Some butterflies depart from the usual lepidopteran show of opaque scales by evolving wings that are partly to fully transparent. Among skipper butterflies of the tribe Pyrrhopygini, many of which are gaudy, certain members of the genus Oxynetra caught our attention owing not only to extensive alar transparency but also to a troubling lack of taxonomic transparency.

The neotropical genus Oxynetra C. & R. Felder, 1862—proposed for O. semihyalina C. & R. Felder, 1862—currently comprises three species: O. semihyalina, O. confusa Staudinger, 1888, and O. hopfferi Staudinger, 1888. Both sexes of O. semihyalina, and especially of O. confusa, have extensively transparent wings, with forewings whose hyalinity is primarily medial but also subapical. In males of O. hopfferi, the medial hyalinity is reduced and the subapical hyalinity is lacking; but in females, there is none at all—the wings are entirely black. Given such extreme sexual dimorphism, it is not surprising that the female was originally described as a distinct species in its own genus, Dis annulatus Mabille, 1889. However, Godman and Salvin (1893) correctly synonymized genus Dis with Oxynetra and argued that the female of O. annulatus might prove to be that of O. hopfferi. Evans (1951) formally synonymized O. annulatus with O. hopfferi, an action that prevails (Mielke 2005).

Unlike its congeners, O. hopfferi is rare in collections, as are females of Oyxnetra generally (e.g., for BMNH, Evans [1951] records 99 males of O. semihyalina and 44 males of O. confusa, mainly from Bolivia and especially Peru, but no females of either species). After 22 years of intensive continuous collecting and rearing in Area de Conservacion Guanacaste (ACG) in northwestern Costa Rica (Burns & Janzen 2001, Janzen et al. 2009), the first O. hopfferi-like specimen surfaced in July 2001, too late to appear in the detailed account (then in press) of pyrrhopygine skippers of ACG (Burns & Janzen 2001). This unforeseen species adds a new larval foodplant, Prunus annularis Koehne, and a new family, Rosaceae, to those eaten by Pyrrhopygini in ACG (Janzen & Hallwachs 2012).

That 2001 specimen, a male (voucher code 01-SRNP-6995), has—in contrast to the multi-orange-banded abdominal pattern of male O. hopfferi—a single orange band on abdominal tergum III. However, one (02-SRNP-23283) of two males reared in 2002, has, in addition to the bold orange band on tergum III, some sparse orange scaling on tergum IV; and the male (03-SRNP-3638) reared in 2003 has a scattering of 23 orange scales there. Despite such individual variation, the general effect is one band.

Multiple rearings of phenotypically similar wildcaught caterpillars unequivocally associate the dimorphic sexes, and DNA barcoding confirms the association. Barcodes of all 10 ACG specimens are identical (Fig. S1 of Janzen et al. 2011).

In early 2011, when Mike Stangeland began preparing photographs of reared Oxynetra males for the Butterflies of America (BoA), website  http://butterfliesofamerica.com (Warren et al. 2012), he noticed that they differed from photographs of O. hopfferi males in other collections and raised the possibility of a new taxon. Subsequent investigation, assisted by many collaborators, indicates that the northwestern Costa Rica specimens represent an undescribed species, closely related to, but distinct from, O. hopfferi.

Discussion

Genitalia. Do male genitalia distinguish O. stangelandi from O. hopfferi? Close, side by side comparison of the dissections that appear in Fig. 3, as well as another, reveals various small differences. These include a larger middorsal knob at the distal end of the tegumen, a more massive, more dentate distal division of the valva, a longer, more dentate pointed projection from the inner side of the valva, and a longer penile hump in O. hopfferi than in O. stangelandi. Judging from over half a century of perusing countless skipper genitalia, Burns cautions that these apparent differences may reflect nothing more than individual variation. No matter how simple their form, genitalia vary appreciably within a population; and individual variation generally becomes more evident as genitalic complexity increases. Given our minimal genitalia samples—two complete dissections of O. stangelandi and one of O. hopfferi, plus photographs of valval exteriors (in situ, and partly exposed by dusting) of two other specimens of O. hopfferi—we cannot identify interspecific differences with certainty. A difference we once considered potentially valuable in telling O. stangelandi and O. hopfferi apart involves the dorsal margin of the valval body, distal to its pronounced hump: in lateral view, this margin makes an even, concave curve down to a slightly upturned, rounded, protruding base in O. hopfferi (Fig. 3b) but a straight line to a sharp, wide, basal angle in O. stangelandi (Fig. 3a). No such difference marks the second example of O. stangelandi.

However, relatively minor genitalic differences may mean more in Oxynetra than they do in many other genera, because the genitalia of all four species are conservative variations on a single complex theme. Even though the facies of O. semihyalina and O. confusa (Fig. 6) differ considerably from those of O. hopfferi and O. stangelandi (Figs. 1, 2), their genitalia (Fig. 4) do not. Features showing interspecific variation between O. semihyalina and O. confusa include the distal division of the valva and the penile hump, which are features noted above that may distinguish O. stangelandi from O. hopfferi. The differences in the distal division of their valvae (compare the oblique views of genitalia in Figs. 3a and 3b) look especially promising for species discrimination. (For an example of intraspecific variation in genitalic form, compare photos of the male genitalia of O. confusa in Orellana 2008:fig. 212 with Fig. 4b.)

Barcode differences and intrageneric phylogeny. We obtained barcode sequences for the entire type series of O. stangelandi and for three O. hopfferi specimens. All 10 of our O. stangelandi sequences are identical within their sequenced length, and all specimens are from the same locality. Sequences of the three O. hopfferi specimens differ from each other by 2 or 4 nucleotides, which amounts to 0.3% and 0.6% intraspecific differences (Fig. 7a); and these three specimens are from different localities within Costa Rica. The O. stangelandi sequence differs from those of O. hopfferi by 17, 19, and 21 nucleotides (2.6%, 2.9%, 3.2% differences). Differences of this magnitude are consistent with interspecific distances found in other similar sets of hesperiid species. Moreover, in some unquestionably valid cryptic species, interspecific distances are much smaller: in large ACG samples of Perichares, for example, P. adela (Hewitson) differs from P. prestoeaphaga Burns by about 0.6%, from P. geonomaphaga Burns by about 0.8%, and P. prestoeaphaga from P. geonomaphaga by about 0.7% (Burns et al. 2008). Indeed, some closely related skipper species may differ by only 1 to 3 nucleotides (Burns et al. 2007).

Fig.1.

Oxynetra type specimens. a, O. stangelandi holotype, male, ACG, Costa Rica, voucher code 02-SRNP-23284. b, paratype, female, ACG, Costa Rica, 02-SRNP-23286. c, O. hopfferi holotype, male, Chiriqui, Panama, leg. Ribbe, Staudinger Collection [ZMHB]; with specimen labels. d, Dis annulatus holotype, female, Chiriqui, Panama, leg. Trötsch, Staudinger Collection [ZMHB]; with specimen labels. Each specimen in dorsal (above) and ventral (below) view.

Fig. 2.

Oxynetra adults. a–h, O. stangelandi. i–q, O. hopfferi. f–l, n–o, q, males (banded wings). a–e, m, p, females (unmarked wings). Subscripts: v, ventral view; l, live individual; a, magnified portion of dorsal hindwing near anal fold. i–k, from Puntarenas, COSTA RICA, l–q, from Chiriqui, PANAMA. n, illustration of O. hopfferi holotype from original description (Staudinger 1888). o–p, illustrations of O. hopfferi and Dis annulatus holotypes from Godman & Salvin (1893). q, illustration of O. hopfferi holotype from Draudt (1921), apparently copied from Staudinger illustration (n) instead of drawn from specimen. Voucher codes of Costa Rican paratypes: a, 02-SRNP-23540; b, 02-SRNP-23109; c, 02-SRNP-23285; d, 02-SRNP-23110; e, 02-SRNP-24529; f, 03-SRNP-3638; g, 01-SRNP-6995; h, 02-SRNP-23283. i, Coto Brus, Sabalito, Cotoncito, 1500 m, 8.943, -82.787, Puntarenas, COSTA RICA, 14 September 2006, R. Gonzalez [INBio]; j, Coto Brus, Sabalito, Estacion Biologica Las Alturas, 1500 m, 8.949, -82.836, Puntarenas, COSTA RICA, 31 January 1992, M. Zumbado [INBio]; k, 97-ZFuentes-055: Peñas Blancas, Monteverde, Puntarenas, COSTA RICA, 1997, Z. Fuentes [USNM].

Fig. 3.

Oxynetra male genitalia. a, O. stangelandi, dissection code X-6958. ACG, COSTA RICA, voucher code 02-SRNP-23283 [USNM]. b, O. hopfferi, dissection code X-6959. Peñas Blancas, Monteverde, Puntarenas, COSTA RICA, 1997, Z. Fuentes, voucher code 97-ZFuentes-055 [USNM]. Left side: genitalia (minus penis) in dorsal and left lateral views. Right side: penis in left lateral view, genitalia (minus penis) in oblique view.

Fig. 4.

Oxynetra male genitalia. a, O. semihyalina, dissection code X-6976. Tingo Maria, 300 m, Huanuco, PERU, 24 June 1982, S. S. Nicolay [USNM]. b, O. confusa, dissection code X-6977. Carpish, 3000 m, 9° 42′ S, 76° 04′ W, Huanuco Dept., PERU, February 1992 [USNM]. Left side: genitalia (minus penis) in dorsal and left lateral views. Right side: penis in left lateral view, genitalia (minus penis) in oblique view.

Fig. 5.

Oxynetra stangelandi female genitalia. Dissection code X-6960. ACG, COSTA RICA, voucher code 02-SRNP-23285 [USNM]. Left: posterior view. Middle: ventral view. Right: right lateral view.

To compare the differences between O. stangelandi and O. hopfferi to differences between them and species that morphologically appear related but more distant, we queried BLAST search < http://www.ncbi.nlm.nih.gov/blast> for publically available sequences most similar to that of O. stangelandi and obtained barcodes of O. confusa and Olafia roscius (Hopffer, 1874) (Fig. 7a). Olafia is the genus most similar to Oxynetra; and its sole species, Ol. roscius, was previously treated as Oxynetra (Evans 1951). The O. stangelandi sequence differs from sequences of O. confusa and Ol. roscius by 35 and 36 nucleotides, respectively (around 5.5%). Though smaller, the barcode distances between O. stangelandi and O. hopfferi are still comparable to those between O. stangelandi and O. confusa, and even to those between Oxynetra and Olafia.

Fig. 6.

Oxynetra adult males. a–b, O. semihyalina, Carpish, PERU, May 1992 [USNM]. c–d, O. confusa, Carpish, 3000 m, 9° 42′ S, 76° 04′ W, Huanuco Dept., PERU, February 1992 [USNM]. a, c, dorsal view. b, d, ventral view.

Although the 654 base pair sequence is generally too short for reliable phylogenetic inference, we applied four standard phylogenetic reconstruction programs offered at  http://www.phylogeny.fr to our genetic data. The resulting trees (Fig. 7b–e) are topologically identical and show O. stangelandi and O. hopfferi as sister species, in agreement with the phenotypic data. Branch lengths in the trees in Fig. 7b, c, and d are to scale and graphically show the extent of intraspecific variation in O. hopfferi and how it scales to interspecific differences.

Distribution, ecology, life history, behavior, and mimicry. So far, O. stangelandi is known only from the 1,000–1,500 m zone of cloud forest covering the top of Volcan Cacao (centered by latitude 10.93328, longitude -85.45729) in central ACG. The enormous rearing project in ACG, which is often coupled with mass barcoding of the reared specimens, has led to the description of many new insect species solely from ACG material; but this does not mean that these species are extremely limited in their geographic distribution.

The shrub-treelet foodplant, Prunus annularis (Rosaceae), occurs throughout the understory and edges of this dense forest and reaches a height of 5 m. Caterpillars of O. stangelandi have been found feeding only on this species; and since almost all other species of plants in this habitat have been intensively searched for over 20 years (Janzen & Hallwachs 2012) for any and all caterpillars (>30,000 records), we assume that O. stangelandi is monophagous. Because it is not known to share this foodplant with other species of Hesperiidae, we infer that empty hesperiid larval shelters on P. annularis belong to this species.

Fig. 7.

DNA barcode sequences and trees. a, positions showing variation in a multiple sequence alignment of the CO1 region from Oxynetra and Olafia species. The aligned 654 positions of Oxynetra CO1 nucleotide sequences correspond to positions 1516–2169 in the Drosophila yakuba mitochondial genome, which is used as the numbering standard, because mitochondrial genomes have different lengths in different taxa. 588 positions that were identical in all four species have been removed for clarity, so that only 66 positions are shown. The uppermost sequence is from the holotype. Species name and voucher code (for our specimens) or Genbank accession code  http://www.ncbi.nlm.nih.gov/genbank (for Oxynetra confusa and Olafia roscius) are shown before and after each sequence, respectively. Drosophila yakuba position numbers are shown above the alignment as columns (e.g., the first shown position is #1528). Hypothetical ancestral sequence is shown below. Nucleotides predicted to be acquired on branches leading to Ox. stangelandi, Ox. hopfferi, Ox. confusa, and Ol. roscius are shown on a red, magenta, green and cyan background, respectively. Changes on the Ox. stangelandi + hopfferi and Oxynetra branches are highlighted in black and gray, respectively. Nucleotides acquired as intraspecific variation are on a yellow background. b–e, Trees obtained from these sequences using different phylogenetic reconstruction methods available at  http://www.phylogeny.fr: b, Neighbor-joining with BioNJ program. c,Maximum likelihood as implemented in PhyML program. d, Bayesian inference as implemented in MrBayes program. e, Maximum parsimony with TNT (branch lengths not computed, only topology is meaningful).

Fig. 8.

Oxynetra stangelandi immatures. a–d, f–m, caterpillars. e, n–z, pupae. Voucher codes: a–d, f–h, n–r, 02-SRNP-23109; e, x, 02-SRNP-23283; i–m, 02-SRNP-23110; s–w, y–z, 02-SRNP-23284 (holotype).

The first individual of O. stangelandi (01-SRNP-6995) was found as a pupa (2 July 2001; eclosed 26 July 2001) in a leaf shelter it had made in the top of an unknown 60 cm tall understory sapling below a 3–4 m tall P. annularis. Skipper caterpillars often construct pupal shelters on plants adjacent to, but different from, their foodplant. A year later, various instars of O. stangelandi were found on P. annularis in the general vicinity (Janzen & Hallwachs 2012), although at least 10 years of search of this foodplant prior to 2001 had not turned up any hesperiid larval shelters. Continued explicit and haphazard search of P. annularis foliage in the same general area from 2003–2011 has yielded only one more caterpillar (03-SRNP-3638), plus one parasitized pupa on a non-foodplant. Oxynetra stangelandi is clearly a very low density species in the 1–3 m height range easily searched by parataxonomists; however, the specimens encountered may represent only the bottom of a higher height distribution of the caterpillars. Dates of adult eclosion range from late July to late December, but the sample is too small to estimate annual phenology. A female caterpillar (02-SRNP-23285) found as second instar on 19 July 2002 and reared in a protective net placed over the foodplant, pupated 30 August 2002 and eclosed 19 September 2002 (at 1220 m elevation).

Oxynetra stangelandi caterpillars have the hairy, strikingly yellow- and black-banded body and hairy head (Fig. 8a–d, f–m) characteristic of many pyrrhopygine skipper caterpillars (Burns & Janzen 2001; Janzen et al. 2009). They are no doubt part of a generally aposematic mimicry system involving hundreds of other ACG caterpillars. Likewise, orange eyespots on the black head of O. stangelandi caterpillars place them in a mimicry complex that comprises hundreds of species of ACG caterpillars exhibiting a great diversity of false eyespot patterns (Janzen et al. 2010). Moreover, the mature pupa is basically white (with small orange and black spots, Fig. 8e, s–z), resembling the white pupae of many other shelter-inhabiting ACG skipper species that mimic pupae attacked by fungi. Pupae that are fungal victims usually become white, highly inedible, and even gustatorially dangerous (see images in Janzen & Hallwachs 2012). In the newly molted O. stangelandi pupa, the head, pronotum, and appendages, including wings, look more or less orange and contrast with dull maroon dorsolateral bands on the cream to tan ground color of the rest of the body (Fig. 8n–r). After about two days, a white waxy layer is exuded over the surface, giving the pupa the white appearance that remains until eclosion.

Although no adult O. stangelandi has been observed in the field, its behavior, when newly eclosed in a large rearing container, is unusual. Newly eclosed ACG hesperiids, in many thousands of cases and across more than 400 species, hold their wings together over their back during hardening, and then usually keep them in that position while walking and perching in the rearing containers. However, both sexes of O. stangelandi harden their wings in the position shown in Fig. 2hl and hold them that way while walking with the characteristic gait and slight up-and-down wing motion of a large ctenuchine arctiid moth (many species of which are diurnal, brightly colored, transparent-winged, and widely believed to be chemistry-based aposematic or Hymenoptera mimics). Though aware that the male in Fig. 2hl must be a hesperid that he had photographed repeatedly as a developing larva and pupa, Janzen had to look twice to reassure himself that it was not a contaminating arctiid whose cocoon had inadvertently come in with food foliage. While the striking colors of the male may serve in courtship and the black color of the female in heat absorption in cold cloud forest, these colors also function in at least two mimicry rings.

History of discovery. We initially addressed Stangeland's suggestion that the reared ACG material might not be O. hopfferi by procuring a specimen whose facies closely match those of the holotype (male) of O. hopfferi (Fig. 1c). The new specimen (voucher code 97-ZFuentes-055 [deposited in USNM]) is a male that eclosed from a pupa collected in 1997 in Peñas Blancas, Monteverde, Puntarenas Province, Costa Rica, which is about 100 km southeast of ACG and considerably closer to the O. hopfferi type locality (Chiriqui). This male has the characteristic O. hopfferi abdomen with five orange bands, as well as a hindwing with a short, narrow white streak in cell 1A+2A–3A. One of its legs, sacrificed to DNA sequencing, yielded a barcode that differed by 2.6% from the uniform barcodes previously obtained for all ten ACG specimens. Subsequent comparison of genitalia gave the uncertain result discussed above.

Our need of a larger sample of O. hopfferi to assess variation was frustrated by the rarity of this species in collections. In addition to the Monteverde specimen (Fig. 2k) and the holotypes of O. hopfferi (Fig. 1c) and D. annulatus (Fig. 1d), we inspected four male specimens, all from Chiriqui, Panama (1, BMNH; 2, ZMHB; 1, MTD). With the help of Kim Garwood and Isidro Chacón, we obtained photographs of four more specimens: three males (from Chiriqui, Panama [OMDZUP], Fig. 2l; and Puntarenas, Costa Rica [INBio], Fig. 2i–j) and one female (from Chiriqui, Panama [OMDZUP], Fig. 2m). The facies of our sample of nine males and two females of O. hopfferi from southern Costa Rica and western Panama do not vary significantly (except that fringe color is dark in the Dis annulatus holotype and largely white in the other female).

Oxynetra annulatus, originally described (in genus Dis) from an almost black female with one orange band on her abdomen, has been synonymized with O. hopfferi. Were O. annulatus the female of the ACG entity, the latter would assume that name. The holotype (Fig. 1d) exhibits a white prothoracic venter devoid of dark scales. The females of O. stangelandi have a fully black thoracic venter, concolorous with the rest of the body (Figs. 1b, 2a_v–e_v). Moreover, they have narrower white areas on the palpi below and a narrower dark middorsal break in the abdominal orange band, which is redder than in the O. annulatus holotype; and they have white wing fringes, in contrast to the mostly dark fringes of the O. annulatus holotype. Holotypes of both O. hopfferi and O. annulatus are from Chiriqui, Panama. We agree that O. annulatus is indeed a female of O. hopfferi. Even without knowing the locality, it is possible to associate the O. annulatus female with the O. hopfferi male, as they both exhibit prominently developed white areas on the venter, contrasting to more restricted distribution of white in both sexes of O. stangelandi; and oranger bands on the abdomen, differing from redder bands in both sexes of O. stangelandi. These differences hold up well in available samples of the two species: 9 males, 2 females of O. hopfferi, and 4 males, 6 females of O. stangelandi.

Oxynetra illustrations in historic works are illuminating. Illustrations in Godman and Salvin (1879–1901) are usually more accurate than are those in Draudt (1921). However, for O. hopfferi, the white streak near the anal fold of the hindwing of the holotype does not appear in Godman and Salvin (vol. 3, pl. 74, fig. 18, reproduced here, Fig. 2o). Although the Draudt illustration (1921: pl. 165, row b, reproduced here, Fig. 2q) of apparently the same specimen depicts this streak clearly, it lacks the small, transparent, triangular wedge at the base of forewing cell M₃- CuA₁ and shows 7 orange bands (6 major and 1 minor) instead of “5 bright orange-red transverse band[s] on the dorsum of the abdomen” as specified in the text (Draudt 1921:849). The Draudt illustration was probably drawn from the illustration in the original description (Staudinger 1888: pl. 99, as felderi, reproduced here, Fig. 2n), because the Staudinger illustration also shows 7 orange bands.

Collaboration among researchers from five countries has resulted in rearing and recognition of a new species of Oxynetra that differs from its sister species, O. hopfferi, in body pattern and perhaps in number of antennal nudum segments and in male genitalia. Barcode differences (2.6% to 3.2%) clearly separate the two species. The male and female associations made through rearing and DNA sequencing of O. stangelandi, and similar sexual dimorphism in the facies of both O. stangelandi and O. hopfferi, indirectly support the synonymy of O. annulatus (described from a female) with O. hopfferi (described from a male).

Acronyms of Museums