INTRODUCTION

Among the orders of holometabolous insects, a lineage renowned as the most diverse radiation of animal life, the snakeflies (Raphidioptera) comprise the least speciose group (ca. 206 modern species) and are among the most distinctive and remarkable of lineages. The order is readily recognizable for the elongate prothorax and long ovipositor in females, among other traits, although true autapomorphies for the group are not immediately evident. Today the order consists of approximately 206 species entirely restricted to the Northern Hemisphere and principally in temperate habitats at latitudes and altitudes that experience a cold winter (Aspöck, 1998; Aspöck, Aspöck, and Rausch, 1991, 1998, 1999a, 1999b; Aspöck, Aspöck, and Yang, 1998; Yang, 1998). A particular concentration of species can be found in Central Asia and around the Tien Shan Mountain range passing from Tadjikistan, through Kyrgyzstan, and extending into western Uzbekistan, southeastern Kazakhstan, and the northwestern borders of China (Aspöck et al., 1999a). Species of the order are arboreal. Larvae are generalist predators and relatively long-lived, principally living under the bark of trees and shrubs (a few raphidiids live in leaf litter at the base of shrubs). Adults are short-lived and, like the larvae, are predaceous, although capable of capturing only weak prey. Species are generally inept flyers and thereby poor dispersers, making the group susceptible to vicariant events and subsequent adaptation and speciation.

Together with Neuroptera and Megaloptera the Raphidioptera comprise the superorder Neuropterida and the living adelphotaxon of the beetles (Coleoptera) (Mickoleit, 1973; Kristensen, 1991, 1999; Whiting et al., 1997; Hörnschmeyer, 1998, Carpenter and Wheeler, 1999; Wheeler et al., 2001). While the overall position of the superorder within Holometabola is of little debate, some controversy remains concerning the relative positions among the three neuropterid orders. Most of the recent studies from both morphological and molecular evidence have converged to support a Raphidioptera + Megaloptera sister-group relationship (e.g., Whiting et al., 1997; Carpenter and Wheeler, 1999). This is somewhat supported by paleontological evidence since for the earliest of fossil snakeflies the hind wing venation is remarkably megalopteran-like while the forewings are typical for primitive Raphidioptera (see Discussion below and under Priscaenigmatidae in appendix 1), tantalizingly suggesting that the Raphidioptera are indeed more closely allied to the Megaloptera and diverged from them in the early Mesozoic.

Fossils of snakeflies have been recognized since the middle of the 19th century. The first fossil was described as Raphidia (Inocellia) erigena Menge (Menge, 1856; Pictet-Baraban and Hagen, 1856) from an individual preserved as an amber inclusion from the Tertiary of northern Europe. Throughout the remainder of the 19th century and well into the early 20th century numerous other Tertiary snakeflies were described from compression fossils in North America, almost exclusively from Florissant, Colorado (appendix 1). In 1925 the great founder of Russian paleoentomology Martynov (1925a, 1925b) discovered the first Mesozoic representatives of the order in Upper Jurassic deposits of Central Asia. Subsequently, Carpenter (1967), Bode (1953), and Martynova (1947, 1961) proposed additional Jurassic and Cretaceous species adding significantly to our knowledge of the group. Modern authors have proposed numerous more Mesozoic species, principally from the Upper Jurassic and Cretaceous of Asia (e.g., Ponomarenko, 1993; Ren, 1997), but also from the Jurassic of Europe (Whalley, 1985; Willmann, 1994) and the Lower Cretaceous of Brazil (e.g., Oswald, 1990); all are preserved as compressions. Lower Cretaceous species are also known from Spain (Martínez-Delclòs, 1991; Peñalver et al., 1999) and Korea (Engel, unpubl. data), but remain undescribed. Although numerous Paleozoic species have been described as plesiomorphic lineages of Raphidioptera, all have subsequently been removed from the order (see Discussion, below).

Until recently, amber snakeflies were restricted to those preserved in middle Eocene (Lutetian) Baltic amber (Carpenter, 1956; Engel, 1995; Weitschat and Wichard, 1998). Grimaldi (2000) described the first Cretaceous-amber snakefly in Turonian amber from New Jersey—the first representative of the extinct family Mesoraphidiidae in amber—as well as a mesoraphidiid larva from the same deposits. Herein a second Cretaceous amber snakefly, also of the Mesoraphidiidae, is described and figured. The new species was discovered in Cretaceous (Cenomanian?) amber from Myanmar and is the first Mesozoic amber snakefly from the Old World. The age and fauna of Burmese amber has most recently been considered by Zherikhin and Ross (2000) and Grimaldi et al. (2002), although several earlier authors have also presented accounts of its age (e.g., Cockerell, 1917). A raphidiopteran larva has also been recovered from these same deposits (Rasnitsyn and Ross, 2000; Engel, personal obs., see below), but is not conspecific with the imago described below and, although perhaps a mesoraphidiid, cannot be confidently assigned to family. Perhaps the most remarkable attribute of the adult specimen considered herein is its diminutive size. The forewing length of the individual is just under 4.5 mm making it the smallest snakefly, living or fossil. This new and particularly noteworthy species is described to make its name available for papers considering the paleontological significance of Burmese amber (Grimaldi et al., 2002) as well as a forthcoming monograph on the neuropterid fauna of Burmese amber (Engel, in prep.). In addition, an overview of the geological history of the order is presented with a taxonomic catalog of fossil snakeflies appended (appendix 1). Morphological terminology follows that of Aspöck et al. (1991).

SYSTEMATIC PALEONTOLOGY

DISCUSSION

To date all “cladistic” studies for Recent and fossil snakeflies have been greatly lacking and the higher phylogeny of the order is confused. The study by Ren and Hong (1994) based on an a priori polarization of six characters for seven terminals is not only the least rigorous but the most fraught with confusion surrounding characters (definitions and codings), identification of snakeflies, and basic systematic principles and theories. Characters from this study are quite dubious, some are clearly continuous without distinctions among states (e.g., prothorax short versus elongate, Rs originating just before or just after wing midpoint) or are based on homology statements that defy logic (e.g., “more crossveins” versus “less crossveins”). No characters were coded as polymorphic for particular lineages when indeed considerable variation exists within the terminals under consideration (e.g., Baissopteridae includes genera with and without pterostigmal veins and there is no clear picture as to which is the ground-plan state for the family, yet this terminal is coded as strictly present). Lastly, these authors claim to have demonstrated that the terminals in their analysis were monophyletic (in this instance, families) and yet, based on their own “data” and “analyses”, two groups lack apomorphies altogether.

The study of Willmann (1994) was certainly more logically founded and more serious. Several familial exemplars were employed, thereby allowing for a more meaningful consideration of the monophyly of particular groups and the obviously erroneously assigned lineages removed. However, the definitions of several characters were similarly continuous and based on unnecessarily complicated transformation series, while some of the characters are simply tenuous. The homologies of particular veins may indeed be of some question, but the comparisons drawn by Willmann (1994) are no more sound than any presently employed. Despite these criticisms, Willmann (1994) has certainly prepared a stronger foundation for the incorporation of paleontological data in Raphidioptera phylogenetics than his predecessors. Although controversial, his hypotheses for relationships among snakeflies are more legitimately based and will require rigorous testing by future cladistic analyses in concert with a critical evaluation of homologies and a broader spectrum of character data. It is to be greatly regretted that the wealth of systematic data present in snakefly genitalia is not available for the fossil lineages.

As one might suspect, the classification of Mesozoic snakeflies is presently an inordinate mess and no clear picture of relationship presently exists. The “distinctive” features of several families are by no means unique to their supposed lineages and, in fact, numerous species are known to intermingle such attributes. The Jilinoraphidiidae and Huaxiraphidiidae are clearly synonyms of Mesoraphidiidae (these families are newly synonymized below; appendix 1). Similarly, although the Sinoraphidiidae is exceptionally poorly documented and described (Hong, 1982, 1992a, 1992b) it is certainly a synonym of Mesoraphidiidae (appendix 1). Despite the removal of these three Cretaceous “families”, considerable confusion persists. The characters separating the remaining families (i.e., Baissopteridae, Mesoraphidiidae, and Alloraphidiidae) are poor. For instance, some of the “distinctive” traits defining the Alloraphidiidae from other families include the strongly distad position of MA relative to the origination of MP, the separation of CuA from M exceedingly close to the stem of R (nearly appearing as a trifurcation), reduced pterostigma, and the greatly elongate wings (Carpenter, 1967, 1992). The elongate wings are truly notable, but they are merely autapomorphic. The former two features are, however, found in the living family Raphidiidae (e.g., some Phaeostigma species) as well as some mesoraphidiids, and for some of these traits there is continuous variation between the states resulting in a confusion concerning the homology of alternate states (i.e., it is difficult to consistently and confidently assign particular taxa to either character-state). Some genera that do appear on the surface to be valid (e.g., Archeraphidia, Pararaphidia) appear to further blend the boundaries of the families Alloraphidiidae and Mesoraphidiidae. The Baissopteridae appears to be more plesiomorphic than the Alloraphidiidae, Mesoraphidiidae, and extant families, assuming that the numerous crossveins and cells are plesiomorphic. The Liassic genera Priscaenigma and Hondelagia, formerly classified in the Baissopteridae, represent a fundamentally different plan of wing venation and one that is plesiomorphic to all other snakeflies. In the forewing of Baissopteridae, Mesoraphidiidae, Alloraphidiidae, Raphidiidae, and Inocelliidae, the subcosta terminates into the costa near the midpoint of the wing. In both Priscaenigma and Hondelagia the subcosta extends to the wing apex and appears to fuse directly with the radial system as in the Megaloptera (hind wings exhibit more typical raphidiopteran venation, e.g., see figures in Willmann, 1994). For this reason, both genera are here placed in a separate family, Priscaenigmatidae (appendix 1), and sister to all other Raphidioptera (fig. 5). The families of Recent and fossil snakeflies are summarized in table 1 while a hypothesis of their relationships and geological distribution is presented in Figure 5.

While the Raphidioptera are undoubtedly ancient, evidence of their occurrence in the Paleozoic has completely eroded (Carpenter, 1967, 1992; Sharov, 1968, 1971; Storozhenko, 1998; Storozhenko and Novokshonov, 1995; Shcherbakov, 1995), leaving the oldest definitive snakeflies as those from the Lower Jurassic. All Paleozoic families of snakeflies (i.e., Fatjanopteridae, Letopalopteridae, Permoraphidiidae, and Sojanoraphidiidae) have recently been removed from the order (although their assignment to Raphidioptera has been suspect for considerable time, e.g., Carpenter, 1967). The Permoraphidiidae, Sojanoraphidiidae, and Letopalopteridae were transferred to the orders Orthoptera (Carpenter, 1967, 1992; Sharov, 1968, 1971), Grylloblattaria (Storozhenko and Novokshonov, 1995), and Protorthoptera (Rasnitsyn in Novokshonov, 1998a, 1998b), respectively. The “ovipositor” of letopalopterids, in particular, was subsequently discovered to be preserved portions of legs, rather than a terminalic structure (Novokshonov, 1998a, 1998b). Lastly, the family Fatjanopteridae was synonymized with the family Ampelipteridae in the Protorthoptera (Shcherbakov, 1995). While I agree that these families are not snakeflies, it must be noted that assignments to orders such as Orthoptera, Grylloblattaria, and Protorthoptera are equally questionable. Particularly dubious are the numerous Paleozoic and Mesozoic winged fossils assigned to Grylloblattaria (today an enigmatic and apterous order); none of these fossils preserve a single diagnostic synapomorphy of Grylloblattaria or even appear to share gross similarities (apomorphic or not) with modern grylloblattids, and thus the conjectures of diversity changes for this lineage (e.g., Storozhenko, 1998; Vršanský et al., 2001) should be considered with serious skepticism. For Paleozoic and Mesozoic fossils the name Grylloblattaria has essentially become a meaningless, receptacle taxon for a polyphyletic assemblage of genera.

Although plesiomorphic with respect to the extant Raphidiidae and Inocelliidae, the earliest snakeflies from the Upper Jurassic already possessed the remarkable synapomorphies for the order. The Raphidioptera, therefore, likely extends well into the Triassic and perhaps even the latest Permian. Although no direct evidence currently supports the conjecture that Raphidioptera is of Paleozoic origin, indirect evidence from the apparent geological ages of related clades (i.e., Neuroptera and Megaloptera) implies that the snakeflies are at least from the Late Triassic and perhaps as old as the Upper Permian (although this latter date is far less likely). Continued paleontological work will undoubtedly refine these estimates as more material becomes available from Triassic deposits. Based on currently available evidence, the snakeflies likely originated and diversified in the earliest Mesozoic after the cataclysmic Permian extinction event.

While modern snakeflies are not found in humid, tropical environments, it is clear that such habitat restrictions have not always been a characteristic of the order. The Cenomanian forests responsible for producing the Burmese amber in which Nanoraphidia occurred were perhaps the most tropical of all the Cretaceous amber localities (Grimaldi et al., 2002). This is also true, although to a much lesser degree, of the amber-producing forests of the Middle Eocene in northern Europe where snakeflies are also known to have occurred (Carpenter, 1956; Engel, 1995; Weitschat und Wichard, 1998). Thus, tropical or subtropical snakeflies persisted at least into the early Tertiary but were perhaps relict and eventually extinguished by the climatic changes marking the Eocene-Oligocene transition. It is also abundantly clear that the order was globally distributed in the past (e.g., species from the Lower Cretaceous of South America) and has undergone significant extinction. The broader distribution and climatic range of extinct snakefly lineages are likely linked. Overall the Cretaceous was much warmer than the Cenozoic (particularly so following precipitous drops in the Paleogene) and climatic changes perhaps drove many Raphidioptera lineages near extinction. As hypothesized by Aspöck (1998, 2000) the decline of the order was perhaps accentuated by the extraterrestrial impact and concomitant climatic disruption at the close of the Cretaceous. Those snakefly lineages not already adapted for colder climates in the early Tertiary would have been quickly extinguished. Extinction left the colder-adapted lineage of the Raphidiidae and Inocelliidae free to diversify in the temperate habitats of the Northern Hemisphere during the latest Paleogene and Neogene, thereby giving us the present distribution of the order.