New mesozoic protomyrmeleontidae (Insecta: Odonatoptera: Archizygoptera) from Asia with a new phylogenetic analysis

journal of Systematic Palaeontology 3 (2): 187-201 doi:io.ioi7/Si4772oi9O5OOi549 Printed in the United Kingdom

Issued 22 June 2005 © The Natural History Museum

NEW MESOZOIC PROTOMYRMELEONTIDAE (INSECTA: ODONATOPTERA: ARCHIZYGOPTERA) FROM ASIA WITH A NEW PHYLOGENETIC ANALYSIS Andre Nel UMR CNRS 5143, Entomologie, M u s é u m National d'Histoire Naturelle, 45, rue de Buffon, F-75005, Paris, France

Julien F. Petrulevicius UMR CNRS 5143, Entomologie, Muséum National d'Histoire Naturelle, 45, rue de Buffon, F-75005, Paris, France and CONICET, Argentina

Xavier Martínez-Delclòs Departament d'Estratigrafia, Paleontologia i Geociències Marines, Facultat de Geologia, Universitat de Barcelona, E-08071, Barcelona, Spain SYNOPSIS The following damselfly-like protomyrmeleontid Odonatoptera are described from the Mesozoic of Eurasia: Ferganagrion kirghiziensis gen. et sp. nov., Paraobotritagrion gen. nov. for Paraobotritagrion tenuiformis (Zessin, 1991), Mongolagrion shartegensis gen. et sp. nov., Protomyrmeleon pumilio sp. nov., Protomyrmeleon kazakhstanensis sp. nov., Protomyrmeleon karatauensis sp. nov., Protomyrmeleon grandis sp. nov. The wing venations of protomyrmeleontid genera are homologised and the high morphological disparity in this family suggests that the Protomyrmeleontidae had very different modes of flight and consequently occupied a wide range of palaeoenvironments. A new phylogenetic analysis suggests that the fossil record is too scarce and incomplete to solve the phylogeny of the Protomyrmeleontidae. In particular, the current division of Protomyrmeleontidae in Triassagrioninae and Protomyrmeleontinae is only weakly supported. KEYWORDS Odonatoptera, Protomyrmeleontidae, Taxonomy, Mesozoic, Asia, Phylogeny

Contents Introduction

188

Material and methods Abbreviations used or names of wing structures

188 188

Systematic descriptions Archizygoptera Handlirsch, 1906 Family Protomyrmeleontidae Handlirsch, 1906 Genus Ferganagrion nov. Ferganagrion kirghiziensis gen. et sp. nov. Genus Paraobotritagrion nov. Genus Mongolagrion nov.

188 188 188 189 189 191 192 192 192 193 194

Mongolagrion shartegensis gen. et sp. nov. Genus Protomyrmeleon Geinitz, 1887 Protomyrmeleon grandis sp. nov. Protomyrmeleon karatauensis sp. nov. Protomyrmeleon kazakhstanensis sp. nov. Protomyrmeleon pumilio sp. nov. Protomyrmeleon sp. indet. Genus Triassagrion Tillyard, 1922

195 197 198 198

188

A.

NEL ET AL

Phylogenetic analysis of the genera of Protomyrmeleontidae

199

Acknowledgements

200

References

200

Appendix 1: List of characters used in the phylogenetic analysis

201

Appendix 2: Data matrix

201

INTRODUCTION The Protomyrmeleontidae, a small family of Odonatoptera belonging to the Protozygoptera, sister group of modern Odonata, ranges in age from Upper Triassic to Lower Cretaceous. Members of this family have a very particular wing venation, characterised by a unique pattern of the radial area with the main veins apparently branching off from RP3/4, instead of RP, with RP3/4 and RP2 being forked, unique among Odonatoptera. Whereas other Protozygoptera disappeared at the end of the Triassic, the Protomyrmeleontidae continued to flourish during the Jurassic and the Early Cretaceous, with numerous genera widespread all over the World. This group of damselfly-like Odonatoptera is even more taxonomically and morphologically diverse than the true Zygoptera during the Late Jurassic. Its extinction during the Early Cretaceous remains unexplained. This family has been the subject of several recent revisions and descriptions of new genera and species (Zessin 1991; Nel 1992; Nel & Henrotay 1992; Ansorge 1996; Martinez-Delclos & Nel 1996; Bechly 1997; Henrotay etal. 1997; Nel & Jarzembowski 1998). We were able to study several new protomyrmeleontid species and genera in the insect collection of the Palaeoentomological Laboratory at the Academy of Science of Moscow (PIN). These new taxa show novel patterns of venation in their radial area, demonstrating morphological diversity higher than currently supposed for this group. Using a new scheme of homologies for the wing venation pattern, we present a phylogenetic analysis of the Protomyrmeleontidae that includes these new genera.

MATERIAL AND METHODS All drawings were made using a camera lucida and binocular microscope. The nomenclature of the dragonfly wing venation is based on the interpretations of Riek & Kukalova-Peck (1984), amended by Nel etal. (1993) and Bechly (1996). For a discussion on the very particular protomyrmeleontid wing venation, see Nel & Henrotay (1992). The protomyrmeleontid fore and hind wings look very similar. Zessin (1991) and Ansorge (1996) considered some isolated wings as fore wings and others as hind wings, but without convincing evidence, because no wings in connection with the body were known. The holotype of Protomyrmeleon grandis sp. nov. (PIN 2066/27), described here, has two wings in connection with the thorax and gives some information on this point. These two wings are very similar except for the length and shape of the bridge (sensu Nel & Henrotay 1992) between RP3/4 and IR2, which is distinctly longer and more closely parallel to RP3/4 ('wing

type A') in one of these wings than in the other ('wing type B'). However, it is not possible to decide which one is the fore wing. Interestingly, a similar long bridge is present in the alleged fore wing of Protomyrmeleon brunonis Geinitz, 1887 and a similar shorter bridge is present in the alleged hind wing of P. brunonis, figured in Zessin (1991: text-figs 1-2). Ansorge (1996: text-fig. 3E) figured an alleged hind wing supposed to belong to P. brunonis but with a long bridge, and another alleged hind wing with a shorter bridge (Ansorge 1996:, text-fig. 3B). However, from the photograph of the same specimen (Ansorge 1996: pi. 1, fig. 1), it is very hard to decide which one is the fore wing. In conclusion, there is a difference between the fore and the hind wings in the Protomyrmeleontidae (at least in the genus Protomyrmeleon), but, based on current knowledge, it is still not possible to know what exactly characterises the fore wing and the hind wing. Abbreviations used or names of wing structures AA, Analis Anterior; AP, Analis Posterior; Axl and Ax2, primary antenodal cross-veins; CuA, Cubitus Anterior; CuP, Cubitus Posterior; IR1 and IR2, intercalary veins of the radial area; MA, Median Anterior; MP, Median Posterior; N, nodus; RA, Radius Anterior; RP, Radius Posterior; ScA, Subcosta Anterior; ScP, Subcosta Posterior.

SYSTEMATIC DESCRIPTIONS ARCHIZYGOPTERA Handlirsch, 1906 Family PROTOMYRMELEONTIDAE Handlirsch, 1906 TYPE GENUS. Protomyrmeleon Geinitz, 1887 (Lower Jurassic to Lower Cretaceous). Triassagrion Tillyard, 1922 (Triassic), Tillyardomyrmeleon Henrotay et al., 1997 (Triassic), Italomyrmeleon Bechly, 1997 (Triassic), Ferganagrion gen. nov. (Triassic), Obotritagrion Zessin, 1991 (Lower Jurassic), Paraobotritagrion gen. nov. (Lower Jurassic), Malmomyrmeleon Martinez-Delclos & Nel, 1996 (Late Jurassic), Mongolagrion gen. nov. (Late Jurassic), Saxomyrmeleon Nel & Jarzembowski, 1998 (Lower Cretaceous). OTHER GENERA.

The unique synapomorphy of this family lies in having vein RP2 secondarily forked into RP2a and RP2b. The other structures of the radial veins are very diverse in this family. The extremely elongated legs could be synapomorphic but this feature is only known in the genus Malmomyrmeleon. REMARKS.

MESOZOIC PROTOMYRMELEONTIDAE: NEW TAXA AND PHYLOGENY

189

Figure 1 A, Ferganagrion kirghiziensis sp. nov., holotype PIN 2069/3266, wing (scale bar = 3 mm); B, Mongolagrion shartegensis sp. nov., holotype PIN 4270/861, wing (scale bar = 2 mm); C, Protomyrmeleon grandis sp. nov., paratype PIN 2066/27, wing (scale bar = 3 mm).

Genus FERGANAGRION nov. After Fergana valley, Kirghizia, where the holotype was discovered. ETYMOLOGY.

side perpendicular to RA and distal side strongly oblique; RP3/4a not zigzagged.

TYPE SPECIES. Ferganagrion kirghiziensis sp. nov.

Ferganagrion kirghiziensis gen. et sp. nov. (Figs lA, 2B)

Wing showing complete fusion of longitudinal part of IR1 with RP2a; abridge between IR2 and RP3/4; RP2 and IR2 separated, with a bridge between them; a long zigzagged pseudo-IRl; an elongate pterostigma, with basal DIAGNOSIS.

ETYMOLOGY

-

After the country Kirghizia.

Holotype and only known specimen- PIN 2069/3266, Paleoentomological Laboratory at the Academy of Science of Russia, Moscow. MATERIAL.

190

A.

N E L ETAL.

Figure 2 Schemata of the wing venation of different genera of Protomyrmeleontidae. A, Tillyardomyrmeleon petermilleri Henrotay ef a/., 1997, holotype, after Henrotay etal. (1997); B, Ferganagrion kirghiziensis gen. et sp. nov., holotype; C, Malmomyrmeleon viohli Martinez-Delclos & Nel, 1996, holotype, after Martinez-Delclos & Nel (1996); D, Obotritagrion petersiZess'm, 1991, holotype, after Zessin (1991); E, Protomyrmeleon quadriordinum (Zessin, 1991), holotype, after Zessin (1991); F, Protomyrmeleon brunonis Geinitz, 1885, specimen figured in Henrotay etal. (1997). Scale bars = 2 mm. For abbreviations used, see the text.

LOCALITY AND HORIZON.

Madygen Formation, later Middle or earlier Late Triassic, Madygen Valley, northern area, Dzhailoucho, Osh Region, Batken District, south Fergana valley, Kirghizia. DIAGNOSIS. AS for genus.

The distal two-thirds of an isolated wing; length of preserved part 18.4 mm, width of wing 3.9 mm; wing narrow; distance from nodus to pterostigma 12.1 mm, from pterostigma to wing apex approximately 1.0 mm, postnodal area very long, with at least 11 postnodal cross-veins and 15 postsubnodal cross-veins; pterostigma sclerotised, long, 4.0 mm long, 0.4 mm wide, covering four cells, with basal side of pterostigma perpendicular to Costa and RA but distal side very oblique; pterostigmal brace absent; nodal DESCRIPTION.

cross-vein and subnodus not preserved; presence of a long bridge between RP3/4 and IR2, approximately 1.0 mm long; true base of IR2 looking like an oblique cross-vein between IR2 and RP, 1.9 mm distal of nodus level; a bridge between IR2 and RP2, 0.8 mm long, thus RP2 apparently branching on IR2, true base of RP2 looking like an oblique cross-vein between RP2 and RP, 4.1 mm distal of nodus level; base of IR1 looking like an oblique cross-vein between RP and RP2a, anterior branch of RP2, 5.1 mm distal of nodus level; distal main part of IR1 completely fused with RP2a; a zigzagged distal pseudo-IRl between RP2a + IR1 and RP1; RP2 divided into two convex branches RP2a and RP2b, 0.6 mm distal of its fusion with the bridge between it and IR2; both RP2a and RP2b nearly straight, with five rows of cells and 4— 5 irregular longitudinal veins between them; one row of cells

M E S O Z O I C P R O T O M Y R M E L E O N T I D A E : NEW TAXA A N D PHYLOGENY

191

Figure 3 Schemata of the wing venation of different genera of Protomyrmeleontidae: small dotted lines correspond to vein IRi; broad bold vein represents the bridge between IRi and RP2; narrow bold vein represents the bridge between RP2 and IR2; dashed line represents vein RP2; alternating dotted and dashed line represents vein IR2. A, Terskeja pumilio Pritykina, 1981, holotype, after Pritykina (1981); B, Triassagrion australiense Tillyard, 1922, holotype, after Tillyard (1922); C, Paraobotritagrion tenuiformis (Zessin, 1991), holotype, after Zessin (1991); D, Mongolagrion shartegensis gen. et sp. nov., holotype; E, Saxomyrmeleon keatingei Nel & Jarzembowski, 1998, holotype, after Nel & Jarzembowski (1998); F, Italomyrmeleon bergomensis Bechly, 1997, holotype, after Bechly (1997). Scale bars, A = 5 mm; B-F= 2 mm. For abbreviations used, see the text.

between RP2b and IR2; two rows of cells between IR2 and RP3/4; RP3/4 divided into two branches RP3/4a and RP3/4b close to posterior wing margin, RP3/4a curved and 1.7 mm long, RP3/4b short, 0.5 mm long; one row of cells between RP3/4 and MA; MA straight; MP basally straight and distally zigzagged; one row of cells between MP and posterior wing margin; CuA if present not preserved. This fossil displays the main synapomorphy of the Protomyrmelontidae, i.e. veins RP2 secondarily forked into RP2a and RP2b (Bechly 1996; Henrotay et al. 1997). It differs from all other Protomyrmeleontidae species in showing complete fusion of the distal longitudinal part of vein IRI with RP2a. DISCUSSION.

Genus PARAOBOTRITAGRION nov. ETYMOLOGY. After para for 'similar to' and Obotritagrion Zessin, 1991. TYPE SPECIES. Obotritagrion tenuiformis Zessin, 1991 (Fig.3C). Short and narrow wing; RP3/4 not divided apically into RP3/4a and RP3/4b; no oblique vein between RP3/4 and RP basal of base of IR2; IR2, RP2 and IRI not partly fused with RP3/4; base of RP2 perpendicular to RP; RP2 partly fused with IR2; a bridge between RP2 + IR2 and IRI; IRI not fused with RP2 or IR2. DIAGNOSIS.

192

A.

NEL

ETAL

Zessin (1991) based this species on a single wing, supposed to be a fore wing without clear argument. He attributed it to the genus Obotritagrion Zessin, 1991. Ansorge (1996: 13-15, text-fig. 4, pi. 1, fig. 4) described and figured another more complete specimen that has the same characters proper to this species. It differs from the type species Obotritagrion petersi Zessin, 1991 (Fig. 2D) as follows: O. petersi has three oblique 'cross-veins' between RP and RP3/4, basal of the apparent branching of IR2 + RP2 + IR1 on RP3/4. These three oblique veins are the bases of the three veins IR2, RP2 and IR1. Thus these veins are partly fused with RP3/4 and fused together. This common 'stem' separates into the anterior IR1 and the posterior IR2 + RP2, then IR2 and RP2 separate one cell distally. In 'Obotritagrion1 tenuiformis, there is no oblique vein between RP3/4 and RP basai of the base of IR2, thus IR2, RP2 and IR1 are not partly fused with RP3/4. The base of RP2 is perpendicular to RP and RP2 is partly fused with IR2. There is a bridge between RP2 + IR2 and IR1, thus IR1 is not fused with RP2 or IR2. These differences are as important as those between other protomyrmeleontid genera. They justify a generic separation between Obotritagrion petersi and Paraobotritagrion tenuiformis. Nevertheless, these two taxa share several characters that are probably related to the small size of their wings, i.e. RP3/4 not divided into RP3/4a and RP3/4b, branches of RP2 shorter than in other Protomyrmeleontidae spp., few postnodal cross-veins, pterostigma short and covering few cells. Similar convergent reductions of the wing venation occur in very different modern families of anisopteran (Libellulidae and 'Corduliidae'). DISCUSSION.

Genus MONGOLAGRION nov. ETYMOLOGY.

After Mongolia and Agrion.

TYPE SPECIES. Mongolagrion shartegensis sp. nov. DIAGNOSIS. In wing: IR2 partly fused with RP3/4; RP2 independent of IR2; IR1 independent of RP2 and IR2; RP3/4a well defined and long; basal stem of RP2 short; cell between IR2 and RP2 rather broad; pterostigma three cells long, only one row of cells in area between distal part of IR2 and RP3/4a, branches RP2a and RP2b long; venation not very dense, with no multiplication of cells in apical part of wing.

Mongolagrion shartegensis gen. et sp. nov. (Figs lB, 3D) ETYMOLOGY.

After the type locality, Shar-Teg.

Holotype and only known specimen: PIN 4270/861, Paleoentomological Laboratory at the Academy of Science of Russia, Moscow. MATERIAL.

Shar-Teg beds, Late JurassicEarly Cretaceous, Shar-Teg Ula Mountain, east of Atas-Bogd Mountain, Gobi-Altai' Aymag, Mongolia. LOCALITY AND HORIZON.

DIAGNOSIS. AS for genus.

A small isolated wing, 20.0 mm long, 4.0 mm wide; distance from base to nodus 5.0 mm; from nodus to pterostigma 12.0 mm, from pterostigma to wing apex about 1.5 mm, postnodal area very long, with 18 postnodal crossveins and 15 postsubnodal cross-veins; pterostigma sclerotised, long, 1.8 mm long, 0.6 mm wide, covering three cells, DESCRIPTION.

with basal and distal sides of pterostigma oblique and parallel; pterostigmal brace absent; nodal cross-vein and subnodus nearly perpendicular to RA and aligned; base of RP3/4 normal, one cell and 0.7 mm distal of subnodus; base of IR2 individualised as an oblique vein between RP and RP3/4, 2 i cells and 2.1 mm distal of subnodus; IR2 fused with RP3/4 for a little more than one cell and 1.1 mm, then directed towards wing apex for 2.1 mm, making a more distal double angle and distally more or less straight; one row of cells between distal part of IR2 and RP3/4a; base of RP2 still individualised as an oblique vein between RP and IR2, 5.6 mm distal of subnodus; a one-celled bridge between IR2 and RP2, 0.4 mm long; stem of RP2 two-cells long; fork of RP2 7.9 mm basal of wing apex; one row of cells between RP2b and IR2; area between RP2a and RP2b with 3-^1 rows of cells in basal part and nine rows of cells along posterior wing margin; base of IR1 probably one cell distal to that of RP2, looking like a slightly oblique vein between RP1 and RP2; a bridge between IR1 and RP2, one cell long; IR1 strongly zigzagged; one row of cells between RP2 and IR1 and between IR1 and RP1; RP3/4 divided into two branches RP3/4a and RP3/4b close to posterior wing margin, RP3/4a slightly curved, not zigzagged and 3.1 mm long, RP3/4b short, 0.9 mm long; one row of cells between RP3/4 and MA; MA and MP straight; two rows of cells between MP and posterior wing margin; CuA not independent of MP; AA not independent of AP. DISCUSSION. This fossil has the main synapomorphy of the Protomyrmelontidae. It differs from all other protomyrmeleontid genera, except Triassagrion and Saxomyrmeleon, in having IR2 partly fused with RP3/4, RP2 independent of IR2 and IR1 independent of RP2 and IR2. It also differs from Triassagrion in the following characters: RP3/4a well defined and long, a short basal stem at RP2, instead of four cells long in Triassagrion between basal part and fork of RP2, presence of a rather broad cell between IR2 and RP2, instead of being closely approximate in Triassagrion, longer pterostigma, venation less dense, with fewer cells in apical part of wing. It differs from Saxomyrmeleon in its distinctly longer pterostigma, three cells long instead of one, and distinctly narrower area between the distal part of IR2 and RP3/4a, with only one row of cells and longer branches RP2a and RP2b.

Genus PROTOMYRMELEON Geinitz, 1887 [= Zirzipanagrion Zessin, 1991] TYPE SPECIES. Protomyrmeleon brunonis Geinitz, 1887 (Lower Jurassic). OTHER SPECIES. Protomyrmeleon pascholdi Zessin, 1991, P. anglicanus Tillyard, 1925a (both Lower Jurassic); P. jurassicus Nel, 1992, P. handlirschi Martynov, 1927, P. angustivenosus Martynov, 1927, P. pumilio sp. nov., P. kazakhstanensis sp. nov., P. karatauensis sp. nov., P. grandis sp. nov. (all Late Jurassic); P. cretacicus Nel & Jarzembowski, 1998 (Lower Cretaceous).

The type specimen of P. brunonis is a single wing that is very poorly preserved and extremely difficult to compare to other protomyrmeleontid species (see Zessin 1991: pi. 1). Ansorge (1996) attributed several other Lower Jurassic specimens, including the type specimens of

REMARKS.

M E S O Z O I C P R O T O M Y R M E L E O N T I D A E : NEW TAXA AND PHYLOGENY

Figure 4

193

Drawings of wings. A, B, Protomyrmeleon grandis sp. nov.: A, holotype PIN 2784/67; B, paratype PIN 2066/27; C, Protomyrmeleon

karatauensis sp. nov., holotype PIN 2239/6. Scale bars = 2 mm. For abbreviations used, see the text.

P. bascharagensis Nel & Henrotay, 1992 and P. grasselensis Bode, 1953, to this species. In fact, there is no way to be certain that these better-preserved fossils are conspecific with the type of P. brunonis. There is a possible difference between P. brunonis and P. bascharagensis in the proximal part of the distal portion of IR2, which is not zigzagged in P. bascharagensis (type specimen and specimen IB000962a, Nel & Henrotay 1992; Henrotay et al. 1997) but is zigzagged in .P. brunonis (type specimen afterthe figure of Geinitz 1887; specimen LDA 325/1 figured in Zessin 1991; and in the specimens figured in Ansorge 1996: fig. 3). Nevertheless, we consider that it is preferable to maintain all these fossils in this species, in order to avoid taxonomic confusion. Only the discovery of further specimens in the various outcrops of the West European Lower Jurassic can help to solve the question.

Protomyrmeleon grandis sp. nov. (Figs IC, 4A, 4B) ETYMOLOGY.

In reference to the large size of the wing of

this species. Holotype: PIN 2784/67, a single wing; paratype: PIN 2066/27, two wings connected to a fragment of thorax, Paleoentomological Laboratory at the Academy of Science of Russia, Moscow. MATERIAL.

Karabastau Formation, CallovianKimmeridgian or Oxfordian-Kimmeridgian (Zherikhin & Gratshev 1993; Mostovski & Martmez-Delclos 2000; Rasnitsyn & Zherikhin 2002), Middle-Late Jurassic, Karatau Mountain Ridge near the village of Mikhailovka (Aulie), Chimkent region, southern Kazakhstan.

LOCALITY AND HORIZON.

DIAGNOSIS. Wing very large and long (more than 30.0 mm); pterostigma covering 4-5 cells; a very long bridge between IR2 and RP3/4 in wing of type 'A' (see above) but distinctly shorter in the wing of type 'B', five cells long in one wing but distinctly shorter in the other; two rows of cells in area

between MP and posterior wing margin; distal free part of IR2 distally zigzagged; three rows of cells in area between RP2 and RP1; one row of cells in area between Costa and RA, distal of pterostigma; 10 rows of cells between RP2a and RP2b along posterior wing margin. Holotype: an isolated very long wing of type 'A', 35.4 mm long, 6.1 mm wide; wing rather narrow; distance from base to nodus 9.5 mm, from nodus to pterostigma 19.5 mm, from pterostigma to wing apex approximately 3.5 mm, postnodal area very long, with at least 19 postnodal cross-veins and 18 postsubnodal cross-veins; pterostigma sclerotised, very long, 4.2 mm long, 0.7 mm wide, covering three cells, with basal and distal sides of pterostigma oblique and more or less parallel; pterostigmal brace absent; nodal cross-vein and subnodus perpendicular to RA, not aligned; RP3/4 beginning one cell distal of subnodus; presence of a very long and straight longitudinal bridge between RP3/4 and distal free part of IR2, five cells and 4.1 mm long, and strictly aligned with the bridge between IR2 and RP2, 1.7 mm long, and followed by a very long zigzagged vein that ends in IR1; IR2 apparently branching on RP3/4, true base of IR2 not clearly individualised; RP2 apparently branching on IR2, true base of RP2 looking like a very oblique cross-vein between RP2 and RP, 8.1 mm distal of subnodus; true base of IR1 not individualised, but IR1 never fused with RP2; two rows of cells between RP2 and IR1 and one row between IR1 and RP1; RP2 divided into two long convex branches RP2a and RP2b, 1.2 mm distal of its fusion with the bridge between it and IR2; both RP2a and RP2b nearly straight, with five rows of cells and 2-3 zigzagged longitudinal veins between them; 2-3 rows of cells between RP2b and IR2; IR2 distally zigzagged and vanishing in area between RP3/4 and RP2; one row of cells between IR2 and RP3/4a; RP3/4 divided into two branches RP3/4a and RP3/4b, with 2-3 rows of cells between RP3/4a and posterior wing margin, RP3/4a slightly zigzagged and 5.8 mm long, RP3/4b rather long, 2.3 mm long; one row of cells between RP3/4 and MA; MA DESCRIPTION.

194

A.

N E L ETAL.

and MP basally straight and apically zigzagged; two rows of cells between MP and posterior wing margin; CuA not independent of MP; AA not independent of AP. Paratype: one wing of the paratype is nearly identical to that of the holotype, especially in its very long bridge between RP3/4 and the distal free part of IR2 (wing of type 'A'); the other wing of type 'B' long and narrow, 32.5 mm long, 5.1 mm wide; distance from base to nodus 10.0 mm, from nodus to pterostigma 17.7 mm, from pterostigma to wing apex about 3.0 mm, postnodal area very long, with at least 25 postnodal cross-veins and 24 postsubnodal cross-veins; pterostigma sclerotised, rather long, 3.0 mm long, 0.7 mm wide, covering five cells, with basal and distal sides of pterostigma oblique and more or less parallel; pterostigmal brace absent; nodal cross-vein and subnodus aligned and perpendicular to RA; RP3/4 beginning two cells distal of subnodus; longitudinal bridge between RP3/4 and distal free part of IR2 of one of the wings short, two cells and 1.4 mm long, and strictly aligned with the bridge between IR2 and RP2, 3.0 mm long, and followed by a very long zigzagged vein that ends in IR1; IR2 apparently branching on RP3/4, true base of IR2 not clearly individualised; true base of RP2 looking like an oblique cross-vein between RP2 and RP, 8.3 mm distal of nodus; true base of IR1 not individualised, but IR1 never fused with RP2; 1-2 rows of cells between RP2 and IR1 and 1-3 rows between IR1 and RP1; RP2 divided into two long convex branches RP2a and RP2b, 0.5 mm distal of its fusion with the bridge between it and IR2; both RP2a and RP2b nearly straight, with 5-6 rows of cells and 12 zigzagged longitudinal veins between them; one row of cells between RP2b and IR2; free part of IR2 basally straight but zigzagged in its very distal part; 1-2 rows of cells between IR2 and RP3/4a; RP3/4 divided into two branches RP3/4a and RP3/4b, with two rows of cells between RP3/4a and posterior wing margin, RP3/4.a poorly defined and strongly zigzagged; RP3/4b rather short, 0.7 mm long; one row of cells between RP3/4 and MA; MA and MP basally straight and apically zigzagged; two long rows of cells between MP and posterior wing margin; CuA not independent of MP; AA not independent of AP. DISCUSSION. This species has the main synapomorphy of the Protomyrmelontidae, i.e. vein RP2 secondarily forked into RP2a and RP2b. It shares with the genera Protomyrmeleon and Zirzipanagrion exactly the same organisation of radial veins. It differs from the latter in the presence of only three rows of cells in the area between RP1 and RP2, instead of four, and in having the basal side of its pterostigma parallel to the distal side, instead of being nearly perpendicular to RA. These differences are very minor. It shares with Zirzipanagrion the presence of two rows of cells in the area between MP and the posterior wing margin. The differences between the two genera Zirzipanagrion and Protomyrmeleon are very few, as already noted by Henrotay et al. (1997). They hardly justify the separation between the two genera. Furthermore, in Protomyrmeleon kazakhstanensis sp. nov., there are specimens with four rows of cells in the area between RP2 and RP1 and one row of cells in the area between MP and the posterior wing margin, but other specimens with only three rows of cells in the former area and two in the latter. Thus, they are intermediate between Zirzipanagrion and the other Protomyrmeleon spp. We, therefore, here synonymise Zirzipanagrion with Protomyrmeleon.

Protomyrmeleon grandis differs from P. anglicanus Tillyard, 1925a and P. pascholdi Zessin, 1991 (both based on incomplete apical halves of wings) in its longer pterostigma, the presence of three rows of cells, instead of two, between RP1 and RP2, and a distally zigzagged IR2, instead of being straight. Protomyrmeleon angustivenosus Martynov, 1927 is based on a fragment of the costo-apical part of a wing. Although it has too few characters preserved to be compared in detail to the other Protomyrmeleon species, it differs from P. grandis in the presence of two rows of cells in the area between Costa and RA, distal of the pterostigma, and the presence of 15 rows of very small cells between RP2a and RP2b, along the posterior wing margin, instead of 10 as in P. grandis. Martynov (1927) also indicated that its wing was probably of the same length as that of P. handlirschi Martynov, 1927, i.e. about 23-24 mm long, thus distinctly shorter than in P. grandis. Protomyrmeleon grandis sp. nov. differs from all the other Protomyrmeleon spp., i.e. P. brunonis Geinitz, 1887 (including P. bascharagensis Nel & Henrotay, 1992 and P. grasselensis Bode, 1953), P. handlirschi Martynov, 1927, P. jurassicus Nel, 1992 and P. cretacicus Nel & Jarzembowski, 1997, in its great size, more than 30.0 mm long, and its very long bridge between IR2 and RP3/4 (five cells long) in its wing of type 'A' (Martynov 1927; Nel 1992; Nel & Henrotay 1992; Ansorge 1996; Nel & Jarzembowski 1997).

Protomyrmeleon karatauensis sp. nov. (Figs 4C, 5A) ETYMOLOGY.

After the type locality Karatau.

MATERIAL. Holotype and only known specimen: PIN 2239/6,

Paleoentomological Laboratory at the Academy of Science of Russia, Moscow. Karabastau Formation, CallovianKimmeridgian or Oxfordian—Kimmeridgian (Zherikhin & Gratshev 1993; Mostovski & Martinez-Delclos 2000; Rasnitsyn & Zherikhin 2002), Middle-Late Jurassic, Karatau Mountain Ridge near the village of Mikhailovka (Aulie), Chimkent region, southern Kazakhstan.

LOCALITY AND HORIZON.

DIAGNOSIS. Wing distinctly shorter than that of P. grandis sp. nov.; a very long bridge between IR2 and RP3/4, five cells long; two rows of cells in area between MP and posterior wing margin; IR2 distally zigzagged; two rows of cells in area between RP2 and RP 1; one row of cells in area between Costa and RA, distal of pterostigma; seven rows of cells between RP2a and RP2b along posterior wing margin; RP3/4a not zigzagged; real base of RP3/4 perpendicular to RP and RP3/4 apparently branching on MA (maybe an individual feature). An isolated and rather short wing of type 'A', 22.5 mm long, 4.1 mm wide; wing rather narrow; distance from base to nodus 6.5 mm, from nodus to pterostigma 13.6mm, from pterostigma to wing apex approximately 1.5 mm, postnodal area very long, with at least 18 postnodal cross-veins and 17 postsubnodal cross-veins; pterostigma sclerotised, rather short, 2.0 mm long, 0.6 mm wide, covering two cells, with basal and distal sides of pterostigma oblique and more or less parallel; pterostigmal brace absent; nodal cross-vein and subnodus perpendicular to RA, aligned; RP3/4 beginning one cell distal of subnodus; real DESCRIPTION.

M E S O Z O I C P R O T O M Y R M E L E O N T I D A E : NEW TAXA AND PHYLOGENY

195

Figure 5 A, Protomyrmeleon karatauensis sp. nov., holotype PIN 2239/6, wing; B, C, Protomyrmeleon kazakhstanensis sp. nov.; B, paratype PIN 2997/4433, wing; C, paratype PIN 2239/5, wing. Scale bars = 2 mm.

base of RP3/4 perpendicular to RP and RP3/4 apparently branching on MA; presence of a very long and straight longitudinal bridge between RP3/4 and distal free part of IR2, five cells and 2.8 mm long, and strictly aligned with the bridge between IR2 and RP2, 1.2 mm long, and followed by a very long zigzagged vein that ends in IR1; IR2 apparently branching on RP3/4, true base of IR2 not clearly individualised; RP2 apparently branching on IR2, true base of RP2 looking like an oblique cross-vein between RP2 and RP, 6.0 mm distal of nodus; true base of IR1 not individualised, but IR1 never fused with RP2; one row of cells between RP2 and IR1 and one row between IR1 and RP1; RP2 divided into two long convex branches RP2a and RP2b, 0.5 mm distal of its fusion with the bridge between it and IR2; both RP2a and RP2b nearly straight, with 3^1 rows of cells and two zigzagged longitudinal veins between them; two rows of cells between RP2b and IR2; IR2 distally zigzagged and vanishing in area between RP3/4 and RP2; one row of cells between IR2 and RP3/4a; RP3/4 divided into two branches RP3/4a and RP3/4b, with two rows of cells between RP3/4a and posterior wing margin, RP3/4a not zigzagged but curved and 2.8 mm long, RP3/4b rather short, 1.1 mm long; one row of cells between RP3/4 and MA; MA and MP basally straight and apically zigzagged; two rows of cells between MP and posterior wing margin; CuA not independent of MP; AA not independent of AP.

DISCUSSION. This fossil wing is very similar to that of P. grandis sp. nov., especially in the presence of a very long bridge between RP3/4 and distal free part of IR2, but it is distinctly shorter (22.5 mm long instead of more than 30.0 mm in P. martynovi sp. nov.), which justifies a specific separation. It differs from P. angustivenosus from the same locality in its smaller number of cells in the area between RP2a and RP2b along the posterior wing margin. It differs from both the poorly known P. anglicanus and P. pascholdi in its IR2, which is distally zigzagged instead of being straight. Protomyrmeleon kazakhstanensis sp. nov. (Figs 5B, 5C, 6A-C) ETYMOLOGY.

After the type region southern Kazakhstan.

Holotype: PIN 2784/70-71 (counterpart and part of an isolated rather short wing of type 'B'). Paratypes: PIN 2997/4433 (isolated incomplete wing of type 'A') and PIN 2239/5 (isolated wing of type 'A'), Paleoentomological Laboratory at the Academy of Science of Russia, Moscow. MATERIAL.

Karabastau Formation, CallovianKimmeridgian or Oxfordian-Kimmeridgian (Zherikhin & Gratshev 1993; Mostovski & Martinez-Delclos 2000; Rasnitsyn & Zherikhin 2002), Middle-Late Jurassic, Karatau LOCALITY AND HORIZON.

196

Figure 6 22

A.

N E L ETAL

Drawings of wings. A-C, Protomyrmeleon kazakhstanensis sp. nov.; A, holotype PIN 2784/70-71, counterpart; B, paratype PIN

3 9 / 5 ; C, paratype PIN 2239/5; D, Protomyrmeleon pumilio sp. nov., holotype PIN 2997/1131; E, Protomyrmeleon sp. indet., specimen PIN

2784/69. Scale bars A-D = 2 mm, E = 1 mm. For abbreviations used, see the text.

Mountain Ridge near the village of Mikhailovka (Aulie), Chimkent region, southern Kazakhstan. DIAGNOSIS. Wing distinctly shorter than that of P. karatauen-

sis sp. nov.; moderately long bridge between RP3/4 and distal free part of IR2, three to four cells long; pterostigma covering four cells; three to four rows of cells in area between RP2 and RP1; area between RP2a and RP2b rather narrow, with about 10 rows of cells along posterior wing margin; RP3/4a long; 21 postnodal cross-veins; IR2 straight except in its apical part. Holotype PIN 2784/70-71: preserved part of wing 19.8 mm long, 3.8 mm wide, preserved part of wing 15.0 mm long, 3.75 mm wide in PIN 2997/4433 (Fig. 5B), wing 17.7 mm long, 3.4 mm wide in PIN 2239/5 (Figs 5C, 6C); wings rather narrow; distance from base to nodus approximately 5.1 mm, and approximately 3.75 mm in PIN DESCRIPTION.

2239/5; distance from nodus to pterostigma 11.5 mm, and approximately 11.0 mm in PIN 2239/5; distance from pterostigma to wing apex 1.9 mm; postnodal area very long, with at least 21 postnodal cross-veins and 19 postsubnodal crossveins; pterostigma sclerotised, rather long, 2.4 mm long, 0.5 mm wide, 2.1 mm long, 0.6 mm wide in PIN 2997/4433; pterostigma covering four cells, with basal and distal sides of pterostigma oblique and more or less parallel; pterostigmal brace absent; nodal cross-vein and subnodus perpendicular to RA, more or less aligned; RP3/4 beginning one cell distal of subnodus; longitudinal bridge between RP3/4 and distal free part of IR2 moderately long, three to four cells and 1.81.9 mm long, and strictly aligned with the bridge between IR2 and RP2, 1.5 mm long, and followed by a very long zigzagged vein than ends in IR1; IR2 apparently branching on RP3/4, true base of IR2 not clearly individualised; RP2 apparently branching on IR2, true base of RP2 looking like a

M E S O Z O I C P R O T O M Y R M E L E O N T I D A E : NEW TAXA A N D PHYLOGENY

very oblique cross-vein between RP2 and RP, 5.5 mm distal of subnodus; true base of IR1 not individualised, but IR1 never fused with RP2; one to two rows of cells between RP2 and IR1 and one to two rows between IR1 and RP1, three to four rows of cells between RP2 and RP1; RP2 divided into two long convex branches RP2a and RP2b, 0.6 mm distal of its fusion with the bridge between it and IR2; both RP2a and RP2b nearly straight, with two to three rows of cells and one to two zigzagged longitudinal veins between them, and three to four rows of cells in PIN 2239/5; one to two rows of cells between RP2b and IR2; IR2 straight but zigzagged in its very distal part, IR2 rather zigzagged and poorly defined (aberrant feature?) in PIN 2239/5; one row of cells between IR2 and RP3/4a, one to two row of cells in PIN 2239/5; RP3/4 divided into two branches RP3/4a and RP3/4b, with two rows of cells between RP3/4a and posterior wing margin, RP3/4a not zigzagged, nearly straight except at its apex, and rather long, 3.5 mm in length, RP3/4b rather short, 1.1 mm in length; one row of cells between RP3/4 and MA; MA and MP basally straight and apically zigzagged; one to two rows of cells between MP and posterior wing margin, two long rows of cells between MP and posterior wing margin in PIN 2239/5; CuA not independent of MP; AA not independent of AP. DISCUSSION. These wings can be attributed to the genus Protomyrmeleon on the basis of the structure of their radial area, as for both P. grandis sp. nov. and P. karatauensis sp. nov. These four wings are of nearly the same size and they share the presence of a moderately long bridge between RP3/4 and the distal free part of IR2, the numerous postnodal cross-veins, the long pterostigma, the rather narrow area between the two branches of RP2, the long RP3/4a, the presence of three to four rows of cells between RP2 and RP1. They differ in the number of rows of cells below MP, the shape of IR2 (maybe an aberrant feature in PIN 2239/5) and the presence of distinctly more numerous cells in the area between RP2 and RP1 in PIN 2997/4433 and PIN 2239/5 than in the holotype. Thus, we provisionally consider that they belong to the same species. Pending the discovery of further material that will help establish the intra-specific variability in these Odonatoptera. Protomyrmeleon kazakhstanensis sp. nov. clearly differs from P. grandis sp. nov. and P. karatauensis sp. nov. in its smaller wings (17.5-20.0 mm long instead of more than 22.5 mm in the two latter species), but mainly in its distinctly shorter bridge between RP3/4 and the distal free part of IR2. It differs from P. handlirschi in having a longer bridge between RP3/4 and IR2, its pterostigma covering four cells and in the presence of only seven to nine rows of cells between RP2a and RP2b along the posterior wing margin, instead of 11-14. This last character also differentiates it from P. angustivenosus. It differs from both P. pascholdi and P. anglicanus in the presence of three or four rows of cells in the area between RP1 and RP2, instead of two. Protomyrmeleon kazakhstanensis sp. nov. shares with P. cretacicus a moderately long bridge between RP3/4 and a distal free part of IR2 but it has distinctly fewer postnodal cross-veins (21 instead of 28). Furthermore, P. kazakhstanensis clearly differs from all other Protomyrmeleontidae species in its distinctly longer bridge between RP3/4 and in having a distal free part of IR2 (three or four cells long instead of two).

197

Protomyrmeleon pumilio sp. nov. (Fig 6D) In reference to the small size of the wing of this new species. ETYMOLOGY.

Holotype and only known specimen: PIN 2997/1131, a single wing, Paleoentomological Laboratory at the Academy of Science of Russia, Moscow. MATERIAL.

Karabastau Formation, CallovianKimmeridgian or Oxfordian-Kimmeridgian (Zherikhin & Gratshev 1993; Mostovski & Martfnez-Delclos 2000; Rasnitsyn & Zherikhin 2002), Middle-Late Jurassic, Karatau Mountain Ridge near the village of Mikhailovka (Aulie), Chimkent region, southern Kazakhstan. LOCALITY AND HORIZON.

Wing short and rather broad, about 18 mm long; distal free part of IR2 strongly zigzagged; short pterostigma covering three cells; postnodal cross-veins relatively few; area between branches of RP2 narrow, with eight rows of cells along posterior wing margin; short bridge between RP3/4 and distal free part of IR2. DIAGNOSIS.

Wing of type 'B', short and rather broad, length of preserved part 17.7 mm long, 3.6 mm wide; distance from base to nodus about 4.6 mm, from nodus to pterostigma 10.6 mm, from pterostigma to wing apex about 1.6 mm, postnodal area long, with 15 postnodal cross-veins and 14 postsubnodal cross-veins; pterostigma sclerotised, rather long, 1.6 mm long, 0.4 mm wide, covering three cells, with basal and distal sides of pterostigma oblique and more or less parallel; pterostigmal brace absent; nodal cross-vein and subnodus aligned and perpendicular to RA; RP3/4 beginning one cell distal of subnodus; longitudinal bridge between RP3/4 and distal free part of IR2 short, two cells and 1.4 mm long, and strictly aligned with the bridge between IR2 and RP2, 1.1 mm long, and followed by a very long zigzagged vein than ends in IR1; IR2 apparently branching on RP3/4, true base of IR2 not clearly individualised; true base of RP2 not clearly individualised; true base of IR1 looking like an oblique cross-vein between RP2 and RP, 5.0 mm distal of nodus; IR1 never fused with RP2; one row of cells between RP2 and IR1 and between IR1 and RP 1; RP2 divided into two long convex branches RP2a and RP2b; both RP2a and RP2b nearly straight, area between RP2a and RP2b long narrow, with one to three rows of cells between them; one row of cells between RP2b and IR2; free part of IR2 strongly zigzagged; one row of cells between IR2 and RP3/4a; RP3/4 divided into two branches RP3/4a and RP3/4b, with one to two rows of cells between RP3/4a and posterior wing margin, RP3/4a rather short and weakly zigzagged; RP3/4b rather short, 1.0 mm long; one row of cells between RP3/4 and MA; MA and MP basally straight and apically zigzagged; one row of cells between MP and posterior wing margin; CuA not independent of MP; AA not independent of AP.

DESCRIPTION.

DISCUSSION. This fossil has the main synapomorphy of the Protomyrmelontidae, i.e. RP2 secondarily forked into RP2a and RP2b. Protomyrmeleon pumilio sp. nov. differs from the other Protomyrmeleon spp. from Karatau in its rather short wings, short pterostigma covering three cells, less numerous postnodal cross-veins, narrow area between branches of RP2, with eight rows of cells along the posterior wing margin, and short bridge between RP3/4 and distal free part of IR2. It

198

A.

NEL

ETAL

Figure 7 A, Tillyardomyrmeleonpetermilleri Henrotay et at., 1997, holotype In 46119, photograph (scale bar = 5 mm); B, C Triassagrion australiense Tillyard, 1922, holotype counterpart In 93544: B, photograph (scale bar = 5 mm);C, drawing (scale bar = 1 mm).

differs from all the Protomyrmeleon spp. in having a strongly zigzagged distal free part to IR2. Protomyrmeleon sp. indet. (Fig 6E) MATERIAL. PIN 2784/69, the distal half of a wing, Paleoentomological Laboratory at the Academy of Science of Russia, Moscow.

Karabastau Formation, CallovianKimmeridgian or Oxfordian-Kimmeridgian (Zherikhin & Gratshev 1993; Mostovski & Martinez-Delclos 2000; Rasnitsyn & Zherikhin 2002), Middle-Late Jurassic, Karatau Mountain Ridge near the village of Mikhailovka (Aulie), Chimkent region, southern Kazakhstan. LOCALITY AND HORIZON.

never fused with RP2; one row of cells between RP2 and IRl and between IRl and RP1; RP2 divided into two long convex branches RP2a and RP2b; both RP2a and RP2b nearly straight, area between RP2a and RP2b long and narrow, with two rows of cells between them for a long distance and only five cells along posterior wing margin; one row of cells between RP2b and IR2; free part of IR2 zigzagged; one row of cells between IR2 and RP3/4a; RP3/4 divided into two branches RP3/4a and RP3/4b, with one or two rows of cells between RP3/4a and posterior wing margin, RP3/4a rather long and straight, 2.9 mm long; RP3/4b rather short, 0.7 mm long; one row of cells between RP3/4 and MA; MA and MP basally straight and apically zigzagged; one row of cells between MP and posterior wing margin; CuA probably not independent of MP.

DESCRIPTION. Length of preserved part of wing 11.9 mm long, 3.75 mm wide; distance from pterostigma to wing apex DISCUSSION. This fossil differs from all other Protomyrabout 1.8 mm, postnodal area probably long, with 12 pre- meleon spp., except the Liassic specimens LGA 621, LGA served postnodal cross-veins and 10 postsubnodal cross- 988,andLGA 1000 that Ansorge (1996: text-fig. 3) described veins; pterostigma sclerotised, rather long, 2.2 mm long, and figured under the name Protomyrmeleon brunonis, in its 0.5 mm wide, covering two cells, with basal side of pter- very narrow area between the branches of RP2, with only ostigma nearly perpendicular to RA and distal side well ob- five rows of cells along the posterior wing margin. As we lique; pterostigmal brace absent; longitudinal bridge between failed to correctly separate the present fossil from these LiRP3/4 and distal free part of IR2 short, two cells and 1.75 assic wings, we prefer to maintain it in open nomenclature mm long, and strictly aligned with the bridge between IR2 in the genus Protomyrmeleon. and RP2, 1.0 mm long, and followed by a very long zigzagged vein than ends in IRl; IR2 apparently branching on Genus TRIASSAGRION Tillyard, 1922 RP3/4, true base of IRl not clearly individualised but maybe present one cell distal of that of RP2; true base of RP2 look- TYPE SPECIES. Triassagrion australiense Tillyard, 1922 ing like an oblique cross-vein between RP2 and RP; IRl (Figs 3B, 7B, 7C)

M E S O Z O I C P R O T O M Y R M E L E O N T I D A E : NEW TAXA AND PHYLOGENY

199

Late Triassic, Denmark Hill, Ipswich, Queensland, Australia.

LOCALITY AND HORIZON.

Although very complete, there are some small errors in the original description and drawing of Tillyard (1922). We had the opportunity to examine the counterpart of the holotype, which is stored at the British Museum of Natural History (specimen In 33544). We amend the original description as follows: only two rows of cells between RA and IR1, opposite pterostigma; RP3/4 not distally divided into two branches; four rows of cells between IR2 and RP2 along posterior wing margin; AA and AP are completely fused (see Fig. 7C). Some other specimens attributable to T. australiense are also stored at the British Museum, i.e. specimens In 33226 (part and counterpart of two superposed wings), In 33519 (part and counterpart of a wing fragment). REMARKS.

PHYLOGENETIC ANALYSIS OF THE GENERA OF PROTOMYRMELEONTIDAE Nel (1992), Henrotay et al. (1997) and Nel & Jarzembowski (1998) have all previously carried out phylogenetic analyses on the genera of Protomyrmeleontidae. Here we perform a cladistic analysis including all the protomyrmeleontid genera listed above. We have chosen the Archizygoptera genera Kennedy a Tillyard, \925b (Kennedyidae Tillyard, 1925ft) and Terskeja Pritykina, 1981 (Batkeniidae Pritykina, 1981) as outgroups. The former belongs to a potential sister group of Protomyrmeleontoidea (= Protomyrmeleontidae + Batkeniidae). The latter belongs to the sister group of the Protomyrmeleontidae, on the basis of the following synapomorphies: IR1 apparently originating on RP2 or even fused to it; RP2 apparently originating on IR2 or even fused to it; IR2 apparently originating on RP3/4 or even fused to it; MA and RP3/4 distinctly curved towards the posterior wing margin, so that both veins are distinctly shortened (Bechly 1996). Our analysis is based on 18 characters of the wing venation (see Appendices), and was performed using the computer software packages Paup 4.0b. 10 and MacClade 3.08, with the Branch and Bound option. We have tested all combinations of outgroup(s). The results were independent of the combinations and the order in which outgroups are listed in the matrix. The characters were first treated as unordered. The analysis with all characters gives 204 most parsimonious trees (length = 43 steps; consistency index (CI) = 0.5349; consistency index excluding uninformative characters = 0.5122; retention index (RI) = 0.5556; rescaled consistency index (RC) = 0.2972). The strict consensus tree is not resolved, with a basal polytomy between all the protomyrmeleontid genera. Thus, several characters are in conflict. A second analysis with the multistate characters 2, 3,4, 5 and 11 ordered gave one most parsimonious tree (length = 48 steps; CI = 0.4792; CI excluding uninformative characters = 0.4565; RI = 0.5536; RC = 0.2653. Fig. 8A), but with the clades poorly supported. Nearly all characters remain homoplastic, except for character 10 (state 1), a synapomorphy of the Protomyrmeleontidae, and character 12 (state 1), a synapomorphy of the clade {Protomyrmeleon (including Zirzipanagrion), Italomyrmeleon, Malmomyrmeleon, Ferganagrion).

Figure 8 A, Unique most parsimonious cladogram obtained with the multistate characters 2-5 and 11 considered as ordered. B, Strict consensus cladogram of the two most parsimonious trees with characters 6, 8-9 and 13-17 excluded, all characters unordered. C, Unique most parsimonious cladogram obtained with the multistate characters 2-5 and 11 considered as ordered and characters 6,8-9, 13-17 excluded.

We also tested the hypothesis that the main evolutionary trend of the Protomyrmeleontidae concerns their highly specialised organisation of the radial veins. Thus, we excluded all the characters that do not concern these structures,

200

A.

NEL

ETAL

namely characters 6, 8, 9 and 13-17. This resulted in two equally most parsimonious trees (length = 24 steps; CI — 0.6667; CI excluding uninformative characters = 0.6522; RI = 0.7143; RC = 0.4762: Fig. 8B). The various clades are better supported than in previous analyses. Nevertheless, the distribution of states of character 3 is surprising, namely the state 'longitudinal bridge between IR2 and RP2 replaced by a partial fusion between IR2 and RP2' (character 3, state 2) supports the clade (Protomyrmeleontidae minus the basal genus Triassagrion), but its apparently less advanced state 'longitudinal bridge between IR2 and RP2 present, but not replaced by a partial fusion between IR2 and RP2' (character 3, state 1) supports the more terminal clade Protomyrmeleontidae minus {Obotritagrion, Paraobotritagrion, Triassagrion). When the multistate characters were ordered, a unique most parsimonious tree was obtained (length = 28 steps; CI = 0.5714; CI excluding uninformative characters = 0.5556; Rl = 0.6923; RC = 0.3956: Fig. 8C). The same topology was obtained when ordering only characters 3,4 and 11. Under this option, character 3, state 2 is convergent between Paraobotritagrion and {Malmomyrmeleon + (Obotritagrion + Italomyrmeleorij). The topology differs in many points from that of the previous option, especially in the positions of Obotritagrion and Malmomyrmeleon. The Protomyrmeleontidae are currently divided into the two subfamilies Triassagrioninae Tillyard, 1922 (which includes only Triassagrion) and Protomyrmeleontinae Handlirsch, 1906 (other genera). Bechly (1996) denned the Triassagrioninae on the following characters: IR2 basally fused with RP3/4, and secondary presence of cross-veins in the space between RP and MA from arculus to midfork (reversal). The first character is also present in Mongolagrion, Saxomyrmeleon, and Obotritagrion, taxa that are not directly related to Triassagrion. The second character is not easy to establish for several fossils where preservation is poor. Furthermore, it is also present in Malmomyrmeleon. Thus, both these characters are not unambiguous autapomorphies of Triassagrion. Bechly (1996) also characterised the Protomyrmeleontinae by the following three putative synapomorphies: IR1, RP2 and IR2 basally fused, and with their common stalk originating on RP3/4, RP3/4 secondarily forked into RP3/4a and RP3/4b, and CuA and AA completely fused. Tillyardomyrmeleon (Figs 2A, 7A) has veins IR1, RP2 and IR2 well separated. Therefore, the current division of Protomyrmeleontidae in Triassagrioninae and Protomyrmeleontinae is weakly supported. In conclusion, only two groupings are consistently recovered, i.e. the clades Protomyrmeleontidae and (Protomyrmeleon + Zirzipanagrion). The available set of wing characters shows high level of homoplasy and is insufficient to solve the phylogeny of this group. Unfortunately, the body structures of Protomyrmeleontidae species are nearly useless because they are rarely and incompletely preserved. The high diversity in the arrangement of radial veins amongst species of Protomyrmeleontidae (veins IR2, RP2, and IR1 partially or totally fused in bridges, see Figs 5 & 6 and Appendix 1) suggests that the fossil record of this group is very incomplete. It also suggests that the Protomyrmeleontidae had very different modes of flight and consequently occupied a wide range of palaeoenvironments. This point is also supported by the very wide geographical and stratigraphical range of this group, together with their presence in very diverse palaeo-

biotas including shallow tropical marine (islands), fluviatile and lacustrine.

ACKNOWLEDGEMENTS We thank first, Vladimir Zherikhin who unexpectedly died 2 years ago, for his help to A.N. and X.M.-D. when they began this work during their visit to the Moscow Laboratory about 10 years ago. We thank Alex Rasnitsyn and Yuri Popov for information about specimens and the current state of collections. We also thank Dmitri Shcherbakov for the photographs; an anonymous referee and Andrew Ross for their kind comments on the first version of this paper and the loan of specimens from the British Museum of Natural History. This research is a contribution from the Spanish Government Scientific Projects PGC 2001-0173 and PGC 2001-0185.

REFERENCES Ansorge, J. 1996. Insekten aus dem Oberen Lias von Grimmen (Vorpommern, Norddeutschland). Neue Pala'ontologische Abhandlungenl: 1-132. Bechly, G. 1996. Morphologische Untersuchungen am Fliigelgeader der rezenten Libellen und deren Stammgmppenvertreter (Insecta; Pterygota; Odonata), unter besonderer Berucksichtigung der Phylogenetischen Systematik und des Grundplanes der *Odonata. Petalura 2: 1-402. (Special Volume). — 1997. New fossil Odonata from the Upper Triassic of Italy, witharedescription of Italophlebia gervasutti, and a reclassification of Triassic dragonflies. Rivista del Museo Civico di Scienze Naturale E. Cajfi 19: 31-70. Bode, A. 1953. Die Insektenfauna des Ostniedersachsischen oberen Lias. Palaeontographica (A) 103: 1-375. Geinitz, F. E. 1887. Neue Aufschliisse der Flozformation Mecklenburgs. IX. Beitriig zur Geologie Mecklenburgs. IV. Jura. Archiv des Vereins der Freunde Naturgeschichte Mecklenburg 41: 194-208. Handlirsch, A. 1906-1908. Die fossilen Insekten und die Phylogenie der rezenten Formen. Ein Handbuchfiir Paldontologen undZoologen. Engelman, V. W., Leipzig, 1430 pp. (Published in parts between 1906 and 1908 as follows: pp. i-vi, 1-160, pis. 1-9 (May 1906); pp. 161320, pis. 10-18 (June 1906); pp. 321^80, pis. 19-27 (August 1906); pp. 481-640, pis. 28-36 (October 1906); pp. 641-800, pis. 37-45 (February 1907); pp. 801-960, pis. 46-51 (June 1907); pp. 961-1120 (November 1907); pp. 1121-1280 (January 1908); pp. vii-ix, 12811430 (July 1908)). Henrotay, M., Nel, A. & Jarzembowski, E. A. 1997. New Protomyrmeleontidae damselflies from the Triassic of Australia and the Liassic of Luxembourg. Odonatologica 26: 395-404. Martinez-Delclos, X. & Nel, A. 1996. Discovery of a new Protomyrmeleontidae in the Upper Jurassic from Germany (Odonatoptera, Odonata, Archizygoptera). Archaeopteryx 14: 67-73. Martynov, A. V. 1927. Jurassic fossil insects from Turkestan. 7. Some Odonata, Neuroptera, Thysanoptera. Bulletin de I'Academie des Sciences de I' URSS, Classe des Sciences Mathematiques et Nature lies 20: 757-768. Mostovski M. B. & Martinez-Delclos, X. 2000. New Nemestrinoidea (Diptera: Brachycera) from the Upper Jurassic - Lower Cretaceous of Eurasia, taxonomy and palaeobiology. Entomological Problems 31: 137-148. Nel, A. 1992. Redescription de la libellule fossile du Jurassique superieur"? Malmagrion eichstaettense (Hagen, 1862)" (Odonatoptera, Odonata, Archizygoptera). Bulletin de la Societe Entomologique deFrance 96: 433^42.

M E S O Z O I C P R O T O M Y R M E L E O N T I D A E : NEW TAXA A N D PHYLOGENY

— & Henrotay, M. 1992. Les Protomyrmemelontidae (Odonatoptera, Odonata, Archizygoptera stat. rest.): etat actuel des connaissances. Annales de Paleontologie (Vertebres - Invertebres) 78: 1—47. — & Jarzembowski, E. A. 1998. New protomyrmeleontid dragonflies from the Lower Cretaceous of southern England. (Insecta, Odonata, Archizygoptera). Cretaceous Research 19: 393—402. —, Martinez-Delclos, X., Paicheler, J.-C & Henrotay, M. 1993. Les "Anisozygoptera" fossiles. Phylogenie et classification. (Odonata). Martinia, Bois d'Arcy, numero hors serie 3: 1—311. Pritykina, L. N. 1981. [New Triassic Odonata of middle Asia.] Pp. 5—42 in Vishniakova, V. N., Dlussky, G. M. & Pritykina, L. N. (eds) [New insects from the territory of the U.S.S.R.]. Trudy Paleontologiceskogo Instituta Akademii nauk S.S.S.R. 183. [In Russian]. Rasnitsyn, A. P. & Zherikhin, V. V. 2002. 4. Appendix: alphabetic list of selected insect fossil sites. 4.1. Impression fossils. Pp. 437-444 in Rasnitsyn, A. P. & Quicke, D. L. J. (eds) History of insects. Kluwer Academic Publishers, Dordrecht, Boston, London. Riek, F. & Kukalova-Peck, J. 1984. A new interpretation of dragonfly wing venation based upon early Upper Carboniferous fossils from Argentina (Insecta: Odonatoidea) and basic character states in pterygote wings. Canadian Journal of Zoology 62: 1150-1166. Tillyard, R. J. 1922. Mesozoic insects of Queensland. 9. Orthoptera and addition to the Protorthoptera, Odonata, Hemiptera and Planipennia. Proceedings of the Linnean Society of New South Wales 47: 447-470. — 1925a. The British Liassic dragonflies. British Museum (Natural History), Fossil Insects, London 1: 1-39. — 1925/). Kansas Permian insects. 5. The orders Protodonata and Odonata. American Journal of Science 9: 40-73. Zessin, W. 1991. Die Phylogenie der Protomyrmeleontidae unter Einbeziehung neuer oberliassischer Funde (Odonata: Archizygoptera sens, nov.). Odonatologica 20: 97-126. Zherikhin, V. V. & Gratshev, V. G. 1993. Obrieniidae, fam. nov., the oldest Mesozoic weevils (Coleoptera, Curculionoidea). Paleontological Journal 27: 50-69.

APPENDIX 1: LIST OF CHARACTERS USED IN THE PHYLOGENETIC ANALYSIS For abbreviations used, see the text. 1. Base of IR1: distinctly oblique (0); the original branching of IR1 on RP is indistinct (1). 2. Base of RP2: the initial branching of RP2 on RP is present (0); the initial branching of RP2 on RP looks like an oblique cross-vein (1); the initial branching of RP2 on RP is completely indistinct (2). 3. Longitudinal bridge between IR2 and RP2: absent (0); present (1); replaced by a partial fusion between IR2 and RP2 (2). 4. Longitudinal bridge between RP3/4 and 1R2: absent, IR2 being basally completely independent of RP3/4 (0); present (1); replaced by a partial fusion between IR2 and RP3/4 (2). 5. RP3/4: not divided into two well-defined branches (0); divided into two well-defined branches RP3/4a and RP3/4b, but RP3/4a is weak and more or less zigzagged (1); RP3/4a is strong, not zigzagged (2).

201

6. Base of RP: nearer to MAb than to base of RP + MA (0); midway between MAb than to base of RP + MA (1). 7. RP3/4: branching normally on RP (0); apparently branching on MA (1). 8. AA and AP: are independent (0); are completely fused (1). 9. Number of postnodal cross-veins: 15 or less (0); more than 15 (1). 10. RP2: not divided into two branches (0); divided into two branches (1). 11. IR1: not prolonged by a vein basal of its base (0); apparently prolonged by a vein basal of distal free part of RP2 but distal of base of IR2 (two rows of cells in area between RP2 and RP) (1); apparently prolonged by a vein basal of distal free part of IR2 (two rows of cells in area between IR2 and RP) (2); IR1 at least partly fused with RP2 (3); IR1 at least partly fused with IR2 (4). 12. Pseudo-IR 1: absent (0); present (1). 13. Pterostigma: very elongate, with its basal side perpendicular to RA and C and its distal side distinctly oblique (0); shorter, with its basal and distal sides more or less parallel and oblique (1). 14. Area between MA, RP and base of RP3/4: with crossveins (0); without cross-veins (1). 15. Number of rows of cells in area between RP2 and RP1 opposite pterostigma: three or less (0); more than three (1). 16. CuA: free (0); never free (1). 17. Area between MP and posterior wing margin: with one row of cells (0); with more than one row of cells (1). 18. Base of IR2: the initial branching of IR2 on RP is present (0); the initial branching of IR2 on RP looks like an oblique cross-vein (1); the initial branching of IR2 on RP is completely indistinct (2).

APPENDIX 2: DATA MATRIX