Introduction

Amber preserves delicate arthropods, mainly insects, with high fidelity, including microscopic features like setae and sensilla on the external cuticle as well as wing and other color patterns (Grimaldi, 1996). In some cases amber preserves soft internal tissues, including cells, organelles, and even endosymbiotic spirochetes and protists (Grimaldi, 1996; Grimaldi et al., 1994; Henwood 1992a, 1992b; Wier et al., 2002).

The most diverse organisms in amber are insects, particularly of the orders Diptera, Hymenoptera, and Coleoptera. Lepidoptera specimens tend to be less common and even rare in amber, the principal record being adult moths. The fossil record of adult butterflies in the world, from the Paleocene to Miocene, contains approximately 50 specimens known, comprising about 33 named species in all living families, but only seven of the specimens have been found in amber, all of them in Dominican amber (Grimaldi and Engel, 2005). For Miocene Dominican amber, the list of fossilized Lepidoptera includes other families: Blastobasidae, Cosmopterygidae, Gelechiidae, Noctuidae, Tineidae and Tortricidae (Poinar [1992]), Tortricidae (Poinar and Brown [1993]), and Oecophoridae (Kristensen and Skalski [1999]). Grimaldi and Engel (2005) figured a caterpillar and two adults of the family Geometridae, an adult of the genus Acrolophus (Acrolophidae), and other adults of Tortricidae and Gelechioidea; an interesting tineoid moth figured by these authors is a case with the caterpillar inside. The scarce butterflies found in Dominican amber have been studied in detail, and are represented as caterpillars of the families Nymphalidae and Riodinidae (DeVries and Poinar, 1997; Hammond and Poinar, 1998) and adults of Riodinidae (Poinar, 1992; Grimaldi, 1996; DeVries, 1997; Hall et al., 2004). Here, we report two additional adult butterfly specimens of the families Riodinidae and Nymphalidae as inclusions in Miocene Dominican amber; the nymphalid specimen is the first adult known of this family preserved in amber.

The family Riodinidae contains about a thousand species, most of which occur in the American tropics and are partly characterized by the greatly reduced male forelegs and the silver or leaden spots in the wings, which is why these butterflies are commonly named metalmarks (see DeVries, 1997). Riodinid caterpillars and pupae resemble those of lycaenids, which is why some authors classify them into a single family (Kristensen, 1976; De Jong et al., 1996), though the families are usually considered sister groups (Ackery et al., 1999). The family Riodinidae contains three subfamilies (Nemeobiinae, Euselasiinae and Riodininae). Riodininae, the largest subfamily, is divided into eight tribal groupings; two of them—Mesosemiini and Eurybiini—possessing five forewing radial veins; the other six possess four forewing radial veins (Hall, 2003). The Riodinidae fossil record is comprised only of a caterpillar specimen (DeVries and Poinar, 1997; see the discussion about other putative riodinid fossils in Hall et al., 2004) and the recently described species Voltinia dramba based on five adult females (Hall et al., 2004), both preserved in Miocene Dominican amber. A new female specimen of Voltinia dramba in the collection of the American Museum of Natural History (AMNH) provides an opportunity to complete some details in the description of this interesting species. The most recent investigations on the phylogeny of Riodinidae are Harvey (1987) in Hall (2003), plus the subsequent contributions by Hall (1998, 1999) and Hall and Harvey (2002). The most recent study on the phylogeny of the subfamily Riodininae, by Hall (2003), is a cladistic analysis of the 16 genera that have five forewing radial veins; this study coded data on the adult ecology, wing venation and pattern, the adult head and body, male and female genitalia, and early-stage ecology and morphology. That analysis confirmed the monophyly of the tribes Mesosemiini and Eurybiini, and established two subtribes for Mesosemiini (Mesosemiina and the new subtribe Napaeina). The fossil species Voltinia dramba belongs to the subtribe Napaeina. For Hall (2003), the best hypothesis about the relationships among the four- and five-radial-vein groups of Riodininae is Mesosemiini + (Eurybiini + all four-radial-vein tribes).

The Nymphalidae is the most diverse family of butterflies, with more than 6,000 described species worldwide. The nymphalid butterflies are found in virtually every terrestrial habitat except Antarctica, and have their greatest diversity in the Neotropics (DeVries, 1987). The nymphalids are characterized by very short forelegs in males and females, on which the males of many species carry a “brush” of long scales. The family contains many colorful, large species, and most of the mimicry complexes in butterflies. Nymphalid larvae feed on dicotyledonous plants, very often toxic ones, and they are also generally adorned with a variety of spines and tubercles (DeVries, 1987; Smith et al., 1994). The species of Nymphalidae have been placed into more than 500 genera classified into 12 subfamilies. The oldest nymphalid fossil is an undescribed form from the Early Eocene of Green River (Colorado), but its classification within the family is unclear, and other Eocene records are several species described from Florissant (reviewed in Grimaldi and Engel, 2005). The systematic relationships among the subfamilies and tribes of Nymphalidae are still poorly known and thus the phylogeny of the family has been frequently discussed. There is some evidence of a sister-group relationship between Lycaenidae and Nymphalidae, and Pieridae is possibly the sister group of the Lycaenidae + Nymphalidae (De Jong et al., 1996). Freitas and Brown (2004) presented a historical overview of Nymphalidae phylogeny. Two recent studies, using the wingless gene (Brower, 2000) and mitochondrial and nuclear genes (Wahlberg et al., 2003), concluded that many of the traditional subgroups are monophyletic. The latest and most complete cladistic analysis to date, using 234 characters from all life stages of 95 nymphalid species, has been published by Freitas and Brown (2004), concluding that the taxa can be grouped into six main lineages and it supports the monophyly and relationships of most presently recognized subgroups. That study also supported the position of the subfamily Libytheinae as the basal group of Nymphalidae.

Due to the poor fossil record of butterflies, they have rarely been used in phylogenetic studies. For this reason it is very important to study all remains found in the fossil record, like the two specimens embedded in Dominican amber described here.

Material and Methods

Dominican amber is fossil resin that was formed by an extinct species of Hymenaea tree (Leguminosae: Caesalpinioidea), and is mid-Miocene in age (Iturralde-Vinent and MacPhee, 1996; see the discussion in Grimaldi, 1995, about the erroneous dating of this amber as Eocene-Oligocene by some authors). This amber derives mainly from outcrops in the mountains in the north and northeast of Santiago. The material studied here was acquired through purchase, and as such its exact provenance within the outcrops of Dominican amber mines is unknown, but its authenticity is certain, based on physical characteristics and typical preservation of the inclusions.

The pieces were polished and some parts of the specimens were studied using sugar gel applied between the amber surface and a slide, which reduces the optical distortion caused by the curvature and imperfections of the amber surface. Photomicrography used the Infinity© K-2 long distance microscope and the MicrOptics© fiber optic flash unit ( http://www.microptics.com). The specimens were drawn using a drawing tube attached to a Zeiss Stemi SV8 stereoscope. Specimens are housed in the amber fossil collection, Department of Invertebrate Zoology, AMNH.

SYSTEMATIC PALEONTOLOGY

CLASS INSECTA LINNAEUS, 1758

ORDER LEPIDOPTERA LINNAEUS, 1758

SUPERFAMILY PAPILIONOIDEA LATREILLE, 1802

FAMILY RIODINIDAE GROTE, 1895

SUBFAMILY RIODININAE GROTE, 1895

TRIBE MESOSEMIINI BATES, 1859

SUBTRIBE NAPAEINA HALL, 2003

GENUS VOLTINIA STICHEL, 1910

Voltinia dramba Hall, Robbins, and Harvey, 2004

Figures 1, 2a–c

Taphonomy

Amber in general preserves insects and other organisms that are uncommon in the fossil record (e.g., Lewis and Grimaldi, 1997), and many specimens are unique, like the nymphalid described here. Normally, the inclusions are small, and on occasion there are preserved large insects like butterflies (Grimaldi, 1996; Grimaldi and Engel, 2005), but most butterflies and moths in amber are small, rarely larger than one centimeter (Weitschat and Wichard, 2002). Small vertebrates, including the remains of mammals, amphibians, and reptiles, have been also found (e.g., Rieppel, 1980; Poinar and Cannatella, 1987; Grimaldi, 1996; MacPhee and Grimaldi, 1996). This indicates voluminous production and exudation of resin by extinct species of Hymenaea during the Miocene, just as occurs with living species of the genus when wounded by storms or boring insects (Langenheim, 2003). There is also evidence that the Tertiary resin was very fluid. The wing surfaces of butterflies easily repel most fluids and are very resistant to wetting, due to the vestiture of imbricate scales. Thus, only a very fluid resin could penetrate such surfaces completely and permit thorough fossilization. The nymphalid specimen has numerous gas bubbles trapped on the wing surfaces, due to the apparently original resistance against the fluids, but the rionidid specimen has no bubbles. Possibly, the scarce record of butterflies in Tertiary ambers is a consequence of a bias against the preservation of large insects, a hypothesis that agrees with the more abundant Tertiary amber record of moths. Moths and butterflies are rarely fossilized in lacustrine deposits perhaps because their broad, scaly wings resist wetting (Martínez-Delclòs et al., 2004).

Wing color patterns are commonly fossilized in both main types of Konservat-Lagerstätten in the insect fossil record, including laminated carbonates and amber. These color patterns have a chemical (pigments) or a physical (e.g., diffractive) origin. Chemical color pattern is preserved in laminated rocks due to anoxic conditions in the bottom of some aquatic environments (Martínez-Delclòs et al., 2004). Fidelity in amber also includes chemical color patterns, surely due also to the anoxic conditions inside resin, but also to the complex mixture of terpenes and other hydrocarbons.

The genus Dynamine is generally recognized by their metallic markings, either on the hind-wing underside or on the upper side of both wings (DeVries, 1987), but these markings are preserved in amber as brown spots, not the original metallic optical effect. Some scales in the Dynamine specimen preserve the original color; for example, on the ventral part of both the abdomen and tibiae are scales that are preserved with their original white color. These white scales are not a taphonomic alteration since such scales are present precisely on the white abdominal and leg parts of extant Dynamine.

Hall et al. (2004) indicated that the five known adults of Voltinia dramba are females, and that caterpillars of Recent species of the subtribe Napaeina feed exclusively on the leaves of epiphytic plants (DeVries, 1997), and thus presumably near resin production in Hymenaea trees where the females oviposited. Subsequently, Vane-Wright (2004) agreed that this record of five females is not a random phenomenon. The new female specimen in the AMNH collection supports this hypothesis of taphonomic bias in the preservation of females. Members of the genus Dynamine are essentially forest butterflies and the extant subspecies D. egaea zetes in Hispaniola has the habit of congregating around trees of different species (Schwartz, 1989). This habit presumably facilitated the entrapment of Dynamine alexae in resin.

Paleoecology

Hall et al. (2004) indicated that caterpillars of V. dramba possibly fed on the leaves of epiphytic Bromeliaceae and Orchidaceae, because this species belongs to a group of genera whose larvae feed on these plants (DeVries et al., 1994). There are other groups of insects found in Dominican amber that today are associated with Bromeliaceae and Orchidaceae, such as stalk-winged damselflies and orchid bees (Engel, 1999), and, in addition, some plant remains found in this amber could belong to orchids and indeterminate epiphytes (Poinar and Poinar, 1999). Minute bromeliads do occur in Dominican amber (AMNH collection).

The caterpillars of extant Dynamine are specialist feeders on flowers or bracts and developing leaves of the genera Tragia and Dalechampia in Euphorbiaceae (Brown and Heineman, 1972; DeVries, 1987). Schwartz (1989) observed D. egaea zetes adults in Hispaniola feeding on the flowers Daucus sp. (Apiaceae) and on Lantana ovatifolia (Verbenaceae), and Alayo and Hernández (1987) cited that caterpillars of D.e.calais can feed on Tragia urens (Euphorbiaceae). Indeed, the adults of Dynamine are generally uncommon except around Dalechampia. In several sites in Ecuador where two or three species of Dynamine are encountered (typically along streams and rivers), nine species were attracted to spots where potted Dalechampia plants were placed in forest gaps (P. J. DeVries, personal commun. 2005). Dominican amber contains abundant plant remains (including myriad flowers), but the plants are very poorly studied, so it is not surprising that Dalechampia is as yet unrecorded from this amber. Poinar and Poinar (1999) cited the families Euphorbiaceae and Urticaceae in this amber, and there is a euphorb flower in the AMNH collection. According to a paleobiogeographic scenario proposed for Dalechampia (Ambruster, 1994), it is probable that this important host plant genus for Dynamine lived during the Miocene in Hispaniola. Extant Dalechampia need resin-collecting bees for its pollination, and such bees (Meliponini: Proplebeia) are common in Dominican amber (Camargo et al., 2000). Lastly, the extant caterpillars of Theope, the other butterfly genus present in Dominican amber, feed on a wide group of plant families, which are: Sterculiaceae, Fabaceae, Bombacaceae, Cecropiaceae, Lecythidaceae, Euphorbiaceae, Convolvulaceae, and Cochlospermaceae (DeVries, 1997).

Paleobiogeography

The fossil butterflies in Dominican amber indicate there was significant extinction in the Antillean fauna, to which we can add another example, Dynamine.

The family Riodinidae is poorly represented in the Antilles, as there is presently only one Recent riodinid species, of the genus Dianesia (Cuba and the Bahamas). Thus the genera Voltinia and Theope, both with fossil specimens in Dominican amber, have no species in the Greater Antilles today. Voltinia has nine living species on the Neotropical mainland; Theope is widely distributed in the Neotropics from Mexico, through Central America to northwestern Peru, throughout the Guianas to southeastern Brazil and Paraguay, and along the eastern slope of the Andes as far south as northern Argentina (Hall, 1999), including some islands. More than half of all named species of Theope are thought to occur only in the Amazon Basin (DeVries, 1997).

Many of the 70 species of nymphalids in the West Indies are endemic, sometimes to a single island; biogeographically there are species with Neotropical and Nearctic affinities and others with relatives on the African continent (Smith et al., 1994). There are about 30 species of Dynamine from the southern United States throughout Central and South America, and the greatest species diversity is in the Amazon Basin; some species are extremely restricted (DeVries, 1987). Central America contains 14 species and there are two species in the Antilles (D. egaea and D. mylitta), both of which are also well represented on the continent (each from Mexico to Paraguay-Argentina). According to Miller and Miller (1989) the Dynamine species in the West Indies are most closely related to those of Mexico and Central America, like most of the butterflies of the Greater Antilles. These authors further consider Dynamine to be good dispersers. Of the 36 species of nymphalids on Hispaniola, only one of them is in the genus Dynamine, specifically D. egaea (subspecies D. e. zetes), which is very similar to the Cuban subspecies D. e. calais. The subspecies D.egaea dyonis, occurring in Central America, more closely resembles the Jamaican subspecies D. e. egaea than the other two Antillean forms, and this has been attributed to the more recent separation of Jamaica from the Yucatán than for the other islands where this species occurs (Smith et al., 1994). We consider a vicariant model for the Jamaican subspecies implausible, given such subtle differences between it and the mainland populations, and the millions of years of geological separation of these landmasses. Dynaminemylitta from Cuba and from the continent are identical, suggesting that they reached the island by dispersal, in contrast to the proposed vicariant distribution and subsequent divergence of island populations of D. egaea.

Miller and Miller (1989) reviewed theories and models for the biogeography of West Indian butterflies, indicating that historically these were dispersalist models. In contrast, Miller and Miller (op. cit.) proposed a combination vicariance/dispersal model, with emphasis on vicariance based on new geological evidence. These authors indicated that it is necessary to analyze the fossil record in order to determine the ages of butterflies, but only a few fossils were available. After study of the riodinid fossils, Hall et al. (2004) proposed that the rarity of Riodinidae in the Antilles today can be attributed to two primary factors; these butterflies are: 1) poorly dispersed over water, and 2) “disproportionately subject to Tertiary extinctions.” Exactly what the “Tertiary extinctions” are is unclear, but presumably includes extinctions due to the insularization of Antillean land masses (there is, actually, one more possibility: that riodinids disperse well but do not colonize well). The finding by Hall et al. (2004) that Voltinia dramba is closely related to V. danforthi from northwestern Mexico actually contradicts the suggestion that a “probable cause” for the genus becoming extinct in the Antilles is xerophytization from Plio-Pleistocene cooling. Voltinia danforthi occurs in the xerophytic states of Sinaloa and Sonora.

We find it most likely that Voltinia's distribution in nuclear America included the Greater Antilles when these islands were closer to or actually fused with the mainland, perhaps in the Early Miocene or Late Oligocene. As the islands became more isolated through drift and/or rising sea levels, Voltinia and scores of other arthropod groups became extinct on the Antilles. Perhaps the best reflection of this is seen in the meliponine bees (Apidae: Meliponini). With the exception of a species living in Cuba that might have been introduced, no meliponines occur in the Antilles, even on the large islands of Trinidad and Tobago close to Venezuela. Meliponines are notoriously poor dispersers, particularly over water, and opposite sides of the Amazon River are known to harbor different meliponine faunas (C.D. Michener to D.G, personal commun.). Yet meliponines were in the Antilles. There are three known species of the extinct genus of stingless bees Proplebeia in Dominican amber (Camargo et al., 2000), whose presence probably reflects a time when Hispaniola was close to the Central American mainland or after they had recently separated. If Hispaniola was directly connected to Central America when Dominican amber was being formed, we suspect the fossil meliponine fauna would be more reflective of continental diversity. We agree with Hall et al. (2004) that the poor dispersal ability (and possibly a poor colonization ability) keeps riodinids virtually nonexistent in the Antilles today, but their extinction from that region was probably due to the same reason why island faunas are generally depauperate.

Comments on the Age of Butterflies

The age of butterflies is becoming a contentious issue. Some entomologists maintain that butterflies evolved in the Cretaceous (e.g., Miller, 1987; Miller and Miller, 1997; Viloria, 1998; Hall et al., 2004; Braby et al., 2005 [see also the citations given therein]). Others maintain that butterflies evolved largely or wholly in the Tertiary (Shapiro, 1994; De Jong, 2003; Vane-Wright, 2004; Grimaldi and Engel, 2005), in approximately half the time of the other estimates.

Hall et al. (2004) supported a Cretaceous origin and divergence for butterflies on the basis of their study of the Voltinia fossil in Dominican amber. Assuming that the extinct Voltinia dramba and living V. danforthi diverged as a result of Caribbean separation from the Central American mainland, they estimated a divergence date of 40–50 million years. First, it is not at all apparent that this divergence coincided with island-mainland drift, even if a geological divergence took place 40–50 mya. It is possible that V. dramba colonized Hispaniola just prior to the time in the Miocene when the amber was formed, either through long-range dispersal or across narrow water gaps during periods of low sea level. Even more obvious, though, is that an Eocene divergence date for these two species is implausibly old. Insects preserved in Eocene Baltic amber, for example, are generally very primitive compared to living genera (Weitschat and Wichard, 2002; Grimaldi and Engel, 2005). Using the 40–50 myo divergence date, Hall et al. (2004) concluded that this provided “additional support for a Gondwanan origin of many butterfly tribes and subfamilies,” citing Miller and Miller (1997) and Viloria (1998), but to which they could apparently now add Braby et al. (2005). All of these authors contend that butterflies evolved as residents of wandering Gondwanan continents, and since Gondwana separated in the Cretaceous, butterflies are at least this old.

The study by Braby et al. (2005) was based on a comprehensive phylogenetic analysis of three genes and the genera of troidine swallowtails (family Papilionidae), which includes the famous bird wing swallowtails. The distribution pattern of troidines in their study appears to be compelling evidence for a Gondwanan distribution. The classic Gondwanan distribution comprises the southernmost, cool temperate forests of Africa, Chile, southeast Australia (including Tasmania), and New Zealand (reviewed in Grimaldi and Engel, 2005). However, no troidines occur in the temperate Austral region. In fact, the areas of endemism defined for each troidine genus by Braby et al. (2005) are unusually widespread, some of which are not even Gondwanan. The lineage that includes Troides and Ornithoptera, for example, extends from southeast Asia to the Solomon islands, New Guinea, and northern Australia, but which they referred to as “Australia”. Likewise, Parides, which is widespread throughout Central and South America, is referred to as “South America”. Another difficulty with the Braby et al. (2005) study is that they used the known ages of landmass separation to calibrate their estimates of taxon divergence, which is circular reasoning if one is testing whether taxa are Cretaceous in age. A third difficulty is that molecular estimates of divergence dates are usually considerably older than evidence based on the phylogenetic study of fossils (Rodríguez-Trelles et al., 2002), but they can also give an illusion of precision (Graur and Martin, 2004). Also, there are Recent insect taxa that show classic austral disjunction, like certain chironomid midges, scolebythid and megalyrid wasps, primitive cyclorrhaphan flies, and others, but for which fossils of many occur in the Northern Hemisphere (Grimaldi and Engel, 2005), indicating formerly widespread distributions. Even if the biogeographic pattern of troidines was compellingly Gondwanan, which we maintain it is not, fossils in general remind us to be cautious about interpreting historical distributions entirely on the basis of the Recent fauna.

Lastly, Braby et al. (2005) make the typical appeal of incompleteness of the fossil record to support their hypothesis of Cretaceous butterflies, namely that Cretaceous butterflies have not been discovered given their rarity.

The fossil record of Papilionoidea, in fact, is quite good among macrolepidopterans. There are approximately 33 named fossil butterflies—more than any other group of macrolepidopterans—and all of them are Tertiary (reviewed in Grimaldi and Engel, 2005). This fact, plus the relatively derived phylogenetic position of butterflies among Ditrysia, makes it unlikely that they are much older than latest Cretaceous. If, according to Braby et al. (2005), troidine swallowtails existed by 90 mya, then the Papilionidae would need to have existed earlier than this, say 100 mya. In this scenario, Rhopalocera (including skippers and Hedylidae) would need to be older still (say 120–130 myo), and Macrolepidoptera and Ditrysia origins would need to have originated in the Jurassic. This scenario defies all fossil evidence. Only small, primitive lepidopterans are known from the Late Jurassic and Early Cretaceous, and only one possible ditrysian (a leaf mine of an apparent gracilliaroid) is known from the Cretaceous (reviewed in Grimaldi and Engel, 2005). It is highly implausible that 80 million years of macrolepidopterans have simply escaped notice.

Tertiary evolution of the butterflies could easily be falsified by the discovery of a Cretaceous specimen, but for the time being we contend that Vane-Wright (2004: 408) was entirely right in saying there is “no existing evidence to suggest that butterflies are older than about 70 million years.”