Facies can fool taxonomists as well as predators. Neoxeniades molion (Godman)—one of the many, large neotropical skippers in the major mimicry group that includes all the cryptic species of the Astraptes fulgerator complex—is misclassified. It really relates to a species of Rhinthon that differs greatly from it in facies and size. In both sexes, the genitalia of Rhinthon molion, new combination, and R. osca (Plötz) are nearly identical, even down to a peculiar titillator on the left side of the aedeagus of the male. Males also share a secondary sex character along vein 2A of the forewing. DNA barcodes unite R. osca and R. molion in a tight, two-taxon cluster: their sequence divergence is about 3.5%. Caterpillars of the two species are alike but distinguishable, and, in Costa Rica, they have been found feeding on various native species of Marantaceae—seven of which are the same—and also on one and the same introduced species of Cannaceae (these are closely related plant families grouped in the order Zingiberales). Because Rhinthon is widespread and relatively speciose in Central and South America, it can no longer be considered a genus primarily of the Greater Antilles. DNA barcodes, which are useful for identifying known species and for indicating possible cryptic species, are useful in this study (in combination with other, more traditional, taxonomic characters) for pulling supposedly unrelated species together into the same genus.
Genitalia and barcodes are once again prime movers. This was true when they catalyzed the shift of the hesperiine skipper Telles arcalaus (Stoll) to Thracides and, as a consequence, synonymized the former genus with the latter (Bums et al. 2009). Telles was in the K Group of Evans (1955) whereas Thracides is in his O Group. What male and female genitalia and DNA barcodes say now is that Neoxeniades molion (Godman) is a species of Rhinthon. Both the general appearance of caterpillars and what they eat confirm this statement. Owing to the independence of these weighty characters, their mutual reinforcement is especially significant.
Data in this study stem from large-scale rearings of wild-caught caterpillars in Area de Conservación Guanacaste (ACG), in northwestern Costa Rica (Janzen & Hallwachs 2008).
Adult Facies and Mimicry (Figs. 1–14). Adult facies mislead: although both Rhinthon molion, new combination, and R. osca (Plötz) are basically brown with prominent hyaline spots in the central forewing, R. osca is a medium-sized skipper with ochreous alar overscaling dors ally and a brownish body ventrally (Figs. 1–4) whereas R. molion is a large skipper, with blue dorsal overscaling and a yellow ventral body (Figs. 5–8). These divergent species share the following forewing hyaline spots: a large one in Cu1–Cu2, an adjacent, smaller, double one in the cell, and tiny ones in R3–R4 and R4-R5. (In R. osca, the two parts of the double cell spot are slightly separate, resembling an = sign, in males; but they are more or less fused in females.) Rhinthon osca always has a tiny, distally displaced, hyaline spot in R5–M1, as well as a small spot in M3–Cu1, neither of which appears in R. molion. (One deviant female out of 95 reared specimens of R. osca even has tiny hyaline spots in M1–M2 and M2–M3.) Tiny, opaque, dorsal and ventral hindwing spots in R. osca are lacking in R. molion. On the other hand, only R. molion has a white costal margin on the ventral forewing and, especially, on the ventral hindwing.
It is this conspicuous, proximal, white costal margin of the ventral hindwing, plus the yellow ventral body color, the blue dorsal overscaling, the absence of the forewing spot in M3–Cu1, and the absence of all hindwing spots, that place R. molion (Figs. 5–8) within a large, widespread, neotropical mimicry complex (28 of whose member species appear, in dorsal view, in Janzen et al. 2009: fig. 4). Rhinthon molion, a hesperiine, is exceedingly similar to the eudamine Astraptes YESENN (Figs. 9–12), which is one of 11 recently discovered, and provisionally named, cryptic species in the Astraptes fulgerator species complex (Hebert et al. 2004; Janzen et al. 2009). Mimetic convergence of R. osca and A. YESENN even includes distal development of opaque white scaling on the ventral forewing in Cu2–2A (Figs. 6, 8, 10, 12).
Except for the degree of white scale development in Cu2–2A, which is greater in females than it is in males, neither of these two mimetic species is sexually dimorphic in color pattern (Figs. 5–12). All three species in Figs. 1–12 express the usual hesperiid sexual dimorphism in wing shape, wherein the wings of females are broader and more apically rounded than are those of males.
Males of R. osca and R. molion share a distinctive secondary sex character, a brand comprising specialized scales (presumably for disseminating pheromone[s]) near the posterior edge of the dorsal forewing, along vein 2A, about halfway between the outer margin and the base of the wing (Figs. 13, 14). This brand is paler and more noticeable in R. osca than it is in R. molion.
Male Genitalia (Figs. 15–18). The genitalia of the Rhinthon males (including even their paired cornuti) look remarkably similar. When individual variation is taken into account, the genitalia are nearly identical (number of dissections compared: 6 R. osca and 5 R. molion). However, there is a distinct and consistent difference in the anterior outline of the juxta in lateral view: rounded in R. osca (Fig. 16) but pointed in R. molion (Fig. 18).
These Rhinthon males have two peculiar genitalic features. The aedeagus is slightly bifurcate at its ventrodistal end, and the left bifurcation sports a long, lightly sclerotized, delicate, dentate titillator that extends posteriad and curves dorsad (Figs. 15–18). The uncus, in a more or less posterior view (Figs. 15, 17), is hollowed out in such a way as to suggest a pitched roof above the gnathos, although, in lateral view (Figs. 16, 18), it appears to be massive and solid at its distal end. The views of the genitalia in Figs. 15 and 17 are purposely oblique, mainly to show both the rooflike form of the uncus (something that the usual dorsal view does not do) and the bifurcation at the distal end of the aedeagus, but also to convey more about the dentation at the distal ends of the symmetrical valvae than is evident in lateral view.
The steeper pitch of the uncal roof in Fig. 15 as opposed to Fig. 17 is not an interspecific difference. It merely reflects slightly different angles of observation and the fact that the uncus of the individual in Fig. 17 happens to be less laterally compressed. These figures are not strictly comparable in some other respects as well, e.g., the right valva of the male in Fig. 17 is, by chance, somewhat splayed.
Female Genitalia (Figs. 19–21). The female genitalia are also similar (number of dissections compared: 6 R. osca and 5 R. molion). A notable common feature is the greatly reduced (and lightly sclerotized) lamella antevaginalis, ventral to the ostium bursae. A more or less distinguishing (though variable) feature is the shape of the posterior edge of the lamella postvaginalis, on either side of its prominent, midventral, V-shaped notch: in outline, that edge is less rounded in R. osca (Fig. 19) than it is in R. molion (Fig. 21).
Because female genitalia are often less fully and less heavily sclerotized than are those of the male, they can be more individually variable. As always, in comparing genitalia in order to distinguish between individual and interspecific variation, the state of the dissection and the viewpoint of the observer are critical. Seeming interspecific differences in the genitalia shown in Figs. 19 and 21 are mostly artifactual. For example, the lamella postvaginalis is, by chance, splayed in Fig. 19 but not in Fig. 21 (moreover, as a result, overlying tergum VIII appears in Fig. 19 but not in Fig. 21). Orientation of dissection X-6399 so as to give a good view of the sterigma (Fig. 21) foreshortens some other features, especially the ovipositor lobes (whose real length is evident in Fig. 20).
The ovipositor lobes, in lateral view (Fig. 20), have a straight distal edge; and their large setae are peripheral rather than generally distributed (Figs. 19–21). Together, the ovipositor lobes present to the outer world a flat face, ringed with long setae.
Larval Facies (Figs. 22–27). Adults of R. osca (Figs. 1–4) and R. molion (Figs. 5–8), with their disparate color patterns, come from caterpillars so similar to each other that they are often confused in the wild, even by experienced collectors. However, grown caterpillars are distinguishable (Figs. 22–27). Basic features of the color pattern shared by last-instar caterpillars are a black nubbly head whose outer edges, in frontal view, present a pair of broad, highly contrasting, light yellow, dorsoventral stripes; and a body that is finely and densely dotted with green. On each side, the black head of R. osca has a second yellow, dorsoventral stripe (Fig. 24), which is so posterolateral in position as to be hidden in most views. Lacking this stripe, R. molion is solidly black in the same area (Fig. 25). In R. osca, but not R. molion, the yellow frontal stripe is actually light brown along most of its inner margin (Figs. 22, 24, 26); and the lateral adfrontal suture (Stehr 1987: 290, fig. 26.1) is also brown (Fig. 26), not black, as it is in R. molion (Fig. 27). The green dots on the body are more prominent in R. molion than they are in R. osca. But laterally, in R. osca, some green dots merge into narrow, irregular strips of solid green that collectively form two wavy, somewhat discontinuous, longitudinal stripes along each side (Fig. 22).
The most notable features of a Rhinthon pupa (Figs. 28, 29) are the long proboscis sheath that extends almost to the tip of the cremaster and the pair of rusty to reddish mesothoracic spiracle covers (MacNeill 1964: 201, fig. 8) that contrast with their surroundings enough to suggest eyes, particularly in frontal view. The pupa and the interior of its shelter are at least partly coated with a flocculent white wax.
Foodplants (Table 1). Rhinthon caterpillars in ACG feed only on plants in closely related families of the order Zingiberales: Marantaceae, Cannaceae, Zingiberaceae, and Heliconiaceae. Most records, by far, are in Marantaceae, and within that family, in genus Calathea. Rhinthon osca and R. molion have similar tastes: they eat the same species of Calathea and, much less often, the same species of Pleiostachya (and both skippers have expanded their diet by using the same introduced species of Canna). The main difference is that R. osca is frequently found on Maranta arundinacea, a plant on which R. molion has never been found. Rearing records for R. osca, but not R. molion, include three more genera of Marantaceae, as well as one species of Zingiberaceae and one of Heliconiaceae. This may merely reflect the fact that the records for R. osca outnumber those for R. molion by 5 to 1.
Rhinthon cubana (Herrich-Schäffer), which some treat as just subspecifically distinct from R. osca (see Mielke 2005: 1269–1270), inhabits the Greater Antilles. Scant foodplant records agree with two from ACG, Costa Rica: in Cuba, Gundlach (1881) found R. cubana feeding on Canna, and Fernández (2001) found it feeding on M. arundinacea.
DNA Barcodes. Neighbor-joining trees derived from DNA barcodes of reared ACG hesperiids have consistently grouped R. osca and R. molion (under the names Rhinthon cubana and Neoxeniades molion) in a tight, two-taxon cluster: interspecific sequence divergence is about 3.5%. To date, 13 adults of R. osca and 16 of R. molion have been barcoded. These Rhinthon species tree far from species that are true Neoxeniades (see Appendix SII in the online version of Janzen et al. 2009).
With the addition of R. molion, Rhinthon can no longer be called a “primarily West Indian genus” (Smith et al. 1994). Rhinthon molion ranges from Mexico to Peru, and R. osca ranges even more widely, from southern Texas and Mexico to Colombia, Ecuador, and Venezuela, plus Trinidad and Tobago, whereas R. cubana occurs sporadically only in Cuba, Jamaica, Hispaniola, and Puerto Rico. A supposed West Indian congener can be discounted: the singular facies of the Hispaniolan skipper R. bushi Watson, which was described from one male from the Dominican Republic, does not appear to fit even the expanded Rhinthon mold (or, indeed, that of other hesperiine genera). Moreover, “further examination of the genitalia of this insect suggests that its placing in Rhinthon was incorrect …” (Smith et al. 1994).
Larval foodplants of Rhinthon in Area de Conservación Guanacaste, northwestern Costa Rica, and number of rearing records for each species of plant.
Now that it includes R. molion, Rhinthon emerges as yet another neotropical genus with one or more species whose adults converge on the flashy color pattern shown in Figs. 5–12. This presumably mimetic convergence is all the more notable because Rhinthon and Astraptes are in different subfamilies (as are some of the other genera involved in this mimicry complex).
Despite the removal of R. molion, Neoxeniades still includes species such as N. luda (Hewitson) and N. pluviasilva Bums with the mimetic color pattern (Burns et al. 2007: figs. 12–15, 24–27; Janzen et al. 2009: fig. 4, photos 23, 24), although it is less exact (especially ventrally) than the resemblance between R. molion and the species of the A. fulgerator complex (Janzen et al. 2009: fig. 4, photos 1–11). But despite the removal of R. molion, Neoxeniades is still polyphyletic (Burns, in prep.).
The genitalia do not suffer the same selection pressures as do facies, and by themselves provide enough information to justify the shift of molion from Neoxeniades to Rhinthon. The genitalia of R. molion mirror those off R. osca, but differ conspicuously, in all of their parts, from those of true Neoxeniades (compare Figs. 15–18 with Burns et al. 2007: figs. 38–41).
The diverse covarying characters—DNA barcodes, larval color pattern, and larval diet—that support the genitalic evidence are no doubt genetically independent of one another; and so, in general, they are potentially of great taxonomic value. In the ease of Rhinthon, all three of these characters are strong. In some cases, however, DNA barcodes are the most useful of these three characters for low-level grouping. Consider, for example, the phylogenetieally compact A. fulgerator complex of 11 species (which has been analyzed with specimens from ACG, although the complex ranges far more widely and includes still more species). The A. fulgerator species complex constitutes a distinct cluster in a neighbor-joining tree derived from barcodes; but the caterpillars of the various species do not (despite some shared basic elements) conform to a single color pattern (Hebert et al. 2004: fig. 2; Janzen et al. 2009: fig. 5, photos 1–11), nor do they all eat plants that are phylogenetically close to each other.
We thank Donald Harvey for dissecting genitalia and Young Sohn for drawing them, Karie Darrow for photographing adults and their secondary sex character and for setting the table and plates, ACG parataxonomists for finding caterpillars and rearing adults, Tanya Dapkey for plucking and shipping LEGS AWAY/FOR DNA (capitals denote the two-line, printed, black-on-yellow label affixed to adult specimens delegged for barcoding), and Olaf Mielke for reviewing the manuscript. Support for this study came from the National Museum of Natural History Small Grants Program (J.M.B.); from National Science Foundation grants BSR 9024770 and DEB 9306296, 9400829, 9705072, 0072730, and 0515699 (D.H.J.); and from grants from the Gordon and Betty Moore Foundation and Genome Canada through the Ontario Genomics Institute (P.D.N.H.).