mounting a common base elevated inside the thorax. Hence, the three main pterygote groups are now Libelluliformes (may- and dragonflies), Gryllones (orthopteroids s.l.), and the cryptosternous assemblage. These are our main reference points in the following considerations.

There is a recent alternative hypothesis claiming holometabolans to be monophyletic with Gryllones rather than with Cimiciformes (Shcherbakov 1999). This is based in part on a superficial venational similarity of some miomopterans and the aberrant roach family Nocticolidae. Otherwise the hypothesis relies on similarity in the characters clearly autapomorphic within at least one of the taxa under comparison. E.g., the incipient cryptosterny is observed in some grylloblattidans which is similar to the hypothetical unrecorded ancestral stage passed by some remote cimiciform and holometabolan ancestor. Another example is the long, thin, flexible ovipositor of snakeflies and grylloblattidan family Sojanoraphidiidae which is highly simplified morphologically at least in Raphidiida. There are arguments based on a dissimilarity, with no evidence proposed against possibility of direct transition between the relevant character states. This concerns devices locking fore wings in their rest position that take place on metascutum in holometabolans and on mesopostnotum in hemipterans. There are further similarities and dissimilarities used by Shcherbakov to support his hypothesis. Some of them (including absence of 3rd axillar sclerite from the mayfly wing, embryological similarity between Gryllones and Scarabaeiformes, and some others) have already been considered and rejected (Rasnitsyn 1969, 1976, 1980, 1998a). Others look unlike but need careful examination (e.g. possibility that the cryptosterny is groundplan character state for insect body segment). In general, the present hypothesis of the insect wing origin and insect phylogeny is not considered as seriously shaken yet.

Due to course of study of the Carboniferous fossils, it was noticed that three above groups (Gryllones, Libelluliformes and the cryptosternous assemblage) differ appreciably in the relative space taken by different vein systems over their wing blades. RS was observed to be dominating (occupying greater wing area compared with M and Cu) in at least more basal taxa of holometabolans (Palaeomanteida and Hymenoptera, possibly in the least advanced neuropterans), of the cimiciform assemblage (Diaphanopterida, Blattinopseida and Caloneurida), and in mayflies and dragonflies. Other taxa in the holometabolan and cimiciform assemblages often have M dominating (psocideans, rhynchotans, mecopteroids), or the main vein systems are comparable in their development (many Dictyoneurida), or else they are highly variable in that respect (Hypoperlida). Anyway, Cu is very rarely dominating there, and apparently only in case of the generally modified venation (some Anthracoptilidae, Hypoperlida). In contrast, Cu is often dominant in the gryllonean orders, or Cu and M are subequal in their development, while RS dominates rarely and only in the advanced forms. This is observed in stoneflies but not in their ancestral grylloblattideans (except the advanced or aberrant Tillyardembiidae, Gorokhoviidae, some Sylvaphlebiidae and Blattogryllidae, Fig. 407), and in the specialised Carboniferous Eucaenus Scudder (Eucaenidae, Fig. 363).

The diagnostic character proposed here does not look very impressive being the evident subject of homoplasy. On the other hand, this character state can be identified in the majority of fossils, unlike the more important characters of the body structure, mode of the wing folding, etc., which are rarely available for the Carboniferous fossils because of the prevailed preservation state (isolated wings and poorly preserved bodies). That is why the careful use of this character, in addition to all others available, is considered justified.

RS is observed to be dominating, besides the above mentioned forms, in the following Carboniferous taxa of debatable affinities: Heterologus Carpenter (see Fig. 81), Heterologellus Schmidt, Kelleropteron Brauckmann et Hahn, Limburgina Laurentiaux, Propachytylopsis Laurentiaux-Vieira et Laurentiaux, Anthraconeura Laurentiaux-Vieira et Laurentiaux, Evenka Rasnitsyn (Fig. 67), Klebsiella Meunier, Anthracotremma Scudder (Fig. 79), Megalometer Handlirsch (Fig. 80), Sypharoptera Handlirsch, Emphyloptera Pruvost, Pruvostiella Handlirsch, Boltonaloneura subtilis Bolton (Fig. 117), Sthenarocera Brongniart, Protokollaria Brongniart, Hapaloptera Handlirsch, Herdina Carpenter et Richardson (Fig. 113), Metropator Handlirsch (Fig. 114), Paoliola Handlirsch (Fig. 115), Cymenophlebia Pruvost, Endoiasmus Handlirsch, Geroneura Matthew (Fig. 116). Noteworthy in this list is that some of these fossils display character states absent in Gryllones. The most important is the roof-like wing position demonstrated by Anthracotremma, Megalometer, Sthenarocera and Sypharoptera. Indicative though not decisive is the similarity of Sypharoptera, Emphyloptera, Pruvostiella, Sthenarocera, Geroneura, and Boltonaloneura to the typical Caloneurida in having elongate wings with narrow costal space and simple, straight CuA and CuP. Approaching to this combination of character states, except for the wider wing, are Metropator, Paoliola and, possibly, brachypterous Herdina. These taxa are probably related to the undoubted Caloneurida [2.2.1.2.1.2].

Anthracotremma and Megalometer are of particular importance in that they display a combination of the roof-like wing position and distant pterothoracic coxae that indicates absence of the cryptosterny (Fig. 79). Therefore they should be attributed, at least preliminary, to the hypothetical stem group of both Libelluliformes and the cryptosternous assemblage (Cimici- and Scarabaeiformes). Still more reasons exist to assign Heterologus, Heterologellus, Evenka and maybe also Stygne Handlirsch (Fig. 82) to this group because of their highly plesiomorphic wings, while the rest taxa, that is, Limburgina, Propachytylopsis, Anthraconeura, Protokollaria, Hapaloptera, Cymenophlebia, Endoiasmus might belong everywhere ranging from the above stem group to vicinity of Caloneurida [2.2.1.2.1.2], and the other three in Hypoperlida [2.2.1.2.2].

Cu-dominated wings are characteristic of other Carboniferous fossils: Daldubidae (Grylloblattida, Fig. 397), some cockroaches (Fig. 367), some Paoliidae and their possible relatives (Holasicia vetula Kukalová (Fig. 77), Prototettix Giebel, Schuchertiella Handlirsch, Merlebachia Waterlot), and the entire eoblattid-spanioderid-cacurgid assemblage (Fig. 353-363). In that assemblage, the order Eoblattida [2.2.2.?.1] was proposed as the stem group of the dictyopteran assemblage (superorder Blattidea [2.2.2.1]), basing on the characteristic form of the fore wing anal area (clavus), and on the gryllonean mode of the wing folding evident in Protophasma Brongniart (Fig. 359). However, Protophasma has not really large clavus, while Eoblatta Brongniart (Fig. 353) is very similar venationally to Stenoneura Brongniart (Fig. 354), Eoblattina Bolton (Fig. 355), Ischnoneura Brongniart, Ctenoptilus Lameere, Ischnoneurilla Handlirsch, which differ considerably from typical blattoids in general appearance as well (have subquadrate to elongate pronotum lacking wide paranota and often longer legs) and sometimes lack the typical blattoid clavus (Ischnoneura, Ischnoneurilla). Additionally, these fossils form rather smooth transition to Spanioderidae (via Ctenoptilus and Cacurgus Handlirsch) and Ischnoneuridae Handlirsch, 1906 (= Aetophlebiidae Handlirsch, 1906, = Narkeminidae Storozhenko, 1996, synn. nov., Fig. 356, probably including Protodiamphipnoa Brongniart, Fig. 357, but not Cnemidolestes Brongniart which is an isolated wing of unknown affinity). All of them lack blattoid clavus as well, and otherwise they show little similarity to cockroaches. Evidently, the wide, lancet-like clavus is not a synapomorphy of Eoblattida and Blattidea.

Hind wing of Spanioderidae is described as lacking large anal lobe (Burnham 1986, Carpenter 1992a): this would make them unlikely to belong to true Gryllones. However, Burnham (1983) has overlooked this lobe in the gerarid hind wing (Fig. 360); the same might be true for Spanioderidae as well. The typically gryllonean anal lobe is definitely developed in the possible eoblattid-spanioderid relatives Ischnoneuridae (Storozhenko 1998 as Narkeminidae). Ischnoneuridae (as Narkemidae or Narkeminidae) are usually attributed to Grylloblattida (Rohdendorf, Rasnitsyn 1980; Storozhenko 1998), but their venational similarity to the eoblattid-spanioderid assemblage is much more deep than to Grylloblattida (see above). Still more similar to Eoblattidae and Spanioderidae are Cacurgus and its relatives. This make us possible to consider all the eoblattid-spanioderid-cacurgid-narkeminid assemblage as the stem group of true Gryllones termed the order Eoblattida [2.2.2.?.1].

Besides the above main assemblages with prevailing RS and Cu, there are many Carboniferous fossils that either have M dominating (e.g. many Geraridae, Fig. 360, some Paoliidae, Fig. 78) or with no evidently dominating system (Kochopteron Brauckmann, Pseudofouquea Handlirsch, Ampeliptera Pruvost, Fig. 122, Heterologopsis Brauckmann, Fig. 134, Protoprosbole Laurentiaux, Aenigmatodes Handlirsch and many others). Many of them can be probably related to one or another of the above main assemblages. Geraridae are undoubted Gryllones, Ampeliptera, Protoprosbole and Aenigmatodes most probably belong to Hypoperlidae, Heterologopsis possibly to stem Dictyoneuridae, Evenka with its deeply reduced Cu possibly related to Caloneurida. This permits us to keep the winged insects segregated into two largest groups, Scarabaeones with the typically (plesiomorphically?) large RS, and Gryllones with the typically (plesiomorphically?) large Cu.

Unlike the earlier version of the pterygote system (Rasnitsyn 1980, Rohdendorf & Rasnitsyn 1980), we cannot include Paoliidae and their relatives (the former order Paoliida or Protoptera) into Scarabaeones. In the above publications Paoliida were hypothesised to be ancestral to all other winged insects, but technically they were placed within Scarabaeones because of absence of the gryllonean anal lobe. The present approach makes this problematic, because now we have an additional diagnostic feature if not the alternative synapomorphies of each Scarabaeones and Gryllones, that is, the different vein systems prevailed. In this character, some paoliids are close to Gryllones in having large Cu, but no one can be related to Scarabaeones because of really large RS. At the same time, paoliids, like plesiomorphic Scarabaeones, keep their wings roof-like at rest (Figs. 74, 75, 76), have no foldable anal lobe in the hind hind wing (Fig. 78), and in some cases (Zdenekia Kukalová-Peck, Fig. 78) closely remind some plesiomorphic Scarabaeones (Ampeliptera, Fig. 122, Limburgina, Propachytylopsis, Heterologopsis, Fig. 134, etc.) in form of their fore wing CuA. In contrast, the general body form of the few paoliids known in that respect (Brauckmann 1991 and unpublished, Figs. 74, 75, 76) is similar to that of Protodiamphipnoa, Protophasma, Gerarus (Figs. 357, 359, 360) and some Spanioderidae (Palaeocarria Cockerell, Fig. 363): all of them are somewhat stick-insect-like in being large and long-legged insects with subquadrate to elongate pronotum without wide paranota. In an extent, however, this can be true for some ancient Scarabaeones as well (Megalometer, Heterologopsis, the Early Permian Strephoneura Martynov, Fig. 131).

The general appearance of paoliids as large and heavy, long-legged insects, probably slow both on wings and feet, is in agreement with our earlier hypothesis of the groundplan pterygote habits, that is, feeding on sporangia mounting branchlets of the Carboniferous gymnosperm plants (Rasnitsyn 1980, Rohdendorf, Rasnitsyn 1980), and so it well might be the groundplan character of the entire subclass Scarabaeona (winged insects). Large body size also fits this hypothesis, because it makes short (comparing the body size) primordial winglets aerodynamically effective in attitude control when jumping from one branch to another (Wootton & Ellington 1991). However, long legs might be obstacles for the effective "pro-flight" of ancestral pterygotes and so they could be acquired later in the evolution, after the insects have got some flight skill. Long legs and/or pronotum, that is, a kind of life form of giraffe and megatherium, are characteristic of many archaic winged insects of the Palaeozoic, both of the scarabaeonean and gryllonean affinities (Figs. 74, 75, 76, 80, 357, 359, 362, 363). At least one of them, short-legged but long-necked Eucaenus ovalis Scudder from the Westphalian D of Mazon Creek, Illinois (Fig. 363), is really found to have several lycopod microspores in its guts (Scott & Taylor 1983), though their quantity cannot exclude the possibility that the spores were occasionally and not intentionally ingested y the insects. This implies that these adaptations, and particularly the long-leggedness, might be synapomorphic for all known winged insects.

Taking the above considerations into account, origin of the winged insects is hypothesised here to start from the bristletail-like habits (that is, the less cryptic one than that in silverfish) and to proceed further through feeding on spores as well as on the gymnosperm plant pollen and ovules, both disperse and enclosed in sporangia, either fallen on the ground or attached to the mother plant. The next stage (Fig. 72) is supposed to be feeding on the sporangia content of gymnosperm trees, with incipient using of movable winglets of the paranotal origin in jumps from one group of terminal sporangia to another, and to escape predators, as mentioned above. At both these stages, the insects are supposed to have large size, relatively short legs and long cerci and paracercus. These assumptions are based, the first, on the results by Wootton & Ellington (1991) that in the insect-like object of larger size (6-10 cm long), even small winglets can be aerodynamically effective in jumping/gliding pre-flight if minimally movable (adjustable against air stream), particularly if assisted by long cerci (and paracercus). The second reason is that a creature with permanently outstretched long winglets (not to say normal wings) would be awkward on feet, both on the ground and among vegetation, and hardly possible to escape extinction unless the superior flying ability is gained (Rasnitsyn 1976, 1980, 1981, 1998a). And the third reason is a number of indication that paranota are in part of the limb origin and so they could retain in their morphogenetic repertoire the relevant mechanisms of limb motility invisible but ready to be re-activated using the customary morphogenetic re-arrangement (Tikhomirova 1991).

At the third stage, when the real flight has appeared and wings grew large, hind margins of wings came into contact while at rest. To make possible using narrow spaces as a refuge from predators, wind, rain, etc., it was important to diminish the transversal size of the insect. To this end, the best way is to lay wings one over another. This makes problem, however, for being ready to flight, because the time is necessary to return the wings into flight position. Therefore the overlapping (flat) wing position at rest is not very likely to be developed until the wing articulation apparatus became reasonably sophisticated. Prior to this, when wings grew wider and met each other along their rear margins while in rest position, it is more likely that the line of wing contact was being pushed upward, and the wings took an oblique (roof-like) rest position. This stage of evolution, like the previous ones, is purely hypothetical.

The next stage, the elongation of legs that permitted insects to behave easily in thicket of branches, is supposedly represented by diverse Carboniferous fossils (Figs. 74, 75, 76, 80, 357, 359, 362, 363). This was probably the key point in the pterygote phylogeny, the starting position for Scarabaeones to begin improving the flight ability, and for Gryllones to enhance adaptations for more cryptic existence.

There are numerous alternative hypotheses about the pterygote origins. Earlier ones are considered elsewhere (Rasnitsyn 1980 and references therein). Worth mentioning here is the recent hypothesis of the insect flight originated in an amphibiotic ancestor through the surface skimming on feet using pre-wings as an air propeller (proposed by Marden & Kramer 1994, Marden 1995, reviewed and further developed by Shcherbakov 1999). This hypothesis does not look appropriate because, the first, it contradicts all basic taphonomical observations [3.2.2.1]. The second, to be efficient, the air-forced locomotory system must rely on far advanced engine and propelling agent. So it can develop from the flight system already evolved, rather than vice versa. It does not look occasional that the man designed air-propelled boats similarly appeared using aeroplane motor and propellers, and not as an aeroplane precursor. Equally this makes inappropriate Shcherbakov's (op. cit.) hypothesis of the ancestral wings modelled after mayflies, that is permanently outstretched and used by an ephemeral adult simply to mate and disperse (even if to let alone that the ephemeral and hence morphologically simplified adult is not the best start for the known variety of adult winged insects).

Taxonomically, we consider the subclass of winged insects as composed of the two infraclasses, Scarabaeones and Gryllones, dealt with below [2.2.1., 2.2.2]. According to the basic taxonomic principle of continuum [1.1c] we try and follow the supposed relatedness and similarity and not the particular characters as such. That is why, the paradigmatic Gryllones (that is, superorders Blattidea, Perlidea and Gryllidea) are complemented here with the fossils displaying either the posterobasal hind wing area tucking down along the line before 2A (Geraridae, Eucaenidae, Ischnoneuridae, Protophasmatidae), or those with prevailing CuA and without evident affinity to Scarabaeones (the eoblattid-spanioderid assemblage).

Those with the evident scarabaeonean connections, or with RS prevailing and without the hind wing anal area foldable anterior to 2A (Anthracotremma, Megalometer, Heterologus, Heterologopsis, Heterologellus, Kelleropteron, Ampeliptera, Protoprosbole, Aenigmatodes, Limburgina, Propachytylopsis, Anthraconeura, Protokollaria, Hapaloptera, Cymenophlebia, Endoiasmus, Sypharoptera, Emphyloptera, Pruvostiella, Sthenarocera, Geroneura, Boltonaloneura, Metropator, Paoliola, Herdina, Evenka) are referred to as Scarabaeones. Paoliidae cannot be assigned to any of above because of combinaton of the roof-like wing rest position and supposedly plesiomorphic Cu prevailence. At the same time, it does not seem wise to create one more infraclass for just a small family, so the order Paoliida is retained for it, unplaced within the subclass Scarabaeona - until better understanding of the insect evolution is gained.

Herbstiala, Schuchertiella, Merlebachia, Hadentomum Handlirsch, Klebsiella, Homoeodictyon Martynov (Fig. 73), and many other obscure genera are considered here as Scarabaeona incertae sedis (winged insects of obscure taxonomic position). Of them Klebsiella and Homoeodictyon are of interest because of their convex MA. However the remaining venation has little in common with Dictyoneuridea, so their position in the stem of that group is not very likely though cannot be ruled out entirely.

The system and supposed phylogeny of the winged insects are shown at Fig. 58.