Redescription of Mymarilla Westwood, new synonymies under Cremnomymar Ogloblin (Hymenoptera, Mymaridae) and discussion of unusual wings

Abstract The monotypic genus Mymarilla Westwood is known only from St. Helena, a remote island in the South Atlantic Ocean. The peculiar species M. wollastoni Westwood (Mymaridae) is redescribed and illustrated from non-type material. Mymarilla is compared with Cremnomymar Ogloblinspp. from the Juan Fernández Islands in the South Pacific Ocean. Stephanodes Enock is shown to be the most likely sister genus to Mymarilla. Nesopolynema Ogloblin, syn. n., Oncomymar Ogloblin, syn. n., Scolopsopteron Ogloblin, syn. n., are placed in synonymy under Cremnomymar and their species transferred as Cremnomymar caudatum (Ogloblin 1952), comb. n., C. dipteron (Ogloblin 1957), comb. n., and C. kuscheli (Ogloblin 1952), comb. n. Wing shape and wing reductions in Mymaridae are discussed in relation to biogeography, particularly with respect island faunas and to four genera, Cremnomymar, Mymarilla, Parapolynema Fidalgo, and Richteria Girault, some or all of whose species have more or less convex fore wings.


Introduction
The small, remote South Atlantic island of St. Helena contains a high proportion of endemic species but among the Hymenoptera only the Formicidae have been systematically studied (Wetterer et al. 2007). Among the insects, one of the most remarkable is Mymarilla wollastoni Westwood (Hymenoptera: Mymaridae). Although the species was described briefly from both sexes (Westwood 1879), it is instantly recognizable from the original illustrations that show the peculiar, strongly convex fore wings. Annecke and Doutt (1961) redescribed the species in greater detail, designated a lectotype, and cleared up the confusion by previous authors over what species should be included in Mymarilla Westwood. Subba Rao (1976) illustrated the species with a habitus drawing. Mymarilla is redescribed from material other than the original type series and is compared with new material of Cremnomymar Ogloblin and its synonyms (proposed below) from the Juan Fernández Islands, about 600 km W. of Chile in the South Pacific Ocean. Wing shape, and wing reduction and/or convexity is discussed in relation to geographical distribution of Mymaridae.

Methods
Morphological terms follow Gibson (1997) and Huber (2012). Measurements are in micrometres (µm). Photographs of critical point dried and card mounted specimens were taken with a ProgRes C14 plus digital camera attached to a Nikon Eclipse E800 compound microscope, and the resulting layers combined electronically using Auto-Montage® or Zyrene Stacker® and retouched as needed with Adobe® Photoshop CS6®. Specimens of Cremnomymar were gold coated for scanning electron micrography, using the techniques described in Bolte (1996).
The confusion in the use of the name Mymarilla and which species should be placed in the genus continued for over 80 years. Heqvist (1960) was the first to treat M. wollastoni as the genotype of Mymarilla, even as he incorrectly placed Mymar species under it, as did previous authors. As Annecke and Doutt (1961) clarified, wherever previous authors use Mymarilla it was clearly in the sense of Mymar, because the species names mentioned are typical Mymar species.
Diagnosis. The combination of smooth, shiny black body, extremely short mesocutum compared to much longer pronotum, and extraordinary convex and densely setose fore wing distinguish the genus and species from any other Mymaridae.
Mymarilla belongs clearly to the Polynema Haliday group of genera within Mymarini sensu Annecke and Doutt (1961). They suggested that Mymarilla was most similar to Oncomymar Ogloblin from the Juan Fernández Islands. Superficially, the most similar genus is Cremnomymar Ogloblin (including Oncomymar Ogloblin and Scolopsopteron Ogloblin, see below), some of whose species also have a convex fore wing and reduced mesoscutum. The two genera are not closely related; their resemblance is due to adaptations to life on remote, presumably wind-swept, oceanic islands. I propose instead that Mymarilla is derived from Stephanodes Enock, likely the most closely related genus. Four features, shared with Stephanodes, suggest this: first, the extremely smooth body without trace of microscupture on the mesosoma (Figs 1,3,4,6,9,10); second, the slightly advanced mesothoracic spiracle about midway between the anterior apex of a notaulus and posterolateral angle of the mesoscutum (Figs 3, 4); third, the presence of a metapleural pit (Figs 9, 10); fourth, the fore wings that are held more or less horizontally. In dead specimens of Stephanodes, the fore wings are often horizontal, crossed scissor-like and covering the body, unlike other, related genera in the Polynema-group where the wings (in dead specimens) are almost always vertical, directed away from the body. The strong convexity of the fore wings of Mymarilla would appear to prevent them from being crossed scissor-like over the body. Yet they are presumably capable of enveloping the metasoma, as pointed out by Westwood who noted ". . . when shut [the fore wings] form a semiglobular dome over the abdomen" when the wasp is at rest. The densely hairy wing membrane with dark base around each microtrichia would allow for maximum heat absorption and retention. Remarks. Lectotype female in OUMNH (not examined), designated by Annecke and Doutt (1961), with catalogue number HYME0029 and labeled: "Holotype", "det. D.P. Annecke 25.vii.1960". Description. Because only one species of Mymarilla is known, the generic and specific features are both described in the species description. Female. Body length 2125 (n=1). Body entirely smooth and shiny except for a few small punctures on pronotum. Colour. Head black except mandibles, mesosoma, metasoma medially, and clava; flagellum, and metasoma basally and apically dark brown; scape, pedicel, mandibles, legs, petiole, and ovipositor brownish yellow; all body setae almost white; wings brown behind venation, membrane beyond venation translucent but membrane around each microtrichia back, so wings generally appearing dark. Head. (Figs 3-6) 1.46× as wide as long, and 1.17× as wide as high (406/227/347) (n=1), in lateral view with anterior surface slightly depressed between toruli then, at level of ventral margin of eye, receding to mouth (Fig. 5). Face measured from eye to eye below toruli about as long as wide, flat above toruli but forming a shallow circular depression medially below toruli; subantennal groove and pits between toruli absent. Torulus about 1.5× its diameter from thin transverse trabecula and about mid height of eye. Preorbital groove against eye almost to ventral margin of eye (Fig. 5), then extending almost straight to anterolateral angle of mouth margin. Eye (Fig. 5) in lateral view 0.75× as long as high, not extending to back of head dorsally and separated by more than its own length from back of head ventrally. Malar sulcus absent and malar space almost 0.6× eye height. Gena narrow dorsally, very wide ventrally. Vertex in lateral view almost flat, sloping anteriorly, forming an obtuse angle to face (separated by transverse trabecula), smoothly merging posteriorly with occiput. Mid ocellus flat in shallow depression, not projecting above surface of vertex and 2× diameter of small lateral ocellus. Ocelli in high triangle, with POL (168), 1.9× LOL, and LOL (89) 3.0× OOL (30). Mandibles normal, overlapping medially when closed, with 3 teeth. Antenna. Scape about 3.9× as long as wide, with both inner and outer surfaces apparently smooth, and with radicle short but distinct; pedicel slightly shorter than fl 1 , about 0.41× scape length; funicle 6-segmented; clava unsegmented, slightly longer than scape. Number of mps of funicle segments and clava uncertain (not clearly visible on card mounted specimen). Measurements lateral view tightly pressed to pronotum laterally and anteriorly, and fused to each other at neck, and, in dorsal view (Fig. 3), slightly extending lateral to pronotum. Neck long (90), clearly separating head from pronotum. Prosternum triangular, strongly appressed laterally to propleura, without median longitudinal line, anterior apex not visible but perhaps closed anteriorly. Mesonotal spiracle ( Fig. 6) small, at end of short tube, midway between posterolateral angle of mesoscutum and anterior apex of notaulus. Mesoscutum (Fig. 3) smooth and shiny, in lateral view slightly convex, in dorsal view short (99), about 0.4× as long as pronotum and 0.4× as long as wide, with slightly diverging, almost straight notauli, each ending anteriorly in a distinct pit. Axillae not advanced (Figs 3,4,9,10). Mesoscutellum (Fig. 3) almost as long as pronotum (267: 246), slightly overlapping metanotum, without trace of frenal line (frenum therefore not distinguishable). Prepectus (Figs 6,9,10) triangular, about 3× as long as dorsal width. Mesopleuron tightly appressed to prepectus, convex, not divided by suture into mesepisternum and mesipimeron. Metanotum small, triangular, separated from propodeum by wide groove. Metapleuron with a large metapleural pit at junction with mesoepisternum. Propodeum (Figs 3, 4, 7, 10) evenly convex, without carinae, with a large pit at anterior margin just anterior to spiracle, with a slightly upturned nucha covering anterior apex of petiole, and with propodeal seta almost at posterior margin. Spiracle small, round, separated by several diameters from anterior margin of propodeum. Wings. Fore wing (Figs 1,7,8) deeply convex, with the anterior and posterior margins strongly curving downward, the wing height in lateral view about 0.7× wing width in dorsal view. Wing behind submarginal vein very narrow, with strongly convex hind margin, abruptly widening beginning at parastigma, generally oval in dorsal view. Entire surface to wing base covered in long microtrichia, those behind venation appressed and those beyond venation semi-erect to erect. Fore wing length 1792 (n=1), width 640, length/width 2.8, venation length 287, about 0.18× forewing length. Submarginal vein black basally, brown apically, and much wider basally than apically; parastigma + stigma black, oval about 1.7× as wide as base of submarginal vein. Costal cell extremely narrow. Hind wing ( Fig. 1) flat to slightly convex, with long marginal setae and erect microtrichia similar to those on forewing, and wing membrane extending almost to base of wing but very narrow behind venation. Hind wing length ca. 1535, width 77, venation black, length about 0.3× wing length. Legs. Long and slender (Figs 1, 7). Metasoma. Petiole length 180, slightly longer than metacoxa, about 6× as long as wide. Gaster (Fig. 1) smooth and shiny, narrowly oval in cross section (as seen in posterior view), wider dorsally, almost knife-like ventrally at ovipositor. Gt 1 about 0.57× gaster length (870) and almost completely covering gs 1 so petiole apparently attached to tergum, gt 2 -gt 5 progressively shorter, gt 6 and syntergum (gt 7 + gt 8 ) each about as long as gt 2 . Spiracle present on gt 6 , small. Ovipositor length ca. 770, slightly down turned apically, as long as gaster but not exserted beyond gaster, about 1.2× metatibia length (640).
Male. Colour as in female but body dark brown (possibly due to fading), scape brownish yellow, pedicel light brown, flagellum dark brown. Body length 1843-1894 (n=2). Fore wing length 1664 (hind wing not measurable on pinned specimens), with the edges almost meeting ventrally in one specimen, giving the appearance of a hirsute cigar when seen end on in posterior view (Fig. 7). Antenna. Scape about 3.8× as long as wide, with both inner and outer surfaces apparently smooth. Measurements Habitat. Westwood (1879) stated that the specimens were swept from low herbage. Annecke and Doutt (1961) suggested that the wings may be used for floating on air currents. Whether individuals are capable of this, let alone normal flight, is uncertain. The greatly reduced mesoscutum suggests that the flight muscles are so reduced they would be incapable of powered flight. The collection of one female in a pitfall trap suggests that M. wollastoni lives near the ground.
Remarks. Ogloblin (1952Ogloblin ( , 1957 described seven species of Mymaride based on eight specimens from the Juan Fernández Islands, all but one from Masatierra Island (Robinson Crusoe Island). Three were described from females only and four from males only. Scolopsopteron dipteron Ogloblin (Ogloblin 1952) has a fore wing of normal length but unusually narrow and a rudimentary hind wing. Oncomymar kuscheli Ogloblin (Ogloblin 1957) has a fore wing of normal width and length but strongly convex (spoon-shaped) and a rudimentary hind wing. Nesopolynema has a normal fore wing (hind wings missing in Ogloblin's specimen). It is improbable that the Juan Fernández Islands would have four endemic genera of Mymaridae, as treated by Ogloblin (1952Ogloblin ( , 1957. Above the species category similarities must be used to define collective groups, not differences. If differences are used, then how different must something be to be placed in a different genus? "Different enough" is not an acceptable answer. If it were, the placement of species in genera of finer and finer difference would be the norm, with the result that the genus category would become almost synonymous with the species category, and each genus would contain only one or two species. Unfortunately, Ogloblin sometimes defined genera based on obvious but superficial differences that resulted in oversplitting. I am confident that all seven of Ogloblin's species from the Juan Fernández Islands represent at most different species within a single genus. Indeed some of them have likely been described twice, once from females and once from males. I examined forty-three specimens (not the types) of Cremnomymar (including the synonyms proposed above), all from Masatierra, the largest of the Juan Fernández Islands. In some of the species the fore wing is fully developed and flat (Fig. 11) but in others it is more or less reduced and convex (Figs 12-14). Depending on the extent of wing reduction the mesothorax is also reduced but the placoid sensilla on the scutellum are always widely separated. The pronotal structure varies from entire, sometimes with indication of a mediolongitutinal line, to being apparently completely divided medially by a complete longitudinal carina. The propodeum in Figure 11. Cremnomymar sp., macropterous female, dorsal (fore wing flat). Scale line = 1000 µm.
short-winged specimens changes in ways that I consider to be at most of species-level significance, from strongly carinate and with a large, sublateral tooth bearing the propodeal seta, as in fully winged Cremnomymar (Figs 11, 12, 32) to almost smooth (Fig. 14) but still with the propodeal seta on a bump or tooth. Therefore, I propose the above generic synonymies and transfer the included species to Cremnomymar as C. caudatum (Ogloblin 1952), comb. n., C. dipteron (Ogloblin 1957), comb. n., and C. kuscheli (Ogloblin 1952), comb. n.   of Robinson Crusoe and 5.8 million years old (Haberle 2009). Over half of the native plants species (126 of 209) are endemic. The climate is subtropical and its climate is affected by the cold Humboldt Current.
Despite being two to three times the age of Santa Clara, the oldest of the Juan Fernández Islands, the remoteness of St. Helena from the nearest continent is likely the main reason why it has only one endemic species of Mymaridae. In contrast, the Juan Fernández Islands have several endemic species, diverse enough to have been placed in separate genera by Ogloblin, who was swayed by the most notable structural differ- One specimen of Parapolynema Fidalgo was seen, collected in Bolivia, La Paz, Sorata environs, 3200m, 20.iv.1997, L. Masner, sweeping open scrub (1 female, CNC). Fidalgo (1982Fidalgo ( , 1991 stated that Parapolymema is most similar to Cremnomymar, based on fore wing structure (Fig. 15) and propodeal seta set on a similar (as in some Cremnomymar), distinct protuberance. But the fore wing also is remarkably similar to that of Richteria Girault (Figs 16-18). The endemic Mymaridae on the Juan Fernández Islands are probably derived from specimens of Parapolynema that arrived there from southern South America but there seems to be a link to the New Zealand/Australian fauna and all three genera may be Gondwanan remnants.

Wing modification in Mymaridae
The wings of Mymaridae vary considerably in shape, perhaps more than in most other families of Chalcidoidea. In a normal, fully developed (macropterous) wing the surface is flat, with the length several times greater than width and with a more or less rounded apex. Variations in shape are due partly to changes in length and width and partly to changes in outline, particularly of the posterior margin and wing apex. Depending on the genus and species a full length fore wing may become narrower so that length becomes even greater relative to width. The extremes are 2.5× as long as wide in Paranaphoidea Girault (Lin et al. 2007, fig. 206) and 30× as long as wide in Cleruchus biciliatus (Ferrière) (Ferrière 1952). Hind wing shape varies only slightly because it is generally linear to begin with. Variation in a full length hind wing is therefore mainly in width, from extremely narrow, thread-like, e.g., Mymar (Huber et al. 2009, fig. 55) to wide, e.g., Par- anaphoidea (Lin et al. 2007, fig. 206); rarely the hind wing may be curved, with the anterior and posterior margins distinctly convex and concave, respectively. Despite all this variation the wings remain two dimensional. Most of the range of wing shapes and sizes is illustrated in generic treatments of Mymaridae by Subba Rao (1983) for the Oriental region, Noyes and Valentine (1989) for New Zealand, Yoshimoto (1990) for the Western Hemisphere, Huber (1997) for North America, Lin et al. (2007) for Australia   Wing reduction (brachyptery, microptery) or complete loss (aptery) usually occurs where there a strong selection pressure against having fully developed wings. Reduction in length may be slight shorter than normal, to extremely short, at least in one sex (usually the female) of at least one species of a genus. Species with wings reduced or absent occur on all continents, mostly in species that search for host eggs in confining places where wings would be a hindrance, e.g., in soil, leaf litter, or in tubules of bracket fungi (some Cleruchus spp.) or in windy places, e.g., small, remote oceanic islands and high elevations on mountains. All three species (100%) of Mymaridae found on small islands (Campbell I., Auckland I., South Georgia) at more than 45° south are wingless or micropterous. Each species is placed in its own genus and each genus is evidently related to a genus on the nearest mainland or is a wingless species with winged congeners in New Zealand, and South America, respectively. For example, on South Georgia (at or near sea level) Notomymar is completely wingless, whereas at 4100 m in Ecuador (Yoshimoto 1990) the species there is micropterous, and at low altitude in Chile, the species (undescribed?) is/are fully winged.
On remote oceanic islands between 30°-45° S Mymaridae are reported so far only from the Juan Fernández Islands (Cremnomymar, 7 described spp., but some probably synonyms), Norfolk Island (Cybomymar Noyes & Valentine [Lin et al. 1997], 1 sp.), and Lord Howe Island (Anagroidea sp., new record). Wingless or short-winged species represent at least 20% of the fauna, excluding genera/species likely introduced over the past few hundred years by humans. In contrast, Mymaridae from continents south of 30° S contain a low percentage of short winged or wingless species. The exception is New Zealand (a large continental island) with 17 of 42 genera that contain at least one flightless species (Noyes and Valentine 1989: 8).
Between the tropics of Cancer and Capricorn no wingless or short winged species have yet been recorded on the relatively well studied but still poorly known (for Mymaridae) Pacific oceanic islands: Micronesia (Doutt 1955), Fiji (Huber 2009), Galapagos Islands (Huber, unpublished), Hawaiian Islands (Beardsley and Huber 2000) and, in the Atlantic Ocean, the Cape Verde Islands (Viggiani and Jesu 1995). Northern Hemisphere islands north of the tropic of Cancer are almost all in the North Atlantic Ocean. They are either too close to continents to develop endemic faunas that include short-winged or wingless species or they are too far north and therefore too cold for most insects to survive. They have generally been poorly explored, except Madeira (Graham 1979(Graham , 1983(Graham , 1988. Curvature of the fore wing so it becomes convex or dome-shaped, i.e., it is no longer flat but clearly three-dimensional, occurs in at least one species of four genera: Cremnomymar Ogloblin (Figs 12, 13), Mymarilla (Figs 1, 7), Parapolynema Fidalgo (Fig. 15), so far known from Argentina and Bolivia, and Richteria (Figs 16-18), so far known from Australia and New Zealand. On wind-swept islands convex wings may perhaps have survival value by enveloping the posterior part of the mesosoma and the metasoma, either for greater heat absorption and retention or to reduce the chance of being blown away. But species of Parapolynema and Richteria do not necessarily occur in harsh environments so there evidently are other reasons for having a more or less convex fore wing. Interestingly, the fore wings of both genera have a double dome, in contrast to the single dome of Mymarilla and some Cremnomymar.
It would be interesting to find out if Mymaridae occur on other remote Southern Hemisphere islands, particularly those south of 30°. Collecting in a diversity of microhabitats using yellow pan traps or sifting of soil and whatever vegetation is present is required because any Mymaridae present are almost certain to be wingless or short winged.