Phylogenetic Position of the Dioptinae

The Dioptinae belong to the largest superfamily of the Lepidoptera, the Noctuoidea, which is estimated to contain 43,000 described species (Heppner, 1991; Minet, 1991; Solis and Pogue, 1999). The group's precise position within the superfamily has long been contentious. Packard (1895) was among the first to explore the phylogenetic position of the “Dioptidae”, though his conclusions were based entirely on study of the California Oakworm, Phryganidia californica, the only dioptid occurring north of the Mexican border (Franclemont, 1970). He placed the Dioptidae, Cyllopodidae (now Geometridae: Sterrhinae) and Geometridae together in a single evolutionary line, but few agreed with his interpretation. Forbes (1923), in the earliest proposed phylogeny for the superfamily Noctuoidea to treat all major family-level groups, placed the Dioptinae (as Dioptidae) at the base of the entire tree. However, various workers had noted characteristics shared by dioptids and the Notodontidae (Bodine, 1896; Prout, 1910), especially apparent in features of the immature stages (Fracker, 1915; Mosher, 1916). Because the adults bear little superficial resemblance to notodontids (Jordan, 1923b), the Dioptidae was nevertheless retained as a separate family by early taxonomists. As the years passed, accumulating morphological evidence highlighted a close affinity between the Dioptidae and Notodontidae (Köhler, 1930; Franclemont, 1970; Brock, 1971) until finally Minet (1983) relegated the group to tribal status, the Dioptini, subsumed within the subfamily Notodontinae. All of these authors observed that dioptine larvae possess reduced anal prolegs, a derived characteristic exhibited by most notodontids.

Spurred largely by the “Dioptidae/Dioptini” controversy, Miller (1991) published a reassessment of the higher classification of the world Notodontidae. An attempt was made to define monophyletic subfamilies and tribes using a cladistic analysis of larval, adult, and later (Miller, 1992b) pupal characters. Nine subfamilies of the Notodontidae were recognized—one being the Dioptinae—and, notwithstanding certain refinements proposed by Nakamura (2007), that classification has remained largely intact until the present time (Kitching and Rawlins, 1999; Fibiger and Lafontaine, 2005). Miller (1991, 1992b) regarded the Dioptinae to be in a derived position on the family phylogeny (fig. 1).

Figure 1

Position of the Dioptinae among major clades of the Notodontidae (from Miller, 1991).

Dioptine adults, characteristically light bodied and of small to moderate size, span a diversity of wing patterns rivaled by few lepidopteran clades. Early authors were doubtful that dioptines formed a natural grouping (e.g., Seitz, 1925; Köhler, 1930), but no one had examined their morphology in detail. Miller (1991) found strong evidence that the Dioptinae are monophyletic, identifying 19 synapomorphies (more than for any other notodontid subfamily)—seven from larvae and 12 from adults. The current study has done nothing to shed doubt on this hypothesis; evidence for monophyly of the Dioptinae is overwhelming. Miller (1991) also described five synapomorphies suggesting that the neotropical subfamily Nystaleinae is the sister group to the Dioptinae. Three of these are from larval morphology, the most unambiguous being the position of seta L2 on abdominal segment 8 (Character 153 in Miller, 1991). The Dioptinae and Nystaleinae are unique among notodontids in that L2 on segment A8 is located anterior to the spiracle, on a line horizontal with it. The L2 seta on A8 is located directly below the spiracle in other Notodontidae.

From a taxonomic standpoint, the Dioptinae are thus poised to become one of the best-understood neotropical moth subfamilies: Monophyly has been confirmed; the clade's position has been established within the context of a higher classification; and a sister group has been identified. The most significant obstacle prior to revisionary species-level work, definition of monophyletic genera, provided the impetus for this paper.

Taxonomic History

The Dioptinae claim an intriguing, if tortuous, taxonomic history. It was Hering's (1925) opinion that no other group of Lepidoptera so thoroughly confounded the early specialists. The vast majority of species were originally described in the geometrid subfamily Cyllopodinae (now Sterrhinae), but during the course of nearly 200 years of descriptive work, dioptines have also been assigned to the Arctiidae, Pyralidae, Zygaenidae, and even the Psychidae (Hering, 1925; Miller, 1987a). Prout repeatedly encountered dioptines incorrectly identified as Geometridae during his taxonomic research. Realizing that these species needed proper placement, Prout (1918) decided to remedy the situation, uniting them in a single group, the Dioptidae, and presenting what he called a “provisional system of classification”.

The family name Dioptidae, first used by Walker in 1862 (Speidel and Naumann, 2005), has been applied throughout the group's history. Josiidae (Piepers and Snellen, 1900) is the only other available family-group name. Based on structure of the metathoracic hearing organ, Kiriakoff (1950) divided the Dioptidae into the Dioptinae and Josiinae, the latter exhibiting a unique kettledrum tympanum and the former showing a range of tympanal types (see also Sick, 1940). This division was later formalized as the tribes Dioptini and Josiini (Miller and Otero, 1994).

Prout (1918) and Hering (1925) relied on a small set of superficial adult traits for taxonomic assessments—notably wing venation, wing pattern, antennal structure, and labial palpus shape—most of which can be evaluated without the aid of a microscope. Considering this fact, the accuracy and insight reflected in their classifications is astounding. Bryk (1930), the most recent taxonomic arrangement for the Dioptinae and the only existing catalogue, made few modifications from Prout and Hering. By 1930, 40 genera and 396 valid species were recognized.

In the years between 1930 and the present, taxonomic contributions on the Dioptinae have been scarce. Hering (1930, 1943) described two additional species and Schaus (1933) described one. Kiriakoff (1950) erected the genus Euforbesia for a single species (unimacula Warren), and Beutelspacher (1986) described Tithraustes watsoni from Mexico; both of these taxa are synonymized in the present work. My own research to date has not focused on dioptine taxonomy, but rather on building our knowledge of the group's morphology and biology (Miller, 1988, 1996; Miller and Otero, 1994). In previous papers I described only four species and one genus (Miller, 1987a, 1989). In a fascinating discovery, Sattler and Wojtusiak (2000) described the first known brachypterous notodontid, a dioptine in the genus Xenomigia. Finally, Rawlins and Miller (2008) described two new dioptine species from the Dominican Republic, each in its own new genus. This paragraph is a complete summary of taxonomic literature on the Dioptinae over the past 80 years.

Prior to the research here, 574 species-group names had been published for the Dioptinae (421 Dioptini, 153 Josiini). The bulk of those are attributable to relatively few taxonomists. Table 1 shows the name totals for all authors who described more than one taxon. These authors comprise a familiar list of historically important players, several of whom are notorious for their extremely brief, often obtuse, species descriptions. Such work has caused relentless headaches for modern students of Lepidoptera taxonomy. It is an understatement to say that some of the taxonomists working near the turn of the 20th century were describing species at a furious pace—William Schaus of the USNM is said to have described roughly 5000 species of Lepidoptera, while Francis Walker at the BMNH described over 10,000 (V. Becker, personal commun.). For the Dioptinae, Warren, Hering, and Druce lead the pack with over 90 names attributable to each (table 1).

Table 1

Number of dioptine species-group names attributable to authors who described more than one taxon (years during which the names were described are in parentheses) Single taxa were described by seven additional authors.

The history of dioptine taxonomy has a positive aspect, however. Six authors—William Warren, Martin Hering, Herbert Druce, Paul Dognin, Louis Prout, and Francis Walker—account for 462 species group names in the Dioptinae, over 80% of the total. An obvious ramification of the limited number of authors is that the vast majority of dioptine type material can be found at one of three museums (table 2): Hering's types are at the ZMH; those of Dognin (and Schaus) are at the USNM; and the huge remainder—the types of Warren, Druce, Prout, and Walker, well over half of those for the subfamily—are housed at the BMNH. Needless to say, I frequented the halls of those museums during the course of this research.

Table 2

Holdings of adult Dioptinae in the world's major collections The number of primary types listed includes those resulting from the current work.

Acknowledgments

The groundwork for this project was laid during my years as an AMNH curator (1987–1999). The most intensive phase, which involved writing the text and completing the illustrations, was undertaken there as well (2004–2008). It is a pleasure to acknowledge the curators, as well as the entire staff of the AMNH Division of Invertebrate Zoology, who, throughout that time, provided an incredibly stimulating working environment. In particular, it is an honor to acknowledge Dave Grimaldi, Lee Herman, and Toby Schuh. Without their encouragement, this project would not have been completed. During the years 2000–2003, Jim Liebherr and Rick Hoebeke of the Department of Entomology at Cornell University kindly provided work space, as well as access to the Cornell insect collections. For logistical support during my return to New York City in 2004, I will forever be grateful to Lee Herman, Valerie Giles, and Andrew Valelly. Between 1991 and 1996, the research was supported by a National Science Foundation grant (BSR-9106517) to JSM. Expeditions to Ecuador (2004, 2006, 2009) were supported by NSF grants DEB-0717173 and DEB-0849361 to Lee Dyer, and by the University of Nevada, Reno. Robert G. Goelet, chairman emeritus and current trustee of the AMNH graciously funded the years 2006 and 2007.

Special thanks must be extended to Steve Thurston (AMNH), whose remarkable expertise and tireless efforts were crucial during the production phase of this paper. He generated digital renderings of the illustrations and produced the final figures and color plates, suffering through innumerable changes and last-minute modifications. It is no exaggeration to say that without Steve's help, the paper would not have been completed; I feel considerable guilt for not making him a coauthor.

Amy Trabka, formerly associated with Cornell University, executed many of the genitalia illustrations (1993–1996), and Sally Goodman (formerly AMNH) inked a series of adult heads (1996). Becky Rudolph and Emily Griffiths of the AMNH Microscopy and Imaging Facility assisted with scanning electron microscopy. Dave Grimaldi kindly provided access to his Microptics-USA system for photographing adult moths and genitalia slides. Tam Nguyen assisted with setup and troubleshooting. On a visit to the Natural History Museum in London (March 2005), Quentin Wheeler made his Microptics system available for photographing BMNH types. Claudia Aragón (Herédia, Costa Rica) rendered the cover for Part 1, and Steve Thurston (AMNH) painted the cover for Part 2.

It is with extreme gratitude that I thank the scientists who reviewed this paper—Vitor Becker (Serra Bonita, Brazil), Andy Brower (Middle Tennessee State University, Murfreesboro), Phil DeVries (University of New Orleans, LA), Ian Kitching (Natural History Museum, London), Don Lafontaine (Canadian National Collections, Ottawa), Carla Penz (University of New Orleans, LA) and Keith Willmott (McGuire Center, Gainesville, FL). Each offered unique insights, drawing on their vast expertise in neotropical Lepidoptera. The submitted version was greatly improved as a result of their remarkable efforts. Special thanks also go to Rob Voss, Lee Herman and Lorenzo Prendini of the AMNH Publications Committee, whose oversight in guiding the manuscript from submission to publication made the process painless. I am especially grateful to Mary Knight, editor of AMNH publications. Her extremely thorough review greatly improved the paper's readability. Furthermore, her extensive knowledge of Latin, Greek, and English grammar proved to be invaluable.

Throughout the years, I have frequently consulted Gerardo Lamas (Museo de Historia Natural, Lima, Peru), whose remarkable knowledge of 19th-century Lepidoptera collectors and their collecting localities was indispensable. He also aided in finding certain extremely rare, usually ancient, literary citations, frequently providing copies from his immense personal collection.

This study would not have been possible without the generosity of the world museum community. It is especially important to acknowledge Martin Honey of the Natural History Museum, London, and Wolfram Mey of the Zoologisches Museum für Naturkunde, Humboldt-Universität, Berlin. Over the years, I have come to count on their amazing knowledge of their respective collections. In addition to facilitating loans, they fielded innumerable questions regarding type specimens and obscure literature citations. I am deeply indebted to both. Dave Furth, collection manager at the United States National Museum, spearheaded the arrangements necessary for a long-term loan of the entire USNM Dioptinae holdings, including types. Daily access to that material was pivotal for understanding the group's systematics.

I am particularly indebted to the numerous private collectors who loaned, or often donated, valuable Dioptinae for this research. I would especially like to thank Vitor Becker (Serra Bonita, Brazil), Bernard Hermier (Cayenne, French Guiana), Francisco Piñas (Quito, Ecuador), J. Bolling Sullivan (Beaufort, North Carolina), Paul Thiaucourt (Paris, France) and Rafael Turrent (Mexico City, Mexico), whose collections were the source of species newly described here.

The following individuals and institutions kindly loaned material for study. Without their generous assistance in providing access to the collections in their care, it would have been impossible to investigate dioptine biodiversity. The abbreviations listed below are utilized throughout this work:

Fieldwork forms the backbone of this project. The following people provided valuable aid, either by accompanying me on expeditions in search of Dioptinae, or by collecting dioptines on their own: from the USA, Lee Herman, Cal Snyder, Andy Brower, Suzanne Rab Green, and Valery Giles; in Costa Rica, Jorge Corrales, Bernardo Espinosa, Isidro Chacón, and Antonio Azofeifa (INBio); in Ecuador, Giovanni Onore, Elicio Tapia, and Mario Tapia; in French Guiana, Bernard Hermier and Frederic Beneluz; in Panama, Annette Aiello and Don Windsor (STRI); in Peru, Gerardo Lamas, Juan Grados, Juan José Ramírez, and Angel Asenjo (MUSM); in Venezuela, Daniel Otero (Universidad de Los Andes). For tireless efforts documenting the natural history of neotropical Lepidoptera, and for generously sharing their findings on the Dioptinae, I would like to thank the following: Dan Janzen, Winnie Hallwachs, and their remarkable field teams at Area de Conservación de Guanacaste (Costa Rica); Harold Greeney, Lee Dyer, Grant Gentry, and the group at Yanayacu Biological Station (Ecuador); Annette Aiello (STRI, Panama); Daniel Otero and Andres Orellana (Mérida, Venezuela). I also gratefully acknowledge the following individuals who donated Dioptinae, collected during their own expeditions, to the AMNH for study: Phil DeVries (University of New Orleans); Dave Wagner (University of Connecticut, Storrs); Tim McCabe (NY State Museum); Gunnar Brehm (Institut für Spezielle Zoologie und Evolutionsbiologie, Friedrich-Schiller-Universität Jena); and Bob Robbins (Smithsonian Institution).

The following agencies issued collecting and export permits, allowing me to collect Dioptinae in their countries: In Panama, STRI and the Instituto de Recursos Naturales Renovables; in Venezuela, the Instituto Nacional de Parques; in Costa Rica, the Instituto Nacional de Biodiversidad and the Ministerio del Ambiente y Energía; in Quito Ecuador, the Ministerio del Ambiente and the Museo Ecuatoriano de Ciencias Naturales; in Lima Peru, the Museo de Historia Natural Universidad Nacional Mayor de San Marcos and the Instituto de Recursos Naturales.

Kurt Picket (University of Vermont), Kip Will (University of California, Berkeley) and Jim Carpenter (AMNH) assisted with cladistic analyses. Andy Brower (Middle Tennessee State University) was instrumental in providing analytical expertise throughout this project, and kindly calculated Bremer support values. For general discussions concerning character coding and phylogenetic methodology, I thank Lee Herman, Kurt Picket, Jim Liebherr, Toby Schuh, and Dave Grimaldi. I thank Andy Warren for providing references on the corethrogyne in skippers, and for facilitating a loan of material from the Alonso and Rafael Turrent Collection in Mexico City.

Methods

The form of this publication takes its inspiration from the landmark monograph by Pedro Wygodzinsky (1966) on the subfamily Emesinae (Hemiptera: Reduviidae), in which he reclassified the world fauna. My project pales in comparison with his; Wygodzinsky monographed nearly 1000 species in 92 genera, twice the size of the fauna treated here. Nevertheless, our goals were the same: to revise the genera and provide species-level taxonomic coverage for a large insect clade, along with identification keys, comprehensive documentation of adult morphology, and geographical distributions.

Pinned adult specimens, assembled through museum loans and from my own fieldwork, provided the raw materials for this study. These were the source for morphological character data, which were then utilized in phylogenetic analyses. A cladistic analysis of all species in the Dioptinae would not have been practical. I instead chose the exemplar approach, a method that has delivered positive results in the past (e.g., Miller, 1987b, 1991, 1996; Miller, Brower, and DeSalle, 1997). The rationale for utilizing a species subset was driven largely by the inordinate rarity of certain taxa. Many are known solely from the type. Likewise, some genera, especially monobasic ones, have been described either from a tiny series of specimens, or from a single specimen. Often, females are unknown. These cases presented two problems. First, when two or fewer specimens represent the sum of our knowledge of a taxon, I was loath to sacrifice one of those by performing a whole-body dissection, necessary for scoring all traits. Secondly, females provide important characters for these analyses; if female specimens were unavailable, that taxon was not included as an exemplar in my cladistic analyses.

In order for the exemplar approach to be successful, it was crucial that the species chosen reflect all or most of the morphological variation that occurs across the subfamily. My approach was to familiarize myself, as completely as possible, with species-level taxonomy and morphology for the entire Dioptinae. In conjunction with this goal, I attempted to examine type material for each of the 574 previously published species-group names. During the course of visits to the museums in London, Berlin, and Washington, I photographed and examined nearly all extant types for the Dioptinae.

Once this groundwork of knowledge had been established, I was able to select exemplar species from a position of strength. My search for characters from skeletal anatomy was exhaustive. Cladistic hypotheses can always be improved upon through the addition of novel characters and more taxa, but given the data set at hand, I place considerable confidence in the results described in these pages. The new generic classification, while not perfect, is a vast improvement over what existed before.

Museums

Dioptinae are poorly represented in collections worldwide (table 2). I was able to locate only 14 collections that contained more than 150 specimens. Interestingly, relatively recent biodiversity sampling programs, such as those at INBio, the LACM, and the CMNH, have produced dramatic results, placing those institutions near the forefront of tropical Lepidoptera collections. Growth of Dioptinae holdings at the AMNH reflects nearly 20 years of focused collecting. When I first arrived there in 1987, the AMNH housed approximately 1000 specimens, whereas it now ranks first in the world, with roughly 3700 specimens. In contrast, the holdings at the BMNH are large, but have grown little over the past 100 years.

The initial phase of this research was carried out during a Smithsonian postdoctoral fellowship in 1986. During that year, I visited major museums, borrowing representative specimens of Dioptinae. Collections studied included those of the AMNH, BMNH, CMNH, CUIC, ZMH, and of course the USNM, the site of my postdoctoral research. Whole-body and genital dissections (males and females) provided an initial morphological survey, and served to highlight problem areas in the generic classification.

A seven-year curatorial fellowship (1987–1994) at the AMNH followed. I continued amassing pinned specimens through loans, and began my own fieldwork on the Dioptinae. During the course of field trips, I studied the following collections: BHC, FPC, INBio, IZA, JCC, MUSM, PUCE, and STRI. Additional museum visits were made to: the CNC, EMEC, MCZ, MNHN, PTC, and the PMNH.

Another phase of loans further increased species representation in my study sample. The following graciously sent me all or most of the Dioptinae in their holdings: ARTC, CAS, CCC, CMNH, EMEC, FML, FNHM, JBSC, LACM, MPM, NMW, SDNH, VOB, and the ZMC.

In an incredibly gracious and helpful gesture, the USNM loaned the entirety of their substantial dioptine holdings, including 92 types, on a long-term basis. This provided daily access to the remarkable historical collections of Dognin and Schaus.

Fieldwork

Dioptines are extremely difficult to collect. They are usually rare, and are always elusive. Members of a few genera, such as Xenomigia, Scotura, and Monocreaga, can be captured at light traps. These taxa thus appear more frequently in museum collections. However, the vast majority of dioptine species are largely or completely diurnal. Many seem to show two active periods each day, flying in late morning, and then again between 4 p.m. and 6 p.m. before sunset. Understory and cloud forest species are active in brief bursts throughout the day. Nevertheless, I have walked particular trails, where dioptines were known to occur, for days without seeing a single individual. A few taxa are locally common. For example, Dolophrosyne coniades (Druce) and Nebulosa yanayacu, sp. nov., fly in abundance at Yanayacu Biological Station in eastern Ecuador. Nonetheless, these two taxa rarely appear in museum or private collections, apparently because butterfly and moth collectors have consistently overlooked them in the field.

My quest to learn more about the biology of these fascinating moths, and to accumulate a strong representation of dioptine specimens for study, has taken me to some of the most beautiful locations on earth. During the course of this project, I made 20 field trips to six different Central and South American countries, searching for dioptine adults and immatures. These expeditions were aimed at visiting areas of high species diversity, based on examination of pre-existing specimen label data. I learned to undertake each trip with a minimum of expectations, and although I was frequently dismayed at the difficulty of finding target taxa, every expedition produced remarkable surprises. These collecting efforts helped to more than triple the number of dioptine specimens housed at the AMNH (table 2), providing a strong species sample for comparative morphological study.

Exemplars

After nearly two decades of accumulated insight regarding dioptine morphology and taxonomy, 115 exemplar species (table 3) were chosen for detailed study—in the species treatments, exemplars are designated by “[EX]” following their name. Cladistic analysis of relationships among this species subset provides the framework for a revised classification of the subfamily. At the outset, I was relatively confident concerning the monophyly of certain pre-existing genera. For example, Erbessa shows a suite of clear-cut morphological apomorphies, obvious even to the early taxonomists. For genera such as this, relatively few exemplars were chosen. On the other hand, my morphological survey suggested that Tithraustes was polyphyletic. In such cases, a much broader sample of representatives was included in the cladistic analyses. Overall, I am confident that the list of exemplar species represents a comprehensive picture of morphological diversity for the Dioptinae.

Table 3

Exemplar species used in the cladistic analyses For previously described species, generic assignments follow earlier classifications (Bryk, 1930; Kiriakoff, 1950; Miller, 1989). Provenance refers to the specimen from which the whole-body dissection was made.

Preparation of specimens for morphological study follows previously described methods (e.g., Miller, 1996). Male and female genital dissections, stained with chlorazol black and mounted in Canada balsam, were made for as many species as possible. In most cases, specimens from more than one locality were dissected. Each slide preparation was given a “JSM” slide number, and these are cited in the individual species treatments. For each exemplar taxon, a whole-body dissection was made by removing the wings and dissolving the body in 10% KOH. Wings were retained with the pin and labels, whereas the bodies were placed in individual vials of 70% ethanol. In certain cases, the right fore- and hind wings, as well as appendages, were stained using eosin Y and mounted in balsam. Altogether, more than 2000 preparations were made during the course of this study—over 1800 genitalia and wing slides, as well as approximately 200 bodies in alcohol.

Morphology

Completion of this study would not have been possible were it not for the fact that adult Dioptinae provide a rich source of morphological characters. Use of pinned museum material allowed for study across the taxonomic spectrum, even for species that had been captured well before the turn of the 20th century but have not been seen since. Interestingly, the Dioptinae seem to offer a broader range of structural variation than most other Lepidoptera groups with which I am familiar. Comparatively speaking, many subfamilies of similar size are morphologically homogeneous. It is not clear why dioptines exhibit such remarkable structural diversity. Perhaps the evolution of a lifestyle novel for the Notodontidae has in turn given rise to a rapid burst of anatomical specializations. Whatever the underlying reason, I am thankful for small favors.

Larval morphology has played an important role throughout the history of notodontid classification, starting with Packard (1895). In previous family-level studies on the Notodontidae, caterpillars have been shown to hold more deep-level phylogenetic information than adults (Miller, 1991). However, I relied exclusively on adult morphology for the cladistic analyses described here. This was of necessity. Tables 4 and 6 summarize our current knowledge of dioptine immatures and host plant relationships. To date, immature stages are known for only 17% of the fauna, and the larvae of certain crucial taxa may never be discovered. However, earlier comparisons showed that adult characters alone are extremely effective for unraveling phylogeny within the Dioptinae (Miller, 1996; Miller, Brower, and DeSalle, 1997).

Table 4

Known host plants of the Dioptini Genera are assigned to plant families according to Soltis et al. (2005) and the Missouri Botanical Garden's angiosperm phylogeny website (Stevens, 2006). For a host-plant database of world Lepidoptera, visit  http://www.nhm.ac.uk/entomology/hostplants/ (Robinson et al., 2007).

My search for informative adult morphological traits was exhaustive. Appendix 1 lists virtually every skeletal character that showed structural variation for the Dioptinae. The 305 characters are described by a total of 938 character states. Difficult character complexes, such as those where character state assignments were somewhat ambiguous, were nevertheless incorporated into the analysis. My rationale was that even these potentially provide useful information regarding common ancestry.

Along with standard references on Lepidoptera morphology (e.g., Common, 1990; Nielsen and Common, 1991; Scoble, 1992; Kristensen, 2003), the important papers of Ehrlich (1958a, 1958b) and Oseto and Helms (1976) proved useful. The latter was especially insightful for understanding sclerites of the head and thorax. Additional discussions focusing on dioptine morphology can be found in my own published work (Miller, 1987a, 1988, 1989, 1992b, 1996; Miller and Otero, 1994). Terminology for microsculpture follows Eady (1968).

Cladistic Analyses

The data matrix (appendix 3) comprises 115 ingroup species and three outgroup species, each scored for 305 morphological characters. Of these characters, 118 are binary while 187 are described by three or more states. Multistates were treated as either unordered (91 characters) or ordered (96 characters); following each character description in appendix 1, these are noted as either [−] for unordered, or [+] for ordered. More complex multistate treatments, for example using character state tree coding, were not employed.

Throughout its construction and development, the matrix was edited using WinClada, Version 1.00.08 (Nixon, 2002). Phylogenetic analyses were performed with a single, straightforward goal in mind—a search for the shortest tree—since this best summarizes available data (e.g., see Wenzel, 1997). Parsimony analyses employed Tree Analysis Using New Technology, Version 1.0, or TNT (Goloboff, Farris and Nixon, 2006). This software combination afforded flexibility for manipulating the large data matrix, and offered powerful tree-searching capabilities. To root the dioptine cladogram, three notodontid species were selected from the Nystaelinae, the sister group of the Dioptinae (Miller, 1991), and scored for all characters. Experimentation utilizing up to five nystaleine taxa produced the same ingroup tree topology, as well as identical character state optimizations.

For tree searches using TNT, the following were specified: First, the settings were modified to allow 100,000 trees to be retained in memory, and General RAM was increased to 100 MB. For the analyses themselves, Random Seed was set to 0; 200 random addition sequences were employed; and the algorithms Sectorial Search, The Ratchet, and Drift and Tree Fusing were utilized simultaneously. Characters were equally weighted. The full analysis took 3 minutes and 2 seconds, and examined over 2.6 billion tree rearrangements. A single shortest tree was found, with a length of 3077 steps, a consistency index of 0.19, and a retention index of 0.71. This tree (fig. 2) forms the basis for the new classification of the Dioptinae (appendix 2). Bremer support (BS) values (Bremer, 1988; Bremer, 1994) were chosen to estimate the robustness of clade support. These were estimated by heuristic anticonstraint searches with 10 random addition sequences (see Brower, 2006). Figure 2 shows BS values for all branches of the dioptine cladogram. Figures 3 and 283 summarize generic assignments for the exemplar species, while the implied cladogram of relationships among dioptine genera is shown in Figure 7.

Descriptive Taxonomy

The primary focus here is to present a revised, and—it is hoped—stable, dioptine classification. However, during the course of these studies, numerous undescribed species were encountered. Sixty-four species of Dioptinae are described here as new. This is a relatively small percentage of the undescribed material residing in collections, and it is only a fraction of the new taxa that will ultimately be recognized. In some genera, such as Xenomigia and Nebulosa, I argue that the described species represent less than a third of those that occur in nature (see Discussion: Biodiversity).

The species described were selected using somewhat arbitrary criteria. A few exhibit novel morphology. The majority were chosen because a sizeable series, usually recently collected, was available for study. Two countries, Ecuador and Costa Rica, are particularly well represented in the list of newly described taxa. Modern collectors have traveled to those nations frequently over the past two decades, and I have visited them on numerous occasions myself. The high number of species described from these two, relatively small neotropical countries, is illustrative. Thorough collecting in larger, ecologically and topologically complex Central and South American countries, such as Guatemala, Nicaragua, Venezuela, Colombia and Peru, will undoubtedly reveal an incredible wealth of undiscovered biodiversity for the Dioptinae (see Discussion: Biodiversity). Choices of species epithets were aided by referring to Brown (1956), with subsequent input from Mary Knight (AMNH). Generic names were checked for availability by searching the online version of Neave (1939) and by searching Google ( http://www.google.com).

Maps showing geographical distributions are provided only for newly described species. Ultimately, it will be important to delimit distributions throughout the Dioptinae. Species distributions for Costa Rica, Ecuador, and Peru are shown using detailed relief maps (figs. 4–Figure 56), while those for species described from other countries are illustrated with simple line maps.

Figure 4

Costa Rica, showing major cities.

Figure 5

Ecuador, showing major cities and river systems.

Figure 6

Peru, showing major cities and river systems.

Figures

Another important focus of this paper is to thoroughly document dioptine morphological diversity. Representative heads, wings, and genitalia are figured for each genus, and usually for each species group where those are defined. When deemed important, variation in particular structures—such as thoracic sclerites, antennae, and legs—is shown. Traditional line drawings are combined with digital photography and scanning electron microscopy (SEM).

Genitalia photographs were taken using a Microptics-USA photomicrographic apparatus, equipped with Infinity K2 optics and a Nikon D1X digital camera. One of three Infinity Photo Optical lenses (CF-1, CF-2, CF-3), each of different magnification, was selected based on the size of the moth's genitalia. Specimens for SEM were sputter coated with gold-palladium and photographed using a Hitachi S-4700 scanning electron microscope.

Figures follow certain conventions throughout. The method for preparing genitalia mounts was to make a slit along the abdomen's left side. Genitalia were then removed and the remaining abdomen was mounted flat, resulting in a dorsal view of Tg8 and a ventral view of St8; whenever these are illustrated, the distal margin is always positioned at the top. In all genital drawings and photographs, the male is shown in posterior view, whereas female genitalia and the aedeagus are illustrated in lateral view. Additional structures, such as heads and thoraces, are also shown in lateral view. Whenever lateral views are shown, anterior is always at the left.

A crucial aspect of Hering's (1925) paper was that he presented renderings for the adults of 271 dioptine species on four, full-color plates. Those illustrations are relatively accurate, and provide important aids for species identification; I have relied on them heavily over the years. However, my goal was to take Hering's contribution a step further. Therefore, color plates depicting an adult specimen representing every described species in the Dioptinae, are included in this paper (pls. 1–35). Locality data for each adult figured are listed in appendix 5. Adult specimens were photographed using the same Microptics system employed for genitalia photographs, except that in this case the lens was a Nikon 60 mm AF Micro Nikkor.

Head

Labial palpi have been used as diagnostic features almost since the beginning of Lepidoptera classification (see Kristensen, 2003). Their early usage in systematics is probably due to the fact that they can be studied at low magnification, sometimes even without the aid of a microscope. Palpus development is highly variable in the Notodontidae; for nearly 200 years, differences and similarities have been providing notodontid taxonomists with useful characters at the subfamilial, tribal, and generic levels (e.g., Stephens, 1829; Packard 1895; Marumo, 1920; Forbes, 1948; Weller, 1992). As in most other Lepidoptera, three segments are typical for notodontids, a notable exception being some members of the subfamily Notodontinae, which exhibit derived reduction to two (Miller, 1991). An analogous situation occurs within the Zygaenoidea, most of whose taxa exhibit three labial palpus segments while a few possess only two (Epstein, 1996). All Dioptinae exhibit three visible segments, but otherwise the diversity of palpus morphology almost matches that observed across the entire Notodontidae. These structures figure heavily in descriptive taxonomy for the Dioptinae (e.g., Warren, 1897), and played a prominent role in the early generic classifications (Prout, 1918; Hering, 1925). Labial palpus morphology provides an extremely useful source of diagnostic characters for dioptine genera.

After considerable trial and error, labial palpus morphology is described using 11 characters (totaling 38 states). These treat overall length, relative length and shape of each palpus segment, vestiture specializations, and scale color. Erbessa, Phaeochlaena, and some Polypoetes exhibit the most striking palpus configurations (figs. 35; 36A; 70A, B; 89B, E; 90A, D; 92C; 94C, D). Here, the extremely long palpi are folded elbowlike over the front. There is an unusual hinged joint between Lp1 and Lp2, with the palpi extending up to the antennal bases, and often well beyond. In these, a portion of Lp2 is usually denuded of scales on its mesal surface (figs. 36C, D; 71B; 82E; 93A; 94D). Among notodontids, this structural type is unique to the Dioptinae, but according to the phylogenetic results described here, it has apparently evolved at least three times within the subfamily. Two clades in the tribe Dioptini exhibit such palpi—one containing Erbessa and Oricia (fig. 31A, E), and the other including Phaeochlaena, Argentala (fig. 84A, D), Pikroprion (fig. 79A, E), and most Polypoetes. Additionally, elongate elbowed palpi appear as a derived trait within the Josiini; those of Phintia (fig. 328A, D) are as long as the ones found in Dioptini. There is sexual dimorphism in length for elbowed-palpus taxa, with palpi of males noticeably longer. The longest palpi of any dioptine occur in Erbessa leechi (Prout) (fig. 35A, D).

When the labial palpus segments are examined separately, Lp1 shows the fewest shape differences, whereas Lp2 exhibits great variation in length, thickness, and degree of curvature. The third segment is frequently short with an acute apex, but morphological differences among groups are extensive. In some species, Lp3 is extremely short (e.g., Cacolyces plagifera; fig. 135B), while in others, such as Tithraustes (fig. 268A, E) and Isostyla (fig. 276A, D), it is greatly elongate. Nystaleinae typically exhibit bullet-shaped or elongate Lp3 types (Miller, 1991; Weller, 1992).

The organ of vom Rath is a sensory pore found at the terminus of Lp3 (vom Rath, 1887; Scoble, 1992). The pore encloses elongate sensilla, which are sensitive to CO₂ levels (Bogner et al., 1986) and various odors (Lee et al., 1985). However, precise answers concerning the organ's biological function have eluded discovery. The organ of vom Rath is a putative autapomorphy for the Lepidoptera (Kristensen, 1984; Kristensen and Skalski, 1999; Kristensen, 2003; Kristensen et al., 2007). In Dioptinae, it is either located apically (e.g., Phanoptis; fig. 64E, G) or on the anterior surface of Lp3 (e.g., Polypoetes nubilosa; fig. 90A). The pore also varies in size and depth; it is shallow and barely visible in Phelina Group species of Dioptis, as well as in D. candelaria (fig. 187G).

A character found in some Dioptinae, and unique to the group as far as I can determine, is the presence of robust spinules located near the labial palpus apex (Character 10). These structures, apparently modified scales, are usually yellowish brown in color, contrasting with the other palpus scales. Presence of palpus spinules in Phaeochlaena, Argentala, Pikroprion, and Polypoetes is one of the synapomorphies uniting these four genera (Clade 6; fig. 7). The spinules occur exclusively on Lp3 in some groups (e.g., Rubribasis Group of Polypoetes; figs. 91D–F), while in others, such as Pikroprion (figs. 79A, 83C–F) and Argentala (fig. 82E, F) they occur distally on Lp2, as well as on Lp3. These structures can be observed with the aid of a dissecting microscope, under the apical scales in pinned specimens, but are most readily seen in material that has been cleared with KOH. The palpus spinules of Argentala subcoerulea are long and thin (figs. 83A, B), whereas those of Polypoetes nubilosa are short and thick (figs. 94E, 94F, 95A). The apex of Lp3 in Erbessa species is thornlike (figs. 35B–E, 36C, 36D), a synapomorphy for that genus. As with most morphological specializations of the Dioptinae, none of these structures have a known function.

Figure 7

Summary cladogram showing relationships among dioptine genera implied from the analysis of 115 exemplar species (figs. 2, 3, 283).

Scale color on the labial palpus provides characters useful for separating species (Miller, 1989), and has proved useful at higher taxonomic levels as well. In many Dioptinae, Lp1 and the base of Lp2 are yellow. Less commonly these are white. Such differences are utilized extensively for the species descriptions in the present paper. Presence of yellow or white scales is a derived trait, the plesiomorphic condition being completely dark palpi. This character is useful for defining clades above the genus level. Differences in the distribution of light-colored scales on Lp1 and Lp2 can be used to separate closely related species.

Most Josiini do not possess scale tufts on the apex of the labial palpus, but in some genera of the Dioptini tufts are highly developed. The scale tufts of Dioptini do not compare with the spectacular palpal androconia found in herminiine Noctuidae (Smith, 1895; Forbes, 1954), but the structures are similar in that they are associated with males. Within Dioptini, the most elaborate tufts on the labial palpus can be seen in males of the Frigida Group of Euchontha, where the terminus of Lp2 bears a wide, fanlike tuft of elongate scales that obscures Lp1 (Miller, 1989). Interestingly, a single josiine—Getta unicolor (fig. 289A)—exhibits analogous conspicuous tufts on its palpi. This modification is strangely absent in other Getta species.

Members of the suborder Glossata, which includes the vast majority of Lepidoptera (Kristensen, 1984), exhibit the haustellate condition in which the paired galeae of the maxilla form an elongate sucking tube, or proboscis. Structure and function of the proboscis are outlined in Scoble (1992) and Kristensen (2003). Our basic understanding of this remarkable organ was laid down in important earlier papers (Börner, 1939; Eastham and Eassa, 1955), and Krenn (1990) detailed its functional morphology.

Proboscis development varies widely in the Notodontidae. Across subfamilies, it can be long, short, or even absent (Miller, 1991; Weller, 1992). Nearly all Dioptinae possess a long tongue, the sole exception being a species in the rare and obscure genus Xenormicola (pl. 22); the proboscis of X. extensa is short (fig. 232B). In the only other described member of the genus, X. prouti, the tongue is long.

Proboscis coloration, a trait easily observed in pinned specimens, provided useful phylogenetic information. In most dioptine genera, and in the outgroup taxa employed for this study, the proboscis is dark brown to blackish brown. On the other hand, some dioptines exhibit a golden-yellow proboscis (Character 14). Proboscis coloration seems to be consistent within genera, but the derived color, a yellow tongue, is found in groups scattered across the dioptine phylogeny, including Oricia, Scotura, Phanoptis, and Xenomigia. In addition, six species formerly placed in Tithraustes are here referred to the new genus Chrysoglossa, so-called because they possess a yellow proboscis. In many Geometridae, the tongue is bright lemon yellow (personal obs.), thus providing a useful field character for separating day flying members of that family from most Dioptinae.

Study of the lepidopteran proboscis using scanning electron microscopy has revealed a range of fascinating characters, many of which provide useful phylogenetic information (e.g., Kitching, 1987, 1988; Ryabov, 1988; Krenn and Kristensen, 2000). Speidel et al. (1996a) recognized two distinct regions of the proboscis: in the proximal part, chaetiform sensilla are present, whereas the distal part is densely covered with styloconic sensilla. These authors found considerable diversity in proboscis surface structure within the Noctuidae for both regions, especially among species of the subfamily Catocalinae. Speidel et al. (1996a) characterized six surface structure types, and showed that some of these define monophyletic entities within the Noctuidae. Miller (1991) documented the utility of similar characters for the Notodontidae. Dioptinae exhibit Type II surface sculpturing: each ring of the proboscis bears a series of longitudinal ridges, so aligned that they alternate with the ridges of adjacent rings in an interlocking configuration (figs. 91B, 91C, 312B). Type II proboscis sculpturing has been found in only two other notodontid subfamilies—the Nystaleinae and Heterocampinae (Miller, 1991).

In Lepidoptera, the distal portion of the proboscis bears a set of large sensilla styloconica (fig. 312B). Typically, each sensillum is columnar, with five or six lateral flutes and a terminal peg (figs. 92F, 216B, 312C). In a remarkable early paper, Guyénot (1912) discussed these proboscis sensilla in detail. Using 70 species representing a dozen moth and butterfly families, he detailed fascinating structural variation among the groups. It is now clear that fluted proboscis sensilla are found almost universally in the Lepidoptera (e.g., see MacIndoo, 1917, 1929; Börner, 1939; Ikeuchi, 1962; Goldware and Barnes, 1973; Städler et al., 1974; Sellier, 1975; Dey et al., 1999; Hallberg et al., 2003). For the Noctuoidea, they have been documented in the Arctiidae (Altner and Altner, 1986), Notodontidae (Miller, 1991), and Noctuidae (Callahan, 1969; Flower and Helson, 1971; Blaney and Simmonds, 1988).

Proboscis sensilla have been a subject of taxonomic interest because morphological differences reflect phylogenetic history in certain groups (Scoble, 1992). Speidel and Naumann (1995) emphasized the utility of proboscis ultrastructure as a source of characters for phylogenetic classification of the Noctuidae. They further summarized the types of sensilla styloconica found across noctuid subfamilies, and provided scanning electron micrographs of representative taxa (Speidel and Naumann, 1995; Speidel et al., 1996a). Of particular evolutionary interest in the Noctuidae are the remarkable modifications of the sensilla and proboscis tip in fruit-piercing and blood-feeding taxa (Darwin, 1876; Wu and Chou, 1985; Bänziger, 1989; Scoble, 1992; Büttiker et al., 1996; Kitching and Rawlins, 1999; Kristensen, 2003).

The sensilla styloconica of the proboscis show a great deal of morphological variation across clades of the Notodontidae (Miller, 1991). Fluted sensilla are the most widely distributed type, but in tear-drinking species (Bänziger, 1988) belonging to the subfamily Dudusinae, the sensilla are cylindrical and smooth (Miller, 1991). Similar smooth sensilla characterize the nymphalid tribe Argynnini (Guyénot, 1912). For this paper, I performed a brief survey of dioptine proboscis sensilla searching for differences, but all the species examined exhibit the typical fluted type; no variation was found.

Other mouthpart structures show minimal variation in the Dioptinae. Pilifers, brushlike lobes arising laterally from the labrum, occur throughout the Heteroneura (Krenn and Kristensen, 2000). They are thought to function in registering the position of the proboscis during feeding (Kristensen, 2003). These structures are absent in some notodontids with a reduced proboscis (Miller, 1991), but are present in all Dioptinae. The pilifers are especially long and narrow in Oricia and some Scotura species (Character 12; figs. 10E, 31A, 31B, 31D). The maxillary palpi of ditrysian Lepidoptera are usually small (Scoble, 1992). For Dioptinae, I scored relatively subtle differences in size and shape (Character 13).

The configuration of the frontal scales provides important phylogenetic information for the Noctuidae (Fibiger and Lafontaine, 2005). Both Prout (1918) and Hering (1925) noted differences in frontal scaling in their generic diagnoses for the Dioptinae. I expanded their interpretations, using three characters (Characters 15–17) with a total of 10 states to describe this variation. Scale configurations range from the downwardly oriented scales of many Josiini (fig. 312A), also typical of Nystaleinae, to the more horizontal arrangement of Dioptis (fig. 188A). In Euchontha and others (Miller, 1989), the frontal scales form an unusual bowl-like structure. In a great many taxa within the Dioptini, such as members of Polypoetes, Nebulosa, and Phaeochlaena, the scales on the lateral regions of the front point upward, forming two elongate tufts that converge near the antennal bases (figs. 91A, 92D, 160B). Color patterns of the front were not used in this study because, though diagnostic for species, they are too variable to enable characterization of genera. On the other hand, the distribution of yellow scales on the vertex and ventral region of the head (Characters 18, 19) provided useful generic traits.

Eye size is apparently correlated with diurnal versus nocturnal behavior in Lepidoptera (Bourgogne, 1951; Powell, 1973), with smaller eyes being typical of day-flying species. Considerable variation in size can occur within clades, as has been shown for the Arctiidae (Ferguson, 1985) and Palaephatidae (Davis, 1986). Day-flying has evolved by convergence many times in the Noctuidae; concordant reduction of the eye has been noted in subclades of the Catocalinae (Speidel and Naumann, 1995), Plusiinae (Kitching, 1987) and Heliothinae (Hardwick, 1970; Matthews, 1987, 1991). Eye size shows considerable variation across subfamilies of the Notodontidae (Miller, 1991; Weller, 1992).

Although various eye-size indices have been developed for Lepidoptera (Powell, 1973; Ferguson, 1985; Davis, 1998b; Kristensen, 2003), I employed a single character with three simple states (Character 20), based on relative width of the postgena, to describe this feature. Variation in eye size within the Dioptinae has been noted in previous works on the group (Prout, 1918; Miller, 1989, 1996). In dioptines where the eye is small, a scaleless, densely spiculate band completely surrounds it (fig. 160A). Based on outgroup comparison, the plesiomorphic state for Dioptinae is large eyes, but reduction has apparently occurred numerous times within the group. The smallest eyes (State 2) are found in Dolophrosyne (fig. 221A–D) and Scoturopsis (fig. 227A, B), as well as certain species of Nebulosa (fig. 159E), Dioptis (187B, E), and Scea. Altogether, these taxa are widely scattered across the dioptine cladogram (fig. 7), suggesting multiple origin of small eyes. Eyes of moderate size (State 1) are observed in numerous groups. Interestingly, certain genera, such as Polypoetes, exhibit variation in eye size, ranging from small eyes in the Rubribasis Group (figs. 89A, C) to large bulging eyes in the Haruspex and Etearchus groups (figs. 89B, 89D, 90A–F). Even closely related species can sometimes vary, as is exemplified by the different eye sizes of Polypoetes approximans and P. subcandidata.

One feature of eye morphology, so-called “hairy eyes”, has been utilized almost throughout the history of noctuoid systematics. For example, present of hairy eyes has been used to define the controversial group Pantheinae (Forbes, 1954; see also Fibiger and Lafontaine, 2005; Mitchell et al., 2005). In Dioptinae, many genera exhibit short microsetae (Character 21), sparsely interspersed among the eye facets (figs. 92E, 188B). This character was difficult to score because the setae are often difficult to see with a dissecting microscope. They are fairly prominent in species belonging to the Rubribasis Group of Polypoetes. Interfacetal setae occur widely across the Dioptini, but appear to be absent in Josiini.

The presence of paired ocelli in adults, located behind the antennae, is apparently part of the lepidopteran groundplan, but they have been lost numerous times within the order (Kristensen, 2003). Presence or absence has been used widely in reference to higher-level relationships among Lepidoptera. For example, it is used to separate subfamilies in the Drepanoidea and Geometroidea (Minet and Scoble, 1999), and figures in the classification of the Zygaenidae (Yen et al., 2005a). Within Noctuoidea, absence of ocelli has been used to define the subfamily Lithosiinae (Arctiidae). However, careful examination shows that they can be present, but extremely small within this group (Kitching and Rawlins, 1999). Ocelli are relatively uncommon in adult Notodontidae, and are absent in all Dioptinae (Miller, 1991).

For nomenclature describing sclerites of the head, I follow Matsuda (1965) and Oseto and Helms (1976). The latter, which provides a detailed morphological analysis for adults of Loxagrotis albicosta (Smith) (Noctuidae), is an extremely useful general reference. Two regions of the head show clear-cut intergeneric differences in the Dioptinae. The gena, a sclerite below the eye, can have a smooth ventral margin or it can be flanged, as in Erbessa and others (Character 22; figs. 35A, 36B). In some genera there is an anterior notch in the gena (Character 23). The submentum in Lepidoptera is the ventral plate of the head to which the labial palpi attach (Snodgrass, 1935). Along its posterior margin, the submentum curves upward to form a secondary sclerite, called the hypostomal bridge (Oseto and Helms, 1976) or gula (Bourgogne, 1951; Matsuda, 1965). The hypostomal bridge, best seen in posterior view, thus forms a sclerotized band along the ventral margin of the occipital foramen. Dioptinae show considerable variation in the size and width of the hypostomal bridge (Character 23). For example, in Xenomigia (fig. 235D) and Brachyglene (143F) it forms a narrow band, while in Erbessa (fig. 35G) and Polypoetes (fig. 90C, 90F) it is wide and robust. Width of the hypostomal bridge may be correlated with labial palpus length; the species with a robust sclerite are roughly the same ones exhibiting long, elbowed palpi (Characters 4 and 5).

Antennae are the primary sensory structures of Lepidoptera (Schneider, 1984), and accordingly they are covered with numerous types of sensilla along their ventral surface. Three morphological regions of the antenna are commonly recognized—the scape, the pedicel, and the flagellum—the latter composed of numerous annulations or flagellomeres (Bourgogne, 1951; Scoble, 1992). Jordan (1898) detailed the comparative antennal morphology of Lepidoptera in a remarkable early paper, and their structure has played a pivotal role in the group's systematics since that time (e.g., Hampson, 1898). Considering that antennal function must be universal, the amount of structural variation observable within certain clades is remarkable.

Scanning electron micrographs of dioptine antennae (Miller, 1987a, 1989, 1996) reveal not only a complexity of sensillum types and distribution, but also vast differences in overall shape. Bipectinate antennae are the most common type (e.g., figs. 82C, 82D, 144A, 144B), but the antennae of Dioptis (fig. 188C–F), Euchontha, and Xenomigia are quadripectinate. The ventral surface of each ramus bears two types of sensilla (fig. 151B)—flattened sensilla auricillica (fig. 151C) and basketlike sensilla coeloconica (fig. 151D). These structures, found throughout the Lepidoptera (Hallberg et al., 2003), were not used as characters in these analyses. In Scotura, Erbessa, and Pseudoricia the antennae are ciliate (figs. 11B, 11C, 36E, 36F, 37B, 37C, 256A), with no rami. Certain clades, such as Argentala (fig. 82A, B), Monocreaga (fig. 216C, D), and the Rubribasis Group of Polypoetes (fig. 92A), are characterized by antennae showing short, transverse flanges where the rami would typically occur. Such antennae are termed subserrate.

Characterizing dioptine genera on the basis of antennal morphology is often problematic; in certain cases, intrageneric differences are remarkable. For example, within the Truncata Group of Oricia, both ciliate and pectinate male antennae occur. The diversity of antennal structure found among Dioptinae easily matches that observed across the entire remainder of the Notodontidae, comprising approximately 2500 species (Gaede, 1934). The five character states used here to describe antennal pectination types (Character 28) are meant to reflect structural similarity above the generic level; they obscure a great deal of variation within genera.

At least among moths, female antennae exhibit less morphological diversity than those of males. The usual explanation for this is that males require more elaborate antennal sensilla with increased surface area so they can actively follow a pheromone trail to the stationary female. In butterflies both sexes are highly mobile. Typically, males actively search for females (e.g., see Silberglied, 1984), and females can travel large distances to find appropriate host plants for oviposition (Chew and Robbins, 1984). Antennae are important in both activities, and butterflies thus show less sexual dimorphism in antennal structure (Jordan, 1898; Chapman, 1899). There has been no research on mating systems in Dioptinae, but one might predict contrasting sensory needs in diurnal and nocturnal species. Interestingly, for most dioptine species in which the males possess pectinate antennae, those of females are pectinate as well (Characters 35, 37). Invariably, the rami are shorter in females (compare figs. 93C, D and 93E), but they are nonetheless present; in some Dioptis species (figs. 188E, F) and in Chrysoglossa phaethon, the female rami are nearly as long as those of males. It will be interesting to explore whether presence of pectinate female antennae is correlated with behavioral specializations.

Numerous, single-species scanning electron microscope studies in Lepidoptera have documented the remarkable diversity of antennal sensillum types (e.g., Faucheux, 1989, 1990, 1997). Particular attention has been paid to crop pests (e.g., Lin and Chow, 1972; Cornford et al., 1973; Cook et al., 1980; Lavoie-Dornik and McNeil, 1987). Flower and Helson (1974) were among the first to develop a system of sensillum nomenclature for the Lepidoptera. Comparative SEM studies among closely related species (e.g., Jefferson et al., 1970; Cuperus et al., 1983; Cuperus, 1985; Faucheux, 1993) reveal a rich source of character information. Such data can also provide important phylogenetic information at higher taxonomic levels, as has recently been shown for the Neopseustidae (Faucheux et al., 2006). As another example, the remarkable ascoid sensilla of Opostegidae and Nepticulidae—primitive members of the clade Heteroneura (Kristensen et al., 2007)—are an important synapomorphy uniting those families (Davis, 1989; Davis, 1998b; Davis and Stonis, 2007).

At least six different sensillum types are found ubiquitously in adult Lepidoptera (Hallberg et al., 2003). Perhaps the most conspicuous of these is the single, large sensillum styloconicum, typically located at the apex of each flagellomere (Fauchaux, 1990). This sensillum has been shown to respond to temperature and humidity (Hallberg et al., 2003). Cursory SEM study of antennae in the Dioptinae reveals interspecific differences in the size, shape and precise location of the sensillum styloconicum (see figs. 11D, 36E, 36F, 37A–D, 82D, 92A, 92B, 189A, 216E, 256B, 313A), variation that could potentially offer fascinating phylogenetic information for the group. However, while a comprehensive study of antennal sensilla in Dioptinae might yield a wealth of characters, such research for a species sample of the size employed here would be a huge undertaking—a study unto itself.

Other than pectination arrangements, I employed relatively few characters to describe differences in antennal structure. These involve traits such as bristle configuration (Characters 29, 38), and the presence of ventral ridges on the flagellomeres (Character 30). All can be observed with the aid of a dissecting microscope. A synapomorphy for the Josiini is presence of an unusual, hingelike junction occurring at the point where each ramus joins the antennal flagellomere (Character 33; fig. 312E, F). In many notodontid subfamilies, the pectinations are abruptly shortened before the antennal apex, with the terminal flagellomeres either simple or with short pectinations (Miller, 1991). This also occurs in several clades within the Josiini (Character 32; fig. 312D). The newly described genus Nebulosa is characterized by bipectinate antennae with long flagellomeres (Character 34), a condition that produces widely spaced pectinations (fig. 160C–E).

Three characters of the occipital region provide character information. One of these, presence of a narrow postocciput (Character 39), occurs in Tithraustes only (fig. 268A–C). Second, in all species examined the occipital condyles meet at the back of the head to form a small notch. The shape of this notch varies among genera (Character 40). For example, it is particularly deep in Chrysoglossa maxima (fig. 150C). Third, some Josiini, such as Ephialtias (fig. 311A, B) and Notascea (fig. 331A–C), possess tiny cuticular projections on the occiput, behind the antennal bases (Characters 41, 42).

The Lepidoptera tentorium, an internal supporting structure of the head joining the anterior and posterior tentorial pits (Snodgrass, 1935), serves as an attachment site for the cibarial and antennal muscles (Oseto and Helms, 1976). Ehrlich (1958a, 1958b), in his pioneering morphological works, showed the utility of tentorium morphology in butterfly phylogeny, and the character has subsequently proved useful in other Lepidoptera groups (e.g., Miller, 1987b, 1991, 1996). Tentorium shape varies in Dioptinae, with some species showing a marked swelling near the midpoint (e.g., Erbessa lindigii; fig. 35E), and others having a narrow, parallel-sided configuration (e.g., Momonipta onorei; fig. 211A). In addition, carinae can occur (e.g., Xenorma cytheris; fig. 53A), sometimes more than one for a given taxon. I used three characters of the tentorium (Characters 43–45) in an attempt to isolate homologous differences. The various carinae, identified based on their location, are often extremely thin, and in small taxa, are difficult to see. Although their utility is not practical from a taxonomic standpoint, for certain genera the tentorium provides excellent diagnostic traits.

Thorax

Considering their complexity and number, the thoracic sclerites of Lepidoptera seem to provide relatively few useful characters. This holds true for the Dioptinae. It might be judged that these structures are conservative from an evolutionary standpoint (Brock, 1971; Kristensen, 1984). General treatments of thoracic morphology include Weber (1924), a beautifully illustrated work, as well as Snodgrass (1935), Matsuda (1970), and Scoble (1992). My analyses show that the thoracic appendages (tegulae, legs, and wings) provide more character information for Dioptinae than the sclerites themselves. Internal structures, such as the furca and discrimen, are particularly complex (Scoble, 1992). Numerous authors have examined these hoping to find information for solving difficult phylogenetic questions in the Lepidoptera. The metafurca, in particular, has been heavily employed in higher-level studies (Ehrlich, 1958b; Brock, 1971; Kristensen, 1984; Davis, 1989). For the Dioptinae, characters of the internal skeleton can be found on all three thoracic segments (Characters 48, 49, 54, 64–66). These differences, almost all involving shape, are difficult to categorize into discrete character states. It is equally difficult to determine state assignments for certain taxa. Nevertheless, I employed these traits as best I could.

Lepidoptera hearing organs were apparently first discovered in the Uraniidae in 1889 (Eltringham, 1923). They show remarkable structural diversity in Lepidoptera, having evolved at least eight times within the order (Grimaldi and Engel, 2005). Presence of a metathoracic hearing organ in Noctuoidea, the largest and most complex superfamily in the order (Wagner, 2001; Mitchell et al., 2005), is often regarded as providing unequivocal evidence for the group's monophyly (Kitching and Rawlins, 1999; Fibiger and Lafontaine, 2005). Over the years, an army of entomologists has studied the morphology, physiology, and taxonomic significance of this structure. Among the earliest papers are those of Forbes (1916) and Eggers (1919). The relatively simple tympanum of Notodontidae, once viewed as a primitive precursor to the more complex organ found in other Noctuoidea (Richards, 1932), is now regarded as a derived specialization (Miller, 1991; Speidel and Naumann, 1995; Speidel et al., 1996b; Minet and Surlykke, 2003). Even in morphologically complex examples, the noctuoid tympanum can be viewed simply as a depression near the dorsal margin of the metepimeron. An auditory cell, or scoloparium, is attached to the surface of the tympanal membrane (Minet and Surlykke, 2003), which spans the upper margin of this depression. Additional structures, such as sound-reflecting cavities on A1 (the “counter-tympanal hood”) are associated with the hearing organ, and these are elaborated to varying degrees across the superfamily (Kitching and Rawlins, 1999).

The metathoracic tympanum has played an illustrious role in the history of dioptine systematics. In one of the first attempts to summarize the phylogeny of the Noctuoidea, absence of a tympanum in Dioptis led Forbes (1916, 1923) to regard the “Dioptidae” as the most primitive noctuoid group (see Character Evolution, under Discussion). Richards (1932) undertook a complex analysis of tympanal morphology to elucidate evolutionary relationships within the Noctuidae. He also looked at the tympana of Erbessa, Josia, and Cyanotricha ( =  Scea), as he assessed structure across additional noctuoid groups. It was he who first noted the unique tympanum in Josiini. Interestingly, Richards followed Forbes' early assessment that the “Dioptidae” were the stem group of the entire Noctuoidea.

Sick (1935) was a proponent of using tympanal morphology in Lepidoptera systematics. Later, it was he who, in a monumental study, first documented extensive variation in tympanum morphology across the Dioptinae (Sick, 1940). He examined the metathorax of 255 dioptine species, dividing these taxa into five groups based on differences in tympanal structure. However, Sick did not formalize his ideas in a revised classification. Using the five groups as terminal taxa, he did however produce a cladogram outlining his hypothesis of tympanum evolution. These theories were later modified by Kiriakoff, who considered the tympanum the most important character, not only for dioptine classification (Kiriakoff, 1950), but for understanding the phylogeny of the entire Noctuoidea (Kiriakoff, 1963). In retrospect, the concepts of Sick (1940) make considerable sense in terms of a modern dioptine classification, whereas Kiriakoff's theories can be largely ignored.

Here, five characters of the metathoracic tympanum are employed (Characters 56–60). Not only does size of the tympanal cavity vary widely, but there are also differences in size and orientation of the membrane itself. The remarkable kettledrum tympanum in Josiini has been the subject of considerable discussion (Richards, 1932; Sick, 1940; Miller and Otero, 1994; Miller, 1996), and needs little further elaboration. On the other hand, the research described here provides novel information regarding tympanum evolution in the Dioptini. For example, my cladistic results (fig. 3) support a broader concept of Dioptis than that of previous authors. According to this new information, the tympanum among Dioptis species shows a complete transformation series, ranging from presence of a well-developed tympanum in Dioptis longipennis to various stages of reduction, with complete loss restricted to one subclade within the genus (Dioptis trailii and relatives).

In other Notodontidae, all of which are nocturnal, the tympanum is thought to function as a system for detecting the echolocation signals of predatory bats (Fenton and Fullard, 1981; Fullard, 1984). Two papers have examined its function in Dioptinae (Fullard et al., 1997; Fullard et al., 2000), which theoretically would not be exposed to bats. The resulting picture is complex. These authors discovered that, while some dioptines possess auditory thresholds typical of Notodontidae, others show reduced hearing at bat-specific frequencies, and still others are completely bat-deaf (Fullard et al., 1997). This range in auditory response occurs in conjunction with a spectrum of flight-activity patterns (Fullard et al., 2000). Rather than being strictly “diurnal”, as they are often characterized, some dioptine taxa fly with considerable frequency at night as well as during the day. Others are indeed restricted to a day-flying habit, where the ears more likely function in intraspecific communication. In many tympanate moth groups where sound receptors originally served to detect bats, such as the Arctiidae, the hearing organs now also function in courtship (Krasnoff and Roelofs, 1990; Conner, 1999; Sanderford, 2009). A fascinating story, involving behavioral evolution in the Dioptinae and possible co-opting of the metathoracic tympanum for novel functions, remains to be told.

The metascutal bulla, a swelling located above the tympanum (figs. 12G, 326B), is found in almost all Notodontidae (Holloway, 1983; Common, 1990; Nielsen and Common, 1991). Brock (1971) considered it to be a diagnostic feature for the family. Miller (1991) highlighted variation in size and development of the metascutal bulla within Dioptinae, following Sick (1940) and Kiriakoff (1950). Four character states (Character 55) are used here to define these differences.

Three additional characters of the metathoracic sclerites were employed: The first (Character 61), probably unique to Notodontidae, is the presence of a sclerotized flange on the ventral portion of the metepisternum (Miller, 1991). This flange is poorly developed in most notodontid species, but can usually be seen upon careful inspection. In Dioptinae, the flange is absent in a few genera (e.g., Phanoptis), but is present in most. Its highest degree of development occurs in the Josiini (Miller, 1991; see fig. 326B). This structure has the general appearance of a sound-producing organ, a phenomenon that has not yet been reported in Josiini. Other Dioptinae are known to make sounds using specialized organs on the forewing (discussed below), but such behavior is probably not related to the metathoracic flange. The second trait (Character 62) occurs only in Josiini. Here, there is a broad, scaleless region immediately behind the tympanal opening (fig. 313B, C). The third character involves the shape of the metameron. This sclerite, usually a broad triangle, is dorsoventrally elongate in Dioptis, as well as in Tithraustes and some species of the closely related genera Isostyla and Stenoplastis (Character 63; fig. 187I).

The remaining thoracic characters I studied involve the various appendages. At the base of each forewing in Lepidoptera there is a moveable sclerite called the tegula. Among other functions, it may assist in wing coupling (Hering, 1958; Scoble, 1992) or to reinforce wing articulation (Matsuda, 1970). There has been historical confusion regarding the name of this structure. Some early lepidopterists called it the patagium (e.g., Schaus, 1894; Hampson 1898), but true patagia are membranous outgrowths of the prothorax (Ehrlich, 1958a; Kristensen, 1984). The tegula is usually large and is almost always covered with long scales, making it highly conspicuous in pinned specimens. A transverse sulcus divides the tegula into two parts, which I term the dorsal arm and the ventral angle. I characterized structural variation of the tegula using three discrete characters (Characters 50–52). Dioptinae exhibit a wide range of tegula size, from large, as occurs in Erbessa (fig. 38C, D, H) and Josiini (fig. 326B), to highly reduced, as in Dioptis trailii (fig. 187I). Intermediate sizes are found in less derived Dioptis species and in Monocreaga (fig. 215E). The tegula is universally large in Nystaleinae. A feature found in many Dioptinae (Character 53) is the presence of contrasting, orange-yellow scales at the tegula base (e.g., Momonipta onorei; pl. 21). This trait is often useful for field recognition, and was noted by many early authors in their species descriptions (e.g., Dognin, 1902; Druce, 1899, 1900; Schaus, 1912, 1913).

The epiphysis on the foreleg tibia, thought to be serially homologous with tibial spurs on the meso- and metathoracic legs (Michener, 1952; Kuznetsov, 1967), is a synapomorphy for the Lepidoptera (Kristensen, 1984; Kristensen and Skalski, 1999). Experiments and field observations confirm that the epiphysis functions as an antennal-cleaning device (Callahan and Carlysle, 1971; ODell et al., 1982; Robbins, 1989). Its surface is covered with spatulate processes (fig. 313D, E). The epiphysis is often smaller in females than in males, concordant with shorter pectinations in female antennae. The structure varies in shape and size throughout the Notodontidae, and provides a valuable taxonomic character for generic diagnoses (Marumo, 1920). It is absent in females of some notodontid genera (Miller, 1991). There is important variation within the Dioptinae (Characters 67, 68). For example, in Erbessa and Oricia the epiphysis is short and teardrop shaped (figs. 37F, 38G), much shorter than the foretibia, while in other taxa it is longer and lanceolate (e.g., figs. 189D, 189E), sometimes extending beyond the tibial apex. This difference might be correlated with antennal development; Erbessa and Oricia have ciliate antennae, and the smaller epiphysis may clean this type more effectively.

The meso- and metathoracic tibial spurs of all Dioptinae occur in the 2–4 configuration typical of Nystaleinae and most other Notodontidae (Miller, 1991), so spur number is not taxonomically informative. However, dioptines do show considerable variation in spur length (Character 69). Tibial spur length was stressed as an important character in understanding phylogeny of the Noctuidae (Speidel et al., 1996b). The sclerotized apex of each tibial spur is serrate in most Notodontidae (Janse, 1920; Jordan, 1923b; Miller, 1991; Weller, 1992), and all Dioptinae (figs. 189F, 301F, 313F).

Wings

Wing venation has been one of the most important character systems in the history of Lepidoptera classification. A possible explanation for this emphasis is that wing veins show a tremendous amount of easily accessible variation, requiring little or no dissection and preparation for study. Herrich-Schäffer (1843–1856) first stressed the utility of wing venation (Scoble, 1992), and devised a standardized system of numbers for each vein. His scheme was adopted by most early taxonomists (e.g., Hampson, 1898). Herrich-Schäffer's pioneering work greatly influenced the use of wing veins as characters, and has been incorporated into modern theories of Lepidoptera classification at the superfamily and family levels. Comstock's (1918) remarkably insightful analysis of insect wing veins has been more influential than any other work on the subject. The sytem developed by Comstock and Needham (in Comstock, 1918) was modified slightly by Kristensen (2003) in the Handbüch der Zoology, the most significant change involving the terms applied to the FW radial sector. The nomenclature of Kristensen (2003) is employed here.

Wing venation has been by far the most heavily utilized character system in development of the dioptine generic classification. Prout (1918) provided characterizations of all genera using wing vein traits, and based his generic key on them. Subsequent additions and changes to the dioptine classification (Hering, 1925; Bryk, 1930) relied on wing veins as well. Unfortunately, these specialists employed different wing-vein terminology. Hering (1925) largely followed Comstock and Needham, but Prout adopted a system of nomenclature different from the others in common usage at that time. Because of the historical complexity of wing-vein nomenclature and the usage of different systems by various authors, the study of dioptine taxonomy, especially interpretation of the early literature, requires an ability to convert from one system to another. Köhler (1930) provided the first illustrations of dioptine wings, showing wing venation for species in three genera. Lima (1950) later figured the wing venation of Erbessa pyraloides in his treatment of the Brazilian fauna.

Many of the wing-vein differences in Dioptinae involve the forewing radial system. As in other Lepidoptera, dioptines exhibit five FW radials, comprising R₁ and four branches of the radial sector: Rs₁–Rs₄ (e.g., fig. 8). Throughout the subfamily, vein R₁ invariably arises from a point well out on the discal cell. The veins of the radial sector, on the other hand, vary widely in their branching patterns. Vein Rs₁ is typically conjoined with Rs₂–Rs₄ (Character 70). In Erbessa (fig. 38A, B, E, F) and Phaeochlaena (fig. 70F), on the other hand, Rs₁ arises from the DC, the placement found throughout the outgroup. Overall, the most common radial vein arrangement in the Dioptinae is exemplified by the forewings of Xenorma (fig. 53F), Phryganidia (fig. 60C), and most Josia (fig. 336F, H). Here, Rs₁ is stalked with Rs₂–Rs₄, and the latter are in the pattern [2+3]+4 (Character 72, State 1). This configuration is not found elsewhere in the Notodontidae (Miller, 1991), but occurs in roughly 70% of dioptine species (Miller, 1987a). Other branching patterns (Character 72, States 2 and 3) are less common, but can provide important diagnostic features for particular dioptine genera, or for subclades within genera. In members of two closely related genera—Dolophrosyne (fig. 221F) and Xenormicola (fig. 232C)—the radial sector is comprised of only three veins (Character 71).

The discocellular veins close the distal margin of the DC. Within Dioptis there is considerable variation in the length of the UDC and MDC relative the length of the LDC (Characters 73, 74; fig. 190D–G). These differences do not occur elsewhere in the Dioptinae. Another apomorphic feature of the FW venation restricted to Dioptis is the presence of a strongly curved radial sector (Character 76).

The FW medial system shows less variation than the radials, but nevertheless offers several important characters. Vein M₁ can arise from near the anterolateral angle of the discal cell (e.g., Nebulosa; fig. 158A, 158C–F), separate from the radials, or it can be stalked with the radials (Character 75), arising well out from the DC (e.g., Proutiella; fig. 284C). This character state is found in many Josiini and provides a useful key character for identifying genera. A unique, and easily recognizable, diagnostic feature for a great many species of Dioptinae is fusion of FW veins M₃ with CuA₁ (Character 85; e.g., figs. 12A–D, 84F, 255F, 306B). This, one of the most useful field traits for distinguishing dioptines from other moths of similar habitus, occurs in 30 genera comprising roughly 75% of the species. Based on outgroup comparison, the plesiomorphic state is where M₃ is approximate to, but not fused with, CuA₁ (e.g., Cleptophasia; fig. 8F). Dioptines in which M₃ is separate from CuA₁ have often been accorded generic status by previous authors on that basis alone. For example, Prout (1918) erected both Xenorma (fig. 53F) and Phanoptis (fig. 64K) for having M₃ and CuA₁ “widely separate” in the FW. Other FW veins are of limited value; the position of FW vein M₂ does not vary, nor does the configuration of the cubital or anal veins.

Nijhout and his colleagues (e.g., Nijhout, 1991, 2003; Nijhout and Wray, 1986, 1988) are largely responsible for spearheading what can only be termed a renaissance in the use of lepidopteran wing patterns for systematic studies. In a recent example, Willmott's (2003) phylogenetic analysis of species in the nymphalid butterfly genus Adelpha employed 114 morphological characters, of which 83 were derived from detailed study of wing pattern. The incredible variety of wing patterns among Dioptinae elicited amazement from early authors (e.g., Köhler, 1930), and the more I learn about the group the more amazed I become. Dramatically different patterns and colors can occur in closely related species, an example being the genus Erbessa, where skeletal morphology is remarkably uniform but wing patterns vary tremendously. As in other diurnal Lepidoptera, the evolution of mimetic resemblance seems to be the driving force behind this phenomenon (Seitz, 1925). One of the examples Bates (1862) used in formulating his theory of mimicry was the near perfect resemblance of some Dioptis species to ithomiine butterflies.

Both day-flying and nocturnal habits occur in the Dioptinae (Fullard et al., 2000), and although experimental work remains to be done, my own observations suggest that wing coloration tracks this trend to some extent. Bold patterns and bright colors tend to occur in the wings of diurnally active species, examples being the yellow transverse bands of Phaeochlaena costaricensis (pl. 9), or the striking orange stripes of some Josiini (e.g., pl. 32), moths collected frequently in the day but almost never at lights. On the other hand, the subdued, intricate patterns of Chrysoglossa (pl. 15) and Xenomigia (pl. 23) seem to indicate nocturnal habits; these taxa are commonly attracted to lights.

In this paper, my goal was to search for generality in wing-pattern characters, rather than focus on the numerous differences. I looked for homologous pattern elements (sensu Nijhout, 1991) using position relative to particular wing veins as a guide. Previous research had shown this to be a successful methodology. For example, superficially identical longitudinal forewing stripes have evolved at least twice by convergence in the Josiini (Miller, 1996). Upon careful inspection, the two pattern types can be distinguished by their position relative to the radial and cubital veins (see Character 82). A similar approach was applied throughout. At first glance, forewing spots seem to occur in an almost limitless number of ways. However, in searching for generalizations, I identified three types of subapical spots on the forewing (Characters 83, 84), using position relative to the radial and M₁ veins in assigning character states. My analysis of dioptine pattern characters is rudimentary. A more fine-tuned study using species-level phylogenies to elucidate transformation hypotheses would be a fascinating endeavor. In my opinion, the Dioptinae provide one of the most intriguing cases of pattern evolution to be found in the Lepidoptera.

An important wing-pattern trait, unique to Dioptinae among notodontids, characterizes a great many species in several genera. Here, the scales along the forewing veins are more lightly colored than the wing ground color (Character 87). Often, as in Brachyglene (pls. 14, 15), Nebulosa (pls. 16, 17), and Xenomigia (pl. 23), the scales along the veins are light brown whereas the ground color is a darker brown. In a large number of Polypoetes species (pls. 10–14), this trend produces striking patterns, where the forewing veins are outlined in orange or yellow scales, while the remaining ground color is blackish brown or black.

The fine structure and taxonomic value of Lepidoptera wing scales have been well documented over the years (e.g., Kellogg, 1894; Mayer, 1896; Kristensen, 1970; Ghiradella, 1984; Scoble, 1992; Simonsen, 2001). A typical wing scale is broad and flat (fig. 290E), with a narrow pedicel at its base that fits into a socket in the wing (fig. 290F). The scale's distal margin is often broadly dentate. Downey and Allyn (1975) devised categories of scale shape, and developed a comprehensive nomenclature to describe scale morphology. Most importantly, they documented the vast array of wing-scale types that occur in the Lepidoptera. These types can provide important phylogenetic information when surveyed across a group. Yen et al. (2005a) provided a particularly elegant use of wing-scale morphology to aid in understanding phylogenetic relationships within the Chalcosiinae (Zygaenidae). In the current study, I limited my focus to wing scales occurring in specialized regions of the wing. For example, some Josiini, such as Polyptychia, possess complex alar androconial systems (figs. 299–301), with associated specializations in wing venation (fig. 298E).

Wing transparency has evolved multiple times in the Lepidoptera (Punnet, 1915; Kristensen and Simonsen, 2003), often by more than one means within any given monophyletic group. Kristensen (1974) documented such an example in the Sesiidae. Within that family the clear-winged trait has evolved via two pathways: In some species, the pharate adult emerges with a full complement of pigmented scales that fall off during the first wing movements. In other sesiids, the wings are covered with scales that are themselves transparent, giving a clear appearance. Similarly, wing transparency in Sphingidae is obtained by two means (Kitching, 2003): Hemaris and Cephonodes exhibit the scale-shedding mechanism, whereas in the Cocytius Group, hind-wing transparency is achieved through narrow, depigmented scales that expose the wing membrane.

The Dioptinae showcase a third means for producing hyaline wings, exemplified by Phanoptis (fig. 65), Dioptis (fig. 191), and Monocreaga (fig. 217). Here, the wings are covered with extremely thin, gently curved scales (figs. 65A, 65B, 191B, 217B, 217C). Furthermore, these are reduced in number and are spaced widely apart. In contrast, the wing veins themselves are lined with broad scales (figs. 65A, 191A, 217A, 217B). This means of achieving hyaline wings occurs throughout the Dioptinae, but appears to have evolved a number of times within the subfamily (see Discussion: Wing Pattern). An intermediate arrangement can also be found whereby the wing surface is covered by long, thin scales intermixed with more typical flat scales, thus producing semitransparent or translucent wings. In all dioptine genera in which transparent wings occur, there are no discernable morphological differences in the specialized scales.

Within the Dioptinae, hyaline wings are found universally in Dioptis (pls. 17–21) and Monocreaga (pl. 21), whereas translucent wings occur in some species belonging to Dolophrosyne and Xenomigia (pls. 21–23). Phanoptis (pls. 8, 9) and Hadesina (pl. 14) are unusual in containing species with clear wings, as well as ones where the wings are fully scaled. In Tithraustes and Isostyla (pls. 24, 25), the wings can be either hyaline or translucent. I employed two derived character states to describe these features in the Dioptinae (Character 94). In one, the covering includes both thin and broad scales, resulting in translucent wings. The second state, which gives rise to hyaline wings, is characterized by the presence of long, thin scales only. The hind wings were used to score this character since, as in Sesiidae (Kristensen, 1974; Edwards et al., 1999), transparency is more pronounced there.

The number of bristles in the female frenulum, part of the wing-coupling mechanism, varies in Lepidoptera (Marshall, 1922; Braun, 1924). These differences hold phylogenetic information for Notodontidae, useful at the subfamily level (Miller, 1991). The basic dichotomy in Dioptinae is presence of two frenular bristles, the plesiomorphic condition, or more than two (Character 100). Only two bristles occur in Nystaleinae, Josiini, and Scotura. Species with the multiple condition, which includes the vast majority of Dioptinae, usually have approximately 10 bristles in the female frenulum. Three unrelated taxa—Euchontha, Scotura transversa, and Phavaraea dilatata—are unusual in possessing three bristles.

Among Notodontidae, the presence of a stridulatory organ in the male forewing (Characters 77–79) is unique to the Dioptinae. In taxa where this structure occurs, veins M₁ and M₂ are swollen in the region immediately beyond the discal cell (e.g., figs. 145F, 162A, 162C–F, 255F), protruding from the wing's ventral surface (figs. 144C, 161C, 256E). The structure was noted by early dioptine taxonomists, and was first discussed in detail by Forbes (1922, 1931, 1939a). Later, Miller (1989) described it using SEM. At that time I suggested two possible functions for this organ—one being that it is androconial in nature since it is found only in males and the associated wing scales are unusual in shape, and the second being that it is used to produce sounds. Careful subsequent reading of Forbes' writings revealed that, during an expedition to Peru, he had witnessed stridulation in the dioptine Euchontha frigida (Forbes, 1922: 72). More recent field observations for Euchontha in Ecuador (J.S. Miller and E. Tapia, personal obs.; F. Sperling, personal commun.) confirm that these moths are capable of producing clicking sounds, undoubtedly using the forewing organ. The clicking sounds in Arctiidae, generated by a metathoracic tymbal organ (Forbes and Franclemont, 1957; Conner, 1999; Kitching and Rawlins, 1999), are a form of auditory aposematism (Conner et al., 2009). Arctiid clicks advertise toxicity, and thus confer protection from nocturnal bat predators (Watson, 1975; Fenton and Fullard, 1981; Dunning and Krüger, 1995, 1996). Sounds in Dioptinae, on the other hand, are most likely used for intraspecific communication, since in all observed cases they are produced during diurnal flight.

Altogether, nearly 180 dioptine species in 22 genera posses a FW stridulatory organ (see Discussion: Character Evolution). This organ thus represents a case analogous to the male forewing organ found in two separate subfamilies of the Noctuidae—the Heliothinae (Matthews, 1987, 1991; Mitter et al., 1993) and Agaristinae (Bailey, 1978; Common, 1990; Rawlins, 1992; Talianchich et al., 2003). In both groups, males stridulate during diurnal, premating, and territorial behaviors.

The dioptine forewing organ varies in its level of morphological complexity. In its simplest form (e.g., Cleptophasia scissa and Xenomigia), the discal cell is not short, but veins M₁ and M₂ are slightly swollen (figs. 8F, 235G, H). In taxa such as Pseudoricia (255F), Euchontha (fig. 249C), and Hadesina (fig. 137F), the stridulatory organ is highly complex, involving several distinctive structural modifications. Here, the forewing discal cell is extremely short—less than one-fourth the FW length; veins M₁ and M₂ protrude markedly from the wing ventral surface; there is a well-developed secondary fold between M₁ and M₂ (fig. 256E, 257C); and the area beyond the discal cell is semitransparent, forming a fascia. In addition, the wing surface of this region is distinctly corrugated (fig. 257C, D). Derived members of Dioptis are unusual in that veins M₁ and M₂ are not swollen, the wing surface shows no signs of corrugation, a fascia is not present beyond the discal cell, but the discal cell is short (fig. 190E, F); more basal members of the genus, such as D. subalbata (fig. 190D), show a typical stridulatory organ. My phylogenetic analyses suggest that this is a case of derived loss of the stridulatory organ. It is interesting that there are Dioptis species in which the membrane of the metathoracic tympanum is completely absent (fig. 189B, C), a situation rare in the Noctuoidea and unique to Dioptis among Dioptinae. Loss of the tympanal hearing organ and the forewing sound-producing structure in the same genus may reflect correlated function. Whether these moths can make sounds is unknown.

Most species of Polypoetes possess a small forewing fascia beyond the discal cell (Character 80). The fascia is covered with thin scales, making it translucent, or, as in P. forficata and relatives, almost clear (pl. 12). My cladistic results suggest that the fascia in Polypoetes is not homologous with the one occurring in species possessing a stridulatory organ. There are morphological differences between these forewing structures; the discal cell is never short in Polypoetes, and veins M₁ and M₂ are never swollen. The functional significance, if any, of the forewing fascia in Polypoetes is not known. It almost always co-occurs with an analogous hyaline area on the hind wing (Character 93, below).

The vast majority of Lepidoptera exhibit heteroneuran wing venation, in which the system of hind-wing veins is reduced compared to that of the forewings (Common, 1990; Scoble, 1992). Concordantly, the hind wing shows less variability in venation. In Dioptinae, the most important hind-wing vein character is whether M₃ and CuA₁ are fused or separate (Character 89). Fusion (e.g., Scotura; fig. 12A–D) is derived, and as in the analogous FW character (Character 85), this provides one of the most useful diagnostic field traits for Dioptinae; it is found widely across the subfamily. The only other useful hind-wing venation characters appear in males of some Josiini, where there is fusion of Rs and M₁ in Getta and relatives (Character 99). In these same taxa, hind-wing shape is modified due to the presence of specialized androconial patches (Miller, 1996; and see below).

A diagnostic feature for some subclades within Polypoetes, such as the Rufipuncta Group, is the presence of a hind-wing fascia (Character 93) located immediately beyond the DC (fig. 95B). The fascia varies in size and shape across species, and is usually visible in pinned specimens. SEM reveals that the scales of this HW fascia are small and somewhat concave (fig. 95C–F), and widely spaced on both the upper and lower wing surfaces. Since this trait occurs in both females and males and lacks associated modifications, it probably does not serve a stridulatory or androconial function.

Androconia, specialized scales in adult males usually involved in the dissemination of close-range scents (Birch et al., 1990; Schneider et al., 1992), are widespread in Lepidoptera (Scoble, 1992). They are especially prevalent among butterflies (e.g., see Scudder, 1877; Eltringham, 1913, 1915; Vane-Wright, 1972; Boppré, 1984; Vane-Wright and Boppré, 1993). Androconia can occur on a wide range of body parts. For example, in Riodinidae they are found on various locations of the wings and abdomen, as well as on the legs and genitalia (Hall and Harvey, 2002). Androconia also show considerable diversity in many moth groups (e.g., Grant, 1971, 1978). The noctuid subfamily Herminiinae is a prime example; there they show dramatic development on the labial palpi and forelegs (Smith, 1895; Kitching, 1984a; Owada, 1987). The spectacular abdominal coremata of male Arctiidae, located between the sterna of segments 7 and 8 and bearing long, hairlike androconia, have been described and figured in numerous publications (e.g., Boppré and Schneider, 1989), and their involvement in courtship has been thoroughly documented (see Weller et al., 1999). In the Notodontidae, elaborate androconia occur on the forelegs of some Nystaleinae (Weller, 1992). Scent scales on the valvae of the male genitalia are found in most notodontid species (Miller, 1991), including the majority of Dioptinae (see discussion of Barth organ, below).

Perhaps the most conspicuous androconial organs in Lepidoptera are those on the wings. Alar androconia have been discussed and figured in considerable detail for certain butterfly groups, such as the pipevine swallowtails (Papilionidae: Troidini) (Zeuner, 1943; Miller, 1987b; Racheli and Pariset, 1992; Racheli and Olmisani, 1998), milkweed butterflies (Nymphalidae: Danainae) (Eltringham, 1913, 1915; Ackery and Vane-Wright, 1984; Boppré‚ and Vane-Wright, 1989), and Riodinidae (Hall and Harvey, 2002). Usually the wing scales within the androconial patch are highly modified, and species-specific differences in shape have been reported (Ackery and Vane-Wright, 1984; Miller, 1987b; Racheli and Olmisani, 1998).

In Dioptinae, alar androconia occur in a single lineage of the Josiini (Clade 19; fig. 7) comprising three genera—Getta, Polyptychia, and Phavaraea—and totaling 12 described species. However, a surprising variety of androconial configurations is exhibited by the members of this small clade (Characters 90, 92, 101). For example, in Getta there are two areas of specialized scales: one on the dorsal surface of the hind-wing anterior margin, and the other on the ventral surface of the forewing posterior margin (fig. 289H). These two patches apparently overlap during flight. Remarkably similar overlap-type alar androconia occur in certain genera of the Arctiidae (Watson, 1975). The closely related josiine genus Phavaraea possesses different alar androconia. Here, the anal margin of the hind wing is expanded into a rolled, dorsal pouch (fig. 306A, B) enclosing short, cream-colored scales and long, bristlelike scales. Superficially at least, this organ is similar to the hind wing androconia of Battus butterflies (Papilionidae: Troidini) (Miller, 1987b). Polyptychia has an androconial fold along the hind-wing anal margin similar to that of Phavaraea, but also exhibits a second pouch located between hind-wing veins M₂ and CuA₁+M₃ (fig. 298E).

Alar androconia in these dioptines are often associated with specialized scales on other body regions. In Phavaraea dilatata the hind-wing anal margin is greatly expanded (fig. 306B), enclosing woolly, deciduous androconia. There is, in addition, a set of robust, black bristlelike scales on the pleuron of A4–A7, as well as a patch of elongate scales on the dorsum of A5–A8. Polyptychia is unique in exhibiting a tuft of long, white hairlike scales on each tibia (pl. 27), in addition to the alar organs (figs. 299–301). In Phavaraea rejecta and P. poliana (pls. 27, 28) the entire venter of the abdomen is covered with a layer of short, iridescent scales, possibly androconial in nature. For at least some of these dioptine species, the nonalar scales may interact physically and chemically with the ones on the wings. Such a situation has been documented for the Queen Butterfly, Danaus gillipus, in which mechanical contact between abdominal hairpencils and hind-wing pockets is required for proper biosynthesis of the male pheromone, danaidone (Boppré et al., 1978).

In summary, Clade 21 of the Josiini has evolved a diversity of androconial morphology that nearly encompasses that observed across the entire Papilionoidea. The biological reason for this phenomenon is unclear. It is curious to note that the caterpillars of these same josiines are the only ones associated with the primitive Passiflora subgenus Astrophea (Miller, 1996; table 6). Further research on this remarkable system might provide insights into the evolution of androconia in Lepidoptera.

Abdomen

(both sexes, A1–A6)

Several specialized features of the notodontid abdomen, such as the sclerotized cup surrounding the spiracle on A1 (Jordan, 1923b; Fullard, 1984; Kitching and Rawlins, 1999) and the cteniophore on sternum 4 (Jordan, 1923a; Weller, 1992; Miller, Janzen, and Franclemont, 1997), are absent in Dioptinae (Miller, 1991). However, an obscure synapomorphy for the subfamily is the presence of a small patch of microsetae, set on a lightly sclerotized transverse bar in the pleuron of A1, anterodorsal to the spiracle. This structure, figured and discussed in Miller (1991), is of unknown function, but seems to occur throughout the Dioptinae (Character 106).

The A1 sternum is highly reduced in primitive Lepidoptera and absent in Heteroneura (Scoble, 1992; Kristensen and Skalski, 1999). Character 102 details various types of “windows”, or fovea, found on the anterior portion of sternum 2, using four states. These are presumably derived modifications of the fully sclerotized condition, but the character shows considerable homoplasy and is somewhat difficult to interpret. Sternum 2 typically bears apodemes on its anterodorsal corners, a feature found throughout the Ditrysia (Kyrki, 1983). Within the Dioptinae, these vary in thickness and shape (Characters 103, 104). The anterior apodemes on sternum 3 are similarly variable (Character 107). Tergum 1 is membranous, but on tergum 2 the anterolateral apodemes vary in their thickness and degree of curvature (Character 105).

Abdominal color patterns provide important characters, especially in the Josiini (Forbes, 1931). For example, presence of a band of orange scales across the dorsum of A1 (Character 108) defines a small subclade of Lyces that includes L. annulata and L. minuta, the latter formerly placed in its own genus—Leptactea. Lateral stripes and dorsal spots occur in Ephialtias (Characters 109, 110), and the pattern on the abdominal venter (Character 111) is an important synapomorphy for the new genus Proutiella.

The final character from the abdomen, other than features of the terminalia, is the shape of the A2–A6 sterna (Character 112), which are trapezoidal and wider at their posterior margin in a clade of high Andean genera that includes Scoturopsis, Dolophrosyne, and Xenormicola.

Terminalia

(A7–A10)

The male and female genitalia of Dioptinae show tremendous structural variation. This is reflected by the large number of genital characters utilized for the phylogenetic analyses in this study; from a total of 305 adult morphological characters, 193 of them, approximately 63%, are genital. Rather than describe the innumerable differences in genitalia morphology across the Dioptinae, below I outline major trends. A more detailed survey can be inferred from the character list (appendix 1), the generic diagnoses, the species descriptions, and the figures shown in this paper. Throughout the history of Lepidoptera systematics, female genitalia have been less frequently utilized than those of males. They provided fewer characters in the present study (68 female, 124 male). Nevertheless, many of the most reliable traits for diagnosing genera and species can be found in the female external genitalia. As is well known to Lepidopterists, while the basic elements of genital morphology can usually be identified with certainty, homology assessments for the myriad of specializations can be extremely problematic (Epstein, 1996). Here, similarity of location (Remane, 1952; Hennig, 1966) was the guiding rule. Regardless, certain areas were difficult to homologize.