Unusual Structures in the Paleozoic Insect Orders Megasecoptera and Palaeodictyoptera, With a Description of a New Family

The order Megasecoptera is a representative of the haustellate paleopterous insects of the evolutionary line that lived during the Pennsylvanian and the Permian. The similarity of wings and body structures, such as mouth-parts and genitalia, indicate very close, relationship with the order Palaeodictyoptera. Both groups are presumed to have emerged sometime during the Mississippian rom a common ancestor. While Palaeodictyoptera are usually larger and more sturdily built, bearing broad wings with a rich venation and prothoracic lobes, Megasecoptera are slender insects with a more delicate appearance, with petiolate wings and simplified venation, with enlarged thorax lacking prothoracic lobes, and with tapering abdomen. The present paper deals with an extraordinary morphological feature--projections of the body cuticle, which occur in m.ost or all Megasecoptera and at least in some Palaeodictyoptera. These are conspicuous processes, which are short to very long, simple or branched, and which are distributed in regular rows on the abdomen and thorax. A fuller understanding of the morphology of this very unusual character resulted from two years of intensive research by Dr. F. M. Carpenter and myself, based upon tossil material of Commentry (Upper Pennsylvanian, France), Mazon Creek (Middle Pennsylvanian, Illinois), Obora (Lower Permian, Czechoslovakia), Elmo (Lower Permian, Kansas), and now also Tshekarda (Lower Permian, Siberia). I am deeply indebted to Professor Carpenter, who was very helpful in the preparation of this study. Until now, the projections have been only poorly known. They were at first mostly interpreted as tracheal gills that persisted into the


24.
Psyche [September adult stage, or later as sho.rt lateral spines on the abdominal segments and thorax. The following is a short account of present knowledge.
The projections on the prothorax are, in some families, long and filiform, but in other amilies rather short, pointed and spine-like.
The abdominal pro.jections are actually filiform, growing in fringelike rows out of the tergites. However, all previous authors observed only the basal parts of several abdominal projections situated laterally, which led to incorrect interpretations. Thus Brongniart (I885, p. 63; I885, p. 658; I89o, p. I54o) considered them. to be branchio-tracheal appendages, which served for aquatic respiration in nymphs and which were carried over to the adults. In his general account on Carboniferous insects of Commentry, he ga.ve a detailed figure (I893, p. 305, p. 298, fig. 5o) of an enlarged "lateral lamelld' with branched "tracheae" in the genus Corydaloides (Mischopteridae). His point of view was followed by Brauer (I886, p. Io7), who classified the projections as "persistent abdominal tracheal gills.".
The gill character of the projections was denied by Lameere I9O8, p. 136; 1917, p. 28; 1917, p. I45), who compared the "lamellae" with the lateral expansions of the Recent mayfly Oniscigaster wakefieldi (N. Zealand). He regarded the projections protruding out from "lamellae" to be backwardly directed spines.
Ma.rtynov I938, p. 25) characterized Megasecoptera as having "lateral expansions of abdominal segments with tooth-like or spinelike outgrowths, homologous with prothoracic spines and prothoracic winglets o Palaeodictyoptera, reduced and modified". Carpenter (1951, p. 353) correctly stated that the projections were extensions of tergites, but also believed them to be short and spine-like in character (Corydaloididae,195 I,p. 351 I951 ). I found that each o the sufficiently preserved megasecopteran bodies (mostly Mischopteridae) had prolonged filaments leading rom the posterior margin of the ab?dominal tergites. The projections were visible only under glycerin, a medium which was obviously not applied to, the ossils by previous students. It should be noted that in the Mischoptera douglassi nymph the bases of the projections give a perfect spine-like appearance, which now seems to be due to incomplete preservation. Recently, Carpenter and Richardson (I97 t) described long filamentous projections in Eubrodia dabasinskasi (Brodiidae) extending posteriorly along the mesotho,rax to almost the end of the body. Summarizing our present knowledge, we can say that the processes or projections are hollow outgrowths of the tergites and are usually arranged into regular transverse rows, are simple or branched, and are sho.rt to very long, according to the particular families. The outgrowths are directed up and backwards from the body, so that they protrude. The ventral side of the projection-bearing bodies is not known. On the thorax, the projections may form spines, or may be filiform, identical to those on the abdomen. The abdominal projections with their superimposed series of fringes, resemble the traditional skirt o the Spanish national emale costume. It is possible that all species of Megasecoptera possessed projections, more or less developed, both adults and nymphs. Projections of the same character occurred in some Palaeodictyoptera, but probably not in all families.

Psyche September
It has taken almost 80 years for acquisition of the above data about these two groups of insects, though they are not rare in Paleozoic deposits. This slow process becomes more understandable after considering the character and nature of preservation of the projections.. Protruding above the body in life, they tended to become hidden by the superimposed sediment rather than become compressed at the same level as the body itself during fossilization. The broken and usually more sclerotized bases of the projections are not distinguishable from spines or tubercles and are mostly inconspicuous. Finally, in a matrix which does not preserve chitin, the imprints of the projections are vague.
During my study, the projections were thoroughly examined for connections with the insect's body. In the matrix capable of preserving chitin (e.g. Commentry shale), the projections have the same brown color as the terga. Their surface is covered with a rugosity similar to that on the body (Mazon Creek, Illinois) or with a minor rugosity and scattered sockets of setae (Obora, Moravia). In Monsteropterum moravicum, the surface of the projections is identical to that of the legs. The arrangement of the projections is regular and probably characteristic for all genera within a family (I have found this to be true so far for Bardohymenidae, Protohymenidae and Mischopteridae). The width of the projections varies with the size of specimens; their arrangement is bilaterally symmetrical. The above mentioned features exclude the possibility that the outgrowths are parasitic organisms, or fungi. These projections in the Megasecoptera and Palaeodictyoptera appear unique among insect orders, and their function remains obscure. However, several features suggest that they might be homologous to certain tergal structures of Odonata. In all Recent Odonata, there is a transverse ridge at each end of the tergum, the anterior and posterior transverse carinae (Walker 953, p. 18). The former is inconspicuous, but the posterior carina is a distinct ridge bearing a row of small tubercles or denticles. By the position and arrangement in rows, the projections in Megasecoptera and Palaeodictyoptera are very suggestive of the prolonged and enlarged carinal denticles of Odonata. Their function, of course, presents a complicated problem, which can hardly be solved with fossil material.  (Carpenter,962), which turns out to be true also for the body. However, the wing venation is less advanced, possessing an MA which is not anastomosed with Rs, and a CuA free from M. Also the general form of the wings is less specialized, as the hind wings are almost equal to the fore wings, not reduced in length as in the Protohymenidae. The body in both families is much alike, possessing a large thorax and tapered abdomen. The bardohymenid body is, in relation to the wings, more heavy. Through the courtesy of Dr. Carpenter I was able to study Protohymen readi Carpenter (933), Protohmen elongatus Carpenter (93o) and Pvotohymen permianus Tillyard (I924) for projections. In all three specimens the bases of projections are present and very similarily distributed as in Bardo.hymenidae. Ho,wever, they merge with the uneven surface of the rock to such an extent that they would be undetectable unless a well preserved specimen, such as the type S. sibiricus n.sp., were available for comparison. By delicate preparation of the surrounding matrix I was able to uncover remnants projections (P. readi, specimen 3257, Museum of Comp. Zoology, Harvard University; P. lermianus, specimen 5o53, Peabody Museum Yale University), which are prolonged and backwardly curved. This a.ct is very significant, because in 8ylvohymen sibiricus n.sp. (Bardohymenidae) the projections continue into the covering matrix and cannot be followed.
The projections in Bardohymenidae and Protohymenidae are arranged in transverse rows. By position and distribution they are very similar to denticles in the transverse carinae of Odonata. In my opinion, these structures may be homologous. Besides, some anisopteran nymphs (for instance ErletogomIhus designatus, Gomphidae) Needham & Westfall, 1955, bear, on several abdominal terga, paired darker pits, loca.ted along the median line precisely like the bases the large paired projections in Bardohymenidae and Protohymenidae. This similarity is suggestive of possible musculature inside the paired projections in Megasecoptera.
Genera included Bardohymen G. Zalessky, I937 (Lower Permian, Barda River, U.S.S.R.) 8lv'ohmen Martynov, 1941  Genus 8ylvohymen Martynov Tylohymen Martynov, 1941: 10;Carpenter, 1947: 31;Carpenter, 1962: 37; Rohdendorf, 1962: 68. Type species: Sylohymen robustus Martynov, 1938 This genus is based upon a distal part o. a wing rom the same locality (Tshekarda) as the presently described specimen o.f Sibiricus, n.sp., which is much more fully preserved. Carpenter (1946, p. 31, fig. 7) described S. ingens from the Lower Permian of Oklahoma, also. based on a distal wing third. The apical parts o the wing in all three specimens resemble each other and the species cannot be separated generically as far as is presently known. Unfo.rtuna.tely, they can also hardly be separated from Bardohymen (this statement Martyonv's reconstruction o Sylvohymen robustsus (1941) showing MA anastomosing with Rs is obviously incorrect, as. the connection o MA with R or Rs by means o a cross vein is characteristic or Bardohymenidae.
Wings long and slender, tapered rather abruptly in the basal third; hind wings slightly longer than ore wings, broader at about midwing; Sc recognizable only in proximal hal o the wing; Rs with 3 main branches; AI S-shaped; po.sterior margin in untapered part o wing almost parallel to the anterior margin.
Body structures: Prothorax with transverse elevations; metathorax broader than mesothorax; first abdominal segment strongly tapering; ovipositor in females stout, covered by hairs; projections probably maximally 4 in number in a row.
As shown previously by NIartynov (I94) and Carpenter (I947, p. 3:), Sylvohymen is closely related to the type genus Bardohymen. The reconstruction of Bardohymen magniennifer (G. Zalessky I937, p. 6o3, fig. is obviously incorrect tor its cubital branches. By the structure of pterostigrnal area the genus Sylvohymen is. related to ztlexahymen n.g., which differs in having a relatively shorter and broader wing with concave posterior margin, small rs area and A parallel with posterior margin.  Figure I, plates and 2 This species is based upon an obverse and reverse of a female with two complete wings and damaged lateral part of body. The thorax and the abdomen are preserved on the dorsal side, while the head presents a composite of dorsal and ventral surface showing bases of stout palpi. The projections, if only the obverse were known, give the appearance of stout tubercles. In the reverse, however, they continue like hollow tubes into the matrix.
For preparing the illustration, both obverse and reverse parts of the specimen were. used.
Wings slightly subequal, the hind pair being longer and broader at about midwing; color markings missing; fore wing length 5o ram, maximum width 9.1 ram, almost equally broad except for the tapered proximal third; anterior margin slightly convex; C bordering the whole wing; apex bent backward and almost pointed; R apicalIy diverging to some extent, with 1-3 terminal twigs; Rs with 3 simple long branches; AI S-shaped with a row of about 9 branches; cross veins about 18 in number; cross vein between RI and Rs forming a heavy bar, thickened at its costal end; hind wing length 51 mm, maximum width 9.7 mm, broadest at the mid-wing; hind wing narrowing proximally less abruptly; anterior margin somewhat straighter; R apically less diverged away from the anterior margin, with only twig.
1This remarkable specimen was turned over to me for study by the Body structures" Length ot? head 1.8 mm, width about 3.6 ram; preserved length of antennae 9 mm, antennae composed of numerous cylindrical segments; segments of maxillary palpi striated, preserved segment length I.I mm, width o.8 mm; median line running along the whole body; prothorax length 4 mm, maximum width 8 mm, provided with two obliquely oriented longitudinal elevations and one elevation located anteriorly and centrally; mesothorax length 4.r ram, probable width 8.8 ram; metathorax length 4.2 mm, maximum width probably 9.2 ram; legs covered by setae; front tibia length 38 mm; hind tibia length about 5o mm; abdomen length I4.5 mm, maximum width 9.4 mm; abdominal segments unequal, length of segments as follows: 1st 2 mm; 2nd o.8 mm; 3rd 1.3 mm; 4th 1.7 mm; 5th 2 ram; 6th 2 mm; 7th o.5 mm; 8th mm; 9th mm; Ioth 1.6 ram; th o.5 mm; each abdominal segment but the Ith has a transverse fiat topped ridge; IIth segment divided by a deep incision into two lobes; ovipositor stout, reaching much beyond the end of the body, The preservation of the holotype is very good, particularly because the veins of the wings have been secondarily penetrated and colored by manganese, which enters also the minute transverse cracks. The body is not tully flattened. The abdomen especially is preserved in its original convexity. Some of the bases o.f the projections are well preserved and only those are introduced in figure I, marked as circles, as they actually appear. The projections were undoubtedly growing out from the tergites in regular rows, but since the surface of the body is uneven, they cannot be distinguished from the irregularities of the matrix. The actual length of the projections could not be followed as they continue inside the reverse of the fossil under an acute angle with the. body. Their position, however, indicates that Psyche [September they ha.ve been sclerotized. In analogy to the closely related Protohymenidae, it is probable that the projections were slightly curved and at least several millimeters long.
Genus Alexahymen, new genus Type species: Alexahymen maruska n.sp., Lower Permian ox Moravia. This genus includes one species,, represented by 3 incomplete wings. Wings shorter than in the related genus 8ylvohymen, broadest behind the midwing, tapering gradually towards the base; Sc distinct to about two thirds o the wing length; R diverging apically slightly from the anterior margin; Rs with two short branches; AI parallel with the posterior margin; posterior margin slightly concave.
Alexahymen differs from other genera of the family Bardohymenidae in its relatively short and broad, gradually tapering wings, posterior margin concavely shaped, A parallel with the posterior margin and sending off a series of numerous twigs, and small rs area.
Alexahymen maruska n.sp. This species is based upon the holotype, represented by a wing without the proximal part, and by two additional, isolated, equally damaged wings. With regard to the close similarity between the ore and hind wings in Bardohymenidae, the po.sition of wings in the pair can be only inferred. However, from analogy with 8ylvohymen sibiricus, the only btrdohymenid with both wings in situ, it seems that the hind wings in this ]amily tended to have a straighter anterior margin and more concave posterior margin. Consequently, the holo- Color markings missing; total wing length about 33-36 ram, maximum width 8.7-9.9 mm; anterio,r margin slightly convex in the distal part; C bordering the wing; apex more or less pointed; R with 4-5 very short terminal twigs; Rs branches short; A sending off 6 branches; cross veins 15-16 in number, mostly in double row. All three specimens of Alexahymen maruska carry the details of venation, described by Carpenter I962, p. 38-39) in ,4ctinohymen russeli, namely flattened C, widening distally beyond the end of and C, So, R-R around midwing to.uehing each other. The cross vein rs-rI orms a heavy bar widened at its costal end in the holotype and specimen 3/972. In the specimen /x972 it is an average, though thick cross vein.
The posterior margin in the holotype is formed in a different way, perhaps as individual variation. It is convexly curved in between the branches, of Ax, media and cubitus, so that the tips of the branches protrude not unlike the fingertips in a bat wing. A similar phenomenon is indicated in 3oravohymen vitreus n.sp. of the related family Moravohymenidae.

Moravohymenidae, new family
Type genus: Moravohymen n.g. This family is based upon a fragment of a single wing (probably hind wing), which seems to combine the features of Bardohymenidae with those of some megasecopterid families of Commentry, France (Carpenter, 1951 ).
Wings broadest at the beginning of the apical third, tapering gradually proximally; Sc remote from the anterior margin and terminating freely in the subcostal area well before apex; R remote from $c, not diverging apically from the posterior margin; NIA connected with R or with the very origin of Rs by a cross vein; stem of NI either clo.se or used with R; AI not parallel with the posterior margin, sending off few irregular branches; cross veins arranged into irregular rows and partly sigmoidal; row of cross veins, in r I-rs area; veins and wing margin provided by setae.
The family Moravohymenidae has very gradually tapering wings with maximum width shifted to the distal third. NIA is connected with R or Rs very much as in Bardohymenidae.  This species is based upon an obverse and reverse of a wing with damaged proximal part. According to the concave shape o{ the posterior margin it might be a hind wing.
Wing ragment: length 21 mm, maximum width 6.9 mm; C, Sc and R distally much thinner veins than in Bardohymenidae; the membrane in the pterostigmal region probably sclerotized; RI sending off 3 terminal twigs to C; r I-rs area broad, with 6 weak cross veins; Rs branches occupying a considerably large area; cross vein connecting MA with R at the origin of Rs is a heavy bar, thickened at its costal end; cross veins in medial area and cubital area slightly sigmoidal; posterior margin with small convex bends, in between the ends of median and cubital branches. This remarkable ossil with ragmentary wings would hardly warrant ormal description because the classification ot the order is based upon the wing venation. However, the specimen shows the inner structure ot the sucking mouth parts, the ventral attachment o the legs to the body, the branched pro:iections of terga with preserved surface and an ovipositor provided by styli. Since this insect is of unusual interest, generic and specific names are being assigned.
Though the wings of the. specimen are tragmentary, there is no doubt about referring them to. the family Homoiopteridae, according to following characteristic features (Kukalov,i 1969, p.  The body is a composite of ventral surxCace and inner structures, which were uncovered by preparation at different levels. With regard to the complexity of the composite, preservation must be discussed next.
The body was preserved while lying on its dorsal side. It was much decomposed betore being covered by sediment. The maxillary palpi disintegrated into single, segments, which were partly displaced.
The legs with some parts o the sterna were shifted towards the head. and tergal elements and is vaguely preserved. It was partly removed to uncover the projections.
Head structures: The character o the palaeodictyopteran mouthparts, which are elongated into a beak consisting o stylets o mandibles and maxillae, have been described in more detail by Crampton (I927) and by Laurentiaux I952, I953). However, there was no evidence about the arrangement of the stylets. Only recently Carpenter and Richardson (1971, p. 28o) described a section of the beak in a strikingly unusually preserved specimen. The pair of some- what larger mandibles is located anteriorly rom the maxillae and there is the fifth stylet, probably derived rom the hypopharynx, between the maxillae and slightly more posterior.
The arrangement of the stylets in Monsteropterum moravicum xully confirms the conclusion of Carpenter and Richardson. The pair o mandibles is superimposed, and slightly longer than the pair of maxillae. The mandibular stylets in all probability partly overlap each other along the inner margin, because the undisturbed width of the let mandibular styler ( fig. 7-Md)  8-E). They orm continuous rows and seem to belong to a. single structural unit, which is a series of short, peg-shaped, perpendicularly oriented pillars, supporting the long hollow mandibular stylets from inside. This assumption is based mainly on the fact that the second row o crescent-shaped pits ( fig. 7-P) passes distinctly under the layer Md ( fig. 7), which is the natural surta.ce of the mandible in contact with the maxillary styler. On the Md layer the pit row continues in the form of oval elevations ( fig. 7-E).
The Iet maxillary stylet is co.mpletely missing; the right is preserved only by the fragment of distal end ( fig. 7-Ma). However it provides information on the morphology of the mouthparts in Palaeodictyoptera: the maxill.ary stylets were located under the mandibular stylets (in fig. 7 reversed because of the ventral view of the beak); they were distinctly shorter than the mandibles; they were Md PE G Because the beak is split along the median plane, it presents an uniquely favorable occasion to study the inner structure. However, it gives little reliable information about the character of the food canals. The transverse section of the beak ( fig. 8) is actually a composite o several inner surfaces of the mandibles, all of them carrying alternating grooves and ridges, and is slightly distorted by an oblique pressure. Because of the uneven level of splitting and slight deformation during fossilization, the food canals are not clearly distinguishable.
The maxillary palpi .are robust and overlap with the beak (fig. 6). All segments carry a flat-topped longitudinal ridge. The surface is covered with a fine rugosity and with occasional irregular grooves.
"The more perfect preservation of the beak in Monsteroterum moracicure brings an explanation of the "protruding needle-like tips" in the beak of Mecynostoma dohrni (Palaeodictyoptera, Meeynostomatidae, KukalovA 1969, p. 210, fig. 28). The protruding structures are undoubtedly also laciniae extending beyond the superimposed pair of mandibles. T) is grown together with the femur and is preserved only as a triangular swelling. The tibia is slightly longer than the femur. The tarsus is five-segmented, with the 3rd and 4th segments distinctly shorter than the remaining ones. The praetarsus bears a pair of lateral claws. The legs are covered by scattered setae and a granular rugosity.
The mesothorax and metathorax are about equal in size, with a broad V-shaped ridge.
With this specimen styli are recognized for the first time for the order Palaeodictyoptera. Within the extant insects,, styli on female genitalia are known in the adult stage only in one order, the Odonata, in which they arise from the second coxopodite. Also, the general appearance of the ovipositor is very much like the endophytic ovipositors of some Odonata.
The projections ( fig. 6-P) are long and very branched, apparently much more than preserved in the f.ossil. They have scattered setae and their surface is finely rugose. The sockets of the setae are deeply incised and their density is about equal to that on the legs. The setae increase in number towards the ends of the branches. The projections are stiffly backwardly curved and were undoubtedly sclerotized. Previously, I have been able to study projections in four families of the order Megasecoptera. In all the projections were simple un- branched outgrowth of the terga. The specimen here described is the first example with projections in the order Palaeodictyoptera and at the same time the first one to show branching. Additional examples should be expected within Palaeodictyoptera, even though their occurrence probably was not as common as in the Megasecoptera.
Psyche [September Original: No. 5/I972 (obverse and reverse of the body and basal parts of wings; separately reverse of the right front leg); Paleontological Institute of Charles University, Prague, Czechoslovakia. Collected in the Lower Permian deposits near Obora, Moravia. SUMMARY Megasecoptera and Palaeodictyptera, two of the three extinct paleopterous haustellate insect orders related to extant Ephemeroptera and Odonata, were found to carry fringe-like projections on the thorax and abdomen. Both orders include mostly large to very large insects with good flying ability, which held the wings outstretched when resting. The nymphs were terrestrial, probably arboreal, and had articulated wing pads which were oriented obliquely backwards (Carpenter and Richardson, x968).
The projections are hollow outgrowth of the terga, forming usually regular rows along the posterior margin or occasionally also on the nota. They are more or less sclerotized, protruding up and backwards from the body. The surface is rugose and with scattered tactile setae. The projections are simple or richly branched and vary from short to very long. They occur in both adults and nymphs.
As far is known, their morpho,logy is characteristic, for separate families. Now, the projections are known in Mischopteridae Aspidothoracidae, Corydaloididae, Brodiidae, Protohymenidae and Bardohymenidae in the order Megasecoptera. and in Homoiopteridae in the order Palaeodictyoptera. In their location, they are homologous with the dentation on the transverse abdominal carinae, present in all extant Odonata and with paired pits on abdominal terga in nymphs of Gomphidae.
Since the projections are obvio.usly a unique and isolated character, their function in two extinct Paleozoic orders is obscure. From the morphology it might be assumed that they were at least to some extent movable and were provided with mechanical sense organs.
The following additional characters have been added to the knowledge of Palaeodictyoptera The beak co.nsists of stylets of mandibles and maxillae, the pair of mandibles being longer and superimposed. Each maxilla is divided into a robust galea and thin lacinia, located at the inner margin and underneath the galea. The lacinia has two, inwardly curving apical lobes and extends beyond the beak. Mandibular stylets partly overlap each other along the inner margin. On contact with the maxillae, they have alternating ridges and grooves, enabling firmer connection. The mandibular stylets are hollow and the cavity is crossed by rows of perpendicularly oriented pillars, providing additional mechanical support.
The legs have short, conically truncate coxa.e. Proximally, on the ventral side, there are two circular structures, which perhaps represent the katapleuran ring separated by a sigmoidal paracostal suture from the anapleuran ring. The trochanter is triangular, grown together with the emur as in Odonata.
The ovipositor is provided by styli as in extant Odonata.