Systematic Entomology 12002) 27, 387±408

Phylogeny and of the and related taxa : Zygoptera: )

MICHAEL L. MAY Department of Entomology, Rutgers University, New Brunswick, New Jersey, U.S.A.

Abstract. The zygopteran genus Enallagma has been the subject of numerous behavioural and ecological studies, but phylogenetic relationships among have been examined only within eastern North America, and even the compos- ition and diagnosis of the genus are unclear on a world-wide basis. Most author- ities currently recognize about seventy species within Enallagma, comprising two major radiations, in North America and Africa. This study, using morphological data, demonstrates that the North American and a few related Palaearctic species form a monophyletic group that is quite distinct from the African species. The latter are themselves divided into at least three, and probably four, separate clades, one of which may be related to E. parvum of India. Consequently, three of Kennedy's long disused genera, , Amphiallagma and 1Kennedy, 1920) are resurrected and two new genera, gen.n. and Pinheyagrion gen.n. are established for the remaining African taxa. Finally, Enal- lagma is divided into two subgenera, Enallagma s.s., the typical `bluets', including many North American, Holarctic and Palaearctic species, and Chromatallagma subgen.n., comprising a group of species of more variable colour that is confined to North America, the Caribbean and northernmost South America.

Introduction These are of considerable interest, not only because they dominate coenagrionid assemblages in large Genus Enallagma, as it is usually understood today, is sections of temperate North America and also include very among the largest in Zygoptera, with about seventy species. common species in Europe and Africa, but also because they It is also a biogeographical anomaly, with two strong, have been the subjects of extensive behavioural, ecological disjunct centres of diversity: in North America 1c. forty and evolutionary study, e.g. adult mating behaviour and species), especially eastern North America, and in sub-Saharan reproductive success 1Bick & Bick, 1963; Bick & Hornuff, Africa 1c. twenty species), with a scattering of species in the 1966; Fincke, 1982, 1985, 1986; Martens & Grabow, 1994; Palaearctic and Oriental Regions. The marked concentra- Samways, 1994), adult dispersal 1McPeek, 1989; Anholt, tion of species in two areas that have been geologically 1990), larval behaviour 1Baker, 1986; Chowdhury & Corbet, separated since around the end of the Cretaceous 1Windley, 1989; McPeek, 1990b), population biology 1Parr, 1976; 1984), and that have no species in common, in itself suggests Macan, 1977; Anholt, 1992), guild and community structure the independent evolution of at least two clades in these 1Johnson et al., 1984; McPeek, 1990a; McPeek & Brown, regions. The principal aim of this study was to investigate, 2000), morphology of the reproductive system 1Srivastava using morphological characters and cladistic methods, & Srivastava, 1986), morphological development of larvae the phylogeny of this genus, with an aim to resolving the 1Pilon, 1982, 1994) and speciation 1McPeek & Brown, 2000). apparent paradox of its distribution and determining, in To place these studies in a sound comparative context, infor- broad outline, the pattern of evolutionary relationships mation on the phylogeny of this group is essential. Brown within it. et al. 12000) proposed a phylogeny including most of the North American species, based on morphology and on sequence data from the mitochondrial genes, COI and Correspondence: Michael L. May, Department of Entomology, COII. The current study extends their results, using morph- Rutgers University, Blake Hall, 93 Lipman Drive, New Brunswick, ology alone, to allow a preliminary understanding of relation- NJ 08901-8524, U.S.A. E-mail: [email protected] ships among most of the species world-wide.

# 2002 The Royal Entomological Society 387 388 Michael L. May

Part of the difficulty in determining the composition of, features that vary consistently among these taxa 1May, 1993), and relationships among, many coenagrionoid genera lies in and certain suites of such characters seem to be very well the facts that their venation is much reduced 1Munz, 1919; correlated with cladistic groupings. Note that characters were Williamson & Williamson, 1924) and many attributes that selected based on personal and others' examination of Enal- have been treated as discrete are continuously distributed lagma, in the broadest sense 1i.e. Bridges, 1994), but not, for 1O'Grady, 1998). This seriously restricts the utility of charac- the most part, on Coenagrionidae generally. Thus, inferred ters that are traditionally used to distinguish damselfly taxa. relationships among species not placed in Enallagma by Enallagma is generally characterized as having the wing Bridges are suspect because characters thought to indicate petiolation ending before Ac, postocular spots present, those relationships, but judged not useful for Enallagma,are females with a vulvar spine on the sternum of abdominal not included. These species were included because the sister segment 8, and M2 separating from M1 near the fifth and taxon to Enallagma is unclear, so it seemed prudent to fourth postnodal crossveins in fore- and hindwings, respect- include several possibly related taxa. ively 1de Selys-Longchamps, 1876; Westfall & May, 1996). The species include thirty-eight North American 1among These are all shared with numerous other taxa, however, which are five principally Central American or Antillean and the last is not constant among Enallagma. In practice, species: E. coecum, E. novaehispaniae, E. rua, E. semicirculare Enallagma is to a large extent recognized by the absence and E. truncatum; and one Holarctic species, the type species, of characters that distinguish related genera, e.g. the E. cyathigerum), one Oriental 1E. parvum) and sixteen dorsoapical prominence of abdominal segment 10 in African species 1including E. granti from Socotra Island) male . Appropriate synapomorphies have not been recognized and attributed to Enallagma by Bridges 11994), recognized that could serve to delineate this genus unam- and nine species not included in Enallagma. Taxonomic biguously. This is the first cladistic analysis to focus on changes subsequent to Bridges' list mostly follow May 11997): phylogenetic relationships among Enallagma and phenetic- E. camerunense is transferred to , E. kauderni is ally similar genera, as well as among species of Enallagma. considered distinct from E. nigridorsum, E. melanotum is The results suggest that the North American assemblage, transferred to Cercion, E. pseudelongatum is considered dis- plus the Palaearctic species closely related to E. cyathigerum, tinct from elongatum and E. strouhali is regarded as a junior constitute a monophyletic group that can be subdivided into synonym of E. risi. I have subsequently examined the para- at least two, and possibly three, clades corresponding more type male of E. caputavis Terzani & Carletti and found that or less to the three species groups proposed by Walker it actually belongs in Pseudagrion, a conclusion with which 11958). The Oriental and African species are probably poly- the describers concur 1Terzani, personal communication). phyletic and, at least in some instances, not closely related to Also, Enallagma brevispina Selys was earlier shown to be a true Enallagma. Kennedy 11920) proposed that the Oriental synonym of 1Rambur) 1Asahina, 1956), species, E. parvum, should be placed in a separate genus, and E. kagiensis 1Matsumura) a synonym of Amphiallagma, and that the African species be assigned to pygmeaea 1Rambur) 1Lieftinck et al., 1984). three different genera, Proischnura, Africallagma and Enallagma vaginale and E. longfieldae are doubtfully dis- Ischnallagma. These genera have not been widely accepted, tinct 1Pinhey, 1963) and are here considered a single taxon, but evidence is provided for the probable validity of all but although insufficient material was available to make a care- the last, and in addition two new genera and a new subgenus ful judgement of their relationship. Likewise the various within Enallagma are proposed. forms of E. coecum 1Westfall & May, 1996) are not distin- guished here, although they may encompass more than one valid taxon. The following species are not included in this Methods study due to lack of material: E. ambiguum Nava s, E. belyshevi Haritonov, E. buchholtzi Pinhey, E. deserti 1Selys), Character and species selection E. insula Fraser, E. maldivensis Laidlaw, E. nigrolineata Belyshev & Haritonov and E. optimolocus Miller & Ivie. Forty-seven morphological and one ecological/behavioural Thermagrion webbianum FoÈ rster was placed by Bridges 11994) character 1Appendix 1) were scored for sixty-six species of and others in Enallagma, apparently based on a speculation Coenagrionidae 1Fig. 1; Appendix 3). Structural characters by Pinhey 11962). The type has not been re-examined to my were examined and photographed using an HitachiTM S510 knowledge, and its whereabouts is unknown 1M. F. O'Brien, scanning electron microscope after coating with gold-palladium. personal communication). The identity of the species is For a number of species, especially from outside North difficult or impossible to determine from the description of America, some characters, mostly of larvae, could not be the single female specimen. Consequently, I see no basis for scored due to lack of material, although in a few cases moving it to Enallagma, and it is not treated here. information was supplemented from the literature. Character states are judged to be qualitatively discrete, although this was confirmed by morphometric analysis only for male stat- Analysis ure 1Appendix 1). Male or female secondary sexual charac- ters are emphasized throughout much of this analysis and in Cladistic analysis based on maximum parsimony was Figs 3±8 because these are among the relatively few structural carried out using heuristic searches 1100±1000 randomized

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 The damselfly genus Enallagma 389

Fig. 1. Phylogram of strict consensus tree summarizing ninety-six shortest 1190.50 steps) trees using `scaled', equally weighted characters 1see text) to estimate relationships among Enallagma and related genera. Character changes are shown as short vertical bars: filled, no homoplasy; half filled, unique appearance of one state of multistate character; unfilled closed, parallel homoplasy elsewhere on tree; open, multiple changes within clade where bar appears 1note that, once a character of this category appears on a branch, subsequent changes are also shown using the open bar symbol, e.g. ch. 2 is also shown as open when changing 0 ! 1 at the base of clade 28±60 and again when reversed in E. eiseni and clade 46±60). Changes are not shown on all terminal branches owing to space limitations. Generic identity of terminal taxa is shown as E. for species retained in Enallagma, `E.' for those transferred in this paper to other genera. Species are numbered consecutively from the basalmost clades, and these identifying numbers are used to designate species in Fig. 2 and Appendix 3. Species compositions of new or redefined genus-group taxa are indicated by vertical bars at right. Asterisk indicates E. cyathigerum, the type species of Enallagma.

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 390 Michael L. May

Fig. 2. Six strict consensus trees showing results of alternative character weightings 1note that, because of different weightings, tree lengths are not directly comparable). A, Weighting and tree topology as in Fig. 1, numbers above and below branches are decay indices and bootstrap values >50%, respectively; B, consensus of six trees, length 110.96, weighting as in A except that characters referring to colour or habitat preference were downweighted fivefold relative to structural characters; C, consensus of 501 trees, length 297, using `unscaled' equal weighting, i.e. giving each character state transition equal weight regardless of the number of states per character; D, consensus of 444 trees, length 164.80, weighting as in C except that characters referring to colour or habitat preference were downweighted fivefold relative to structural characters; E, consensus of sixty-three trees, length 54.07, starting with the ninety-six trees summarized by Fig. 1 and iteratively reweighting each character by its RCI until constant length was attained; F, consensus of fifty trees, length 126.04, starting with 100 bootstrap trees reweighted according to RCI using `best' option with multiple trees 1Kjer et al., 2001). Small arrowheads indicate position of Enacantha caribbea, vertical line in F indicates the unusual position of Africallagma cuneistigma, A. rubristigma, A. sapphirinum and A. vaginale.

replicates) on PAUP 4.0b4a 1Swofford, 1999). Although ing characters based on coloration and habitat 1because these several weightings were tested, except when otherwise noted, are often variable within species) by a factor of five, iterative each character was weighted equally regardless of character reweighting 1Farris, 1969) based on rescaled consistency type and the number of character states 1`scaled' equal index, and reweighting using the method of Kjer et al. 12001). weighting). Because some characters have multiple states, All characters are unordered, i.e. change is equally likely state transitions were, in effect, downweighted in proportion between any pair of states. In each analysis, trees were to the number of states for these characters relative to binary rooted using only conditum as the outgroup. characters. This results in fractional tree and branch lengths. Species with character scoring identical to another species Other weightings examined included equal weighting for were omitted to speed analysis without changing tree each state change 1`unscaled' equal weighting); downweight- lengths; these species included E. boreale, E. circulatum,

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 The damselfly genus Enallagma 391 and E. risi 1identical to E. cyathigerum), E. davisi 1branch lengths approximately proportional to the number 1ˆ E. recurvatum), E. doubledayi 1ˆ E. civile), E. praevarum of character state changes) in Fig. 1. Bars to the right indi- and E. rua 1ˆ E. anna) and E. truncatum 1ˆ E. signatum). cate hypothesized clades that I suggest should receive taxo- Decay indices 1Bremer, 1988) were calculated for selected nomic recognition 1see Discussion). Clearly, none of the branches by searching for the shortest trees that excluded major clades is strongly supported by either decay index that branch, i.e. prevented monophyly of the clade originat- or bootstrap values 1Fig. 2A), but note that this is true also ing from the ancestor represented by that branch. Support of basal branches where Ischnurainae 1Davies & Tobin, for separation of certain groups of taxa was evaluated by 1984; spelling, see Fet & Bechly, 2000) separate from pre- finding the shortest trees in which pairs of those groups sumably much more distantly related taxa. Confidence in were constrained to be monophyletic and comparing tree the reality of several important clades is substantiated by lengths with that of the shortest unconstrained trees. In their refractoriness to various alternative weightings addition, bootstrap analyses 1100 bootstrap replicates, five 1Fig. 2). random replicates per bootstrap replicate) were performed North American Enallagma species appear to form a for the scaled equal weighting trees. monophyletic group, except for the inclusion in the clade of Enacantha caribbea. Uncertainty exists about the place- ment of Enacantha 1Fig. 2B,F), but otherwise this conclu- Results sion is supported by all weightings except unscaled equal weighting, which leaves almost the entire tree unresolved A strict consensus tree of the ninety-six shortest trees 1Fig. 2C). The possibility clearly must be considered that 1length 190.50) is presented in the form of a phylogram Enacantha should be subsumed within Enallagma. The male

Fig. 3. Scanning electron micrographs of penes and male caudal appendages of putative outgroup genera. A, Oblique dorsolateral view of cerci and paraprocts of Chromagrion conditum; B, lateral view of distal portion of penis of C. conditum; C, ventrolateral view of distal portion of penis of Enacantha caribbea; D, oblique internal view of left cercus of E. caribbea; arrow indicates scalariform cuticle; E, detail of C. showing apparent scalariform cuticle 1arrow). Scales ˆ 0.05 mm 1E); 0.25 mm 1B,C); 0.50 mm 1A).

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 392 Michael L. May

Fig. 4. Scanning electron micrographs of penes and male caudal appendages of representative Enallagma, including Chromatallagma, subgen.n. A, Dorsal view of caudal appendages of E.1Enallagma) aspersum; B, oblique internal view of left cercus of E. aspersum showing scalariform cuticle 1arrow); C, detail of scalariform cuticle from B; D, as in A, E.1Enallagma) cyathigerum;E,asinB,E. cyathigerum, with transversely elongated terminal hook of reduced dorsal arm and scalariform cuticle overlapping terminal hook 1arrows); F, extreme tip of right cercus of E. cyathigerum showing portion of terminal hook and overlapping scalariform cuticle 1arrows); G, lateral view of male caudal appendages of Enallagma eiseni, with short, dorsally directed paraproct 1arrow); H, as in B, E. eiseni 1cf. Fig. 3D); I, oblique internal view of right cercus of E. 1Chromatallagma) signatum, inferior branch fused to superior branch 1arrows), scalariform cuticle covering medial surface of both; J, as in B, right side, E. 1Chromatallagma) antennatum; K, detail of J., scalariform cuticle absent on inferior branch of cercus; L, ventral view of distal portion of penis of E. cyathigerum, showing distolateral and lateral lobes 1arrows); M, as in L, E. antennatum; N, lateral view of distal portion of penis of E. signatum; arrow indicates large bladelike internal spine. Scales ˆ 0.01 mm 1C); 0.10 mm 1E,K); 0.25 mm 1B,L,M); 0.50 mm 1A,D,F,G±J,N).

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 The damselfly genus Enallagma 393

Fig. 4. Continued. cerci are very similar, especially to Enallagma eiseni monophyletic with respect to the other so-called Enallagma, 1Figs 3D, 4H) although, as noted by Donnelly & Alayo although they may be close to E. parvum and the 11966), the penis of Enacantha is highly autapomorphic E. nigridorsum species group. They are unique in possessing 1Fig. 3C) and the male pterostigmata are bicoloured. For- scalarifom cuticle 1Fig. 4B,C,E,F; secondarily lost in cing monophyly between North American clade 1excluding E. antennatum, Fig. 4K), a characteristic hook terminating Enacantha) and the various African and Oriental taxon the superior arm of the male cercus, and in penis structure groupings increases tree length by one to four steps 1Fig. 4). The morphology of the cercus is shown very clearly 1Table 1). North American species are thus probably by E. aspersum 1Fig. 4A±C) and in considerably modified

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 394 Michael L. May

Table 1. Added steps required to force the taxon sets indicated in columns 1 and 2 into monophyly, assuming `scaled', equal character weighting, as in Fig. 1.

Merged taxon sets Added steps

All African `Enallagma'N.A.Enallagma 4.00 `E.' angolicum N. A. Enallagma 4.12 `E.' subfurcatum group N. A. Enallagma 2.46 `E.' glaucum group* N. A. Enallagma 2.62 Extended `E.' glaucum group** N. A. Enallagma 2.50 `E.' nigridorsum group N. A. Enallagma 1.12 `E.' parvum N. A. Enallagma ‡ Enacantha 0.33 E. signatum group Bluet group 5.71 E. signatum group Bluet group ‡ E. durum 3.83 E. signatum group ‡ E. basidens Bluet group 5.00 E. signatum group ‡ E. basidens Bluet group ‡ E. durum 3.46 E. divagans ‡ E. traviatum groups ‡ E. antennatum Bluet group 4.88 E. divagans ‡ E. traviatum groups ‡ E. antennatum Bluet group ‡ E. durum 4.33 E. coecum group Bluet group ‡ E. durum 2.00 E. coecum group E. signatum ‡ E. divagans ‡ E. traviatum groups 0.00

*Excludes E. cuneistigma, E. rubristigma, E. sapphirinum and E. vaginale. ** Includes the four species above. form by E. cyathigerum, in which the superior arm is greatly Proischnura, are also well separated from their supposed reduced 1Fig. 4D±F) and E. signatum, in which the superior congeners and closely grouped with the heretofore enig- and inferior arms are fused 1Fig. 4I). matic E. polychromaticum 1Fig. 1; in fact the penes and Among North American species, two primary clades are male cerci are rather similar, Fig. 5), with the former having evident, one including the typical bluets, E. cythigerum, triangular projections just before the distal flexure of the E. civile, etc. 1Appendix 2), the second a visually diverse penis. This general placement is also supported by most species assemblage including a number of relatively slender weighting assumptions 1Fig. 2), and is not inconsistent blue species, plus the very distinctive orange E. signatum with the others. Enallagma angolicum also often is grouped species group. Only the E. signatum group has a relatively with these three species 1Fig. 2), with which it shares high decay index and bootstrap support. Of uncertain subterminal processes on the penis 1Fig. 5), but it differs placement are E. eiseni, E. coecum ‡ E. novaehispaniae and markedly in size, coloration, cercal structure and the form E. durum 1Figs 1, 2). The bluet group itself has relatively of the lateral penile projections. It seems safe to say that weak support and is not recovered by all weightings, but E. angolicum is not related to North American Enallagma other evidence 1see Discussion) allows for some confidence nor, in all probability, to other African species outside the in its monophyly. It is characterized in most species by E. subfurcatum group. All these species have very well extensive dorsal blue colour of the male abdominal seg- developed lappets on the male mesostigmal plates, as in ments and by the characteristic penis shape 1Fig. 4G; also the E. glaucum group 1see below) but unlike any North see Donnelly, 1963). American Enallagma. The second 1`non-bluet') clade is recovered by all weighting Two additional clades are recovered consistently. The schemes except unscaled equal weighting. In some cases E. glaucum group 1E. glaucum, E. subtile, E. elongatum, 1Fig. 2), E. coecum and E. novaehispaniae 1the `coecum- E. pseudelongatum, E. sinuatum) appears under all weight- group') are included at the base of this clade. Salient char- ings, except that E. vaginale is included within it in one acters include a dark male abdomen and, again, distinctive instance 1Fig. 2D). A somewhat more heterogeneous group, penis morphology 1Fig. 4L,N; shared with E. durum and consisting of E. cuneistigma, E. vaginale, E. rubristigma E. eiseni; Donnelly, 1963). Forcing the major subgroups of and E. sapphirinum, usually appears as a sister clade the non-bluet clade into monophyly with the bluets increases to the E. glaucum group 1but see Fig. 2F). All of these tree length by about 3.5±5.5 steps; some minimum-length species except E. cuneistigma share among themselves, and trees place the E. coecum group in monophyly with the with E. angolicum and the E. subfurcatum group but not non-bluets whereas at least 2.0 additional steps are required with any North American species, the possession by males to generate a clade grouping the former with the bluets of paraprocts that are clearly bifurcated in lateral view 1Table 1). 1Fig. 6) and all have distinct, raised ridges or lappets on the All species that were placed outside Enallagma by Bridges posteromedial corner of the mesostigmal laminae, that may 11994), except Enacantha caribbea, group well away from serve to prevent intrasexual tandem pairing 1Fig. 8; con- North American Enallagma. Of the African Enallagma, sidered to be present only if this area was distinctly more E. subfurcatum and E. rotundipenne, placed by Kennedy in elevated in males than females). Most of the E. glaucum

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 The damselfly genus Enallagma 395

Fig. 5. Sketch and scanning electron micrographs of penes and male caudal appendages of species formerly placed in Enallagma but here transferred to Pinheyagrion and Proischnura. A, Lateral view of male caudal appendages of Pinheyagrion angolicum, arrow indicates pad of flexible cuticle; B, oblique ventrolateral view of distal portion of penis of P. angolicum, with bifurcated spine 1arrow); C, lateral view of caudal appendages of Proischnura subfurcatum; D, oblique internal view of left cercus of P. subfurcatum; E, detail of D showing scalariformlike cuticle and glabrous thumblike projection 1arrows); F, ventral view of distal portion of penis of P. subfurcatum, with tonguelike projection 1arrow); G, as in E, P. polychromaticum;H,asinF,P. polychromaticum. Scales ˆ 0.1 mm 1E,G); 0.2 mm 1B,F,H); 0.5 mm 1D); 1.0 mm 1C).

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 396 Michael L. May

Fig. 6. Scanning electron micrographs of penes and male caudal appendages of species formerly placed in Enallagma but here transferred to Africallagma: A, Lateral view of male caudal appendages of , arrow indicates superior branch of paraproct; B, oblique internal view of left cercus and paraproct of A. glaucum showing scalariformlike cuticle 1arrow); C, as in A, A. elongatum;D,asinB, A. elongatum; E, ventrolateral view of distal portion of penis of A. glaucum showing distolateral and lateral lobes 1arrows); F, lateral view of distal portion of penis of A.1?)rubristigma showing distolateral lobe 1arrow). Additional figures of cerci in Fraser 11947: Fig. 1). Scales ˆ 0.5 mm.

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 The damselfly genus Enallagma 397 group are also characterized by a low anteromedial well resolved tree 1Fig. 1), I suggest the equally weighted trough and flange on the female mesostigmal plate tree using scaled characters as most likely to reflect the true 1Fig. 8). phylogeny, but attention to other weighting possibilities is The other African species, comprising the E. nigridorsum important. For example, downweighting colour and eco- group 1May, 1997), plus E. parvum, apparently form a clade logical characters by up to fivefold relative to structural with the North American species but as a group are sister to ones does not change the main relationships, although it the North American species. Although decay index support reduces resolution within some groups 1Fig. 2), nor does is low, the African species are united by the very similar and specifically upweighting penis shape fourfold. distinctive structure of the male cerci 1Fig. 7), with three The North American portion of the tree receives very longitudinal lobes, the upper one hooked medially and the substantial support from the work of Brown et al. 12000). middle one of rugose, slightly flexible cuticle 1usually pale in Although they examined fewer species, their data included a life). Enallagma parvum is usually placed as sister taxon to slightly different morphological dataset as well as sequence the E. nigridorsum group but sometimes 1Fig. 2B) to North data from COI and COII mtDNA, and maximum likeli- American Enallagma plus Enacantha. The cercal structure hood as well as parsimony analysis. Their results very largely of E. parvum is rather different from the African support those reported here, and in particular support the E. nigridorsum group 1Fig. 7). All of these species agree monophyly of the bluet clade, including E. durum, very with North American Enallagma in that male paraprocts strongly. Their results also suggest recent, very rapid mor- are not bifurcated 1a probable plesiomorphy). Lappets on phological evolution within this clade, which may have the mesostigmal laminae are vestigially present 1Fig. 8) made it more difficult to recover it consistently using mor- except in E. parvum, where they are absent. phological characters alone 1Fig. 2). More recent unpublished molecular data from these mitochondrial genes 1McPeek, personal communication) Discussion places Palaearctic specimens of E. cyathigerum, as well as Palaearctic species not included here 1E. deserti Selys, A glance at RCI values in Appendix 1 clearly indicates one E. risi, E. circulatum), well within a monophyletic group reason for the low branch support provided by conven- including the North American bluets. Furthermore, the tional measures À most characters are subject to extensive African E. glaucum, E. nigridorsum and E. pseudelongatum homoplasy or loss, or both. Perhaps the most characteristic are all basal to Ischnura ramburii and I. verticalis in a tree identifying feature of North American Enallagma is the rooted by resolutum, and thus they may be even presence on the inferior branch of the cerci 1Fig. 4) of more widely separated from North American Enallagma `scalariform cuticle' 1Jurzitza, 1975; this is probably an adap- than suggested here. tation for gripping the female prothorax during tandem The phylogenetic conclusions presented here suggest pairing), yet this appears to have been lost in E. antennatum, several taxonomic changes. For example, E. subfurcatum, perhaps shifted to the superior branch in E. eiseni, and E. rotundipenne and E. polychromaticum form a valid genus to be present in much reduced form in E. basidens.It Proischnura that is distinct from Enallagma s.s.; they are should be noted at this point that many coenagrionids have clearly united by the extremely similar penile morphology areas of modified cuticle on the cerci that may function in and structure of the cerci 1Barnard, 1937; Robertson & tandem linkage 1the `middle lobe' on the cerci of the Patterson, 1982). Although E. angolicum could be included E. nigridorsum group of species is a likely example; Fig. 7). here, it is phenotypically so distinct that a very close Scalariform cuticle, however, is specifically characterized as relationship appears unlikely; it clearly is not an Enallagma consisting of an array of small protrusions, usually closely but resembles in some respects and spaced 1Fig. 4C), by being relatively pale and flexible, at related species. I propose here a new, monotypic genus, least in freshly mature specimens, and in its position on Pinheyagrion gen.n., for this species 1see Appendix 2 for the inferior cercal branch or a structure identifiably derived diagnosis). from this branch. Substantial justification exists for separating the other In addition to the problem of homoplasy, scoring of some non-North American species at the generic level. In par- characters is uncertain. The broad triangular projections of ticular, the E. glaucum group is well separated phenotypically the middle penis segment of E. subfurcatum are scored as the and cladistically from true Enallagma. Thus, I propose that same character state as the bifurcated ones of E. angolicum this group be recognized as genus Africallagma Kennedy 1of 1Fig. 5), or the spines in this position in Ischnura verticalis which Ischnallagma becomes a junior synonym). The relation- 1see, e.g. Robinson & Novak, 1997), but there surely is room ships of the four outliers to this group., i.e. E. cuneistigma, for doubt about the homology of these structures. E. vaginale, E. rubristigma and E. sapphirinum, to one Character weighting is also problematic. Several charac- another and to the E. glaucum group are uncertain. In ters have numerous possible states, notably thoracic colour these species the male cerci or the penes, or both, differ and shape of the terminal penis segment. These could, of somewhat from the core E. glaucum group, although these course, be resolved into sets of binary characters, but the characters are somewhat variable within the latter as well result would be, in effect, to weight these characters very 1e.g. cerci of E. glaucum vs. E. elongatum, Fig. 6; in most highly. For this reason, and because the result gives a very species the cerci are more similar to those of E. glaucum).

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 398 Michael L. May

Fig. 7. Scanning electron micrographs of penes and male caudal appendages of species formerly placed in Enallagma but here transferred to Amphiallagma and Azuragrion. A, Dorsal view of male caudal appendages of Amphiallagma parvum, showing dorsal tubercle 1arrow); B, lateral view of distal portion of penis of A. parvum showing distolateral lobe 1arrow); C, lateral view of male caudal appendages of Azuragrion nigridorsum showing superior branch and inferior protrusion of paraproct 1arrows); D, oblique internal view of left cercus and paraproct of A. nigridorsum showing subterminal dorsal hook, middle lobe, and inferior arm of cercus 1arrows); E, lateral view of distal portion of penis of A. nigridorsum showing distolateral lobe 1arrow); F, as in C, A. somalicum 1specimen is uncoated); G, as in D, A. somalicum 1uncoated). Additional figures in May 11997). Scales ˆ 0.2 mm 1B,E); 0.5 mm 1A,C,D,F,G).

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 The damselfly genus Enallagma 399

Fig. 8. Scanning electron micrographs of dorsal views of mesostigmal plates and posterior margin of pronotum. A, Africallagma glaucum male showing posteromedial lappet 1arrow); B, A. glaucum, female showing anteromedial trough and flange 1arrow); C, Azuragrion nigridorsum, male showing vestigial posteromedial lappet 1arrow); D, A. nigridorsum, female; E, Enallagma 1Enallagma) aspersum, female showing pronotal pits 1arrow). Scales ˆ 0.5 mm.

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 400 Michael L. May

I suggest that all these species be placed provisionally Acknowledgements in Africallagma, comb.n., pending a more detailed study. Support for separation of the E. nigridorsum group is less I am grateful to Bill Mauffray and Minter Westfall 1Florida strong, but in addition to the evidence based on the phyl- State Collection of ), O. S. Flint and Nancy ogeny itself, the biogeography supports independent evolu- Adams 1Smithsonian Institution, National Museum of Nat- tion. This is especially true because the Enallagma that are ural History), Margie Cochrane 1South African Museum), geographically nearest to E. nigridorsum and its relatives, Don Azuma 1Academy of Natural Sciences of Philadelphia) and that are clearly closely related to North American and Rudo Sithole 1National Museum of ) for the species 1e.g. E. deserti in North Africa and south into loan of specimens in their care, and to David Goodger for ; D'Andrea & Carfi, 1994), are probably among the assistance in allowing me to examine the collection at The most recently evolved species 1Brown et al., 2000) and are Natural History Museum, London. Special thanks are due very different from all species of the E. nigridorsum group in Viola Clausnitzer for providing numerous African speci- coloration and genital structure. A similar argument for mens, Wolfgang Schneider for the generous gift of a male removal from Enallagma applies to E. parvum, for which and female E. granti and Michael Samways for the loan of the genus name Amphiallagma Kennedy is available. reared material of African species. I am also indebted to Whether or not the African E. nigridorsum group should Kiyoshi Inoue and Shigeru Tsuda for valuable references on also be placed in Amphiallagma is problematic. The very synonymy. This is New Jersey Agricultural Experiment close similarity of the male cerci within this group, and the Station Publication D-08155-12-01. relative distinctiveness of those of E. parvum, support the monophyly of the African species convincingly. Given that fact, the geographical gap between the latter and E. parvum, References and the relatively weak support for the monophyly of E. parvum plus the E. nigridorsum group, I suggest that the Anholt, B.R. 11990) Sized-based dispersal prior to breeding in a latter be placed in a new genus Azuragrion 1see Appendix 2 damselfly. Oecologia, 83, 385±387. for diagnosis), with E. nigridorsum designated herewith as Anholt, B.R. 11992) Sex and habitat differences in feeding by an the type of the genus. Clearly, information about the poorly adult damselfly. Oikos, 65, 428±432. known species, E. insula Fraser 1mainland India) and Asahina, S. 11956) An annotated list of Japanese Odonata, with E. maldivensis Laidlaw 1Maldive Archipelago), might shed brief descriptions of their larvae. 3. Agrionidae, 2. Shin-Konchu, light on relationships of the African and Indian taxa, but 9, 47±51 1in Japanese). these are known only from the type series, which are Baker, R.L. 11986) Effects of density, disturbance, and waste badly damaged 1E. maldivensis; Laidlaw, 1919; Olsvik & products on growth of larval Enallagma ebrium 1Hagen) HaÈ maÈ laÈ inen, 1992; personal observation) or of uncertain 1Odonata: Coenagrionidae). Canadian Entomologist, 118, whereabouts 1E. insula). 325±328. Barnard, K.H. 11937) Notes on dragon-flies 1Odonata) of the The taxonomy of North American species also deserves S. W. Cape with descriptions of the nymphs and of new species. examination. The bluet and non-bluet species groups seem Annals of the South African Museum, 32, 169±260. clear and consistent enough to deserve recognition at the Bick, G.H. & Bick, J.C. 11963) Behaviour and population structure subgeneric level. I therefore propose the name Chromatal- of the damselfly, Enallagma civile 1Hagen) 1Odonata: Coena- lagma subgen.n. for the latter, in reference to the bright and grionidae). Southwestern Naturalist, 8, 57±84. variable colour of these 1see Appendix 2 for diag- Bick, G.H. & Hornuff, L.E. 11966) Reproductive behaviour in the nosis and species composition); Enallagma signatum is here- damselflies 1Hagen) and Enallagma with designated the type species. Enallagma coecum and exsulans 1Hagen) 1Odonata: Coenagrionidae). Proceedings of E. novaehispaniae probably also belong here. It may event- the Entomological Society of Washington, 68, 78±85. ually prove justifiable and useful to divide the new subgenus Bremer, K. 11988) The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution, 42, 795±803. further, particularly in view of the strong differentiation of Bridges, C.A. 11994) Catalogue of the Family-Group, Genus-Group the E. signatum group and the molecular support 1Brown and Species-Group Names of the Odonata of the World, 3rd edn. et al., 2000) for the sister-group relationship of the latter Privately published, Urbana, Illinois. to E. antennatum and the E. divagans and E. traviatum Brown, J.M., McPeek, M.A. & May, M.L. 12000) A phylogenetic groups. perspective on habitat shifts and diversity in the North American The placement of E. eiseni remains uncertain, largely Enallagma damselflies. Systematic Biology, 49, 697±712. because of the unusual colour pattern, and because the Chowdhury, S.H. & Corbet, P.S. 11989) Feeding behaviour in is unknown. This may account for its position basal larvae of 1Charpentier) 1Zygoptera: to Enacantha with some weightings. Clarification of its Coenagrionidae). Odonatologica, 18, 285±288. status requires further information on larval and molecular D'Andrea, M. & Carfi, F. 11994) Annotations on a small collection from Ghana, West Africa, with six new species for the characters. In view of the strong phenotypic differentiation national fauna. Opuscula Zoologica Fluminensia, 124, 1±7. of Enacantha caribbea and the instability of its position, Davies, D.A.L. & Tobin, P. 11984) The of the world: a particularly with respect to E. eiseni, under different char- systematic list of the extant species of Odonata, Vol. I. acter weightings 1Fig. 2), I propose that this monotypic Zygoptera, Anisozygoptera. Societas Internationalis Odonatolo- genus continue to be recognized as such. gica Rapid Communications 0Suppl.), 3, 1±128.

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 The damselfly genus Enallagma 401

Donnelly, T.W. 11963) Possible phylogenetic relationships among Abstracts of Papers. XII International Symposium of Odona- North and Central American Enallagma. Proceedings of the tology, Osaka. North Central Branch, Entomological Society of America, 18, May, M.L. 11997) The status of some species of Enallagma 116±119. 1Zygoptera: Coenagrionidae). Entomological News, 108, 77±91. Donnelly, T.W. & Alayo, P. 11966) A new genus and species of May, M.L. & Corbet, P.S. 12001) Occurrence and taxonomic damselfly from Guatemala and Cuba 1Odonata: Coenagrioni- significance of a palpal spine in larvae of Enallagma and other dae). Entomologist, 49, 107±117. genera 1Odonata: Coenagrionidae). International Journal of Farris, J.S. 11969) A successive approximations approach to Odonatology, 4, 41±49. character weighting. Systematic Zoology, 18, 374±385. McPeek, M.A. 11989) Differential dispersal tendencies among Fet, V. & Bechly, G. 12000) Ischnurainae Fraser, 1957 1Insecta, Enallagma damselflies 1Odonata) inhabiting different habitats. Odonata): proposed conservation as the correct spelling of Oikos, 56, 187±195. Ischnurinae to remove homonymy with Ischnuridae Simon, 1879 McPeek, M.A. 11990a) Determination of species composition in the 1Arachnida, Scorpiones). Bulletin of Zoological Nomenclature, Enallagma damselfly assemblages of permanent lakes. Ecology, 57, 26±28. 71, 83±98. Fincke, O.M. 11982) Lifetime mating success in a natural McPeek, M.A. 11990b) Behavioral differences between Enallagma population of the damselfly, 1Walsh) species 1Odonata) influencing differential vulnerability to 1Odonata: Coenagrionidae). Behavioural Ecology and Socio- predators. Ecology, 71, 1714±1726. biology, 10, 293±302. McPeek, M.A. & Brown, J.M. 12000) Building a regional species Fincke, O.M. 11985) Alternative mate-finding tactics in a non- pool: diversification of the Enallagma damselflies in eastern territorial damselfly 1Odonata: Coenagrionidae). Behav- North American waters. Ecology, 81, 904±920. iour, 33, 1124±1137. Munz, P.A. 11919) Venational study of Zygoptera with keys to Fincke, O.M. 11986) Lifetime reproductive success and the genera. Memoirs of the American Entomological Society, 3, 1±78. opportunity for selection in a non-territorial damselfly O'Grady, E.W. 11998) A cladistic analysis based on morphological 1Odonata: Coenagrionidae). Evolution, 40, 791±803. characters for subfamilies within Coenagrionidae 1Zygoptera: Fraser, F.C. 11947) The synonymy of Enallagma elongatum Odonata). MSc Thesis, Rutgers University, New Brunswick, 1Martin) and E. pseudelongatum Longfield and their differenti- New Jersey. ation from a new African species 1order Odonata). Proceedings of Olsvik, H. & HaÈ maÈ laÈ inen, M. 11992) Dragonfly records from the the Royal Entomological Society, London, Series B, 16, 143±148. Maldive Islands, Indian Ocean 1Odonata). Opuscula Zoologica Johnson, D.M., Bohannan, R.E., Watson, C.N. & Martin, T.H. Fluminensia, 89, 1±7. 11984) Coexistence of and Enallagma Parr, M.J. 11976) Some aspects of the population ecology of the traviatum 1Zygoptera: Coenagrionidae) in Bays Mountain Lake: damselfly Enallagma cyathigerum 1Charpentier) 1Zygoptera: an in situ enclosure experiment. Advances in Odonatology, Coenagrionidae). Odonatologica, 5, 45±57. 2, 57±70. Pilon, J.-G. 11982) Notes on oviposition and embryonic develop- Jurzitza, G. 11975) Rasterelektronenmikroskopische Untersuchen- ment of Enallagma ebrium 1Hagen) and E. vernale Gloyd in gen an dem Appendices und der Laminae mesostigmales eineger Quebec 1Zygoptera: Coenagrionidae). Odonatologica, 11, 45±52. Enallagma-Arten 1Odonata, Zygoptera). Forma et Functio, 8, Pilon, J.-G. 11994) Morphologie larvaire d'Enallagma carunculatum 33±48. Morse 1Zygoptera: Coenagrionidae). Odonatologica, 23, 133±147. Kennedy, C.H. 11920) Forty-two hitherto unrecognized genera and Pinhey, E.C.G. 11962) A descriptive catalogue of the Odonata of subgenera of Zygoptera. Ohio Journal of Science, 21, 83±88. the African continent 1up to December 1959). PublicacËoÄes Kjer, K.M., Blahnick, R.J. & Holzenthal, R.W. 12001) The Culturais da Companhia de Diamantes de Angola, 59, 1±323. phylogeny of Trichoptera. Systematic Biology, 50, 781±816. Pinhey, E.C.G. 11963) Some anomalous types of African Odonata Laidlaw, F.F. 11919) A list of the dragonflies recorded from the and the description of a new species. Journal of the Entomo- Indian Empire with special reference to the collection of the logical Society of Southern Africa, 26, 146±160. Indian Museum, Part II. Records of the Indian Museum, 16, Pinhey, E.C.G. 11971) Odonata of Fernando Po Island and of 169±195. neighbouring Cameroons territory. Journal of the Entomological Lieftinck, M.A., Lien, J.C. & Maa, T.C. 11984) Catalogue of Society of Southern Africa, 34, 215±230. Taiwanese Dragonflies 0Insecta: Odonata). Asian Ecological Robertson, H.M. & Patterson, H.E.H. 11982) Mate recognition Society, Taichung, Taiwan. and mechanical isolation in Enallagma damselflies 1Odonata: Longfield, C. 11931) A list of the Odonata of British Somaliland Coenagrionidae). Evolution, 36, 243±250. and a description of a new species of the genus Enallagma. Robinson, J.V. & Novak, K.L. 11997) The relationship between Entomologist, 64, 274±278. mating system and penis morphology in ischnuran damselflies Longfield, C. 11936) Studies on African Odonata, with synonymy 1Odonata: Coenagrionidae). Biological Journal of the Linnean and descriptions of new species and subspecies. Transactions of Society, 60, 187±200. the Royal Entomological Society, London, 85, 467±498. Samways, M.J. 11994) `Sailing' on the water surface by male Macan, T.T. 11977) A twenty year study of the fauna in the Enallagma nigridorsum Selys 1Zygoptera: Coenagrionidae). vegetation of a moorland fishpond. Achiv fuÈr Hydrobiologie, 81, Odonatologica, 23, 175±178. 1±24. de Selys-Longchamps, E. 11876) Synopsis des agrionines, Martens, A. & Grabow, K. 11994) Males of Enallagma vansomereni cinquieÁme le gion: Agrion 1suite). Le genre Agrion. Bulletin Pinhey settling on water surface 1Zygoptera: Coenagrionidae). de l'AcadeÂmie Royale des Sciences de Belgique 02), 41, 247±322, Odonatologica, 23, 169±174. 496±539, 1233±1309 1reprint 1±199). May, M.L. 11993) Understanding the taxonomic significance of the Srivastava, V.K. & Srivastava, B.K. 11986) Morphology of the male caudal appendages of Zygoptera ± the case of Enallagma. male reproductive organs of Enallagma parvum Selys

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 402 Michael L. May

1Zygoptera: Odonata). Proceedings of the National Academy of Williamson, E.B. & Williamson, J.H. 11924) A remarkable new Science of India, 56, 13±19. genus of Coenagrionidae from Brazil, with descriptions of three Swofford, D.L. 11999) PAUP*. Phylogenetic Analysis Using Parsi- new species. Occasional Papers of the University of Michigan mony and Other Methods, Version 4. Sinauer Associates, Museum of Zoology, 154, 1±24. Sunderland, Massachusetts. Windley, B.F. 11984) The Evolving Continents. 2nd edn. John Wiley Walker, E.M. 11953) The Odonata of Canada and Alaska, Vol. 1. & Sons, New York. University of Toronto Press, Toronto. Westfall, M.J. & May, M.L. 11996) Damselflies of North America. Scientific Publishers, Gainesville, Florida. Accepted 29 October 2001

Appendix 1. Character list. Colour 0females)

Characters illustrated by Westfall & May 11996) are indi- 14. Humeral stripes: 10) entire; 11) longitudinally divided cated as `W & M: Fig. ± ' ; other illustrations are in the for less than half their length; 12) longitudinally divided present paper except for character 47. Note that character for most of length; W & M: Fig. 127. CI ˆ 0.400, state designations have no significance with respect to RI ˆ 0.500. polarity. Consistency 1CI) and retention 1RI) indices are 15. Black on dorsum of abdominal segment 1: 10) covering modal values for the ninety-six shortest trees. more than basal two-thirds; 11) less than basal two- thirds. CI ˆ 0.500, RI ˆ 0.833. 16. Abdominal segment 8 dorsally: 10) mostly pale; 11) about Colour 0males) equally pale and dark; 12) mostly dark. CI ˆ 0.400, RI ˆ 0.182. 1. Pale thoracic colour:10)blue;11)yellow/orange; 17. Abdominal segment 9 dorsally: 10) mostly pale; 11) about 12) green; 13) violet; 14) otherwise. CI ˆ 0.444, equally pale and dark; 12) mostly dark. CI ˆ 0.400, RI ˆ 0.583. RI ˆ 0.647. 2. Black on dorsum of abdominal segment 2: 10) extending 18. Abdominal segment 10 dorsally: 10) mostly pale; 11) about more than half segment length; 11) less than half equally pale and dark; 12) mostly dark. CI ˆ 0.167, segment length. CI ˆ 0.125, RI ˆ 0.632. RI ˆ 0.600. 3. Abdominal segments 3±6 dorsally: 10) extensively pale; 11) intermediate; 12) almost entirely dark; W & M: Fig. 12. CI ˆ 0.250, RI ˆ 0.625. Secondary sexual characters 0males) State 2 includes species with extensive lateral encroach- ment of pale colour but with a distinct middorsal dark 19. Cerci with separate ventrobasal lobe: 10) present; stripe the length of all or most segments. 11) secondarily fused; 12) absent; Figs 4±7. CI ˆ 0.400, 4. Lateral or ventrolateral black markings on abdominal RI ˆ 0.667. segment 8: 10) present; 11) absent; W & M: Fig. 13. 20. Cerci with dorsal lobe: 10) hooked; 11) not hooked; CI ˆ 0.571, RI ˆ 0.250. Figs 4±7. CI ˆ 1.000, RI ˆ 1.000. 5. Abdominal segment 8 dorsally: 10) mostly pale; 11) about 21. Cerci with scalariform or scalariformlike cuticle: equally pale and dark; 12) mostly dark. CI ˆ 0.273, 10) present on ventrobasal lobe; 11) present elsewhere; RI ˆ 0.529. 12) absent; Figs 4±7. CI ˆ 0.333, RI ˆ 0.486. See text 6. Abdominal segment 9 dorsally: 10) mostly blue; 11) mostly for explanation. orange; 12) mostly dark. CI ˆ 0.800, RI ˆ 0.750. 22. Middle lobe of cerci: 10) present; 11) absent; Figs 4±7. 7. Abdominal segment 10 dorsally: 10) mostly pale; CI ˆ 0.500, RI ˆ 0.800. 11) about equally pale and dark; 12) mostly dark. 23. Paraprocts: 10) distinctly bifid; 11) slightly bifid; 12) not CI ˆ 0.333, RI ˆ 0.333. at all bifid; Figs 4±7. CI ˆ 0.286, RI ˆ 0.750. 8. Middorsal carina of male: 10) almost entirely pale; 24. Paraprocts: 10) very short and directed upward; mainly dark 11). CI ˆ 0.0.357, RI ˆ 0.357. 11) not so; W & M: Figs 143 and 148. CI ˆ 0.333, 9. Humeral stripe constricted before mesopleural fossa: RI ˆ 0.333. 10) commonly; 11) never. CI ˆ 0.167, RI ˆ 0.444. 25. Lateral spines just proximal to distal flexure of penis: 10. Postclypeus: 10) mostly pale; 11) about equally pale and 10) present; 11) absent; Figs 4±7. CI ˆ 0.500, RI ˆ 0.750. dark; 12) mostly dark. CI ˆ 0.429, RI ˆ 0.273. 26. Penis with terminal segment: 10) quadrilobate distally; 11. Postfrons: 10) mostly pale; 11) about equally pale and 11) crescentic distally; 12) with earlike distolateral lobes; dark; 12) mostly dark. CI ˆ 0.300, RI ˆ 0.731. 13) with short, digitate distolateral lobes; 14) broad with 12. Pale spots anterior to lateral ocelli: 10) present; subquadrate distolateral lobes; 15) with distolateral 11) absent. CI ˆ 0.0.250, RI ˆ 0.647. lobes recessed, long and spatulate; 16) with distolateral 13. Area between ocelli: 10) pale; 11) dark; W & M: Fig. 128. lobes filamentous; 17) very narrow, abruptly expanded CI ˆ 1.000, RI ˆ 1.000. distally to reflexed, undulate distolateral lobes; 18) very

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 The damselfly genus Enallagma 403

short and broad with broadly rounded distolateral lobes; 42. Posterolateral margin of head: 10) prominent; 110) not Figs 4±7. CI ˆ 0.667, RI ˆ 0.867. prominent; W & M: Fig. 134. CI ˆ 0.0.167, RI ˆ 0.583. The large number of character states partly reflects the 43. Numerous stout setae on lateral carinae of abdomen: diversity of penis shapes among outgroups; shapes were 10) present; 11) absent; W & M; Fig. 135. CI ˆ 0.222, not combined into one state when very distinct to avoid RI ˆ 0.875. providing artificial support for clades with disparate shapes. 44. Lateral carinae of abdomen: 10) projecting abruptly 27. Shaft of penis: 10) with 1 or 2 large, bladelike internal toward posterior end; 11) not so; W & M: Fig. 135. spines; 11) with several small spines; 12) without spines or CI ˆ 1.000, RI ˆ 1.000. with small denticles only; Figs 4±7. CI ˆ 0.250, RI ˆ 0.400. 45. Extensive and well defined areas of extra-tracheal 28. Penis with distinct lateral lobes: 10) present; 11) absent; pigmentation on gills: 10) present; 11) absent; W & M: Figs 4±7. CI ˆ 0.143, RI ˆ 0.538. Fig. 136. CI ˆ 0.333, RI ˆ 0.778. 29. Mesepisternal lappets: 10) present; 11) vestigial; 46. Well developed series of antenodal setae on median gill: 12) absent; Fig. 8. CI ˆ 0.400, RI ˆ 0.850. 10) present; 11) absent; W & M: Fig. 136. CI ˆ 0.250, RI ˆ 0.824. 47. Spine on labial palp: 10) present; 11) absent; May Secondary sexual characters 0females) & Corbet, 2001: Figs 1±5. CI ˆ 0.333, RI ˆ 0.600.

30. Distinct pits on middle lobe of prothorax: 10) present; 11) absent; Fig. 7; W & M: Fig. 130. CI ˆ 0.250, RI ˆ 0.786. Habitat 31. Posterior margin of pronotum with median tubercle: 10) well developed; 11) vestigial; 12) absent; W & M: 48. Primarily inhabiting: 10) ; 11) streams. CI ˆ 0.250, Fig. 132. CI ˆ 0.250, RI ˆ 0.625. RI ˆ 0.500. 32. Posterior margin of pronotum; 10) entire; 11) notched or undulating; Fig. 8. CI ˆ 0.286, RI ˆ 0.808. Appendix 2. Diagnoses of adults of genus-group taxa 33. Posteromedial tubercle of mesostigmal plates: introduced or treated in detail here. 10) present; 11) absent; W & M: Fig. 132. CI ˆ 0.333, RI ˆ 0.692. The following characters apply to all the taxa described below 34. Mesostigmal plates: 10) very broad and posteriorly and distinguish them from many other genera of Coenagrio- convex;11)notso;W&M:Figs151and152. nidae: frons rounded in profile; postocular spots present, at CI ˆ 1.000, RI ˆ 1.000. least in males; thoracic colour pattern with distinct dark 35. Anteromedial trough and flange on mesostigmal plates: humeral stripe overlying the mesopleural suture, occasionally 10) present; 11) absent; Fig. 8. CI ˆ 0.333, RI ˆ 0.600. longitudinally divided or reduced to hairline within meso- 36. Ovipositor valves: 10) short, not extending caudad as pleural suture, especially in females; venter of metathorax far as end of paraprocts; 11) of medium length, without distinct, moundlike prominence; tibial spurs at most extending beyond paraprocts but not to end of cerci; only slightly longer than intervening spaces; inferior tooth of 12) long, extending beyond ends of cerci. CI ˆ 0.750, each tarsal claw normally developed; anal vein diverging from RI ˆ 0.933. posterior wing margin proximal to Ac; pterostigmata of front 37. Vulvar spine: 10) present; 11) absent; W & M: Figs 7 and and hind wings usually similar in colour and shape in both 50. CI ˆ 1.000, RI ˆ 0.000. sexes; quadrangle with posterodistal corner moderately acute; dorsum of abdominal segment 10 of males without a marked, bifurcated prominence 1sometimes with low protuberance or Miscellaneous adult structures small, rounded distal scale on each side of midline, or both) and without tori 1Westfall & May, 1996); dorsum of abdom- 38. Vein Ac: 10) distinctly basal to midway between inal segment 10 of females not divided medially; male epi- antenodal crossveins; 11) at or distal to midway proct small, not visible in dorsal view; female with vulvar between antenodal crossveins. CI ˆ 0.600, RI ˆ 0.826. spine; penis without elongate distal flagellae. 39. Stature of males: 10) slender, ratio of width to length of Characters that discriminate these genus-group taxa, and abdominal segment 3 < 0.12; 11) intermediate, this ratio the proposed composition of each taxon, are detailed below. between 0.12 and 0.14; 12) robust, this ratio > 0.14; Parenthetical question marks after species names indicate CI ˆ 0.182, RI ˆ 0.609. that I question whether these species actually are within a monophyletic group with the type species or I am uncertain of the identity of the species. Larval structures

40. Second and third antennal segments: 10) relatively long; Africallagma Kennedy 11920) 11) short; W & M: Fig. 134. CI ˆ 0.750, RI ˆ 0.933. 41. Distinct dark pattern on eyes: 10) present; 11) absent; Medium sized to very slender and elongate, male abdom- W & M: Fig. 134. CI ˆ 0.333, RI ˆ 0.889. inal segments 3±6 largely dark dorsally, at least along and for

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 404 Michael L. May a short distance on either side of the midline. Pale thoracic Azuragrion, gen.n. colour usually blue but sometimes very pale tan, almost whitish, black humeral stripe never divided longitudinally, Very small to large 1A. granti), male abdominal segments sometimes reduced to hairline in males and females. Ptero- 3±6 largely dark dorsally except largely blue in somalicum. stigmata pale grey to nearly black, vein Ac at or distal to Pale thoracic colour mostly blue in males, pale bluish green midway between antenodal crossveins. Abdominal segment to brown in females, black humeral stripe never divided 10 of male without posterodorsal protuberance. Male para- longitudinally, sometimes reduced to hairline in female. Ptero- procts distinctly bifid in lateral view except in A. cuneistigma, stigmata grey to brown, vein Ac usually basal to midway with inferior arm much broader and usually shorter than between antenodal crossveins, sometimes at or distal to this superior; cerci variable but lacking a distinct inferior arm or point in A. somalicum and A. granti, especially in forewing. middle lobe, generally curved ventromedially in distal third, Abdominal segment 10 of male without posterodorsal pro- or with recurved process at one-third to two-thirds length tuberance. Male paraprocts not distinctly bifid but with a directed ventromedially, sometimes with scalariformlike cut- low ventral protrusion as well as an elongate dorsal process icle 1presumably not homologous with that of Enallagma visible in lateral view; cerci with superior sclerotized arm with because of its very different position on the cercus and acute terminal or subterminal medial hook, middle lobe of because it appears to be normally sclerotized) on mediodistal about same length consisting of pale, flexible cuticle, inferior surface. Penis with large rounded or recurved earlike sclerotized arm from as long as to about twice as long as distolateral lobes 1narrow and digitate in A. rubristigma), dis- superior arm, scalariform cuticle absent. Penis with small to tinct lateral lobes present, spines or projections not present large quadrate or triangular earlike distolateral lobes, distinct just proximal to distal flexure. Male mesostigmal plates with lateral lobes usually absent 1present in A. nigridorsum), spines prominent posteromedial lappets; female plates usually with or projections not present just proximal to distal flexure. wide anteromedial trough and flange 1absent in A. sapphir- Male mesostigmal plates with very low, vestigial postero- inum, condition in A. rubristigma not known), posterior mar- medial lappets; female plates without wide anteromedial trough gin of pronotum sinuate, with distinct median prominence, andflange1sometimeswithdeeptransversetroughandele- or with median notch, middle lobe of pronotum never with vated anterior margin), posterior margin of pronotum with well defined pits. Ovipositor valves of medium length or long, distinct median prominence, middle lobe of pronotum never sometimes strikingly so. Figs 6, 8. with well defined pits. Ovipositor valves of short or medium Type species: Agrion glaucum Burmeister, 1839. Other length. Figs 7, 8. species: Enallagma cuneistigma Pinhey, 1969 1?); Ischnura Type species: Enallagma nigridorsum Selys, 1876. Other elongata Martin, 1907; Enallagma pseudelongatum Long- species: Enallagma buccholzi Pinhey 11971) 1see Pinhey, field 11936); Enallagma rubristigma Schmidt, 1951 1?); Enal- 1971), Ischnura granti McLachlan, 1903; Ischnura kauderni lagma sapphirinum Pinhey, 1950 1?); Enallagma sinuatum SjoÈ stedt, 1917; Enallagma somalicum Longfield 11931); Ris, 1921; Enallagma subtile Ris, 1921; Enallagma vaginale Enallagma vasomereni Pinhey, 1955. SjoÈ stedt, 1917 1includes E. longfieldae Fraser, 1947) 1?). Etymology: Named for the azure blue colour of males of most species.

Amphiallagma Kennedy 11920) Chromatallagma, subgen.n. of Enallagma Very small, male abdominal segments 3±6 largely dark dorsally. Pale thoracic colour mostly blue, black humeral Medium sized to fairly large 1E. daeckii), males usually stripe never divided longitudinally, sometimes reduced to slender, male abdominal segments 3±6 largely dark dorsally. hairline in 1immature?) female. Pterostigmata tan to black, Pale thoracic colour blue, violet, yellow, orange or red in vein Ac at or distal to midway between antenodal cross- males, pale blue, greenish, tan or dull yellow to orange in veins. Abdominal segment 10 of male without posterodor- females, black humeral stripe sometimes divided longitu- sal protuberance. Male paraprocts not bifid in lateral view; dinally, occasionally reduced to hairline in males and cerci smoothly rounded distally, somewhat excavated dorso- females. Pterostigmata pale tan to nearly black, vein Ac medially, with very small conical tubercle along dorsal usually basal to midway between antenodal crossveins. ridge at about one-third length, cuticle probably somewhat Abdominal segment 10 of male without posterodorsal pro- flexible dorsally on distal surface but scalariform cuticle tuberance but usually with rounded distal scale on each side absent, middle lobe absent, each cercus with dorsomedial of midline. Male paraprocts not bifid in lateral view; cerci projection at extreme base. Penis with small, quadrate ear- with a superior arm ending in a glabrous hook whose sharp like distolateral lobes, distinct lateral lobes absent, spines or edge is somewhat elongated, distinct inferior arm bearing projections not present just proximal to distal flexure. Male sclariform cuticle 1absent in E. antennatum, reduced in mesostigmal plates without posteromedial lappets; female E. basidens), this arm sometimes fused with superior arm. plates without wide anteromedial trough and flange, poster- Penis with quadrate distolateral lobes produced at corners, ior margin of pronotum undulate, middle lobe of pronotum distinct lateral lobes present, spines or projections not pre- never with well defined pits. Ovipositor valves short. Fig. 7. sent just proximal to distal flexure. Male mesostigmal plates Type species: Enallagma parvum Selys, 1876. without posteromedial lappets; female plates without antero-

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 The damselfly genus Enallagma 405 medial trough and flange, posterior margin of pronotum Enallagma opitimolocus Miller & Ivie, 1996 1?); Agrion usually smoothly rounded but sometimes with distinct praevarum Hagen, 1861; Davis, 1913; median tubercle, middle lobe of pronotum often with well Enallagma risi Schmidt, 1961; Enallagma rua Donnelly, 1968; defined pits. Ovipositor valves usually short. Fig. 4. Selys, 1876; Type species: Agrion signatum Hagen, 1861. Other species: Gloyd, 1943. Except for E. eiseni, the species in this group Agrion antennatum Say, 1839; Enallagma basidens Calvert, whose names are followed by a question mark are of uncer- 1902; Agrion coecum Hagen, 1861 1?); Enallagma concisum tain identity or validity. They are retained provisionally in Williamson, 1922; Telagrion daeckii Calvert, 1903; Enal- Enallagma strictly for nomenclatorial convenience, not on lagma divagans Selys, 1876; Root, 1924; the basis of evidence about their relationships. Agrion exsulans Hagen, 1861; Calvert, 1907 1?); Root, 1923; Morse, 1895; Agrion pollutum Hagen, 1861; Pinheyagrion, gen.n. Enallagma sulcatum Williamson, 1922; Enallagma traviatum Selys, 1876; Enallagma truncatum Gundlach, 1888; Enal- Very small, male abdominal segments 3±6 largely dark lagma vesperum Calvert, 1919; Byers, 1927. dorsally. Pale thoracic colour mostly blue or greenish in Etymology: Named for the bright and varied colours of males, sometimes orange brown in females, black humeral males within this group. stripe never divided longitudinally, sometimes reduced to hairline in 1immature?) female. Pterostigmata of male bright pink, those of female brown, vein Ac at or distal to midway Enallagma Selys, sensu stricto between antenodal crossveins. Abdominal segment 10 of male without posterodorsal protuberance. Male paraprocts dis- Medium sized, males usually robust, male abdominal seg- tinctly bifid in lateral view, the inferior branch much longer ments 3±6 usually largely blue dorsally. Pale thoracic colour than superior; cerci with 2 branches fused for about half their blue in males, pale blue, greenish or tan in females, black length, no middle lobe, the superior branch ending in an humeral stripe never divided longitudinally, not reduced to acute, spinelike, but not hooked, tip directed posterodorsally, hairline. Pterostigmata pale tan to nearly black, vein Ac the inferior asymmetrically rounded distally, somewhat almost always basal to midway between antenodal cross- dorsoventrally flattened and inclined slightly upward laterally, veins. Abdominal segment 10 of male without posterodorsal the ventral surface covered with padlike cuticle 1probably protuberance but often with rounded distal scale on each side somewhat flexible), scalariform cuticle absent. Penis with of midline. Male paraprocts not bifid in lateral view; cerci small, quadrate earlike distolateral lobes, distinct lateral with a superior arm ending in 1or sometimes reduced to) a lobes absent, with prominent bifurcated spine on each side glabrous hook whose sharp edge is somewhat elongated, just proximal to distal flexure. Male mesostigmal plates with distinct inferior arm bearing scalariform cuticle, this arm very prominent posteromedial lappets; female plates without sometimes partly fused with superior arm when latter is wide anteromedial trough and flange, posterior margin of reduced, so scalariform cuticle overlaps the latter. Penis pronotum undulate, middle lobe of pronotum never with usually with crescentic distolateral lobes 1quadrate in well defined pits. Ovipositor valves long. Fig. 5. E. durum and E. eiseni), distinct lateral lobes present, spines or Type species: Enallagma angolicum Pinhey, 1966. projections not present just proximal to distal flexure. Male Etymology: Named for the late Dr Elliot Pinhey in mesostigmal plates without posteromedial lappets; female recognition of his many contributions to the taxonomy of plates variable but usually without anteromedial trough African Odonata. and flange, posterior margin of pronotum usually smoothly rounded 1sinuate in E. eiseni) and without median tubercle, middle lobe of pronotum sometimes with well defined pits. Proischnura Kennedy 11920) Ovipositor valves short. Figs 4, 8. Type species: Agrion cyathigerum Charpentier, 1840. Other Small to medium sized, male abdominal segments 3±6 species: Enallagma ambiguum Nava s, 1936 1?); Enallagma largely dark dorsally. Pale thoracic colour blue or blue- anna Williamson, 1900, Agrion aspersum Hagen, 1861; green to violet in males, bluish green to tan in females, Enallagma belyshevi Haritonov, 1975; Aenallagma boreale black humeral stripe never divided longitudinally. Pterostig- Selys, 1875; Enallagma carunculatum Morse, 1895; Enallagma mata of both sexes grey, brown or black, usually paler in circulatum Selys, 1883; Agrion civile Hagen, 1861; Enallagma female, vein Ac at or distal to midway between antenodal clausum Morse, 1895; Enallagma davisi Westfall, 1943; crossveins. Abdominal segment 10 of male with low, bifid Agrion deserti Selys, 1871; Agrion doubledayi Selys, 1850, posterodorsal protuberance. Male paraprocts distinctly or Agrion durum Hagen, 1861; Agrion ebrium Hagen, 1861; slightly 1P. polychromaticum) bifid in lateral view, the superior Enallagma eiseni Calvert, 1895 1?); branch longer; cerci with wide superior branch, broadly Kellicott, 1895; Agrion hageni Walsh, 1863; Enallagma insula rounded distally with low nipplelike protrusion at tip, small Fraser, 1920 1?); Morse, 1895; Enallagma glabrous thumblike process on dorsomedial edge, otherwise maldivense Laidlaw 11902) 1?); Morse, with dorsomedial area usually bearing numerous stiff 1895; Enallagma nigrolineata Belyshev & Haritonov, 1975; setae except for small distal area of scalariformlike cuticle

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 406 Michael L. May

1presumably not homologueous with that of Enallagma posterior margin of pronotum undulate, medially tuberculate because of its very different position on the cercus), narrow or with prominent median notch, middle lobe of pronotum digitate inferior branch projecting ventromedially from base of never with well defined pits. Ovipositor valves of medium superior branch. Penis with elongate crescentic distolateral length. Fig. 5. lobes just beyond a distinct constriction, distinct lateral lobes Type species: Enallagma subfurcatum Selys, 1876. Other present or absent, with prominent triangular or tonguelike species: Enallagma polychromaticum Barnard 11937); projection on each side just proximal to distal flexure. Male Enallagma rotundipenne Ris, 1921. mesostigmal plates with prominent posteromedial lappets; female plates without wide anteromedial trough and flange,

# 2002 The Royal Entomological Society, Systematic Entomology, 27, 387±408 # 02TeRylEtmlgclSociety, Entomological Royal The 2002

Appendix 3. Character matrix. Numbers for characters and character states correspond to those in Appendix 1; species numbers correspond to those of Figs 1 and 2.

Characters Species 1 2 3456 7 8 9101112131415161718192021222324252627282930313233343536373839404142434445464748

Chromagrion conditum 002100 2 1 ? 221101 2 2 20121211821202111101 1 2 0 101100 1 1 010100 2 1 1 221101 2 2 20121011521212111111 1 2 0 110110 1 0 Ischnura verticalis 202100 2 1 1 2211010,20,2221210106212020111100,120 010100 1 0

ytmtcEntomology Systematic Pinheyagrion angolicum 002120 2 1 1 221101 2 0 20121010421002111120 1 ? ? ?????? ? ? quadratum 002100 2 1 1 221101 2 0 02121211721202111100 1 0 0 000110 0 0 Argentagrion cheliferum 002120 0 1 1 221121 2 0 00111011201102111?00 0 ? ? ?????? ? ? Cyanallagma interruptum 212120 0 1 1 221101 2 0 02121011221101111100 1 ? ? ?????? ? ? Proischnura subfurcatum 002100 2 1,01 211100 2 2 20111010401002111110 1 1 ? ?????? ? ? P. polychromaticum 302000 2 1 1 001100 2 2 20110110420000111110 1 2 ? ?????? ? ? P. rotundipenne 002100 2 0 1 201100 1 2 20111010???001011??0 1 ? ? ?????? ? ? Africallagma 1?) cuneistigma 0021000,21 1 221100 2 0 0202121????002111000 1 ? ? ?????? ? ?

, A. 1?) rubristigma 002110 1 1 1 221100 2 1 021110113000???????010? ?????? ? ? 27 A. 1?) vaginale 002122 0 0 1 201100 2 1 021210112200???????010? ?????? ? ? 387±408 , A. 1?) sapphirinum 000100 0 0 0 201100 2 2 12121111220001111110 0 2 ? ?????? ? ? Africallagma glaucum 002100 2 1,01 200100 2 2 20121011220001111010 1 1 ? ??????1? A. subtile 002100 0 1 1 000100 1 0 00121011220002111010 1 0 ? ?????? ? ? A. elongatum 002100 2 1,00 000100 1 0 00121011221002111020 1 0 ? ?????? ? ? A. pseudelongatum 012100 2 0 0 200100 2 0 00121011220002111020 1 0 ? ?????? ? ? A. sinuatum 402100 2 0 0 000100 2 0 001210112200???????010? ?????? ? ? Argentagrion ambiguum 002120 2 1 1 221120 2 0 00121011220102110110 0 ? ? ?????? ? ? Homeoura nepos 002120 2 1 1 221120 2 2 20111011200102111100 0 ? ? ?????? ? ? Amphiallagma parvum 002100 0 1 1 201100 2 2 20021211221202111100 0 0 ? ?????? ? ?

Azuragrion somalicum 000100 0 1 0 001100 2 1 0002021120010211?1000,10? ?????? ? ? genus damselfly The A. granti 002120 2 0 1 200100 2 2 00020111220202111100 1 0 ? ?????? ? ? A. nigridorsum 002110 2 1 1 201100 2 2 20020111220101111110 0 0 ? ??????1? A. kauderni 002110 2 1 1 221100 2 2 20020111221102111110 0 0 ? ?????? ? ? A. vansomereni 002100 2 1 1 201100 2 2 00020111221102111110 0 0 ? ?????? ? ? Enacantha caribbea 010100 2 1 1 200100 1 2 20001211021202011110 0 ? 0 111110 1 0 Enallagma eiseni 001100 2 0 1 101100 2 2 20001201020202111110 0 2 ? ?????? ?0 E. durum 010100 2 0 1 120100 2 2 00001211010202011100 0 20,1100110 0 0 nlam 407 Enallagma 408 ihe .May L. Michael

Appendix 3. Continued.

Characters

Species 123456 7 8 9101112131415161718192021222324252627282930313233343536373839404142434445464748

E. civile 010100 2 1 1 221100 2 2 20001211120202011100 0 2 0 1101100,10 E. minusculum 010100 2 1 1 221100 2 2 20001211120202011100 0 2 0 110111 ? 0 E. semicirculare 310100 2 1 1 201101 0 2 200012011202020111000,120 110110 0 0 E. carunculatum 010100 2 0 1 221101 2 2 20001211120202011100 0 2 0 110110 ? 0 E. anna 010100 2 0 0 221101 2 2 20001211120202011100 0 2 0 110110 0 1 E. clausum 010100 2 0 0 221101 0 2 20001211120212011100 0 2 0 110110 ? 0 E. cyathigerum 010100 2 0 0 2211010,2220001211120202011100 0 2 0 110110 0 0 E. aspersum 012100 2 1 1 221101 1 2 20001211120212011100 0 1 0 110110 0 0 E. geminatum 002000 2 1 1 221101 1 2 20001211120212011100 0 1 0 110110 0 0

# E. ebrium 010100 2 1 1 221101 2 2 20001211120212001100 0 2 0 110110 0 0 02TeRylEtmlgclSociety, Entomological Royal The 2002 E. hageni 010100 2 1 1 221101 2 2 20001211120212001100 0 2 0 110110 0 0 E. laterale 010000 2 1 1 221101 2 2 20001211120212011100 0 2 0 110111 0 0 E. recurvatum 010100 2 1 1 221101 1 2 20001211120212011100 0 2 0 110111 0 0 E. coecum 312000 2 1 1 001101 0 2 20001201120202010100 0 0 1 001111 0 1 E. novaehispaniae 312000 2 1 1 001101 0 2 20001201120202011100 0 0 1 001111 0 1 E. antennatum 202120 2 0 1 221110 2 2 20021211020200010100 1 2 1 001111 0 0 E. daeckii 002100 0 0 1 200001 2 0 00001211120202011100 1 00,10011110,10 E. pallidum 012100 0 1,01 000000 0 0 00001211120202010100 1 0 1 001111 0 0 E. traviatum 002100 2 1 0 000000 1 0 00001211120202010100 1 0 1 001111 0 0 E. divagans 002100 2 1 1 2011200,2000001211120200010110 1 0 1 001111 0 1 E. exsulans 002120 2 1 1 201120 2 0 00001211120200011100 1 0 1 001111 0 1 E. weewa 302120 2 1 1 2211200,20,200001211000200011100 1 0 1 001111 0 1 E. basidens 000100 2 0 1 200120 2 2 000012110212000101000,121 001001 0 0 E. sulcatum 102120 2 1 1 200100 2 1 00001211020211010100 0 0 1 0110010,10 ytmtcEntomology Systematic E. vesperum 102120 2 1 0 200100 2 1 00001211020211010100 0 0 1 0110010,10 E. pollutum 102121 2 1 1 200100 2 2 00001211000211010100 0 0 1 011001 0 0 E. signatum 102121 2 1 1 200100 2 2 01001211000211010100 0 0 1 0110010,10 E. dubium 102122 2 1 1 221100 2 2 210012110002110101000,101 011001 0 0 E. concisum 102122 2 1 1 000100 2 2 21001211000211010100 0 0 1 011001 0 0 E. pictum 102122 2 1 1 000100 2 2 21001211000211010100 0 1 1 011001 0 0 , 27 387±408 ,