1. Systematic palaeontology

Order Odonata Fabricius, Reference Fabricius1793

Suborder Anisoptera Selys in Selys & Hagen, Reference Selys-Longchamps and Hagen1854

Superfamily Aeshnoidea Leach, Reference Leach and Brewster1815

Family Gomphaeschnidae Tillyard & Fraser, Reference Tillyard and Fraser1940 (sensu Bechly Reference Bechly1996)

Genus Oligoaeschna Selys, Reference Selys-Longchamps1889

Oligoaeschna? anglica Cockerell & Andrews, Reference Cockerell and Andrews1916

(Plate 1, figs 1–4, text-figs 1–3)

1. 1916 Oligoaeschna anglica Cockerell & Andrews, p. 89, pl. 2, fig. 1.

2. 1986 Oligoaeschna anglica Wighton & Wilson, pp. 507, 515.

3. 1994 Oligoaeschna(?) anglica Nel et al., p. 163.

4. 1999 ‘Oligoaeschna’ anglica Nel & Jarzembowski, p. 198.

5. 2005 ‘Oligoaeschna’ anglica Nel et al., pp 150–152, fig. 3.

Plate 1 (1–4) Oligoaeschna? anglica Cockerell & Andrews, Reference Cockerell and Andrews1916, holotype: (1) part, NHMUK I.8649. Scale bar=3 mm; (2) counterpart, NHMUK In.17266. Scale bar=3 mm; (3) part, NHMUK In.17353. Scale bar=3 mm; (4) counterpart, NHMUK In.17353. Scale bar=3 mm. (5–6) Anglogomphaeschna eocenica gen. et sp. nov.: (5) holotype, NHMUK In.8793, forewing. Scale bar=10 mm; (6) paratype, NHMUK In.17368, hindwing, male. Scale bar=10 mm.

Text-figure 1 Oligoaeschna? anglica Cockerell & Andrews, Reference Cockerell and Andrews1916, holotype, NHMUK I.8649. Scale bar=3 mm.

Text-figure 2 Oligoaeschna? anglica Cockerell & Andrews, Reference Cockerell and Andrews1916, NHMUK In.17353. Scale bar=3 mm.

Text-figure 3 Oligoaeschna? anglica (?), NHMUK In.24756, forewing. Scale bar=3 mm.

Holotype. NHMUK I.8649/In.17266 (Brodie/Smith coll., Plate 1, figs 1–2, Text-fig. 1), Insect Limestone, northwest Isle of Wight.

Other material. Specimen NHMUK In.17353 (Smith coll., Plate 1, figs 3–4, Text-fig. 2) is a more complete wing apex that we attribute to the same species. Nel et al. (Reference Nel, Petrulevičius and Jarzembowski2005) described a nearly complete hindwing.

The original description of Cockerell & Andrews (Reference Cockerell and Andrews1916, p. 89, pl. 2, fig. 1) was based on the holotype. A. J. Ross found the counterpart, which is more complete. The Recent genera Basiaeschna Selys, Reference Selys-Longchamps1883, Oligoaeschna Selys, Reference Selys-Longchamps1889 and Linaeschna Martin, Reference Martin1908 have very similar wing apices, with a posteriorly curved Rspl, two rows of cells between Rspl and IR2, IR2 unforked and weakly curved, three rows of cells between IR2 and RP2, RP2 with a distinct curve opposite the base of pterostigma, a short IR1, and a pterostigma covering 2–3 cells (Martin Reference Martin1908). The hindwing described by Nel et al. (Reference Nel, Petrulevičius and Jarzembowski2005) is very similar to that of a Recent Oligoaeschna, but the generic attribution of this species remains uncertain because of the lack of body characters (see also Nel & Papazian Reference Nel and Papazian1983; Nel et al. Reference Nel, Martinez-Delclòs, Escuillié and Brisac1994).

Specimen NHMUK I.8934/In.24756 (part and counter-part, Brodie/Hooley coll., Text-fig. 3) is the basal two thirds of a forewing that could also belong to the same species, because its dimensions are compatible with those of the wing apices attributed to Oligoaeschna? anglica, and its wing venation is consistent with the genus Oligoaeschna.

Genus Anglogomphaeschna gen. nov.

Type species. Anglogomphaeschna eocenica sp. nov.

Etymology. After Anglia, Latin name for England and Gomphaeschna.

Diagnosis. Wing characters only; antenodal cross-veins numerous; ‘cordulegastrid gap’ present; median space free; space between MP and CuA nearly as wide at basal third as at distal third in hindwing; MA and RP3/4 parallel to each other up to wing margin, MA without concave bend; Mspl and Rspl nearly parallel to MA and IR2 respectively; AA & AP and AA2 parallel before anal angle; anal triangle present; cubito-anal area broad; anal loop large and broad, transverse; two cross-veins in submedian area; discoidal triangle divided into five cells in forewing and three cells in hindwing.

Anglogomphaeschna eocenica sp. nov.

(Plate 1, figs 5–6, 2, figs 1–3; Text-figs 4–7)

Text-figure 4 Anglogomphaeschna eocenica gen. et sp. nov., holotype, NHMUK I.8793, forewing. Scale bar=3 mm.

Text-figure 5 Anglogomphaeschna eocenica gen. et sp. nov., paratype, NHMUK In.17368, hindwing, male. Scale bar=3 mm.

Text-figure 6 Anglogomphaeschna eocenica gen. et sp. nov., paratype, NHMUK In.17330, hindwing, female. Scale bar=5 mm.

Text-figure 7 Anglogomphaeschna eocenica gen. et sp. nov., paratype, NHMUK In.24649, hindwing. Scale bar=3 mm.

Holotype. Specimen NHMUK I.8793 (Brodie coll., base of forewing, Plate 1, fig. 5, Text-fig. 4), Insect Limestone, northwest Isle of Wight.

Paratypes. NHMUK In.17368 (Smith coll., base of male hindwing, Plate 1, fig. 6, Text-fig. 5), NHMUK In.17330 (Smith coll., base of female hindwing, Plate 2, fig. 1, Text-fig. 6), NHMUK In.24649 (base of a hindwing, Text-fig. 7), NHMUK I.10390/In.24652(1) (base of a hindwing, on same piece as scarabaeid beetle abdomen, Plate 2, figs 2–3).

Plate 2 (1–3) Anglogomphaeschna eocenica gen. et sp. nov.: (1) paratype, NHMUK In.17330, hindwing. Scale bar=10 mm; (2) paratype, NHMUK In.24652, hindwing, part. Scale bar=10 mm; (3) paratype, NHMUK In.10390, counterpart. Scale bar=10 mm. (4–7) ‘Oplonaeschna’ vectensis Cockerell & Andrews, Reference Cockerell and Andrews1916: (4) holotype, NHMUK I.8572. Scale bar=5 mm; (5) paratype, NHMUK I.8590, hindwing, part, probably related to Oligaeschna wedmanni. Scale bar=5 mm; (6) paratype, NHMUK In.17369, counterpart, probably related to Oligaeschna wedmanni. Scale bar=5 mm. (7) paratype, NHMUKI.8636, probably related to Oligaeschna wedmanni. Scale bar=5 mm.

Etymology. After the Eocene period.

Diagnosis. As for genus.

Description. NHMUK I.8793. Forewing 14·3 mm wide, distance from wing base to arculus 8·1 mm, from arculus to nodus 25·8 mm, from base to Ax1 5·0 mm, from Ax1 to arculus 2·8 mm, from arculus to Ax2 4·8 mm; no secondary antenodal cross-vein basal of Ax1, four between Ax1 and Ax2, 14 distal of Ax2; 14 cross-veins between RA and RP basal of subnodus, with last cross-vein missing (‘cordulegastrid gap’); median area free; submedian+subdiscoidal area crossed by two veins; hypertriangle longer than anterior side of discoidal triangle, with three cross-veins; discoidal triangle elongate, with anterior side 8·3 mm long, posterior side 7·6 mm long, basal side 2·8 mm long, divided into five cells; two rows of cells in anal area; four rows of cells in cubito-anal area; postdiscoidal area with two rows of cells distal of triangle; Mspl straight, with two rows of cells between it and MA; opposite base of Rspl two rows of cells between MA and RP3/4, but these veins remain parallel; oblique vein just distal of base of RP2.

NHMUK In.17330. Hindwing. Distance from wing base to arculus 7·3 mm, from arculus to nodus 15·3 mm, from base to Ax1 4·6 mm, from Ax1 to arculus 1·4 mm, from arculus to Ax2 3·8 mm; no secondary antenodal cross-vein basal of Ax1, two between Ax1 and Ax2, eight distal of Ax2; six cross-veins between RA and RP basal of subnodus, with last cross-vein missing (‘cordulegastrid gap’); median area free; submedian+subdiscoidal area crossed by two veins; hypertriangle longer than anterior side of discoidal triangle, with at least one cross-vein; discoidal triangle elongate, with anterior side 6·0 mm long, posterior side 5·0 mm long, basal side 2·8 mm long, divided into three cells; space between MP and CuA nearly as wide at basal third as at distal third, with one row of cells; anal loop transverse, clearly defined, divided into five cells; cubito-anal area with at least five rows of cells between CuA and posterior wing margin; no anal angle and no anal triangle (female); 13 postnodal cross-veins; two rows of cells between Mspl and MA and Rspl and IR2; Mspl and Rspl straight.

NHMUK In.17368 mainly differs from In.17330 in the presence of an anal triangle and anal angle (male).

Discussion. The attribution of these forewing and hindwing fragments to the same species is based on the following characters: similar wing length; the differences between NHMUK I.8793 (forewing basal half) and NHMUK In.17330, NHMUK In.17368 and NHMUK In.24649 (hindwing basal halves) can be attributed to the normal differences between the forewings and the hindwings within the same species. They share in particular the same Mspl, distal part of area between MA and RP3/4; costal and subcostal area, median and submedian spaces. Anglogomphaeschna gen. nov. has a ‘cordulegastrid’ gap, viz. the distal part of antesubnodal area between RA and RP is free of cross-vein, which is an apomorphy of the clade Gomphaeschnidae, more pronounced in the numerous Mesozoic taxa, in the Paleocene genus Alloaeschna Wighton & Wilson, Reference Wighton and Wilson1986, Gomphaeschna Selys, Reference Selys-Longchamps1871 and Sarasaeschna Karube & Yeh, Reference Karube and Yeh2001 than in Oligoaeschna, and absent in Linaeschna Martin, Reference Martin1908 (Wighton & Wilson Reference Wighton and Wilson1986; Bechly et al. Reference Bechly, Nel, Martínez-Delclòs, Jarzembowski, Coram, Martill, Fleck, Escuillié, Wisshak and Maisch2001, Ellenreider Reference Ellenrieder2002). Anglogomphaeschna differs from the Mesozoic Paramorbaeschna Bechly et al., Reference Bechly, Nel, Martínez-Delclòs, Jarzembowski, Coram, Martill, Fleck, Escuillié, Wisshak and Maisch2001, Progomphaeschnaoides Bechly et al., Reference Bechly, Nel, Martínez-Delclòs, Jarzembowski, Coram, Martill, Fleck, Escuillié, Wisshak and Maisch2001, Plesigomphaeschnaoides Bechly et al., Reference Bechly, Nel, Martínez-Delclòs, Jarzembowski, Coram, Martill, Fleck, Escuillié, Wisshak and Maisch2001, Gomphaeschnaoides Carle & Wighton, Reference Carle, Wighton and Grimaldi1990 and all the Recent gomphaeschnid genera in its larger and distinctly transverse anal loop. It differs from Sinojagoria Bechly et al., Reference Bechly, Nel, Martínez-Delclòs, Jarzembowski, Coram, Martill, Fleck, Escuillié, Wisshak and Maisch2001 in its forewing discoidal triangle divided into five cells. It differs from Alloaeschna in its more numerous antenodal cross-veins (21 instead of 13).

Anglogomphaeschna differs from all the genera of the ‘Boyeria’ clade and the ‘Brachytron’ clade (sensu Ellenreider Reference Ellenrieder2002) in a series of characters depending on the genus, i.e. median space free; submedian space with only two cross-veins; space between MP and CuA nearly as wide at basal third as at distal third in hindwing; cubito-anal area broad.

Among the genera included within the Aeshnoidea, but still of uncertain position, Austroaeschna Selys, Reference Selys-Longchamps1883, Acanthaeschna Selys, Reference Selys-Longchamps1883 and Planaeschna McLachlan, Reference McLachlan1896 also have a relatively similar wing venation, but they differ from Anglogomphaeschna in the absence of the ‘cordulegastrid’ gap, narrower cubito-anal area, less transverse anal loop, different anal triangle and presence of four cross-veins in submedian area. Antipodophlebia Fraser, Reference Fraser1960 and Telephlebia Selys, Reference Selys-Longchamps1883 have their median space crossed. Austrophlebia Tillyard, Reference Tillyard1916 has a very well-defined fork of IR2.

Lastly, Anglogomphaeschna differs from the third clade of Ellenreider (Reference Ellenrieder2002) (Aeshninae clade) in its MA without concave bend, and Mspl and Rspl straight.

Family Aeshnidae Leach, Reference Leach and Brewster1815

Genus undetermined

‘Oplonaeschna’ vectensis Cockerell & Andrews, Reference Cockerell and Andrews1916

(Plate 2, figs 4–7, 3, fig. 1; Text-figs 8–9)

1. 1916 Oplonaeschna vectensis Cockerell & Andrews, pp. 89–90, pl. 2 figs 2–4.

2. 1986 Oplonaeschna vectensis Wighton & Wilson, p. 508.

3. 1994 Oplonaeschna vectensis Nel et al., pp. 170, 174.

4. 1999 ‘Oplonaeschna’ vectensi' Nel & Jarzembowski, p. 198.

Text-figure 8 ‘Oplonaeschna’ vectensis Cockerell & Andrews, Reference Cockerell and Andrews1916, holotype, NHMUK I.8572. Scale bar=3 mm.

Text-figure 9 ‘Oplonaeschna’ vectensis Cockerell & Andrews, Reference Cockerell and Andrews1916, paratype, NHMUK In.17334 (conterpart), probably related to Oligaeschna wedmanni. Scale bar=4 mm.

Holotype. Specimen NHMUK I.8572 (Brodie coll., Plate 2, fig. 4, Text-fig. 8), Insect Limestone, northwest Isle of Wight. The type specimen of this species described by Cockerell & Andrews (Reference Cockerell and Andrews1916, pp 89–90, pl. 2, fig. 2) is too fragmentary to be accurately attributed to a precise genus, according to Nel & Jarzembowski (Reference Nel and Jarzembowski1999). Furthermore, the paratype NHMUK I.8590/In.17369 (Brodie/Smith coll., Plate 2, figs 5–6) (figured in Cockerell & Andrews Reference Cockerell and Andrews1916, pp 89–90, pl. 2, fig. 3) is a costo-basal part of a hindwing that has strictly no structure in common with the holotype. Its attribution to the same species is impossible to justify (Plate 5, figs 2–3). Also the attribution of paratype NHMUK I.8636/In.17334 (Brodie/Smith coll., Plate 2, fig. 7, Plate 3, fig. 1, Text-fig. 9) (figured in Cockerell & Andrews Reference Cockerell and Andrews1916, pp 89–90, pl. 2, fig. 4) to the same species as the holotype cannot be demonstrated. This species is related to Oligaeschna wedmanni sp. n. (see below).

Plate 3 (1) ‘Oplonaeschna’ vectensis Cockerell & Andrews, Reference Cockerell and Andrews1916, paratype, NHMUK In.17334, probably related to Oligaeschna wedmanni. Scale bar=5 mm. (2) Aeschnophlebia andreasi Nel et al., Reference Nel, Petrulevičius and Jarzembowski2005, NHMUK In.17395, part. Scale bar=5 mm. (3–4) Oligaeschna wedmanni sp. nov.: (3) holotype, IWCMS 2012.576, forewing. Scale bar=10 mm; (4) paratype, NHMUK I.10284. Scale bar=10 mm. (5–6) Neophya legrandi sp. nov.: (5) holotype, NHMUK In.24606, hindwing. Scale bar=5 mm; (6) paratype, NHMUK II.2851a. Scale bar=5 mm. (7) Neophya legrandi sp. nov., paratype, NHMUK In.24641. Scale bar=5 mm.

Clade ‘Brachytron’ (sensu Ellenrieder Reference Ellenrieder2002)

Genus Aeschnophlebia Selys, Reference Selys-Longchamps1883

Aeschnophlebia andreasi Nel et al., Reference Nel, Petrulevičius and Jarzembowski2005

(Plate 3, fig. 2; Text-figs 10–11)

1. 2005 Aeschnophlebia andreasi Nel et al., pp 348–350, fig. 2

Text-figure 10 Aeschnophlebia andreasi Nel et al., Reference Nel, Petrulevičius and Jarzembowski2005, NHMUK In.17371. Scale bar=2 mm.

Text-figure 11 Aeschnophlebia andreasi Nel et al., Reference Nel, Petrulevičius and Jarzembowski2005, NHMUK In.17395. Scale bar=2 mm.

Holotype. Specimen IWCMS 2004.52, Insect Limestone, northwest Isle of Wight.

Other material. NHMUK In.17371 (Text-fig. 10) and NHMUK In.17395 (Plate 3, fig. 2, Text-fig. 11) (both wing apices, Smith coll.).

Description of the new specimens. NHMUK In.17371. Pterostigma 5·4 mm long, 1·0 mm wide, covering six cells; pterostigmal brace oblique; postnodal cross-veins numerous, more than 24; IR1 base in a very basal position, 10 cells basal of pterostigma; RP2 with a weak curve just basal of level of pterostigmal brace; IR2 with a distinct but weak anterior branch, three rows of cells between branches of IR2; Rspl straight, with two rows of cells between it and IR2.

Discussion. The attribution of NHMUK In.17395 (forewing apex) to the same species is based on its great similarity to NHMUK In.17371, except for the shorter basal part of IR1. The structures and dimensions of these two wing apices are nearly identical to those of the holotype (Nel et al. Reference Nel, Petrulevičius and Jarzembowski2005).

Clade ‘Aeshninae’ (sensu Von Ellenrieder Reference Ellenrieder2002)

Genus Oligaeschna Piton and Théobald, Reference Piton and Théobald1939

Oligaeschna wedmanni sp. nov.

(Plate 3, figs 3–4; Text-figs 12–13)

Text-figure 12 Oligaeschna wedmanni sp. nov., holotype, IWCMS 2012.576, forewing. Scale bar=4 mm.

Text-figure 13 Oligaeschna wedmanni sp. nov., paratype, NHMUK I.10284. Scale bar=3 mm.

Holotype. Specimen IWCMS 2012.576 (forewing, Plate 3, fig. 3, Text-fig. 12), Insect Limestone, northwest Isle of Wight.

Paratype. Specimen NHMUK I.8636/In.17334 (forewing?, paratype of ‘Oplonaeschna’ vectensis). A further specimen NHMUK I.10284 (Brodie coll., forewing, Plate 3, fig. 4, Text-fig. 13) and the paratype of ‘Oplonaeschna’ vectensis NHMUK I.8590/In.17369 probably belong to the same species.

Etymology. Named after our colleague Dr Sonja Wedmann.

Diagnosis. Large wing; five rows of cells between RP2 and IR2 along wing margin; pterostigma covering three cells; one row of cells between MA and RP3/4 from the level of base of Rspl and posterior wing margin.

Description. Forewing hyaline, 50·0 mm long, 10·8 mm wide; distance from base to arculus 5·8 mm, from arculus to nodus 19·2 mm, from nodus to pterostigma 15·8 mm, from pterostigma to wing apex 5·7 mm; pterostigma 4·7 mm long, 0·8 mm wide, covering three cells; pterostigmal brace distinctly oblique; one antenodal cross-vein basal of Ax1, Ax1 is 4·6 mm from wing base and 1·7 mm from arculus, Ax2 3·7 mm distal of arculus; five antenodal cross-veins between Ax1 and Ax2; at least 15 secondary antenodal cross-veins distal of Ax2; 17 postnodal cross-veins; median space free of cross-veins; submedian space+subdiscoidal space crossed by five cross-veins (including CuP); hypertriangle 7·7 mm long, 0·8 mm wide, with four cross-veins; anterior side of discoidal triangle 6·3 mm long, posterior side 5·5 mm, basal side 2·5 mm, divided into five cells; 2–3 rows of cells in postdiscoidal area distal of triangle; Mspl curved with 3–4 rows of cells between it and MA; area between CuA and posterior wing margin 2·8 mm wide, with four rows of cells; MA with a slight curve opposite base of Rspl; at this level and for a distance of three cells, two rows of cells between MA and RP3/4, then one row until the posterior wing margin; Rspl strongly curved, with four rows of cells between it and IR2; IR2 with a distinct posterior curve; three rows of cells between IR2 and RP2 in the broadest part and five rows of cells along posterior wing margin; RP2 with a distinct but smooth curve below pterostigma.

Discussion. The attribution of the two wings IWCMS 2012.576 and NHMUK In.17334 to the same species is based on their nearly identical wing apices and similar dimensions. The possible attribution of NHMUK I.10284 is based on the very similar wing bases.

This fossil species clearly belongs to Aeshnidae and to the clade ‘Aeshninae’ sensu Ellenrieder (Reference Ellenrieder2002, fig. 20) after combination of the following apomorphic characters: MA and RP3/4 not parallel, MA with a marked concave bend; Mspl with a concave bend in its distal portion (homoplasy); Rspl not parallel to IR2, with a concave bend. It can be attributed to the group of the three genera Oligaeschna, Kvacekia Prokop & Nel, Reference Prokop and Nel2002 and Oplonaeschna Selys, Reference Selys-Longchamps1883 on the basis of the following characters: pterostigmal brace clearly oblique; median space free; areas between Rspl and IR2 and between Mspl and MA with four rows of cells; absence of very strong bend of RP2; fork of IR2 absent. Oligaeschna and Oplonaeschna differ in the length of pterostigma, distinctly longer in the former than in the latter, and structure of discoidal triangle, longer and divided into five or more cells in Oligaeschna (except for O. saurai Peñalver et al., Reference Peñalver, Nel and Martínez-Delclòs1996), instead of four in Oplonaeschna (González-Soriano & Novelo Gutierrez Reference González-Soriano and Novelo Gutierrez1998). Kvacekia differs from Oligaeschna and Oplonaeschna in the pterostigma covering 5–6 cells, and cubito-anal area of forewing with more than nine rows between CuA and posterior wing margin (Prokop & Nel Reference Prokop and Nel2002). On the basis of the combination of the above-mentioned characters we attribute our fossil to Oligaeschna.

All the fossil species currently attributed to the Recent genus Oplonaeschna are incertae sedis (Prokop et al. Reference Prokop, Přikryl, Dostál and Nel2007 and see above). The genus Oligaeschna comprises seven described species from Oligocene and Miocene Eurasian and North American deposits. O. jungi Piton & Théobald, Reference Piton and Théobald1939 (Oligocene of Puy-de-Dôme, France, Oligocene of Enspel, Germany, Early Oligocene of Provence, France, Wedmann Reference Wedmann2000; Nel et al. Reference Nel, Petrulevičius and Jarzembowski2005) and O. palaeocoerulea (Timon-David Reference Timon-David1946) (Oligocene, Bouches-du-Rhône, France) have circa ten rows of cells between RP2 and IR2 along wing margin instead of the five in O. wedmanni sp. n., and two rows of cells between MA and RP3/4 from the level of base of Rspl and posterior wing margin, as in O. lapidaria (Cockerell & Counts in Cockerell Reference Cockerell1913) from the Late Eocene of Florissant (Colorado, USA) (Cockerell Reference Cockerell1913). O. moravica Prokop et al., Reference Prokop, Přikryl, Dostál and Nel2007 (Middle Oligocene of northern Moravia), O. separata (Scudder Reference Scudder1890) (Florissant), and O. ashutasica (Martynov Reference Martynov1929) from the Paleocene to Oligocene of Ashutas mount (East Kazakhstan), differ from O. wedmanni in the presence of five cross-veins between RA and RP1 below pterostigma (Scudder Reference Scudder1890; Martynov Reference Martynov1929; fig. 4; Prokop et al. Reference Prokop, Přikryl, Dostál and Nel2007). O. saurai Peñalver et al., Reference Peñalver, Nel and Martínez-Delclòs1996, from the Early Miocene of Ribesalbes (Castellón, Spain), is a distinctly smaller species than all the other species (hindwing less than 30 mm long), and has two rows of cells between MA and RP3/4 from the level of base of Rspl and posterior wing margin.

Subfamily ‘Gynacanthinae’ Cockerell, Reference Cockerell1913

‘Gynacanthinae’ species A

(Text-figs 14–15)

Text-figure 14 ‘Gynacanthinae’ species A, NHMUK In.24638. Scale bar=3 mm.

Text-figure 15 ‘Gynacanthinae’ species A, NHMUK In.24751. Scale bar=3 mm.

Material. NHMUK In.24638 (Hooley coll., Text-fig. 14) and NHMUK In.24751 (Hooley coll., Text-fig. 15), Insect Limestone, northwest Isle of Wight.

Description. It is not possible to decide if these fossils are fore- or hindwings. Wing hyaline; length of fragment NHMUK In.24638 30·0 mm, width of wing 17·5 mm; MA and RP3/4 parallel to each other up to wing margin, with two rows of cells between them distal of nodus level; Rspl with a strong curve, four rows of cells between Rspl and IR2, IR2 with a very strong fork; six cells basal of pterostigma, area between the branches of IR2 with 3–4 rows of cells, RP2 with a strong curve opposite base of pterostigma; one row of cells between IR2 and RP2; pterostigma 5·0 mm long, 1·0 mm wide, rather long, covering three cells; pterostigmal brace distinctly oblique and long; postnodal and postsubnodal cross-veins numerous.

Discussion. These two fossils can be attributed to the same species because their common wing venation structures are identical. Few Recent genera have a similar wing venation, viz. the ‘gynacanthine’ Tetracanthagyna Selys, Reference Selys-Longchamps1883, Triacanthagyna Selys, Reference Selys-Longchamps1883, Plattycantha Förster, Reference Förster1908, Gynacantha Rambur, Reference Rambur1842, Neuraeschna Hagen, Reference Hagen1867, Agyrtacantha Lieftinck, Reference Lieftinck1937 and Staurophlebia Brauer, Reference Brauer1865: MA and RP3/4 parallel to each other up to wing margin, with two rows of cells between them distal of nodus level; Rspl with a strong curve, four rows of cells between Rspl and IR2, IR2 with a very strong fork basal of pterostigma, area between the branches of IR2 with 3–4 rows of cells, pterostigma rather long (Martin Reference Martin1908; Lieftinck Reference Lieftinck1937; Ellenreider Reference Ellenrieder2002). Neuraeschna and Staurophlebia could be excluded because their space between the branches of IR2 is narrower and the area between Rspl and IR2 is broader than in our fossil. The Recent Gynacantha also have a narrower space between the branches of IR2, and there is only one row of cells between MA and RP3/4. Agyrtacantha has a distinctly shorter pterostigma, its fork of IR2 is in a more basal position, and it has only one row of cells between MA and RP3/4. Tetracanthagyna would differ from our fossil in the distinctly broader areas between Rspl and IR2 and between the branches of IR2. All these differences are weak and are not sufficient to exclude one of these genera. It would be necessary to have information on the basal part of the wings of this fossil species. At this stage, it is not possible to attribute it to a precise ‘gynacanthine’ genus. Nevertheless, this fossil species is of great interest because all its closest living relatives live in warm to hot climates in Indo-Malaysia, South America and Australasia.

Very few fossils have been attributed to the ‘Gynacanthinae’ (Nel et al. Reference Nel, Martinez-Delclòs, Escuillié and Brisac1994). Gynacantha polydore (Heer Reference Heer and Englemann1849) from the Miocene of Oeningen (Germany) needs revision. Martynov (Reference Martynov1927) described Heliaeschna parallela from the Miocene of Azerbaijan. Nel & Petrulevičius (Reference Nel and Petrulevičius2010) synonymised the fossil genus Epacantha Martynov, Reference Martynov1929 (one species E. magnifica from the late Oligocene of Kazakhstan) with the Recent genus Epiaeschna Hagen, Reference Hagen1875. Lastly, the fossil genus Necracantha Martynov, Reference Martynov1929 (N. composita Martynov, Reference Martynov1929 and N. proxima Martynov, Reference Martynov1929 from the Late Oligocene of Ashutas mount (East Kazakhstan)) differs from our fossil in the straight Rspl and very long pterostigma covering six cells or more (Martynov Reference Martynov1929). Necracantha should be revised, because its wing venation has strong similarities with Epiaeschna.

Clade Cavilabiata Bechly, Reference Bechly1996

Family Cordulephyidae Tillyard, Reference Tillyard1917

Subfamily Neophyinae Tillyard & Fraser, Reference Tillyard and Fraser1940

Genus Neophya Selys, Reference Selys-Longchamps1881

Neophya legrandi sp. nov.

(Plate 3, figs 5–7; Text-figs 16–18)

Text-figure 16 Neophya legrandi sp. nov., paratype, NHMUK II.2851a. Scale bar=3 mm.

Text-figure 17 Neophya legrandi sp. nov., paratype, NHMUK In.24641. Scale bar=3 mm.

Text-figure 18 Neophya legrandi sp. nov., paratype, NHMUK In.24757. Scale bar=2 mm.

Holotype. Specimen NHMUK In.24606 (hindwing base, Hooley coll., Plate 3, fig. 5), Insect Limestone, northwest Isle of Wight.

Paratypes. Specimens NHMUK II.2851a,b (14/11/04) (hindwing base, Ross coll., Plate 3, fig. 6, Text-fig. 16), NHMUK In.24641 (costo-apical part of a wing, Hooley coll., Plate 3, fig. 7, Text-fig. 17), NHMUK In.24757 (wing apex, Hooley coll., Text-fig. 18).

Etymology. Named after our colleague, odonatologist Prof. Jean Legrand.

Diagnosis. Wing characters only; vein separating hypertriangle and discoidal triangle ending in MA at distal angle of discoidal triangle; only four antenodal cross-veins; two cross-veins in area between RA and RP basal of subnodus.

Description. Probable total length of hindwing 26·0 mm; Distance from wing base to arculus 3·3 mm, from arculus to nodus 7·4 mm, from nodus to pterostigma 12·3 mm; pterostigma 1·8 mm long, 0·5 mm wide, covering two-third of a cell; pterostigmal brace weak, not aligned with basal side of pterostigma; four antenodal cross-veins aligned with the subantenodal cross-veins; arculus between the two basal antenodal cross-veins, closer to first than to second; sectors of arculus stalked and strongly curved (costal side of hypertriangle very convex); trigonal vein that is separating the hypertriangle from the discoidal triangle, is distinctly curved; vein separating the hypertriangle and the discoidal triangle ending in MA at the distal angle of discoidal triangle; discoidal triangle nearly equilateral; median space free; submedian space crossed by CuP; anal loop elongate, with two rows of cells and a median rib (Cuspl); first cross-vein between MA and RP short and ending in anterior side of hypertriangle; two cross-veins in area between RA and RP basal of subnodus; one Bqr cross-vein; oblique vein ‘O’ just distal of base of RP2; postdiscoidal area with one row of large transverse cells distal of triangle; Mspl poorly preserved but probably rudimentary; Rspl absent; RP2 and IR2, as well as RP3/4 and MA probably distally converging, after their general shape and direction; IR1 short, with its base just distal of apex of pterostigma; eigth postnodal cross-veins not aligned with the four postsubnodal cross-veins.

Discussion. These three specimens can be attributed to the same species because of the identity of the shared parts of wings and very distinctive venation. The main remarkable structure of Neophya legrandi sp. nov. is the distinctly curved anterior side of hindwing discoidal triangle. Only few groups within the Cavilabiata have this character, viz. some Libellulidae, the Mesozoic Araripelibellulidae: Araripelibellulinae Bechly, Reference Bechly1996 (especially the English Wealdian Cretaneophya strevensi Jarzembowski & Nel, Reference Jarzembowski and Nel1996), and the Recent Neophya rutherfordi Selys, Reference Selys-Longchamps1881 (Cordulephyidae Tillyard, Reference Tillyard1917 of the subfamily Neophyinae Tillyard & Fraser, Reference Tillyard and Fraser1940). The relevant Libellulidae (Tetrathemis Brauer, Reference Brauer1868, Nannophlebia Selys, Reference Selys-Longchamps1878) have this vein less curved than in our fossil and a well defined Rspl, unlike our fossil (Ris Reference Ris1909–1913). The main difference between Cretaneophya and Neophya is in the size and shape of the anal loop; 2–3 cells in Cretaneophya instead of transverse elongate with a median vein as in Neophya and in this fossil.

Furthermore, N. legrandi has the wing venation synapomorphies of the Neophyinae, i.e. sectors of arculus stalked and strongly curved (costal side of hypertriangle very convex); trigonal vein that is separating the hypertriangle from the discoidal triangle, is distinctly curved; Rspl absent; RP2 and IR2, as well as RP3/4 and MA distally converging. All the other characters present in N. legrandi are identical to those of N. rutherfordi, except for the vein separating the hypertriangle and the discoidal triangle ending in MA at the distal angle of discoidal triangle instead of before this angle as in Neophya, the presence of only four antenodal cross-veins instead of six as in N. rutherfordi, and two cross-veins in area between RA and RP basal of subnodus, instead of four as in N. rutherfordi (pers. obs.; Martin Reference Martin1906).

The Palaeogene Argentinean Palaeophya argentina Petrulevičius & Nel, Reference Petrulevičius and Nel2009 shares with N. legrandi the presence of only four antenodal cross-veins, but the former differs from the latter in the presence of four cross-veins in the area between RA and RP basal of the subnodus, instead of two, and of two rows of cells in postdiscoidal area opposite the base of RP3/4, instead of only one as in N. legrandi.

Clade Italoansida Bechly, Reference Bechly1996

‘Corduliidae species A’ sensu Nel & Jarzembowski (Reference Nel and Jarzembowski1999)

(Text-figs 19–20)

Text-figure 19 ‘Corduliidae species A’, NHMUK In.24654. Scale bar=3 mm.

Text-figure 20 ‘Corduliidae species A’, NHMUK In.24760. Scale bar=3 mm.

Material. This fossil ‘species A’ was originally described on the basis of an incomplete forewing (PORT 842245). A further fragmentary forewing can be attributed to the same species (NHMUK In.24654/In.24760, part and counterpart, Hooley coll., Text-figs 19–20, Insect Limestone, northwest Isle of Wight).

Discussion. This fossil species belongs to the Italoansida Bechly, Reference Bechly1996 (Corduliidae+Anauriculida Bechly, Reference Bechly1996) (synapomorphies of forewing: discoidal triangle divided in the groundplan into two cells by a longitudinal cross-vein; subdiscoidal triangle divided into three cells in the groundplan; wings with a Mspl, that is still somewhat indistinct and more or less parallel to MA in the groundplan). The exact affinities of this fossil species cannot be ascertained within this group however, because we lack the hindwing structures and those of the distal third of the forewing (especially the Rspl). It is an Italoansida of uncertain position, probably not Libellulidae because its Mspl is not well defined.

‘Corduliidae’ species B

(Text-fig. 21)

Text-figure 21 ‘Corduliidae’ species B, NHMUK In.17370, hindwing, female. Scale bar=3 mm.

Material. Specimen NHMUK In.17370/In.24763 (Smith/Hooley coll., Text-fig. 21), Insect Limestone, northwest Isle of Wight.

Discussion. This fossil could well correspond to the forewing of ‘Corduliidae species A’, because they have similar dimensions, but it is not possible to confirm this because we lack the wing apices of these fossils. The hindwing also fits into the Italoansida for the following reasons: hindwing discoidal triangle recessed to level of arculus; wings with a Mspl, that is still somewhat indistinct and more or less parallel to MA in the groundplan; hindwing with elongated and boot-shaped anal loop that nearly reaches the hind margin of the wing with its ‘toe’, and with a straight (non-zigzagged) and forked midrib (Cuspl) (Bechly Reference Bechly1996). Its rather poorly developed Mspl and sectors of arculus basally separated suggest that this fossil is not a Libellulidae, but rather a ‘Corduliidae’ or a Hemicorduliidae Bechly, Reference Bechly1996. The development of the anal loop and cubito-anal area of the Recent corduliid genus Epitheca Charpentier, Reference Charpentier1825 is comparable to that of this fossil (Martin Reference Martin1906, fig. 59).

‘Corduliidae’ species C

(Plate 4, figs 1–2; Text-figs 22–23)

Plate 4 (1–2) ‘Corduliidae’ species C: NHMUK In.17324, hindwing, part. Scale bar=5 mm; (2) NHMUK In.24771, counterpart. Scale bar=5 mm. (3–4) Eomacrodiplax incompleta gen. et sp. nov. 1: (3) holotype, NHMUK In.24770, forewing. Scale bar=5 mm; (4) paratype, NHMUK In.17333. Scale bar=10 mm. (5) ?Bolcathore species A., NHMUK In.24648. Scale bar=5 mm. (6) Dysagrionini species A, NHMUK I.9866. Scale bar=5 mm.

Text-figure 22 ‘Corduliidae’ species C, NHMUK In.17324, hindwing. Scale bar=3 mm.

Text-figure 23 ‘Corduliidae’ species C, NHMUK In.24771, hindwing. Scale bar=3 mm.

Material. Specimen NHMUK In.17324/In.24771 (Smith/Hooley coll., Plate 4, figs 1–2, Text-figs 22–23). Insect Limestone, northwest Isle of Wight.

Discussion. This hindwing base fits in the Italoansida and rather a ‘Corduliidae’ or a Hemicorduliidae for the same reasons as above for species B, except for those concerning the vein Mspl, unknown in this fossil. It corresponds to a different species because it is distinctly smaller, with a narrower cubito-anal area with only three rows of cells between the anal loop and the posterior wing margin, instead of four as in species B.

Family Urothemistidae Lieftinck, Reference Lieftinck1954

Genus Eomacrodiplax gen. nov.

Type species. Eomacrodiplax incompleta sp. nov.

Etymology. Named after Eocene and Macrodiplax.

Diagnosis. Forewing base characters only. Subdiscoidal triangle divided into three cells; basal side of subdiscoidal triangle not perpendicular to anterior side; distal side of discoidal triangle distinctly angular; primary and secondary antenodal cross-veins identical; sectors of arculus not stalked; cubito-anal area broad with three rows of cells.

Eomacrodiplax incompleta sp. nov.

(Plate 4, figs 3–4; Text-figs 24–25)

Text-figure 24 Eomacrodiplax incompleta gen. et sp. nov., holotype, NHMUK In.24770. Scale bar=3 mm.

Text-figure 25 Eomacrodiplax incompleta gen. et sp. nov., paratype, NHMUK In.17333. Scale bar=3 mm.

Holotype. Specimen NHMUK In.24770 (forewing base, Hooley coll., Plate 4, fig. 3, Text-fig. 24), Insect Limestone, northwest Isle of Wight.

Paratype. Specimen NHMUK In.17333 (broken wing with separate base and apex, Smith coll., Plate 4, fig. 4, Text-fig. 25).

Etymology. Named after the incomplete state of preservation of the holotype forewing.

Diagnosis. As for the genus.

Description. Forewing hyaline, width of wing 8·0 mm; distance from base to arculus 3·0 mm, from arculus to nodus 12·3 mm, from arculus to base of RP3/4 7·7 mm, from wing base to Ax1 3·7 mm, from Ax1 to arculus 1·3 mm, from arculus to Ax2 1·5 mm; primary antenodal cross-veins identical to the secondaries; at least five antenodal cross-veins, probably seven or eight in the complete wing; only two cross-veins between RA and RP basal of subnodus; one Bqr cross-vein; oblique vein ‘O’ just distal of base of RP2; sectors of arculus not stalked; median area free of cross-vein; submedian area with only CuP; subdiscoidal area subdivided into three cells; hypertriangle free, 4·4 mm long, 0·5 mm wide; discoidal triangle free, with distal side with a strong angle; two rows of cells in postdiscoidal area distal of triangle; Mspl straight, with one row of cells between it and MA; anal area with two rows of cells; cubito-anal area with three rows of cells.

Discussion. The wing base of NHMUK In.17333 is identical to that of the holotype. The presence of the characteristical shape of Rspl, curved and distally rejoining the IR2, a distinct Mspl and the subdiscoidal triangle divided into three cells support affinities of this fossil with the Italoansida. Within this clade, affinities with the ‘Corduliidae’ and Hemicorduliidae are excluded because of the distinctly angular distal side of discoidal triangle. Its attribution to the Libellulida is supported by the two very distinct primary antenodal brackets Ax1 and Ax2. Affinities with the Eulibellulida are excluded because the sectors of arculus of this fossil are not stalked. The remaining family Urothemistidae Lieftinck, Reference Lieftinck1954 comprises the four Recent genera Aethriamanta Kirby, Reference Kirby1889, Macrodiplax Brauer, Reference Brauer1868, Urothemis Brauer, Reference Brauer1868 and Selysiothemis Ris, Reference Ris1897. Eomacrodiplax gen. nov. shares with Macrodiplax and Selysiothemis the strong angle of distal side of discoidal triangle, but differs from Selysiothemis in its subdiscoidal triangle divided into three cells. Eomacrodiplax differs from Macrodiplax in the basal side of its subdiscoidal triangle not perpendicular to its anterior side and its cubito-anal area broader with three rows of cells (Ris Reference Ris1909–1913).

The Urothemistidae were previously unknown in the fossil record (Nel & Paicheler Reference Nel and Paicheler1993b, Reference Nel and Paicheler1994a). Despite the incomplete state of preservation of the type specimen, we consider this is the first fossil representative of this family and thus worthy of being named. The Urothemistidae are currently considered as the sister group of the Libellulida, whose oldest known representative is Turonian (Fleck et al. Reference Fleck, Nel and Martínez-Delclòs1999). Thus the Urothemistidae are probably also at least Late Cretaceous, and their presence in the Palaeogene is not surprising. The Recent Urothemistidae are known from the Palaearctic, Nearctic, Oriental, Afrotropical and Australasian regions. It is not possible to infer significant palaeoenvironmental information from their discovery in the Late Eocene of the Isle of Wight.

Suborder Zygoptera Selys, Reference Selys-Longchamps1854

Clade Eucaloptera Bechly, Reference Bechly1996

Subclade Calopterygiformia Bechly, Reference Bechly1996

Specimen NHMUK In.24759 (Hooley coll., Text-fig. 26, Insect Limestone, northwest Isle of Wight) belongs to the clade Eucaloptera (Amphipterygida and Calopterygomorpha), because of its very numerous cells and secondary longitudinal veins. It is not possible to ascertain its affinities.

Text-figure 26 Eucaloptera incertae sedis, NHMUK In.24759. Scale bar=3 mm.

Subclade Calopterygomorpha Bechly, Reference Bechly1996

Superfamily Polythoroidea Münz, Reference Münz1919 (sensu Gentilini Reference Gentilini2002)

Family Bolcathoridae Gentilini, Reference Gentilini2002

Genus Bolcathore Gentilini, Reference Gentilini2002

?Bolcathore species A.

(Plate 4, fig. 5; Text-fig 27)

Text-figure 27 ?Bolcathore species A., NHMUK In.24648. Scale bar=3 mm.

Material. Specimen NHMUK In.24648 (Hooley coll., Plate 4, fig. 5, Text-fig 27), Insect Limestone, north-west Isle of Wight.

Remark. This fragment is a part of the medio-cubito-anal area, dark brown red. Although very incomplete, this fragment can be accurately attributed to the Eucaloptera because of its very numerous cells and secondary longitudinal veins. It has the typical polythoroid fork of CuA into CuAa and CuAb, with a concave intercalary vein between these branches (triadic branching), and the very broad cubito-anal area of the Bolcathoridae (Bechly Reference Bechly1996; Gentilini Reference Gentilini2002). This family is recorded from the Middle Eocene of Monte Bolca (Italy) by B. colorata Gentilini, Reference Gentilini2002. Our fossil is too fragmentary to be accurately compared to the Italian species. The Recent Polythoridae occur in the warm and humid Neotropical forests.

Subclade Amphipterygida Bechly, Reference Bechly1996

Superfamily Amphipterygoidea Tillyard, Reference Tillyard1917

Family Thaumatoneuridae Tillyard & Fraser, Reference Tillyard and Fraser1938

Subfamily Dysagrioninae Cockerell, Reference Cockerell1908

Dysagrionini species A

Material. Specimen NHMUK I.9866/I.9718 (part and counterpart of two wing bases, Hooley coll., Plate 4, fig. 6, Text-fig. 28), Insect Limestone, northwest Isle of Wight.

Text-figure 28 Dysagrionini species A, NHMUK I.9866. Scale bar=1 mm.

Description. Wing fragment hyaline, 8·7 mm long; distance from wing base to arculus 3·7 mm, from arculus to subodus 4·0 mm, from wing base to Ax1 2·5 mm, from Ax1 to Ax2 1·3 mm; Ax2 just distal of arculus; subnodus nearly vertical.

Discussion. Even if it is only a fragment of a wing base, this fossil can be attributed to the Dysagrioninae on the basis of the following characters: cubito-anal area very broad (inferred after the general shape of the wing); base of IR2 below nodus; base of RP3/4 between arculus and nodus, close to nodus; discoidal cell broad but broader in its distal part than in its basal part. The Petrolestini Cockerell, Reference Cockerell1927 (Petrolestes Cockerell, Reference Cockerell1927 and Congqingia Zhang, Reference Zhang1992) differ from this fossil in the base of IR2 in a more basal position. Several Dysagrionini Cockerell, Reference Cockerell1908 are known from the Palaeogene of North America and Eurasia (Dysagrion Scudder, Reference Scudder1878, Phenacolestes Cockerell, Reference Cockerell1908, Primorilestes Nel et al., Reference Nel, Petrulevičius and Jarzembowski2005, Electrophenacolestes Nel & Arillo, Reference Nel and Arillo2006) (Nel & Paicheler Reference Nel and Paicheler1994b; Bechly Reference Bechly1996; Nel et al. Reference Nel, Petrulevičius and Jarzembowski2005; Nel & Arillo Reference Nel and Arillo2006). Our fossil differs from Electrophenacolestes in its subnodus nearly vertical. Our fossil is too incomplete to be compared with the three other genera. We consider it as a Dysagrionini of uncertain affinities. It confirms the presence of this group in the Late Eocene of Western Europe.

Remark. The fossil wing fragment MNEMG 1998.35 (=BLS 86a,b) that Nel & Jarzembowski (Reference Nel and Jarzembowski1999, pp 199–200, fig. 7) described under the label ‘Odonata suborder uncertain, species A’ could well belong to the same species as specimen NHMUK I.9866, because they have very similar discoidal and subdiscoidal cells, probably deformed in the latter.

Clade Coenagrionomorpha Bechly, Reference Bechly1996

Family Megapodagrionidae Calvert, Reference Calvert1913

Genus Oligoargiolestes Kennedy, Reference Kennedy1925

Oligoargiolestes oligocenum Kennedy, Reference Kennedy1925

1. 1916 Megalestes anglicus Cockerell & Andrews, p. 90, pl. 2 fig. 5.

2. 1925 Oligoargiolestes oligocenicum Kennedy, p. 299.

3. 1994b Oligoargiolestes oligocenicum Nel & Paicheler, p. 46.

4. 1999 Oligoargiolestes oligocenicum Nel & Jarzembowski, pp 195–197, figs 2–3.

Holotype. Specimen NHMUK I.8548 (Brodie coll., Plate 5, fig. 1), Insect Limestone, northwest Isle of Wight.

Plate 5 (1) Oligoargiolestes oligocenum Kennedy, Reference Kennedy1925, holotype, NHMUK I.8548. Scale bar=3 mm. (2) Anglohypolestes fasciata gen. et sp. nov., NHMUK In.17373. Scale bar=5 mm. (3) Eohypolestes hooleyi gen. et sp. n., holotype, NHMUK In.24653. Scale bar=5 mm. (4) ‘Enallagma’ oligocena Cockerell & Andrews, Reference Cockerell and Andrews1916, holotype, NHMUK I.8631. Scale bar=5 mm. (5) ‘Enallagma’ oligocena Cockerell & Andrews, Reference Cockerell and Andrews1916, paratype, NHMUK I.8647. Scale bar=5 mm. (6) Coenagrionidae genus and species B, specimen X.50140.143. Scale bar=5 mm. (7) Coenagrionoidea undetermined, NHMUK In.24642. Scale bar=5 mm. (8) Coenagrionoidea undetermined, NHMUK In.17372. Scale bar=2 mm.

Other specimens. MNEMG 1998.36 figured in Nel & Jarzembowski (Reference Nel and Jarzembowski1999, pp 195–196, fig. 2), NHMUK I.8890 (Brodie coll., Text-fig. 29), and NHMUK In.24645 (Hooley coll., Text-fig. 30).

Text-figure 29 Oligoargiolestes oligocenum Kennedy, Reference Kennedy1925, NHMUK I.8890. Scale bar=3 mm.

Text-figure 30 Oligoargiolestes oligocenum Kennedy, Reference Kennedy1925, NHMUK In.24645. Scale bar=3 mm.

Description of new material (NHMUK I.8890 and NHMUK In.24645). Wing width 5.5 mm (I.8890), 5.2 mm (In.24645); distance from base to arculus 4.0 mm, from arculus to nodus 3.8 mm, from nodus to pterostigma 15.0 mm, pterostigma 2.1 mm long, 0.7 mm wide, covering 1.5 cell; pterostigmal brace oblique; basal and distal sides of pterostigma oblique; nodal Cr and subnodus distinctly oblique; Ax1 0.5 mm basal of arculus; Ax2 nearly opposite arculus; no secondary antenodal cross-vein; no antesubnodal cross-vein; median space free; CuP distal of base of AA; discoidal cell elongate, 1.7 mm long, 0.5 mm wide, with its anterior and posterior sides parallel, distal side oblique, 0.7 mm long; base of RP3/4 near subnodus, 1.0 mm basal of it; base of IR2 opposite subnodus; base of RP2 at about 5–6 cells distal of subnodus; base of IR1 7–8 cells distally; one row of cells between RP1 and IR1; three rows between IR1 and RP2, between RP2 and IR2, and between IR2 and RP3/4, five rows between RP3/4 and MA, one row between MA and MP, one row between MP and CuA and between CuA and posterior wing margin.

Discussion. Cockerell & Andrews (Reference Cockerell and Andrews1916, p. 90, pl. 2, fig. 5) figured this fossil under the name ‘Megalestes anglicus’ Kennedy (Reference Kennedy1925, pp 296–297, 299) created the genus Oligoargiolestes Nel & Paicheler (Reference Nel and Paicheler1994b, p. 46) and Nel & Jarzembowski (Reference Nel and Jarzembowski1999, pp 195–196, figs 2–3) discussed its family attribution and considered it could belong to the Megapodagrionidae or to the Pseudolestidae. This species can be separated from the Hypolestidae of the Isle of Wight (Late Eocene) on the basis of its very oblique distal side of pterostigma (see below). Both the type specimen and MNEMG 1998.36 are wing apices, but the two new specimens NHMUK I.8890, and NHMUK In.24645 are more complete wings. The attribution of these new specimens to the same species as the type is based on the exact identity in all their shared structures.

This species can be considered as a Coenagrionomorpha Bechly, Reference Bechly1996 because of the presence of their synapomorphies in the wing venation sensu Bechly (Reference Bechly1996), i.e. pterostigma shortened; postnodal and postsubnodal cross-veins aligned; lestine oblique vein absent; basal closure of discoidal cell including the development of a dorsal arcular bracket. Within the Coenagrionomorpha, the Hypolestidae could be excluded because of their amphipterygid type of pterostigma (basal margin strongly slanting). The Coenagrioniformia Bechly, Reference Bechly1996 could also be excluded because of their suppressed intercalary veins (except IR1 and IR2) and the presence of only two rows of cells in the total wing space between RP1 and RP2 that are separated by the IR1. The last group within the Coenagriomorpha are the Megapodagrionidae. They have no known synapomorphies in the wing venation, after Bechly (Reference Bechly1996), and could well be paraphyletic (Groeneveld et al. Reference Groeneveld, Clausnitzer and Hadrys2007).

Racenis (Reference Racenis1959) proposed a division of the Megapodagrionidae into four subfamilies: Argiolestinae; Megapodagrioninae; Philosininae Ris, Reference Ris1917; and Dysagrioninae Scudder, Reference Scudder1878 (transferred into the Thaumatoneuridae by Bechly Reference Bechly1996). Davies (Reference Davies1981) added the Tactonemidinae. Affinities with this last subfamily are excluded because they have veins RP3/4 and IR2 arising distal of the subnodus. Oligoargiolestes shares with Philosina Ris, Reference Ris1917, the unique representative of the Philosininae, the structures of the petiole and anal and CuP veins, but differs in its straight CuA with a narrow cubito-anal area, and its distinctly shorter pterostigma (Ris Reference Ris1917).

Oligoargiolestes could have some affinities with the Argiolestinae because of its longitudinal wing veins curved in their posterior part (convex), thus somewhat shortened (especially MA, MP and CuA) and the branches of RP (RP3/4 and IR2) distally diverging (Bechly, Reference Bechly1996). This author also considered that the character of ‘more than one row of cells between CuA and the hind margin (reversal)’ is a synapomorphy of the Argiolestinae, but this character is absent in some argiolestine genera as Trineuragrion and Rhinagrion. Oligoargiolestes shares with these last genera the presence of one row of cells between CuA and the hind margin. Nevertheless, all the argiolestine genera have their base of free AA opposite or distal of the arculus, unlike Oligoargiolestes.

The Megapodagrioninae have no known synapomorphies in the wing venation after Bechly (Reference Bechly1996). Only the representatives of the tribe Megapodagrionini Racenis, Reference Racenis1959 have their CuP veins proximal of the arculus and distal of the base of AA (see list in Nel et al. Reference Nel, Martinez-Delclòs, Papier and Oudard1997). Amongst them, Megapodagrion Selys, Reference Selys-Longchamps1885 and Priscagrion Zhou & Wilson, Reference Zhou and Wilson2001 have a CuP in a similar distal position relatively to base of AA, but Priscagrion has secondary antenodal cross-veins (Zhou & Wilson Reference Zhou and Wilson2001; De Marmels Reference De Marmels2002). Oligoargiolestes differs from Megapodagrion in its area between RP3/4 and IR2 being greatly broadened distally. Also, all the Recent Megapodagrioninae have their longitudinal veins (especially MA, MP and CuA) not curved in their posterior part.

Nel et al. (Reference Nel, Martinez-Delclòs, Papier and Oudard1997) listed and commented on the fossil megapodagrionid genera. Amongst them, Oligoargiolestes differs from Eopodagrion Cockerell, Reference Cockerell1920 in that its subnodus is not vertical and it has more numerous postnodal cross-veins. Melanagrion Cockerell, Reference Cockerell1907 has two rows of cells in the cubito-anal area and in the area between C and RA distal of the pterostigma, and its wing is petiolated distal of the CuP. Miopodagrion Kennedy, Reference Kennedy1925 has two rows of cells in the area between C and RA distal of the pterostigma. Vulcagrion Nel & Paicheler, Reference Nel and Paicheler1994b has its veins IR2 and RP3/4 fused basally distal of the subnodus and the wing petiolated distal of the CuP vein, as in Lithagrion Scudder, Reference Scudder1890. Cerdanyagrion Nel et al., Reference Nel, Arillo and Martínez-Delclòs1996 differs from Oligoargiolestes in the presence of double rows of cells in the area between RA and RP1 and between C and RA (Nel et al. Reference Nel, Arillo and Martínez-Delclòs1996). Oligoargiolestes differs from Thanetophilosina Nel et al., Reference Nel, Martinez-Delclòs, Papier and Oudard1997 in the number of postnodal cross-veins (about 16 instead of 30) (Nel et al. Reference Nel, Martinez-Delclòs, Papier and Oudard1997). Oligoargiolestes differs from Furagrion Petrulevičius et al., Reference Petrulevičius, Wappler, Wedmann, Rust and Nel2008 in having its pterostigma covering less than two cells and with a more oblique distal side. It differs from Eckfeldia Petrulevičius et al., Reference Petrulevičius, Wappler, Wedmann, Rust and Nel2008 in that its pterostigma covers less than two cells and it has a more oblique distal side (Petrulevičius et al. Reference Petrulevičius, Wappler, Wedmann, Rust and Nel2008). Oligoargiolestes differs from Electropodagrion Azar & Nel, Reference Azar and Nel2008 having in its vein CuP in a more distal position, and in the presence of several secondary longitudinal veins in the areas between the branches of RP (Azar & Nel Reference Azar and Nel2008).

Family Hypolestidae Tillyard & Fraser, Reference Tillyard and Fraser1938

Genus Anglohypolestes gen. nov.

Type species. Anglohypolestes fasciata sp. nov.

Etymology. Named after Anglia, Latin name for England and Hypolestes.

Diagnosis. Wing characters only; IR2 and RP3/4 bases basally recessed midway between nodus and arculus; no straight intercalary veins between RP1 and IR1; IR1, RP2 and IR2 nearly straight; a cross-vein between RA and RP just basal of subnodus; no cross-vein between basal side of pterostigma and costal margin; pterostigma covering three cells; CuA very long, weakly zigzagged; anal area present below discoidal cell; five rows of cells in area between IR2 and RP3/4, area between MP and CuA with one row of cells; area between IR1 and RP2 with four rows of cells along posterior wing margin; area between RP2 and IR2 with a very short secondary longitudinal vein; area between MA and MP broad, with six rows of cells along posterior wing margin.

Anglohypolestes fasciata sp. nov.

Holotype. Specimen NHMUK In.17373 (Smith coll., Plate 5, fig. 2, Text-fig. 31), Insect Limestone, northwest Isle of Wight.

Text-figure 31 Anglohypolestes fasciata gen. et sp. nov., NHMUK In. 17373. Scale bar=3 mm.

Etymology. Named after the dark zone crossing the distal part of wing.

Diagnosis. As for the genus, a dark zone crossing distal part of wing.

Description. Wing hyaline except for a dark zone crossing distal part of wing; preserved part of wing 15·0 mm long, width 4·6 mm; distance between arculus and nodus 3·1 mm, between nodus and pterostigma 8·1 mm, between pterostigma and apex 2·1 mm; pterostigma 1·9 mm long, 0·5 mm wide, covering three cells, with basal margin strongly slanting; pterostigmal brace not very oblique; 11 postnodal and postsubnodal cross-veins aligned; no oblique vein ‘O’; Ax2 probably aligned with arculus; discoidal cell elongate, 0·37 mm wide and probably 0·8 mm long, with anterior side nearly as long as posterior side; subdiscoidal cell and anal area present below discoidal cell; subnodus oblique; one cross-vein in area between RA and RP basal of subnodus; bases of RP3/4 and IR2 midway between arculus and nodus; base of RP2 three cells distal of subnodus; base of IR1 two cells distally; all longitudinal veins straight or weakly curved and/or weakly zigzagged; distribution of intercalary rows of cells and veins between main veins as in diagnosis.

Discussion. This fossil could be considered as Coenagrionomorpha Bechly, Reference Bechly1996 because of the presence of their synapomorphies in the wing venation sensu Bechly (Reference Bechly1996), i.e. pterostigma shortened; postnodal and postsubnodal cross-veins aligned; lestine oblique vein absent. Within the Coenagrionomorpha, the Coenagrioniformia Bechly, Reference Bechly1996 could be excluded because of their suppressed intercalary veins (except IR1 and IR2) and the presence of only two rows of cells in the total wing space between RP1 and RP2 that are separated by the IR1. The Megapodagrionidae Calvert, Reference Calvert1913 have no known synapomorphies in the wing venation (after Bechly Reference Bechly1996), but their IR2 and RP3/4 bases are below the nodus, not basally recessed as in this fossil. It shares with the Hypolestidae Tillyard & Fraser, Reference Tillyard and Fraser1938 the amphipterygid type of pterostigma, with basal margin strongly slanting. The Heteragrioninae Racenis, Reference Racenis1959 (Heteragrion Selys, Reference Selys-Longchamps1862) and the Philogeniinae Racenis, Reference Racenis1959 (type genus: Philogenia Selys, Reference Selys-Longchamps1862, other possible genus Paraphlebia Hagen, Reference Hagen1861) are excluded due to the positions of the bases of IR2 and RP3/4. Anglohypolestes gen. nov. belongs to the remaining subfamily Hypolestinae, in which the midfork is at least somewhat recessed basally, and the subnodus is located between the bases of RP2 and IR2 that are widely separated.

Within this subfamily, the Lestoideini Münz, Reference Münz1919 (Lestoidea Tillyard, Reference Tillyard1913) can be excluded, because Anglohypolestes has numerous intercalary longitudinal veins between the main veins, but lacks the two characteristic straight intercalary veins between RP1 and IR1.

The Hypolestini Tillyard & Fraser, Reference Tillyard and Fraser1938 (Hypolestes Gundlach, Reference Gundlach1888) have a characteristic distal curvature of the veins IR1, RP2 and IR2, not present in Anglohypolestes. Also our fossil has a cross-vein in the area between RA and RP, just basal of the subnodus, which is absent in Hypolestes.

The Philosinini Kennedy, Reference Kennedy1925 comprise the type genus Philosina Ris, Reference Ris1917 and the sister-genera Lestomima May, Reference May1933 and Rhipidolestes Ris, Reference Ris1912, and maybe Eolestes synthetica Cockerell, Reference Cockerell1940 and Prohypolestes dauphinensis Nel & Paicheler, Reference Nel and Paicheler1994b (Bechly Reference Bechly1996). Lestomima and Rhipidolestes differ from Anglohypolestes in the shape of the CuA and the pterostigma (Münz Reference Münz1919; May Reference May1933). The base of IR2 is below the subnodus in Philosina, unlike in Anglohypolestes. Eolestes has one row of cells in the area between IR2 and RP3/4, and a very broad area between MP and CuA; the main differences from Anglohypolestes (Cockerell Reference Cockerell1940). Differences from Prohypolestes are fewer, but nevertheless justify a generic separation; i.e. the area between IR1 and RP2 with two rows of cells along the posterior wing margin in Prohypolestes, instead of the four in Anglohypolestes; the area between RP2 and IR2 with a longer secondary longitudinal vein in Prohypolestes than in Anglohypolestes; the area between MA and MP narrower in Prohypolestes than in Anglohypolestes; and the pterostigma covering two cells in Prohypolestes instead of three as in Anglohypolestes (Nel & Paicheler Reference Nel and Paicheler1994b).

Genus Eohypolestes gen. nov.

Type species. Eohypolestes hooleyi sp. nov.

Etymology. Named after Eocene and Hypolestes.

Diagnosis. Wing characters only; IR2 and RP3/4 bases basally recessed midway between nodus and arculus; no straight intercalary veins between RP1 and IR1; IR1 zigzagged; RP2 and IR2 distally curved; no cross-vein between basal side of pterostigma and costal margin; pterostigma covering two cells; seven rows of cells in area between IR2 and RP3/4, area between RP3/4 and MA with six rows of cells; area between IR1 and RP2 with three rows of cells along posterior wing margin; area between RP2 and IR2 with two zigzagged secondary longitudinal veins.

Eohypolestes hooleyi sp. nov.

Holotype. Specimen NHMUK In.24653 (Hooley coll., Plate 5, fig. 3, Text-fig. 32), Insect Limestone, northwest Isle of Wight.

Text-figure 32 Eohypolestes hooleyi gen. et sp. nov., holotype, NHMUK In.24653. Scale bar=3 mm.

Paratype. Specimen NHMUK In.24769 (Hooley coll.).

Etymology. Named after Mr Hooley, collector of the type specimen.

Diagnosis. As for the genus.

Description. Both specimens are distal two-thirds of wings. Wing hyaline (a dark zone is present in the holotype along the posterior wing margin, but this is probably due to taphonomy); preserved part of wing 19·3 mm long, width 5·9 mm; distance between nodus and pterostigma 11·7 mm, between pterostigma and apex 2·3 mm; pterostigma 2·1 mm long, 0·7 mm wide, covering two cells, with basal margin strongly slanting; pterostigmal brace not very oblique; 11 postnodal and postsubnodal cross-veins aligned, but no postsubnodal cross-vein corresponding to the most distal postnodal cross-vein; no oblique vein ‘O’; subnodus oblique; bases of RP3/4 and IR2 midway between arculus and nodus, but well apart; base of RP2 three cells distal of subnodus; base of IR1 five cells distally; cubito-anal area with three rows of cells between CuA and posterior wing margin (visible in paratype); shape of main longitudinal veins and distribution of intercalary rows of cells and veins between main veins as in diagnosis.

Discussion. This fossil can be attributed to the Hypolestinae for the same reasons as above. Furthermore, affinities with the Lestida Bechly, Reference Bechly1996 are very unlikely, because this fossil has its postnodal and postsubnodal cross-veins well aligned and the bases of IR2 and RP3/4 more distant than in taxa of this clade. The Lestoideini, Philosina, Lestomima and Rhipidolestes, are excluded for the same reasons as above (the shape of the pterostigma in the cases of Lestomima and Rhipidolestes). The area between MA and MP is much narrower in Prohypolestes than in Eohypolestes gen. nov. Eolestes has one row of cells in the area between IR2 and RP3/4, unlike in Eohypolestes.

Eohypolestes shares with the Hypolestini and Hypolestes the characteristical distal curvature of RP2 and IR2, but not IR1, which is zigzagged in Eohypolestes. Other differences with Hypolestes are as follows: base of RP2 three cells distal of subnodus in Eohypolestes, instead of two cells; base of IR1 five cells distal of that of RP2 in Eohypolestes, instead of two; area between RP3/4 and MA with six rows of cells along posterior wing margin, instead of only one; area between IR2 and RP2 with four rows of cells and two long intercalary veins in Eohypolestes; three rows of cells in cubito-anal area. Thus we tentatively attribute Eohypolestes to the Hypolestini, but to a different genus from Hypolestes.

Remark. The family Hypolestidae was already known in the fossil record from several larvae and a possible adult in the Eocene Baltic amber, Eolestes synthetica Cockerell, Reference Cockerell1940 (Eocene, USA) and Prohypolestes dauphinensis Nel & Paicheler, Reference Nel and Paicheler1994b (Oligocene, France) (Nel & Paicheler Reference Nel and Paicheler1994b; Bechly & Wichard Reference Bechly and Wichard2009). Thus the present discovery of two new genera and species is not surprising.

Clade Coenagrioniformia Bechly, Reference Bechly1996

Family Coenagrionidae Kirby, Reference Kirby1890

Genus Enallagma Charpentier, Reference Charpentier1840

‘Enallagma’ oligocena Cockerell & Andrews, Reference Cockerell and Andrews1916

1. 1916 ‘Enallagma’ oligocena Cockerell & Andrews, pp. 90–91, pl.2 figs 6,7.

2. 1993 ‘Enallagma’ oligocenica Nel & Paicheler, p. 60.

3. 1999 ‘Enallagma’ oligocena Nel & Jarzembowski, p. 197.

Note. This species was originally described on the basis of the holotype NHMUK I.8631 (Brodie coll., median part of two wings, Reference Hooker, Collinson, Grimes, Sille and MatteyPlate 5, fig. 4, Text-fig. 33, Insect Limestone, northwest Isle of Wight) (Cockerell & Andrews Reference Cockerell and Andrews1916, pp 90–91, pl. 2, fig. 6), and paratype NHMUK I.8647 (Brodie coll., a wing apex, Reference Hooker, Collinson, Grimes, Sille and MatteyPlate 5, fig. 5, Text-fig. 34) (Cockerell & Andrews Reference Cockerell and Andrews1916, pp 90–91, pl. 2, fig. 7). Nel & Paicheler (Reference Nel and Paicheler1993a) and Nel & Jarzembowski (Reference Nel and Jarzembowski1999) discussed the possible affinities of this taxon and indicated that the second specimen could well belong to a different species. ‘E.’ oligocena is a Coenagrionidae of uncertain affinities.

Text-figure 33 ‘Enallagma’ oligocena Cockerell & Andrews, Reference Cockerell and Andrews1916, holotype, NHMUK I.8631. Scale bar=3 mm.

Text-figure 34 ‘Enallagma’ oligocena Cockerell & Andrews, Reference Cockerell and Andrews1916, NHMUK I.8647. Scale bar=3 mm.

Coenagrionidae genus and species A

Remarks. Nel & Jarzembowski (Reference Nel and Jarzembowski1999) described and figured this Coenagrionoidea on the basis of the distal two-thirds of a wing (MNEMG 1998.33, Insect Limestone, northwest Isle of Wight), mainly characterised by the shortened MP and a very short pterostigma covering less than one cell. Its affinities remain uncertain.

Coenagrionidae genus and species B

Remarks. This taxon is based on the basal two-thirds of two wings attached to a thorax (CAMSM X.50140.143, Smith Coll., Reference Hooker, Collinson, Grimes, Sille and MatteyPlate 5, fig. 6, Insect Limestone, northwest Isle of Wight), and differs from genus and species A in its elongate CuA and MP. It differs from ‘Enallagma’ oligocena in its discoidal cell with a shorter anterior side. Nevertheless, its incomplete state of preservation does not allow further taxonomic placement within the Coenagrionidae.

Coenagrionidae genus and species C

Remarks. This taxon is based on the median half of a wing (specimen NHMUK In.24762, Hooley coll., Text-fig. 35, Insect Limestone, northwest Isle of Wight). It differs from the other coenagrionoid taxa from the Isle of Wight in the presence of two rows of cells in the distal part of the cubito-anal area.

Text-figure 35 Coenagrionidae genus and species C, specimen NHMUK In.24762, drawing. Scale bar=2 mm.

Remarks. Several other fragments of wings attributable to the Coenagrionoidea have been found in the Isle of Wight outcrops (specimens NHMUK In.17372, In.17385 (Smith coll.), NHMUK In.24528, In.24640, In.24642, In.24755, (Hooley coll.) and MIWG 3602) (Reference Hooker, Collinson, Grimes, Sille and MatteyPlate 5, figs 7–8, Reference Hooker, Grimes, Mattey, Collinson, Sheldon, Koeberl and MontanariPlate 6, fig. 1, Text-figs 36–41), but their exact affinities and identity cannot be determined.

Plate 6 (1) Coenagrionoidea undetermined, NHMUK In.17385. Scale bar=3 mm. (2) Angloprotoneura eocenica gen. et sp. nov., holotype, NHMUK In 17246. Scale bar=5 mm. (3) Lestes aff. regina Théobald, Reference Theobald1937, NHMUK In.17408a, part. Scale bar=5 mm. (4) NHMUK In.17408b, counterpart. Scale bar=5 mm. (5) NHMUK In.24639. Scale bar=5 mm. (6) ‘Megalestes’ anglicus Cockerell, Reference Cockerell1915, holotype, USNM No. 61449. Scale bar=4 mm.

Text-figure 36 Coenagrionoidea undetermined, NHMUK In.24640. Scale bar=2 mm.

Text-figure 37 Coenagrionoidea undetermined, NHMUK In.24755. Scale bar=3 mm.

Text-figure 38 Coenagrionoidea undetermined, NHMUK In.24642. Scale bar=3 mm.

Text-figure 39 Coenagrionoidea undetermined, NHMUK In.17372. Scale bar=2 mm.

Text-figure 40 Coenagrionoidea undetermined, NHMUK In.24528. Scale bar=2 mm.

Text-figure 41 Coenagrionoidea undetermined, MIWG 3602. Scale bar=3 mm.

Superfamily Platycnemidoidea Jacobson & Bianchi, Reference Jacobson and Bianchi1905

Family ‘Protoneuridae’ Jacobson & Bianchi, Reference Jacobson and Bianchi1905

Genus Angloprotoneura gen. nov.

Type species. Angloprotoneura emilielacroixi sp. nov.

Etymology. Named after Anglia, Latin name for England and Protoneura.

Diagnosis. Wing characters only; CuA lacking; AA lacking; only seven postnodal cross-veins; MP ending three cells distal of level of subnodus; pterostigma covering one cell and longer than broad; base of RP2 three cells distal of subnodus.

Angloprotoneura emilielacroixi sp. nov.

Holotype. Specimen NHMUK In.17246 (Smith coll., Reference Hooker, Grimes, Mattey, Collinson, Sheldon, Koeberl and MontanariPlate 6, fig. 2, Text-fig. 42), Insect Limestone, northwest Isle of Wight.

Text-figure 42 Angloprotoneura eocenica gen. et sp. nov., holotype, NHMUK In.17246. Scale bar=1 mm.

Etymology. Named after the late Emilie Lacroix, friend of one of us (GF).

Diagnosis. That of the genus.

Description. An isolated wing, 12·3 mm long, 2·5 mm wide, probably hyaline, with dark pterostigma, the traces of coloration of some veins being probably due to diagnenetic process; pterostigma 0·7 mm long, 0·25 mm wide, longer than broad, covering nearly one cell; petiole elongate and narrow, 1·75 mm long, 0·5 mm wide, but with its posterior part partly destroyed; nodus in the basal 32% of wing length, distance from wing base to arculus 1·75 mm, from arculus to nodus 2·3 mm, from nodus to pterostigma 6·5 mm, from pterostigma to wing apex 1·5 mm; antenodal area partly destroyed, with antenodal cross-veins not preserved; nodal Cr and subnodus distinctly oblique; seven postnodal cross-veins, aligned with postsubnodal cross-veins, except for the two distals; base of RP3/4 below nodus; base of IR2 aligned with subnodus; base of RP2 three cells and 2·4 mm distally; base of IR1 three cells and 2·3 mm distally; IR1, RP2, RP3/4, and MP nearly straight; IR2 and MA distally zigzagged; cross-veins of the radial area nearly aligned in transverse rows; discoidal cell partly destroyed but clearly narrow rectangular and elongate, 0·1 mm wide, 1·1 mm long; one row of cells in area between MP and posterior wing margin; MP ending three cells distal of level of subnodus, nearly below base of RP2; CuA lacking, AA lacking.

Discussion. The vein CuA completely fused with the posterior wing margin, thus only retained as subdiscoidal vein, is a specialised character that convergently developed between the Triassic Protozygoptera: Batkeniidae, and among the true Zygoptera, in the Platystictidae Laidlaw, Reference Laidlaw1924, Disparocypha Ris, Reference Ris1916, the Lestoideini Münz, Reference Münz1919 (Lestoidea Tillyard, Reference Tillyard1913) and the Protoneuridae Jacobson & Bianchi, Reference Jacobson and Bianchi1905 (Bechly Reference Bechly1996).

The Lestoideini have completely different positions of the bases of veins RP3/4 and IR2, well basal of nodus and a very short MP, not reaching the level of nodus (Fraser Reference Fraser1957). The rhinocyphine genus Disparocypha also has its bases of RP3/4 and IR2 basally recessed. The Platystictidae have their wings with a falcate apex, longitudinal veins RA, RP1, IR1, RP2 and IR2 strongly converging to the apex, the costal edge of the pterostigma much shorter than the radial edge, and the nodus in a very basal position (about 22% of wing length), unlike Angloprotoneura gen. nov. (Bechly Reference Bechly1996).

Bechly (Reference Bechly1996) characterised the Protoneuridae by the following wing venational synapomorphies: ‘discoidal cell rectangular (convergent to Lestoideini, Platystictidae and most Caloptera; maybe synapomorphic with Platycnemidinae); discoidal bracket (dorsal sclerotisation on MAb and the subdiscoidal vein) strongly reduced (convergent to Perilestidae and Anisoptera); CuA completely fused with the hind margin, thus only retained as subdiscoidal vein (convergent to Batkeniidae, Disparocypha, Lestoideini and Platystictidae).’ None of these characters is strictly present in the Protoneuridae, but their combination present in Angloprotoneura gen. nov. occurs only in this family.

Fraser (Reference Fraser1957) divided the Protoneuridae into Protoneurinae, Caconeurinae Fraser, Reference Fraser1957, Disparoneurinae Fraser, Reference Fraser1957, and Isostictinae Fraser, Reference Fraser1955. Watson (Reference Watson1992) synonymised the Disparoneurinae with the Protoneurinae. Later, Bechly (Reference Bechly1996) divided the Protoneuridae into Protoneurinae (incl. Caconeurinae and Disparoneurinae), and Isostictinae, but he proposed no precise wing venational synapomorphy to support the Protoneurinae, and only one for the Isostictinae, i.e. arculus secondarily shifted distal of Ax2, but indicating it may be a synapomorphy with certain Protoneurinae. Pessacq (Reference Pessacq2008) concluded after a new phylogenetic analysis that the Protoneuridae as a whole is probably polyphyletic. Therefore, we have to compare Angloprotoneura gen. nov. to all the genera. The list of protoneurid genera up to 1993 is given in Bridges (Reference Bridges1993).

Angloprotoneura differs from Chlorocnemis Selys, Reference Selys-Longchamps1863 and Proneura Selys, Reference Selys-Longchamps1889 in its lacking CuA vein (Münz Reference Münz1919, figs 130–131). It differs from Elattoneura Cowley, Reference Cowley1935, Junix Racenis, Reference Racenis1968, Proneura Selys, Reference Selys-Longchamps1889, Disparoneura Selys, Reference Selys-Longchamps1860, Peristicta Hagen, Reference Hagen and Selys-Longchamps1860, Neoneura Selys, Reference Selys-Longchamps1860, Caconeura Kirby, Reference Kirby1890, Esme Fraser, Reference Fraser1922, Phylloneura Fraser, Reference Fraser1922, Arabineura Schneider & Dumont, Reference Schneider and Dumont1995, Lamproneura De Marmels, Reference De Marmels2003 and Idioneura Selys, Reference Selys-Longchamps1860 in lacking its AA vein (Münz Reference Münz1919; Fraser Reference Fraser and Stephenson1933; Cowley Reference Cowley1935; Schneider & Dumont Reference Schneider and Dumont1995; De Marmels Reference De Marmels2003).

Peristicta, Caconeura, Phylloneura and Esme also differ from Angloprotoneura gen. nov. in their great number of postnodal cross-veins (more than 12, instead of seven). Ellatoneura has more than nine postnodals (Cowley Reference Cowley1936). Junix and Neoneura have a very short MP, not surpassing the level of the nodus (Racenis Reference Racenis1968). Lamproneura has also a very short MP, unlike Angloprotoneura.

Among the taxa with AA completely lacking, Angloprotoneura gen. nov. differs from Roppaneura Santos, Reference Santos1966 in its pterostigma of ‘normal’ size and shape, covering one cell and longer than broad; in its longer MP, reaching the posterior wing margin 3–4 cells distal of the level of the nodus, instead of 2–3 cells; and in its base of RP2 three cells distal of the subnodus, instead of two (see Santos Reference Santos1966, figs 1, 3–5). Angloprotoneura gen. nov. differs from Idioneura Selys, Reference Selys-Longchamps1860, Epipleoneura Williamson, Reference Williamson1915, Epipotoneura Williamson, Reference Williamson1915, Nososticta Selys, Reference Selys-Longchamps1860, Prodasineura Cowley, Reference Cowley1934, Psainoneura Williamson, Reference Williamson1915, Protoneura Selys, Reference Selys-Longchamps and de la Sagra1857, Phasmoneura Williamson, Reference Williamson1916, Forcepsioneura Lencioni, Reference Lencioni1999 and Drepanoneura Ellenrieder & Garrison, Reference Ellenrieder and Garrison2008 in its distinctly longer MP, reaching the posterior wing margin 3–4 cells distal of the level of the nodus, instead of none, one, two, or three (Münz Reference Münz1919; Cowley Reference Cowley1936; Lencioni Reference Lencioni1999; Machado Reference Machado2004; Ellenrieder & Garrison Reference Ellenrieder and Garrison2008). Angloprotoneura gen. nov. differs from Microneura Hagen, Reference Hagen and Selys-Longchamps1886, Melanoneura Fraser, Reference Fraser1922 and Amazoneura Machado, Reference Machado2004 in its smaller number of postnodal cross-veins (around 20 in the latters) (Fraser Reference Fraser and Stephenson1933; Cowley Reference Cowley1935).

The genus Eoprotoneura, originally attributed to the Protoneuridae: Eoprotoneurinae Carle & Wighton, Reference Carle, Wighton and Grimaldi1990, and latter transferred to the Isostictinae: Eoprotoneurini, differs from Angloprotoneura in its very short MP, just reaching the nodus level, base of IR2 well distal of the subnodus, and in its pterostigma covering one cell and a half (Carle & Wighton Reference Carle, Wighton and Grimaldi1990; Bechly Reference Bechly1996).

The fossil record of the small damselfly family Protoneuridae is so far reduced to the sole genus Eoprotoneura Carle & Wighton, Reference Carle, Wighton and Grimaldi1990 from the Early Cretaceous of Brazil. Piton (Reference Piton1940) described a very poorly preserved and enigmatic fossil from the Paleocene of Menat (France), which he attributed to the Ephemeroptera family Protoneuridae (sic), under the name Archaeophlebia enigmatica. Nel & Roy (Reference Nel and Roy1996) revised this taxon and concluded that it is not an Odonata or an Ephemeroptera, but a Mantodea of uncertain family, probably a Chaeteessidae.

Clade Lestomorpha Bechly, Reference Bechly1996

Family Lestidae Calvert, Reference Calvert1901

Genus Lestes Leach, Reference Leach and Brewster1815

Lestes aff. regina Théobald, Reference Theobald1937

(Reference Hooker, Grimes, Mattey, Collinson, Sheldon, Koeberl and MontanariPlate 6, figs 3–5; Text-figs 43–44)

Text-figure 43 Lestes aff. regina Théobald, Reference Theobald1937, NHMUK In.17408. Scale bar=3 mm.

Text-figure 44 Lestes aff. regina Théobald, Reference Theobald1937, NHMUK In.24639. Scale bar=3 mm.

Previously described material. Specimen BMB 018808/-9, Insect Limestone, northwest Isle of Wight.

New material. NHMUK In.17408 (Smith coll., Reference Hooker, Grimes, Mattey, Collinson, Sheldon, Koeberl and MontanariPlate 6, figs 3–4, Text-fig. 43); NHMUK In.24639 (Hooley coll., Reference Hooker, Grimes, Mattey, Collinson, Sheldon, Koeberl and MontanariPlate 6, fig. 5, Text-fig. 44).

Discussion. Nel & Jarzembowski (Reference Nel and Jarzembowski1999, pp 193–195, fig. 1a–b) described and figured the specimen BMB 018808/-9. They considered that this species is very close, if not identical, to Lestes regina from the Late Eocene of Gard (France). The new specimens can be attributed to the same species as BMB 018808/-9 on the basis of the same wing venation, especially the presence of three rows of cells in the cubito-anal area (first group of Nel & Paicheler Reference Nel and Paicheler1994b). NHMUK In.17408 and NHMUK In.24639 are slightly larger wings (respective widths 5·7 mm and 5·3 mm, instead of 4·9 mm for BMB 018808/-9).

Note. The type specimen of ‘Megalestes’ anglicus Cockerell, Reference Cockerell1915 (USNM No. 61449) is the middle part of a wing (see Reference Hooker, Grimes, Mattey, Collinson, Sheldon, Koeberl and MontanariPlate 6, fig. 6) that could correspond to the same species as Anglohypolestes fasciata, to Eohypolestes hooleyi, or to another taxon belonging to a different family (Cockerell Reference Cockerell1915, pp 498–499, pl. 61, fig. 9; Nel & Paicheler Reference Nel and Paicheler1994b, pp 3–4, fig. 2; Nel & Jarzembowski Reference Nel and Jarzembowski1999, p. 195). It is a Zygoptera Lestiformia, or a Coenagrionomorpha of uncertain affinities.

2. Palaeoclimatic implications

With nearly 20 different species distributed in 11 different families, the Isle of Wight fauna is one of the most diverse of the European Palaeogene, comparable to that of Aix-en-Provence. The family profile of the Isle of Wight differs strongly from those of the Oligocene of France, in the presence of Cordulephyidae, Urothemistidae, Bolcathoridae and Protoneuridae; but also in the absence of the Libellulidae. This last family dominates the diversity of nearly all the Middle to Late Oligocene outcrops of France and Germany. The Libellulidae are recorded from the Late Cretaceous and the Early Eocene. It is possible that they began to diversify only after the Eocene but, unfortunately, the present knowledge on the odonatan fauna of the Palaeocene–Eocene remains very incomplete.

The ‘Gynacanthinae’ species A is of great interest because all its closest living relatives live under warm to hot climates in Indo-Malaysia, South America and Australasia. Neophya rutherfordi is an Afrotropical taxon, living in forests along rivers and small streams in the warm to hot intertropical zone. This discovery of a Neophya is the second case of the presence of an Afrotropical insect genus in the Western European Palaeogene. The first one was the presence of the calopterygid genus group Sapho+Umma in the Late Oligocene of Aix-en-Provence and Armissan (Nel Reference Nel1987; Nel & Paicheler Reference Nel and Paicheler1992; Nel & Petrulevičius Reference Nel and Petrulevičius2010). The presence of these Afrotropical insects in the Palaeogene of Western Europe could correspond to multiple contacts between these areas during the Cretaceous–Palaeogene (Gheerbrant & Rage Reference Gheerbrant and Rage2006). The presence of Palaeophya Petrulevičius & Nel, Reference Petrulevičius and Nel2009 in the Palaeogene of Argentina rather supports a great antiquity for the Cordulephyidae. The presence of Palaeophya in Argentina and of a Neophya species in the UK suggests relict taxa of a widely distributed family during the Late Cretaceous–Palaeogene.

The ‘Protoneuridae’ is a circumtropical family, known from the Neotropical, Afrotropical (also a genus from Arabia), Australo-Papuan and Indo-Malaysian regions. As the exact affinities of Angloprotoneura cannot be accurately established, its main interest is to demonstrate the presence of this ‘group’ of intertropical Zygoptera in Europe during the Palaeogene. The Recent representatives live in a warm climate, but not always very humid, and their larvae inhabit small streams and rivers, suggesting a warm palaeoclimate for the Isle of Wight Eocene palaeobiota.